Pathogen prevalence is associated with cultural changes in gender equality

  • Nature Human Behaviour 1, Article number: 0003 (2016)
  • doi:10.1038/s41562-016-0003
  • Download Citation
Published online:


Gender equality has varied across time, with dramatic shifts in countries such as the United States in the past several decades. Although differences across societies and changes within societies in gender equality have been well documented, the causes of these changes remain poorly understood. Scholars have posited that such shifts have been driven by specific events (such as Title IX and Roe versus Wade), broader social movements (such as feminism and women’s liberation) or general levels of social development (for example, modernization theory1). Although these factors are likely to have been partly responsible for temporal variations in gender equality, they provide fairly intermediate explanations void of a comprehensive framework. Here, we use an ecological framework to explore the role of key ecological dimensions on change in gender equality over time. We focus on four key types of ecological threats/affordances that have previously been linked to cultural variations in human behaviour as potential explanations for cultural change in gender equality: infectious disease, resource scarcity, warfare and climatic stress. We show that decreases in pathogen prevalence in the United States over six decades (1951–2013) are linked to reductions in gender inequality and that such shifts in rates of infectious disease precede shifts in gender inequality. Results were robust, holding when we controlled for other ecological dimensions and for collectivism and conservative ideological identification (indicators of more broadly traditional cultural norms and attitudes). Furthermore, the effects were partially mediated by reduced teenage birth rates (a sign that people are adopting slower life history strategies), suggesting that life history strategies statistically account for the relationship between pathogen prevalence and gender inequality over time. Finally, we replicated our key effects in a different society, using comparable data from the United Kingdom over a period of seven decades (1945–2014).

There is a growing view in the behavioural sciences that the ultimate sources of cultural variations in psychology, behaviour and institutions may lie in systematic variations in ecology across societies2,​3,​4,​5. Ecological frameworks have provided theory-driven novel explanations for cultural variations in phenomena ranging from collectivism5 to homicide6. Ecology has also been used to explain not only patterns of cultural variation but also, more recently, changes in various cultural practices and values such as individualism and contempt7. We apply this approach to cultural changes in gender equality, focusing on four dimensions of ecology: pathogen prevalence, resource scarcity, external threat and climatic stress8.

The threat posed by infectious disease has been implicated in a host of cross-cultural differences in phenomena4. These cultural variations have been conceptualized as adaptive responses to varying pathogen loads, part of a behavioural immune system that reduces the chance of individuals becoming ill9. Higher levels of infectious disease seem to lead societies and individuals to adopt more traditional attitudes and norms. Higher levels of pathogen prevalence are associated with more conformity10,​11,​12, more authoritarianism13, less openness14 and tighter social norms15. Given that most human societies have relatively patriarchal traditions, it seems possible that higher pathogen prevalence might also promote behaviours, norms and values that foster gender inequality. That is, as men have traditionally held more power in most societies, if pathogens push people towards more traditional values, then one might expect individuals and institutions to accord women fewer rights and opportunities when pathogen loads are higher. Furthermore, cross-sectional research suggests that, in countries where infectious disease is more prevalent, men and women tend to diverge more in the qualities they seek in potential mates, with women showing a relatively larger preference for intelligence and status in a potential mate and men showing a relatively larger preference for physical attractiveness16. This also suggests that higher pathogen prevalence might increase inequality among the sexes by increasing pressure on men and women to emphasize and develop more traditional attributes and skills. In fact, empirical work has shown that cross-cultural differences in gender inequality are associated with higher pathogen levels16,17. Such cross-sectional observations (for an independent replication, also see Supplementary Fig. 1) raise the question as to whether changes in the levels of infectious disease may lead to changes in gender equality.

Testing longitudinal associations between pathogens and gender inequality can also help address the issue of possible underlying mechanisms. Pathogens may affect gender inequality because they may cue more generally conservative orientations4. Another possibility is that pathogens impact gender inequality because higher levels of infectious disease cue people to adopt faster life history strategies16,18, mating earlier, having more children and focusing on short-term goals as opposed to long-term personal investments such as education19. For example, a recent study found that people who perceived themselves as more vulnerable to disease show more impulsivity, greater delay discounting and a reduced ability to delay gratification18, other work has found that teen birth rates are higher in US states with higher levels of pathogen prevalence20. Thus, because higher levels of pathogens cue people to adopt faster life history strategies, and because earlier mating reduces the ability to pursue education, career and status, women may be less likely to seek or attain positions of economic, social and political power in times where infectious disease is more prevalent. Conversely, when levels of infectious disease are low, people are more likely to adopt slower life history strategies. For women, this might mean delaying reproduction in favour of pursuing education and careers, thus, one might expect gender inequality to decrease with pathogen levels.

