Fig 1.
Treatment effects of the positive and negative frames compared to the control group.
Estimates based on OLS regression with country dummies and post-stratification weights. Standard errors clustered at the country level (N = 14,017). Outcome: latent measure of attitudes about Qatar underlying respondents’ agreement (1 = fully disagree, 7 = fully agree) with the following three items: (i) “The decision to award the FIFA World Cup tournament to Qatar was reasonable”, which asks about the actual decision to hold the event in Qatar. (ii) “The FIFA World Cup will be well-organized by Qatar”, which refers to Qatar’s organizational efficiency. (iii) “Qatar’s reputation will improve due to hosting the FIFA World Cup.” Figure based on regression results from Model 1 in Table S3 in the S1 File.
Fig 2.
Distribution of responses on the assessment of the human rights situation in Qatar by country.
The x-axis plots the share of respondents with negative (left of 0) and positive assessments (right of 0) respectively on a 7-point scale where 1 = “not respected at all” and 7 = “well-respected” for each country (y-axis). Percentages per country indicate the aggregate share of all respondents with a negative (below 4) and with a positive assessment (above 4) respectively. Number of valid responses = 11774. Missing values (“don’t know”, “don’t want to respond”) omitted from the graph.
Fig 3.
Heterogeneous treatment effects.
Heterogeneity of framing effect strength across countries with treatment effect size on the y-axis and the baseline (intercept) of the control group on the x-axis. Data on media pluralism taken from the SGI [42], color-coded according to level of pluralism with light gray indicating lower and dark gray higher levels of media pluralism (1 = lowest, 10 = highest). Effect sizes and figure based on regression results with post-stratification weights from Table S4 in the S1 File.
Fig 4.
Assessment of heterogeneous treatment effects at the individual level.
Predicted values of attitudes towards Qatar as the World Cup host (y-axis) by respondents’ individual characteristics (on respective x-axes): A: Ideological self-placement from left to right. B: Authoritarian values from low to high. C: General interest in football from low to high. Predicted values based on OLS regressions with post-stratification weight, country dummies, and clustered standard errors on the country level. Figure based on regression results from Table S10 in the S1 File.
Fig 5.
Principal component analysis along (i) the level of democracy (x-axis) and (ii) the quality of the football team (y-axis) with percentages of explained variance in parantheses. Countries participating in the 2022 FIFA World Cup are color-coded in dark gray, non-participants in light gray.
Table 1.
Overview of selected countries.
Table 2.
Experimental setup and issue frames.