^{1}

^{2}

^{*}

^{1}

^{2}

Edited by: Ronnel B. King, University of Macau, China

Reviewed by: Manuel Soriano-Ferrer, University of Valencia, Spain; Hanke Korpershoek, University of Groningen, Netherlands

This article was submitted to Educational Psychology, a section of the journal Frontiers in Psychology

This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

Although much research has found girls to be less interested in mathematics than boys are, there are many countries in which the opposite holds. I hypothesize that variation in gender differences in interest are driven by a complex process in which national culture promoting high math achievement drives down interest in math schoolwork, with the effect being amplified among girls due to their higher conformity to peer influence. Predictions from this theory were tested in a study of data on more than 500,000 grade 8 students in 50 countries from the 2011 and 2015 waves of TIMSS. Consistent with predictions, national achievement levels were strongly negatively correlated with national levels of math schoolwork interest and this variation was larger among girls: girls in low-achievement, high-interest countries had especially high interest in math schoolwork, whereas girls in high-achievement, low-interest countries had especially low interest in math schoolwork. Gender differences in math schoolwork interest were also found to be related to gender differences in math achievement, emphasizing the importance of understanding them better.

Children and young adolescents are typically obliged to go to school and must take part in schoolwork even if they do not find it interesting. Nonetheless, it is preferable that students are interested in their schoolwork, both because they are likely to experience more satisfaction in school and because they are likely to achieve better (e.g.,

The theoretical idea I propose is that gender differences in interest in schoolwork may be influenced by a societal factor—the achievement culture, which tends to drive interest in schoolwork down—in combination with a gender difference in conformity, with girls tending to conform more than boys. The outcome, I argue, would be a specific, complex pattern. In high-achievement cultures, it would be common for students to have a low level of interest in math schoolwork and, due to conformity, a low level of interest would be especially common among girls. In low-achievement cultures, by contrast, it would be common for students to have a high level of interest in math schoolwork and, again due to conformity, a high level of interest would be especially common among girls. Thus, high-achievement cultures would exhibit gender gaps in math schoolwork interest that favor boys, while gender gaps would be reversed in low-achievement cultures. Below I develop this novel hypothesis in greater detail, grounding its assumptions in previous literature.

The achievement culture of a society may be an important factor behind how interested students are in mathematics schoolwork. When comparing across countries, it is well-known that a high average level of student achievement in mathematics and science is related to a range of negative outcomes, including more negative attitudes to math and science (

The hypothesized path from high-achievement culture to gender differences in interest in math schoolwork. The first arrow represents the hypothesis that high-achievement nations will have lower average student interest in math schoolwork. The second (bold) arrow represents the hypothesis that differences in average national interest will be amplified among girls.

It is well-known that students’ motivation may be heavily influenced by their peers, both positively and negatively, and both intentionally and unintentionally (

Based on this previous literature I expect that within-society conformity with respect to interest in math schoolwork will be more accentuated among women than among men (through mechanisms such as gender differences in confidence, risk aversion, and friendships). However, the focus of the present research is on

This hypothesis is illustrated in the diagram in

Gender differences in math schoolwork interest are important not least because they are likely to impact on the math achievement of boys and girls. At the individual level, interest in math schoolwork is thought to be conducive to learning (

Above I have outlined a theory about antecedents and consequences of societal levels of math schoolwork interest among girls and boys. To fully test claims of causality would require experimental or, at least, longitudinal data, neither of which are available. Instead, I here make do with analyzing cross-sectional data provided by TIMSS. The theory predicts certain statistical patterns to arise in such data and the aim of the empirical part of this study is to examine whether these patterns can indeed be observed. It is an important first test of the theory to see whether it correctly predicts several non-trivial features of a complex dataset, even though alternative causal accounts cannot be excluded.

RQ1. The hypothesis of high-achievement culture impacting on students’ math schoolwork interest yields the first prediction to be examined: Is there a negative correlation between national levels of achievement and math schoolwork interest?

RQ2. The hypothesis of a difference between boys and girls in peer influence on math schoolwork interest yields a suite of testable predictions: (a) Is within-society variation in math schoolwork interest smaller among girls than among boys? (b) Is between-society variation in math schoolwork interest larger among girls than among boys? (c) Is there a positive correlation between national levels of math schoolwork interest and gender gaps in math schoolwork interest favoring girls? (d) Do national levels of math schoolwork interest mediate a negative correlation between national levels of achievement and gender gaps in math schoolwork interest favoring girls?

RQ3. The hypothesis that gender differences in math schoolwork interest has an independent impact on the gender gap in math achievement also yields a testable prediction: Does the gender gap in math schoolwork interest account for some of the variance in the gender gap in math achievement, over and beyond the variation already accounted for by gender egalitarian values?

