Gender- and SES-Specific Disparities in Shadow Education: Compensation for Boys, Status Upgrade for Girls? Evidence From the German LifE Study

In the present article, we draw on social reproduction theories to explain the increase in the use of “shadow education” (SE) in Germany over the last two decades as a status-based, gender-specific investment strategy of families. Thus, we ask whether investing in private tutoring for both girls and boys alike serves to maintain or improve their status position, or whether gender-specific investment strategies exist. Our hypotheses are quantitatively tested by means of logistic regression using data of the 2012 German Life Courses into Early Adulthood (LifE) study. In contrast to prominent beliefs, our results show that SE in Germany does not function as a tool to promote social inequality. Instead, SE is used largely independent of social origin to achieve higher educational credentials. We found that particularly boys from non-academic but high-income families use SE, whereas girls seem unaffected by social origin entirely. Our findings call for further strengthening the gender dimension in existing inequality theories.


Editorial Throwing Light on Shadow Education
This special issue is concerned with the growing global phenomenon of supplementary tutoring, which takes distinct forms in different societies and sometimes even transcends national boundaries. As the title of this issue we chose a metaphor wellknown in the field − shadow education − to highlight how private tutoring often shadows or mimics the operation of the formal school system (see the origin of the term in Marimuthu et al., 1991;Stevenson & Baker, 1992;Bray, 1999), we do however recognise that sometimes the content it covers, its aims and purposes, may not coincide with those of formal education. The belief that schools alone − and formal education in general − cannot fulfil all of a student's learning needs, creates the demand for tutoring worldwide. Much of student learning takes place outside of traditional schools, some may be of an academic orientation, while other times it may focus on non-academic activities or offer a hybrid model. We recognise that the reasons behind this growing phenomenon are not merely educational. It depends, to a significant degree, on the socio-cultural, political and economic context in which it operates; on the identities of the providers and consumers of the service; on educational structures, opportunities and barriers existing within a given context; on the ambitions, aspirations, social and economic capital of the parents; on the educational and cultural values embodied in different societies; on the readiness and willingness of students to participate in such educational opportunities; on the various academic and social pressures that students experience throughout the process, among other things.
In addressing these important aspects of the topic, this special issue is a renewed response to the call to turn our attention towards educational markets broadly speaking and private tutoring, more narrowly. UNESCO's upcoming Global Education Monitoring Report (GEM) 2021 on non-state actors in education also recognises the increased role that private providers play in the public education system. The growing interest in private tutoring among scholars around the world indicates that this is an important area of inquiry. At the same time there is an ever-increasing body of literature on the privatization of education and the conceptualization of public and private spaces of learning, often with blurred frontiers. Bibliographic research demonstrates that scholars have generated significant research literature not only in Asia, which has traditionally been a major area for tutoring, but also in Europe (Bray, 2020). This positive tendency has led scholars to develop comprehensive thematic literature in the form of special issues in scholarly journals, such as the Journal of Education Research Online (Guill & Spinath, 2014), the Asia Pacific Journal of Education (Manzon & Areepattamannil, 2014), the East China Normal University Review of Education (Zhang & Bray, 2019), or the European Journal of Education (Győri, 2020).
While these special issues have been valuable in the analysis of the forms, practices and meanings of shadow education, they have focused on particular countries and jurisdictions, methodologies and approaches. There remains a need to find new ways of conceptualizing what shadow education is and is not, going beyond the metaphor, filling in the empty places in the world geography and expanding and clarifying the boundaries of meaning. New, robust studies in the field provide opportunities to address old questions and to raise new ones. New approaches and conceptualizations are especially important in the light of the changing educational situation following the COVID-19 pandemic, which has altered the context of education globally, including private tutoring. This special issue builds on the growing academic interest in shadow education, and benefits from many previous contributions by scholars on the subject. We hope the four studies that we have included in this special issue will open up meaningful scholarly conversations and advance the field.

Special issue highlights
To begin with, all four of the articles, authored by both experienced as well as emerging scholars, draw attention to the continued significance and importance of shadow education in the lives of students, teachers and parents. These studies recognize that the demand for shadow education has only been increasing and that there is a need to understand the underlying forces, reasons and motivations.
By situating the analysis of tutoring (known as Nachhilfe) in a German context, Entrich and Lauterbach engage in a discussion of the relationship between tutoring and students' socio-economic status (SES), something which has been considered responsible for exacerbating social inequalities. This paper challenges the prominent assumption that shadow education serves economically advantaged families as an instrument of social exclusivity, instead positing that tutoring provides a compensatory mechanism to improve low or average academic performance. The authors found differentiated patterns of use according to gender and social origin, namely boys from non-academic, but high-income families are more likely to use shadow education, whereas girls seem unaffected by social origin. This paper shows that group-differentiated in-depth analyses of quantitative data may unveil hidden patterns that are not obvious at a first glance, because the predictors may (inter) act differently for different groups.
While many previous papers in the field of shadow education often tended to stress the problematic aspects of the shadow education phenomenon, Jansen, Elffers and Volman highlight the ways in which it helps Dutch students' learning and strengthens their sense of well-being; the authors argue that tutoring, existing between the school and home, stimulates students' attention-spans and learning productivity, and thus maintains a symbolic meaning for students as a "third place" (Oldenburg, 1999). The conceptualization of shadow education as a third place in the lives of students provides a fresh perspective and evidence of positive experiences, because it contrasts with so much of the previous work that portrayed shadow education as exam-focused and stress-related.
Teaching and learning English − today's lingua franca − as a foreign language, is an important topic that receives considerable attention from scholars worldwide, including those focusing on the specifics of private tutoring of the English language. By retrospectively studying the biographical experience of Czech learners of English, covering the participants' lives from early childhood to their entry into tertiary education institutions, Černá has demonstrated the importance of the private tutoring phenomenon throughout the course of a person's life, showing that the roles, functions and nature of private tutoring change as the learner's life situation changes. Whilst many studies in the field of shadow education focus on the characteristics of the phenomenon at one specific time point, Černá's paper is unique in the sense that it takes into account a long-term perspective on learning English and accessing private tutoring.
Khaydarov has provided nuanced and contextualised insights into the way shadow education has become an integral part of the lives of students and teachers at academic lyceums in Samarkand, and demonstrates how private tutoring is embedded within the situated context of Uzbekistan. The author finds contrasting evidence: on the one hand, tutoring serves as a catalyst for teachers' professional development, on the other hand, it also drains energy from these same teachers, who sometimes prioritise tutoring over lyceum teaching due to its lucrative nature. This is an important finding, demonstrating the clash between teachers' professional values and market values. Crucially, the paper also touches the as-of-yet uncharted phenomenon of "shadow education inside schools", i.e. the institutionalized and paid tutoring in academic subjects which is provided officially by the school (schoolteachers) within the school premises. This is a good example of the "hidden privatization in education" (Ball & Youdell, 2007) that has been documented in other parts of the world (Bray, Kobakhidze, Zhang, & Liu, 2018).
After outlining the highlights of these papers in our special issue, in the following paragraphs we would like to position them within the wider context of shadow education as a field of study with regards to their geographic coverage, methodology and central concepts.

Geographic coverage
In terms of geographic groupings, three studies in this special issue represent the European region (Germany, Netherlands and the Czech Republic) and one study represents Central Asia (Uzbekistan). If we take a historical perspective, the Czech Republic and Uzbekistan may be grouped under post-Soviet or post-Socialist countries. While the Czech Republic, formally known as Czechoslovakia, was a satellite state of the Soviet Union, Uzbekistan constituted an integral part of USSR as the Uzbek Soviet Socialist Republic until the collapse of the Soviet Union in 1991. Variability in the socio-economic, cultural and political contexts influenced the phenomenon of shadow education with regard to its forms, aims, purposes and driving forces. As is evident from the articles in this special issue, private tutoring took a very distinct form in each country. We are delighted to bring to you new research evidence from countries in which shadow education is not yet well documented, i.e. the Netherlands and Uzbekistan, thus reducing the number of "uncharted territories" on the global map of shadow education research.

Methodological considerations
Researching shadow education is not easy given its sensitive nature, and the related challenges are evident from the articles in this special issue.
Investigating existing datasets has many advantages, but at the same time many limitations, because the phenomenon of shadow education is not always the main focus of the research studies these datasets come from, and they may therefore lack information on important aspects of shadow education. In this special issue, two of the four papers employed secondary data analysis.
Entrich and Lauterbach located and analysed a rich quantitative dataset from a longitudinal German study that linked together information from parents and their children. This allowed them to get a more complex picture of the issue under investigation, while at the same time being able to distinguish between paid and unpaid tutoring, something other quantitative datasets (such as PISA or TIMSS) rarely allow. The authors did, however, recognise minor limitations originating in the temporal structure between dependent and independent variables.
While secondary analyses of quantitative data are becoming relatively common in the shadow education literature, re-analysing qualitative data is still something of an exception, in part also due to the "intimate" nature of the data and related ethical considerations (Ruggiano & Perry, 2019). In this context, Černá's study is novel in its approach, utilising qualitative data that were not collected primarily with the intention of investigating the shadow education phenomenon, but instead focused on the broader topic of learning English. Knowing the context and background of the primary study and being one of its data collectors helped Černá overcome some of the limitations of secondary qualitative data analysis. Readers will also learn how the implementation of strategies such as reinforcement feedback, attention-focusing devices and various probes helped her collect narratives from participants.
Khaydarov noticed that among respondents, some teachers and school principals were unwilling to provide honest answers and avoided the questions. He explained this by providing the historical and socio-cultural reasons behind such behaviours and also by elaborating on how issues related to teacher salaries, education quality and private tutoring can be politically sensitive in Uzbekistan. At the same time, the study points to the value of inside-outside legitimation (Onwuegbuzie & Johnson, 2006), showing the importance of not being "caught" in the "insider" perspective, while stressing the advantage of being confronted by an "outsider" during every phase of the research.
The importance of a specific physical place was also stressed by Jansen, Elfers, and Volman − not only in relation to tutoring − this was also projected in their methodological approach. In order to strengthen a feeling of safety in their informants, they always chose a "neutral ground" for the qualitative interviews, never interviewing people in their own home or school. Nevertheless, they reported problems with very short answers from participating students and so developed interactive probing techniques to create prompts to help the students speak up.
As can be seen from the examples above, the papers in this issue, although not primarily concerned with the methodological aspects of shadow education research, also contribute to the wider methodological literature and may help beginning as well as experienced scholars in designing and conducting their investigations of shadow education.

Definitions and foci
There are important questions to be addressed and much ground yet to cover, however one of the central concerns still lies with the question of definition. The diversity of educational experiences under the broad umbrella of shadow education or tutoring creates challenges when classifying and categorising its multiple forms.
What is and what is not shadow education? While some authors understand it merely as tutoring in academic school subjects in addition to regular school instruction for a fee, others may also include non-academic subjects, fee-free tutoring or activities other than tutoring, such as learning from pre-scripted online tutorials without the assistance of a tutor or learning using Artificial Intelligence (Kobakhidze & Suter, 2020).
Papers in this special issue explore the "traditional" shadow education forms and types. All four papers focus on tutoring provided "live" by a tutor either to individual students, or in small or larger groups. To distinguish these two cases, Dutch language even uses a special terminology − bijles for one-on-one tutoring and huiswerkbegeleiding for group sessions, which also offer homework support for students. Černá focused almost explicitly on "classic" private tutoring lessons in English, but also mentioned other ways her informants learned English (e.g., finding a pen friend to practice English with), some of which could fit into "extended" definition of shadow education. Something that is particularly challenging to define and conceptualize may be shadow education provided by "unusual" subjects. As already mentioned, Khaydarov found that some paid tutoring lessons were provided officially by the lyceum which was his research site. However, as the lyceum was a public school, it was unclear how appropriate it would be to call these lessons "private tutoring"? All four papers focus (explicitly or implicitly) on tutoring in academic school subjects, leaving other extracurricular activities (such as hobbies or sports) aside. And finally, all four studies deal primarily with paid tutoring, although some (e.g., Entrich and Lauterbach) make references to unpaid tutoring as well.

Research sheds light, much still in the dark
While the research on shadow education was previously slow to evolve, we are now pleased to see more sustained attention and a breadth of new works. The articles in this special issue take the readers through a wide range of educational practices and contexts which illuminate, or shed new light on shadow education, helping to fill some of the gaps in our knowledge, understanding and geographic coverage of shadow education research. Many of the findings in this issue's papers also have direct implications for various stakeholders, including policymakers, 1 and we hope this will be brought to their attention.
The special issue also includes two informative thematic book reviews by Novotná and Bhorkar. The former analysed a monograph (Kim & Jung, 2019) that takes a global approach and discusses the intersection of shadow education and curricula, the latter reviewed a book that explored the policy implications of private tutoring in Myanmar (Bray, Kobakhidze, & Kwo, 2020).
In the future, it would be interesting to explore how different types of shadow education (Manzon & Areepattamannil, 2014) enable or hinder students from navigating their future educational pathways and aspirations by following longitudinal, ethnographic designs and other forms of long-term studies.
Much more research is needed to get a clearer picture of the ways shadow education can cohabit alongside public schooling without compromising its reputation, or creating extra pressure on teachers, students and families. We need more insight into the micro-social dynamics, meso institutional structures and macro global shifts and patterns that shape students' experiences of shadow education. In particular, with the current uncertainties of the global economy due to the COVID-19 pandemic, there will perhaps be a need for a class-based analysis of shadow education to reveal any inherent inequalities. Attention must also be given to emerging new forms of tutoring such as "education pods", "Zutors" (i.e. Zoom tutors) and "microschools", which have been pushed by parents' initiatives worldwide as a result of the pandemic. Additionally, the field could benefit from studies focusing on the technological advances the tutoring industry has been employing, such as AI-enabled platforms, virtual and augmented reality technologies, gamification and adaptive learning.
As is obvious from the previous paragraphs, this issue of Orbis scholae managed to cast light on only a section of the field of shadow education. We sincerely hope this special issue will inspire a wider academic discussion and turn scholarly attention towards further investigation of the subjects broached by the studies presented herein.
Pleasant reading! Ruggiano, N., & Perry, T. E. (2019). Conducting secondary analysis of qualitative data: Should we, can we, and how? Qualitative Social Work,18(1), 81−97. Stevenson, D. L., & Baker, D. P. (1992). Shadow education and allocation in formal schooling: Transition to university in Japan . American Journal of Sociology,97(6), 1639 In Germany, the proportion of 17-year olds who received shadow education (SE) (Bray, 1999(Bray, , 2017Byun, Chung, & Baker, 2018;Stevenson & Baker, 1992), 1 has rapidly increased from 27 percent in the early 2000 years to 47 percent in the early 2010 years (Hille, Spieß, & Staneva, 2016, p. 116). 2 Although participation during primary school remains low (reaching 8% in fourth grade), it increases rapidly following the transition to secondary school, with every fifth student receiving SE in grades nine and ten (ibid.). It is estimated that each year about 1.1 million students subscribe to programs in the private tutoring market (Klemm & Klemm, 2010), which is dominated by 4,500 tutoring schools, mostly founded since 1992 (Birkelbach, Dobischat, & Dobischat, 2017, pp. 59−62). Annual profits of up to 1.5 billion Euros are the result (Klemm & Klemm, 2010). Thus, SE is a widespread family strategy to promote educational attainment of children. In the present 1 In Germany, any kind of tutoring is called "Nachhilfe" (extra-help). We focus our analysis on private, fee-based, commercial tutoring, which fits the formal definition of shadow education (SE) by Bray (2017). The terms shadow education and private tutoring are used synonymous.
article, the observed increase in the use of SE in Germany is theoretically derived and empirically examined as a consequence of status-based, gender-specific investment strategies of families. Previous German and international studies on SE generally agree that (a) students from families with higher socioeconomic status (SES) use SE more frequently than lower SES students, thus indicating that SE exacerbates social disparities in educational attainment (e.g., Byun et al., 2018;Dohmen, Erbes, Fuchs, & Günzel, 2008;Entrich, 2020;Guill & Lintorf, 2019;Hille et al., 2016;. Only few studies provide (b) empirical evidence suggesting that SE may be used independent of SES, simply to compensate low school performance (see Abele & Liebau, 1998;Guill & Bonsen, 2011;Luplow & Schneider, 2014;Seiyama & Noguchi, 1984). International research also reported (c) a higher likelihood of girls to use SE in several countries, (e.g., USA: Buchmann, Condron, & Roscigno, 2010;Japan: Entrich, 2015;Korea: Lee & Shouse, 2011), but not in others (e.g., Germany: Hille et al., 2016;China: Liu & Bray, 2016;Poland: Safarzynska, 2013). However, for Germany we find (d) gender specific investment by school subject (Dohmen et al., 2008, p. 40). Furthermore, recent research (e) stressed that families generally make gender-specific educational investments, which are confounded by their SES (e.g. Becker & Müller, 2011;Breen, Luijkx, Müller, & Pollak, 2012;Hadjar & Berger, 2010;Lühe, Becker, Neumann, & Maaz, 2016. Whether family SES explains differences in the gender specific use of SE has never been in the focus of any study. Gender − if considered at all − has always been treated as a control variable or complimentary factor in past research. Based on these findings, we invest in three questions: Is SE more likely in high SES families, or is it used across all social strata to compensate for underperformance in school? Second, are there gender differences in the use of SE, differentiated by SES? Third, how do both findings affect social reproduction mechanisms? The last question is significant, as the frequent use of SE can lead to dislocations in the educational process of children and youth, as shown in studies concerned with shadow education in East Asia (e.g. Byun, 2014;Stevenson & Baker, 1992), for example.
In order to answer these questions, we follow previous research findings and draw on social reproduction theories (Boudon, 1974;Breen & Goldthorpe, 1997;Lucas, 2001). We ask whether investing in SE for girls and boys serves to maintain or even increase their status position, or whether there exist status-specific gender preferences for SE investment. The derived hypotheses are quantitatively tested by means of logistic regressions using data of the 2012 German Life Courses into Early Adulthood (LifE) study (Lauterbach, Fend, & Gläßer, 2016).

Shadow education as SES-specific investment?
International research confirmed that underperformance in academic school subjects (language, math, English, and MINT-subjects) is the main motive for SE attendance in most countries (Baker, Akiba, LeTendre, & Wiseman, 2001;Byun et al., 2018;), including Germany (Birkelbach et al., 2017Guill & Lintorf, 2019;Jürgens & Diekmann, 2007;Luplow & Schneider, 2014;Schneider, 2005). The same research argues that high SES (measured in terms of parental education, occupation and income) families are generally more likely to invest in SE than low SES families.
But for Germany the influence of socio-economic factors was only partially proven. Some studies report considerable effects of household income on SE participation (Hille et al., 2016;Schneider, 2005), while others report no such effects (Abele & Liebau, 1998;Guill & Lintorf, 2019;Luplow & Schneider, 2014). Similarly, some studies found that academic parents tend to not send their children to SE in Germany (Entrich & Lauterbach, 2019;Hille et al., 2016;Luplow & Schneider, 2014;Schneider, 2005), while others reported a significantly higher likelihood of children with academic parents to obtain SE (Guill & Lintorf, 2019;Guill, Lüdtke, & Köller, 2020). We suspect that the differences in the effect of the SES of the students are related to the measurement of the variables and the construction of the models. It is striking, for example, that the influence of educational variables on the use of SE only shows positive significant effects in studies that do not differentiate between paid and unpaid tutoring. Moreover, the measurement and number of SES variables included in the anlysis vary from study to study.
In addition to SES, exaggerated educational aspirations of upper strata parents are constantly discussed as a cause for the increased use of SE in Germany. It is assumed that academically educated parents from high-income households in particular demand tutoring to provide their children with a competitive advantage in the form of above-average grades (Hollenbach & Meier, 2004;Klemm & Hollenbach-Biele, 2016;Koinzer, 2013;Schlösser & Schuhen, 2011). This way, high SES families try to make sure that their children achieve above average school leaving degrees, enter tertiary education and get access to high-rewarding jobs. Several studies from international research on SE support this prominent claim (see Bray, 1999. Accordingly, the more cost-intensive lessons specializing on entrance exam preparation as the most prominent way to achieve competitive advantages for high performing students (enhancement features) are more frequently pursued by high SES students (e.g. USA: Buchmann et al., 2010;South Korea: Byun, 2014;Japan: Entrich, 2018;England: Ireson & Rushforth, 2011). However, reliable studies on the effectiveness of SE in Germany indicate that low-performing students may not catch up to their peers through additional tutoring (Entrich, 2014a;Ömeroğulları, Guill, & Köller, 2020). Empirical evidence refuting the assumption that well-performing high SES students use SE to gain competitive advantages for educational placement is still pending, though.

Shadow education as gender-specific investment?
Figure 1 shows gender specific participation in SE in Germany from 2000 to 2015 against the background of educational expansion. Findings show that girls more often achieve the highest formal school degree available in Germany, the Abitur (2015: 58.6% girls; 49.4% boys), and enter universities more frequently nowadays (2015: 60.5% girls; 56.1% boys). 3 Alongside this development, there has been a clear upward trend in the demand for SE between 2000 and 2010. Since 2010, the SE experience of 17-year-olds stabilized well beyond 40 percent for both male and female students (2015: about 42% for boys and 43% for girls).

