We now have added "Informational Posts" which are tidbits of information that may come in handy at some point.


2001 (SO Policy):

This paper explores the impact of sex offender policies by examining sex offenses through time series analysis. Using monthly count data of rapes aggregated at the state level, this analysis uses Box-Jenkins autoregressive integrated moving average (ARIMA) models to conduct ten separate intervention analyses on the enforcement of Megan’s Law. The results of the interrupted time series analyses are mixed with regard to whether the enforcement of sex offender registration had a statistically significant effect on the number of rapes reported at the state level. Although several states showed a nonsignificant increase in the number of rapes, only three states had a significant effect on decreasing the number of rapes. Policy implications of this research are discussed in terms of the efficacy of sex offender registration and whether changes in these laws should be considered.


Sex offender registration and notification policies are a relatively recent development in the criminal justice system. As such, there has been neither a great deal of methodological development nor many empirical findings. This research has attempted to overcome this lack of research by empirically exploring the effect of sex offender registration and notification laws on the number of sex offenses, measured by rape, committed over time across the United States. Utilizing time series analysis (ARIMA), we examined the incidence of rape in a sample of 10 states to see if the sex offender laws had any noticeable influence on sex offenses.

The key finding of this research is that the passage of sex offender registration and notification laws have had no systematic influence on the number of rapes committed in these states as a whole. Most of the states in our sample (five of ten) showed no significant differences (increase or decrease) in the average number of rapes committed before and after the sex offender laws. These non-significant findings masked the fact that roughly half of these non-significant changes were actually increases in the average incidence of rape within a given state. Of the states (four of the ten states) that did indicate statistically significant findings, three of these states experienced a decrease in the number of rapes after the implementation of the laws. The fourth state, however, actually illustrated a steep, statistically significant increase in the number of rapes committed.

These findings beg the question of what is occurring concerning the relationship between sex offender registration and notification and subsequent sex offending. It is possible, as Sample’s (2001) research indicated, that this was knee jerk legis lation based simply on keeping the constituency happy. If that is true, sex offender registration and notification laws are failed legislation. As discussed above, there are alternative explanations of what could be occurring. These include a deterrent nature to these laws, a heightening of police awareness, both, or the rape counts simply decreased at the same time as the introduction of the law and therefore our model could be misspecified (i.e. confounded). We offer several suggestions of how these differing explanations may be producing the findings of this research.

First, sex offender registration and notification laws are based on the assumption that sex offenders are more likely to recidivate than other offenders. The research on the validity of this assertion is very mixed. While the nature of our data does not permit us to comment one way or the other in this area, sex offenders are assumed to be an identified group. As such, the rationale behind sex offender laws would assume that a majority of the offenses committed within the confines of any year would actually be sex offenders who are re-offending. A potential alternative assumption is that in those states with increases (roughly half of the states examined here), it would be expected that these offenders would be first time offenders or offenders who have never been caught before. This is supported by Walker and McLarty (2000) who found that 73% of sex offenders in their study committed a sex offense as their first offense. If this is the case, it would be
expected that sex offender registration and notification is not able to control this population because a substantial number of them are committing sex offenses as their first offense. Hence, there is no name on the register and no way to inform the community.

Second, these laws may be having no influence on the number of sex offenses because communities are not actively using sex offender registries to protect their communities to the fullest extent possible. Community members may be desensitized to different notification strategies. For instance, Walker et al.’s (2001) research showed that in many cases in Little Rock, Arkansas, registered sex offenders in a majority of cases seemed to choose homes that clustered tightly around day cares, schools, and parks. This means that a potential exists for a motivated [sex] offender to come into contact with a suitable victim (children, in this instance) where there is a lack of some suitable guardian (teachers or parents) according to Cohen and Felson’s (1979) work on the routine activities perspective. The significance of this finding is that the community is apparently doing nothing about these offenders.

Finally, increases in the average number of sex offenses may reflect an increased scrutiny from both communities and the police who are continually updated on the presence of sex offenders. Since there is an increase in the average number of sex offenses in half of the states examined here, police practices, in concert with community support, may now be focusing on more sex offenders. This would lead to an increase in the average number of sex offenses since law enforcement effort is now focused on a predetermined population that is relatively easy to find.

Based on the findings of this study and the potential conflicting explanations, future research on sex offender registration and notification policies should explore several different paths. First, a more appropriate unit of analysis would be the city. Cities may be preferable because aggregate level time series data suffers greatly due to binning. As such, using smaller “bins,” such as cities, might give more insight than using state level data, which loses detail in comparison. Second, because, sex offender registration is mainly addressing repeat offenders, future research should focus on sex offender recidivism before and after the enforcement of sex offender laws. Finally, future research should address sample size. While we intended to conduct time series analysis on all 50 states, after getting data from states with adequate follow- up time and excluding state data that did not conform to time series analysis, our final sample was only 10 states. Data mining techniques are constantly improving, and the follow-up periods in states now allow for at least a three year follow- up. Future time series research on sex offender policies should also explore these states.

For remainder see study by Jeffery T. Walker, PhD, University of Arkansas at Little Rock and Sean Maddan, Statistical Systems, Inc. and Bob E. Vásquez, University of Albany, SUNY and Amy C. VanHouten, Statistical Systems, Inc. and Gwen Ervin-McLarty, Director Statistical Analysis Center Arkansas Crime Information Center

Note: Chart showing when registries were implmented.

No comments: