Numerous studies have highlighted a statistical relationship between homosexuality and some outcomes. These outcomes range from indicators of mental and physical health to health behaviors (see for example, Lee at al., 2017, about prevalence of depression; Fricke and Sironi, 2017, about sleep disorders). However, it is difficult to understand if homosexuality induces the outcomes, or if there are other (unmeasured) variables that induce both homosexuality and outcomes.

With reference to male homosexuality, we propose to face this problem by means of instrumental variables; that is, inserting into a regression model (homosexuality à outcomes) some independent variables that should influence outcomes (e.g. depression) only through homosexuality.

As instrumental variables, we ​​propose to use the number of siblings and the number of male siblings. Some studies have shown that these two variables have a direct influence – justified from genetic and biological reasons – on the probability of developing homosexuality in the male population (Bogaert et al., 2017; Balthazart 2018; Camperio Ciani and Pellizzari 2012).

This approach is tested on the SELFY survey, which involved more than 7,000 Italian university students in 2017. The data includes some strict proxies of the two tentative instrumental variables, together with different measures of homosexuality.