1Among marked shifts in the nature of educational provision during recent decades has been the rise of private supplementary tutoring. The phenomenon has a long history – perhaps almost as long as the history of schooling – but until recently it had limited scale and was restricted to a minority of wealthy families. Now private tutoring has become widespread, in some countries enrolling over 70% of the students in some grades, and serving families of all income groups. On the positive side, private tutoring promotes learning and provides employment; but it demands substantial household expenditures, maintains and exacerbates social inequalities, and can have a backwash on schooling.

2For this issue of the journal focusing on data in education, the article commences by sketching what we know about the scale and nature of private supplementary tutoring around the world. The short answer is that we know the general contours, with more detail in some countries than others. However, we also have major gaps. The article therefore explains why the matter is important, and then highlights data needs and ways to bridge the gaps.

3To frame what follows, a definition is needed. The focus is on fee-charging supplementary tutoring in academic subjects outside school hours at the levels of primary and secondary education. Parts of the literature describe such tutoring as shadow education because much of its content mimics schooling: as the curriculum changes in the schools, so it changes in the shadow (Bray, 2023; Zhang, 2023).

The scale and nature of private tutoring

4East Asia is the world region most noted for the scale of private tutoring. Thus, Japan is known for its jukus and South Korea for its hagwons which provide institutional forms of out-of-school support. These institutions grew in number and significance during the second half of the 20th century, and have become a permanent feature in their educational ecosystems. The phenomenon also has a long history in such countries as Egypt, Greece and Mauritius, and more recently has emerged almost everywhere else. Snapshots from around the world (Bray, 2024; Galinié and Heim, 2016; Zhang, 2023) include the following:

• Asia: In South Korea, 63% of students across all grades received private tutoring in 2022. In Japan, 2015 national data indicated that 48% of Grade 6 and 61% of Grade 9 students received tutoring in jukus or with individual tutors. In Cambodia, a 2019/20 national dataset indicated enrolment rates of 24% among primary, 57% among lower-secondary, and 77% among upper-secondary students. An Indian survey in rural West Bengal found in 2022 that 63% of Grade 1 students were receiving shadow education, with the proportion reaching 76% in Grade 8; and urban enrolment rates were presumably even higher.

• Europe: In England, 11% of students aged 11-16 surveyed in 2022 were receiving shadow education, and 30% (46% in London) had done so at some time. In France, among 35,000 Grade 6 students sampled in 2008, 9% were receiving tutoring. Then, following the same sample in 2011 when the students had reached Grade 11, 14% (but not necessarily the same students) were receiving tutoring. In Greece, 90% of Grade 12 students surveyed in 2017/18 were receiving tutoring.

• Latin America: In Brazil, 2014 national data from 5.4 million candidates for the High School National Examination provided information about supplementary support (including some that was fee-free) and other matters. Among the candidates, 19% had received support for the selection examination at some stage in their school careers, 17% had received support in foreign languages, 62% in computer science, and 34% in other subjects.

• Middle East and North Africa: In Egypt, a 2018 national survey found that 62% of students across all grades received tutoring, commencing with 40% in Grade 1 and reaching a peak at 73% in Grade 9. In Qatar, a 2018 survey found shadow education enrolment rates of 38% in Grade 8, and 56% in Grade 12.

• Sub-Saharan Africa: In Mauritius, 81% of Grade 6 students were estimated in 2013 to be receiving shadow education, while proportions in Botswana and Malawi were 35% and 14%, and there is every reason to assume that such rates have been maintained and even increased. Comparable 2019 figures were 22% in Senegal, 20% in Cameroon, and 16% in Togo.

5This mapping shows significant enrolment rates not only in high-income countries such as England and Qatar but also in low-income ones such as Cambodia and Senegal. A fundamental underlying force is social competition, exacerbated by population mobility, globalisation, urbanisation, population expansion, and massification of education systems.

