Error Patterns in the Prediction of Fertility Behavior1

Authors


  • 1

    This research was supported by Research Grants HD-07225001A1 and HD-07024-01, 02, 03 from the National Institute of Child Health and Human Development, Center for Population Research. The authors are grateful to their research collaborators, Frederick Campbell, James Jaccard, and Brenda Townes for their helpful insights and suggestions. We also thank Shirley Feldman-Summers, Gary McClelland, and Harry Triandis for their comments on an earlier version of this article.

Requests for reprints should be sent to Andrew R. Davidson, Department of Psychology (NI-25), University of Washington, Seattle, WA 98195.

Abstract

Two separate three-wave, 2-year, longitudinal studies investigated error patterns in two decision models' predictions of couples' birth-planning decisions. One study tested the subjective expected utility model with a sample of married couples from the state of Washington and the other study tested the Fishbein model using a sample of married women from Illinois. Consistent with the proposal that predictions of deviation from the status quo result in a higher proportion of prediction errors than predictions of nondeviation from the status quo, both studies demonstrated that when couples were predicted to decide not to become pregnant, predictive accuracy was very high, but when couples were predicted to decide to become pregnant, predictive accuracy was significantly lower. Two hypothesized explanations for the biased error pattern were considered and received some support. The first, resembling the concept of inertia, is that a stronger motivational force is required to compel a person to elect a change from the status quo than is required to maintain the status quo. The second focuses on changes in decisions as a function of time and events. The heuristic value of directly examining prediction errors, in contrast to the more prevalent practice of reporting only one summary index of goodness of fit, was discussed.

Ancillary