Resource scarcity has also been linked to more anti-egalitarian values1 and promotes faster life strategies21. Thus, gender inequality may be higher in times when economic hardship is more prevalent.

Predictions about the effect that climatic stress and external threats such as wars may have on gender inequality are less straightforward than those for pathogen prevalence and resource scarcity. According to climato-economic theory, more demanding climates should lead to less broadly egalitarian norms and values in places where resources are scarce but to more broadly egalitarian norms and values where resources are abundant5. This might lead one to predict that higher levels of climatic stress would be associated with less gender inequality in affluent Western societies (such as the United States or the United Kingdom).

Finally, external threats such as wars may also affect levels of gender inequality, yet predictions are conflicting. On one hand, men typically fight wars and women take on more diverse societal roles during such times, especially in prolonged conflicts. Thus, one might expect war to be negatively associated with gender inequality. On the other hand, external threats (including armed conflicts) have been linked to tighter social norms in cross-cultural work15, suggesting that prevalence of armed conflicts might be associated with an increase in gender inequality.

In study 1, we used archival data from over six decades and tested the role of these four types of ecological threats/affordances in variations over time in levels of gender inequality within US society. We did so using cultural-level indices of gender inequality in the social and political domains, in temporal linguistic trends and in attitudes, as well as a range of archive-based indicators of (social) ecology: pathogen prevalence, resource scarcity, warfare and climatic stress. Subsequently, in study 2, we aimed to replicate the main patterns from study 1 in another society, using comparable archival data.

In study 1, we observed that gender inequality declined dramatically over time (Pearson coefficient of correlation, r = –0.92, P < 0.001; Fig. 1). To test the role of ecology, we first examined zero-order correlations between our four ecological predictors and our gender inequality index. Pathogen prevalence was positively correlated with gender inequality over time (r = 0.77, P < 0.001) and unemployment was negatively correlated with gender inequality (r = –0.29, P = 0.05). Neither war (r = –0.18, not significant (ns)) nor climatic stress (r = 0.15, ns) was significantly correlated with gender inequality. Next, we entered our predictors simultaneously into a regression model with gender inequality as the dependent variable. Results indicated a good model fit (R 2  = 0.70). Pathogen prevalence remained a significant predictor of gender inequality (β (standardized regression coefficient) = 0.75, |t| (Student’s t-test) = 8.26, P < 0.001, rpart (partial correlation coefficient) = 0.69), as did unemployment (β = –0.22, |t| = 2.56, P < 0.02, rpart = –0.21). Climatic stress was also a significant predictor in this model (β = 0.20, |t| = 2.27, P < 0.03, rpart = 0.19) but war was not (β = –0.08, |t| = 0.87, ns, rpart = 0.07).

Figure 1: Gender inequality over time in the United States.
Figure 1

We then used cross-correlation function analysis to assess directionality. We focused our analysis on pathogen prevalence because it was by far the strongest predictor of temporal variation in gender inequality (see Supplementary Figs 3–5 for plots for unemployment, climatic stress and war). This analysis produced a plot that allowed us to assess the directionality of a relationship by comparing the magnitude of correlations between the two variables in a time-lagged fashion. For this analysis, greater values on the left of the graph indicate that changes in pathogen prevalence precede shifts in gender inequality, whereas a skew to the right indicates that the opposite is the case. The results indicate that pathogen prevalence in the United States was positively associated with gender inequality, with the largest correlation observed between pathogen prevalence and subsequent gender inequality 15 years later. The latter observation suggests that changes in pathogen prevalence preceded changes in gender inequality (Fig. 2).

Figure 2: Cross-correlation function analysis of the relationship between pathogen prevalence and gender inequality over time in the United States.
Figure 2

The time series trends are nonlinear, therefore, the gender inequality index is Box–Cox transformed27,28 to improve time series estimates. Analyses with raw data yield a similar picture, with pathogen prevalence slightly preceding gender inequality (see Supplementary Fig. 2). Values outside the blue dashed horizontal lines are significant, P < 0.05.