Figure 1
Percentage of Abitur graduates, entrants to universities and participation in paid tutoring in Germany, according to gender (in %, 2000−2015) Notes: Abitur graduation rates as a proportion of the respective age population of the corresponding year; university enrolment rates as a proportion of first-year students of the population of the corresponding year of birth; Nachhilfe experience rate based on the item: Did you ever obtain paid Nachhilfe? (Yes/No), showing the weighted mean of 17-year-old participants according to birth cohorts (2000: born 1982−84; 2005: born 1987−89; 2010: 1992−94; 2015: 1997−99), SOEP adolescent survey. Sources: Own calculations based on Helbig (2012);Statistisches Bundesamt;SOEPv36. 3 It has to be noted that not all individuals who achieve the Abitur enter university afterwards, many also enter vocational programs instead. The high entrance rates are thus also an outcome of greater openness in access to university, since individuals without the Abitur are increasingly allowed to enter a range of study programs at university if they have completed a related dual vocational training program of generally three years instead of the Abitur, for example. In line with research on differences in subject-specific performance by gender (e.g. DiPrete & Buchmann, 2013;Hadjar, 2011;Helbig, 2012), gender differences in the use of tutoring according to school subjects are often reported. Existing primarily descriptive findings indicate that boys show on average lower achievement in languages (German and English) and are more likely to use tutoring in related subjects, whereas girls show on average lower performance and thus higher demand for tutoring in mathematics (Dohmen et al., 2008;Guill & Bonsen, 2011;Hollenbach & Meier, 2004;Jürgens & Diekmann, 2007;Rudolph, 2002). Yet, studies explicitly examining the gender-specific use of private tutoring are not available.

Intersectionalities between gender and SES in shadow education investment?
Subject domain-specific differences in academic achievement by gender are often confounded with parental SES. Using data from the Berlin study for fourth grade students, Lühe et al. (2016) point to systematic SES-specific differences in performance between boys and girls in elementary school. Boys show greater variation in academic performance measured by grades across social strata than girls. In all three investigated domains, i.e. reading, German and mathematics, boys from high SES families showed higher performance compared to girls from similar strata. The opposite was found for low SES boys. The authors attribute this mainly to strong traditional concepts of masculinity, which are more prevalent in low SES households. In a different analysis based on national representative data from the 2006/07 TIMSS survey, the authors showed that there are only marginal performance differences between boys and girls, differentiated by social origin. Striking, however, is the high gender-independent performance among high SES groups (Lühe et al., 2017).
Past research has also shown that parents' educational aspirations for children's secondary school choice have largely converged for girls and boys with similar performance (Helbig, 2012). In addition to the "de-traditionalization" of gender roles, evidence suggests that educational aspirations of girls and parental aspirations for daughters have increased, partly favoring investment in the education of daughters over sons (Helbig, 2013). Whether parents favor girls or boys in their private tutoring investment is not clear and was hardly investigated. One exception is the study by Seiyama and Noguchi (1984), which demonstrated gender-and SES-specific differences in the likelihood to obtain tutoring in Japan. Findings show that high SES families invest more in the SE of boys than girls, especially if boys are high performers. Girls received SE independent of their grades and family background, and gained significant performance improvement − boys did not.
The SES-specific differences in the appreciation and attribution of traditional gender roles, including resulting gender-specific educational aspirations, which partly explain the higher performance and success rates of girls in high schools and universities, suggest SES-specific investment in private tutoring by gender in Germany as well.

Theoretical considerations
In sociological educational research, rational choice theories (Boudon, 1974;Breen & Goldthorpe, 1997) are often used to understand social inequality in educational attainment (e.g. Baumert, Maaz, & Trautwein, 2010;Becker & Lauterbach, 2016). Two effects of social stratification have been identified as causing social reproduction: First, SES-specific differences in learning habits and family support often lead to differences in academic achievement. Students from low SES families often show lower performance than privileged students (primary effect). Second, based on SES-specific educational aspirations, parents seek to ensure that their children achieve at least a position in the same class as themselves (secondary effect). All families are anxious to avoid downward social mobility, particularly through investment in education (status maintenance motive). However, the actual risk taking based on the risk aversion level varies across social strata. High SES parents are more likely to choose educational programs promising high returns despite having high risk aversion; under the same circumstances, low SES parents tend to choose programs with lower returns. Hence, different decisions are made by different strata, even if the performance of the students is the same (Breen & Goldthorpe, 1997).
According to effectively maintained inequality (EMI) theory, the central status maintenance motive involves exhausting all possible resources to avoid a decline in status. Two major dimensions of education affect status stability: a quantitative (e.g., obtained number of years of education) and a qualitative dimension (e.g., pursued study program/track) (Lucas, 2001). If the majority of students attend upper secondary school, for example, high SES families will try to place their children in more advantageous degree programs. However, if, as happened in Germany, more than every second student achieves the most advantageous school leaving degree, the Abitur, the value of this certificate is no longer measured solely by its "possession", but in relation to the number of students who also earn this degree (Shavit & Park, 2016). The more students achieve the Abitur, the more students will compete for coveted positions in the labor market and for seats at university, which increases the cohort competition. Especially high SES parents will then seek new opportunities for their children to secure competitive advantages for status maintenance, shifting their focus to increasing the relative value of the degree, i.e. its grade point average.
From this perspective, the investment in SE is a rational decision of forward-looking parents to support the school performance of their children, considering the related costs and benefits Guill, 2012;Luplow & Schneider, 2014;Schneider, 2005). The use of SE can serve two functions related to status maintenance and upgrade motives: (1) compensation of below average performance to prevent dropping out from the competition for educational qualifications; and (2) enhancement of average/good performance to attain a competitive advantage within the same school degree program, i.e. improvement of grades above the average (Entrich & Lauterbach, 2019).
If high SES families make more use of one or both functions of SE than low SES families and given that these investments show positive effects, social inequality would be strengthened. However, if we take into account the findings from research on the effects of SE in Germany, it appears doubtful that children gain any significant long-term performance improvement from their use of tutoring. Instead, reliable research based on national representative (panel) data shows that students achieve nearly no improvement in performance through tutoring − tutees remain at their respective performance levels. Students continue to show low performance even if they receive tutoring (Dohmen et al., 2008;Entrich, 2014a;Guill & Spinath, 2014;Hosenfeld, 2011;Ömeroğulları et al., 2020). This indicates a primarily compensatory function of SE: SE helps students to keep their (often low) performance level, achieve a certain degree and avoid dropping out of school. Since high SES parents can provide better support for their children, children from these families show on average better performance and have also less need for compensatory tutoring, though. In order to function as a status maintenance tool for high strata, students from high SES families would need to use SE significantly more frequently to maintain their (often high) performance. Only then, social inequality would be positively affected.
However, we doubt that parents' socioeconomic background and their aspirations show similar strong effects on SE attendance in Germany as found in other national settings (e.g. East Asia; . Due to the educational expansion in Germany, more students from disadvantaged SES backgrounds achieve a university entrance qualification (Abitur) instead of the less prestigious lower secondary and intermediate degrees at the Haupt-and Realschule, respectively. 4 But, since low and middle SES parents seldom possess the Abitur themselves, they often lack the competence to support their children in achieving this degree. Thus, the increased demand for SE may be caused especially by students from non-academic families and their higher need for compensatory tutoring due to higher requirements at higher levels/tracks of schooling. If low SES strata make substantial use of SE, this may compensate the SES gap in average school performance and cognitive ability enough to cope with the higher requirements of more demanding tracks. This would enable these students to achieve higher degrees than their parents, thereby weaken social reproduction in educational attainment. This argumentation leads to the first set of hypotheses: Contrasting to arguments from social reproduction theories, we expect that lower SES families are generally more likely to use SE in Germany. More specific, 4 The German secondary school system provides students with the opportunity to choose between three general tracks of different length, curriculum and expectable returns: (1) The upper secondary school degree (Abitur, 12 to 13 years), traditionally obtained at the Gymnasium, provides students with the opportunity to enter university and thus gain access to the most prestigious and rewarding jobs. (2) The intermediate degree (10 years; Realschule) provides access to those parts of the dual vocational training system leading to white collar jobs. (3) The least demanding lower secondary degree (9 years; Hauptschule) enables graduates to enter those parts of the dual vocational training system that lead to blue collar or unskilled jobs (Weiss & Schindler, 2017). to achieve status upgrading by attaining the highest school leaving certificate, the Abitur, non-academic parents are more likely than academics to invest in the SE of their children (hypothesis 1-1). We further expect a strong association between the students' performance level and their experience with SE. Specifically, we expect that families are more likely to invest in SE when their children are in danger of early dropout from school due to poor performance (compensation strategy). The empirically proven rather marginal effects of SE on academic achievement indicate that low performing students may keep their grades at a similar level without significantly improving them. Hence, students with below average performance should show the highest likelihood of having obtained SE (hypothesis 1-2). Moreover, we expect that students who continue their school career to upper secondary education and/or enter qualitatively more demanding secondary school tracks, i.e. the Abitur program, are generally more likely to obtain SE (hypothesis 1-3). Since students "might choose divergent educational pathways [and gain] more influence over the decision for shadow education as they grow older" (Entrich, 2015, pp. 212−213), we also expect that − aside the parents' educational aspirations − students' own aspirations positively influence their likelihood to obtain SE (hypothesis 1-4).
In a second step, we extend the above arguments to gender disparities. Following Boudon (1974), gender disparities in educational attainment can be explained analogously to class-specific disparities. For example, stereotypically gendered upbringing and socialization experiences influence students' interests, behaviors, and motivations and affect their subject-specific performance and educational progress (primary effect) (Hadjar & Berger, 2011;Lühe et al., 2016Lühe et al., , 2017. Traditionally, parents favored an investment in the education of boys over girls because, in line with traditional role models and the male breadwinner model, they viewed the education of boys as more important for later income and status. Educational decisions have therefore always been influenced by the gender of the child, with lower educational aspirations for girls compared to boys (secondary effect). Cost-benefit calculations of parents have traditionally turned out to be to the disadvantage of girls, because parents favored their sons to achieve a high education and subsequent status to follow into their (father's) footsteps (ibid.). The status maintenance motive in social reproduction theories thus stronger referred to sons than daughters. However, the recent de-traditionalization of gender roles concomitant with a preference of investing in girls (Helbig, 2012) indicate that families may also favor investing in the SE of their daughters for status maintenance or upgrading.
Because traditional gender role expectations are still more prevalent in low SES families, especially low educated, non-academic families, and least prevalent in high SES families with advantaged educational backgrounds, we suspect that parental status upgrade and maintenance motives are still more traditional in non-academic families. Thus, these families should favor investments in the education of boys over girls for status maintenance and upgrade. Hence, we expect interaction effects between students' gender and parental educational background. Specifically, we expect that boys from non-academic backgrounds are more likely to obtain SE than girls from the same background (hypothesis 2-1). Furthermore, we expect larger effects of other SES factors and aspirations on the likelihood that boys from non-academic strata obtain SE. Especially the financial situation of the household should affect whether non-academic families, i.e. low and middle SES strata, can afford tutoring 5 (hypothesis 2-2). We expect that because non-academic families favor the support of boys over girls in academic matters, girls from non-academic families are less likely to receive tutoring if they show below average grades than girls from academic families. In academic families, children are more equally treated in academic matters, wherefore we expect no concrete differences in the reception of tutoring by performance level between education strata for boys (hypothesis 2-3).

Data: The German LifE study (1979−2012)
The German "Pathways from Late Childhood to Adulthood" (LifE) study started as a longitudinal youth study in Hesse, one of the federal states of Germany. Annual samples of approximately 2000 children and adolescents of the birth cohort 1967 were collected from 1979 to 1983. 12-year-old students were questioned up to five times during this time period, their parents were questioned two times. The youth study initially focused on urban and rural environments, as well as different school contexts. The study covered students from the former tripartite and comprehensive system ('quasi experimental system'). Thus, an urban region (Frankfurt am Main) of former "West" Germany and adjacent rural areas (Odenwald, Bergstrasse) with different school systems were chosen. The study also focused on juveniles coping with developmental tasks during adolescence, and on the development of personality. As far as representativeness is concerned, it can be said that the families with children aged 12 to 16 years in the years 1979 to 1983 in West Germany are well represented in the sample, because children in this age group were focused at that time (Tsching, Berichte, & Fend, 1983). The LifE study continued to accompany the former students and collected data again in the years 2002 (age ~35) and 2012 (age ~45) (Lauterbach et al., 2016).
SE related items were introduced for the first time in 2012. The representative original student sample in 1979−1983 and the now 45-year olds (N=1,359 participants) was supplemented by an additional independent sample of their children, 6 hence providing us with valuable information on a third generation (N=581). So in 5 Even though the German government issued a country-wide voucher program ("Bildungsgutschein") in 2011, allowing all low SES households receiving social welfare services ("Hartz 4") to apply for fee-based tutoring, if the child shows low performance in school (grades 4 or worse), the use of these services will not be reflected in our data.

6
Only the first-born child in each family aged 12 to less than 18 years was considered.
contrast to studies where only one cohort was sampled and either the parents or the students had to make assumptions on key variables of the other cohort (such as social background estimated by students or school performance estimated by parents; e.g., Entrich, 2014b;Guill & Lintorf, 2019), we are in a position to use reliable statements of both cohorts and bring them together for our analysis of SE determinants. More important, we are able to differentiate paid from unpaid tutoring in our analyses, which remains a major shortcoming in many major studies on SE in Germany (e.g., Behr, 1990;Guill, 2012;Guill & Lintorf, 2019;Ömeroğulları et al., 2020). The specific design of the complete survey, of the youth study (1979−1983) and the renewed questionnaire surveys in 2002 and 2012 (approximately 30 years later), has a positive selection of the cohort according to following aspects: migration background, school degree and divorces (Lauterbach et al., 2016, pp. 32, 36, 40). Concerning childlessness, there were fewer childless women and men than the average population at the age of 45 years in 2012 and significantly more families with two children (Lauterbach et al., 2016, p. 38). So far, we have not weighted our findings because we were interested in correlations and predictions and not in extrapolating our results to the corresponding population sizes. Due to the uniqueness of the study on the use of private tutoring, this study is very well suited to test the usage behaviour of children according to performance and social background.

Variables
For the 2012 LifE survey, we asked parents the following three questions concerning SE: "Has your child ever received private tutoring during his/her school life? If yes, how often?"; "Did you pay for this tutoring or was it mainly free of charge?"; and "What prompted you to organize private tutoring for your child?" According to our data, in 2012 44.1% of the 12-to 17-year-old children reported to have received paid tutoring at some point, whereas another 8.2% obtained free of charge tutoring. Half of all SE recipients (51.4%) used these lessons to improve their grades in school, 38.4% pursued SE to prepare for upcoming tests, exams or classes, and 28.1% demanded individual support, wanted to close gaps of knowledge, or practiced new learning strategies. To investigate whether SE may contribute to educational and social inequalities, we focus our analysis on paid tutoring and encode unpaid and no tutoring as a reference category.
To test our hypotheses, on the parents' side the SES and their educational motivation are used as decisive determinants for an investment in tutoring. In order to take into account all the relevant resources of the family, apart from the highest parental education (1 = academic, i.e. one or both parents possess a university degree; and 0 = non-academic, i.e. parents do not possess university degrees), especially economic (household net equivalent income: 1 = more than 150% of the average; 2 = average; and 3 = less than 70% of the average) and social dimensions of origin (class according to the European Socioeconomic Classification, i.e. ESeC: 1 = salariat; 2 = intermediate; and 3 = working class). 7 In addition, parents' post-secondary educational aspirations for their children (1 = university and 0 = vocational training/no university degree) are included in the analysis.
On the part of the student, gender is of primary interest, which is included as a binary variable (1 = female, 0 = male) in our analyses. In order to answer our hypotheses, students' academic achievement level in school is important. Thus, we classified students into three different types of performers based on their grades of the last school report in the subjects mathematics, German and English collected at the time of the 2012 survey: above-average (grades 1, 2); average (grade 3 = reference); and below-average performers (grades 4, 5, 6). To reflect the requirements at school, the attended school level (1 = upper secondary education, grades 10 to 12; 0 = lower secondary education, grades 5 to 9) and the school track (1 = Abitur track; 0 = Real-or Hauptschule degree programs) are included. Furthermore, we control for the students' own post-secondary educational aspirations (1 = university degree, 0 = vocational training / no university degree). 8

Temporal structure between dependent and independent variables
Although the data do not accurately capture the time structure between the use of tutoring and the measurement of performance, we can state that there is no bias in the calculation of effect sizes. The majority of the students were interviewed in grades 7 to 10 and research shows that students take SE mostly in these grades (Hille et al., 2016). In addition, research shows clearly that pupils achieve little, if any, improvement in performance through tutoring and generally remain at their respective performance levels, i.e. show high, average or low performance (Dohmen et al., 2008;Entrich, 2014a;Guill & Spinath, 2014;Hosenfeld, 2011;Ömeroğulları et al., 2020). This indicates that SE primarily takes on a compensatory function, i.e. helping students in each performance group to maintain their level of performance, but not to increase it significantly. In order to be used as a status maintenance tool for high strata, students from high SES families would have to have used SE in the course of their school career to maintain their 7 The ESeC is used to classify European societies into nine categories, ranging from higher professions and management occupations to unskilled workers. We recoded these nine groups into three broad classes: the salariat (categories 1 and 2), the intermediate (categories 3, 4, 5, and 6), and the working class (categories 7, 8, and 9) (for a detailed overview see Wirth & Fischer, 2008). 8 The LifE study also includes data on the parents' and students' idealistic and realistic educational aspirations for the highest school leaving degree. Based on our theoretical arguments we found it more promising to focus on post-secondary aspirations instead, because we believe that these measures better relate to status upgrade and maintenance motives of families. Supplementary analyses not shown here support this notion. We tested the effects of post-secondary and school-leaving degree aspirations both separately and together in all our models and found robust effects of post-secondary aspirations, but no effects for school-leaving degree aspirations. high performance. When exactly this was the case and for how long is of secondary importance, because we are not interested in testing whether the students' academic achievement has led to the use of tutoring at a certain point in time. Rather, we investigate whether students with different levels of performance (high, medium, low) are more likely to have obtained tutoring and how this is related to their gender and social origin, which allows drawing conclusions about specific investment strategies of families.

Methods and analysis strategy
We first look at descriptive data, before conducting multivariate binary-logistic regression analyses to predict students' likelihood to receive paid SE in secondary school by social origin and gender. 9 To statistically limit the conditions under which families choose to invest in SE for their children, first, we present a model including all relevant variables step by step. Second, we differentiate the complete model by gender, and third, by the educational background of parents. Instead of reporting logistic coefficients or odds ratios, we show average marginal effects (AME). AME show how many percentage points the average probability of the represented group of one variable is different from the probability in the reference group, while ensuring comparability between models and groups (Mood, 2010). Finally, to test whether the interactions between gender and educational background are significant, we performed logistic regressions interacting all predictor variables with the respective subsample variables (gender and/or educational background).

Missing data
Overall, the proportion of missing values for our sample was very low. Yet, besides gender, there are variables with some missing values, varying between 1.9% (income) and 7.2% (school track). Our sample has 549 cases (285 girls and 264 boys), for which the following missing values remained: 2.7% (N = 15) for parental class; 0.6 % (N = 3) for educational aspirations of parents; 1.8% (N = 10) for grades in math and German, respectively; 2.7% (N = 15) for grades in English; 3.6% (N = 20) for students' school level; 7.5% (N = 41) for school track; and 2% (N = 11) for educational aspirations of students (see also Table 1). To avoid further reduction of our sample and biased parameter estimates due to missing values, we carried out multiple imputation (Schafer & Graham, 2002). We used the multiple imputation chained equations (MICE) routine implemented in STATA 14 for imputing the missing values (Marchenko, 2011). In total, we imputed 10 data sets including all covariates, the outcome variable, and additional auxiliary variables in the predictor models. Table 1 shows descriptive findings. 10 According to our data and in line with national representative findings, there are no significant gender differences in the use of SE (girls 42.1% / boys 46.2%, t-test: .291). In contrast to prominent assumptions, we find hardly any differences in the use of SE according to class, household income or educational aspirations of the parents. Parental education level is even negatively associated with tutoring: only 34.8% of the children from academic families, but almost every second child from non-academic families (49.5%) obtained paid tutoring, which is statistically significant. Reinforcing social inequality through SE is unlikely on the basis of these findings. Rather, students from different SES families seem to receive SE similarly often.

