Acceptance of Prostitution and Its Social Determinants in Canada

The Theoretical Significance of Prostitution

Prostitution is a thorny issue not only in Canada but also in the world (Cao & Maguire, 2013). The public opinion of prostitution is intriguing to study for several reasons. First, it is a sensitive issue because it deals with an unconventional behavior or lifestyle that is deemed by some as abnormal, sinful, or wicked, stemming from society’s historical attempt to control sexual expression (K. Davis, 1966; G. Rubin, 1993). Most people seem to be morally ambiguous to its regulations. On one hand, prostitution represents a sexual alternative to marriage (Cao & Stack, 2010; Liang & Cao, 2013). Its existence represents a symbolic rejection of the prevailing norm of marriage between one man and one woman. On the other hand, despite thousands of years’ legal suppression, prostitution as a part of larger sex industry still exists and even has a tendency to thrive in some areas (Attwood, 2009; Brents & Sanders, 2010; Zheng, 2014). Public condemnation of prostitution coexists with clandestine visits to prostitutes.

Second, the topic blurs the line between the personal and the social. It is personal because it concerns two willing individual adults, and it is a social issue because all its related activities are illegal in Canada. Unlike in the United States, prostitution itself is technically legal in Canada, but all or most peripheral activities are penalized (Barnett, 2008; S. Davis, 1994; Morton, Klein, & Gorzalka, 2012), including clients of prostitutes in the new law (Armstrong, 2014). Keeping a common bawdy house, procuring, and living on the avails of prostitution are some examples of the type of activities that are criminalized. When there are laws regulating a private behavior between two consenting adults, the behavior is no longer personal, and it has, thus, become social. Giddens (1992) further elevated the issue of sexual partners to the politics of democracy as he regarded sexual intimacy as democracy because “it is oriented to control of distributive power” (p. 197).

Third, opposition to prostitution can be supported by both a conservative and a liberal ideology for different reasons. The left-wing people oppose it for its exploitation of women bodies whereas right-wing folks resist it from the religious tradition. The issue also splits feminism movement (Irvine, 1998; Jolin, 1994; Meier & Geis, 1997; G. Rubin, 1993; Weitzer, 2000). Since the 1970s, there have been animated debates between “pro-sex” feminists who believe that any sexual regulation limits women’s rights (G. Rubin, 1993) and “anti-sex” feminists concerned about the sexual subordination of women (Dworkin, 1997; MacKinnon, 1982). This divide is particularly evident in debates about prostitution, with some feminists affirming the rights of sex workers to practice their profession without being criminalized or stigmatized, and others operating from the perspective that most prostitutes are coerced and exploited.

Finally, it is a theoretically important issue for scholars because its acceptance seems to be related to the rise of cosmopolitan culture of more progressive societies (Giddens, 1992; Inglehart, Nevitte, & Basanez, 1996; Stack, Adamczyk, & Cao, 2010; Welzel & Inglehart, 2005; Welzel, Inglehart, & Klingemann, 2003). As a consequence of the ever-more radical break of the institutions of modernity with tradition, and growing intrusion of its system of control across preexisting external boundaries of social action, “plastic sexuality,” which is decentered sexuality, freed from the needs of reproduction (Giddens, 1992), has become popular since the 1960s. Inglehart et al. (1996) argued that changes in how men and women interact as sexual partners are part of the transformational processes of modern society, in which traditional norms and practices are no longer able to resist the rise of new cosmopolitan culture. In a new version of modernization theory, Welzel and Inglehart (2005) and Welzel et al. (2003) contended that the essence of human progress is the process of enlarging human autonomy and choices through freedom rights. Freedom rights give people legal space to exert autonomous choices in the private and public activities. These theories all suggest that there is a trend toward more tolerance of different lifestyles in more progressive societies, for which Canada is one of them (Cao & Selman, 2010).

