According to the Situational Action Theory, the “perception-choice process” is the explicit mechanism explaining how the interaction between an individual and a setting might render in a rule-breaking act. One way of studying the process has been through a randomized vignette approach. Prior research has generally found that crime prone individuals assessing criminogenic settings are more likely to choose rule-breaking alternatives. However, less is known if this association also holds when comparing an individual with him-/herself over two time points. Using data from the longitudinal project Malmö Individual and Neighbourhood Development Study (a replication of PADS+), this study examines if individuals who report stability [or change] in level of crime propensity also report stability [or change] in the assessment of vignettes. Furthermore, scholars have recently challenged the view that self-reported accounts of attitudes do not generalize to actual behaviour, proposing a Bayesian approach. This study follows the suggestions, and results are discussed against a backdrop of theoretical implications.