To answer the research questions, the current study analyzes TIMSS data. TIMSS is an excellent resource for comparative research as it uses large representative national samples of students from many countries. Details on the design are provided by the International Association for the Evaluation of Educational Achievement (

Data from the 2011 and 2015 waves of TIMSS were downloaded from IEA^{1}. Data were available for a total of 50 countries, out of which 35 countries had participated in both waves, 10 countries had participated only in the 2011 wave, and 5 countries only in the 2015 wave. See

TIMSS sample sizes and key measures.

Country | 2011 |
2015 |
Math schoolwork interest |
Mean math |
Gender |
||||

Sample size | Sample size | Girls |
Boys |
Achievement |
Egalitarian values | ||||

M | SD | M | SD | Girls | Boys | ||||

Armenia | 7,556 | 10,338 | 3.52 | 0.63 | 3.45 | 0.69 | 472.8 | 464.8 | −0.91 |

Australia | 4,640 | 4,918 | 2.70 | 0.79 | 2.79 | 0.78 | 502.6 | 508.2 | 1.09 |

Bahrain | 5,846 | 5,060 | 3.08 | 0.76 | 3.05 | 0.84 | 446.4 | 417.3 | |

Botswana | 5,400 | 5,964 | 3.36 | 0.68 | 3.31 | 0.71 | 401.8 | 385.4 | |

Canada | 8,757 | 2.97 | 0.74 | 3.01 | 0.74 | 525.8 | 530.8 | 1.31 | |

Chile | 5,835 | 4,849 | 3.07 | 0.78 | 3.12 | 0.79 | 414.0 | 430.1 | 0.31 |

Chinese Taipei | 5,042 | 5,711 | 2.41 | 0.74 | 2.51 | 0.81 | 605.8 | 602.7 | |

Egypt | 4,035 | 3.50 | 0.69 | 3.43 | 0.74 | 396.6 | 387.4 | −1.87 | |

Finland | 4,266 | 2.42 | 0.72 | 2.40 | 0.73 | 516.6 | 512.1 | 0.62 | |

Georgia | 4,563 | 4,155 | 3.39 | 0.65 | 3.33 | 0.70 | 441.7 | 442.7 | −0.85 |

Ghana | 7,812 | 3.53 | 0.58 | 3.58 | 0.55 | 318.7 | 342.2 | −0.65 | |

Honduras | 7,323 | 3.53 | 0.65 | 3.51 | 0.68 | 327.8 | 351.0 | ||

Hong Kong | 4,418 | 4,893 | 2.58 | 0.75 | 2.73 | 0.81 | 590.2 | 590.0 | |

Hungary | 4,015 | 6,130 | 2.74 | 0.77 | 2.74 | 0.81 | 505.8 | 513.3 | 0.52 |

Indonesia | 5,178 | 3.14 | 0.44 | 3.10 | 0.48 | 392.4 | 379.5 | −0.56 | |

Iran | 5,795 | 4,704 | 3.16 | 0.75 | 3.24 | 0.76 | 424.6 | 426.5 | −1.10 |

Ireland | 5,512 | 2.75 | 0.81 | 2.80 | 0.82 | 520.8 | 526.3 | 1.43 | |

Israel | 6,029 | 4,481 | 2.91 | 0.82 | 2.88 | 0.85 | 515.1 | 512.6 | 0.68 |

Italy | 4,699 | 4,745 | 2.82 | 0.70 | 2.85 | 0.74 | 491.9 | 500.5 | 0.87 |

Japan | 3,979 | 4,887 | 2.26 | 0.66 | 2.39 | 0.72 | 576.8 | 578.4 | −0.13 |

Jordan | 4,414 | 7,865 | 3.47 | 0.65 | 3.43 | 0.71 | 407.6 | 384.3 | −1.55 |

Kazakhstan | 4,390 | 5,309 | 3.42 | 0.55 | 3.31 | 0.58 | 508.6 | 506.2 | 0.13 |

Korea | 7,694 | 4,503 | 2.28 | 0.65 | 2.38 | 0.71 | 607.4 | 611.3 | −0.23 |

Kuwait | 3,873 | 3.01 | 0.79 | 3.23 | 0.76 | 396.0 | 388.7 | −1.39 | |

Lebanon | 5,166 | 4,347 | 3.30 | 0.76 | 3.30 | 0.76 | 442.3 | 449.9 | −0.32 |

Lithuania | 3,974 | 9,726 | 2.91 | 0.73 | 2.92 | 0.75 | 508.5 | 505.3 | −0.09 |

Macedonia | 4,747 | 3.23 | 0.79 | 3.23 | 0.78 | 429.6 | 422.6 | 0.09 | |

Malaysia | 5,733 | 3,817 | 3.18 | 0.63 | 3.04 | 0.67 | 459.5 | 445.5 | −0.98 |

Malta | 13,035 | 2.80 | 0.81 | 2.93 | 0.83 | 495.1 | 492.7 | ||

Morocco | 8,986 | 8,883 | 3.49 | 0.64 | 3.46 | 0.66 | 378.5 | 377.8 | −0.97 |

New Zealand | 9,542 | 8,142 | 2.72 | 0.78 | 2.85 | 0.77 | 486.8 | 494.6 | 0.49 |

Norway | 5,336 | 4,697 | 2.70 | 0.77 | 2.78 | 0.78 | 493.6 | 493.1 | 1.90 |

Oman | 3,862 | 5,403 | 3.52 | 0.56 | 3.31 | 0.71 | 408.4 | 361.0 | |

Palestine | 4,422 | 3.57 | 0.57 | 3.38 | 0.72 | 415.3 | 392.2 | ||

Qatar | 5,523 | 4,780 | 3.01 | 0.81 | 3.08 | 0.83 | 427.8 | 419.1 | −1.48 |

Romania | 4,893 | 3.05 | 0.79 | 2.98 | 0.81 | 463.7 | 452.8 | −0.13 | |

Russia | 4,344 | 3,759 | 3.09 | 0.69 | 3.09 | 0.70 | 535.9 | 541.0 | −0.30 |

