Correlation vs. causation: Family dining edition

The New York Times makes a classic error, portraying the correlation between eating together as a family and various positive outcomes as a causal relationship:

The family dinner table, meanwhile, has mostly managed to retain something of a sanctified aura - the last meeting ground left in a world of working parents, relentless afterschool activities and an array of solitary technological amusements.

Science backs tradition: numerous research studies have shown that eating together as a family does everything from raise test scores to reduce the risk of behavioral problems, drug addiction and depression. One recent University of Michigan study found the communal dinner had more influence on a child's development than church-going or studying.

It's more likely, of course, that eating together is a proxy for various unmeasured characteristics of the family -- unless someone wants to claim that the meal itself is driving all of those factors? I doubt it.

Comments

[brent]

Corr != caus and all that, but the real error is the lack of citation. No doubt researchers at the University of Michigan are hip to causality and research design. They either exhibited some controls to test causality, made highly qualified claims of causality, or said nothing about causality and were misrepresented by the article.

It's possible that dining together imparts the sense of coherent family structure on young minds, allowing them more secure development and hence successful lives. Wouldn't be surprizing; I'd like to see some evidence for or against that (I have kids).

A citation, even partial (author name, journal and issue, even the actual year of publication!), would allow clarification.

[Michael Heinz]

I think the point is that eating together as a family encourages conversation and involvement - but, as you say, if parents are trying to create communication and involvement, dinnertime is merely one tool of many.

[brent]

One tool of many, perhaps. But still causal. Pushing a glass off a table will cause it to break, even though a physicist understands that its mass, susceptibility to gravity, fragility, and forceful interaction with the hard floor are more proximate culprits.

If communication and involvement are the key (just a hypothesis, and a testable one), they may explain how family dinner causes successful children (again assuming it does). But that doesn't diminish the causal statement.