Levels of reliability in fertility survey data.

A number of factual and attitudinal questions asked in the 1973 Taiwan KAP-4 survey were repeated in a postenumeration survey one month later in order to assess the reliability of responses of the 286 women reinterviewed. The level of reliability is found to vary depending on the measures used and on whether the focus is aggregate data or individual responses. Analysis of consistency of responses shows that while overall reliability at both aggregate and individual levels is reasonably good, there is greater reliability for factual than for attitudinal data. Nevertheless, consistency of responses on factual questions varies considerably depending on the salience of the topic to the respondent. Estimates of reliability are shown to depend on the measure used and on the skewness of the distributions of the responses.