The findings show that questions in both surveys showed some change in the types of information they sought at the two time periods, which means they did not measure the same type of respondent perceptions/feelings at two time periods. This was true for approximately two out of four questions. This variation in what was measured thus rendered invalid any conclusions about comparisons of these measurements at the different times. The authors suggest possible ways of dealing with such variations in measurements, but still advise that future research could result in more specific guidelines. One of the datasets assessed was the Belgian Safety Monitor, which is a Federal victim survey repeated every 2 years through computer-assisted telephone interviewing. In addition to seeking information on victimization, the survey contains questions on crime prevention, police functioning, perceived neighborhood problems, and fear of crime. The current analysis used the 1998 (n=6,535), 2000 (n=6,090), and 2002 (n=5,777) waves of the survey. The second dataset examined was the APS survey (Administration Planning and Statistics) of the Flanders Authority. Since 1996 this survey has been used as an annual measure of sociocultural changes in Flanders. It involves only one segment of the Belgian population. Some of the question blocks rotate over each year; this is the case for the set of questions on fear of crime. This assessment examined fear-of-crime questions for the 1999 (n=1,186), 2000 (n=1,163), and 2002 (n=1,325) questionnaires. 4 tables, 6 figures, 20 notes, and 28 references