Longitudinal analyses It is known that risk perception measures prospectively predict risk-taking behaviour (Floyd, Prentice-Dunn, & Rogers, 2000; Harrison et al., 1992; van der Pligt, 1998). To further assess the validity of study measures, a similar analysis was conducted on the currentdata set. Dangerof 70km/h ina 60zoneat T1 was correlated with speeding at T2 (rð253Þ ¼ 2:30, p , :01), as was T1 danger at 80km/h (rð253Þ ¼ 2:18, p , :05), and T1 risk beliefs (rð253Þ ¼ 2:35, p , :01). A hierarchical regression was conducted to predict T2 speeding from T1 risk perceptions. At Stage 1, age, gender, and T1 speeding were added. This resulted in an R 2 (df¼3,251) of .300. The three risk perception items were added at Stage 2. A significant increase in R 2 was noted (DR2 ð3;248Þ ¼ :034, p , :01). Due to the T1 correlations between the risk perception items, the only significant multivariate predictor was risk beliefs (standardized b ¼ 20:16, p , :05). This offers some support for the validity of measures used in the study. Table 3 shows univariate and multivariate predictors of T2 risk perceptions. Univariate correlations showed that both T1 speeding and T1 to T2 speeding change inversely predicted danger at 70km/h and risk beliefs, whilst T1 speeding inversely predicted danger at 80km/h. Three hierarchical regressions were used to predict T2 risk perception variables controlling T1 scores on the relevant variable, age and gender. Logistic regression was used for the danger 80km/h analysis, linear regression for the others. T1 scores, age, and gender were entered first, followed by T1 speeding and speedingchange.Ageandgenderwerenotsignificantpredictorsandwereremovedfrom all analyses. The first step was significant in all equations. The addition of the step 2 variables produced significant model improvements for all equations. Both T1 speeding and the speeding change score were significant inverse predictors of all three T2 risk perception scores. Details of these analyses can be found in Table 3. Table 1 Item Means and SD at Time 1 and Time 2, and Changes from T1 to T2 (N ¼ 255).