6Building on the simple enrolment rates are questions about the modes, intensities, and locations of tutoring. Concerning modes, tutoring may be delivered one-to-one, in small groups, and/or in full classes. In societies as diverse as Egypt, Hong Kong and Sri Lanka, tutoring may be in large lecture theatres, delivered by ‘star’ tutors. Increasingly, tutoring is also delivered online. This had already become common in the 2010s, facilitated by technological developments, and expanded during the Covid-19 pandemic when direct teaching in both schools and tutorial centres was prohibited. The intensity of tutoring generally increases as the grades progress, peaking just before high-stakes examinations especially at the end of secondary schooling. Some students receive tutoring for just a few hours a week immediately before the examinations, while others receive greater intensity perhaps throughout the academic year and including vacations. Much tutoring, especially one-to-one and in small groups, is received in the homes of students or tutors, or in public spaces such as libraries and cafés, while other tutoring is in the premises of tutorial centres.

7Diversities are exemplified by a 2018 survey in the United Arab Emirates (UAE) of 3,929 parents of children in Grades 5, 9, 10 and 12 in 85 schools (Rocha & Hamed, 2018).

• Modes: Among Emirati students, 62% received one-to-one tutoring, 24% in groups of two to five students, 11% in groups with more than five students, and 2% online. Corresponding numbers for non-Emirati students were 45%, 37%, 14%, and 4%.

• Intensities: The numbers of hours per week varied by subject. Among sampled students, 83% received tutoring in mathematics; and among these, 78% did so regularly for 1-3 hours per week while 22% only did so before examinations. For Arabic, only 44% of students received tutoring, among whom 82% did so regularly and only 18% before examinations. Further variations were evident for science and English.

• Locations: Overall, 59% of parents reported that tutoring was received in their homes. Other tutoring was in the tutors’ homes (11% among Emiratis and 29% among non-Emiratis). Approximately 10% of tutoring was in designated centres, and 6% in schools.

8Research in other contexts might show different proportions and further diversity, e.g. in mixed offline and online tutoring, across grades and perhaps genders, and in greater variety of locations. Research could also focus on content. Thus while some tutoring, especially for slower learners, repeats the content of school lessons, other content stretches higher achievers by build on schooling for elaboration and enrichment.

9Then, turning to the suppliers, questions arise about the identities, ages and qualifications of the tutors. Tutoring may be provided by companies, teachers earning extra incomes, and informal providers such as university students, self-employed tutors and retirees. Patterns vary around the world, with the South Korean provision dominated by companies and little provision of tutoring by serving teachers since they are paid well and do not see the need for extra incomes. In Cambodia, by contrast, most tutoring is provided by teachers since their salaries are low. Locations with large numbers of university students providing tutoring have bias towards provision by young people without professional training, but companies commonly employ older people including personnel who used to teach in schools. Such people are commonly trained as teachers, though not necessarily as tutors requiring slightly different orientations, and in general the industry has few regulations on this matter. The UAE study cited above found that, depending on the subject, about 13% of children were tutored by their own teachers, 12% by other teachers in their schools, 47% by teachers from other schools, and 29% by personnel in tutorial centres or elsewhere (Rocha and Hamed, 2018).

Why does it matter?

10Private tutoring can and should promote learning. Yet whether it actually does so depends strongly on the motivations of the students, the skills of the tutors, the curricula, and many other factors possibly including time of day, day of week, season of year, location, social class, age, and gender. For such reasons evaluations have shown mixed findings, which might particularly worry parents who are investing considerable time and money.

12Also of great importance are broader social issues. Since private tutoring requires fees, lower-income families cannot afford either the quantities or qualities of the tutoring consumed by higher-income families. Again in Egypt, only 32% of students in the lowest income quintile received private tutoring, rising to 55% and 67% in the two highest quintiles. The enrolments are raised by social pressures when everybody else seems to be investing in tutoring. In France and elsewhere, analysts have linked such matters to the marketisation of education (Bisson-Vaivre et al., 2023).