Finally, we assessed potential mechanisms for the link between pathogens and gender inequality. Separate regression analyses showed that pathogens remained a significant predictor of levels of gender inequality when controlling for collectivism (R2 = 0.49, β = 0.69, |t| = 6.27, P < 0.001, rpart = 0.60) and for conservatism (R2 = 0.64, β = 0.72, |t| = 5.29, P < 0.001, rpart = 0.71). Pathogens remained a significant predictor of gender inequality when controlling for teenage births, but the effect decreased (R2 = 0.62, β = 0.52, |t| = 4.15, P < 0.001, rpart = 0.33). We tested for mediation by each of these variables using separate non-parametric bootstrapping analyses with 5,000 resamples. Collectivism was not a significant mediator of the effect of pathogens on gender inequality, as the bias-corrected and accelerated 95% confidence interval (CI) for the indirect effect included zero (95% bias-corrected and accelerated CI –0.0003–0.0007); nor was conservatism a significant mediator of the effect of pathogens on gender inequality (95% bias-corrected and accelerated CI –0.0011–0.0044). However, we did observe significant mediation of the effect of pathogens on gender inequality with fast life history strategies (teenage birth rate) (95% bias-corrected and accelerated CI 0.0005–0.0022).

In study 2, we sought to replicate in the United Kingdom the central findings of study 1. Specifically, we examined whether pathogen prevalence was associated with gender inequality and whether this relationship was mediated by life history strategies.

Gender inequality in the United Kingdom has declined over time (r = –0.87, P < 0.001). Furthermore, it was positively associated with pathogen prevalence (r = 0.33, P = 0.005), and cross-correlation function analysis showed that changes in pathogen prevalence precede changes in gender inequality; we observed the strongest correlation at a lag of 25 years (Fig. 3). Moreover, the effect of pathogens on gender inequality was reduced when we controlled for teenage births (R2 = 0.61, β = 0.26, |t| = 2.34, P = 0.02, rpart = 0.20). Non-parametric bootstrapping analyses with 5,000 resamples revealed that this mediation was significant (95% bias-corrected and accelerated CI 1.246–3.4648).

Figure 3: Cross-correlation function analysis of the relationship between pathogen prevalence and gender inequality over time in the United Kingdom.
Figure 3

Values outside the blue dashed horizontal lines are significant, P < 0.05.

Our results show that cultural changes in the United States in gender inequality are associated with pathogen prevalence. Levels of gender inequality over time were most strongly correlated with levels of pathogen prevalence. This relationship was robust, holding when we simultaneously controlled for other ecological dimensions. Furthermore, the results of cross-correlation function analysis suggest that changes in pathogen prevalence precede changes in gender inequality, suggesting a causal direction from shifts in pathogens to shifts in gender inequality does exist. Moreover, similar patterns of results emerged in the United Kingdom, suggesting that this effect is not confined to the United States. Taken together, these findings suggest a novel explanation for variations in how men and women are treated in different time periods—levels of infectious disease.

Our results also suggest that, rather than reflecting more general effects of pathogen levels on broadly conservative or traditional attitudes and norms, pathogen levels have specific effects on gender inequality. Furthermore, the results of mediation analyses in both countries suggest that links between pathogen prevalence and gender inequality may be partly due to women adopting faster life history strategies in response to higher levels of infectious disease prevalence. Collectivism did not significantly mediate this effect, nor did conservatism, providing further evidence that this effect is not due to more general links between levels of infectious disease and more traditional attitudes and norms. Rather, these findings are consistent with the notion that life history strategies may provide a mechanism for the link between pathogens and gender inequality.

Before concluding, we note that the longitudinal data analysed in the present studies do not enable definite inferences concerning causality. Although there are indications of the direction of the relationship between pathogens and gender inequality, further approaches involving experiments and agent-based modelling of societal change via computer simulations could help further corroborate causal inferences. This limitation notwithstanding, the present research systematically explored the role of ecologically derived explanations for the forces that may cause societies to change in terms of levels of gender inequality. The results suggest a crucial role for pathogen prevalence. Beyond their theoretical implications, these findings also have practical implications of interest to policymakers. Our results suggest that efforts to reduce infectious diseases, such as vaccinations, free health care, public sanitation and water treatment, might also increase equality among the sexes.


