Fig. 1

Results of subgroup analyses, with logistic regression models testing associations between exposures and the odds of attending arts events separately in each survey year (1993 n = 1590, 1998 n = 1432, 2002 n = 1355, 2012 n = 2838, 2016 n = 1469). Odds ratios and 95% confidence intervals are displayed. For associations between sex and arts events, the odds ratio represents attendance in females compared to males. For associations between race/ethnicity and arts events, White is the reference category. Associations were estimated in the full logistic regression models (including all exposures as shown in Table 2), but only results for sex and race/ethnicity are presented