Insufficient equity in education is recognized as one of the major challenges in school systems around the world (OECD, 2007; Ainscow et al., 2012). Countless studies have documented that students with richer and more well-educated parents tend to perform better in school (White, 1982; Sirin, 2005; Harwell et al., 2017). Although this phenomenon appears to be quite universal, the size of the achievement gap between students with high and low socioeconomic status (SES) is not constant. For example, the gap appears to change with time (Chmielewski, 2019) and to be moderated by school factors (Gustafsson et al., 2018). Here we are concerned with another potential moderator of the SES achievement gap: gender.

More than 30 years ago, a Swedish study found that SES gaps in scores achievement tests were larger among boys than girls (Fischbein, 1990). Consistent with this finding, several authors have proposed theoretical arguments predicting larger SES gaps in achievement among boys than girls (Connolly, 2006; Entwisle et al., 2007; Auwarter and Aruguete, 2008; Autor et al., 2019). There is some empirical support for this prediction. In addition to the Swedish study, larger SES achievement gaps among boys than girls have been observed in several studies in several Western countries: Denmark, the United Kingdom, and the United States (Entwisle et al., 2007; Penner and Paret, 2008; Mensah and Kiernan, 2010; Glaesser and Cooper, 2012; Brenøe and Lundberg, 2018; Autor et al., 2019). However, this finding was not replicated in several other studies, similarly conducted in Western countries: New Zealand, Norway, Ireland, the United Kingdom, and the United States (Gibb et al., 2008; Fryer and Levitt, 2010; Strand, 2014; Lenes et al., 2022; McGinnity et al., 2022). Thus, while theorists have predicted larger SES achievement gaps among boys than girls, empirical studies in Western countries have not robustly obtained this finding.

It is well-known that psychological and behavioral findings obtained in Western countries are not necessarily representative of humanity at large (Henrich et al., 2010). Might gender be a stronger moderator of SES achievement gaps in other parts of the world? This question is not answered in prior studies. Indeed, we only found two non-Western studies of how gender moderates the SES achievement gap (Zhu et al., 2018; Liu et al., 2022), and their findings are difficult to interpret because they used SES data reported by students. Student-reported SES data has the drawback that gender differences in SES achievement gaps are conflated with gender differences in data reporting accuracy. Indeed, we have found in ongoing work that boys tend to report SES data less accurately than girls do, causing greater attenuation of SES achievement gaps among boys so that they look narrower than they are (For this reason, we did not include a handful studies that use student-reported SES data in our above review of Western studies: Deslandes et al., 1998; Connolly, 2006; McGraw et al., 2006; Glaesser and Cooper, 2012; Contini et al., 2017).

In short, it is currently unknown whether the Western pattern of results, with no robust gender difference in the size of SES achievement gaps, holds universally or whether culture plays a role. Next, we examine prior theories in order to analyze whether their applicability may differ between cultures.

International large-scale assessments of student achievement tend to target either fourth-year students (around age 10) or adolescents (around age 14–15). Our hypothesis is applicable to both age groups. However, the quality of SES data (i.e., parents’ education and occupation) differs between the two age groups. SES data for adolescents is reported by students, and student-reported SES data is not very reliable (Lien et al., 2001; Avvisati, 2020). Moreover, the reliability of student data varies across countries (Jerrim and Micklewright, 2014), and there may well be gender differences in reliability too. To avoid these problems we will instead focus on assessments of fourth-year students, in which SES data are provided by parents. Specifically, we use data from the Progress in International Reading Literacy Study (PIRLS) and the Trends in International Mathematics and Science Study (TIMSS), which are organized by the International Association for the Evaluation of Educational Achievement (IEA). We use data from 36 countries that participated in 2016 PIRLS as well as 2019 TIMSS (the most recent waves of these assessments) and that included the SES measures in the questionnaires to parents. This study was not preregistered.

Descriptive statistics of the country-level estimates of the SES-gender interaction and, for completeness, the main effects of SES and gender are presented in Table 2. While the main effects are not our focus of interest, note that there is a robust positive effect of SES on achievement, whereas the gender effect varies across assessments between a consistent disadvantage for boys in PIRLS (i.e., in reading) and, on average, a slight advantage for boys in TIMSS (i.e., in mathematics and science). The focus of our interest is the SES-gender interaction, which is on average slightly positive. For example, when SES is measured by parents’ highest education the SES-gender interaction in PIRLS has a mean value of 2.5, while the mean value of the main effect of SES is about 10 times larger. This means that the effect of parents’ education on reading achievement is roughly 10% larger among boys than girls in the average country.

