Abstract: | Four causal models describing the longitudinal relationships between attitudes and achievement have been proposed in the literature. These models feature: (a) cross‐effects over time between attitudes and achievement, (b) influence of achievement predominant over time, (c) influence of attitudes predominant over time, or (d) no cross‐effects over time between attitudes and achievement. In an examin‐ation of the causal relationships over time between attitudes toward science and science achievement for White rural seventh‐ and eighth‐grade students, the cross‐effects model was the best fitting model form for students overall. However, when examined by gender, the no cross‐effects model exhibited the most accurate fit for White rural middle‐school girls, whereas a new model called the no attitudes‐path model exhibited the best fit for these boys. © 2002 Wiley Periodicals Inc. J Res Sci Teach 39: 324–340, 2002 |