Tag Archives: correlation is not causation

Chocolate, Correlation, Causation, and Depression

A recent study published in the Archives of Internal Medicine has attracted a lot of media attention. ‘Chocolate and Depressive Symptoms in a Cross-sectional Analysis‘ has been written up in a number of newspapers here in the United States, often in the ‘oddly enough’ news sections, which is where ‘whimsical’ news about scientific studies often seems to end up.

There’s kind of a long history of media outlets writing up studies on food and twisting or ignoring the conclusions of those studies. Contradictory information about food is readily available through the media and sometimes there are very serious problems with the framing of these ‘oddly enough’ food stories. Like the claims that red wine is good for your heart, which ignored the fact that most studies supporting this claim pointed out that you would have to drink a lot of red wine to achieve cardiovascular benefits.

Two chocolate bars, a Richfield's Dark with Raspberries, and a Whittaker's Dark Caramel, shown in their wrappers.

The chocolate study involved 1018 subjects from San Diego, 931 of whom were not taking antidepressants. Researchers used the Center for Epidemiologic Studies Depression Scale (CES-D) to gauge levels of depression among the participants, and then collected information about chocolate consumption. What they found is that people with higher scores on the Scale tended to consume more chocolate. For people with very high scores, chocolate consumption was doubled.

The conclusion?

Higher CES-D depression scores were associated with greater chocolate consumption. Whether there is a causal connection, and if so in which direction, is a matter for future prospective study.

The media jumped on this study with headlines like ‘Study links chocolate and depression.’ ‘Study: Chocolate and depression go hand in hand.’ ‘Say it ain’t so: Study links chocolate to depression.’ And so on. Some articles went right ahead and used the ’cause’ word.

These articles theorize that either depression triggers chocolate cravings, or chocolate makes people depressed. They’ve made the classic mistake of conflating correlation and causation. abby jean has talked about this problem elsewhere on FWD, discussing the problem of bias in studies and bias in the way people talk about scientific studies. Recently, I discussed some of the problems with tests administered to measure levels of depression in ‘Internet Use and Depression.’

There are a couple of questions which should be asked about a study like this. The first is how the sample was chosen. Because I can’t access the full text, I can’t answer this question (someone who has read the study is welcome to chime in in comments). One thing which is notable is that this study took place in the United States. Here in the States, eating chocolate to  improve mood is a learned behaviour, and thus I can’t say that I was terribly surprised to  learn that people in the US who appear to be depressed are eating more chocolate. Since this study took place in a country where people believe that there is a link between chocolate and mood, it’s possible that some observer bias was involved in this study.

People believe that chocolate will make them feel better when they’re down1. In other cultures, there are other comfort foods which people eat to make themselves feel better. It would be interesting to see a cross-cultural study. As it stands now, we know that chocolate consumption appears to be higher in depressed residents of San Diego, California.

Another question about this study is whether the researchers controlled for other dietary factors or behaviours. Apparently they did control for ‘common’ factors like caffeine and carbohydrates, but it’s unclear how far they drilled down. Is it possible that people with depression also eat a lot of, say, peas? Brazil nuts? It’s impossible to control for everything, of course, but food and diet are complex issues and we’re still learning a lot about how food works, the compounds in food, and how different foods interact.

Furthermore, the study was not long term. More meaningful results might be obtained with a more extended study tracking people over months or years. It would also be immensely helpful to have subjects keep complete food and activity diaries to generate a spread of data which could be used to look at other things which might be correlated with depression.

Chemically, chocolate is incredibly complex, especially when you start talking about all of the varieties of chocolate available. There could be something going on here, but it would take a lot more research to find out.

  1. The jury is out on this one science wise. Some studies suggest that compounds in chocolate are mood elevators although they may work only in the short term. Others suggest that there’s no effect on mood. Others suggest that compounds in chocolate might actually work the other way around. And, of course, many studies like to take the time to point out that chocolate makes people fat and fat is bad, ergo, people who eat chocolate will be sad because they are fat.