Are Our Samples Representative?
The samples discussed in this report represent the results of a complicated process of selection. Our samples were drawn from the foster children of carers who had responded to a questionnaire (the General Questionnaire) sent to all foster carers in the seven authorities and who agreed to answer a further questionnaire on at least one foster child. However, the questionnaire about a specific child was sent some six months after the General Questionnaire, and so the sample had to be selected from those who were still fostering and had a child at that date.
As can be seen, this process allows for a variety of sources of bias. The final sample is influenced by response rates to the General Questionnaire, the proportion agreeing to go forward to the next sample, the proportion of those who had an eligible foster child at the appropriate date, the proportion of those sent a questionnaire and so on. The sample will be biased if these processes depart from random selection–for example, if older carers were less likely to respond to the first questionnaire than younger ones. This appendix examines these possible sources of bias and their relevance to the questionnaires returned by foster carers, social workers and children.
Before discussing these issues we should point out that according to usual criteria our samples pass as highly representative. As has been seen (Chapter 2), they reflect almost exactly the national picture in terms of the age and sex of the children, the length of time they have been in the care system, and their legal status.
We showed in Appendix 1 of our first book (Sinclair etal. 2004) that, as far as could be judged from national statistics, our local authorities were, taken as a whole, highly representative. They had almost exactly the right proportion of children looked after per 10,000 population, and the right proportions of these fostered, in children's homes or placed for adoption.
We showed in the same appendix that our sampling frame may have excluded some carers who were fostering relatives and were supported by area social workers