Speaking of character: Character strength references in movies and presidential nomination speeches

ABSTRACT

One way in which language can be used for persuasive purposes is through references to personal character, or the lack thereof. A series of lexica reflecting 24 character strengths were developed and combined with existing lexica reflecting virtues and morality. Two studies were conducted to evaluate these lexica. The first compared words in presidential candidate nomination speeches from 1864 to 2016 to election outcomes. Results indicated that candidates’ use of terms related to hope was a better predictor of election outcomes than incumbency. In the second study, both the highest grossing and highest rated films of 2013–2016 included fewer words representing seven of the character variables than a normative database, while the highest grossing films contained more dialog referencing creativity and leadership. These results indicate that lexical analysis can provide a potentially useful tool for understanding the cultural use of positive character terms.

KEYWORDS:

• Positive psychology
• VIA-IS (Values in Action Inventory)
• virtue
• hope
• presidential
• speech
• movies
• language
• election

Acknowledgments

The authors are grateful to the VIA Institute on Character for their support for the current research.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Grover Cleveland was anomalous in that he completed two non-successive terms. We identified him as an incumbent in both his second and third runs for the presidency.

2. Given the small sample size, analyses were replicated using cross-validated lasso multiple and logistic regression, which adds bias to coefficient estimates in return for reduced sampling error in coefficient estimates when the ratio of participants to predictors is low (e.g. Zou & Hastie, 2005). These analyses were conducted using the glmnet package (Friedman, Hastie, & Tibshirani, 2010) available in R, testing 100 estimates for the bias coefficient with the goal of minimizing the resulting bias. In these analyses, all predictors were associated with non-zero coefficients. However, standardized regression coefficients for the hope variable were consistently larger than those for incumbency, supporting its superiority as a predictor of the two outcomes. Authenticity also proved superior to incumbency as a predictor of popular vote.