Consistency of Retrospectively Reported Individual Fertility Data Across Time

点击翻译按钮获取中文摘要

Project Abstract The conditions in which men and women have children are widely studied, with two distinct bodies of research: (1) predicting who has births and the characteristics of those births, and (2) analyzing how birth characteristics influence later outcomes. Both areas are highly dependent on survey data, with individuals retrospectively reporting about their births. Birth characteristics range from subjective conception factors (such as relationship status or whether the pregnancy was intended) to behaviors during pregnancy (such as smoking or prenatal care) to more objective data (such as gestational age or birth weight). These characteristics are then used as either dependent or independent variables to identify the predictors and consequences, respectively, of various birth characteristics. Rarely, though, have researchers considered whether this retrospectively-reported data – data produced by asking individuals to recall statuses, feelings, behaviors, and details of past events – is reliable. If input data is flawed, then research using such data risks drawing inaccurate conclusions. The limited research directly investigating retrospectively-reported fertility survey data has indeed found problems – some men’s births are not reported at all, and a substantial minority of young women change how they retrospectively characterize the intendedness of a birth when asked about the same birth at different points in time. This suggests that retrospective reports for other birth characteristics may also be prone to inconsistencies over time. The current project thus investigates an important but untested assumption: that mothers and fathers are consistent in how they report about past fertility. The project tests this assumption with the only two longitudinal datasets (the National Longitudinal Study of Adolescent to Adult Health and the Toledo Adolescent Relationships Study) that collect full fertility histories at multiple waves, which means that individuals have multiple opportunities to report characteristics for the same birth(s). The proposed research will accomplish two goals. First, it will establish whether respondents report consistently about their births, investigating a wide range of birth measures. This will a) identify which characteristics are most likely to be reported consistently, b) leverage differences in surveying to determine how survey frequency and structure influence consistency, and c) establish whether there is systematic variation in who reports consistently. Second, this project will answer a more fundamental question: does inconsistency affect the conclusions drawn about the relationship between birth characteristics, on the one hand, and potential precursors and outcomes, on the other? In parallel analyses using birth characteristics as either the dependent or independent variable – and varying the waves from which these characteristics are drawn – models will be compared to identify whether the associations between precursors/outcomes and birth characteristics are sensitive to inconsistent reports. This fits squarely within the mission of the NICHD Population Dynamics Branch Fertility and Infertility Program and has both practical and substantive implications for fertility research and public health programs.