Overclaimed housework

Why do both partners' honest estimates of housework add up to far more than everything done?

Ask two people who share a home, separately and in good faith, what percentage of the housework each of them does. One says seventy. The other says sixty. The house does not contain a hundred and thirty per cent of a house.

The obvious reading is that at least one of them is exaggerating, and each partner will privately settle on which one. But run the experiment across many couples and the pattern holds almost everywhere, in both directions, including for people who have every reason to be scrupulous. A bias that reliable is not a character flaw distributed at random. It is telling us something about how the estimate is produced.

Separate two quantities that the argument keeps welding together. There is the work done — a real amount, in principle countable, that a stopwatch and a clipboard could settle. And there is each partner's estimate of their share, which is not read off the world but constructed on the spot, in a few seconds, from whatever memory offers up.

So ask the mechanical question: what does memory offer up? To answer "how much of the laundry do I do?", nobody consults a ledger. They try to recall instances, notice how readily instances come, and convert that ease into a proportion. Whose instances come more readily?

Your own, and not by a small margin. You were present for every load you carried and for none that you did not. Your own effort arrives with the effort attached — the aching back, the interrupted evening, the resentment of doing it at eleven at night — while your partner's arrives, if at all, as a finished result: the basket is empty, and an empty basket carries no timestamp and no cost. Retrieval is doing the work of measurement, and retrieval samples your own contributions almost exhaustively and theirs sparsely.

That is a decent story, but it is also exactly what a self-serving liar's behaviour would look like from the outside, so it needs a test that separates the two. Here is the test, and it is the reason this case is worth reasoning through at all. If the inflation is self-flattery, then people should overclaim good contributions and underclaim bad ones. If it is a retrieval effect, they should overclaim both, because a bad thing you did is still a thing you were present for.

Ross and Sicoly ran essentially that study in 1979, having married couples estimate their responsibility across a list of household activities. Self-estimates summed to more than the whole, as expected. The decisive part is that the same overclaiming appeared for undesirable items, such as responsibility for causing arguments. People took more than their share of the blame by the same mechanism that gave them more than their share of the credit.

That result does not abolish self-serving motives — they are real and they surely ride along. What it does is establish that a purely motivational account is not necessary, because a purely informational one already predicts the sum exceeding a hundred per cent.

Push the informational account one step further and it explains the specific flashpoints. Some domestic work leaves a visible trace: a mown lawn, a cooked meal, a clean floor. Some leaves almost none — noticing that the school form is due, holding the shape of the week in your head, remembering which relative was owed a call. Work of the second kind is nearly unobservable even to an attentive partner, so it is not merely undersampled but structurally invisible, while being fully available to the person doing it. If overclaiming is driven by asymmetric access to evidence, then the partner carrying more of the invisible work should feel the largest gap between what they do and what is credited — which is the complaint that gets voiced most often.

Three refinements matter.

First, this is not a claim that the underlying division is fair. Time-use surveys across many countries find genuinely unequal splits, and the perception effect sits on top of that reality rather than replacing it. Two things are true at once: the split may be unequal, and both partners' estimates of it are inflated. Using the second to dismiss the first is a misuse of the finding.

Second, the effect is not confined to couples. Ask any group of collaborators for their percentage contributions to a joint project and the total will exceed a hundred, and the excess grows with group size — more collaborators means more contributions that any one member never observed. Housework is a special case of a general accounting problem.

Third, the practical implication is not "try to be more honest", because honesty was never the missing ingredient. Anything that supplies the missing evidence does more than exhortation: keeping a shared list for a fortnight, splitting by task rather than by feel, or — the cheapest version — describing the invisible work out loud, which converts something structurally unobservable into something observable.

• Whether the same asymmetry explains disputes over emotional labour and household planning, where the trace left behind is thinnest.
• How the effect changes with more household members, given that the surplus grows with the number of contributors.
• What time-use diary studies show about the gap between measured hours and reported share, and how large that gap typically is.

Every term the collection defines is gathered in the glossary.