Descriptive statistics
In contrast, very much as expected, students' school achievements in mathematics, German and English are obviously related to SE. Twice as much low performing students (about 60%) obtained SE compared to high performing students (about 30%). The percentage of average performers using SE is considerable, though (about 50%). The highest SE demand show those already attending the upper secondary school level (50.7%) and aspiring to achieve the Abitur (46.3%). This supports our assumption that paid tutoring is often used by students to compensate for the higher requirements towards the end of their studies and generally when entering qualitatively more demanding school tracks. Students' own educational aspirations show hardly any differences.
Differentiating the use of SE by gender partially questions these findings. We found significant differences in the use of tutoring by parental education and aspirations for boys, of whom a much higher proportion from non-academic backgrounds (about 54%) and with parents without tertiary education aspirations (about 50%) obtained SE compared to those from academic families (about 29%) and without parents with tertiary education aspirations (about 37%). This already indicates a higher overall need to compensate low performance of boys from non-academic educational backgrounds. In contrast, there are no significant gender differences in SE use evident for children from high SES families, i.e. where parents possess a university degree, belong to the high-income group and to the salariat, and have high aspirations. 10 Additionally, we tested for statistical significance in the difference between variables using t-tests. Our data also show differences in the use of SE by gender according to school grades, the school level and track. Girls with above-average grades in mathematics took less SE (about 25%) than boys (about 32%). For the domains German and English, the opposite is found. These differences between genders are non-significant, though. Both genders more frequently obtained tutoring if they attended the upper secondary school level and/or pursued the Abitur. However, only girls obtained significantly more often SE if they were enrolled in the Abitur track or at the upper secondary school level. Significant gender differences in students' own post-secondary aspirations are not discernible.
The correlations between the predictor variables are mostly weak. Parental education is positively correlated with household-net-income (.26) and class (.24), with the latter two being also weakly correlated with each other (.18). Notable correlations also exist for educational aspirations of parents and students (.36) as well as for students' grades in the three different subject areas German, mathematics and English (.34 to .56). Table 2 summarizes the results of the stepwise logistic regressions to test our first set of hypotheses. Model 1 examines the influence of parental SES on the likelihood of students to have obtained SE. As expected in hypothesis 1-1 and indicated by our descriptive findings, parental education is negatively associated with SE (−15% probability if parents possess university degree). Other SES factors show no significant effects. In Model 2 we test whether the overall performance level of the students affects their involvement with SE. Results show that − despite possible endogeneity and a lack of time structure − performance is the most significant predictor for SE, reporting the highest R 2 (.08) among the first set of separated models. Compared to average performing students, above average performers in math and English are significantly less likely to have obtained SE, whereas no significant differences are found for the performance in German. Conversely, this supports hypothesis 1-2: Students with below-average (and average) performance are far more likely to have experience with SE. Model 3 then tests school institutional factors' influence on the likelihood to have obtained SE: Students pursuing the Abitur (+9%, significant at the 10% level) are more likely to seek SE (supporting hypothesis 1-3). Model 4 then tests the influence of parents' and students' post-secondary aspirations showing unexpected results: If parents wish for their children to enter tertiary education, the children will be significantly less likely to have obtained SE (contrary to hypothesis 1-4). Model 5 focuses on the association between gender and SE, showing that a student's gender does not show any obvious effect on tutoring attendance. In Model 6 we included all predictor variables and find robust effects of parental education, performance level, and school track. Parental educational aspirations are now mediated through their education background and reduced to insignificance, whereas students' own aspirations become more important (+13%). Even though the R 2 is quite low for SES variables (Model 1: .02), educational background remains a robust factor for SE participation. These first results support the notion that SE is primarily used by non-academic families to compensate low performance rather than serving status maintenance of high SES families.

Multivariate results
To test for interaction effects between gender and educational background, in Table 3, we first differentiated our analyses by students' gender (Model 7) and found astounding differences: While parental education level shows a massive impact on the boys' SE attendance (−27% probability if parents possess university education), there are no effects for girls. This difference in average probability between genders is highly significant (see M7 Diff. P > |t|; confirming hypothesis 2-1). Hence, there exist clear intersectionalities between gender and parental education in SE investment. Even though there seem to exist differences in the effects of performance, school level, and aspirations on SE between gender, these differences are statistically non-significant with the exception of school level (at the 0.1-level). Thus, girls attending the upper secondary schooling level are significantly more likely than boys to have obtained SE.

Model 3 School
Model 4 Asp.

Model 6 All
Parents AME SE AME SE AME SE AME SE AME SE AME .02

SE
.08 .01 .01 .00 .12 Note. AME = Average Marginal Effects; SE = Standard Error (in brackets) ***p < 0.001; **p < 0.01; *p < 0.05; +p < 0.10; n.s. = not significant Source: LifE 2012, own calculation. Note. AME = Average Marginal Effects; SE = Standard Error (in brackets); Diff. P>|t| = Difference between predictors across models: significant or not. ***p < 0.001; **p < 0.01; *p < 0.05; +p < 0.10; n.s. = not significant Source: LifE 2012, own calculation. Note. AME = Average Marginal Effects; SE = Standard Error (in brackets); Diff. P>|t| = Difference between predictors across models: significant or not. ***p < 0.001; **p < 0.01; *p < 0.05; +p < 0.10; n.s. = not significant Source: LifE 2012, own calculation. Finally, in Model 8 we further differentiated our analyses according to parental educational background. First, we take a look at the displayed effects within the models before investigating whether the found effects differ significantly between the four sub-groups. As for children from academic backgrounds, it appears that girls are significantly more likely to have obtained SE if they show below average performance in English and are enrolled at the upper secondary school level. For boys from the same backgrounds we find a lower likelihood of SE if their parents want them to enter university in the future. These gender differences are significant in the case of English performance: Among academic strata, girls are more likely than boys to have obtained SE if they show below average performance in English (see M8-1 Diff. P > |t|). Girls' support in foreign language proficiency in English seems to be more highly valued among academic families.

Model 8 Parents' Education x Students' Gender
In the case of children from non-academic families, we find that boys' likelihood to have obtained SE is significantly higher (+23%) if they come from high-income families, or show high post-secondary aspirations (+20%), thus supporting hypothesis 2-2. However, we cannot confirm that this effect is significantly different from other subgroups (see M8-2/3/4 Diff. P > |t|). Girls from non-academic backgrounds show a lower likelihood to obtain SE if they come from the intermediate class, are enrolled at the upper secondary school level and/or pursue the Abitur, and show above average performance in math and English or below average performance in German. Significant gender differences in the effects of the predictor variables for students from non-academic backgrounds are not found (see M8-2 Diff. P > |t|).
However, the last result for girls is particular interesting, especially if compared with the results for girls from academic backgrounds (see M8-3 Diff. P > |t|). Comparing genders across different educational backgrounds shows that academic parents are significantly more likely to invest in the SE of their low-performing daughters (in German and English) than non-academics. These results thus support hypothesis 2-3. As for boys, we find a significant difference in parental aspirations. Thus, boys from academic backgrounds are less likely to obtain SE if their parents wish for them to enter university in the future. But no such relationship exists for boys from non-academic backgrounds.

Discussion
In the present work, we questioned the causes for the strong increase in the use of SE in Germany over the past two decades. We attempted to explain this development through gender-and SES-(education) specific familial investment strategies. In contrast to international findings and prominent theses, our results show that SE in Germany does not serve as a tool to promote social inequality. SE is more prevalent among lower performing male students from non-academic family backgrounds. This holds especially for the student group who intend to achieve the Abitur and enter tertiary education afterwards (thus confirming the first set of hypotheses). The lack of (positive) parental aspirations towards the university entrance of their children further strengthens the view that primarily status upgrade motives of non-academic families have caused the higher SE investment in Germany -not status maintenance motives of high SES families. SE in Germany seems to compensate challenges at school for students from non-academic families instead of promoting social distinction.
Differentiated regression models showed that gender plays a significant role in the reception of tutoring, suggesting different educational investment strategies based on gender-specific status attainment goals. We found considerable interaction effects between gender and parental education on SE investment. First of all, boys from non-academic families are most likely to obtain SE. Secondly, these boys are also significantly more likely to have obtained SE than girls, for whom no influence of parental education or other SES factors could be verified (confirming hypothesis 2-1). In addition, higher income and aspirations (of the students themselves, not the parents) seem to play a more important role for SE attainment of boys from non-academic strata compared to other students (confirming hypothesis 2-2). Third, girls' SE attainment seems unaffected by their SES in general, but there are significant differences in the likelihood of low performing girls to have obtained SE according to parental education. Girls with below average performance in languages are more likely to experience SE if they come from academic instead of non-academic families (confirming hypothesis 2-3). In these cases, status maintenance rather than upgrade motives drives the girls' SE attainment. For girls' performance in mathematics no significant differences were found, though.
In conclusion, this first-time investigation of SES-and gender-specific differences in SE investment reveals some interesting differences in family investment behavior. Not only could we show that tutoring in Germany depends less than expected on the socio-economic situation of the household; boys from disadvantaged educational backgrounds are even more likely to obtain SE, thereby possibly counteracting a widening of the SES-and gender-achievement gap. Our findings indicate first concrete intersectionalities between gender and SES in SE use for Germany insofar as non-academic families are more likely using SE for compensatory purposes for boys, whereas highly educated families also use SE for girls for status maintenance. Although variations in the quantity and quality of SE may largely affect these findings, because then the financial resources of the family become more important, existing surveys actually show that the bulk of spending on tutoring in Germany is rather moderate in nature and hardly comparable to spending in countries with sophisticated tutoring systems (Birkelbach et al., 2017).
Overall, it seems as if there still exists a lack of incentives for high SES families in Germany to use tuition for status-promotion, such as "gatekeeper" exams that determine transitions to the upper secondary or tertiary education levels. 11 Still, this paper provides a good start for future research, which should further clarify how SES affects differences in SE reception of different quantity and quality. So far, however, the prominent assumption that SE serves high strata as an instrument of social exclusion proves to be untenable on the basis of the presented findings, especially for boys. Families primarily try to ensure that their children compensate the increasing requirements as they went through school. Especially if the demands in school are high, for example if students aspire to graduate with the Abitur, tutoring might serve to avoid dropping out by compensating the higher requirements of this degree track. Even if it remains doubtful that SE possesses the power to significantly improve the performance of students from below average to average or even above average performance, German families may well use SE to increase the chances of their children to achieve the highest school degree and through this achieve a status upgrade. Future research should thus also investigate the actual effects of SE on allocation to the upper secondary and tertiary education levels. Research in this direction is common in many countries, but non-existent for Germany.
Based on our findings, the general increase in the use of SE in Germany should be understood as a consequence of educational expansion. One the one hand, the lack of remedial, individual support for the increasing number of students (esp. girls) entering higher school tracks (e.g., leading to the Abitur) caused demand for individual support through supplementary tutoring. On the other hand, the high ambitions of students from non-academic backgrounds (esp. boys) to achieve higher educational credentials than their parents (status upgrade motive) furthered SE participation.
The significant differences in the general and education background-specific use of tutoring by gender and the implicit effects on student achievement call into question the assumptions of prominent inequality theories. The here presented rational choice and effectively maintained inequality approaches better explain why boys from low SES families would receive tutoring for status upgrade. However, why the investment in girls' SE is largely unaffected by parental SES in general and why lower performing girls from non-academic backgrounds are less likely to have obtained SE than girls from academic backgrounds is up to now not well explained. That's because evidence suggests that SE provides its customers in Germany with no enhancement features similar to those found in other national settings (e.g., exam preparation in the United States or in East Asia). Thus, incentives for high SES families to invest in SE to achieve advantages in the competition for credentials are scarce at best. SES and gender are largely intertwined when it comes to educational investment strategies of families. A fact that needs to be better investigated in future research. Even though our findings are far from conclusive since we cannot test for long-term effects of SE on boys' and girls' educational attainment and progress, our findings point to the need of developing social reproduction theories to stronger address gender disparities in educational decisions.
Our findings call for similar approaches in other national settings to explore the role of family status motives for the increasing demand for tutoring and its implications for inequality persistence and gender disparities. In this regard, systematic research on the relationship between gender-specific investments in SE and the recent trend of female educational advantage in many schooled societies should be investigated throughoutly, as already suggested by . Also, a future review of our results over time (trend analysis) and the analysis of long-term effects of SE, especially for school allocation and transition to university seems promising.  When the school bell rings in the Netherlands, an increasing number of students head to private supplementary tutoring (CBS Statistics Netherlands, 2020), mirroring a pattern of increasing participation in so-called "shadow education" that is found in many countries across the globe. Like a shadow following a person or an object, shadow education follows mainstream education by providing students with formal, private, out-of-school learning activities to enhance their academic performance (Kim & Jung, 2019). In some countries, mainly in East Asian societies which value Confucian traditions of effort and achievement, shadow education is regarded as a standard educational practice for most school-aged children ). In other countries, particularly those with well-established, state-funded educational systems like the Netherlands, shadow education has long been a relatively marginal phenomenon, yet seems to be emerging as a more regular educational practice nowadays. For instance, in 2018, parents in the Netherlands spent a total of 284 million euros on shadow education, a marked increase from 26 million euros in 1995 (CBS Statistics Netherlands, 2020).

References
Given its increasing status, shadow education in the Netherlands is prone to varying interpretations of what goals shadow education caters to and what shadow education offers that regular educational settings may not, or cannot, offer. Policy makers, for instance, often interpret shadow education as an indication of the underperformance of schools, inferring that parents and students use shadow education to compensate for failing schools . Parents, in particular those in school systems where students are separated based on ability, can seek competitive advantage for their children through shadow education (Matsuoka, 2019). As less affluent parents may not be able to afford shadow education, concerns are that the growing use of shadow education enhances socioeconomic disparities in educational trajectories. Such concerns may fuel the need for empirical efforts exploring to what extent, and why, shadow education possibly competes with, supplements, or replaces learning activities at school (Elffers et al., 2019;. Whereas existing explorative research in the Netherlands focuses on how policy makers, parents, and educators interpret shadow education (Bisschop, van den Berg, & van der Ven, 2019), scant attention has been paid to how students interpret their shadow education attendance. Therefore, this paper aims to explore the place shadow education, as a learning space, has in students' daily academic lives. We conceptualise shadow education both as a physical place that students can turn to next to their school and home environment, as well as a symbolic place to which student attach certain meanings that may or may not be unique to shadow education as a place for learning and support. By asking students to reflect on the goals they pursue and the benefits they perceive from attending shadow education, we try to unravel what it is that students seek and find in shadow education in comparison to the home and school context. Our examination of students' goals and benefits not only builds upon and extends earlier student-centred work on shadow education in Asian and North-American contexts (Chan & Bray, 2014;Forsey, 2013;Hajar, 2018Hajar, , 2019Kim & Jung, 2019;Mahmud, 2019), but can also enhance our understanding of the relatively unknown position shadow education occupies in students' academic lives.

Conceptual framework
In pursuit of the above-mentioned aim, we apply an institutional approach, whereby we explore the place of the shadow education institution in the academic lives of students through interviews with such students (Burman & Miles, 2020;Yamato & Zhang, 2017). Rather than providing an exhaustive list of factors that may or may not influence students' goals and experienced benefits (cf. Guill, Lüdtke, & Schwanenberg, 2019;Yung & Chiu, 2020), in this section, we examine what students' goals and experienced benefits can be, and how such goals and benefits are indicative of interrelationships between institutions (Figure 1).

Students' goals
Previous studies have tried to map students' goals of attending shadow education by asking students about their motives for tutoring participation. Often, such motives are a reflection of what caused students to turn to shadow education in the first place. Hajar (2018), for example, found that students attended shadow education with the relatively proximal motive of passing school exams, and these findings were consistent across interview studies in Bangladesh (Mahmud, 2019), China , Israel (Addi-Raccah, 2019), and Japan (Ozaki, 2015).
Other researchers have found that students can also have less tangible goals for attending shadow education. , for instance, noted that some students in Japan mostly attend private tutoring to meet peers, a finding echoed by Chan and Bray (2014). In their qualitative study in Hong Kong, Chan and Bray (2014) found that rather than focusing on academic motives, students saw tutoring as a context for communicating and interacting with their peers or tutors. Taken together, these studies indicate that students can ascribe a range of goals to shadow education.

Experienced benefits
In previous research, tutored students mainly reported gains in terms of knowledge (Kim, 2016), skills Nam & Chan, 2019), and attitudes (Hajar, 2018). Despite these initial empirical findings, existing research has not yet provided a clear picture of the experienced benefits of shadow education attendance, particularly because such benefits differ from student to student. For instance, some students may quickly grasp the tutor's explanation, already be familiar with independent studying (Chih-Hao, 2019;de Guzman et al., 2018) or put more effort into the tutoring . Given such possible heterogeneity in experienced benefits, we are particularly interested in students' own descriptions of what shadow education offers them individually.

Shadow education adding to or substituting for home and school
In their discussion of marginalised educational spaces, Burman and Miles (2020) point to shadow education as an institution that is supposed to add to mainstream schooling, yet in practice ends up replacing some school and parental functions. If students, for instance, seek to refine already-learned study skills through shadow education, the practice adds to school. If, on the other hand, students seek extensive remedial lessons through shadow education Chan, 2019), the practice may compensate for shortcomings in teaching. Studying what students seek from shadow education, that they do not find at school or home, can thus be a way to explore the place of shadow education in students' academic lives. In what follows, we describe three channels − parents, teachers, and tutors − through which shadow education can either add to or substitute for the functions of home and school. Parents, as navigators between the different institutions in students' academic lives, have been found to influence schooling in various ways (Jerrim & Sims, 2019;Matsuoka, 2019). Parents with higher levels of schooling are, by comparison with their less-educated counterparts, more likely to be able to discuss education with their child and monitor their educational progress (Park, Byun, & Kim, 2011). Such parents tend to also, as argued by , have relatively high aspirations for their children's lives after school, which may trickle down to a student's perception that shadow education may be a necessary supplement to school (Jerrim & Sims, 2019;Matsuoka, 2019), for instance, to improve their self-esteem Otto & Karbach, 2019). Students in Hong Kong  and Bangladesh (Mahmud, 2019) have been found to perceive parents as busy individuals who have high expectations of them but do not have time to help them with schoolwork, nor possess the ability to do so. Whereas some students may have access to academic support at home , shadow education may be the only go-to academic support for other students (Matsuoka, 2019;Ozaki, 2015). Other research, investigating the relationship between shadow education and the home institution, points to shadow education providing parental relief, as parents trust their children to have completed their schoolwork during shadow education. In such instances, shadow education may result in fewer conflicts among family members about school work (Otto & Karbach, 2019), possibly improving students' satisfaction with family life (Guill, Lüdtke, & Köller, 2019), resulting in shadow education providing a benefit that the school and home cannot provide. In this study, we initiate the conversation on the role of parents by asking students to reflect on why they believe they (or their parents) signed (them) up for tutoring.
Teachers can constitute another channel through which shadow education can add to or substitute for the functions of schooling, for instance, when shadow education plays a role in providing more or better instruction to students who attend underperforming schools. Some students experience academic problems at school for which they blame their teachers' instruction (Mahmud, 2019), and resort to after-school tutors to find better instruction (Hartmann, 2008). The ethnographic study of Paramita (2014), for instance, shows that students saw tutors − and not teachers − as their main sources of knowledge accumulation. Given its role in knowledge accumulation, which may result in boosting their school performance, students can see shadow education as a tool that provides an advantage in competitive school systems with high-stakes testing and tracking , albeit labelled by them as sometimes unfair (Hajar, 2018). In the absence of school frustrations, students can also perceive shadow education as a service providing additional instruction that builds upon the instruction of their own teachers (de Guzman, Rodriguez, & de Castro, 2018;Forsey, 2013).
Whether shadow education adds to or substitutes teachers' instruction may also depend on the instructional quality of shadow education (Chan, 2019;Zheng, Wang, Shen, & Fang, 2020). Guill, Lüdtke, and Köller (2019) elaborate on three dimensions of instructional quality of shadow education: structure, challenge, and support. Structure refers to the presentation of learning content. Challenge and support refer to the way the learner is activated and the tutor-student relationship, respectively. Research on students' experienced benefits of attending shadow education has revealed varying preferences in terms of tutor focus. Whereas some students appreciate that tutors focus on content-related subject knowledge (Nam & Chan, 2019;Paramita, 2014), others value tutors that focus on memorising answers to questions listed in textbooks (Tsai & Kuo, 2008), or predicting exam questions (Nam & Chan, 2019). In most cases, students seem to experience the benefits of shadow education when tutors focus on lesson revision or on handling examination time and questions . There are, however, also students who mainly appreciate tutors that are funny, inspirational, or who offer a relaxed atmosphere .
Shadow education may also present its services by explicitly relating these to the home and school, possibly trying to ease students' school frustrations by promising to offer what students do not find in school (Hallsén & Karlsson, 2019;Holloway & Pimlott-Wilson, 2019). Students may, for instance, seek a distraction-free study atmosphere (Hunter & Cox, 2014;Kim, 2016), and tutors can focus on creating such an atmosphere. Tutors can also present themselves as providers of academic and social support, or as knowledgeable mentors that provide support that teachers may not be able to provide (Forsey, 2013). Tutoring that is presented as a service to allay school frustrations can be found in advertisements in Sweden (Hallsén & Karlsson, 2018), Israel (Addi-Raccah, 2019), Hong Kong (Trent, 2016), and Bangladesh Mahmud, 2019).

Research questions
In this study, we explore the place of shadow education in students' academic lives. As the only data on the use of shadow education in the Netherlands comes from register-based household expenditure on tutoring (CBS Statistics Netherlands, 2020) or government-commissioned research (Bisschop et al., 2019), our paper is the first empirical examination of students' experiences of tutoring in the Netherlands. Drawing on the research outlined above, we examine the place of shadow education in students' academic lives by asking students to reflect on the goals they attach to shadow education, and on the benefits they experience. We invite students to relate these goals and benefits to their school and home context . We formulated three research questions (RQ1−3): • RQ1: What are students' goals when attending shadow education? • RQ2: What benefits do students experience from attending shadow education? • RQ3: How do students relate their goals and experienced benefits to their home and school context?