Study 1

We gathered cross-temporal data covering a six-decade period in the United States (1951–2013). We created an index of gender inequality using data on indicators of political representation (the number of women in Congress22), wage inequality (male:female wage ratio based on data from the US Women’s Bureau and the National Committee on Equal Pay, retrieved from http://www.infoplease.com/ipa/A0193820.html), linguistic representation in cultural products (use of male versus female pronouns in published books23) and sexist work attitudes (percentage of respondents in Gallup polls preferring a male boss24). These variables were standardized and averaged to create an overall gender inequality index. The items were highly intercorrelated, and the index had high internal reliability (0.80 <  Pearson coefficients of correlation, rs ≤ 0.95, Cronbach’s α = 0.95). This index is conceptually similar to indices such as the United Nations Gender Empowerment Measure and the Global Gender Gap Index, which are used to assess the levels of gender inequality across countries and include measures of political and financial gender parity.

We took pathogen prevalence and climatic stress data for this period from Grossmann and Varnum7. Their pathogen index7 was based on the prevalence of nine of the most common infectious diseases based on data from the US Census Bureau and the Centers for Disease Control, and their climatic stress index7 was derived by calculating the absolute deviations of average temperatures from 22.22 °C in January and July in a year using data from the National Climatic Data Center. We operationalized resource scarcity as the percentage of the population who were unemployed in a given year according to the US Department of Labor. We operationalized war as a binary variable indicating whether the United States was engaged in a major armed conflict.

We also tested two potential mechanisms by which pathogens might be linked to the levels of gender inequality: more broadly traditional views and norms, and fast life history strategies. We operationalized traditionalism in two ways, first by using the index of collectivist themes by Grossman and Varnum7 in Google Books. We also operationalized traditionalism using Gallup survey data on the percentage of Americans who self-identify as conservatives using data from the earliest year available (1992) until 2013. To assess fast life history strategies, we gathered data on the annual rates of teenage births per 1,000 women aged 15–19 years25.

Study 2

We gathered data from the UK over seven decades (1945–2014) on markers of gender inequality: wage inequality (the gender wage gap as a percentage of median male wage according to data from the Organization for Economic Cooperation and Development), use of male versus female pronouns in British books in the Google Ngrams database and number of women in parliament (retrieved from http://www.ukpolitical.info/FemaleMPs.htm), which were standardized and averaged to form a single index (0.71 <  rs  ≤ 0.90, Cronbach’s α = 0.92). We computed an index of pathogen prevalence during this period using data on mortality rates per 100,000 due to tuberculosis and measles based on data from the UK Office for National Statistics. We also gathered data on annual teenage birth rates per 1,000 women aged <20 years from the UK Office for National Statistics and from Wellings and Kane26. A complete list of data sources as well as notes on the data can be found in Supplementary Tables 1 and 2.

Code availability

The code used to generate the data is available at osf.io/s3pft. All regression analyses were performed in IBM SPSS (version 22). All graphs and CCF analyses were performed using R language for statistical programming.

Data availability

All data presented in this paper are available through the Open Science Framework at osf.io/s3pft

Additional information

How to cite this article: Varnum, M. E. W. & Grossmann, I. Pathogen prevalence is associated with cultural changes in gender equality. Nat. Hum. Behav. 1, 0003 (2016).


  1. 1.

    & Rising Tide: Gender Equality and Cultural Change Around the World (Cambridge Univ. Press, 2003).

  2. 2.

    & in Advances in Culture and Psychology Vol. 3 (eds Gelfand, M. J. et al. ) 78–123 (Oxford Univ. Press, 2014).

  3. 3.

    Socioecological psychology. Annu. Rev. Psychol. 65, 581–609 (2014).

  4. 4.

    & The Parasite-Stress Theory of Values and Sociality: Infectious Disease, History and Human Values Worldwide (Springer, 2014).

  5. 5.

    Climato-economic habitats support patterns of human needs, stresses, and freedoms. Behav. Brain Sci. 36, 465–480 (2013).

  6. 6.

    & Parasite stress promotes homicide and child maltreatment. Phil. Trans. R. Soc. B 366, 3466–3477 (2011).

  7. 7.

    & Social structure, infectious diseases, disasters, secularism, and cultural change in America. Psychol. Sci. 26, 311–324 (2015).

  8. 8.

    & Socio-ecological changes are linked to changes in the prevalence of contempt over time. Behav. Brain Sci. (in the press).

  9. 9.

