Personality Traits and Body Weight: Evidence from the Health and Retirement Study

Although psychologists have long been interested in the impact of personality on social outcomes, economists have traditionally placed more research attention on intelligence or ability, as measured by IQ tests and educational attainment, and have only recently begun to carefully consider a broader set of factors which encompass attributes such as personality.  Much of the economic work on personality to date has focused on labor market outcomes and academic performance, while a handful of recent studies have examined health outcomes and behaviors. 

In this study we examine the importance of personality traits, as measured by the Big Five taxonomy, for body weight among older adults using data drawn from the Health and Retirement Study.  Older adults, defined as those 50 years of age and above, comprise an increasingly large share of the United States population and are thereby an important group to study.  In 2010 there were 99 million (32%) older adults, an increase of 24% from 1980.  As the population ages, health care costs are high and escalating rapidly: currently the U.S. spends over $2.7 trillion each year on medical services.  Because health care costs increase as individuals enter older ages these two trends raise concerns among fiscal policymakers.  Little economic research has investigated the importance of personality for older adult health behaviors, however.

We measure body weight with the body mass index (BMI), and dichotomous indicators for underweight (BMI < 18.5) and obesity (BMI ≥ 30).  We find evidence that the Big Five personality traits are strong predictors of body weight.  Overall personality traits appear to be more relevant for female than male body weight, and agreeableness, openness, extraversion, and conscientiousness show the strongest relationship with body weight.  Traits of agreeableness and openness have a positive association with BMI and obesity. Extroversion and conscientiousness are negatively associated with BMI and obesity.  Findings from unconditional quantile regressions demonstrate a substantial degree of heterogeneity between the traits and BMI in that the associations are generally largest in the right tail of the BMI distribution.  Our results indicate that personality is associated with weight outcomes for older adults, suggesting different types of weight loss programs may be more appropriate for successful weight reduction based on personality traits. 

In an extension we contribute methodologically to the health economics literature by building on seminar work by Cawley (2004) and Cawley and Burkhauser (2006), and develop and test a correction algorithm for measurement error in self-reported weight and height specific to the older adult population.  Self-reported weight and height variables are known to contain substantial systematic reporting error, but exiting correction algorithms are designed for the working age population exclusively.  Given differences between older adults, in particular the elderly, and the general population of adults it is not clear whether it is appropriate to apply existing algorithms to this specific sub-population.  Moreover, we show that failure to correct for errors in self-reported data among the elderly can lead to substantial errors in underweight and obesity prevalence estimates.