Cognitive style influences entry into the sciences and humanities
Updated: Dec 30, 2018
Prior literature has suggested that females tend towards empathy while males towards systemic behavior. However, Billington et al. (2007) proposed that these differences arise due to variation in empathetic/systematic behavior, rather than pure gender differences in alone. Being one of a handful of studies that suggested the role in alternative factors in subject selection, it was of interest to determine the extent to which cognitive approach influences subject choices in a global, high-school setting.
Aim: To determine whether empathy levels correlate with one's gender and/or subject choices.
Method: A cross-cultural, etic questionnaire was sent out as a Google Form through Quora, Reddit, Instagram, Twitter and various Facebook groups to International Baccalaureate (IB) or prospective IB students. The questionnaire collected one's grade (out of grade 9, 10, 11 or 12 in high school), gender, and current or intended IB Higher Level (HL) subjects. Subjects' grade was collected in order to reduce demand characteristics. Participants were then asked to take the Empathy Quotient (EQ) Questionnaire as developed by Dr. Baron-Cohen at the Autism Research Center, and input their EQ score on the form.
A total of 481 respondents answered. Among these, answers that did not fall between the EQ score range of 0-80 were removed, leaving 474 respondents. For gender analysis, answers that did not identify as either male or female were excluded. For subject-choice analysis, respondents' IBHL subject choices were used to categorize them as either 'science' (s) or the 'humanities (h). Among their HLs, the majority that lied in either the sciences or the humanities was used to classify them accordingly. The definition of 'sciences' and 'humanities' followed the categorization as used by the IB. 'Languages', 'The Arts' and 'Individuals and Societies' (formerly known as 'humanities) fell under 'humanities', while 'Science' and 'Mathematics' characterized the 'science' classification.
Following this, a T-test was used to calculate the significance difference in EQs between males/females, and between the science/humanities. A T-test was used because the sample mean and variance was unknown: Thus, analysis looked for a significant difference between the two groups.
Results: Quantitative analysis revealed a significant difference (p < 0.00001) between males (mean EQ = 37.61) and females (mean = 43.46) in EQ. This supports prior research that females tend to have higher empathy than males. More interestingly, 'science' students had lower EQs (mean = 39.10) as opposed to 'humanities' students (mean = 42.75). These results were also statistically significant (p = 0.12865), although less significant as compared to gender. Using the chi-squared test for correlation between two groups, there was statistically significant (p=0.016141) evidence suggesting that males tended to study the sciences while females tend to pursue the humanities. There was also a significant indication (p=0.02595) that people with a lower EQ pursued the sciences, and vice versa. Qualitatively, it was noted that more females responded to the survey as males (253 female vs 214 male respondents, and 7 identifying as 'other').
Conclusion: These results provide support to Billington's hypothesis that one's cognitive style (empathetic versus systematic) may be a predictor for future subject choice. However, gender and subject-choice was more strongly correlated than EQ and subject choice, indicating that gender still played a large role, thus disagreeing with Billington's findings. Furthermore, gender and EQ was more strongly correlated than subject-choice and EQ, suggesting that gender (as opposed to subject choices) more strongly influences empathy. On a side note, was interesting to see that the higher responsiveness of females to the questionnaire could be an indication of greater agreeableness in women, a trait noted in several previous studies.
Further Steps: It would be of interest to use Billington et al. (2007)'s analysis and classify students into 'Extreme science' and 'Extreme humanities' profiles, to see the extent to which subject-choice and empathy correlate. To test the predictability of EQ as opposed to gender in predicting subject choices, a predictive logarithmic/linear regression model for each factor could be created and their accuracies compared.
Discussion: The differences in empathy can be explained via evolutionary means, as females consistently maintained the role of child-rearers and caregivers. This personality trait reveals to contribute to the gender-imbalance in STEM. Is such a characteristic of empathy inborn between boys and girls, or acquired due to childhood exposure and parental upbringing? Interestingly (and rather prescriptively), research has shown that 3-year old boys tended to outperform girls in a spatial/analytical task, but girls attained the same level when given hints. Moreover, there is evidence to suggest genetic bases for differences in empathy. This suggests that personality differences are developed at an incredibly early age, or possibly even inborn. Nevertheless, more research is required to clearly elucidate the role of cognitive tendencies in one's career path.
As all correlational studies, one must be careful before assigning deterministic or causation of one attribute to an other. It is, theoretically, equally likely that a science student developed a higher EQ as a result of his/her study. This could be further tested by analyzing whether Grade 12 students had more extreme EQs as opposed to Grade 9/10 students. In addition to this warning, one must always be aware of confounding factors - cultural dimensions, family environment, and possibly age. However, students' educational environment and degree type were well controlled. Additionally, it could be speculated that the higher female response-rate is due to researcher effects (the experimenter was female), however the effect of this would be minimal because a majority of the responses came from an anonymous forum (Reddit). Moreover, there is always the possibility of bias in self-report through questionnaires, but the large participant sample, as well as careful analysis to exclude suspicious data, should largely minimize their effects on the results.