Prostitution is part of the much larger “sex industry” (Weitzer, 2000). Prostitution refers to the direct-contact sexual services performed for remuneration. The sex industry includes erotic massage, erotic dancing, telephone sex operation, escorting, domination services, pornography as well as prostitution. In addition, when the public think of prostitution, they generally equate it with street prostitution, which is estimated to account for only about 20% of prostitution activity in Canada (Cool, 2004). The official Canadian stand on prostitution is to condemn the exploitation of prostitutes, not prostitution itself. The persistence of prostitution is associated with traditionally culturally supported attitudes that encourage men to feel entitled to sexual access to women and to feel superior to women (Cotton, Farley, & Baron, 2002). Its persistence is also related to “the marketplace exchange of money for the unemotional provision of sexual gratification with no strings attached” (Meier & Geis, 1997, p. 31).

Prior Research on the Public Opinion of Prostitution

Prostitution is generally perceived as a moral issue. It embodies moral beliefs about individual autonomy, democratic equality, and unambiguous and nondiscretionary fulfillment of obligations. The use of survey data to study public opinions is not new in Canada, but Canadian academics have not paid any attention to analyze data on prostitution. The public opinion is a complex mélange reflecting a variety of political and cultural forces and institutions (Parkin, 1971). The legal status of prostitution, however, is dependent to a substantial degree upon public opinion (Burstein, 2003; Cao & Maguire, 2013; Stack, Cao, & Adamczyk, 2007). Public attitudes toward prostitution and their determinants are severely under-researched in Canada. This study focuses on prostitution not only because it touches one of more fundamental yet sensitive and ambiguous topics but also because there is a lack of understanding on the social determinants of this attitude in Canada.

The prevalence of prostitution is roughly known (Barnett, 2008; Cool, 2004), but the accuracy of such information is always questionable. There are only a few published academic studies on public opinion concerning prostitution in Canada. Marwick and Partners’s (1984) report is quite comprehensive, and the public attitudes toward prostitution are a part of a wide range of legal issues regarding regulations of prostitution. The questionnaires are not well designed and there are many morally loaded wording. For example, the questionnaire contains a number of questions about the “harms of prostitution.” The techniques of analyses are limited to a series of correlation matrix and percentage tables, and the information is out of date. Van Brunschot (1994) reported the results of Gallup polls in Alberta. He focused on the issues of whether prostitution was a serious problem, whether tougher laws were needed, and whether prostitutes should be confined within red-light districts. The percentages were reported, but no statistical tests were done on whether these percent differences were significant or not. More recently, Lowman and Louie (2012) reported the latest public opinion on prostitution law reform in Canada. They presented many polls on a series of issues of regulating prostitution in Canada, but no analyses were done on the social determinants of the attitudes toward prostitution except gender. It is found that females are more likely to support the idea of making prostitution illegal while males are more likely to choose the option of decriminalizing some illegal acts to allow consenting adult prostitution.

More detailed analyses of the public attitudes toward prostitution are conducted in the United States and elsewhere. Cao and Stack (2010) found that authoritarianism, family values, feminism, and tolerance are related to prostitution acceptance in China. Peracca, Knodel, and Saengtienchai (1998) utilized 88 adults in Thailand in a focus group design, and their study showed that a majority of participants thought that it was acceptable for sex workers to marry. Rasanen and Wilska’s (2007) student sample in Finland reveals a strong gender difference in accepting prostitution: male students are more accepting than their female counterparts. In testing a version of postmaterialist theory with data from 32 nations, Stack et al. (2010) reported that survivalist culture is the strongest predictor of prostitution acceptance.

