Researchers from the University Medical Center Groningen calculated a gender index for adult Lifelines participants based on baseline questions on a broad range of topics, in order to facilitate the research into the role of gender versus sex in healthy ageing.
Notably, biological sex and psychosocial gender are not the same concepts. Sex refers to the biological characteristics, included but not limited to genes, anatomy, hormones of males and females. Gender, in contrast, entails the embodiment of different roles, behaviors, identities, and relationships of men and women prescribed by societal norms, that result in different social expectations, opportunities and experiences for women and men in a given society. Gender and its embodiment is dynamic over time, while biological sex is mostly stable.
The gender index is a continuum, ranging from 0% (masculinity) to 100% (femininity). A gender index of 50% indicates androgyny, where equal levels of feminine and masculine characteristics are present (sections: demographics (sex & gender) and secondary & linked variables).
The gender index can be requested in the Lifelines catalogue or by mail (data@lifelines.nl). When this data has been used in your research, you will have to include a reference to the following paper:
Participants had to meet several criteria in order to be included in the calculation:
This selection resulted in the inclusion of N=13,321 adult Lifelines participants in a subsample, which was used to calculate the gender index.
All psychological and social variables are included in the model to construct the gender index, as far as these meet both of the following criteria:
In total, 153 variables have been included in the calculation of the gender index. These variables cover three of the four gender aspects (i.e. gender roles, gender dynamics and institutionalized gender). More detailed information on the selection process can be found in the reference article.
To construct the gender index, 245 (dummy) variables, derived from the 153 selected variables were entered into an elastic net regularized generalized linear model, with sex as a dichotomous outcome. The participants were randomly assigned to a trainingset (80%) and a testset (20%). The trainingset was used to estimate regression coefficients, whereas the testing set was used to calculate the developed model’s predictive accuracy. Ultimately, an alpha = 1.0 was found to be the optimal regularization parameter, equaling a LASSO logistic regression. Of the initial 245 potentially sex-related variables, 92 were excluded from the model and 153 (dummy) variables remained representing 85 unique variables. The model’s AUC was 92%.
The estimates of the coefficients obtained through the aforementioned regression formed the basis of the composite gender index that was applied to each adult Lifelines participant.
Derivative variable
The resulting variable from this analysis is the gender index. This variable is a continuum, ranging from 0% (masculinity) to 100% (femininity). A gender index of 50% indicates androgyny, where equal levels of feminine and masculine characteristics are present.