Background:
In the 1970s:
- Masculinity and femininity were conceptualized as opposites on a scale.
- Having a personality not matching your gender was considered unhealthy.
Psychologist Sandra Bem thought it was possible to be both masculine and feminine and that it was healthy to express both.
She made the Bem Sex Role Inventory to study gender differences in personality. Stereotypical gender traits and items were found by surveying Stanford Students.
The Open Sex-Role Inventory (OSRI) is an updated alternative to Bem’s test with items like:
- I take lots of pictures of my activities
- I have played a lot of video games
Also, shyness and being un-athletic are no longer seen as female traits and there is a third option for gender.
Personally, this interested me because growing up, my parents always chided me for being ‘too sensitive’ and not interested in stereotypical boy things.
Some Quick Definitions
Items can depict either masculinity or femininity:
- Masculine items depict “assertive-dominance” and “instrumentality”
- Feminine items depict “expressiveness” and “nurturing-interpersonal warmth”
Participants can have one of four categorizations:
Fun fact:
The Dataset
The OSRI dataset can be found here. It has 316,256 responses to 44 beta items that participants rated 1–5.
Participants also gave demographic data on:
Age, gender, education, orientation, race, religion, English proficiency and which hand they used for writing.
Research Question
What differences are there in the percentage of categorizations based on gender? And by proxy, which genders are healthier? Assuming expressing both feminine and masculine traits is healthy.
Methodology
I started by cleaning the data and then organizing the average responses by gender which left us with 314,005 responses. The OSRI 1.0 only used 20 of the 44 beta items.
These are the masculine item questions and their mean responses.
Masculine Sex-Role
I have daydreamed about saving someone from a burning building.
I think a natural disaster would be kind of exciting.
I have thrown knives, axes or other sharp things.
I like guns.
I have considered joining the military.
I have taken apart machines just to see how they work.
I have set fuels, aerosols or other chemicals on fire, just for fun.
I have been very interested in historical wars.
I have studied how to win at gambling.
I have burned things up with a magnifying glass.
There isn’t as much of a difference between genders as I thought. Other is closer to male for items 1,3,6 and 10 but closer female for 4,5, 8 and 9
And for the feminine items and responses.
Feminine Sex-Role
I like dancing.
I give people handmade gifts.
I leave nice notes for people now and then.
I take lots of pictures of my activities.
I bake sweets just for myself sometimes.
I decorate my things (e.g. stickers on a laptop).
I jump up and down in excitement sometimes.
I wear a blanket around the house.
I have kept a personal journal.
I think horoscopes are fun.
Other very closely follow Female for all items except maybe F1 and F4. Also, there is a much wider gap between male and female ratings compared to the Male Items.
I used normalized heat maps to see these distributions against each other. Here’s the combined heatmap.
And here are the heatmaps by genders.
Nothing too surprising given the previous bar charts.
Conclusion
There are more androgynous Female and Other. This makes sense since Males are less likely to express feminine traits than vice versa.
Other’s distribution is reasonably similar to the female distribution.
I would have expected more Other gender who are above average in masculinity and average in femininity given that there are fewer in the top middle compared to Female.
To conclude, there are differences, mostly accentuated by Males being comparatively less comfortable expressing feminine traits.
Assuming it is healthy to express both and that results in the top right quadrant are healthier than women are men as on average.
There are lots of studies that show men have higher suicide rates, lower life expectancies, fewer friends (at a glance, people high in feminine traits seem more pleasant to be around) so this difference in gender expression might be part of it.
Further Research
This dataset can answer tons of questions with its demographic data. For example, here is a rough chart showing average masculinity/femininity by age.
A sharp decline coincides with people born in the 70s when Bem started challenging the status quo about masculinity and femininity. Given the background, in retrospect, this would have made more sense to explore first.
This project was done as part of my first sprint as part of Lambda School Data Science Cohort 10. The notebook can be viewed here. Feel free to make a copy and play around with it yourself.