Method
We conducted individual interviews with a reflective nature Pessoa, Harper, Santos, & Gracino, 2019), in which interviewees were invited to share the goals and benefits of shadow education participation.

Research context
The study was conducted in the Netherlands where, upon completion of primary education, students are assigned to one of seven secondary education tracks: basic practical vocational training (praktijkonderwijs), or pre-vocational education (VMBO BBL, VMBO GL, VMBO KBL, VMBO TL), general (havo), or pre-university (vwo) education. The latter two tracks last five and six years, respectively, while pre-vocational education lasts four years. Policy makers, educators, and researchers in the Netherlands tend to make a distinction between one-on-one tutoring (bijles), and private tutoring in group sessions, which are often focused on generic skills and homework (huiswerkbegeleiding). A recent exploration of Dutch tutoring practices reveals that most shadow education companies adhere to this distinction, but sometimes also offer hybrid forms of shadow education that contain elements of both homework support and private tutoring (Elffers et al., 2019). We used the list of cases developed by Elffers et al. (2019) to select cases in different regions of the Netherlands. From the institutes that were willing to participate, we selected three cases that only focus on homework support, three cases that focus on private tutoring, and two cases that offer a general form of support. All cases featured a physical classroom. As shown in Table I, we made sure to include respondents attending a variety of tutoring practices. Note. Type (students used the terms huiswerkbegeleiding and bijles interchangeably), Size (Small = fewer than 50 enrolled students; Medium = between 50 and 100 enrolled students; Large = more than 100 enrolled students), Tutors (NFQ = No formally qualified teachers).

Respondents
The 37 respondents were all enrolled in secondary education. These students were approached through the tutoring agency in which they were enrolled. The majority of these students were boys (n = 23) or attended the middle-stream havo track (n = 19) in schools that offer more than one track. Respondents' ages ranged from 11 to 18 years old. The duration of students' participation in shadow education varied from two weeks to three years. As can be seen in Table 1, there was also heterogeneity in terms of the size of the tutoring institutes, and in the qualification of tutors.

Data collection and analysis
With ethical approval from the Research Ethics Review Committee at the University of Amsterdam, the first author of this paper collected data from January to June 2019. We contacted providers of shadow education directly to collect basic descriptive information on the types of tutoring offered. Data were collected using semi-structured interviews. To ensure students felt safe to speak freely, we interviewed them in an empty classroom or the corner of a library, but never in their own home or school, or in the presence of tutors, teachers, parents, or other adults. Whereas some students provided elaborate answers, others were very brief, mostly providing single-word answers, resulting in interview lengths ranging from 10 to 40 minutes. Five of the final 42 students included in our sample did not return the parental consent form, and as a result, had to be excluded from the analysis. Students and parents were ensured anonymity and could withdraw from the research at any time. The interviews were conducted in Dutch, and the final sample consisted of 37 students, as shown in Table 1. An interview protocol (see Appendix A) was developed for the semi-structured interviews. The questions were about students' goals (e.g., What made you feel that you need to attend tutoring?), as well as experienced benefits (e.g., What do you know now that you did not know before and that you now know because of this tutoring?). Furthermore, the interviewer asked respondents to describe the learning activities in the physical tutoring space (e.g., Take me with you as you enter this building. What do you do here?), after which the respondent was asked to reflect upon tutoring in relation to the home and school context (e.g., What do you do or learn here that you do not do or learn at school/home?). We ensured respondents had ample time to reflect upon their own experiences.
In this study, experienced benefits refer to the value judgments students make regarding the advantages of attending tutoring (Nam & Chan, 2019). Unlike students' goals, which are based on hopes and expectations regarding the tutoring, experienced benefits are based on actual experiences. Such experiences may, however, influence students' reflection on the goals of shadow education. We account for such conceptual overlap by asking specific questions about the utility of shadow education to the individual student (e.g., How and in what ways is this tutoring useful to you?).
To evaluate the quality of the interview protocol, we conducted a small pilot. The pilot revealed that students needed additional probing, which is why in the final interviews blank cards were used for note-taking of oft-repeated words during the interview. Afterwards, which students were asked to rank the cards according to what they considered most (to least) important. Examples of words used by students to describe experienced benefits that were written down on the cards include "better in exams" or "improve a grade" for students' goals and "planning skills" or "peace at home". We controlled for the subjectivity of the interviewer by making literal notes of what students said.
Interviews were first transcribed and then analysed with ATLAS.TI using thematic analysis (Braun & Clarke, 2013). The first author of the paper read and re-read the transcripts to allow for familiarisation with the data. After such familiarisation, the first author coded the data to generate initial codes indicating students' goals and experienced benefits. The data was also simultaneously coded using a predefined coding scheme (Appendix B). Such a dual-coding process resulted in subthemes and higher-order themes (Figure 2). To ensure the reliability of the coding, two additional coders coded two interview transcripts and reviewed the potential themes. The coders discussed the coding process and the results. Figure 2 shows the thematic map extracted from the data. As the figure illustrates, students mainly expressed proximal goals (self-study, performance boost) and distal goals (career support), and formulated the benefits of attending shadow education in terms of gaining more planning skills and participating more in the classroom. Students related those goals and benefits to their school and home context mainly in referral to the amount of distractions at home, the focus of their teachers on class success instead of their individual success, and the unclear instruction of teachers. Yet, they identified their school as their primary learning context, and the home as their secondary learning context, both supplemented by shadow education. A more detailed account of students' comments is given in the following sections.

RQ1: What are students' goals when attending shadow education?
Some students expressed that they attend shadow education for independent (focused) studying, not necessarily in the presence of − or guided by − a tutor. As Max and Sebastian stated: By providing students with a relatively silent, distraction-free place to study, the physical place in which students are tutored can serve as a cue for them to get to work. Over half of the respondents indicated using tutoring for self-study. Respondents appreciated such goals of shadow education and, in some instances, referred to shadow education using the Dutch proverb stok achter de deur ('stick behind the door', an extra inclination to work), highlighting that coming to the tutoring location may serve as a signal to complete a set of homework tasks.
Alongside turning to shadow education for self-study, students turn to shadow education to boost their performance; for instance, by preparing for an upcoming exam. Such performance-oriented goals are driven by parents' perspectives, but also come from the students themselves, as can be observed in the interactions below: Researcher: Why do you think your mother signed you up for this tutoring? Toby: Purely because I need a lot of help with language subjects other than English because I know English very well. So, she thought things were going even worse this year, and that's why I'm here.
[…] I have a very noisy mother. She talks too much about my schoolwork. It's also nice that I can stay here every day until 6 o'clock so she can't whine all the time. (Toby, vwo, eighth grade)

Figure 2
Thematic map depicting themes and subthemes (solid arrows represent specifications from individual research questions to codes in the data, and dashed arrows represent possible relationships between goals, experienced benefits, and students' home and school context) Whereas the goal of boosting performance refers predominantly to short-term proximal goals (e.g., obtaining a satisfactory mark for English or attending a specific track), wishing to receive career support refers to students' perception that shadow education should bring them distal outcomes. One student, for instance, attended shadow education to seek tutoring with a view to qualifying as a medical doctor. Other students refer to such coaching in relation to attending university, as the following student commented: I think my plan for the future is set; I just want to get my secondary education to go to university and work, so that is all fine. It's not like I'm someone who doesn't know what he wants to do, then I would be less motivated. I think the homework guidance will help me […] this will play a role in my future. (Gabe, havo, eleventh grade) Gabe expresses a wish to attend university and believes shadow education will help in achieving this goal. Students with a wish for career support often mentioned their tutors as individuals who are constantly available, responding promptly to instant messages or coming to the students' home to ensure they feel supported to achieve long-term goals.

RQ2: What benefits do students experience from attending shadow education?
Respondents unanimously stated that shadow education helps to develop exam-related skills, particularly planning skills. Furthermore, respondents perceived shadow education as equipping them to participate more actively in the classroom. When asked what students would find difficult to do on their own if they did not attend shadow education, all but one respondent referred to planning for an upcoming exam or a school task. As the following two students commented: […] I also think that's due to tutoring, through the learning strategies that I have learned. You keep your planner more up to date, they [tutors] also see all the tests you have, so they will remind you of that. And yes, then I usually just start in advance for those tests, and I get high marks. (Tim, vwo, eighth grade) In the second quote, Tim explicitly says that he scores better because of tutoring and the learning strategies he is taught at shadow education. In further conversation with Tim, he mentioned that such learning strategies include the use of sticky notes to memorise parts of the school subject or using a to-do list to work on a set of tasks during the day. In addition, improving their planning skills enables students to relax at home. Students often stressed that well-executed planning results in a sense of relief for having completed their homework during the day, as is illustrated by the following quote from Suzan: I feel pretty proud when I go home, because I've really been productive for a couple of hours. Normally I am productive, but it all takes a long time and so on. So, I'm usually proud when I go home, and I'm also glad it's just done, because I have a lot of friends, for example, who don't start until 8 o'clock, while I start here at 4 o'clock. So, I always like that a lot, because I can't do anything in the evening, for example. I'm always very happy when I'm home. Then I can just eat and just go to bed.
[…] So, I always like that very much. I always feel very satisfied when I go home. (Suzan, vwo, ninth grade) Other benefits that were mentioned by students indicate that, due to their shadow education attendance, students find themselves participating more in the classroom, and may be more inclined to ask their teachers questions on matters they do not understand, as the following comment elucidates: Steven: On my own, I don't ask as many questions but only when I need to. Here, I do. They [tutors] make sure you start asking questions, so because of my interaction with them, I do. I don't know; it's hard to explain. […] I've learned here to ask questions, and because of that, I raise my hand sooner in class. Researcher: Why is that? Steven: I think that's because you start to see that asking for help is going to help you. (Steven, havo, tenth grade) Echoing several other respondents, Steven explicitly mentioned that asking more questions in class occurs because of their interactions with tutors. Students, for instance, expressed that during a tutoring session tutors inspire them to ask questions on unclear matters. In some instances, tutors share their personal stories of how asking questions helped them to succeed, resulting in tutors serving as role models to the students, possibly provoking the students to set out − and work towards − certain goals.
Possibly due to their perceived improvements in planning and classroom participation, students find, since enrolling in shadow education, that they have become more effective in test-taking preparation for an upcoming exam. Various students mentioned that when an exam is nearing, tutors focus heavily on such exams; for instance, by quizzing the student. As the following comments illustrate, Tim and Thomas value such quizzing:

RQ3: How do students relate their goals and experienced benefits to their home and school context?
When asked about their goals and experienced benefits in relation to their home and school context, respondents described both the physical and symbolic place that shadow education occupies in their life. In doing so, students reflected on the amount of distractions at home and their teachers at school. Physical place. The students expressing a self-study goal were also the ones reflecting on their rooms at home as unsuitable places to study, for instance, due to associations with or experiences in their room, as the following two comments from Suzan and Karen elucidate: I've heard that it's much better for you to have, say, a place where you're really going to learn and that that's not your room, for example, because you also get negative feelings from your room or something. So, it's just nice to have a place where you can really learn or something. When I come in here, I just know, I have to work. When I enter my room, I think oh maybe I should clean up or sleep. So, I do that. (Suzan, vwo, ninth grade) At home, it's very busy, and then I ask my parents to please be a little more quiet because I'm busy with my homework. Then they are, for example, watching TV very loudly.
[…] I often have visits from nephews and nieces; they come and if I have homework, they are running around. That's not nice. I find it better here, […] it's quiet, and I can ask the questions I want. (Karen, havo, eighth grade) Suzan and Karen appear to value shadow education, particularly as a physical place. They appreciate the relaxed and home-like atmosphere at the tutoring place. One tutoring company's website explicitly states that a shadow education institute should feel like home. The website mentions that if students arrive after a whole day at school, tired and worn out, they should find a place where they can relax first. The tutor welcomes the students, then offers them a cup of coffee or tea, often with a snack. The students then go to study rooms, which are set up in such a way that each student place is shielded off by partitions, providing a silent place where students can study.
In their description of the physical place, some students also described their shadow education institute as a place where you sit at tables that are set up in the room, usually with a tutor providing group-based or individual instruction. Other students described and valued their shadow education institute as a place that is similar to a classroom, with tables arranged in a way that also allows for one-onone interactions between tutors and students. Figure 2 shows three illustrations of students' descriptions of shadow education as a physical place that is different from the school and the home.
Symbolic place. In addition to providing a physical place to study, shadow education appears to occupy a symbolic place in students' lives, as a meaningful place where they find the instruction and support they seek but do not find in school or at home. When asked what the difference is between shadow education and their school, one student commented: "It is actually the same as what you do at school, but then you get a more specific explanation." This comment illustrates that students may perceive that shadow education and their regular school have the same educational purpose, however in some cases tutoring is better in realising this purpose. Indeed, some tutoring places present themselves as solving the problems that students encounter at school. Alongside solving their schooling problems, students reflected on the tutors as individuals who are always there to help, even during late-night hours. Constant availability of support is unlikely to be offered by the student's school or parents, and shadow education could fill this gap in students' academic lives.

Discussion
This study explored students' goals and the experienced benefits of their participation in shadow education. By asking students what they hope to get out of shadow education (goals) and what it offers them in practice (benefits), and by asking them to relate these goals and benefits to their home and school context, we tried to unravel what it is that students seek and find in shadow education that they do not seek or cannot find in their home or school context.
With regard to students' goals when attending shadow education (Research Question 1), our data show that respondents turn to shadow education in most cases with the goal to self-study, to boost their performance, or to receive career support. In terms of the benefits of shadow education that students experience (Research Question 2), our study shows that respondents experience improvement of skills such as planning, but also that shadow education equips them to participate more actively in the classroom. Respondents' reflection of these goals and benefits in relation to their home and school (Research Question 3) indicated that respondents assign shadow education its own particular place in their academic lives. Students indicated that shadow education caters to goals that the school and home are less able to serve (e.g. distraction-free studying), and that the experienced benefits of shadow education are unique to the institution (e.g. individual support). Given that students perceive shadow education as beneficial and catering to their goals, the practice becomes meaningful to them, both in the physical sense − as a place to study without distractions and in the symbolic sense − as a place where they can find the support and encouragement they seek.
The distinction between the physical and symbolic place provides indications that, in the Netherlands, shadow education does not replace the place of schools or homes, but occupies its own position in students' academic lives. Such position of shadow education as a physical and symbolic place bears a resemblance to the ideas of Oldenburg (1999), who investigated the emergence of recreational spaces like cafés and libraries. Such locales, which he labels "third places", possess certain distinctive qualities that the home (first place) and work (second place) do not possess. As Oldenburg and Brissett (1982) stated, the ingredients of third places are "elusive and emergent, changing with the shifting patterns of lifestyle" (p. 270). Physical places are believed to obtain meaning (i.e. to become symbolic places) to their users if such places are conducive to their goals (Purnell, 2015). Given that the students in our study expressed that shadow education caters to their goals, we argue that shadow education can be conceptualised in line with the ideas of such a "third place".
The conceptualisation of shadow education as a third place builds upon, but does not echo, student-centred work on shadow education; for instance in the United Kingdom (Hajar, 2018(Hajar, , 2019, Bangladesh (Mahmud, 2019), and Israel (Addi-Raccah, 2019). Whereas in such previous studies students report relatively negative student experiences regarding shadow education and its examination-orientedness, none of the participants in our study expressed such negative experiences. The difference in results may be explained by the different schooling systems in which the research was conducted, but also by the fact that our phrasing of the questions was open, whereas previous research intentionally asked students to recall negative experiences. Our study points out that, from students' perspectives, shadow education can offer meaningful places to self-study, to receive career support, and to improve performance. Although students had ample opportunities to voice negative experiences, even discontent, with their shadow education, all respondents did not. Thus, instead of referring to their experiences with shadow education as positive or negative, students mostly embrace its benefits.
To fully understand the role of shadow education in students' academic lives, we also need an examination of the experiences of students who are not enrolled in shadow education, which we were not able to do in this study. Some of the participants did express emotions such as relief and satisfaction as a result of their shadow education attendance. Future research could examine the way that non-tutored students feel about their educational careers and opportunities in comparison to tutored students. Non-tutored students can, for instance, feel insecure about their performance and abilities when they notice that other students enrol in shadow education. To further understand such experiences, interview studies among non-tutored students would be needed. Such research can further our knowledge about shadow education as an increasingly prominent, and potentially influential, practice in students' schooling.

Conclusion
Overall, this study strengthens the idea that shadow education is becoming a reality in students' academic lives, a process that policy makers, educators, and parents do not always know how to respond to. By identifying shadow educations' distinctive qualities, our study can inform and guide educational practitioners, policy makers and parents in their response to shadow education as an emerging practice that is increasingly finding a manifest place in the academic lives of students in the Netherlands, and worldwide. Learning nowadays happens in various contexts. It is "far more individualistic, in terms of its form and its content and purpose" (Biesta, 2006, p. 18). This individualisation of learning has led to "a rapidly growing market for nonformal forms of learning" (Biesta, 2006, p. 18), which, in terms of English, is coupled with its global spread as a lingua franca of international communication. Consequently, the English language is found to be the most popularly enrolled subject, especially in countries where it is learnt as a second or foreign language (Yung, 2019), with the Czech Republic being no exception. Hamid et al. (2009) even propose that private tutoring in English (PTE) is "a special and important subclass of private tutoring […] a macro-phenomenon of modern education" (p. 282). Languages have always had their economic value; Díaz (2015) argues, however, that the scope of the commodification of languages has increased in the world since languages have become a benefit or an asset to job seekers globally. Analysing recent developments in France, Díaz (2015) uncovers the commodification of English through advertising campaigns − English is represented by images that link knowledge of this language to success, i.e. to a higher salary and better job. Referring to PTE in Bangladesh,  propose "a silent competition between the curricular and non-curricular sectors" (p. 871), the former being challenged in maintaining the status quo as the latter gains more ground. These examples from two different countries provide evidence that PTE has become a significant part of educational systems that cannot be ignored by educational researchers and policymakers. This article aims to explore private tutoring in English (PTE) in the Czech Republic through the eyes of its recipients. First, the topic is linked to the broader context of shadow education; then the research on PTE is reviewed, followed by an introduction to the Czech context of the study and the study itself. Finally, conclusions shedding some light on PTE are drawn.

Private tutoring
Around the world, a significant number of students receive some type of private tutoring (Yung, 2019). This widespread phenomenon is referred to as a shadow education system. 1 Bray (1999) justified the use of the metaphor of a shadow in the following way: First, private supplementary tutoring only exists because the mainstream education exists; second, as the size and shape of the mainstream system change, so do the size and shape of supplementary tutoring; third, in almost all societies much more public attention focuses on the mainstream than on its shadow; and fourth, the features of the shadow system are much less distinct than those of the mainstream system (p. 17).
Private tutoring is a complex phenomenon since it may have various forms, involve numerous agents, and occur in a variety of contexts. Consequently, various definitions of the term are available in the literature on shadow education. In this article, private tutoring is understood as "tutoring in an academic school subject (e.g., mathematics, history, or English), which is taught in addition to mainstream schooling for financial gain" (Silova, Būdiene, & Bray, 2006, p. 29). This definition of private tutoring includes private tutoring lessons (offered by individuals) and preparatory courses (offered by institutions). In the present study, however, the focus is on the former.
Concerning geographical dispersion, private tutoring is determined especially by cultural, educational, and economic factors. Bray (1999) suggests that private tutoring is likely to be widespread in Asian cultures, which place a strong emphasis on effort as a factor explaining and determining success. In contrast, European and 1 Apart from this expression, which is used in Western countries, there are country-specific terms traditionally denoting private tutoring: Buxiban in Taiwan, Hagwon or cram schools in South Korea, Yuku in Japan (Chung, 2019).
North American cultures are more likely to emphasise ability and thus private tutoring is not so prominent there. Relatedly, Hamid et al. (2009) propose that in East Asia private tutoring assumes a position as an essential part of education, culture, and society, while in Anglophone and European countries it is principally remedial and supplementary in nature. Recent developments, however, show that, for example, private tutoring in Australia has grown immensely in response to neoliberal educational reforms, which have intensified competition in schooling (Sriprakash et al., 2016).

