What Facebook can tell us about the preference differences between women and men

Ángel Cuevas Rumin, Ruben Cuevas Rumin, Klaus Desmet, Ignacio Ortuño-Ortin 08 January 2022

a

A

Do preference differences between men and women become attenuated or accentuated in more gender-equal societies? The answer to this question might seem straightforward: equal opportunities and equal access to resources should make men and women more similar in their preferences. However, the answer turns out to be more complex. For example, the gender gap in engineering is particularly large in some of the world’s most gender-equal countries, such as Finland and Sweden (Stoet and Geary 2018). Of course, this so-called gender-equality paradox need not generalise to the universe of preferences, from politics and religiosity to travelling, fitness, and movies. 

In fact, there are two theories on the question of gender equality and preference differences between men and women. On the one hand, evolutionary psychology posits that gender equality allows everyone to be their true self, so that both women and men can more freely express their innate predispositions. This tends to amplify differences between men and women. On the other hand, social role theory posits that men and women differ in their preferences mostly because of stereotypes, socialisation and discrimination. With gender equality removing those stereotypes, the gender gap tends to narrow, making men and women more similar in their interests and preferences. 

Although these theories point in opposite directions, they need not be mutually exclusive. Evolutionary psychology applies mostly to preferences that are innate, whereas social role theory applies mostly to preferences that are socially constructed. If we could find a way to distinguish between both types of preferences, we could test whether both theories might be correct. In contrast to most studies that have focused on very specific preferences (Falk and Harmle 2018), this would require comprehensive data on a broad set of preferences. 

Facebook: The world’s largest database on preferences

In a recent paper (Cuevas et al. 2021), we use Facebook’s Marketing application programming interface (API) to get information on the shares of male and female Facebook users that are interested in 45,397 different topics, and we do this for all countries of the world. For example, this tells us how many men and how many women are interested in Lionel Messi in Nigeria, or how many men and how many women are interested in poetry in Bangladesh. Because of the large number of interests, it covers almost everything, from cuisine, psychology, and romantic comedies to football, family, and organic food. 

How does Facebook identify its users’ interests and preferences? Facebook not only observes what you explicitly like, but also what you read, share, and download. And it does so not just on its own platform, but also on all of the world’s webpages where it has a presence. In addition, Facebook also observes much of your offline activity, as long as it has access to your GPS location. If you go jogging every day, visit the pub every night, or attend church on Sundays, Facebook is likely to know. In that sense, Facebook plays the role of the ethnographer, but on a massive scale: it peers into the lives of billions of people, and it unobtrusively observes what interests them. 

When assessing how reliable this information is, it is useful to remember that Facebook’s business model depends on making correct inferences about its users’ true preferences. This allows Facebook to show you more relevant posts, thus incentivising you to spend more time on the social network. This ultimately translates into Facebook being able to show you more ads. As such, Facebook’s bottom line depends crucially on correctly identifying its users’ true preferences. 

Consistent with this, in related work we have shown that distances between countries based on Facebook interests correlate strongly with distances between countries based on surveys that enquire about people’s values, norms, attitudes, and preferences (Obradovich et al. 2020). 

Facebook has the additional advantage of taking a bottom-up revealed preferences approach: it identifies whatever people find interesting, rather than what the social scientist who designs surveys might find interesting. Again, this is similar to the ethnographer’s approach: Facebook observes people go about their daily lives, without forming any prejudices about what it is important and what is not. 

Gender-related and non-gender-related preferences

With data on 45,397 interests for men and women across most countries of the globe, we distinguish between two types of interests. We call an interest ‘gender-related’ if it exhibits the same gender bias across the globe. For example, in essentially all countries of the world, men are more interested in cars, video games, and football, and women are more interested in spas, romance novels, and children. And we call an interest ‘non-gender-related’ if it does not show the same gender bias across the globe. For example, travelling, horses and eBooks are more prevalent among men in some countries, and more popular among women in other countries.  

