The objective of the proposed research is to apply new methods of analysis to improve the utility of attitudinal data in predicting fertility-related behaviors. Demographers have struggled for decades with the difficulty of predicting fertility at the individual level using attitudinal measures. A key question is to what extent weak associations between attitudes and behavior result from a true disjuncture between cognition (what people think) and behavior (what they do), and to what extent they are an artifact of inadequate measurement of cognition. Our project focuses on the second possibility. In this project, we use an innovative conceptualization of fertility-related attitudes that draws on advances in psychology regarding cognition and behavior. Specifically, we take attitude measures as meaningful in relation to other attitude measures, and we consider patterns of relationships between attitude measures as a proxy for patterns of cognitive associations. The specific aims of the proposed research are (1) to use novel methods for analyzing fertility-related attitudinal data to group respondents who think similarly about fertility, and to describe the patterns of cognitive associations and sociodemographic characteristics of the groups identified; and (2) to determine whether these methods predict behavioral outcomes (in this case, contraceptive use) better than do conventional methods of analyzing attitude data. This project addresses recent calls by prominent demographers to use models of cognitive processes and demographic behavior that reflect advances in scientific knowledge about cognition and behavior (e.g., Bachrach 2014; Bachrach and Morgan 2013; Johnson-Hanks et al. 2011). Our analyses will use the rich data from the Relationship Dynamics and Social Life study, a longitudinal survey of a population-based sample of young women. The data contain multiple measures of attitudes in many domains related to fertility behavior, including attitudes about motherhood, relationship sequences, contraception, timing of childbearing, career, and education. The data also include detailed measures of contraceptive use, allowing us to relate attitude structures to behavioral outcomes. We will use two types of methods, Latent Class Analysis (LCA), a measure that creates latent groups based on sharing similar patterns of responses across a series of variables, and Relational Class Analysis (RCA), which identifies sub?groups of respondents who share similar relationships among their responses. These methods allow us to distinguish among respondents who give the same answers to some items but who think about fertility in fundamentally different ways. This research will test theories regarding the role of attitudes in fertility behavior and provide replicable procedures for analyzing the predictive value of survey-based measures of attitudes.