Many attempts have been made to examine the determinants of victimisation and the fear of crime,
often guided by social disorganisation theory and environmental criminology. However, there
have been only a handful of studies that have been carried out in East Asia so far. Consequently,
it is unclear whether those factors which are reliably associated with higher or lower levels of
crime and fear of crime in Western societies are generalisable to the dissimilar context of Japan.
Against this backdrop, this thesis is concerned with the patterns and predictors of victimisation,
repeat victimisation, fear of crime and perceived risk of victimisation in Japan, drawing on
repeated cross-sectional data collected as part of a nationally representative household survey “the
Japanese Public Safety Survey” (JPSS) and the census.

Study 1 is concerned with the patterns and predictors of household property crime. Exploratory
factor analysis was first performed to reveal the factor structure of eleven perceived
neighbourhood disorder variables used in the JPSS. A series of multilevel logistic regression
models demonstrated that the year variables were found to be negatively associated with
household property crime risk. Detached house, homeownership, social support, and the presence
of community policing were found to be associated with household property crime risk.

Study 2 examined the patterns and predictors of repeat victimisation of residential burglary and
vandalism. In contrast with what was found in Study 1, the survey year variables were not
correlated with the risk of repeat residential burglary and vandalism victimisation. Social support
and university degree were found to be the factors which distinguish repeat residential burglary
victims from other groups. Social support and social disorder were found to be the factors that
distinguish repeat vandalism victims from other groups. Social support and high ratio of manufacturing industry were found to be the factors that distinguish repeat residential burglary
victims from single victims. Social disorder was found to be the factor that distinguishes repeat
vandalism victims from single victims.

Study 3 examined the patterns and predictors of fear and perceived risk of household property
crime. The results of multilevel regression models revealed that, at the individual/householdlevel,
experiencing previous victimisation, being older, living in a detached house and having
higher annual household income were associated with increased fear of household property
crime. At the neighbourhood-level, the presence of social disorder and community policing
were statistically related to the levels of fear of household property crime. There was a
statistical association between prior victimisation and perceived risk of victimisation, and
different predictors were found to be associated with fear of crime and perceived risk of
victimisation. The survey year variables were not found to be associated with fear of and the
perceived risk of household property victimisation. The findings from the analysis furthered
support the three models of fear of crime.

In summary, the findings of three empirical studies yielded both consistencies and inconsistencies
with the relevant literature derived mainly from studies conducted in Western industrialised
countries, showing some applicability of the criminological theories to Japan. The thesis
demonstrated the usefulness of multilevel modelling and multiple secondary data sources, and the
importance of introducing measures dealing with neighbourhood social disorder, and crime
prevention measures which reflect the crime trends or related problems of each municipality.