Results from the first twenty years of the Trends in International Mathematics and Science Study (TIMSS; Mullis et al., 2016) show that more countries have increased than decreased their mean achievement in both subjects and at both grades, i.e. fourth and eighth. Another favorable trend is that the gender gap in mathematics achievement is decreasing. Meinck and Brese (2019) investigated the tails of the achievement distributions to provide a differentiated picture of potential gender differences. They found a persistent trend in the TIMSS twenty-year data of an overrepresentation of male students in the group of high-achievers in both subjects. In addition, male and female students are often unequally represented at the lower end of the ability distributions with differing patterns across countries and cycles.

However, as Charles et al. (2014) pointed out, there is an inverse relationship between students’ attitudes toward mathematics and the economic development of their country. The authors used TIMSS data from 2003 to 2011 and the Human Development Index (HDI) for national economic development. They found that attitudes toward mathematics are significantly more negative in high HDI countries; and that this result holds for both gender but it is stronger for females. Breda et al. (2020) in a recent study investigated the gender-equality paradox, i.e., that gender segregation across occupations is more pronounced in more egalitarian and more developed countries. The authors composed a measure for the internalization of the stereotype that “math is not for girls” at the country level by using data from the Program for International Student Assessment (PISA) collected in 2012. Their results show that the gender stereotypes are stronger in more egalitarian and more developed countries. Pennington and Heim (2016) recently studied the effects of stereotype threat on women’s mathematical performance and their findings show that gender-homogeneous environments may constitute a practical intervention to mitigate stereotype threat effects for females but contribute to a paradoxical effect on mindset. The authors argue for teaching students about the pervasive effects of stereotype threat and the direct influence on performance.

Against this background, the present study presents gender differences in the long-term trends of mathematics achievement and attitude, i.e. motivation for learning mathematics. We are interested in the trends in students’ motivation for learning mathematics; whether these trends are different for female and male students; if it is possible to capture students’ internalized gender stereotypes with TIMSS data; and if these stereotypes are related to achievement.

We expand the TIMSS twenty-year-trend data by including the Second International Mathematics Study (SIMS), collected between 1980 and 1982, administered by the International Association for the Evaluation of Educational Achievement (IEA). With including SIMS in the analyses, there is a unique opportunity in an international large-scale assessment (ILSA) data to explore students’ responses to explicit questions about gender stereotypes. At the same time, it is possible to explore them in relation to the endorsement of general statements like “My parents really want me to do well in mathematics”. According to Kifer and Robitaille (1989), the responses to explicit stereotypical questions reflect the culture, in which the school is located, more than any other answers on the attitude scales in SIMS. However, we need to take the effects of social desirability (Edwards, 1953) into account when looking at these items.

We address the following research questions:

1. Does gender affect the trends in students’ motivation for learning mathematics?
2. What is the relationship between gender differences in mathematics motivation and gender differences in mathematics achievement over time?
3. Can we capture students’ internalized gender stereotypes with TIMSS data?
4. Do students’ gender stereotypes influence their learning mathematics (achievement and motivation)?

We have selected six educational systems for the analyses: England, Hong Kong, Hungary, Japan, Sweden, and the United States. These countries participated at all times points except for Sweden that did not take part in 1999. We have decided to include Sweden because according to The Gender Equality Index developed by the European Institute for Gender Equality, Sweden currently is the leading country in Europe in the progress to gender equality. It is interesting to see the long-term trends of Swedish students’ attitudes towards mathematics with achievement and to compare them to those of other western and non-western cultures with lower gender equality index. We explore the relationship of intrinsic and extrinsic motivation (i.e. non-cognitive outcomes) with achievement (i.e. cognitive outcome) with focusing on gender differences. In our analyses, we used data from grade eight or equivalent. We make use of the previously established (Majoros et al., 2021) long-term scale of mathematics achievement spanning the period from 1964 (the First International Mathematics Study) to TIMSS 2015 using data from England, Israel, Japan, and the United States. We put the data of Hong Kong, Hungary, and Sweden on this scale by equating procedures at each time point. We refer to intrinsic motivation when individuals engage in an activity because they are interested in and enjoy the activity, while when individuals engage in activities for instrumental or other reasons, such as receiving a reward, they are extrinsically motivated (Eccles & Wigfield, 2002). We tested measurement invariance across countries. The linking procedures of the motivation scales were performed using IRT modeling and single-group concurrent calibration. The effect of gender on country-level trends was explored with difference-in-differences analyses. Furthermore, following the approach proposed by Meinck and Brese (2019), we explored the gender composition of the groups of students comprising the 20th highest and lowest percentiles for both types of outcomes, for each country. The previously mentioned studies highlight the gender-equality paradox in relation to economic development. Therefore, we used GDP per capita as a measure of economic development. Finally, to explore gender stereotypes we selected questionnaire items that were similar to those used in the studies of Breda et al. (2020) and Charles et al. (2014). Moreover, we used the explicit stereotypical attitude scale in SIMS. Measurement invariance across countries was tested for this scale.

