The demography of intolerance: communist legacy, demographic threat, and attitudes toward homosexuality

ABSTRACT

We analyse attitudes toward homosexuality in postcommunist- and Western Europe, focusing on two distinctive postcommunist demographic trends: dramatic and rapid declines in fertility, which may spark fears about the “future of the nation” (fertility threat) and persistently elevated mortality rates, which evoke feelings of insecurity and sustain “survival values” that are less tolerant of homosexuality (mortality threat). We probe this theorisation's plausibility using attitudinal data from 1989 through 2022, finding that mortality threat contributes to regional differences in tolerance but little evidence that fertility threat does. Our analysis refines theorising about communist legacies and demographic alarmism in populist rhetoric.

KEYWORDS:

• Communist legacy
• mortality rate
• fertility rate
• attitudes toward homosexuality
• modernisation theory

Acknowledgments

We thank Andrew Rosenberg, Valerie Sperling, Matthew Stenberg, the European Politics Working Group at UC, Berkeley, as well as the reviewers and editors at East European Politics for their very helpful suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Notes

1 The rate of natural increase is defined as the crude birth rate minus the crude death rate (Mole 2011, fn 4).

2 Each wave denoted on the x-axis corresponds to the following periods and surveys: 1989-1994, WVS II & EVS II; 1995-1998, WVS III; 1999-2004, WVS IV & EVS III; 2005-2010, WVS V & EVS IV; 2010-2014, WVS VI; and 2017-2022, WVS VII. See Online Appendix A for a list of the countries included.

3 To classify the postcommunist democracies, we use Freedom House’s list of electoral democracies (https://freedomhouse.org/report-types/freedom-world).

4 For the countries included in Figures 2 and 3, see Online Appendix B.

5 The World Bank data description defines male adult mortality rates as the probability expressed as the number of people per 1,000 male adults “of dying between the ages of 15 and 60 – that is, the probability of a 15-year-old dying before reaching age 60, if subject to age-specific mortality rates of the specified year between those ages” (http://databank.worldbank.org/data/home.aspx).

6 The opioid epidemic in the US is changing this picture (Case and Deaton 2020), not to mention Covid.

7 The data are taken from the World Bank database.

8 Note that this should not be confused with an interactive effect between demographic changes and the postcommunist dummy. Our estimation strategy is not to see if demographic variables have varying effects on tolerant attitudes between the two regions. Instead, our goal is to explain what gave rise to the regional differences in levels of tolerance by finding a underlying factor driving the relationship between the region and attitudes, that is, a postcommunist deficit in tolerance. This can be achieved statistically if we can identify a third factor that affects both the regional dummy and tolerance, thereby nullifying the significant effect of the former on the latter. The same logic applies to the fertility hypotheses.

9 The first wave was not used for the analysis as the survey was conducted during the communist era.

10 The full list of country years in the sample is found in Online Appendix C.

11 We acknowledge that the justifiability survey item may be an imperfect measure of tolerance insofar as tolerance is an active behavior and justifiability is a passive opinion. An alternative would be to use the WVS item asking whether homosexuals make undesirable neighbors. However, since this is a binary measure, it fails to capture the more detailed variation in the level of tolerance that we aim to explain. Furthermore, since there is no good reason to believe that the WVS’s justifiability item leads to biased results (Anderson and Fetner 2008), it has been widely used in previous studies: using it here enhances the comparability of our findings.

12 We use highest education level attained (8-point scale) as a measure of education, but most of the WVS II studies were missing for the variable. To get around this, as a substitute, the rescaled item measuring the age of completion of education (10-point scale) is employed for the WVS II subsample.

13 Using the item asking the respondent’s job profession, we categorised those employed as a ‘foreman and supervisor,’ ‘skilled manual worker,’ ‘semi-skilled manual worker,’ and ‘unskilled manual worker’ into the working class.

14 Following Pop-Eleches and Tucker (2013), we treat ages six through seventeen as one’s formative years.

15 Those who answered they are either Protestants, Christians, Evangelicals, Baptists, Presbyterians, Methodists, or Lutherans, are grouped as Protestants.

