Response reliability in a longitudinal survey in Thailand.

The two rounds of the National Longitudinal Study in Thailand provide a useful opportunity to explore response reliability in a large-scale social and demographic survey in a developing country. The results indicate that nonrandom reliability at the individual level ranged from quite high (for several straightforward, facual questions) to quite low (for most attitudinal questions). There was considerable distributional stability, however, even for many of the variables with low individual-level reliability. In terms of its response reliability, the Thai study compares reasonably well with several leading US fertility surveys. However, in both countries response reliability at the individual level for attitudinal questions is distressingly low. This clearly should be a matter of major concern for social scientists using survey results.