|dc.description.abstract||This research aimed to determine whether sex trafficking-related offenses in Austin, Texas, clustered geographically and, if so, whether such patterns could be predicted by the spatial distribution of other prostitution and drug-related crime, proximity to highways, and the presence of facilities known to be associated with the sex trade (ATMs, gas stations, and cheaper hotels), as well as neighborhood variables capturing the extent of social disorganization.
A variety of methods were employed, including nearest neighbor analyses, kernel density estimation and thematic maps, non-parametric correlations, and a hurdle regression model. Data for all sex trafficking and compelled prostitution crimes recorded by Austin Police Department (APD) during 2013-2015 were used as the outcome variable. Data for the predictor variables came from APD and several public sources.
Most of the analyses employed the census block group as the unit of analysis, of which there are 519 in the research area.
Results showed sex trafficking-related offenses significantly clustered in space. Bivariate correlations revealed sex trafficking activity to be moderately correlated with prostitution and drug-related crime, and weakly associated with most of the remaining variables. In the multivariate models, however, only drug-related offenses and the percentage of households on public assistance were shown to be significant predictors of whether a census block group were affected by sex trafficking (i.e., at least one offense recorded; binary logistic regression); in affected block groups, the actual incidence of sex trafficking was predicted by prostitution- and drug-related offenses, the combined distance from the block group to the three highways considered, and the number of ATMs and cheaper hotels (<$100) weighted by their proximity to highways (Poisson regression).
The limitations of police-recorded sex trafficking data, and how these more likely represent the transaction stage of trafficking (when sex/victims is/are sold), may explain the weak/nil results obtained, as some of the predictors considered might be more closely associated with other stages of the trafficking process such as recruitment and harboring. In any case, the significant effects detected are consistent with existing theory and evidence, and the spatial concentrations observed call for law enforcement and other agencies to target their efforts in highly affected areas.||