    & The behavioral immune system (and why it matters). Curr. Dir. Psychol. Sci. 20, 99–103 (2011).

  10. 10.

    , & & On the origins of cross-cultural differences in conformity: four tests of the pathogen prevalence hypothesis. Pers. Soc. Psychol. Bull. 37, 318–329 (2011).

  11. 11.

    & Threat(s) and conformity deconstructed: perceived threat of infectious disease and its implications for conformist attitudes and behavior. Eur. J. Soc. Psychol. 42, 180–188 (2012).

  12. 12.

    Frontiers, germs, and non-conformist voting. J. Cross-Cult. Psychol. 44, 832–837 (2013).

  13. 13.

    , & Pathogens and politics: further evidence that parasite prevalence predicts authoritarianism. PLoS ONE 8, e62275 (2013).

  14. 14.

    & Pathogens, personality, and culture: disease prevalence predicts worldwide variability in sociosexuality, extraversion, and openness to experience. J. Pers. Soc. Psychol. 95, 212–221 (2008).

  15. 15.

    et al. Differences between tight and loose cultures: a 33-nation study. Science 332, 1100–1104 (2011).

  16. 16.

    , & Evolutionary foundations of cultural variation: evoked culture and mate preferences. Psychol. Inq. 17, 75–95 (2006).

  17. 17.

    , & Parasites, democratization, and the liberalization of values across contemporary countries. Biol. Rev. 84, 113–131 (2009).

  18. 18.

    , & Vulnerability to disease as a predictor of faster life history strategies. Adapt. Hum. Behav. Physiol. 2, 116–133 (2016).

  19. 19.

    , & in The Handbook of Evolutionary Psychology 2nd edn, Vol. 1 (ed. Buss, D. M.) Ch. 2 (Wiley, 2015).

  20. 20.

    Sources of regional variation in social capital in the United States: frontiers and pathogens. Evol. Behav. Sci. 8, 77–85 (2014).

  21. 21.

    et al. When the economy falters do people spend or save? Responses to resource scarcity depend on childhood environments. Psychol. Sci. 24, 197–205 (2013).

  22. 22.

    & Women in the United States Congress, 1917-2004: Biographical and Committee Assignment Information and Listings by State and Congress. Report RL30261 (Congressional Research Service, 2014).

  23. 23.

    , & Male and female pronoun use in US books reflects women’s status, 1900–2008. Sex Roles 67, 488–493 (2012).

  24. 24.

    & Gallup Poll Social Series: Work and Education (Gallup, 2014).

  25. 25.

    , & National and State Patterns of Teen Births in the United States, 1940–2013 Vol. 63, No. 4 (National Vital Statistics Reports, National Center for Health Statistics, 2014).

  26. 26.

    & Trends in teenage pregnancy in England and Wales: how can we explain them? J. R. Soc. Med. 92, 277–282 (1999).

  27. 27.

    Nonlinear time series models in economics. J. Econ. Surv. 5, 215–242 (1991).

  28. 28.

    & Does the Box–Cox Transformation Help in Forecasting Macroeconomic Time Series? (Univ. Library of Munich, 1991).

Download references


This research was supported by the Insight grant no. 435-2014-0685 from the Social Sciences and Humanities Research Council of Canada (to I.G.) and by a grant from the John Templeton Foundation ‘Prospective Psychology Stage 2: A Research Competition to Martin Seligman’ (sub-grant awarded to I.G.). The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the John Templeton Foundation. Neither funder had any role in the study design, the data collection and analysis, the decision to publish or the preparation of the manuscript.

Author information


  1. Department of Psychology, Arizona State University, 950 S. McAllister, Tempe, Arizona 85287, USA

    • Michael E. W. Varnum
  2. Department of Psychology, University of Waterloo, 200 University Avenue West, Waterloo, Ontario N2L3G1, Canada

    • Igor Grossmann


  1. Search for Michael E. W. Varnum in:

  2. Search for Igor Grossmann in:


M.E.W.V. developed the original study concept. M.E.W.V. and I.G. gathered and analysed the data and drafted and revised the manuscript.

Competing interests

The authors declare no competing interests.

Corresponding author

Correspondence to Michael E. W. Varnum.

Supplementary information

PDF files

  1. 1.

    Supplementary information

    Supplementary Methods, Supplementary Results, Supplementary Figures 1–5, Supplementary Tables 1,2 and Supplementary References