Multivariate Model Details

In Chapters 5 and 6, I included multivariate regression models of various outcomes related to wealth ownership using the 1979–2004 NLSY. I use data prior to 1985 to create background variables; and I use data from 1985–2004 to create other measures including dependent variables. For all models, I created pooled cross-section time series data with the 1985–2004 data; collection of wealth information began in 1985 when the youngest respondents were in their twenties. In the multivariate models, person-years is the unit of analysis, allowing me to take advantage of the longitudinal nature of the data. The data included one observation per respondent per year, and both the dependent and independent variables were able to vary yearly for each respondent. Table 5.9 includes logistic regression models of father's and mother's educations. The dependent variable is whether the respondent's biological father (model 1) and biological mother (model 2) had a bachelor's degree or more. Table 5.10 includes a model of another dichotomous outcome: whether or not the respondent has ever received an inheritance. Because the outcome in each of these models is dichotomous, I use logistic regression to estimate the models. Tables 5.9 and 5.10 use the full NLSY sample; Table 6.1 repeats these logistic models for white, nonimmigrant respondents to the NLSY who reported religion as Roman Catholic in 1979. Otherwise the models are identical.