In the United States, Abrams and Della Fave (1976) found that authoritarianism and religiosity are related to attitudes toward legalization of victimless crime, prostitution included. Cotton et al. (2002) discovered that there were significant differences between males and females about prostitution myths, but not between johns and other males. Basow and Campanile (1990) found that feminism and authoritarianism are linked to prostitution acceptability. Cosby, May, Frese, and Dunaway (1996) reported that males, Whites, and Catholics are more likely than females, non-Whites, and Protestants to support the legalization of prostitution. Surveying only males who have used prostitutes, Sawyer, Metz, Hinds, and Brucker (2001) observed that “johns” are a heterogeneous group, and they tend to be more supportive of legalizing prostitution and more endorsing of personal support for the prostitutes.

Beyond simple correlations and nonrandom samples, Cao and Maguire (2013), May (1999), and McCaghy and Cernkovich (1991) used multiple regression analyses. Relying on the data from the telephone interviews in Toledo (Ohio), McCaghy and Cernkovich found that older respondents are more approving of legalization and the less educated are the most approving of the suppression strategy. Based on the national random sample of telephone interviews of adults residing in the 48 continuous states in 1995, May’s analyses show that educational level, age, males, Whites, non-Protestants, and the unmarried are significant predictors of prostitution acceptability. In addition, May reported that tolerance, measured by a single item of tolerance of gambling, is related to the attitude toward prostitution. Those who are more tolerant of gambling are also more accepting of prostitution. May’s study does not examine the effects of authoritarianism and religiosity on acceptance of prostitution. Relying on the national probability sampling over 20-year span, Cao and Maguire found that the public acceptance of prostitution in the United States has been steadily increasing from 1981, 64% of respondents said prostitution was never justified, to 48% in 2000. In addition, they tested a comprehensive model of prostitution acceptability and found that increased age and education, together with tolerance, are associated with more acceptance of prostitution while religiosity and authoritarianism tend to reduce the acceptance. Class and gender are not statistically significant in their model.

This literature review indicates that public’s attitudes toward prostitution in Canada is understudied. The rich information in public opinion polls has not been utilized fully. Many theoretically important variables, such as religiosity and authoritarianism, have not been linked to the attitude toward prostitution in Canada. Considering the importance of religion in the Canadian life, this neglect is problematic. In addition, the effect of gender found in Lowman and Louie’s (2012) study has not been subject to more rigorous multivariate test. It is quite possible that the relation is spurious and once the proper control variables are introduced, the direct relationship between gender and prostitution may not hold. Finally, Canadian specific situation, such as Quebec, has not been taken into consideration.

According to Lipset (1990), there has been a so-called “Quiet Revolution” of value transformation in Quebec, Canada, beginning in the 1960s. This revolution is reflected by the decline of the traditional influence of the Roman Catholic Church, expansion and development of the economic structure, modernization of the educational system, and the rise to power of more left-leaning political parties. As a result of this revolution, Quebec is now seen as a more liberal region of Canada (Cao & Selman, 2010). Indeed, regarding the prostitution policy, the Quebec government issued a report in 2004, which recommends a repeal of the communication laws against both prostitutes and their clients in addition to sweeping changes that tackle the important issues surrounding prostitution—poverty, schooling, health care, drug treatment, community policing, and so on (see Barnett, 2008). Therefore, we expect that residents in Quebec are more accepting of prostitution than their English counterparts elsewhere.

The Conceptual Model

In light of the limitations of the extant literature, the current investigation extends the study of acceptance of prostitution to Canada. We first evaluate the public attitude toward prostitution over 25 years and see whether there is a trend toward more tolerance of prostitution as described by Giddens (1992), Inglehart et al. (1996), and others. Then we investigate the social determinants of the attitude toward prostitution in Canada with data from 2006. Using SEM, we develop the most comprehensive model derived from the model in the United States (Cao & Maguire, 2013) and test it with Canadian data. Although there were similarities between America and Canada in terms of social liberalism—prostitution acceptance is part of this index—the effect of nations was found insignificant after controlling for the effects of races and regions (the South and Quebec) (Cao & Selman, 2010). The situated knowledge, therefore, is necessary to better understand Canada.