However, there is considerable variation across countries; for example, the effect of parents’ education on reading achievement is almost 300% larger among boys than girls in Saudi Arabia, which is among the least gender-equal countries in this study according to the GGGI measure (Supplementary Table S2). Table 3 reports how each of the estimated effects correlated with the level of gender equality of countries. In support of the hypothesis of the current study, gender equality is negatively correlated with SES-gender interactions (whether or not we control for the development level of countries). In other words, it is especially in less gender-equal countries that the effect of SES on achievement is larger among boys than girls. The scatterplot in Figure 1 illustrates this finding by showing how the most gender-unequal countries tend to have the largest SES-gender interaction, that is, the greatest gender difference in the SES effect. The hierarchical linear regression in Table 4 demonstrates that the level of gender equality in countries explains their SES-gender interaction above and beyond the level of development.

There are several theoretical reasons to expect SES achievement gaps to be larger among boys than girls, but prior studies have not found a robust gender difference in the size of SES achievement gaps. As prior studies were almost exclusively conducted in Western countries, we considered the possibility that different results may be obtained in other parts of the world.

From two international large-scale assessments we obtained data on SES, gender, and achievement that allowed us to estimate the SES-gender interaction in 36 countries across the world. Consistent with prior studies, estimates of the SES-gender interaction in Western countries did not have a consistent sign. However, in certain non-Western countries—like Bahrain, Qatar, Oman, Saudi Arabia, and the United Arab Emirates—the effect of SES on achievement was considerably larger among boys than girls. Moreover, these countries all have low levels of gender equality (as measured by the World Economic Forum). The moderating effect of gender equality was found regardless of whether SES was operationalized by parents’ education or parents’ occupation, and whether achievement was measured in the reading domain (PIRLS) or in the math-science domain (TIMSS).

Our interpretation of these results is that SES achievement gaps to some extent depend on how children are treated. In more gender-equal societies, boys and girls are treated more equally, which would explain why there are little gender differences in the SES achievement gaps in these societies. Unequal treatment of boys and girls may manifest in, say, differences in parental investments in boys’ and girls’ education, or differences in parental expectations on boys and girls. Prior theories of SES-gender interactions in student achievement assume that such differences in how boys and girls are treated may interact with socioeconomic differences (Connolly, 2006; Entwisle et al., 2007; Auwarter and Aruguete, 2008; Autor et al., 2019). The novelty in the current study lies in that we consider how gender differences in treatment may vary with the level of gender equality of the society.

When estimating the SES-gender interaction, we also obtained estimates of the main effects of SES and gender on achievement. Country variation in SES achievement gaps was not related to gender equality. Other studies have examined other country factors that may explain why SES achievement gaps vary in size (e.g., Marks, 2005; Bodovski et al., 2017). By contrast, there was a strong correlation in our data between the main effect of gender on achievement and gender equality; boys achieve better, relative to girls, in more gender-equal societies. For a more detailed examination of this phenomenon, see Eriksson et al. (2020).

A limitation of our study is that it only covers 36 countries. The world is large and it would be interesting to see how SES achievement gaps differ between boys and girls in, say, African countries. Another limitation is that there may be omitted variables that confound the effect of gender equality. In the current study we controlled for the overall development level of the countries, but there may be other important variables that we did not control for. To examine whether the level of equality in the treatment of boys and girls has a causal effect on gender differences in SES achievement gaps, intervention studies could be conducted. It would be valuable to know whether SES achievement gap among boys can be reduced in size through measures that address specific ways in which boys and girls are treated differently. Another possibility is to study how the SES-gender interaction varies across ethnicities in the same country (Strand, 2014), to see whether it is related to ethnic differences in the level of gender egalitarianism.

Gender equality has increased globally for many decades (Inglehart and Norris, 2003). If there is a causal connection between low gender equality and the size of the SES-gender interaction in student achievement, we should therefore expect that SES-gender interaction has decreased over time. Thus, our findings motivate future longitudinal studies of the same topic.

In conclusion, this study has documented that socioeconomic achievement gaps are often larger among boys and girls, and especially in non-Western societies with high levels of gender inequality.

The raw data used in step 1 are publicly available at IEA’s website (iea.nl). All country measures used in step 2 are available at OSF (https://osf.io/exkcm/).

Ethical approval was not required for the study involving humans in accordance with the local legislation and institutional requirements. Written informed consent to participate in this study was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and the institutional requirements.

KE conceived of the study, performed the analysis, and wrote the paper. JL conducted the survey of the literature and provided critical feedback on the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by the Swedish Research Council under Grant 2014–2008.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyg.2023.1120211/full#supplementary-material