Providers and recipients 2
Reflecting the changes in the mainstream education system, shadow education providers make tutoring readily available to potential recipients. The providers typically offer a number of modes of tutoring. The four main modes are as follows: one-on-one tutoring, small-group tutoring, lecture-type tutoring, either live or video-recorded, and online tutoring (Zhan et al., 2013, p. 498). The present study focuses on oneon-one tutoring, i.e. on situations in which a single tutor works with one tutee at a time.
Diversity may be found in the qualifications of tutors (Silova, Būdiene, & Bray, 2006). Teachers are the main providers of private tutoring. Bray (1999, p. 37) distinguishes two situations: (1) tutors are teachers in the mainstream system who tutor their own students and receive additional payment; (2) teachers tutor students for whom they otherwise do not have any responsibility. Obviously, if the first situation is the case, it raises serious ethical issues. In recent studies, depicting such circumstances in Bangladesh and India, Hamid et al. (2009) and Chatterjee (2018) explain that poor teacher salaries are the reason for such practices since private tutoring yields an extra income. In some other countries, for example in Singapore or Morocco, tutoring one's own students is forbidden (Bray, 1999). Contrastingly, in the Czech Republic, the shadow education system is ignored by educational policy (Šťastný, 2016).
Apart from teachers, there is a considerable variety of tutors in the shadow education system in terms of age, qualification, employment status, workload, etc. An important cohort of tutors is that of university students, who, according to Zhan et al. (2013), often work as part-time tutors; they commonly supplement their incomes by tutoring secondary school students (Bray, 1999).
Private tutoring may be offered to recipients through a company or a self-employed tutor. Given the individualistic nature of learning nowadays (Biesta, 2006), some students may choose to respond to the supply and buy the service, i.e. private tutoring in a particular subject. The services that professionals provide are characteristically different from the goods that are sold by a manufacturer or a retailer in 2 Bray (1999) refers to the two groups of agents as "producers and consumers" (p. 37). that they are intangible and the purchaser has to take them on trust (Macdonald, 1995).
Regarding the choice to use private tutoring, Jokić et al. (2013) established a conceptual framework which places the previously defined decision concerning the use of private tutoring at the centre of the system. This ecological system […] is divided into five socially organised subsystems, representing a set of nested and interconnected structures that all have a potential to influence the decision and upon which this decision can have a reciprocal effect. These structures range from the immediate individual characteristics of the pupil to the most remote setting of the larger society (Jokić et al., 2013, p. 28).
The system is illustrated by five concentric circles; annuli represent individual subsystems: pupil, parents, school, educational policy, and society. In his theoretical study, Šťastný (2015, pp. 60−61) discusses this system and summarises the authors' assumptions leading to placing the previously defined decision at the centre of the scheme: (1) the decision usually depends on pupils or their parents; and (2) it is tightly linked to their attitudes to the educational system and educational aspirations. Following the framework of Jokić et al. (2013), the decision concerning the use of private tutoring is likely to be the result of complex interactions of the five subsystems.
Research findings show that those who engage in private tutoring are mainly students at the secondary level; the dominant group comprises students whose performance is already good, and who want to stay competitive (Bray, 1999). Frequently reported reasons for private tutoring include remedial or enrichment purposes (e.g. Zhan et al., 2013), and, also, securing an educational advantage (Sriprakash et al., 2016). The next part deals specifically with PTE; it shows which aspects of current PTE are of interest to researchers and provides a mosaic of findings.

Private tutoring in English
Yung (2019) asserts that research in the field of shadow education, particularly in language learning, is still in an infant stage. In 2009, Hamid et al. reported only two research studies that focused exclusively on PTE. Considering recent studies in this area, two major strands of research may be observed. The first one includes studies which investigate the effects of PTE on students' achievement in English (Andabati et al., 2018;Chang, 2019;Chaterjee, 2018;Hamid et al., 2009;Kim et al., 2016). These quantitative studies addressing the effectiveness of PTE relate students' participation in PTE to their academic achievement, i.e. their performance on various standardised tests. The studies investigate the phenomenon in the countries where shadow education has traditionally been an established enterprise 3 (e.g. Hong Kong, Korea, Taiwan) or in less affluent countries, where it is believed to be a necessity 3 Research papers report considerable investments by families in private tutoring in wealthier East Asian countries, which amount to billions of dollars annually (Bray, 1999;Hamid et al., 2009). to compensate for the inadequacies of formal tuition in English (e.g. Bangladesh, Uganda). Interestingly, all the studies confirmed a positive effect of PTE and identified some other influential factors (Andabati et al., 2018;Chang, 2019;Hamid et al., 2009;Kim et al., 2016).
For example, Hamid et al. (2009) conducted a mixed-methods study in which they investigated PTE in Bangladesh, more specifically tenth-grade students' perceptions of PTE. In the quantitative part of their study, they analysed questionnaire data and the students' test results (n = 228). Multivariate logistic regression confirmed a considerable impact of PTE and disclosed the significance of some other variables (the mother's level of education and gender). Chang (2019, p. 86) emphasised that the strength of PTE became minimal when the self-study variables were added to the research − the time spent on self-study activities had a significant and positive effect on Taiwanese students' performance. Kim et al. (2016) investigated the effects of the time and cost of private tutoring on Korean students' achievement in English and mathematics. Concerning English, they concluded that in the Korean context it is likely to be effective to invest in the quality of tutoring, i.e. in a more costly service, since the cost of tutoring was significant in all models when the tutoring time, i.e. average hours a week spent for attending English and math tutoring, was held constant. Contrary to that, the tutoring time was not statistically significant (Kim et al., 2016, p. 253). Using structural equation modelling, Andabati et al. (2018) investigated the significance of private tutoring in improving English language literacy in learners in Uganda. They confirm the positive impact of PTE but also suggest several additional factors, such as age, gender, pre-school education, type of school, and household structure, which influenced the level of literacy.
The second strand of research in PTE aims at investigating learners' experience, attitudes, and motivations. There are few studies providing a better understanding of English language learners' participation in private tutoring. In the qualitative part of their study Hamid et al. (2009) interviewed 14 selected students out of a larger sample (n = 228). The analysis of the data uncovered the fact that the students perceived the quality of English teaching at schools as being poor, which made success in learning impossible to achieve without private lessons. Consequently, they saw PTE as imperative; in other words, PTE was not only socially desirable but also inevitable and, for some learners, even preferable since they did not believe in the quality of formal education.
In his study, Lee (2010) examined the English-learning experiences of the students in a university programme in detail and focused on the ways in which they were similar and different in terms of their English-learning backgrounds and proficiency. On the basis of the survey and in-depth interviews, seven cases out of 43 were selected for the study. The findings of the study suggest that the students' levels of English proficiency varied immensely even within the same programme because of their different private English learning experiences. Furthermore, these students reported diverse motivations and strategies for learning English outside school.  investigated Chinese learners' accounts of their PTE experiences and reflections on those experiences through narrative inquiry. The findings of the study suggest that the 14 participants upheld the importance of achieving communicative competence through PTE but at the same time accepted the fact that PTE was better equipped to address examination-oriented needs. In conclusion, they were comfortable with this seemingly contradictory view of PTE.
Another study by Yung (2019) is more specific since it focuses on the dynamic nature of student motivation during PTE. The study is the first attempt to analyse the L2 motivational selves (Dörnyei, 2009) of PTE learners through their language learning narratives. The findings suggest that learners receiving PTE possessed dominant ought-to L2 selves, suppressed ideal L2 selves, and insecure actual L2 selves. Their low self-efficacy and/or high aspirations to perform well in the public examination in their actual L2 selves created a discrepancy between their current and future L2 selves (Yung, 2019, p. 129).
Nevertheless, none of the studies which were discussed above investigated oneon-one tutoring; all of them focused on other types of private tutoring, especially on group formats including lecture-type supplementary tuition. This study focuses on one-on-one tutoring because of its significant features (Bleistein & Lewis, 2015), namely providing personalised learning that cannot be achieved in larger language classrooms, addressing an individual's special strengths in a way that is impossible in larger classes, and creating space for learner agency. One-on-one tutoring is an organisational form that enables the implementation of teaching practices helping students to enact and enhance their agency in ways discussed, for example, by Larsen-Freeman (2019) including learner-driven feedback etc. From the perspective of PTE in the Czech Republic, the study explores an under-researched area (part 2.1). Another asset of the study is that it provides insights into PTE in the European context, which is also rare.

Context of the study
In the Czech Republic, English maintains the status of a foreign language, which means that learners of a foreign language learn a non-native language in the environment of their native language (Gass & Selinker, 2008). Therefore, the characteristics of English language learning is this country correspond to those of foreign language learning as proposed by Cameron − "very little experience of the language outside the classroom, and encounters with the language (…) through several hours of teaching in a school week" (2001, p. 11). These characteristics apply to the majority of learners, though, obviously, there may be inter-individual differences.
Concerning formal education in the Czech Republic, the curricular reform in the first decade of the 21st century established a new system of curricular documents (Greger & Walterová, 2007) including Framework Education Programmes for pre-school, basic 4 (MŠMT, 2017), and secondary education (e.g. VÚP, 2007). Schools in the Czech Republic are autonomous institutions; they construct their own School Education Programmes, which are in alignment with the respective frameworks. English, as a Foreign Language, falls within the educational area called Language and Communication through Language. The position of English among other foreign languages is prominent since it should be the first foreign language to be taught. Compulsory education in English starts in grade three, though it is possible to start earlier, depending on a school's vision and resources (MŠMT, 2017). Most commonly, there are three mandatory lessons of English per week in each year from grade three of basic school until grade four of secondary school. Thus, a typical secondary school graduate, excluding, for example, graduates from international schools, bilingual secondary schools, etc., has experienced three lessons of English a week for a period of 13 years before taking the school-leaving exam. The exam is known as the maturita exam. 5 It is not competitive, since it does not regulate entry to prestigious study programmes. Czech universities often rely on other test providers, e.g. the SCIO company, 6 administer their own tests, or do not require a test to be passed as a criterion for admission. In 2019, 84.5% of all applicants for Czech universities were accepted for the 2019/2020 academic year (MŠMT, n.d.).

PTE in the Czech Republic
As regards the Czech Republic, little is known about the shadow education system in general, including PTE. Šťastný (2016) reports a certain lack of interest in this topic on the part of researchers, which is in contrast with recent trends in Asian countries and elsewhere in Europe. Therefore, his research project represents a substantial contribution to the body of research on shadow education in the Czech Republic. Though his research does not target PTE only, it sheds some light on private tutoring in this subject-specific domain.
The study by Šťastný (2016) uncovers the fact that 59% of the 465 secondary school students who took private lessons had experienced private tutoring in foreign languages, mainly in English. The students reported various reasons for participating in private tutoring, 7 the most frequent being dissatisfaction with school results, preparation for the school-leaving exam, parents' decisions, and a desire to learn something beyond the curriculum (Šťastný, 2016). The reasons for students' participation in private tutoring were also the focus of Černá et al. (2016), who investigated 4 Basic education in the Czech Republic comprises primary and lower-secondary education (ISCED 1 and ISCED 2 levels). 5 Since 2011 the reformed maturita exam has been implemented. Nowadays it consists of a common (state) and a profiling (school) part. For the 2019/2020 school year the common part includes a mandatory exam in the Czech language and a mandatory exam in the first foreign language (i.e. mostly English) or mathematics, from which every testee is obliged to choose one. 6 More than 50 faculties in the Czech Republic use SCIO tests for admission procedures (SCIO, 2019). 7 Here private tutoring means tutoring in all kinds of subjects, foreign languages (English) and mathematics being the most frequent ones. the learning histories of Czech speakers 8 of English and their achievement in selected communicative language competencies. The quantitative part of this mixed-methods study yielded data concerning individual students' participation in non-formal learning while at basic and secondary school. As regards individual students' involvement in PTE while at basic school, 28.6% of the 228 students took courses and/or private lessons. As for the reasons for their participation in PTE, the most frequent ones were related to the pleasure of learning English, remedial needs, advancement needs, and parental decisions. While at secondary school, 22.2% of all the students participated in PTE. The recurrent reasons for their involvement were related to exam preparation, remedial needs, and the pleasure of learning English (Černá et al., 2016, pp. 79−83).
Both studies confirmed the influence of parents' educational status on the involvement of their children in private tutoring (Černá et al., 2016, p. 88; Šťastný, 2016, p. 180). Otherwise, the findings of the two studies can hardly be compared because of conceptual differences; nevertheless, there is a considerable level of correspondence between the reasons for private tutoring, irrespective of the subject and the reasons for PTE.
Šťastný (2016) also explored the use of various organisational forms of private tutoring in the Czech context; the results indicate that individual lessons prevail (78%) over small groups (22%); however, the occurrence of small groups in PTE was much higher compared to mathematics (32% vs. 15%). The author attributes the difference to the nature of the subject matter.
Furthermore, Šťastný (2016) also disclosed who the providers of private tutoring in the Czech Republic were. Most frequently, a teacher from a different school provided private tutoring (36%). Among other providers, he also identified a native speaker and language agency tutor (6%), who are specifically related to foreign language education.
In conclusion, PTE in the Czech Republic remains very much in the shadows. Therefore, this study aims to provide insights into the perceptions of those who consume PTE, its recipients.

Method
The present study is linked to previous research on individual learning histories in that it analyses a segment of data obtained earlier in the context of a large interdisciplinary study. More specifically, the study is based on recollective data elicited by means of narrative interviews (Creswell, 2011;Hendl, 2012) within an earlier research project (Černá et al., 2016). This article reports a secondary analysis of the data from a new perspective, i.e. a supplementary analysis (Heaton, 2008). In this 8 First-year students in the English language teacher education study programmes (i.e. successful learners of English) at three Czech universities (Palacký University in Olomouc, University of South Bohemia České Budějovice, and University of Pardubice) constituted the research sample. study an emergent issue of the data that was only partially addressed in the primary study, i.e. one-on-one tutoring in English, is analysed in greater depth.
The aim of this study is to uncover the nature of the PTE that the Czech students of English experienced from early childhood to their entry to the university and the perceived personal significance of this type of education in their lives.
Consequently, the following research questions were formulated: 1. What is the nature of the PTE that the students experienced? 2. How do the students perceive the role of private tutoring in their learning English?
The respondents in this study are Czech speakers of English at the onset of their university studies of English, i.e. on entry to bachelor's study programmes preparing students for English language teacher education study programmes at three Czech universities. The research sample included 20 respondents (Table 1), 13 females and seven males, 9 who were purposefully selected from a larger group of 228 students according to the nature of their learning experiences. The selection procedure was based on questionnaire data, which enabled the researcher to divide the students into four groups. The first group included the students who learnt English mainly in a formal educational context; non-formal and informal learning was marginal in their lives. The second group involved the students who used English extensively outside the school for various purposes. The students in the third group learnt the language in a variety of contexts; they learnt English at school, attended private lessons and courses, used English outside the school, and travelled to English-speaking countries. Lastly, the fourth group comprised the students with exceptionally long residence in an English-speaking country. Moreover, each group had two sub-groups: the students whose attitudes toward English and learning English were positive all the time and the students whose negative attitudes had undergone considerable modifications in the course of their lives. After the students had been grouped, 20 respondents to be interviewed were chosen randomly from each group. The selection procedure was influenced by the unwillingness of some students to cooperate and by students' availability in the time period allocated to the interviews in the individual towns. After conducting a pilot interview, one researcher, the author of this paper, interviewed all the respondents at their universities in June and July 2014 and in February and March 2015. All the interviews were conducted in the Czech language, 10 the students' mother tongue, in order to prevent potential problems stemming from the respondents' inability to produce an extensive narrative in English. All the students who were interviewed were given nicknames to protect their identity.  Apart from following general guidelines for interviewing, the researcher respected recommendations for conducting narrative interviews (Elliott, 2005;Hendl, 2012). After a short introduction, the researcher asked questions to elicit the main narrative, i.e. questions directed to specific times (the pre-school, basic school, secondary school, and post-secondary periods), and to specific situations (learning English in various contexts, including a non-formal context). If necessary, the researcher asked a question to learn more about a particular period in the interviewee's life, about a topic of interest, etc. In order to maintain the momentum of the interviews, the researcher implemented some strategies, e.g. using reinforcement feedback, attention-focusing devices, or various probes, recommended in the relevant literature (Dörnyei, 2007, pp. 142−143). All the interviews were audio-recorded; the average length of an interview was 50 minutes and 19 seconds.
Concerning the analysis, potential limitations of the secondary analysis (Johnston, 2014) were minimised considerably by the author's involvement in the primary study. The author of this paper was familiar with its context and procedures and collected and analysed the data. The process of the qualitative data analysis contained all the steps recommended by Creswell (2011). The recordings of the interviews were transcribed using commented transcription (Hendl, 2012). A single researcher using open coding supported by the Atlas.ti7 software conducted a subsequent analysis: after a preliminary exploratory analysis, the researcher analysed the data using the constant comparative method (Silverman, 2001). Categories were developed inductively through "comparing incidents in the data to other incidents, incidents to categories, and categories to other categories" (Creswell, 2011, p. 434). As regards the trustworthiness of the study, it builds on the credibility of the primary study, which was ensured by having an external audit conducted by an expert from outside the research team.

Results
The analysis of the interviews confirmed the individualised nature of learning nowadays. The study, however, identified a group of students for whom PTE was not directly relevant. None of them ever mentioned considering PTE as an educational option, even though indirect questions addressed the issue. 11 Out of twenty informants, five people had used PTE in some period of their lives. The nature of the experience, however, differed in many aspects. The next part presents accounts of the individual respondents' experience of PTE in order to illustrate its nature, and then the answers to the research questions are provided.

Students' experience of PTE
Facing the school-leaving exam, which included English, Alice needed encouragement and reassurance: "…before the maturita exam I felt respect; I was afraid" (2:2). A series of ten lessons with a private tutor, which she called a course, was intended to help her gain confidence. Though she did not learn any new content − "we revised this and that … and he did not teach me anything new, we just revised" (2:5) − she obtained the support that she needed. Concerning PTE, it was just an episode in her life: "It was the only occasion on which I needed such help… otherwise there was no need for it" (2:7).
Similarly, approaching the completion of secondary school encouraged Frances to use PTE, though her reasons were different. One year before the school-leaving exam, Frances made the decision to study English at university. At the same time, she felt dissatisfaction with the English lessons at school: "… the teacher taught us just basic grammar… she did not explain any exceptions to the rule, […] tests were very simple […], during the lesson she asked us to talk in pairs and the students talked in Czech and about something totally different; she walked among them and said nothing, […] she gave us easy tasks…" (7:7). Consequently, she realised that the school she was studying at would not prepare her for university: "I knew that with what we were learning at school I would not make it" (7:1). Then she found a private tutor, a university student in the study programme that Frances planned to apply for, to get her ready for the entrance exam. She attended private lessons twice a week for a year.
In her learning history, Cindy underwent several episodes of motivational fluctuations; whenever her motivation declined, it was in response to the harmful effects of the teacher or the learning experience (Černá, 2015). When this happened, her mother always did something to counterbalance her loss of motivation. Apart from assisting her daughter with home assignments or finding a penfriend, the main intervention strategy that she used was hiring a private tutor. It first happened when Cindy's marks in English became average in grade five after the arrival of a new teacher whose teaching style differed considerably from that of the previous one. Cindy was desperate about her school results, which defined the main aim of private lessons: "[the tutor] tried to prepare me for those tests in order to have better marks" (4:7).
Later, while at an eight-year secondary school, 12 Cindy experienced a large class of learners. She felt that she had a limited chance to be active in English lessons: "My turn never came so that I could practise, so I had private lessons to improve in the language" (4:8). Initially, Cindy felt that she was being pushed too much by her mother to attend private lessons, but later started to appreciate her private lessons: "Without the extra lessons … I wouldn't have been able to make progress" (4:4). At the same time, however, she perceived differences in the quality of individual tutors' lessons: "it seemed to me that she [the second tutor] didn't know what to do with me … as if she was waiting for me to tell her what I needed, but I didn't know it at that time" (4:10). Another tutor was much more successful in addressing Cindy's needs: "She prepared various tasks for me; I liked it; she also prepared listening tasks, which I didn't do with the previous tutor" (4:13).
Concerning Wendy, she started to attend private English lessons as a pre-schooler together with the son of her mother's friend. The lessons lasted for years and provided space for enrichment of the curriculum or for remedial practice: "In those private lessons we continued [with the subject matter] on and on, and we learnt something new and when there was a problem at school, we dealt with the problem" (21:7). Interestingly, she developed a pleasant friendship with her 'classmate', which motivated her to attend the private lessons: "We used to go to her [tutor's] place, and we both liked it, and we looked forward to seeing each other very much after a week" (21:10). While at secondary school, Wendy attended the lessons on her own, since there was a need to link the content of PTE to the lessons at school: "…we mostly did the same as we did at school to deepen my knowledge" (21:8).
Though different in many aspects, the experience of the four students with PTE shares a common characteristic: there is almost no mention about who the tutors were. The students mostly refer to them as "someone", "a man", "a woman", "a colleague of my parents", "an acquaintance of my parents", "a (university) student". The only exception was Frances, who, having found her tutor herself, knew exactly who she was. None of the respondents used the word "teacher" when explaining who provided the private lessons.
Unlike the previously mentioned students, George went through a substantially different experience with PTE, since his tutor was a native speaker of English, more specifically a middle-aged American woman. George is a successful learner of English who started to learn the language as a pre-schooler and also went on a number of courses. While at an eight-year secondary school, he was critical of his English teachers and their lessons. At the same time, he studied English autonomously and even provided PTE to his peers. He was active in searching for opportunities to communicate in English and, therefore, he hired the native speaker tutor. His aims were quite clear; he sought to improve his speaking, especially pronunciation and pragmatic aspects of communication. Furthermore, he was interested in learning about the culture of the US.
During an introductory lesson, however, the intended private tutoring changed into a reciprocal service. "I attended the first lesson of conversation. I was there and we talked and enjoyed each other's company […] then it progressed in the way that I taught her Czech and she taught me English, so nobody paid anything" (8:3). George appreciated this arrangement since paying by this reciprocal service enabled him to save a sum of money per lesson which he considered relatively high at that time. He was the only respondent who mentioned financial aspects of PTE.
In the course of time, George's relationship with the American tutor changed into a friendship: "And then we made friends, also with her family, with her son, who is also an American, and we sometimes visited each other, went for dinner, or what-ever…" (8.4). Thus, George's contacts with the American family became motivated by social relationships, which the English language mediated: "We often used to go out together […] so, we spent a lot of time together and he was a native speaker, whom I used to see a lot… not as a teacher, but it was conversation, wasn't it?" (8:5). Apart from having a companion, George acknowledged that "thanks to him I certainly learnt a lot" (8:6).

