Figures & data
Figure 1. The combined within-subject and between-subject design. At selection (t1), all respondents were applicants and went through the same admission testing situation. At retest (t2), the same applicants were by now incumbents, and were randomly assigned to one of three response conditions.
Figure 2. The multigroup means and covariance structure model, containing one-factor models at selection (t1) and at retest (t2), and the covariances. The λs denote the factor loadings, the τs the item intercept, the θϵs the residual variances, the rcs the residual covariances, the ξs the latent trait means, the Φs the latent trait variances, and rs the latent trait covariance.
Table 1. Model fits and model comparisons for the comparability of the three conditions at selection (t1), the measurement invariance (MI) across selection (t1) and retest (t2), and the comparison of latent retest correlations of the scale activity in familiar communicative situation.
Table 2. Model fits and model comparisons for the comparability of the three conditions at selection (t1), the measurement invariance (MI) across selection (t1) and retest (t2), and the comparison of latent retest correlations of the scale preventive health behavior in response to warning signals.
Table 3. Model fits and model comparisons for the comparability of the three conditions at selection (t1), the measurement invariance (MI) across selection (t1) and retest (t2), and the comparison of latent retest correlations of the scale self-confidence in test situation.
Table 4. Comparison of selected unstandardized parameter estimates of selection (t1) to the three conditions at retest (t2) for all three scales. For selection (t1), the unstandardized parameters are given. For re-text (t2), either the unstandardized parameters are given, or = t1 in case of invariance across measurement points.
Table 5. The latent retest correlations of the final models of all three scales.