Once we distinguish between these two types of interests, we find contrasting results in their relation to gender equality. For gender-related interests, such as football or children, the difference between men and women is greater in more gender-equal countries. For non-gender-related interests, such as travel and fitness, the opposite is true: men and women are more alike in more gender-equal countries. Because different interests may sometimes reflect the same underlying preferences, we use singular value decomposition – a technique similar to principal component analysis – to identify the main latent preference dimensions. When differentiating between the main gender and non-gender dimensions of preferences, we confirm our findings: more gender-equal societies display greater differences between women and men in gender-related preferences but smaller differences in non-gender-related preferences. 

Evolutionary psychology and social role theory

How are these results related to what we know from evolutionary psychology and social role theory? Recall that the former applies to innate preferences, whereas the latter applies to socially constructed preferences. There is of course no foolproof way to differentiate innate preferences from socially constructed preferences. However, we take the view that for an interest to be innate, it is reasonable to expect that it should exhibit the same gender bias across the globe. Suppose an interest, such as travelling, is more popular among men in some countries and more popular among women in other countries. In that case, it is unlikely to be innate. However, suppose an interest, such as sports or war, is more popular among men the globe over. In that case, it has the potential to be innate. 

Adopting this interpretation, our results concur with both theories. For potentially innate preferences (i.e. those with the same gender bias everywhere, such as sports and war), men and women are more different in more gender-equal societies. This is precisely what evolutionary psychology predicts. But for non-gender-related interests, such as travelling, horses, and eBooks, the opposite holds. This is precisely what social role theory predicts.

A word of caution is needed here. We refer to gender-related interests as potentially innate, because we cannot discard the possibility that some gender-related interests are socially constructed. Take the example of the stronger interest of males in football the world over. This male bias might be due to hormonal differences that make men more interested in anything to do with competition, including football. In that case, the preference would be innate. However, it is also possible that the male love for football was socially constructed in some societies, to then conquer the world through cultural diffusion. That said, it is likely that at least a subset of gender-related preferences is effectively innate, whereas non-gender-related preferences are most likely not innate. 

Gender policy

Do these findings have any policy implications? Policies that reduce barriers and provide equal opportunities to men and women are likely to improve welfare, partly because it allows everyone to live up to their own interests and aspirations and partly because it leads to a better allocation of talent (Hsieh et al. 2019, Chiplunkar and Goldberg 2021). Lowering barriers is good, independently of whether preferences are socially constructed or innate. When it comes to more specific policies, such as gender quotas, the distinction between different types of preferences becomes more relevant (Bagues and Campa 2017). If differences are socially constructed, it would seem desirable to have as many female computer scientists as male, and as many male physicians as female. In contrast, if differences are innate, it becomes less obvious that complete gender equality is what we should pursue. 

References

Bagues, M and P Campa (2017), “Electoral gender quotas fail to empower women”, VoxEU.org, 9 September.

Chiplunkar, G and P Goldberg (2021), “Aggregate implications of barriers to female entrepreneurship”, VoxEU.org, 19 April. 

Cuevas, A, R Cuevas, K Desmet, and I Ortuño-Ortín (2021), “The gender gap in preferences: Evidence from 45,397 Facebook interests,” CEPR Discussion Paper No. 16740. 

Falk, A and Hermle, J (2018), “Relationship of gender differences in preferences to economic development and gender equality”, Science 362(6412): eaas9899.

Hsieh, C-T, E Hurst, C I Jones, and P J Klenow (2019), “The allocation of talent and U.S. economic growth”, Econometrica 87(5): 1439-1474.

Obradovich, N, Ö Özak,, I Martín, I Ortuño-Ortín, E Awad, M Cebrián, R Cuevas, K Desmet, I Rahwan, and A Cuevas (2020), “Expanding the measurement of culture with a sample of two billion humans”, CEPR Discussion Paper No. 15315. 

Stoet, G and D C Geary (2018), “The gender-equality paradox in science, technology, engineering, and mathematics education”, Psychological Science 29(4): 581-593.

a

A

Topics:  Gender

Tags:  gender, Gender preferences, Facebook, social role theory, systemic gender bias

Ramón y Cajal Fellow, Department of Telematic Engineering, University Carlos III de Madrid

Associate Professor, Telematic Engineering Department, University Carlos III de Madrid

Altshuler Professor of Cities, Regions and Globalisation, SMU; CEPR Research Fellow; NBER Research Associate

Professor of Economics, Universidad Carlos III de Madrid

Events

CEPR Policy Research