What makes us happy?: predicting general happiness from the general social survey

Abstract

This investigation explored the salient variables that predict general happiness, as well as those variables’ relative importance, as indicated by the General Social Survey (GSS). The study applied the Least Absolute Shrinkage and Selection Operator (LASSO) regression method to explicitly select the most important variables and remove the others. Additionally, a Random Forest (RF) was applied to the variables to create a large number of models with random subsets of the variables included in each. All coefficients were reduced to zero within the LASSO regression, and the relative importance of each variable were diminutive. Due to unexpected results, follow-up correlations and a forward regression were conducted. All predictor variables were significantly, positively, and, at best, moderately correlated with the outcome variable. Within the forward regression, the second model, fit the data the best, accounting for approximately 39.8% of the variance with only two variables. This study concludes that an immediate sense of happiness and satisfaction in and with life best determines general happiness within United States’ citizens.