Rank effect on education VOX, CEPR Policy Portal

There are many contemporary policy-related debates in education, such as whether high-powered peers influence student achievement and how more female students can be encouraged to choose STEM programs in colleges and universities. Interestingly, how we think about these issues can determine whether students are more affected by their academic rank than other children in their class or year group. When studying academic rank, researchers usually evaluate students based on their performance in the aptitude or achievement test – keeping everyone else equal, including the absolute level of aptitude or performance – having a higher rank leads to better results for later students.

The effect of academic rank is a special kind of peer effect that depends on both the level of the student’s own human capital and the level of their peers. The empirical challenge is to distinguish rank effects from other potential peers and classroom influences while allowing rank correlation with absolute achievement and with student characteristics such as gender or color.

Important papers such as Murphy and Weinhardt (2020), Cicalo et al. (2018), and Elsner and Isphording (2017, 2018) have recently kickstarted a major literature on rank effects; We have learned much about the size of these effects and their relevance to understanding broader issues in the economics of education. Our recent survey paper (Devereux and Delaney 2022) reviews this recent literature, describing difficult identification and measurement problems, literature-based estimates and methods, and empirical results. In this column, we describe why rank is important and outline some key outcomes, but we also focus on what insights the existence of rank provides to other educational phenomena, such as how students benefit from high-powered peers and attendance. Gender gap in STEM.

Why can academic rank matter?

Student rankings can influence subsequent results through many different channels. Having a high rank can boost confidence and make students more likely to study and do better, hang out and be influenced by other high-ranked students, or be less likely to engage in disruptive behavior. In addition, teachers and parents can provide more support and encouragement to higher-ranked students, and this can lead to better performance. Similarly, a higher position in a subject can convince a student that their comparative advantage lies in that area and later choose it as a specialization in college or university. The effects of rank can be somewhat mechanical even if universities are more likely to admit students with higher rank in secondary school. Although there are many possible mechanisms, evidence shows that higher rankings increase students’ confidence in their abilities, which in turn has a beneficial effect on their behavior and expectations.

The main result about the rank

Studies that calculate strengths or accomplishments across different countries, ages, and types of educational institutions have generally found that students who rank higher in their class or grade have better subsequent results. Using UK data, for example, Murphy and Weinhard (2020) found that students who got higher grades in primary school achieved higher test scores during high school. In the United States, Elsner and Isphording (2017, 2018) found that higher-ranked students were more likely to finish high school and go to university, and they found that rank had a negative effect on the likelihood of smoking, drinking, and unprotected sex. , And involved in violence. The size of the effect is usually larger in terms of rank than the amount available for other factors and interventions. For example, using U.S. data, Denning et al. (2021) showed that moving from 25th to third grade to 75th percentile (year four) increased income by about 7% at the age of 23-27 in the rank, a general estimate of income return in the extra year of schooling.

Impact for peer group influence

If students benefit from higher rankings, it may offset or deny any advantage of having a high-editor peer, as more high-power peers will lower their class rank. Therefore, the effects of rank may justify the general failure to find the strong positive effects of high-power counterparts when the assumed rank is not taken into account (Cooles et al. 2019). Indeed, Bertoni and Nistico (2019) show that the approximate effect of high-performance peers is much more positive when regression control for rank is introduced. For example, the results obtained from the literature on rank have important implications for how we think about peer influence in general.

The effect of gender gap on STEM

Women are still widely represented in university science, technology, engineering, and mathematics (STEM) programs (Kugler et al. 2017, Delaney, and DivX 2019), and field choices could have a significant impact on future labor market success (Dahl et al. 2020). Research has linked this gender gap to a number of factors, including the comparative advantage of girls in verbal skills rather than math. However, students may not know their subject-specific abilities and may estimate the relative advantage across subjects from their rank in these subjects in their school or class. At any perfect level of mathematical and verbal ability, it may be that students in higher positions in mathematics are more likely to apply to university STEM programs than in verbal skills.