13On a different tack, research commonly shows that children residing in urban areas are more likely than their rural counterparts to receive tutoring. In France, Galinié and Heim (2016) found the highest enrolment rates in shadow education in Paris and in cities with over 100,000 inhabitants. Next on the list were enrolment rates in medium-sized cities (50,000 to 100,000 inhabitants), and lowest were enrolment rates in rural areas. Such statistics may have reflected income levels and perhaps the more competitive nature of urban societies, but also reflected availability of tutoring. The picture has counterparts in most other countries around the world.

14Of further importance is the backwash of tutoring on schooling. When teachers provide tutoring alongside their main jobs, they may neglect their main responsibilities in order to devote more effort to the private activities. Especially problematic are circumstances in which teachers tutor their existing students, since the teachers may be tempted deliberately to cut content from their regular classes in order to promote demand for the private ones. Such patterns have been reported in Egypt and also, for example, in Cambodia, India and Turkey. Other forms of backwash arise around the world when students learn materials first in tutoring and are then bored and perhaps disruptive of lessons during their regular schooling. Also, when high-achieving students receive tutoring on top of schooling, teachers face widened disparities in their classrooms.

What are the main data gaps and why?

16By contrast, South Korea stands out for its comprehensive data on private tutoring enrolments and expenditures collected annually with systematic coverage of different geographic regions and educational levels. This exceptional situation reflects the government’s longstanding concerns, and the capacity to collect data facilitated by an advanced economy with financial and human resources. No other government has commitment resembling that of South Korea. This is partly because many Ministries of Education are in effect Ministries of Schooling, and view private tutoring as beyond their remit. International bodies such as Unesco and the OECD do not strongly encourage national governments to collect such data, and domestic pressure even in such countries as Brazil, Egypt, Greece, Japan and the UAE is insufficient to push governments to collect detailed data.

Cette année scolaire, environ combien d’heures par semaine passez-vous à étudier les matières suivantes en plus du temps prévu dans votre horaire de cours obligatoire ? (Comptez le nombre total d’heures que vous passez à faire vos devoirs, à suivre des cours supplémentaires et à étudier.) [underlining and italics in the original]

18Analysts desiring data on private tutoring were frustrated that it was merged with homework and private study, and that the question did not ask whether the additional instruction was fee-charging or fee-free. Also, while the question focused on “Cette année scolaire”, it did not allow for variations in the examination season, mid-term break, vacations, etc. This arrangement allowed for inconsistent and confusing reporting. Other ambiguities arose in a subsequent question about types of individual instruction, with “Cours donnés par une personne” overlapping with “Cours particuliers entre moi seul(e) et la personne qui le donne”; and ambiguities in the original English-language version were compounded by problems of translation into other languages (Bray et al., 2020).

19Extending this illustration, challenges may be noted in the design of PASEC questions. The opening question in the relevant section of the 2014 survey was about devoirs, which is not a synonym for supplementary tutoring and oriented the respondents in a different direction. Further, the question asked about such work done “à la maison”, whereas tutoring could be received in many other locations including the teachers’ homes, public venues, and company premises. Then, it asked “Si quelqu’un t’aide à la maison pour les devoirs, qui t’aide ?”. The options had ambiguities because tutrice/tuteur can mean either a guardian or a tutor. These and related problems led to revision in the 2019 version of the questionnaire, but did not fully resolve the issues. Consequently, although the data provided some indications, they could not be trusted for precision (Bray and Baba Moussa, 2023).

20A further reason for inadequate data arises from the potential inabilities and/or unwillingness of respondents to provide data. Thus, PASEC collects data from children in Grade 6, from teachers, and from parents; but Grade 6 children cannot easily answer questions on complex domains, especially since many of the African schools on which PASEC focuses suffer qualitative challenges. Teachers have greater competence but may avoid questions on private tutoring, particularly when they are themselves providers in circumstances of potential social disapproval and perhaps contravention of government policies. Parents may also avoid such questions, especially when tutoring might seem either to label their children as underperformers or as receiving unfair advantages. Such factors underline the importance of (realistically) assuring respondents that the data will be treated confidentially. Respondents should see that the questionnaires are anonymous, and administrators will need the personal skills to be convincing in their assurances. This may require briefing and training of survey administrators and data processors.

