| Abstract: | Summary The validity and usefulness of 'desired additional children' and 'ideal family size' as predictors of fertility are analysed in this paper on the basis of longitudinal survey data from Thailand. First, the extent of measurement error in these variables is considered, and it is concluded that the error variance and the true variance are of similar orders of magnitude. Secondly, the changes in attitudes subsequent to births and deaths of children are investigated. It is found that the number of additional children desired is decreased by births and increased by deaths, but less than would be expected if 'desired additional children' represented an unchanging target family size. 'Ideal family size' is almost unaffected by births and deaths. Thirdly, the contribution of attitudinal variables to behavioural models is examined. It is found that desired fertility is explained no better than fertility in a standard economic model. A birth function separating desired children from identifiable physiological factors as explanatory variables indicated that the former was just significant. A model of contraceptive acceptance also found desired fertility to be a significant determinant. Thus, desired fertility can be successfully integrated into behavioural models. But on the whole, its explanatory power was weak, and it was concluded that the independent use of this variable does not significantly improve on models which relate fertility to socio-economic variables directly. |
