Recently, there has been a transition in the survey that measures perceived and actual safety in the Netherlands. This transition has caused significant discontinuities, especially for attitude questions related to the perception of the neighbourhood and the evaluation of the police performance. Diffrent influences of the response order effect and social desirability bias and their interaction may explain these.

Before the transition, the Safety Monitor (SM) was conducted annually with a net sample size of about 20,000 respondents nationwide. Parallel to the SM local authorities collected data on the same topics at a regional level, which gave rise to inconsistent regional and national data about safety feelings and crime victimization. Because of this, the SM and the regional surveys were combined into one new survey, the Integrated Safety Monitor (ISM). Apart from other differences between the methodologies of the two surveys, the ISM employs Web Survey and Self Completion Paper Questionnaires in addition to Computer Assisted Personal Interviewing and Computer Assisted Telephone Interviewing as used by the SM. In autumn 2008, the ISM was performed for the first time with a net sample size of 62,803 respondents.

To assess the effect of the modified methodology, the SM has been executed in parallel with a limited net sample size of 6,113 respondents. Comparison of ISM with SM results reveals that the survey transition has caused discontinuities in some key safety indicators that can be traced back in part to the different sets of response modes employed in the two surveys and their related effects.The role of selection effects may be limited.