21For the above reasons, specialised research, even if smaller scale and locally focused, may be more illuminating than the PISA, PASEC and similar surveys that must cover multiple topics with restricted depth. In local contexts, researchers may be better able to gain the confidence of the data providers and to adapt their questions to those contexts.

22However, the question still arises how best to secure the necessary data. For expenditures on private tutoring, household surveys are more reliable than school-based surveys since children often do not know precisely what their parents spend. Household surveys can also secure data on enrolment rates across the grades of schooling, and alongside Egypt and India has been an important statistical source in Pakistan. Again, a major challenge is to include questions with sufficient detail, given that many other priorities demand attention. Other domains for research, as noted above in connection with the effectiveness of tutoring, might focus on the motivations of the students, the skills of the tutors, the curricula, and many other factors. These require their own specialist methodological approaches.

*
**

23Most of the statistics presented above are simply snapshots taken at specific points in time with limited samples determined by the photographer, the visible scenery in front of the photographer, and the chosen angle of vision. Some of the statistics are over a decade old, and thus unable to reflect fast-changing social and economic dynamics including, for example, technological developments and the impact of Covid-19. As mentioned, online tutoring expanded when schools were closed by the pandemic, and multiple forms of tutoring have been encouraged in the post-pandemic era not only by entrepreneurs but also by some governments as a way to compensate for perceived learning loss (see e.g. Hatch, 2023). Given the global growth of shadow education and its far-reaching social, educational and economic implications, it seems obvious that more and better data are needed.

24Returning to the 1.6% of GDP in Egypt consumed by private tutoring as reported by Moreno (2022), it is noteworthy that the author did not specify the year to which the number applied. His article referred to a 2015 source, which itself referred to a 2007 source, which referred to a 2003 source, which referred to a 2002 source, which referred to a 2000 source, that specified 1997 as the year of the data. Moreno has respected expertise, having been a senior World Bank employee and having focused in detail on Egypt among other countries. The fact that even he could not secure an updated statistic for his article underlines the point about missing data. In practice even casual observation would have shown that household expenditures on private tutoring were high in 1997 and remained high when Moreno wrote his 2022 article. Yet perhaps even the 1997 data might have needed checking for reliability.

25The question then is why we do not currently have better data. More than any other factor, the answer lies in prioritisation, since to date the significance of private tutoring has been under-recognised across the globe. Ironically, this is chiefly because the data have not been collected: politicians, administrators and the general public are not adequately informed about the scale, nature and implications of the phenomenon. Yet experiences in Egypt, Greece, India, Mauritius, South Korea and elsewhere show that once shadow education becomes engrained in the culture, it is very difficult to remove. The actors in the ecosystem become habituated to it, and those with vested interests resist threats to their mode of operation. Most obvious among stakeholders with vested interests are the tutors who earn substantial revenues. Yet ambitious parents who view tutoring as a path to secure high social status may also resist threats to tutoring supply and to forms of regulation. Indeed, even schools may covertly support the existence of shadow education since they commonly claim the credit for students’ academic achievement even if much of that credit is due to the tutors.

26Once there is concern about the matters, the technical issues can be solved relatively easily, at least insofar as they relate to basic indicators on enrolments, costs and suppliers for learners in different grades, ability levels, locations, and social groups. Beyond such basic indicators, of course, are many refinements needing data not only from quantitative surveys but also from qualitative studies. Insofar as in some settings shadow education has enrolments and expenditures representing at least two thirds of those in schooling, arguably the research on shadow education should quantitatively attract the equivalent of at least two thirds of the research on schooling. In the immediate future, that possibility seems distant. However, at least private tutoring is now gaining more attention than it received in the past, and the research that does exist helps to take it out of the shadows. This is important not only for governments but also for schools, families, community advisers, and even tutorial companies and self-employed tutors.