Tag Archives: Bottom-Up

Personality and Health- Satisfaction in the SOEP

Research on personality and health has a long tradition. Some research provides support for the concept of a hypochondriac personality. That is, some people react to physical symptoms with extreme distress and they tend to exaggerate the severity of symptoms or their consequences (see Seinfeld episode as an example, clip). Watson and Pennebaker (1989) found that neuroticism is consistently related to subjective health perceptions. Not surprisingly, neuroticism is also a predictor of health-satisfaction ratings (Brief et al., 1993).

One open question in wellbeing science is how personality is related to domain satisfaction (Diener, Lucas, & Oishi, 2018). One possibility is that personality traits like neuroticism influence global life-satisfaction judgments and that global life-satisfaction influences satisfaction with specific life domains. According to this model, life-satisfaction would mediate the relationship between neuroticism and health satisfaction (Heller, Watson, & Illies, 2004). The alternative model assumes that personality influences domain satisfaction and that satisfaction with important life domains leads to higher overall life-satisfaction (Brief et al., 1993; Schimmack, Oishi, & Diener, 2002). So far, empirical studies have been unable to settle these opposing views of the relationship between life-satisfaction and domain satisfaction. The SOEP data provide a unique opportunity in making progress on this front because personality, life-satisfaction and domain satisfaction have been assessed in three waves over an eight-year period.

I already posted analyses of life-satisfaction and job satisfaction (Schimmack, 2019a, 2019b). Here, I present the results for health satisfaction. These results will be used to build a larger model with multiple domains in a single model. The model is identical to the model that was used to analyze life-satisfaction (see OSF for code and detailed results; https://osf.io/vpcfd/ ). Model fit was acceptable, CFI = .97, RMSEA = .022, SRMR = .030.


Observed stability was r = .54 from 2005 to 2009, r = .54 from 2009 to 2013, and r = .48 from 2005 to 2009. It is remarkable that the retest correlation spanning 8 years is just slightly lower than the 4-year retest correlations. Using Heise’s formula, this implies low reliability and high stability; REL = .54*.54/.48 = .61, 8-year stability = .48/.61 = .79. The reliability estimate is consistent with estimates based on annual assessments (Schimmack, Schupp, & Wagner, 2008). Thus, health satisfaction is rather stable and it is worthwhile to examine the predictors of stability in health satisfaction.

Personality measured at Time 1 was used as a predictor of health satisfaction at times 1 to 3. If personality contributes to stability in health satisfaction, personality traits should predict health satisfaction concurrently and prospectively. The results in Table 1 show that this was the case for neuroticism. The remaining personality traits were weak predictors of health satisfaction. Halo bias also predicted stability in health satisfaction but the effect was small and decreased over time. Overall, these results are consistent with the hypochondriac hypothesis.


Table 2 examines whether changes in personality predict changes in health satisfaction. To do so, health satisfaction was regressed on the residual variances in personality at times 2 and 3.

Extraversion 0.010.06

As before, neuroticism and halo bias were the only notable predictors of change in health satisfaction. The results for halo bias show that health satisfaction ratings change as respondents tendencies to respond positively change. The results for neuroticism are more dfficult to interpret. Maybe changes in health status produce changes in neuroticism or changes in neuroticism produce changes in health perceptions. More complex models are needed to disentangle these complex relationships.

The final result was the stability of the residual variance in job satisfaction that is not explained by personality – as measured in the SOEP. Stability estimates were r = .86 and r = .86 over the 4-year intervals with an implied stability of r = .75 over the 8-year interval. Thus, personality is just one predictor of stability in health satisfaction and it contributes a relatively small amount to stability in job satisfaction. Other factors like objective health status may also contribute to stability in health satisfaction.


The results are largely consistent with previous evidence that neuroticism is the main predictor of health satisfaction (Brief et al., 1993). The results show that this relationship holds concurrently and prospectively over an eight-year period and that it holds while controlling for shared method variance in personality and health ratings. These results will be used for a more complex model that can distinguish between top-down and bottom-up effects of health satisfaction and life-satisfaction (Diener et al., 2018).