Prevalence of Male Clients of Street Prostitute Women in the United States

Human Organization, Fall 2008 by Brewer, Devon D, Roberts, John M Jr, Muth, Stephen Q, Potterat, John J

Survey estimates of the prevalence of clients of prostitute women are biased because men underreport sex with prostitutes. We conducted capture-recapture analyses of prostitution arrest records in several United States metropolitan areas and found that about two to three percent of adult male residents patronized local street prostitutes during observation periods of two to five years. An estimate from Colorado Springs, based on the prevalence of local prostitutes, the mean number of their client sex partners, and clients' mean number of prostitute sex partners, showed a client prevalence of 3.5 percent for a one-year period. These prevalence estimates were almost twice as large as those based on self-reports in the General Social Survey. There was no increasing or decreasing trend in client prevalence over time. Furthermore, almost three-quarters of clients identified in a Colorado Springs study patronized on the street, suggesting that off-street activity accounts for a fairly small portion of prostitution in that city.

Key words: prostitution, capture-recapture, measurement, self-report, survey

Introduction

Prostitution is often associated with illegal drug markets, crime, violence, other negative impacts on neighborhoods, and sexually transmitted disease. Despite the problems of public safety, order, and health linked to prostitution, there is little good empirical evidence on the extent of prostitution as indicated by the prevalence of clients of prostitutes. Capture-recapture and census methods have been used for many decades to estimate the prevalence of prostitute women (Brewer et al. 2000; Symanski 1981; Watts 1994; Woolston 1921). In the last 20 years, the prevalence of their male clients has also been studied systematically in scores of countries worldwide, but only with surveys. Unfortunately, survey estimates of client prevalence are plagued by underreporting. Randomized experiments have repeatedly shown that men substantially underreport contact with prostitutes in self-administered paper questionnaires, face-to-face interviews, and interviewer-administered telephone interviews (as used in most surveys to date) compared to computer-assisted self-interviewing, which is thought to promote accurate reporting (Des Jarlais et al. 1999; Lau, Thomas, and Liu 2000; Lau, Tsui, and Wang 2003; Rogers et al. 2005; Turner et al. 1998; van Griensven et al. 2006). Additional evidence for underreporting comes from the observation that men are more likely to acknowledge patronizing prostitutes on repeated questioning in surveys (Brewer et al. 2000).

Given the difficulty of assessing client prevalence with surveys, we sought to use alternate methods, such as capture-recapture techniques, for this estimation problem. Capture-recapture methods are often used to estimate the size of populations that are difficult or impossible to find and count (Sudman 1988). One basic capture-recapture approach involves observing one sample of individuals over a period of time and noting the number of times each individual in the sample is encountered or "captured." Features of the frequency distribution of captures can indicate the number of individuals in the population not observed, thus providing the means to estimate the overall population size. Patronizing arrest data constitute this sort of one-sample capture-recapture data. That is, they include information with which to construct a frequency distribution of client "captures" (number of times a client was arrested for patronizing, defined as buying or attempting to buy sex).

In this paper, we report capture-recapture estimates of client prevalence in several United States metropolitan areas, and assess variation in prevalence across time and space. We supplement the capture-recapture estimates with a prevalence estimate from Colorado Springs derived algebraically from the unique combination of local data on the prevalence of prostitutes, their number of clients, and clients' number of prostitute sex partners. We also compare these estimates to those from the General Social Surveys (GSS), a regular national probability sample household survey. Moreover, we estimate the fraction of the estimated number of clients who are arrested.

Methods

Inclusion Criteria

For meaningful application of capture-recapture methods to arrest data, several requirements must be met. To be analyzed, the arrest records must include data on: uniquely identified arrestees; prostitution arrests in law enforcement jurisdictions that comprise all or nearly all (>90%) of the prostitution arrests in the metropolitan area; arrests that have not been filtered by judicial processing (e.g., conviction or court appearance), as such procedures likely produce subsets of clients who differ in some ways from arrested clients overall; and arrestees' residential locations.

Search for Data Sets

Between 2000 and 2005, we sought prostitution arrest data from approximately thirty selected local jurisdictions across the United States. In addition, we requested statewide prostitution arrest records from the central computerized criminal history (CCH) repository for each of the 50 states (excluding those states with statutes that explicitly forbid disclosure of arrest data for this purpose). For Washington state, we requested arrest records from all cities with populations greater than 25,000 residents and counties with an incorporated city with more than 15,000 residents) because misdemeanor arrests are not reliably reported to the state CCH repository.