The impact of attending classes with smaller sized groups constitutes one of the most widely researched themes in education economics while also among the least consensual.
In searching for correlations in student results from the PISA – Program for International Student Assessment test and the average school class size, we do not encounter any defined pattern between these two variables. (*)
Hence, the objective involves identifying programs and experiments reducing the number of students per class and informing on their short and long term impacts.
Joshua Angrist, who was recently awarded the Nobel Prize for Economics, in conjunction with his co-author, Victor Lavy, were among the pioneers in studying this question. In a 1999 work, the authors deployed an unusual characteristic of the Israeli education system inspired on a Talmudic principle from the 6th century that discusses the appropriate ratio of students and teachers for studying the sacred text. According to Rabbi Maimonides, writing in the 12th century, this principle required interpreting as follows: a group of up to a maximum of 25 students should be allocated to a single teacher; when there are between 25 and 40 students, the teacher requires an assistant in the classroom; when the group contains over 40 students, the class is to be divided. This then became known as the Maimonides Rule.
This same rule is applied to structure part of the mechanism allocating students to classes in Israel. Correspondingly, a class with 40 students remains intact while another with 41 students is then split into two. The impact of attending small classes may thus be measured through the differences between these two groups of students. This led to evaluating the impact of this mechanism for structuring classes in the 3rd, 4th and 5th years of schooling in accordance with the student results obtained in the Hebrew and Mathematics tests taken in 1991 and 1992 across a universe of around 62,000 children. This approach reported that students in smaller classes had benefited from positive impacts in terms of their test results and with these higher in the 5th year than the 3rd year and the 4th year, respectively.
More recently, in 2019, Angrist and Lavy revisited this methodology based upon more recent results, and from between 2002 and 2011 for more than 200,000 5th year students and around 9,000 classes. This time around, there were no impacts on the results of students attending smaller classes. In their study, the authors propose that the positive effects found in the older data were less precise and abnormally positive.
Another of the best known programs reducing class sizes is the famous Student/Teacher Achievement Ratio (STAR), launched in 1985 in 79 schools in Tennessee, the United States of America, involving 11,600 students in the final year of pre-school through to the 3rd year of schooling. During the first four years, students were randomly assigned to small classes (with between 13 and 17 students, and regular classes (between 22 and 25 students), but with the latter including a subgroup designed to provide additional support in the classroom. The teachers were also randomly assigned to these classes. The randomness of the distribution of students and teachers made STAR an example of the feasibility of isolating the effect of belonging to different class sizes through excluding potential bias caused by other factors or characteristics of the students or teachers on this allocation and would otherwise limit the interpretation of this program. Another particularly important aspect derives from the range of the data as the study accompanied these students throughout their lives and thereby observing the impacts of exposure to smaller class sizes during infancy over both the short and long term.
One of the first studies to emerge from this program, by Alan B. Krueger, published in 1999, analysed the impacts of reducing the number of students in classes as measured by tests taken between the 1st and 3rd years of schooling. This calculated that students from the smaller classes saw their Reading and Mathematics test results rise by four percentage points at the end of the first year with this effect diminishing in the following years to reach a maximum equivalent to one percentage point. For students attending regular sized classes but with extra support from the teacher, the findings state they did not benefit from any particularly positive impact except in cases from socioeconomically disadvantaged households.
Another 2001 study, by Krueger and Whitmore, focused on the impacts more distant in time and correspondingly illustrated how belonging to the smaller classes under the auspices of the STAR program raised the likelihood of students sitting university access tests by four percentage points. This also registered a particularly significant drop in the level of inequality prevailing between white and black students and also returning a small but positive impact on the results obtained in these tests.
Another relevant study, carried out by Raj Chetty and a six-member team in 2011, reached further in observing the outcomes of STAR project students in university and following their entrance into the employment market. The findings state that attending smaller classes during the first four years of schooling reflected in a rise in the likelihood of attending university by the age of 20 by 1.8 percentage points and by 1.6 percentage points through to 27 years of age. This higher level of education generated a small average gain in earnings of 109 dollars among study participants, with the average annual wage of around 15,000 dollars.