The nature of PTE
In response to the first research question, the nature of the PTE that the respondents experienced in their lives will be discussed in terms of intensity, reasons, recipients, providers, and the decision to use it.
The intensity of the PTE varies from short-term (a series of lessons), to long-term events (years) depending on the reasons individuals have for using it. Those may be clearly defined, e.g. preparation for an exam (Alice and Frances), or may be variable and responsive to changing circumstances (Cindy and Wendy). The reasons for PTE, as they were identified in the respondents' narratives, reflected their cognitive, affective, and social needs in a particular phase of formal education. Cognitive needs seem to have been more relevant for Frances, Cindy, and Wendy while she was a secondary school student. Affective needs are discernible among Alice's reasons for PTE. Social needs gradually gained importance for Wendy, when a social relationship with another pupil motivated her to participate in PTE, and for George, whose relationship with the family of his American tutor outweighed the learning outcomes.
Both basic and secondary school learners become recipients of PTE, Cindy and Wendy started to take private lessons as primary 13 school learners.
Concerning the providers of PTE, the analysis uncovered the fact that they are very much hidden in the shadows; little is known about their qualifications and expertise since the respondents mostly used various indefinite expressions to refer to the tutors and did not provide specific information. This may have several causes; for example, for a variety of reasons, the respondents did not possess information about the tutors' qualifications and/or educational background or, given the nature of the study, they did not remember it. The question is whether the parents who initiate the use of PTE consider providers' qualifications or not. In spite of being in the shadows, the providers enjoy the trust of those who buy their services − they are called to action to remedy problems which originate within the mainstream system (see Cindy's reasons for PTE), or to substitute for some of the responsibilities of the mainstream system (e.g. preparation for the school-leaving exam, as in Alice's case).
Native speakers of English represent a group of tutors that is specific to the field of foreign languages. Considering George's reasons for finding a native speaker tutor, they perfectly match the assets that native speakers bring to foreign language classrooms, i.e. communicative competence in English and knowledge of the target culture.
As regards the decision to use private tutoring, in the present study, it was the parents who decided to use PTE when it concerned a primary school pupil. This was evident in Cindy's and Wendy's accounts. While at secondary school, the students made the decision themselves, presumably with their parents' financial and other support − with the exception of George, nobody even mentioned the costs of the service.

The role of PTE in the students' lives
Concerning the second research question, it may be concluded that the Czech students in the present study who experienced PTE perceive its role as positive, with occasional reservations, and appreciate its impact on their learning of English and also on their social relationships.
Furthermore, PTE assumed different roles in the respondents' lives. PTE is supplementary in the sense that it provides what the mainstream system fails to provide (Alice, Frances, Cindy, Wendy) or supplementary, indicating that it provides additional space for learning which is not, for some reason, available in the mainstream system (Wendy, George).
Noticeably, the five students' participation in PTE is related to formal learning of this language. In the cases of Alice, Frances, Cindy, and Wendy their involvement in PTE really shadowed the mainstream education system: the shadow system reacted flexibly to students' needs that were triggered in the formal system. For George, self-initiated PTE opened up a space for progress which the mainstream system could not provide.
There were students, however, in whose lives PTE did not assume any direct role. On the basis of the similarity of their histories, two groups of students may be identified.
Rose, Lara, and Paul learnt English predominantly at school. Rose enjoyed every aspect of her schooling, something she repeated many times during the interview. She considered herself a successful learner of this language since her success in communication with foreigners outside school confirmed her competence. Similarly, Paul succeeded in learning English at school and he was able to communicate with native speaker teachers. Even before the maturita exam he passed an international exam, which certified the B2 level of proficiency in English. Lara was different in that she was not satisfied with her teachers of English. Her dissatisfaction was not limited to English but concerned the school she was attending at that time. Her parents took action and encouraged her to attend a different school, which brought about a change for the better, including English.
The rest of the group perceived English as a medium of communication or a tool to achieve something rather than a school subject to learn. This was reflected in their use of English in their everyday lives in a range of activities. While some of them were successful learners at school, some others were not, but it did not seem to matter. Some students were proactive in finding opportunities to communicate in English face to face or online. For Irene, Bill, and Simon, what was substantial in their lives was communication with native speakers. Bill and Simon, similarly to George, developed personal relationships with their former native speaker teachers. Irene enjoyed her lessons with a native speaker teacher on an intensive course after leaving her secondary school. Being able to communicate gave her confidence and provided motivational incentives for further study of English at a university, which she had never planned before. The priority of some other students was to use English for accomplishing different goals. For example, Harry, David, and Victor constituted a group of online gamers − their goal was to gain access to a more prestigious online group of gamers thanks to a better command of English. Tess and Kate went through extensive periods of exposure to English when they watched online videos and TV series in response to their personal interests. Jane and Eve linked their learning English with staying abroad. Following her sister, Eve left for England and spent several months and then several years working as an au pair and learning English there. Enchanted by a school trip to Scotland, Jane repeatedly returned there for summer.
The analysis of these students' narratives suggested that either the formal system did not trigger incentives that would give rise to the decision to use PTE or the action that was taken pursued a different direction. This concerns Lara, whose dissatisfaction with the way her teachers taught had a negative influence on her attitudes towards English and learning English, which might have been detrimental for learning in the future. Though there was an option to hire a private tutor, neither Lara nor her parents made this decision. They considered her English-related problems in the context of her schooling and encouraged Lara to apply for a different school, which eventually solved the problem. The second example is Eve's story. Similarly to Lara, she was not satisfied with the turnover of English teachers and with their teaching styles. This situation, in her opinion, made her an unsuccessful learner. While Cindy's mother immediately hired a private tutor in such a situation, Eve's parents did not. Because of their level of education, their expectations concerning their daughter's school achievements might have been lower. The significant figure in Eve's life was her sister − she always inspired Eve, who followed the steps she took, including the decision to travel to England and to live there.

Discussion
The findings of the study concerning the decision to use private tutoring correspond to the framework of Jokić et al. (2013) or, more specifically, to the underlying assumptions upon which the framework is built: the decision usually depends on pupils or their parents and is closely linked to their attitudes to the education system and educational aspirations.
The study uncovered a group of students who decided not to use PTE as an educational option because of the lack of reasons or pursuing different paths. The reasons for using PTE of those who engaged in it correspond with the literature on shadow education, which distinguishes remedial or enrichment purposes of private tutoring (e.g. Zhan et al., 2013). However, securing an educational advantage (Sriprakash et al., 2016) did not appear among them. It might be linked to the fact that the maturita exam is not competitive and it may even have a negative backwash effect on a segment of the secondary school population because the exam is at a lower level than the expected outcomes of some programmes. Thus, secondary grammar school students, unlike students of other types of secondary schools, aim at the B2 level (VÚP, 2007) while the maturita exam in English is at the B1 level (CERMAT, 2014).
Private tutoring is rather associated with secondary school students (Bray, 1999;Šťastný, 2016). Two students in this study became recipients of PTE as primary school learners. This may be attributed to the specific features of learning English, because parents, who usually help their children while they are at primary school, do not feel confident enough to assist them with English. This necessitates external tuition. Cindy's mother, a beginner in English, took some lessons together with her daughter. Being aware of the role of English in today's society, Wendy's parents initiated PTE as a free time activity, which later served remedial purposes. They were not able to help their daughter since they spoke other languages but not English.
The findings of the study show how specific and important native speaker tutors are. They are typically hired because of the assets they bring to teaching and learning English (Medgyes, 1999), which match those described by George. Native speakers tend to establish a less formal tutor-learner relationship, which was apparent in George's story as well as in those of Bill and Simon. Though they did not engage in PTE, they developed a personal relationship with their former native speaker teachers. They would meet them informally to talk about their shared interests. For Irene, her native speaker teacher on a post-maturita course was an important source of the motivation for learning English that she eventually found. The students perceived the ability to communicate with native speakers as a kind of certificate of their communicative competence. Furthermore, they appreciated the opportunity to develop pragmatic aspects of communication in English (e.g. using different registers, discourse management strategies).
The outcomes of the study may be compared to the findings of those studies which also focus on individuals. Hamid et al. (2009) conclude that students, being critical of the formal educational system, considered it impossible to achieve success in English without PTE. In the Czech context, however, all the students managed to be successful − all of them were accepted to study English at a university. Some of them benefited considerably from PTE either in the short term or long term. PTE reacted flexibly to the students' needs and provided individualised instruction, which is one of the main assets of one-on-one tutoring (Bleistein & Lewis, 2015). In Cindy's case, one tutor failed in this respect, as she seemed unsure what to do. Cindy, a lower-secondary learner at that time, did not know it herself. This could be caused by unshared differing expectations on the part of both parties, which is a challenge for one-on-one tutoring (Bleistein & Lewis, 2015). The space for learner agency, i.e. for actively guiding the learning process, remained unused simply because Cindy was not ready for it. This shows how a potential benefit of one-on-one tutoring may easily become a drawback.
This instance of a tutor-learner mismatch may also be discussed in terms of tutors' qualifications. Unfortunately, details about this particular tutor are unknown, but her not knowing what to do to help the learner may be attributed to her lack of communicative competence in English, lack of knowledge about the language, about teaching and learning, etc. (cf. Shulman, 1987). Thus, incompetent action or none at all makes tutees vulnerable, especially in an emergency when tutoring for remedial purposes is needed. Bleistein and Lewis (2015, p. 23) propose that qualifications for tutoring comprise three components: formal qualifications, experience, and personal qualities. Concerning formal qualifications, they do not provide any normative suggestions; their recommendations emphasise the importance of a tutor-learner match for a particular context. This was obviously not catered to in the above-mentioned case.
The other qualitative studies that were mentioned in this article (Lee, 2010;Yung, , 2019 focus on different aspects of PTE than this study, which makes it impossible to compare the results. At the same time, they provide inspiration for future research (Yung, 2019).

Conclusion
The study sheds some light on PTE in the Czech context by uncovering the perspectives of its recipients. The findings of the study suggest that PTE supplements formal education in the course of the respondents' lives, including early childhood. Furthermore, the findings show that children's involvement in supplementary PTE is initiated either by their parents or by the children themselves for a variety of purposes that reflect their cognitive, affective, and social needs. The shadow education system is able to react flexibly to those needs when they arise. Concerning those who provide PTE, they constitute a diverse group that includes both native and non-native speakers of English; nevertheless, this study did not uncover anything about their qualifications, educational background, and/or expertise, or about the quality of their services. According to the study, the nature of tutoring tends to be considerably different if the tutor is a native speaker − apart from the perceived impact on their communicative competence, some learners, for example, built a strong relationship with the tutor, which they eventually valued more than the learning gains. In the other learners' lives, the role of supplementary PTE is linked to remedial or enrichment purposes.
The study identified two specific features of PTE. First, there is a demand for private tuition in English for primary learners for both remedial and enrichment purposes, which is not so common for other subjects (Bray, 1999;Šťastný, 2016). This is because English is part of the primary curriculum and not all parents are able to assist their children if necessary. In addition, attending extra lessons in English is a popular free-time activity. Second, there is the native speaker tutor phenomenon. Native speakers constitute a specific group of English tutors, their main contribution being their communicative competence and knowledge of the target culture. Given the current emphasis on communication, this is especially valued, together with the lower formality of the tutoring situation.
The study did not shed light on the providers of PTE, who remain hidden in the shadows. Causes may be searched for among the limitations of the secondary analysis (Johnston, 2014); specific information concerning tutors was not collected for the primary analysis since PTE was not the focus − it emerged as an interesting issue to explore. Furthermore, deploying narrative interviews as a means of data elicitation presupposes a focus on the respondents' perception of events rather than on mere facts. Moreover, the participants are likely to recall recent events more precisely than those that happened in the remote past. Therefore, designing a study with a focus on private tutors of English is necessary in order to gain insights into who they are.
Another reason for not learning much about private tutors may be related to the fact that they are not in the focus of the Czech society, including, for example, educational policy, the media, and those who purchase private tutoring. This corresponds to Bray's (1999) claim that "in almost all societies much more public attention focuses on the mainstream than on its shadow" (p. 17). Consequently, it raises a significant issue to consider, the issue of trust. While teachers in the mainstream system are often subject to systematic criticism, which often tends to slip into denigrating the representatives of the teaching profession and the system (Tůma et al., 2020), eroding trust does not concern private tutors, who are protected by the shadows. Thus, addressing this issue might be another desirable direction for future research.

Sherzod Khaydarov
Westminster International University in Tashkent Abstract: Private tutoring is a globally widespread phenomenon which can be associated with marketization and privatization. After the breakdown of the Soviet Union, the transition to a free-market economy in Uzbekistan facilitated the rise of informal private sector in public education. This in turn has affected mainstream schooling in many ways and changed the face of education. Considering socio-economic transformations in the country, the theory of hidden privatisation by Ball and Youdell (2007) is employed to analyse different facets of informal privatization in the form of shadow education. Drawing on 24 face-to-face interviews with teachers, (vice) principals and students and student questionnaires, this article examines the nature and scale of private tutoring as well as teachers' perspectives on the influence of tutoring on teaching and learning process. The findings demonstrated that teachers' overall attitudes towards private tutoring were positive and teachers mostly considered tutoring as an indispensable part of teaching and learning process. The most significant finding was that the scale of tutoring was exceptionally high (95%) in academic lyceums. The pervasiveness of the phenomenon is associated with entrance examinations, which increased the dependency of teachers and students on tutoring. The article concludes by elucidating how the emergence of shadow education in academic lyceums resulted in the change of the nature of mainstream schooling and the transformation of teacher identities.
Keywords: shadow education (private tutoring), teachers' perception, academic lyceum, hidden privatization, Uzbekistan Private tutoring (PT), as a new form of privatization of education, has grown at a fast pace over the few decades all over the world and has become a billion-dollar business in the global market (UNESCO, 2017). The trend is especially significant in some East Asian countries, where the phenomenon has been known for longer periods as well-structured establishments with specific names, such as 'juku' in Japan, 'hagwon' in South Korea, 'buxiban' in Taipei (China), and 'kewai fudao' in Mainland China Liu, 2015). Although diverse tendencies are visible globally, Bray, Kwo, and Jokić (2015) asserted that research on PT is less progressed than the actual development of the issue. The investigation of the phenomenon is largely focused in East and Southeast Asia, such as Hong Kong SAR, China, South Korea, Taiwan, Japan, Singapore, Sri Lanka, and India. Studies and investigations have also spread across African, European, North and South American countries as well as other parts of Asia (Wang & Bray, 2016), although the research is not as widespread. Silova and Bray (2006a) identified different trends of PT across former socialistic countries, grouping them according to their similarities: Central and South-Eastern Europe, the Soviet Union, and Mongolia. The market for PT among these countries developed vastly after the collapse of socialism in the 1990s, following political, economic, and social transformations (ibid). Amid post-Soviet countries, similar tendencies can be observed in the Caucasus and Central Asian countries. However, there are only a few researchers who have touched upon the phenomenon in these countries (Bagdasarova & Ivanov, 2009;Kalikova & Rakhimzhanova, 2009;Kazimzade & Lepisto, 2010;Kodirov & Amonov, 2009;Pussurmanova, 2017;Silova, 2010) and none of these studies have addressed the issue in Uzbekistan. Silova (2009b) emphasized the scarcity of data on PT in her research on Central Asian countries and despite her attempts, she could not gain access to data collection in Uzbekistan (p. 47). Bray and Kwo (2014) and Silova (2009a) provided a few references about the country, but those data are not based on actual research. Thus, there is a need to explore this phenomenon empirically in the context of Uzbekistan.

Background
PT has attained a great significance in public education especially in countries which experienced a transitional economy. Following the breakdown of the Soviet Union, the transformational recession in Uzbekistan had serious repercussions on social life. The reduction of funding in education as well as high rate of inflation was reflected in the decrease of teacher salaries and the quality of education. The financial hardships forced teachers to engage in PT to supplement their salaries and this enabled hidden privatization to enter the mainstream schools in the form of shadow education. This eventually caused the gradual expansion of PT market in the country and similar trends were observed in neighboring countries, such as Kazakhstan, Kyrgyzstan, and Tajikistan (Silova, 2009a).
Despite its short history, the spread of PT has been rapid and enrolment rates have significantly increased in recent years. Outside school PT has become a main competitor of public education, while inside school premises, the trend for teachers has shifted from a survival mechanism to an imperative part of teacher identity. For parents and children, on the other hand, PT has become the only option to meet their individual needs. Considering these changes, the study intends to explore how a fast-growing shadow education market influences mainstream schooling in Uzbekistan.
The study aims to identify the scale, types, intensity, mode, and costs of PT as preliminary data and explore the potential effects of PT on the process of teaching and learning from the viewpoint of teachers. As  asserted, there are few studies only addressing teacher identity and challenges associated with PT from teachers' perspectives. With this in mind, this research considers the dual roles of teachers and their contradicting beliefs: professional/moral values versus market values as perceived by teachers and teacher-tutors.
The study is concerned with academic lyceums (affiliated with higher education institutions) at upper-secondary education level in which the phenomenon of PT is more visible, since lyceums prepare students for the entrance exams at higher education (HE) institutions. Studies also have shown that tutoring is more widespread in urban areas (see Bray & Kwok, 2003), hence it is worthwhile to study the phenomenon in the context of lyceums as they are mostly located in big cities. Grade 12 students were selected for this study as it is more likely that students at transitional periods take more PT .
Academic lyceums were formed in 1997 as three-year specialized secondary education which were supposed to play the role of 'elite schools' to assist bright students to get into higher education. Secondary school students (after finishing Grade 9) enrol in academic lyceums through the entrance examinations held by State Testing Centre. Since the academic year of 2017/18, the period of schooling at lyceums has been reduced to two years following the shift from K12 to a former 11-year education. The number of lyceums has also been reduced more than twofold and currently 68 lyceums are in operation in the country. 1

Definition and conceptual framework
The definition of shadow education is based on Bray's (1999) interpretation which is associated with three characteristics of PT: being supplementary, fee-charging and academic. The metaphor of 'shadow education', which is widely acknowledged among researchers, refers to the content in the 'shadow' that mimics the curriculum of conventional teaching at schools (Bray, 1999).
Investigating the phenomenon of shadow education is rather complex due to its delivery on an informal basis (Bray, 2006). The trend of the shadow market in newly emerged countries, such as Uzbekistan, can be quite similar to the patterns observed worldwide because of the influence of globalization and free marketization. However, it is likely that the perceptions of stakeholders might vary from the existing literature of global discourses owing to cultural, political, economic and social settings.
Bearing in mind these factors, the study employs the theory of 'hidden privatization in education' by Ball and Youdell (2007). In their study, the authors classified privatization into two types: endogenous and exogenous (See Figure 1). In the endogenous type, new concepts, techniques, and practices are introduced into public education from the private sector and the public sector performs "more like business or business-like" (p. 16). Whereas, in the exogenous type, the private sector itself participates in the provision of education along with designing and managing programs. The authors categorized PT as an exogenous type of privatization and defined as a mechanism which complements the public sector by providing education for private profit. The authors paid very little attention to PT and did not discuss the 1 See https://edu.uz/uz/pages/secondary-special-and-vocational-education.  The education of all children is held to be intrinsically of equal worth The education of children is valued in relation to costs and outcomes type of tutoring that is provided by teachers. Nevertheless, their concepts can be similarly applied to evaluate the impact of tutoring on teachers' teaching activity in the context of lyceums. In particular, the concept of changing teacher professional values into market values (Table 1) can be utilized as a framework to assess teachers' attitudes regarding their teaching and tutoring roles.

