Is it possible to predict current voting intentions by Big Five personality domains, facets, and nuances? A random forest analysis approach in a German sample
Voting decisions of individual voters have the power to influence political developments by enabling certain parties and politicians to govern their country. Therefore, it is of tremendous importance to understand what is driving individual differences in voting intentions and decisions. This study investigated the predictability of voting intentions in a German sample by Big Five of personality domains, facets, and nuances/items; thereby overcoming several shortcomings of previous studies descriptively associating Big Five domains and voting behaviors.A dataset of N = 4,997 people (48.19% men) was investigated. Random forest models were trained and tested at different split ratios of training-to-test datasets 1,000 times. Mean classification errors and variable importance scores across runs at each split were extracted to predict i) voting versus non-voting, ii) voting for specific parties, iii) voting for left- versus right-from-the-center parties.Only voting and non-voting could be predicted with classification errors consistently below 10% by Big Five domains, facets, and items. Openness was one important domain in this prediction. Investigating facets and nuances did not enhance prediction over investigating domains.