Reliability estimates of some survey data on family planning

Summary A follow-up study employing the panel design which aimed to evaluate the temporal stability of rural respondents' answers to KAP and fertility-related questions after a lapse of nearly five months revealed that reports by females on factual data dealing with socio-economic and demographic topics are as highly reliable as are those of males. The study based on 79 males and 81 females in Haryana further supports the hypothesis that re-test reliabilities of factual data such as respondent's age, religion, occupation, etc. are substantially higher than those of knowledge about contraceptives. Attitude items showed very low re-test reliabilities although the internal consistency reliabilities of the two attitude scales were found to vary within the range of 0.72 to 0.88. Both attitude scales were found to have a very high reproducibility coefficient, all higher than 0.95, for both the interview sessions. Re-test reliabilities of data on contraceptive use were also found to be moderately high. However, the rates of loop insertion and condom use as reported by female respondents are likely to show a substantial amount of gross error as is evident from their low re-test reliabilities. The findings of the present study cast serious doubt on the assumption that married males are not capable of accurately reporting the pregnancy histories of their wives. Several implications of the findings are discussed. From the methodological point of view they suggest a need to develop improved techniques for assessing attitudes toward family planning programmes and abortion as well as to knowledge of contraceptive methods and desired family size. The findings also justify the inclusion of eligible male respondents in KAP and fertility surveys.