The proposed conceptual model highlights the importance of two intermediating variables, authoritarianism and religiosity, which are hypothesized to have negative direct effects on acceptance of prostitution (see Figure 1). These two variables represent traditional controls or moral regulation of sexual behavior (Abrams & Della Fave, 1976; Cao & Maguire, 2013). Authoritarianism is one aspect of conventionality with an aversion to anything that appears odd or different. It is therefore expected that highly conventional individuals will regard visiting prostitutes as deviant and will disapprove of it. For religiosity, Christian belief, historically, has tended to equate the term “immorality” with illicit sexual behavior, prostitution included. Almost half (48%) of Canadians surveyed in 2006 said that they are active members of a religious organization. Although the percentage of Canadians who identify themselves as religious has dropped, those who regard themselves as spiritual remains high (Carlson, 2012). Religion remains a salient factor in life as Canadians sing their national anthem, “O Canada”: “God keep our land glorious and free!” In addition, 11 exogenous variables—age, class, education, employment, gender, gender equality, marital statuses, marriage-out, Quebec, race, and tolerance—are hypothesized to have direct effects on acceptance of prostitution and indirect effects through authoritarianism and religiosity. ² According to the common practice of SEM, circles in Figure 1 represent a latent construct that is defined by multiple items while squares are defined by a single item.

OPEN IN VIEWER

Figure 1.

Conceptual model predicting acceptance of prostitution.

The Sample

The current study uses data from the World Values Surveys (Inglehart et al., 2000; World Values Survey Association, 2014). The World Values Surveys grew out of a study launched by the European Values Survey (EVS) group under the leadership of Jan Kerkhofs and Ruud de Moor in 1981. It was designed to facilitate cross-national comparisons of basic values and attitudes in a wide range of concerns. The surveys have evoked such widespread interest that it is replicated in more than 80 countries. We made use of four waves of Canadian data in 1981, 1990, 2000, and 2006. The first three waves of data were collected by Gallup-Canada, and the samples sizes are 1,254 in 1981, 1,730 in 1990, and 1,931 in 2000, respectively. For our SEM, only the Canadian sample in 2006 was used. It was collected between February 14 to April 8, 2006, and the sample size is 2,164 respondents. The principal investigator is Neil Nevitte, Department of Political Science, the University of Toronto. Using a nationally representative multistage probability sampling frame, the sample was designed to be representative of the entire adult population aged 18 years and older. All surveys were conducted via face-to-face interviews with respondents. For further discussion of the surveys, see European Values Study Group and World Values Survey Association (2005).

The Outcome Variable

Prostitution acceptability is measured with the item on a 10-point continuum ranging from never justified (1) to always justified (10). The respondents were asked, “Please tell me for each of the following statement whether you think it can always be justified, never be justified, or something in between.” Prostitution is among many issues listed. Other issues include claiming government benefit, take a bribe during the course of work, divorce, euthanasia, suicide, and so on. Note that the tone of the item is quite neutral. For the 2006 sample of Canada, the responses to the prostitution item are normally distributed with the skewness score of .869. A total of 91 respondents did not answer or said that they did not know. These cases and a few others were dropped in the final analyses, and the final sample size was 2,027.

Intermediate Variables

According to our conceptual model, there are two intermediate variables: authoritarianism and religiosity. For authoritarianism, the respondents are asked whether they think that greater respect for authority is good, neutral, or bad. We reverse the original order so that those who say bad are coded as 1 and those who do not mind are coded as 2 and those said good are coded as 3. The variable is a measure of respect for authority, but it could also be a proxy for authoritarianism (Cao & Maguire, 2013; Cao & Stack, 2010) because it taps one key element of authoritarianism: conventionality—an attachment to socially accepted modes of behavior coupled with an aversion to anything that appears different (Abrams & Della Fave, 1976).

