The Verdict Problem

I have spent five years reading research on housework, money, rest and desire. In that time I have cited dozens of studies in these notes. This one is about what those studies can and can't tell us, and about a temptation I have felt myself.

Most of this research is correlational

Most of the findings I've written about show that two things tend to go together. When women carry more of the household's mental work, they report more stress. When housework is shared more equally, couples report more satisfaction or more sex. When women see their partners as dependent, their desire tends to be lower.

These are associations. They don't show which way the arrow runs, or whether something else is driving both. A couple who share housework equally may also communicate better, earn more, or have fewer children under five, and any of those could explain their happier relationship.

Researchers know this, and the good ones say so. Lucia Ciciolla and Suniya Luthar's 2019 study of mothers, Emily Harris and colleagues' 2022 study of desire, and Daniel Carlson and colleagues' 2016 study of housework and sex are all careful about their limits. The trouble usually starts later, when findings are summarised.

The temptation

This subject invites moral readings more than most. When a study links an unequal division of labour to lower satisfaction, it is very easy to read it as a verdict: he isn't doing enough, and this is the proof.

I have felt that temptation in these notes. The findings often match what women tell me, and they match my own sense of what is fair. But a correlation is not a judgement on any particular man or marriage. It describes a pattern across many people. It can't tell you what is happening in one kitchen.

Samples matter too

Many of the studies I rely on come from wealthy Western countries, mostly from different-sex couples, often from educated, middle-class samples. Some are small. Maume and colleagues' 2010 study of sleep involved 25 couples. Beban and Roberts' recent study of fair but unequal arrangements in same-sex couples involved 10. Small studies can reveal processes. They can't tell us how common something is.

Some famous findings have also been questioned since. Parts of the predictive claims associated with John Gottman's laboratory work on couples, which I have drawn on, have been challenged by other researchers. That doesn't make the work useless. It means it should be held lightly.

Why honesty about limits helps

It might seem that admitting all this weakens the argument. I think it does the opposite.

When a finding is presented as a verdict, the person on the other side of the argument has every reason to reject it. When it is presented as a pattern, with its limits, it becomes something two people can look at together. That is the conversation I described in Fairness, Felt and Measured.

So here is my commitment for the rest of these notes. When I cite a study, I will say what kind of study it was and what it can't show. Where the evidence is thin, I'll say so. And I will try not to use research as a weapon in arguments it was never designed to settle.

The aim was always to describe something carefully. That is harder, and it is more useful, than delivering a verdict.

Questions people ask

What is the difference between correlation and causation in relationship research?

A correlation shows that two things tend to occur together. Causation means one produces the other. Most research on housework and relationships shows correlations.

Can studies prove that unequal housework harms relationships?

Most can't prove cause. They show consistent associations between unequal arrangements and lower satisfaction, which are useful but should be read with their limits.

Why do sample sizes matter in this research?

Small studies can reveal how something works but not how common it is, and many studies come from wealthy Western, mostly different-sex samples.