Further reflections on changes in fertility expectations and preferences

In this paper we have shown that the statements made in our original paper and disputed by Sloane and Lee are not ". . . . .at least partially in error." Concerning the first two areas of disagreement, Sloane and Lee's analysis did not address the conclusions that we reached and, therefore, could not refute them. In the first instance the observations of Sloane and Lee are useful in understanding how one of the effects we observed occurred, but in the second area the usefulness of their analysis is not at all clear. The third area of disagreement resulted because Sloane and Lee paid insufficient attention to the observed sample relationships and used a statistical test that is inefficient for the type of data used. Of course, our disagreement is not with log-linear methodology, but with its inappropriate application and interpretation by Sloane and Lee. In two instances they used results from the log-linear procedures to address issues that could not be addressed by those particular results, and in another instance they used a log-linear test of statistical significance when a more efficient alternative provides a different picture. The problems associated with the Sloane and Lee analysis illustrate the importance of precise specification of the substantive issues being investigated and the careful interpretation of the substantive implications of the original data and statistical results.(ABSTRACT TRUNCATED AT 250 WORDS)