The religiosity measure is based on four items, with an emphasis on issues such as salience, membership, beliefs, and attendance (Cao & Maguire, 2013; Zhao & Cao, 2010). Items include the following: “How important is God in your life?” “Independently of whether you go to church or not, would you say you are a religions person, not a religious person, or a convinced atheist?” “Apart from weddings, funerals, and christenings, about how often do you attend religious services these days?” and “Are you an active member, an inactive member or not a member of Church or religious organization?”

Exogenous Variables

The proposed model includes seven exogenous variables: age, class, education, employment, gender, marriage, and race. These variables have been used in the previous literature on prostitution. Age is measured as the respondent’s actual age in years at the time of the survey. Class is assessed with the subjective feeling about one’s own relative standing among one’s nation, where 1 = lower class and 5 = upper class (after reversing the original order). Educational attainment is assessed with nine ordinal categories: 1 = no formal education, 2 = incomplete primary school, 3 = complete primary school, 4 = incomplete technical/vocational school, 5 = complete technical/vocational school, 6 = incomplete secondary school (less than 12 years), 7 = complete secondary school (12 years), 8 = some university-level education without degree, and 9 = university-level education with degrees. Employment status is measured as a dummy variable where 1 = employed and students and 0 = all others (part-time, self-employed, unemployed, housewives, retired, and others). Gender is a binary variable where male = 1 and female = 0. Race is a dummy variable with Whites as 1 and all Others as 0. None of racial or ethnic subgroups in Canada has more than 2.2% of the sample. Marriage is a dummy variable where 1 = married and widowed and 0 = singles/never marriage and others (divorced, separated, and living together as married). It is expected that age and marriage are negatively associated with acceptance of prostitution because these characteristics represent the social establishment within a society and, as control theory suggests, that people with intact or strong social bonds are less likely to be deviant (Stack et al., 2010). Education is expected to be positively related to accepting prostitution because its effect is often regarded as a proxy for social liberalism, a global symbolic orientation that signifies one’s openness to new concepts and ideas (Stack et al., 2007). Being males and Whites are expected to be positively related to acceptance of prostitution because they are the beneficial of the services provided by prostitutes. Although it is possible for both males and females to be prostitutes, the overwhelming majority of prostitutes are female, and the problem of prostitution is largely regarded as the exemplars of gender subordination and exploitation (Jolin, 1994).

Four additional variables are also included: gender equality, marriage-out, Quebec, and tolerance. For gender equality, respondents were asked about their attitudes toward several statements that concern gender equality: when jobs are scarce, men should have more right to a job than women, men make better political leaders, university is more important for a boy than for a girl, and men make better business executives than women do. For the first statement, the answer was coded as from 1 to 3, with 1 indicating agree and 3 indicating disagree. The rest of the answers were coded as 1 = strongly agree and 4 = strongly disagree. High scores on the final combined index represent more support for gender equality. For marriage-out, the item asks whether family is an outdated institution, and those who agree are coded as 1 and those who do not agree are coded as 0. Quebec is a binary variable with those who live in Quebec coded as 1 and those who do not were coded as 0. The latent concept of tolerance was created based on four items. The respondents are asked with the lead question, “Could you please sort out any that you would not like to have as neighbors?” Then the following items are displayed: people who are immigrant or foreign workers, people of a different religion, unmarried people who are living together, and people who speak a different language. The answer yes is coded as 1 and no is coded as 0. Higher scores on the final combined index represent more tolerance.

Estimation Methods

SEM is used for the analyses because it has the advantage of using latent variables derived from observed measures, and it can model the structural relations pictorially to enable a clear conceptualization of the theory under study (Byrne, 2010). SEM allows researchers to rule out spurious effects while elaborating and testing direct and indirect effects of exogenous variable (through mediating variables) on the outcome variable, and it can address one major drawback of traditional regression analyses: the assumption that independent variables are perfectly reliable (Geiser, 2013). By specifying a regression analysis at the latent level, SEM allows researchers to correct for the measurement error for the latent factors.