PT in the context of Uzbekistan: History and regulations
The extent of PT is largely dependent on the principles and regulations of the education system. The absence of legal guidelines and lax controls may hinder education at mainstream schools as this enables teachers to engage with tutoring themselves.
Concerning the history of PT in Uzbekistan, initial forms of tutoring existed even during the Soviet Union in a hidden form among few families, but it was muted as the private sector was strictly banned by officials (Silova & Bray, 2006b). However, the market for PT became noticeable when the current centralized test-based high-stakes entrance examination was introduced in 1994 2 to replace pre-existed university-based admissions. The majority of students were unfamiliar with testbased examinations and this shift generated the demand for PT as students assumed tutoring to be efficient for coping with transformations in the education system (Silova, 2009c).
Similarly, after the independence, the transition from socialism to a free market economy facilitated the emergence of the private sector in education as well. However, establishing private schools were banned in 1993 amid the fear of arising extremist groups (Hays, 2008). To sustain the needs of students, test-oriented preparatory courses were established under each university which were coached by university teachers (Silova, 2009b). The fees were considerably low, usually contracted for the whole year for certain examination subjects relevant to each university. Tutoring centres, by contrast, were very rare. PT was therefore done mainly by teacher-tutors.
Currently, there are no formal regulations concerning teachers' engagement in PT in Uzbekistan and this jurisdiction characterizes laissez faire practice in which "[t]he school and education authorities do not have policies on the matter, leaving decisions to the teachers themselves and to their clients" (Bray & Kwo, 2014, p. 45). PT is considered as a form of entrepreneurship and officially teachers are neither restricted nor imposed a tax for offering tutoring services. Thus, most teachers provide tutoring on an informal basis. According to the previous Education Law (1992), teachers were given the right of performing "an individual pedagogical activity" (Article 32), but the current Educational Law does not include any reference regarding private tutoring. However, following the Presidential Decree No. 1875 on December 12, 2012, the Department of Foreign Languages Proficiency Assessment was established as part of the State Testing Centre and it started to license both teachers and tutoring centres with qualification of tutoring foreign languages. It was officially announced that teachers of foreign languages are required to obtain language proficiency certificates to offer tutoring services, however this regulation is rarely observed in practice. Therefore, the engagement of teachers in PT remains hidden and unnoticed.

Studies on teachers' perspectives
PT has been examined by different authors for varied purposes. A considerable amount of studies on PT addressed the nature, demand, drivers, impacts, and regulations of the phenomenon. Through the literature review, a few studies were found on teachers' perspective on shadow education Kobakhidze, 2014Kobakhidze, , 2018Popa & Acedo, 2006;Sun, 2017;Wang and Bray, 2016) which illustrate insights of the phenomenon in three contexts only, namely in Romania, Georgia and Hong Kong. Popa and Acedo (2006) investigated teacher professionalism in relationship with tutoring in Romania and their interviews with high school teachers illuminated how teachers' engagement in tutoring helped regenerate professional identity and reestablish teacher authority. By contrast, Kobakhidze (2014) studied corruption risks associated with tutoring, dual roles of teachers, challenges and moral dilemmas related to tutoring their own students in the context of post-Soviet Georgia.
Unlike the studies mentioned above, Wang and Bray (2016) examined the viewpoints of teachers regarding the role of supplementary PT as received by their existing students in Hong Kong. Their study revealed significant findings which demonstrated opposing attitudes of teachers towards PT and their feelings on social stratification against holistic education in the context of examination-driven culture. In a similar context, but at the primary school level, Sun (2017) studied Chinese and English teachers' attitudes towards PT taken by their students. Although the study found some interesting aspects of teacher attitudes on the implications of PT, the reasons behind conflicted attitudes were not questioned fully. Kobakhidze's (2018) study, on the other hand, explored the phenomenon from the teacher-tutors' perspective focusing on teachers' professional beliefs and values regarding morality and corrupted practices. She found sociological concepts more suitable to study the behaviors of teacher-tutors due to the informal nature of PT in Georgia. One of her major contributions was to explore the shadow market from the perspective of economic sociology, which helped to draw a comprehensive picture of the phenomenon with contextual nuances in post-Soviet Georgia.

Research method
The study employed a mixed method to examine the nature of PT and teachers' perceptions of the phenomenon. For data collection, the priority-sequence model by Morgan (1998, as cited in Teddlie & Tashakkori, 2003) was used as a framework. Morgan introduced two groups of decision rules in his model to merge quantitative and qualitative methods for information gathering during the fieldwork. One is deciding the priority of the methods and the other is deciding on the sequence of the methods. For studying the implications of PT, the researcher adopted the framework which comprises 2 phases.
For the quantitative part, a two-page questionnaire was employed to elicit preliminary data about the enrolment rates in PT and identify the scale of teachers who are tutoring their students. The scarcity of the information on PT in Uzbekistan necessitated the inclusion of quantitative data to illustrate contextual features of the phenomenon and corroborate the findings of the qualitative part.
A substantial part of the data concerning PT experiences was gathered through semi-structured individual interviews in the Uzbek language. Separate interview protocols were prepared for each target group with organizational and preliminary notes. Teachers were a priority group; therefore, teachers' interview protocol had an extensive list of open-ended questions that are comprised of nine thematic topics. By contrast, interview questions for students and principals were limited to five topics. Interview questions were mainly adapted from  and some sections that are related to teacher-tutors were borrowed from Kobakhidze (2018).
The author mostly followed the interview guide approach (Patton, 1980, as cited in Cohen, Manion, & Morrison, 2007, in which the topics and research concerns were identified in advance and questions were asked flexibly during the interview. This approach maintained the richness of the data as interviews were relatively conversational and it gave the researcher more flexibility whether to add or omit questions due to the experience of interviewees (ibid).
This study is chiefly based on teacher interviews. Yet, the data were triangulated with interviews of principals, vice-principals and sampled students. For research validity, the author grounded the study on Onwuegbuzie and Johnson's (2006) inside-outside legitimation (p. 58). The author received assistance from peers and an MEd supervisor from the University of Hong Kong, which helped to observe the phenomenon from an outsider perspective, while being an insider at the research site helped the researcher understand the phenomena from a local perspective. The convergence of these etic and emic viewpoints corroborated research outcomes.

Selection of research site
The target population of the study was students and teachers of selected academic lyceums in Samarkand city. Samarkand, the author's hometown, was chosen as a research site for two reasons: first, the setting was convenient in terms of time and financial aspects; secondly, the viability of gaining access was an important factor.
According to the regional educational board, 14 academic lyceums were operating in Samarkand region as of 2017. 3 The total population of lyceums amounted to 10,402 students (3,340 of them were Grade 12 students) and 935 teachers in the 2017−2018 academic year. Bearing in mind the availability of limited time, the study is restricted to two lyceums, which can represent general features in the region. The lyceums were selected based on convenience sampling: the first lyceum was the researcher's former workplace, while the principal of the second lyceum was his former colleague.
Academic lyceums had been administered by regional Educational Boards by the year of 2017. However, following the President's Decree No. 2829 in 2017, the task of supervising academic lyceums was transmitted to the respective HE institutions they had been bound to. As for selected lyceums, both lyceums have a few distinct features, which are presented in Table 2.

Sampling and recruitment
The research utilized a multilevel sequential design for sampling based on Onwuegbuzie and Collins' (2007) mixed sampling framework. This framework is suitable for selecting two or more sets of samples for quantitative and qualitative phases. In the current research, the quantitative part involved the sampling of students, while the qualitative part utilized the sampling of teachers, principals, vice-principals as well as students.
For the questionnaire, Grade 12 classes were randomly selected from each lyceum and passive consent forms were distributed a week before the survey. Then, based on the results, 50 to 60 students from each lyceum were selected based on cluster sampling. In total, 109 students were involved in the survey.
For interviews, teachers were drawn from the same lyceums through purposeful (to get broad range of teachers) and snowball sampling. Considering the characteristics of teachers, such as gender, experience, age and teaching subjects, eight teachers were selected from each lyceum for individual interviews. For the cross-verification of teacher interviews, four students from each lyceum were selected from the questionnaire sample. Finally, a principal and vice-principal of each lyceum were interviewed to reinforce the data. In total, 28 individuals were interviewed during the survey.

The scale and nature of PT
The findings of the questionnaire showed that 95% of lyceum students were receiving or had received PT over the last 12 months prior to the fieldwork. There was very little difference in the participation rates of PT between the two sampled lyceums, indicating that the phenomenon could be significant in other lyceums too. Although lyceums are often regarded as institutions with high academic excellence, nearly all students had to regularly attend tutoring classes after the lessons and this means that "PT is actually becoming a substitute for regular schooling rather than just a shadow" (Bray and Kobakhidze, 2014, p.15).
As for the delivery of PT, the questionnaire findings indicated that most students received tutoring in large (65%) and/or small (63%) groups, while one-to-one tutoring (only 17% of respondents) was much less common. Popular PT subjects among students consisted of Uzbek language, English, Mathematics, and History, whereas science subjects such as Physics, Chemistry, Biology were less popular. This is largely associated with the demand of students and "[t]he subjects in greatest demand are ones required by examination systems" (Bray, 2006, p. 520). Therefore, as Bray (2006) stated, tutoring in subjects such as national languages and Mathematics were more common.
Regarding the dominant type of PT, all the teachers, except one, specified lyceum teachers as primary providers of PT for the students of academic lyceums. The findings of the questionnaire also indicated that students predominantly received tutoring from lyceum teachers, however, the percentage of tutoring centres was a bit higher in AL2. Although students did not specify the name of tutoring centres they were going to, some of the teachers remarked that students were mostly attending the tutoring centre that was newly established within the premises of the lyceum. The administration also approved that the students were receiving tutoring either from teachers or the tutoring centre of the lyceum.
The share of other tutoring centres or independent tutors was also significant. Half of the interviewees remarked on the growing influence of tutoring centres, due to their advertisements and promotional deals. One of the teachers asserted that tutoring centres were trying to entice students in many ways in response to stiff competition with other providers of tutoring. Likewise, another one stressed that tutoring centres were becoming their main competitors in terms of tutoring, but claimed that most students still chose lyceum teachers.
One of the interesting findings was that nearly all students (97%) received PT to prepare for national entrance examinations. Students mostly received tutoring in two or three subjects as examinations are based on a combination of three different subjects. This suggests that students took PT primarily for an 'enrichment' strategy (Bray 2006;Bray & Kobakhidze, 2014;Silova, 2009b) to manage with high-stakes examinations. Additionally, teachers in the study asserted that students did not participate in PT for 'remedial' or 'reinforcement' purposes (Baker & Le Tendre, 2005;Bray, 2006;Silova, 2009b) which was in line with the students' responses in the questionnaire and interviews. These findings suggest that students are less concerned about their lyceum lessons but concentrated more on PT to pass entrance examinations.

Teacher attitudes
When teachers were asked about their feelings regarding students receiving PT in the subjects that they teach, almost all teachers felt 'positive', 'happy' or 'glad', while the rest did not mind their students seeking PT. The teachers approved their students' use of PT either because they thought PT was more helpful in supporting low-achieving students or because they regarded PT as a necessity for students' academic achievement. No one mentioned negative aspects of PT and teachers were likely to be supportive of PT as they were providing tutoring too.
Many teachers felt sympathetic towards students due to the increasing competition and complexity of tests. Even as parents, the teachers indicated that they would encourage their children to receive PT. Although it was a slight exaggeration, a Physics teacher claimed that students were capable of solving not more than 10% of entrance examination tests without PT. Similarly, a History teacher, responded to the question as follows: In the current situation, I regard PT as positive and I think it is highly necessary. The knowledge given in lyceum is not enough and I don't believe the students who rely on regular schooling can pass entrance examinations. An Uzbek language teacher was even curious about how many students were taking PT from her subject: Whenever I am for the first time in a class I ask how many students are receiving PT and try to find out the reasons if any students are not taking tutoring. If most students are receiving, I'll become happy, because PT makes them busy with studying and they won't forget the subject.
Likewise, a Geography teacher, who held relatively negative opinions of PT, expressed that he would be delighted to see students receiving PT from his subject, not because he thought PT was helpful for his teaching, but because students would show an interest in studying his subject rather than regarding it as a secondary subject.
Only one teacher reported that he encouraged students to study independently and believed that students could manage without PT if they kept studying regularly over the period. However, he asserted that not all students would study diligently by themselves, therefore he thought PT to be helpful for those who could not cope with regular lessons.

Teachers as tutors
Although most teachers maintained a positive stance on PT throughout the interview, they had difficulty articulating their feelings about their tutoring roles. Some teachers considered providing PT a 'good thing', a 'normal situation', an 'easier' task or even an 'inspiring job' and they were pleased to hold the role of tutors. To some extent, they had enough reasons to claim so, as they pointed out financial advantages, professional development and enhancing their reputation as factors of a stimulus.
However, by no means were all teachers satisfied with their double roles, as some of them considered holding two roles time-consuming, exhausting and stressful. For instance, a History teacher specified her thoughts on tutoring as follows: I think the motivation for doing PT derives from financial needs, in the first place. I hear most teachers complaining about their wages. They say 'If we had enough salary, we would not spend so a great deal of time and channel our efforts into PT, instead we would work on ourselves and get prepared for lessons'. And I wish the same too… She was not only the one who mentioned the salary as the primary driving force of doing tutoring. A few other teachers and even principals acknowledged that teachers needed to provide tutoring to augment their meager income. However, in return, they had to sacrifice their strength and free time (and even partially, their working time) for the sake of PT.
Half of the teachers expressed some regret over the time spent providing PT. A History teacher reported that he sometimes felt embarrassed when he could not make time for teachers' and parents' meetings or other kinds of organizational work, whereas an Uzbek language teacher, who seemed to 'obtain energy from tutoring', instead worried more about her family time as she explained: "The only thing I regret about PT is my family time. I could have spent that time with my children. They need more care." By contrast, an English language teacher shared her feelings about the complexities of performing mixed roles: Sometimes, as a teacher, I feel bad for it because there were times I could not prepare for lyceum lessons. … I try to keep my teaching and tutoring roles in balance. If you disregard your main duty, it certainly affects your reputation, and eventually, it influences your tutoring too, … Sometimes, it depends on my mood and the environment in the classroom. There were times I paid more attention to PT. But in general, I put lyceum lessons in the first place. Kobakhidze (2018) elucidated some subtle nuances of teacher and tutor roles and identities in the post-Soviet context of Georgia. The teachers in her study often expressed confusion when it came to defining their identities as they had formed a new identity with 'blurred boundaries between teaching and tutoring' (p. 114).
In the context of Uzbekistan, signs of the same tendency were visible, but teachers failed to discern negative aspects of their acquired new role: If to talk about PT, of course, it may have many negative aspects, but in the current situation, as teachers, we don't realize them, we don't see them. We only see positive sides, because we mostly look at the results in students' performance. (Geography teacher) The teachers reflected in the same way when they were asked about tutoring their own students. Among 16 teacher-interviewees, only 2 of them were not engaged in tutoring and all of the remaining teachers acknowledged that they tutored their students. Half of the teachers considered tutoring their students as normal or positive and asserted that it did not have any negative impact. Others did not know how to respond and avoided commenting. Only one teacher considered such a case as 'not a good thing' and explained that "it causes irresponsibility and inattentiveness in students who are receiving tutoring from their teachers, and their attention will be on PT as they know that the teacher will explain the topic broadly during PT".
These responses show that there is a lack of awareness on teacher and tutor identities and most teachers viewed tutoring as part of their main job. However, the main challenge was how to keep the balance between the two roles as the majority of teachers were "juggling between tutoring and teaching" (Kobakhidze, 2018, p. 157) but not caring much about the consequences.

Influence of PT on teaching
With regards to the influence of PT, half of the teachers responded that tutoring made their teaching easier, four teachers stated that it caused difficulties, two teachers remarked there was no influence, while the other two teachers had mixed feelings.
Majority of teachers regarded PT as a means of developing their professional competences, i.e. fostering knowledge, increasing pedagogical experience, and enhancing test solving skills. Therefore, they had professional satisfaction from PT. This was clearly elaborated by an Uzbek language teacher: PT makes my job easier because I work a lot on myself. I prepare only lesson plans for lyceum lessons. When I see the topic, I know everything, what to teach and how to approach, so there is no need for the revision.
Seemingly, it was a common perception among teachers to regard PT as a part of the preparation for regular lessons. Similarly, PT was viewed as a way of preparing for the tests of State Inspection for the Quality in Education under the Cabinet of Ministers. Principal interviews also confirmed that teachers with PT were more successful in attestation screenings.
Furthermore, some teachers noted that PT enhanced the learning environment in the classroom. For instance, a Mathematics teacher asserted that PT improved students' understanding of the subject and helped his lessons go smoother. Other teachers observed that students became more active and demanding after receiving tutoring, and therefore accepted that PT was beneficial to their teaching.
However, the interviews also illuminated multiple problematic cases related to PT. The respondents provided some undesirable examples from their own experiences that provoked the discussion of the issue. To illustrate, an Astronomy teacher opposed the trend of studying only two or three subjects to gain admission into HE institutions. He criticized students focusing only on specialized subjects, which fostered the notion in students that other subjects were as if "secondary subjects". He also criticized students trying to do their tutoring assignments in a hidden way during the class.
Another experienced Mathematics teacher shared her experience of tensions that arose with her students when tutoring was provided by other tutors: When I started teaching Mathematics, I realized that most of the students didn't understand the concept of the subject. They were taught to find the answer with a formula book. They were able to solve problems with shortcut tips but could not explain the process. I struggled a lot and I had to teach the basics first. … Some students didn't like my approach and we became in conflict. If I had taught myself, it'd have been easier… It seems not only experienced teachers but also young teachers could not tolerate when students prioritized their tutors' teaching practices. An English language teacher, who was ambivalent about PT, had quite similar experiences with her students. She noticed her students evaluated her lessons and compared her teaching with their tutors'. They then became negligent or inattentive when they mastered the topics in PT ahead of time. She also faced difficulties with planning lessons and materials that matched the different levels of students and thought that PT expanded the gap between students' knowledge even more.
Similarly, an English language teacher expressed mixed feelings towards PT. She considered PT to be positive in terms of strengthening students' knowledge but admitted negative impacts of tutoring on regular classes too in terms of additional tasks: "As a teacher and a tutor I am aware if I give numerous tasks to students during PT, they'll do these tasks during lyceum lessons. I feel a bit guilty for that." What is interesting in this data is that teachers provided more negative aspects regarding the effects of PT on teachers' time and effort. A majority of teachers conceded that they had to spend a good deal of time and energy for PT which in turn affected their preparation for lyceum lessons. Although teachers avoided evaluating their own experience, a few of them admitted that they had heard of other teachers who prioritized PT and saved their energy during regular lessons.
Students' interviews indicated that students were mostly content with their core subjects. However, one student reported that teacher-tutors did not have a full dedication to teaching during regular lessons as in PT, while teachers of non-specialized subjects were not concerned much about teaching. He explained that teachers of 'secondary subjects' did not bother working with students the whole lesson, instead teachers allowed them to do their tasks or tutoring homework after explaining a new topic which can be indications of teacher malpractices such as energy-saving and professional negligence in regular classes.
In short, part of the teachers sometimes felt 'demoralized' or 'stressed out' when they encountered conflictual situations with students and thought they were teaching 'in vain' when PT 'complicated their teaching practices' or could not find a solution for ineffective teaching practices. However, the rest of the teachers did not suppose PT to be an obstacle for their teaching, instead, they regarded tutoring as a mechanism to consolidate the knowledge taught at lyceum and strengthen their professional skills.

Impact of PT on students' learning
All teachers believed that PT was efficient to increase students' knowledge and observed as most of their students improved their academic grades. The respondents enumerated several factors why students learned better in PT. First of all, teachers noted that students had stronger motivations for learning as they had certain ambitions to receive PT, such as entering universities, going abroad or attaining certain qualifications. Moreover, they believed that students were more disciplined and had a higher responsibility owing to their parents' spending. Lastly, PT was deemed to have a more flexible learning environment to meet each student's individual needs. This factor was reiterated in students' interviews too, in which students stated that more individual time with the teacher/tutor compared to classroom environment better suited their learning and understanding, therefore they expressed satisfaction from their tutoring classes.
However, not all teachers agreed that PT was efficient for learners. One teacher, for instance, stated that she had students with different learning abilities and motivations, therefore she thought it was rather challenging to achieve each student's progress. Some teachers reported that they would quit tutoring with students who lacked interest in a subject or were not capable of mastering given tasks because teachers did not want parents to waste their money when tutoring did not fit with their children's capacities.
A few other teachers thought that tutoring was ineffective when parents pushed students to take extra classes. For example, one teacher disapproved of parents who forced their children to go for tutoring and blamed tutors when the results fell short of their expectations. An English language teacher, on the other hand, argued that tutoring was not always as efficient as she expected. She stated that low-achieving students mostly performed in the same way as in regular lessons and PT had no significant effect on their academic grades. She also noted that "they usually go for tutoring, for their parents want them to do so, or they follow their friends who are performing well in lyceum lessons". This indicates that parental factors and peer influence also had an important role in students' choice of PT.
Moreover, most teachers indicated that PT negatively affected students' grades in non-core subjects, since their attention was directed at studying two or three specific subjects only. A good reference was made by an Astronomy teacher, who noted that "some may excel at Mathematics, but they lag way behind in other subjects". Interviews with students also corroborated this, as half of the students experienced a drop in their grades from non-core subjects. One student's comment on the quality of regular lessons, for instance, was quite surprising: "Students are concerned about three subjects from their specialization, as for other subjects it is not even interesting for them whether they are satisfactory or not." Similarly, another student also asserted that students "partly study non-core subjects only for grades, not for acquiring knowledge". Because 'secondary-level' subjects were weighted less than core subjects, students did not bother to study them which eventually influenced their grades too.