We expect differing patterns across cultures over time, in terms of gender differences in both outcomes and their relationship in line with previous research (e.g. Meinck & Brese, 2019; Michaelides et al., 2019). The explicit gender stereotype scale is expected to show less distinct cultural differences because of the social desirability of the responses. We expect to face challenges with the cross-cultural measurement of non-cognitive constructs, such as method-, construct-, and item bias (Van de Vijver, 2018). In contrast to the four-point Likert scale questionnaires in the TIMSS cycles, in SIMS, the students had a fifth, neutral response option. Preliminary analyses show that the popularity of the middle option in Japan in SIMS was consistently and considerably high, indicating a possible cultural difference compared to the other countries. Measurement invariance of the attitude scales is not expected to hold for all countries. Bearing in mind that the relative proportion of females choosing a mathematical track in upper secondary and higher education is still unreasonably low and unrelated to mathematics achievement in many countries, the information gained by these analyses has nevertheless potential significance. By using comparable data from four decades, our contribution to the discourse about the gender-equality paradox and gender stereotypes may shed light on long-term processes related to gender differences in learning mathematics.

Breda, T., Jouini, E., Napp, C., & Thebault, G. (2020). Gender stereotypes can explain the gender-equality paradox. Proceedings of the National Academy of Sciences of the United States of America, 117(49), 31063–31069. https://doi.org/10.1073/pnas.2008704117 Charles, M., Harr, B., Cech, E., & Hendley, A. (2014). Who likes math where? Gender differences in eighth-graders’ attitudes around the world. International Studies in Sociology of Education, 24(1), 85–112. https://doi.org/10.1080/09620214.2014.895140 Eccles, J. S., & Wigfield, A. (2002). Motivational beliefs, values, and goals. Annual Review of Psychology, 53(1), 109–132. https://doi.org/10.1146/annurev.psych.53.100901.135153 Edwards, A. L. (1953). The relationship between the judged desirability of a trait and the probability that the trait will be endorsed. Journal of Applied Psychology, 37(2), 90–93. https://doi.org/10.1037/h0058073 Kifer, E., & Robitaille, D. F. (1989). Attitudes, preferences and opinions. In D. F. Robitaille (Ed.), International studies in educational achievement II: Contexts and outcomes of school mathematics (pp. 178–208). Oxford: Pergamon Press. Majoros, E., Rosén, M., Johansson, S., & Gustafsson, J.-E. (Accepted for publication). Measures of long-term trends in mathematics: Linking large-scale assessments over fifty years, Educational Assessment, Evaluation and Accountability Meinck, S., & Brese, F. (2019). Trends in gender gaps: using 20 years of evidence from TIMSS. Large-Scale Assessments in Education, 7(1). https://doi.org/10.1186/s40536-019-0076-3 Michaelides, M. P., Brown, G. T. L., Eklöf, H., & Papanastasiou, E. C. (2019). Motivational profiles in TIMSS mathematics, IEA Research for Education (vol.7). Springer, Cham. https://doi.org/10.1007/978-3-030-26183-2 Mullis, I.V.S., Martin, M.O., & Loveless, T. (2016). 20 Years of TIMSS: International Trends in Mathematics and Science Achievement, Curriculum, and Instruction. Chestnut Hill, MA: Boston College. Pennington, C. R., & Heim, D. (2016). Creating a critical mass eliminates the effects of stereotype threat on women's mathematical performance. The British Journal of Educational Psychology, 86(3), 353–368. https://doi.org/10.1111/bjep.12110 Van de Vijver, F. J. R. (2018). Towards an Integrated Framework of Bias in Non-cognitive Assessment in International Large-Scale Studies: Challenges and Prospects. Educational Measurement: Issues and Practice, 37(4), 49-56. https://doi.org/10.1111/emip.12227