Birth order and family size have been studied in a representative sample (10,000) of the adult population of Great Britain. This was done to test theoretical predictions about the ways in which the birth-order distribution in a population sample will be biased by secular changes in family size. The principal predictions were that, in a sample born during a period when family size was decreasing, the distribution of birth ranks in the larger families would be biased towards an over-representation of later born subjects (i.e. those lower in birth order), while in smaller families there would be an under-representation of later born. In a sample born during a period when family size was increasing, the reverse effects would occur. It was also predicted that among subjects from large families there would be an excess of females; and that the extent of the bias due to decrease in family size would be greatest in the unskilled and least in the highly skilled social classes.
The findings confirmed these predictions. The sample also showed an un-predicted excess of subjects in the middle ranks of families. It is suggested that the hypothesis of birth rate bias may be used for predicting the type of birth-order distribution to be expected in a sample of any population.