Discussion
The main objective of this study was to determine the teachers' overall attitudes towards PT. The findings showed that all teachers, including principals, supported PT and thought that it was useful and necessary for students' academic achievements. Even the teachers who were not serving as tutors were pragmatic and encouraged their students to take additional classes. Interviews indicated that the primary concern of teachers was the progression of students to HE institutions. These findings are consistent with those of Altinyelken (2013), Wang and Bray (2016), and Sun (2017), in which teachers were in support of students' taking tutoring because of stiff competition in examinations.
In contrast to Turkey or Hong Kong, teachers in Uzbekistan are engaged in tutoring themselves and their dual roles inevitably affect their attitudes too. According to Ball and Youdell (2007), when the private sector is brought into public schools it forms "a new moral environment for both consumers and producers" (p. 52). This means that both teachers and students will face a "culture of self-interest" or "survivalism" (ibid.) and particularly teachers are likely to experience transformations in their identities as teachers' professional values clash with market values. This notion was echoed in the interviews which revealed that teachers' attitudes were different towards students with tutoring and without. To be precise, they perceived students taking PT as more 'active' and 'diligent' compared to students who did not. This suggests that teachers may not be treating all students equally regardless of their mixed ability (commonality value).
Similarly, a majority of teachers considered tutoring their students as a normal or positive procedure. In previous studies, this type of tutoring was referred to as educational corruption (see Brehm & Silova, 2014;Dawson, 2009). However, it is difficult to assess teachers' tutoring their students to be completely unethical, because many teachers noted that they refrained from treating students better for receiving tutoring and called such acts as 'unfair' or 'wrong'. Similar patterns were observed in Kobakhidze's (2014) study in Georgia, where teachers' tutoring their own students became a typical practice like in Uzbekistan, and she argued that such practice is "not necessarily a form of corruption" as there were many teachers who were fully dedicated to their profession (p. 455).
Additionally, a sympathetic attitude of teachers towards students from disadvantaged families can be construed as an example of moral values. Nearly all teachers stated that they tutored students from low-income families free of charge or with discounts. Some teachers even noted that they did tutoring not only for making a profit but also 'out of the goodness of their heart'. These findings are consistent with the findings reported by Kobakhidze (2014) in which she categorized such teachers as "Robin Hood teachers" (p. 465).
By contrast, the teachers of non-core subjects and those who were not tutoring presented contrasting viewpoints towards PT. Although they considered that PT is necessary for the current situation due to the complexity of examinations, they were more concerned about its negative effects on students' attitudes toward non-specialized subjects as they became more disruptive and negligent during their classes.
The scale of tutoring was another important aspect of the phenomenon: overall 95% of the sampled students responded that they received PT over the last 12 months. This includes students who were currently not taking any tutoring but had attended during the 12 months prior to the survey. The intensity and frequency of tutoring might vary across different groups, yet on the whole the attendance was remarkably high in both lyceums. Although the data are incomparable due to the size of the sample and differences in target population, they provide some insights about the pervasiveness of tutoring in the context of Uzbekistan. Similar trends were observed in the findings of Silova and Kazimzade's (2006) survey in Azerbaijan which indicated that 90% of first-year university students were found to have received private tutoring (including preparatory courses) before entering the university. Likewise, the figures in neighboring countries, such as Kazakhstan (64.8%), Tajikistan (60%) and Kyrgyzstan (52.5%), were relatively high (Silova, 2009d).

Infl uence of PT on teaching and learning
The fi ndings highlighted that teachers mostly considered PT to be useful for teaching and learning even though they disclosed multiple issues related to tutoring which have negative effects on the quality of education. Nevertheless, it is important to bear in mind the possible biases in the responses of teachers as their viewpoints are contingent upon their perceived 'new identity'. As discussed above, teachers' dual roles, as both teachers and tutors represent two different (professional and market) values, and the mixture of these affected their way of thinking and beliefs. For instance, half of the teachers believed that PT made their teaching easier because they thought the tutoring role enabled them to expand their knowledge, develop their test-solving skills and strengthen their pedagogical experience. Some teachers even identifi ed PT not as an obstacle, but as part of the preparation for regular lessons. Again, this manifests teacher-tutors' acquired 'culture of self-interest' − an "orientation toward the internal well-being of … its members and a shift away from concern with more general social and educational issues within [community]" (Ball & Youdell, 2007, p. 52

Figure 2
Infl uence of PT on teaching and learning Bray (2011) remarked that the practice of teachers' involvement in PT, particularly teachers' tutoring their students, might cause the risks of decreased motivation and effort in teaching at mainstream schools, favoritism, and other unethical practices. Although the teachers ignored the issues of favoritism and corruption in their practices, a few teachers did not deny the existence of such malpractices in their examples of other teachers. In the same way, teachers refused to prioritize PT, but they admitted to preparing additional materials for tutoring, such as test banks, audio-visual aids, and manuals.
Likewise, Loveluck (2012) asserted that teachers' engagement in PT deters teachers to fulfil their lesson plans. This was consistent with the findings of current research which showed that teachers spent a great deal of time and effort towards PT and could not prepare well for lyceum lessons. Some teachers were reported to save energy during regular lessons while some others intentionally decreased teaching hours to have more time for tutoring. This suggests that tutoring had a significant impact on teachers' teaching practices, even though they were reluctant to admit the negative influences.
More issues were visible in non-core subjects. Due to the complex nature of entrance examinations, students' dependence on PT increased. This might have enhanced students' learning from specialized subjects but it caused a loss of interest in learning other subjects. In the same way, PT adversely affected the quality of teaching in non-core subjects. Student interviews unveiled that teachers of these subjects usually explained the new topic in the first half of the lesson and then allowed students to do their tutoring assignments for the remaining time. This can be explained as an 'erosion' of teachers' professional values since they became less considerate about the repercussions of PT (Figure 2).

Emergence of formal shadow education within mainstream schooling
One of the significant findings to emerge from this study was the establishment of a tutoring centre inside the academic lyceum (as was observed in AL2). The distinctive feature of this tutoring centre was that it operated within the premises of the lyceum and under the management of the principal, and the tutoring was delivered mainly by lyceum teachers. This can be regarded as the confluence of mainstream education with shadow education or the emergence of 'formal' shadow education within mainstream schooling. The study refrains to discuss the legal bases of establishing a tutoring centre within the public sector because it was beyond the research purpose. However, the teachers' involvement in the running of the tutoring centre might affect their commitment to the teaching profession. For instance, some teachers stopped being employed as teachers in the lyceum and were fully employed as tutors in the tutoring centre, while some others became part-time teachers to increase their tutoring hours.
On the one hand, the establishment of the tutoring centre provided the students with official tutoring services within the academic lyceum. On the other hand, it enabled the teachers to engage with tutoring in the first half of the day when they did not have lessons. This may lead to the prioritization of tutoring and decreased devotion towards the teaching profession. The integration of private sector in public education, in this sense, can be a good example of Ball and Youdell's (2007) privatization concepts on "transformation of education from a public good to private commodity" (p. 53) as well as transformation of identities: Principal to manager, and a teacher to a technician (p. 45).

Dependence on PT
The teachers' overall attitudes towards PT indicates teachers' increasing reliance on PT rather than their teaching capacities or professionalism. This was evident in teachers' supporting students to take extra classes from their subjects. Even, teachers as parents encouraged their children to receive PT. However, it is worth noting that students' dependency on PT was more psychological. Among students, it became a common belief that without PT they cannot succeed in entrance examinations. Therefore, nearly all students received PT irrespective of their level of knowledge. Kwo and Bray (2011) also mentioned teachers' partial reliance on tutoring to consolidate some parts of the curriculum in Hong Kong. In Cambodia, on the other hand, public schools were found to cover only the part of the national curriculum during school time, which allowed teachers to supplement the remaining content during tutoring classes for a fee (Brehm & Silova, 2014), which illustrates an explicit form of teacher corruption. In Uzbekistan, however, teacher tutors reported that they fully covered the state curriculum and even utilized their tutoring materials in classroom lessons as additional sources. Therefore, PT was viewed as a tool to supplement the lyceum program with the syllabus of secondary schools and vocational colleges. It seems the dependence on PT in academic lyceums derived from the complexity of examination tests and divergence of the lyceum curriculum from entrance examinations. Student interviews also revealed their satisfaction from lyceum lessons but they thought that the curriculum was not adequate for high-stakes examinations.

The shift of the role from mainstream to shadow: Examination-oriented education
In the current context of academic lyceums, PT has taken a form of alternative education to substitute mainstream education in preparing students for high-stakes examinations. There is a general assumption among parents that the quality of education at lyceums is substantially higher as compared to secondary schools or vocational colleges, and lyceum teachers are deemed the most qualified of all. Notwithstanding, many teachers regard lyceum lessons as not very efficient in preparing students for entrance examinations, owing to limited teaching hours, an excessive number of students, and a diversity of students' levels. Given the shortcomings of academic lyceums in meeting individual needs for learning, PT has taken the role of preparing students for entrance examinations. Therefore, for students, the mastery of the required knowledge has become viable only through PT.
During a 2013 TED Talk speech titled How to Escape Education's Death Valley, renowned educationalist Sir Ken Robinson stated, "the dominant culture of education has come to focus on not teaching and learning, but testing". Indeed, the pervasiveness of testing culture has been visible in education systems around the world, particularly in East Asian countries, such as China, South Korea, and Japan. In these countries, tutoring centres are considered a major competitor of mainstream schooling by promoting examination-oriented education . In the context of Uzbekistan, similar trends have become visible after the introduction of a testing system for university admissions. However, unlike the aforementioned countries, in Uzbekistan mainly teachers took the role of providers of tutoring. This enabled hidden privatization to creep into mainstream schooling in the form of shadow education and "contribute[d] to the privatization of paths to educational attainment" (Dawson, 2010, p. 22). These statements verify that the landscape of public education had shifted under the influence of the market economy and neoliberal ideologies which focus mainly on examination results or performance outcomes rather than the broader goals of schooling. Therefore, a fundamental dilemma remains: whether or not students should keep receiving PT in line with mainstream schooling in academic lyceums, the place whose mission is to give deeper knowledge in specialized fields and prepare students for the next pillar of education.

Conclusion
PT is widespread globally and it has a growing nature which can be associated with marketization and privatization. This study outlined how the blurring of the private sector (shadow education) with the public sector affected the teaching and learning process and the attitude of educators. The findings from this study suggest that the emergence of shadow education in academic lyceums changed the nature of public schooling as well as transforming the identities of main stakeholders such as teachers, principals, and students. The prevalence of PT was strongly linked to the complexity of current entrance examination tests which increased the dependency of both teachers and students on PT. Despite multiple negative aspects of PT, teachers remained positive in their opinions and thought PT to be essential both for learning and teaching. For them, PT served as a survival mechanism, which brought certain benefits along with exacerbating existing problems.
Finally, number of important limitations need to be acknowledged. First, the current study has only examined the perspectives of lyceum teachers in Samarkand region, and with a small sample size, caution must be applied, as the findings might not be transferable to the country level. The data of the quantitative component is not representative of the whole upper secondary level in the region, because the focus of the study is restricted to Grade 12 students in two academic lyceums (vocational colleges are excluded).
The study has also limitations associated with timing. The data was collected in 2017, in a period of political transfer, when radical reforms were introduced in public education system by the current President Shavkat Mirziyoyev. The shift to former 11-year education system from K12 education was completed in 2019. The quantitative data was grounded on the previous system (Grade 12 students), nevertheless the challenge of PT is still applicable to the current system due to the growing scale of the phenomenon. In addition, communication with lyceum principals and tutoring teachers were also filled with issues in guaranteeing an accurate description of the issue. As topics of private tutoring, education quality and teacher salary were politically sensitive, in some occasions administrators and participants might have been unwilling to provide a fully accurate picture or did not want to disclose information during the interview. Therefore, the researcher added 2 vice-principals and 2 teachers who did not engage in PT to the initial sample size to maintain the balance and objectivity in data aggregation and compare the variations.
In spite of these limitations, this research serves as the first attempt to empirically explore the nature and implications of private tutoring in Uzbekistan. It is a baseline study. Thus, the findings of the study may serve as valuable information for stake-holders and policy-makers in the country to evaluate the implications of private tutoring in a rational way when making education policy decisions. Almost a decade ago, Bray, accredited with drawing academic attention to the field of shadow education, analogised private tutoring research to assembly of a jigsaw puzzle with most of the pieces missing because of its nascent stage of research. He and his team of researchers have continued their pursuit to fill in missing pieces of the puzzle, resulting in a wide body of literature that has conceptualised and unravelled the private tutoring phenomenon in several regions of the world. In the book under review, Bray, along with his co-authors Kobakhidze and Kwo, shines the light on Myanmar, providing a very useful and rich account of private tutoring there. Drawing on their expertise and knowledge of private tutoring in other countries, the authors have collected rich data with the support of local teams, and have reported contextual findings to serve their purpose of identifying policy implications.
The book begins with a briefing on international and comparative private tutoring research. It then lays out the context of the political, economic and educational transformations in Myanmar, before delving into its private tutoring specificities. The first chapter on the scale and nature of tutoring around the world orients the reader to reflect on similarities while taking into account differences. It also helps in unpacking the heterogeneity of the phenomenon by pointing out the different formats of tutoring, the types of tutors, and the positive and negative outcomes for families, schools and the societies at large. The second chapter apprises the reader of the geographical and administrative features of contemporary Myanmar and portrays how recent political and economic reforms have influenced education for the masses. It also provides a layout of the schooling system along with statistics on literacy and enrolment rates. The system of assessments and examinations, which is closely associated with tutoring, is described in detail. The authors then specify the regulatory legal frameworks operating in the domain.
The methodology chapter builds upon Bray, Kwo, and Jokić's (2015) Researching Private Supplementary Tutoring: Methodological Lessons from Diverse Cultures, which is the first of its kind to offer detailed insights into the methodological aspects of shadow education research. One major challenge in the field is collection of data that is often considered sensitive because of perceived threats to the "business" of tutoring, complicated further by the power-plays between the multiple actors from both the formal and informal spheres of education. The research design of the study in Myanmar elaborated by the authors shows how they navigated the methodological and linguistic challenges by embarking on collaborative efforts across multiple organisations and further enlisting students and researchers from their own institution in Hong Kong. Another difficulty highlighted in the book is pertaining the restriction of research sites to urban and peri-urban regions of Yangon, pointing to the need for national-level education surveys to include details of tutoring and overcome limitations of time and budget often faced by research teams.
The choice of mixed methods presents broad contours of the features of tutoring while at the same time, includes actors from government, NGOs and professional groups of teachers and tutors in addition to household and school level stakeholders. Repeat interviews and workshops on preliminary findings to validate the findings, deepen the analyses and provide directions for further interviews is another valuable technique employed by the researchers. This participatory approach enabling immediate dissemination of findings and interchange of knowledge elevates the research process by providing opportunities to participants and stakeholders to reflect on the tutoring process at an early earlier stage of research.
The chapters on findings commence with the demand-side perspectives of students and parents and then turn to the supply-side perspectives of teachers. They systematically show the complexity of private tutoring and ways in which it is inextricably linked with aspects of mainstream schooling. This approach also explicates the larger political economy in which education is embedded. In the details regarding the main drivers of tutoring, time and money spent and the subjects for which tutoring is received, the higher tutoring rates of high-achieving students comes across as a significant finding. The authors have skilfully brought out the contrasting perceptions of the teachers, some of whom believe that high-achieving students need not seek tutoring as the low-achieving students are more likely to benefit from it. These findings nevertheless indicate how tutoring serves the enrichment purpose more than the remedial purpose of learning, which the authors state match with findings in several other places. With this, the authors point out the association of the examination system with private tutoring usage. The authors raise the two fundamental questions of firstly, the perceived effect of tutoring in shaping one's educational and career path and secondly the (in)sufficiency of schooling in providing the required competitive edge. These questions are relevant to the wider education research community that is grappling with issues of poor learning outcomes, restrictive assessment formats, and emphasis on instrumental goals over the intrinsic values of learning.
The final chapter moves to the policy implications, which is also the key focus of the book. Several of the authors' earlier books and papers provide directions for regulating the private tutoring sector, indicating approaches that have been tried in the other countries and drawing upon those lessons to suggest innovative, realistic and multi-pronged approaches. Grounding the implications on Myanmar's economic situation and education budget, the authors make several institutional and national-level recommendations. At the same time, they acknowledge numerous predicaments. Some are to do with students and parents, who may not only want to receive tutoring to better their life chances but may also consider it as a personal right. From the teacher's point of view, tutoring is a way of increasing their earning potential. For teachers not working as tutors, tutoring can be seen as sharing their burden of teaching. The government may consider it as contributing to improving learning outcomes.
Thus, the authors acquiesce that the need for tutoring can neither be eliminated nor is it something that can be easily regulated. Alongside, the readers are presented with evidence of countries where having a regulation has not been adequate for controlling problems arising from the tutoring sector. Yet despite these roadblocks, novel suggestions of self-regulation by the tutoring industry and extending the scope of tutoring to provide holistic learning experiences to students have been made.
An emerging interest in the field of education is the role of non-state actors, exemplified by its selection as the focus of UNESCO's 2021 Global Education Monitoring Report. This book therefore is extremely timely, and will interest academics, educational policy planners and research students alike. Each of the works on shadow education from the Comparative Education Research Centre at the University of Hong Kong contribute significantly to understanding of the tutoring phenomenon, and this book is a valuable addition to that literature. It provides a critical analysis of private tutoring situated at the intersections of economics, sociology and politics to move further beyond problematising private tutoring to recommending practical and implementable solutions. The limitations of the study are explicitly stated and so are further directions to continue assembling the jigsaw puzzle of tutoring. As pointed out by the authors, it is paramount to engage actively in dealing with the negative facets of private tutoring before tutoring reaches massive proportions and becomes deeply entrenched in the educational process.
Shalini Bhorkar shalini.tti@gmail.com in the past, but conclude that nowadays this is no more the case. Thus, it should no longer be framed as a "shadow" of education, but rather as another legitimate form of education (p. 25). Nevertheless, they still use the term "shadow education" throughout the book. Furthermore, five basic forms of shadow education are presented: home-visit private tutoring, private tutoring institutes, subscribed learning programs, internet-based private tutoring, and after-school programs. 1 Each form is described with examples from different (mostly East Asian) countries, subdivided into types, and summarized briefly at the end of each section.
Chapters four, five and six deal with three different uses of shadow education. First, to prepare and succeed at school and college admissions, second, to remedy student's learning problems. The authors speculate that mathematics is "arguably one of the most popular subjects for shadow education worldwide", since many students find it difficult (p. 105). Subsequently, a typology of mathematics shadow education is described in depth, although the authors admit that only their research in South Korea revealed those forms and other countries may have other variations. Third, shadow education is also used by gifted and highly motivated learners, whose needs are supposedly often ignored by schools and therefore they are often engaged in shadow education, too (p. 127).
The seventh chapter discusses the definition of shadow curriculum, which the authors understand as "supplementary curriculum out of schooling provided by educational business industries that is intended to improve academic success among individual students in formal education" (p. 149). According to the authors, shadow curriculum: • is based on students' academic needs, • is oriented towards student academic success, • is focused on school grades and exam preparation, • accelerates different learning opportunities based on family investment, • is oriented toward personalized learning, • exacerbates the competitive aspect of education.
The last two chapters deal with the role of shadow curriculum in a society and in research. The present time is seen as a postmodern schooling era. The authors assert that "…the status of schooling is eroding: It is competing with shadow education, and in some places, shadow education is winning" (p. 162), therefore, theoretical discussions about shadow curriculum are of great importance. Suggestions on what kind of curricular approaches should be applied to further study shadow curriculum and many specific questions in nine different areas of research are then provided by the authors.
The book is well structured and clearly arranged − in the introduction to a chapter, the topics are elaborated, at its end there is a short summary, a list of references, and at the end of the book there is an alphabetical index. The main arguments are supported by references from international literature, including various quotes from case studies carried out by the authors. A major part of the book is a quality compilation of existing literature from around the world, laying the ground for a more precise definition of shadow curriculum theory and helping the reader to understand the matter.
However, in some cases, the authors have misinterpreted the primary sources or supported their claims with non-relevant literature. 2 Many facts are mentioned in the chapters of the book repeatedly, 3 for example various definitions of shadow education, the metaphor of a shadow, Gangnam moms. 4 In such cases, mere references to the relevant chapters and sections might be sufficient. Chapters 2−6 seem to be written in a slightly different style from the rest of the book and have a different arrangement and structure than the other chapters. It raises the question whether the two authors worked together or whether they wrote parts of the book separately and merged them later. Moreover, chapters 2−6 offer a great number of typologies and their subtypes, which sometimes seem unnecessary and puzzling for the reader. A more careful selection of information based on its importance would also be desirable. For example, there are many redundant pieces of information (such as prices for individual types of tutoring in different countries) included in the main text. Some tables may seem dispensable, too.
To sum up, Shadow Education as Worldwide Curriculum Studies presents a useful overview of existing research on shadow education with a focus on its types, objectives, and developments worldwide. For scholars who are beginning to familiarize themselves with shadow education, the book may be a useful first step; however, for advanced shadow education researchers, it does not provide much new information. The situation in East Asian countries (with the focus on South Korea) is well described in the book and supported by many examples based on various (also the authors' own) studies. In some instances the authors describe a certain phenomenon comparing it to "Western countries". However, this comparison does not occur systematically and seems quite random. Primarily, the authors theorize the concept of shadow curriculum and suggest that it should be considered as important as school curriculum. They support this statement with evidence and offer many avenues for future research.