Modelling domestic work time

What variables should be used as regressors in models of the length of time which people spend doing unpaid domestic work? To most economists, the answer would be straightforward: use the variables which are implied by a theoretical model of household time allocation (e.g. Becker's). This paper shows that this strategy has not been followed, explores why this is so, and makes some recommendations about variable specification and the treatment of paid market work time in particular. The arguments are illustrated using regressions based on UK time budget data for the mid-1980s.