In this study, all analyses were conducted by using Mplus Version 7, and parameters were estimated using weighed least square mean variance adjusted (WLSMV) algorithm. The sample was weighted based on their demographic characteristics. The absolute fit index includes chi-square (χ²) statistics, where χ² is the likelihood ratio statistic used to test whether there is a statistically significant discrepancy between the correlation matrix and the model-implied matrix. Although a nonsignificant result is preferred, a significant chi-square value is acceptable when there is a large sample size. Another absolute fit indicator is the root mean square error of approximation (RMSEA) which takes the error of population approximation and degrees of freedom into account and characterizes the lack of fit of the hypothesized model to the population covariance matrix. The cutoff point of RMSEA is below 0.06 (Hu & Bentler, 1999; also see Steiger, 2007) to constitute a good fit.

Among incremental fit indices are a group of measures that attempt to answer the question of how well is the model doing, compared with the worst model that there is. The comparative fit index (CFI) assesses “the fit of a user-specified solution in relation to a more restricted, nested baseline model,” in which the “covariances among all input indicators are fixed to zero” positing no relationship among variables (Brown, 2006, p. 84). It ranges from 0 to 1.00, with values greater than 0.95 indicating a reasonably good fit between the hypothesized model and the empirical data (Hu & Bentler, 1999). Simiarly, a cutting point of .95 of Tucker–Lewis Index (TLI) indicates a good fit.

Missing data are not a serious problem for this research. In general, the missing rates of most variables do not exceed 10%. However, to take the most advantage of the data, we used the full information approach in the current research. This method is one of the most popular ways to address missing values because it does not impute any values nor discard any data (Acock, 2012; Enders, 2010).

Partially Structural Regression Analysis With WLSMV Estimator

In examining the correlation matrix (see the appendix), correlations (r) among exogenous variables range from .013 to .350 and multicollinearity is not a problem. As the outcome variable is ordinal in nature, probit regression is applied. Using SEM, this study ran the comprehensive models with the weights added in calculating the percentages for inferential purpose. This variable was created to make the sample characteristics, such as age and sex, consistent with the Canadian national population characteristics. Because our model contained both manifest and latent variables, we used partially latent structural regression analysis to predict an ordinal outcome. With CFI = .957, RMSEA = .036, and TLI = .948, our data fit the model well. The effects of class and employment were not statistically significant, and we dropped them in our final model. The rest of exogenous variables on acceptance of prostitution were mediated by moral regulation variables (authoritarianism and religiosity) except education (see Figure 3). People who have more respect for authorities and are more religious have lower level of acceptance of prostitution. Respondents with higher level of education are more likely to accept prostitution ceteris paribus. The direct effect of religiosity is the strongest (−.314), followed by authoritarianism (−.138) and education (.099).

OPEN IN VIEWER

Figure 3.

Final model of prostitution acceptance with standardized probit coefficient.

The effects of Quebec and tolerance on acceptance of prostitution are totally mediated by religiosity. All effects on authoritarianism and on religiosity are in the expected directions. Overall, about 17% of variance of prostitution acceptance is explained by the model. Table 2 presents all standardized probit coefficients shown in Figure 3.

OPEN IN VIEWER

Table 2.

Model Direct and Indirect Effects on Prostitution Acceptance With Standardized Values (n = 2,027).

	Direct effect	Indirect effect via authoritarianism	Indirect effect via religiosity
Authoritarianism	−.138***	—	—
Religiosity	.314***	—	—
Age	—	−.013**	−.072***
Education	.099***	.030***	.025**
Gender (male = 1)	—	.009*	.042***
Gender equality	—	.014**	.040**
Married (yes = 1)	—	−.012*	−.029**
Marriage as an outdated institution	—	−.012**	−.074***
Quebec	—	—	.031**
Race (White = 1)	—	.013**	.061***
Tolerance	—	—	.035*

*

p < .05. **p < 01. ***p < .001.