16 For the construction of the index, we consult diverse sources such as equaldex.com, Amnesty International, and the International Lesbian, Gay, Bisexual, Trans and Intersex Association (ILGA).

17 We acknowledge that EU membership and the legal environment regarding LGBTQ rights may not be independent because (1) EU membership requires an expansion of minority rights, including LGBTQ rights, and (2) a given country’s legal environment might be influenced by other EU members once that country gains membership. However, our dataset shows that there is still variation in LGBTQ legal environments across countries and years, even when considering the period before and after EU entry. Additionally, in our dataset the Pearson’s correlation between the two is 0.42, indicating a modest but not a strong correlation. Moreover, running models excluding each of the two variables separately produced consistent results with those including both variables.

18 Of course, as Guasti and Bustikova (2020, 240) argue, EU policies expanding LGBTQ rights, especially rulings by the European Court of Justice regarding recognition of same-sex marriage, are also a source of fear for domestic illiberal elites. In general, though, the literature on the EU’s impact on LGBTQ politics in postcommunist countries has emphasized an overall improvement in minority recognition, even if it follows an initial wave of backlash (Ayoub 2016; O’Dwyer 2018).

19 We use World Bank GDP data.

20 As SWIID arguably provides much better data coverage and comparability across times and spaces than alternative sources (Solt 2020), we believe that our analysis uses a more solid measure for income inequality than Anderson and Fetner’s (2008), in which they relied on multiple different sources and interpolated missing observations by averaging the values in “the years immediately prior and following” (947).

21 The countries in our sample are categorised as: (1) Catholic: Croatia, France, Hungary, Italy, Lithuania, Poland, Slovakia, Spain; (2) Protestant: Finland, Norway, Sweden, United Kingdom; (3) Mixed Christian: Australia, Canada, Czech Republic, Germany, Netherlands, New Zealand, Slovenia, Switzerland, United States; (4) Others: Albania, Bosnia and Herzegovina, Bulgaria, Cyprus, Estonia, Georgia, Latvia, Macedonia, Moldova, Montenegro, Romania, Russia, Serbia, Ukraine.

22 In multilevel settings, addressing group-specific effects is crucial. A random-intercept multilevel model may not be suitable when trying to identify within-group causes because it requires including all potential group-level factors that could determine the outcome—which is impractical. In such cases, fixed effects are preferred. However, if the goal is to examine the effects of specific group-level factors, a random effect model is more appropriate as it allows testing the group-level variables of interest while treating unexplained effects as a variance component. Since we aim to examine what national-level factors account for individual-level differences across the regions, we employ a random effect model. We also considered Bryan and Jenkins (2016) suggestion of utilizing a two-step model designed to address the estimation issue of group-level effects with a small number of groups, but find it unnecessary as our final sample includes 101 WVS studies, which cannot be regarded as small.

23 The results of the surveyed years are not reported. The result table with all variables is found in Online Appendix D.

24 Following Gelman (2008), we standardise the independent variables by subtracting the mean from all non-binary variables and dividing them by two standard deviations, leaving binary variables unchanged. This standardises the variables for comparability in terms of their substantive effect.

25 Because modernisation processes involve various economic and political changes linked to fertility, we also excluded the other country-level variables from the model and ran it only with fertility to focus on its individual effect. The results did not meaningfully change.

26 See endnote 25.

Additional information

Notes on contributors

Dong-Joon Jung

Dong-Joon Jung is Associate Professor of social studies education at Inha University in Incheon, South Korea ([email protected]). With a regional focus on postcommunist Europe and South Korea, his research interests center upon democratisation and regime transitions, comparative political behaviour, parties and elections, civil society, gender and political attitudes.

Conor O’Dwyer

Conor O’Dwyer is Professor of political science at the University of Florida ([email protected]). He specialises in comparative politics, with a regional emphasis on postcommunist Europe and thematic interests in social movements, LGBT politics, political parties, and, most recently, postcommunist urbanism. He is the author of Coming Out of Communism: The Emergence of LGBT Activism in Eastern Europe (NYU Press, 2018).