Childhood experiences, parental attachment relationships, and gender identity of men

Previous research has suggested that early recollections of parent-child relations may vary across groups of men with and without cross-gender identity (i.e. transsexuals). The goal of the present study was to further clarify the relationship between gender development and attachment relationships in men. The Internet was used to recruit a sample (n = 221) of male transvestites, male-to-female transsexuals, non-gender dysphoric male homosexuals, and non-gender dysphoric male heterosexuals. A battery of questionnaires were administered, measuring cross-gender behavior, mental health, social desirability, and parent-child relationships.;In regards to group differences on childhood experiences, a series of MANOVAs demonstrated that heterosexuals differed from homosexuals, transvestites, and transsexuals on many of the maternal and paternal relationship variables. However, these difference were attributable to self-reports cross-gender behavior in childhood. Regardless of group, men reporting high levels of childhood cross-gender behavior, as measured by the Feminine Gender Identity Scale - Part A (Freund, et al., 1977), reported low levels of parental warmth and support and high levels of parental overprotection, rejection, and dependency. Men reporting high levels of cross-gender behavior in childhood also reported higher levels of abuse (physical and emotional) than their low cross-gender behavior comparison group. These data suggest that differences in parent-child relationships are related to cross-gender behavior in childhood more so than adult gender identity.;One concern was that an Internet-recruited sample would be peculiar (e.g., high levels of mental distress) or deceitful (e.g., falsely report transgendered behavior). These data demonstrated that this Internet-recruited male sample differed on behaviors that have differentiated groups recruited in more traditional settings (e.g., clinic), scored within the appropriate range on a mental health screen, and had low correlations between most subscale scores and a measure of other-deception. These results are argued to support the generalizability of our research.