Could Depression Influence Grocery Choices?

Households with at least one person with depression did demonstrate some differences in shopping patterns. Compared with other households, they spent about 5 percent less on groceries annually and about 40 percent more at convenience stores. They also devoted nearly 20 percent less of their shopping carts to fresh produce, 70 percent more to tobacco products, and 20 percent less to alcohol. Spending on junk food such as salty snacks, bakery items, and candy was roughly the same.

Those links are simply correlations—they don’t prove that depression causes differences in shopping, the researchers caution. “Depression could be correlated with shopping behavior either through a direct causal link between depression and choice process or through incidental correlation with a third, unmeasured variable,” Meckel and Shapiro write.

Other factors appear to be at play, they suggest. The relationships largely disappeared when the researchers took other variables such as household income, homeownership, and age into account, which suggests that those variables are more likely than depression to explain shopping differences. The researchers also find that when the depressed person in the household got treatment, the shopping patterns didn’t change from one year to the next.

Meckel and Shapiro note that there could still be a relationship between depression and grocery shopping that simply takes longer than four years to observe or that requires more nuanced data than the study captures—such as whether participants’ depression is short term or chronic. More research is needed to understand how, when, and to what degree depression affects food choices, they write.

What was more certain was that households with depression dropped out of the study more frequently than households without. And this is the part that the researchers are concerned could affect the the findings of marketing studies that rely on surveys to understand shopping behaviors. Researchers have to figure out how to address this issue, as it could create bias in their samples and unreliable findings, Meckel and Shapiro write.

They advise that researchers need a strategy specifically for taking into account the selection of depressed households out of a sample. In particular, a cross-sectional design, which involves collecting data from many people at a single point in time or during a single year, rather than requiring people to participate in a study that spans years, should provide a more representative pool of respondents. This comes with its own challenges, the researchers note, outlining additional steps to mitigate those. However, “the more you require of your survey respondents, the more people will eventually churn from the sample, and that churn is nonrandom with respect to depression,” says Shapiro.

In the meantime, grocery sellers, at least, may take comfort in the fact that shopping behavior is relatively stable even when certain factors change. The research, Meckel and Shapiro write, “adds to our understanding of how sticky shopping behavior can be.”