Time-use surveys — where respondents record what they did in fixed intervals across a day — are among the more reliable instruments in social science, because they measure behaviour rather than recollection.
They consistently find that women perform more unpaid domestic and care work than men, in every country where the data has been collected, with the size of the gap varying substantially.
What is less frequently noted is that the same surveys find the gap present in childhood.
The childhood data
Time-use studies including children find girls performing more household work than boys of the same age, with differences appearing in the primary school years and widening through adolescence.
The composition differs as well as the quantity, in the pattern described in this site's article on allowances: girls' tasks are weighted towards recurring indoor work and care of siblings, boys' towards episodic outdoor work.
The distinction between recurring and episodic is the one that persists into adulthood and does most of the work in explaining why the adult gap is so resistant to change.
Why recurring work is different
Episodic work has a completion state. It is noticed when done and produces a visible result.
Recurring maintenance work has no completion state. A kitchen cleaned today requires cleaning tomorrow. Its output is the absence of a problem, which means it is noticed only when it stops.
This has three consequences that appear throughout the literature.
It is systematically undercounted, including by the people doing it, because there is no artefact to point at.
It cannot be scheduled or deferred in the way project work can, which constrains everything else in a way that episodic work does not.
And it generates no credit, which matters in any setting where contribution is assessed.
The parallel with the non-promotable-task research in workplaces is close enough to be worth stating: the same distinction between visible episodic work and invisible recurring work, with the same distribution and the same consequence for who accumulates a record.
The management layer
Beyond the tasks themselves is the work of knowing what needs doing — tracking, anticipating, remembering, delegating.
This is difficult to capture in time-use surveys, because it does not occupy discrete time blocks and is frequently performed simultaneously with other activity.
Research using other methods consistently finds it distributed more unevenly than the tasks themselves, and finds it cited as the more burdensome component by those carrying it.
The reason it is more burdensome is that it cannot be delegated without being performed. Asking someone to do something requires having already noticed it needed doing, decided who should do it, and initiated it — which is most of the work.
Adolescents encounter this early. A young person who is asked to help is in a different position from one who is expected to notice, and time-use research suggests the expectation of noticing is unevenly applied.
What the adult data shows about persistence
The gap has narrowed over recent decades, mostly because women's hours of domestic work have fallen rather than because men's have risen substantially.
The narrowing has slowed. Analyses of trends in several countries find convergence stalling.
The gap persists when both partners work full time. It persists when the woman earns more. Studies examining this have found that the relationship between relative earnings and domestic work distribution is weaker than economic bargaining models predict, which is one of the more robust anomalies in this literature.
That anomaly is the main reason purely economic explanations are considered insufficient, and the main reason attention has turned to norms formed earlier.
The consequences that are not about fairness
Unpaid work reduces available hours for paid work, which affects earnings, pension accrual and career continuity.
It is unrecorded in national accounts, meaning a substantial portion of economic production is invisible to policy.
And it affects who can accept opportunities that require unpredictable time — which is a large proportion of the opportunities that lead to advancement.
What can actually be changed
In households with children: audit the actual distribution rather than the perceived one, because both parties systematically misestimate. Rotate the recurring invisible tasks rather than only equalising totals. Assign the noticing, not only the doing — being responsible for a domain including knowing when it needs attention is a different assignment from being asked to help with it.
In policy: parental leave design has the best evidence. Non-transferable leave reserved for fathers has been associated with increased paternal involvement in several countries that have implemented it, and the effects appear to persist beyond the leave period in some studies.
Measurement matters too. Countries that conduct regular time-use surveys have better information about a form of work that is otherwise invisible, and several have begun including estimates in satellite national accounts.
None of this is dramatic. But the finding that the distribution is established in childhood implies that the intervention point is childhood, and almost nobody treats a household chore rota as a policy instrument.