Evaluating eight-year trajectories for sense of purpose in the health and retirement study

Why we asked

Feeling that your life has direction and meaning is linked to healthier aging, but earlier cross-sectional work suggests that a sense of purpose declines from middle age into later life. We wanted to know two things. First, do all older adults decline, or do some hold steady or even increase? Second, can we tell who is more likely to decline? We also suspected a measurement problem. Many purpose items ask about carrying out daily plans, and a person with a job may read those items differently from a retired person.

What we did

We used data from 4,234 participants in the Health and Retirement Study (mean age about 65) who answered a seven-item purpose scale in 2006, 2010, and 2014. We first tested whether the scale measured the same thing at each wave and for four groups: people who kept working, people already retired in 2006, people who retired during the study, and people who had never worked. We then fit latent growth models within each group and asked whether demographics, self-rated health, and the Big Five predicted where people started and how they changed.

What we found

The scale behaved consistently across waves but not across work groups. In other words, the items did not load onto purpose in the same way for workers and retirees, so we modeled the groups separately. Purpose declined on average in every group except people who kept working. Only the group already retired in 2006 showed reliable differences between people in how much they changed.

Within that group, older age was associated with steeper declines. Better health and more education were associated with less decline. Being married was associated with more decline, which may reflect higher starting levels leaving less room to rise (this is our speculation). When we entered all five traits together, none uniquely predicted change, although conscientiousness, extraversion, and emotional stability were associated with higher starting levels.

What it means

The model fit was only moderate, and assessments were four years apart, so we could not look at events like widowhood. This summary is based on the preprint, which is the version we host.

Researchers should check whether a purpose scale means the same thing to the people they compare.