Students exposed to more informative grading are less likely to graduate from high school
In education, it is commonplace for students to receive feedback in the form of grades. Yet, despite the widespread use of grading feedback in today's schools, we have little knowledge about how the type and, in particular, the precision of feedback affects performance. The findings of a new PhD thesis from Lund University show that exposure to a more granular grading scale has negative consequences for students. Students exposed to a more granular grading scale are less likely to graduate from high school, from an academic high school track and from STEM high school tracks. These results appear to be driven by a negative shock to students’ self-belief and increased stress.
By Jonas Lundstedt
A more granular grading scale provides a more accurate assessment of one's achievement, providing more informative feedback to students on their performance. There are many reasons why more informative grading could matter for an individual’s educational outcomes. On the one hand, more informative grading should provide students with more accurate information on their strengths and weaknesses, helping students to find better matches in education and the labor market (Bobba and Frisancho, 2019; Chevalier et al., 2009). More informative feedback reduces information frictions not only for students but also their parents, allowing parents to make more appropriate investments in their children’s human capital (Cobb-Clark et al., 2021; Dizon-Ross, 2019). On the other hand, if people are overconfident, like a large literature suggests, more informative grading may lead to reduced self-belief (Bobba and Frisancho, 2020), lower effort (Fisher and Sliwka, 2018, Kőszegi et al., 2022) and increased dropout rates (Stinebrickner and Stinebrickner, 2014). The effect of more informative grades on educational outcomes is therefore an empirical question.
Studying the effect of more informative grading is challenging. Comparing school systems with different types of grades is not informative, as there is a myriad of institutional, cultural and demographic factors affecting outcomes that vary and may be difficult to account for. To understand how more informative grades affects student outcomes, Collins and Lundstedt (2023) instead examines a reform to the compulsory school grading system in Sweden in 2011. The reform replaced the previous grading scale, which included three passing grades, with a more granular scale with five passing grades. As children in Sweden are assigned to school cohorts based on their date of birth, the study exploits that children born just after January 1st, 1997 were exposed to more informative grading, while those born just before were not.
If we were to compare the difference in outcomes between those born just before and just after the admissions cut-off of January 1st, 1997, the estimated effect of more informative grading would however be confounded by school-starting age effects. By comparing the difference in outcomes between those born just before and just after the admissions cut-off of January 1st, 1997 to the difference in outcomes of those born just before and just after the January 1st admissions cut-offs from 1992 to 1996, we are able to separate the confounding effect of school starting age and identify the effect of exposure to more informative grading.
The research design is presented graphically in figure 1. The figure shows the observed share of individuals born in each month in the window surrounding January 1st who graduated from an academic high school track (as an example outcome). The left panel shows a clear discontinuity in the outcome at January 1st for the pooled comparison cut-offs. Over this period, students born just after January 1 exhibit a higher likelihood of graduating from an academic high school track than those born just prior to January 1st. This is the school starting age effect. The right panel shows that this discontinuity does not occur at the treatment cut-off (i.e., 1996/97). The difference between these two estimated discontinuities represents the treatment effect. As can be seen, there is a large reduction in the discontinuity in the outcome at the treatment cut-off, relative to the pooled comparison cut-offs, indicating that the grading reform had a negative effect on the likelihood of graduating from an academic high school track which is approximately equal in magnitude to the school starting age effect.
Figure 1 Difference-in-discontinuity visualisation
Notes: This figure plots graduation from an academic high school track by month of birth. The left panel shows a clear discontinuity at January 1st, which represents the school starting age effect present at the 1991/92, 1992/93, 1993/94 and 1995/96 cut-offs. The right panel shows the 1996/97 cut-off, corresponding to the introduction of more informative grading, where there exists no clear discontinuity.
Overall, we find that exposure to more informative grading has negative consequences for educational outcomes, with treated students being 3.3% less likely to graduate from high school and 7.6% and 11.9% less likely to complete an academic track or an academic STEM track in high school. While we do not find evidence of reduced university enrollment, we do find a 16.7% decrease in the likelihood of continuing to a STEM track in university. We also estimate that exposure to more informative grading leads to a 1.2% reduction in average yearly earnings between ages 28-30.
To investigate potential mechanisms underlying our results, we use survey data from an internationally representative sample of school children to compare treated Swedish students to their counterparts in Denmark and Norway, finding that treated children are less likely to see themselves as performing well or very well in school. Treated children are also more likely to report feeling pressure from their school work and to report experiencing sleeping difficulties. This suggests that the reductions in high school graduation and STEM participation are due to reduced self-efficacy and increased stress.
Our results give insights into how receiving more informative feedback in one's formative years can have large and persistent impacts on outcomes, with important implications for the framing of feedback in the education setting. While our results do not speak to whether children should be graded at all, when it comes to the information content of grades, less is more.
 So-called school-starting age effects refer to the phenomenon that in countries where children are assigned to school cohorts according to their year of birth, children born early in the calendar year typically perform better than those born later in the calendar year (Black et al. 2011; Fredriksson and Öckert, 2014).
More from the author
This research was undertaken as part of the author’s PhD studies at the department of Economics at Lund University (Lundstedt, 2023). The thesis includes two additional studies examining various topics in the economics of education. One chapter examines students’ expectations about the signaling value of the institutional seal on their diploma. The result suggest that students’ educational decisions are affected by a demand for signaling and not only a human capital investment motive – which, in turn, can result in productivity losses on the individual and aggregate level. The other chapter investigates the effect of student aid to parents on participation in formal education, finding that a lower financial threshold for university increases enrollment and completion rates among parents, especially for mothers.
Black, S. E., Devereux, P. J., Salvanes, K. G. (2011). Too young to leave the nest? The effect of school starting age. The Review of Economics and Statistics, 93(2):455-467.
Bobba, M. and Frisancho, V. (2019). Perceived ability and school choice. TSE Working Paper.
Bobba, M. Frisancho, V. (2020). Self-perceptions about academic achievement: Evidence from Mexico City. Journal of Econometrics, 231(1):58-73.
Cobb-Clark, D. A., Ho, T., and Salamanca, N. (2021). Parents responses to children’s achievement test results. IZA Discussion Papers, No. 14663.
Collins, M. and Lundstedt, J. (2023). The long-term consequences of more informative grading. Working Paper.
Chevalier, A., Gibbons, S., Thorpe, A., Snell, M., and Hoskins, S. (2009). Students’ academic self-perception. Economics of Education Review, 28(6):716-727.
Dizon-Ross, R. (2019). Parents’ beliefs about their children’s academic ability: Implications for educational investments. American Economic Review, 109(8):2728-2765.
Fisher, M. and Sliwka, D. (2018). Confidence in knowledge or confidence in the ability to learn: An experiment on the causal effects of beliefs on motivation. Games and Economic Behavior, 111:122-142.
Fredriksson, P. and Öckert, B. (2014). Life-cycle effects of age at school-start. The Economic Journal, 124(579):977-1004
Kőszegi, B., Loewenstein, G., and Murooka, T. (2022). Fragile self-esteem. The Review of Economic Studies, 89(4):2026-2060.
Lundstedt, J. (2023). Essays in education economics. Lund Economic Studies Number 235.
Stinebrickner, R. and Stinebrickner, T. (2014). Academic performance and college dropout: Using longitudinal expectations data to estimate a learning model. Journal of Labor Economics, 32(3):601-644.