Literacy and Voting Behavior: A Bivariate Probit Model with Sample Selection

Abstract A bivariate probit regression with sample selection was used to explore the contribution of selected demographic and sociopolitical variables on registering to vote and on voting. The database for this study was a subset of the respondents to the Young Adult Literacy Survey (YALS), administered in 1985 on a stratified national sample of 21- through 25-year-olds. Separate probit regressions showed that all selected independent variables were significant predictors for both registration and voting. However, the model implied by separate probit regressions fails to take into account that only those who actually register to vote can vote. When a more appropriate model, bivariate probit regression with sample selection, was applied to these data only amount of hard news read and years of education were significant predictors of voting, given registration. For registration, these variables, plus ethnicity/race, television viewing, and degree to which one keeps up with governmental affairs were all significant predictors. We argue that a bivariate probit model with sample selection is an appropriate and essential approach for modeling voting behavior. We then conclude that increased education as well as increased attention to hard news reading within school and adult education curricula could lead to increased propensities to vote. The limitations of the present research as well as suggestions for improving the database required for voting studies are also given.