Saudi Arabia | 5,927 | 6,116 | 3.21 | 0.74 | 3.16 | 0.81 | 388.1 | 373.5 | −1.79 |

Singapore | 4,415 | 4,257 | 2.89 | 0.68 | 2.94 | 0.72 | 620.5 | 611.7 | 0.20 |

Slovenia | 11,969 | 12,514 | 2.55 | 0.69 | 2.55 | 0.75 | 508.8 | 512.4 | 0.86 |

South Africa | 5,573 | 4,090 | 3.41 | 0.65 | 3.38 | 0.68 | 364.8 | 359.5 | 0.04 |

Sweden | 4,413 | 6,482 | 2.61 | 0.71 | 2.73 | 0.74 | 492.5 | 493.8 | 1.66 |

Syria | 6,124 | 3.55 | 0.60 | 3.48 | 0.69 | 374.9 | 384.7 | ||

Thailand | 14,089 | 18,012 | 3.29 | 0.55 | 3.24 | 0.60 | 437.3 | 419.7 | −0.20 |

Tunisia | 5,128 | 3.47 | 0.65 | 3.43 | 0.68 | 416.8 | 433.6 | ||

Turkey | 6,928 | 6,079 | 3.11 | 0.65 | 3.06 | 0.71 | 459.0 | 451.6 | −0.17 |

UAE | 3,378 | 7,822 | 3.15 | 0.73 | 3.15 | 0.77 | 467.7 | 453.1 | |

United Kingdom | 4,062 | 10,221 | 2.71 | 0.76 | 2.83 | 0.77 | 514.7 | 511.5 | 1.25 |

Ukraine | 10,477 | 3.30 | 0.66 | 3.27 | 0.69 | 477.6 | 481.0 | −0.25 | |

United States | 3,842 | 4,814 | 2.80 | 0.83 | 2.83 | 0.83 | 512.9 | 515.5 | 1.18 |

The TIMSS datasets come with appropriate sampling weights, which were used when calculating the below measures. Missing data (less than 3% of data) were ignored. Preliminary analyses revealed that country measures were highly consistent across the two waves. For the below analysis we therefore pooled the individual data from the two waves.

The student questionnaire in the 2011 and 2015 waves of TIMSS included three items bearing explicitly on interest in math schoolwork: “I am interested in what my teacher says,” “My teacher gives me interesting things to do,” and “I learn many interesting things in mathematics.” For each item, students gave their response on a four-point scale:

TIMSS provides ready calculated national average scores for girls’ and boys’ math achievement, which were downloaded using the International Data Explorer of the National Center for Education Statistics^{2} (see

Following

Country levels of math schoolwork interest, general affect toward math, and math achievement, were calculated as the averages of the corresponding mean values for girls and boys. Gender differences for the same variables were similarly calculated as the mean value for girls minus the mean value for boys.

Descriptive statistics for country levels and gender differences of the measures in

Variable | Min | Max | ||

Country level of math schoolwork interest | 3.06 | 0.34 | 2.33 | 3.55 |

Country level of math achievement | 464.3 | 70.3 | 330 | 616 |

Gender difference in math schoolwork interest | −0.01 | 0.09 | −0.22 | 0.20 |

Gender difference in math achievement | 2.9 | 12.7 | −24 | 47 |

The first research question concerns the prediction of a negative correlation between country levels of achievement and math schoolwork interest. In line with the prediction, a very strong negative correlation was observed,

In

^{2} = 0.133, which is 36% higher than the corresponding country variance among boys, σ^{2} = 0.098. To estimate the statistical significance, we use the Morgan-Pitman test for difference in variance in paired data. This test assumes normally distributed data, and Kolmogorov-Smirnov tests indicated that the country level data on math schoolwork interest did not deviate from a normal distribution either for girls or boys,

Country variation in math schoolwork interest is amplified among girls. In high-interest countries, the interest levels of girls (black dots, solid line) tend to be higher than the interest levels of boys (circles, dashed regression line), while in low-interest countries girls tend to have lower interest levels than boys do.