Applying data from a European education system, the 2018 study, by Bethlehem Argaw and Patrick Puhani, focused on the German state of Hesse, included a sample of around 250,000 students, and reported new, non-definitive impacts from reducing the number of students per class. In this German state, as in the Israeli case, there are rules determining the maximum and minimum class sizes and enabling the identification of students attending larger and smaller classes. The authors measure the consequences of smaller class sizes on the probability of students being recommended to attend academic education or professional training options from the age of ten, before concluding that, on average, this is either only small or close to zero. However, there is heterogeneity across the results. For example, for boys, attending classes with less than ten students boosts the likelihood of being recommended for an academic option by three percentage points.
Should we move the focus away from policies and onto individual experiences, a study by Ludger Woessmann and Martin West, in 2006, analysed a vast set of countries based on data contained in the Trends in International Mathematics and Science Study (TIMSS) for the 1994/1995 academic year. Taking a total of 11 countries into consideration and evaluating students aged 13 (between the 7th and 8th years), the authors conclude that the consequences of attending smaller classes are less than those returned by the results of the STAR project and without any generalised evidence of positive effects. Given this study took into account a range of different education systems, we may conclude that the effects of reducing the class size tend to be greater when the average quality of teaching staff is lower, thereby serving as a possible factor of compensation. This explains, for example, the differences in the positive effects from lowering the class sizes in countries such as Greece contrasting with the total lack of any influences in Singapore.
The discussion around the reduction in class sizes should also extend to the positive impacts over the short and long term, contrasting these impacts with the scope of the variation in student numbers. The positive effects of this policy need to take on a sufficiently large magnitude to offset the costs inherent to contracting more teachers and/or the need for larger facilities in schools. When all is said and done, this analysis also needs to provide a cost-benefit relationship for these alternative policies. Therefore, this requires contemplating whether there might be other policies capable of obtaining the same results at far lower costs.
(*) OECD, Pisa 2015 Database, Tables I.2.3 and II.6.26. http://dx.doi.org/10.1787/888933436345
Angrist, J. D., & Lavy, V. (1999). Using Maimonides' Rule to Estimate the Effect of Class Size on Scholastic Achievement. The Quarterly Journal of Economics, 114(2), 533–575. http://www.jstor.org/stable/2587016
Angrist, J. D., Lavy, V., Leder-Luis, J., & Shany, A. (2019). Maimonides' Rule Redux. American Economic Review: Insights, 1 (3): 309–24. https://doi.apa.org/doi/10.1257/aeri.20180120
Argaw, B. A, & Puhani. P. A. (2018). Does class size matter for school tracking outcomes after elementary school? Quasi-experimental evidence using administrative panel data from Germany. Economics of Education Review, 65, 48–57. https://doi.org/10.1016/j.econedurev.2018.06.005
Chetty, R., Friedman, J. N., Hilger, N., Saez, E., Schanzenbach, D. W., & Yagan, D. (2011). How Does Your Kindergarten Classroom Affect Your Earnings? Evidence from Project Star. The Quarterly Journal of Economics, 126(4), novembro, 1593–1660. https://doi.org/10.1093/qje/qjr041
Krueger, A. B. (1999). Experimental Estimates of Education Production Functions. The Quarterly Journal of Economics, 114(2), 497–532. http://www.jstor.org/stable/2587015
Krueger, A. B., & Whitmore, D. M. (2001). The Effect of Attending a Small Class in the Early Grades on College-Test Taking and Middle School Test Results: Evidence from Project STAR. The Economic Journal, 111(468), 1–28. http://www.jstor.org/stable/2667840
Woessmann, L., & West, M. (2006). Class-size effects in school systems around the world: Evidence from between-grade variation in TIMSS. European Economic Review, 50 (3), 695-736. https://doi.org/10.1016/j.euroecorev.2004.11.005