In Delaney and Devereux (2021), we use Irish administrative data on university applicants to investigate whether mathematics affects the choice of a university principal, in grades in English, and on the condition of perfect achievement at the end of high school. We find that higher rank in mathematics increases the chances of applying to a university’s STEM program, whereas higher rank in English reduces the chances. Figure 1 illustrates this search using a flexible specification that includes the indicator variable for each ventilation of the rank distribution and the indicator for being the top person (s) in the school-cohort subject, the 10th ventilation exclusion section. Controls regression for test scores in math and English, as well as for other school- and cohort-level variables. Coefficient estimates and 95% confidence intervals have been plotted. The effect of math rank on STEM is almost linear. Students with higher math rank are more likely to apply for STEM. In contrast, the negative impact of English rank on STEM is evident in the upper half of the English rank distribution; The relationship distribution is quite flat in the lower half.

Figure 1 Influence of Rank Ventile and Top Person in Mathematics and English as a First Choice Study

Comments: Estimates from regression where subject rank is entered in the ventilator, with an additional section for the top person (s). Excluded section is 10th ventil. The difference between point estimates and 95% confidence is shown.
Source: Delaney and Devereux (2021).

In mixed-sex schools, males tend to rank higher in mathematics and lower in English than female students (Delaney and Devereux 2021). We estimate that the preferred gender gap for STEM at university will be reduced by about 4% if mixed-sex schools have the same rank for both boys and girls, while maintaining their absolute level of academic achievement. Rank effects may partly explain why Irish mixed-sex schools have a wider gender gap in STEM university applications than gay schools: in gay schools, the average rank of boys and girls, by definition, is the same in each subject. Overall, different ranks in mathematics and English may explain a small but significant part of the gender gap in the decision to choose STEM as the head of a college or university.

The comment is final

Evidence of the long-term impact of academic rankings raises questions about how well students know their rankings and whether they should be given more information. Students can use their class rank in a subject as a measure of how competent they are in that subject, which may be a suboptimal career choice if they are in a school or class that is unusual in allocating skills to students. If students are influenced by their rank, it may be helpful to provide information about their qualifications or achievements in a variety of subjects compared to the general population. Given the evidence that class rank is influential but sometimes confusing, this policy intervention can give students the ability to make better decisions about their university major and possibly other academic and career choices.


Bertoni, M and R Nisticò (2019), “Ordinary Rank and Peer Composition: Two Aspects of the Same Currency?”, IZA Discussion Paper No. 12789.

Cicala, S, RG Fryer and JL Spenkuch (2018), “Self-selection and comparative advantage in social interaction”, Journal of the European Economic Association 16 (4): 983–1020.

Cools, A, R Fernández, and E Patacchini (2019), “Girls, boys, and high achievers”, VoxEU.org, 30 August.

Dahl, G, D Rooth and A Stenberg (2020), “Payment of Earnings in Different Areas of Secondary School Education”, VoxEU.org, 10 November.

Delaney, JM and PJ Devereux (2019), “It’s not just for boys! Understanding Gender Differences in STEM “, VoxEU.org, 19 April.

Delaney, JM and PJ Devereux (2021), “High school placement in math and English and gender gap in STEM”, Labor economy 69

Delaney, JM and PJ Devereux (2022), “The Impact of Rank in Education: What Do We Know So Far?”, CEPR Discussion Paper 17090.

Denning, JT, R Murphy and F Weinhardt (2021), “Class Rank and Long-Term Outcome”, NBER Working Paper 27468.

Elsner, B and IE Isphording (2017), “A Big Fish in a Small Pond: The Dignity of Power and the Investment of Human Capital”, Journal of Labor Economics 35 (3): 787–828.

Elsner, B and IE Isphording (2018), “Rank, Sex, Drugs and Crime”, Human Resources Journal 53 (2): 356–381.

Kugler, A, C Tinsley and O Ukhaneva (2017), “Why there are more women in STEM fields”, VoxEU.org, 2 November.

Murphy, R. and F. Weinhard (2020), “Top of the Class: The Importance of Ordinal Rank”, Review of Economic Studies 87 (6): 2777–2826.

Leave a Reply

Your email address will not be published.