In line with the prediction, there was a strong positive correlation between the total country level of math schoolwork interest and the gender difference in math schoolwork interest,

Gender differences (favoring girls) in math schoolwork interest correlate with the national level of interest. The Pearson correlation is

In line with the prediction, there was a negative correlation between national levels of math achievement and gender differences in math schoolwork interest,

Mediation diagram. The negative correlation between the national math achievement level and the gender difference in math schoolwork interest was mediated by the national level of math schoolwork interest. All variables were standardized prior to the mediation analysis. Bootstrapped 95% confidence intervals in brackets.

There was a positive correlation between gender differences in math schoolwork interest and gender differences in math achievement,

Consistent with prior research (

The present paper studied the difference between girls and boys in their interest in math schoolwork and how it varies across countries. A theory was proposed according to which national culture promoting high math achievement drives down interest in math schoolwork, but more among girls than among boys due to conformity to peer influence being stronger among girls. Moreover, I argued that gender differences in math schoolwork interest are important because they will contribute to gender differences in math achievement. In the absence of experimental data, I tested the predictions this theory makes about statistical observations in cross-sectional data, provided by TIMSS. Results were consistent with predictions, as detailed below.

First, an extremely strong negative correlation between national levels of achievement and math schoolwork interest was observed. This finding, which is well in line with prior research on the relation between national achievement levels and attitudes to math and science (

Second, several findings were consistent with the hypothesis that conformity to peer influence on math schoolwork interest is higher among girls than among boys. In almost all countries in the study, within-society variation in math schoolwork interest was smaller among girls than among boys, thus indicating greater female conformity. Because societies vary in their average level of interest in math schoolwork, the effect of peer pressure will vary too. Consistent with greater susceptibility to peer influence among girls, between-society variation in math schoolwork interest was larger among girls than among boys (

Taken together, my theory proposes a pathway in which high-achievement culture drives down schoolwork interest, which through differential peer influence creates gender gaps in interest disfavoring girls. Consistent with this pathway, I found a negative correlation between national levels of achievement and gender gaps in math schoolwork interest favoring girls, and this correlation was mediated by the national level of math schoolwork interest.

Why is it important how girls’ and boys’ interest in math schoolwork vary across countries? For one thing, it is theoretically important to realize that the variation is substantial. In countries like Japan, Hong Kong, Sweden, and New Zealand, the interest level of the average girl was about 0.2 standard deviations lower than the interest of the average boy. These findings are consistent with research arguing for a fundamental gender difference in subject interest (e.g.,

Gender differences in math interest may also have real-life implications by influencing how girls achieve in mathematics relative to boys in the same country. Consistent with this hypothesis, I found that variation in the gender gap in math schoolwork interest accounts for part of the proportion of variance in the gender gap in math achievement that is not explained by variation in gender egalitarian values (

This study is an example of the benefits of using big data from large-scale assessments of student achievement to examine phenomena in educational psychology. A limitation, inherent in the reliance on cross-sectional data, is that directions of causality are not established. The findings are consistent with the proposed theory, but they could also have arisen from other mechanisms. It is helpful to consider what these alternative mechanisms could be. With respect to the strong negative correlation between national levels of achievement and schoolwork interest, it seems implausible that it would arise from low interest levels having a positive effect on achievement levels. Following

Similar reasoning applies to the amplification among girls of national variation in schoolwork interest. I proposed that this arises from differential conformity to peer influence, but it cannot be excluded that there is some alternative societal factor that causes girls’ interest levels to be more extreme than the interest levels of boys. An interesting possibility for future research would be for large-scale assessments to provide some direct measures of peer influence (see also

The idea of conceiving of high-achievement culture as a factor behind gender differences has a precedent. In a study of PISA data,

To conclude, the present study has contributed to scientific understanding of gender differences in interest in mathematics schoolwork by, first, proposing a theory of why such gender differences would arise and vary across countries, and second, testing several theoretical predictions in a large cross-national dataset. Results were consistent with both key components of the theory: high-achievement culture may be detrimental to interest in schoolwork and this effect may be amplified among girls due to their higher conformity to peer influence. These positive findings motivate further study of the validity and scope of the proposed mechanisms.

Publicly available datasets were analyzed in this study. This data can be found here:

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

KE performed the statistical analysis and wrote the manuscript.

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

Ksenia Startseva-Lora provided helpful preliminary analyses.