Cross-Cultural Invariance and Gender Bias when Measuring ‘Fear of Crime’

Abstract

The persistent questioning within the ‘fear of crime’ literature of its own conservative measurement, has resulted in a growing tendency towards using scaling techniques, as a far better way of measuring a complex and multidimensional concept like ‘fear of crime’. Choosing more complex measures, however, gives rise to a number of ‘new’ issues related to measurement error, which have remained mostly undiscussed in the ‘fear of crime’ literature. Cross-cultural measurement invariance is such an issue; in order for cross-cultural comparisons of fear of crime to be meaningful, the instruments used to measure the theoretical constructs have to exhibit adequate cross-cultural equivalence. Gender bias is another potential danger to measurement invariance; a substantial part of the ‘fear of crime’ literature has empirically found and discussed different ‘fear of crime’ levels between men and women. Without testing the assumption of measurement invariance it is unclear whether these gender differences in ‘fear of crime’ are a reflection of true differences, or the result of a different understanding of the central concept between men and women. Using exploratory factor analysis and structural equation modeling, both cross-cultural invariance and gender equivalence will be assessed in this contribution, based on the 1999 General Election Study of Belgium (N = 4239).