NOT YOUR TYPICAL "PRETTY WOMAN": FACTORS ASSOCIATED WITH

Wendi E. Goodlin

A Dissertation

Submitted to the Graduate College of Bowling Green State University in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

December 2008

Committee:

Stephen A. Cernkovich, Advisor

Christopher S. Dunn Graduate Faculty Representative

Jorge M. Chavez

Alfred DeMaris

Peggy C. Giordano ii

ABSTRACT

Stephen Cernkovich, Advisor

Criminologists have long been interested in , particularly prostitution.

However, the research in this area has been very uneven and has produced conflicting results.

The goal of the current research is to move forward in filling the gaps in our knowledge of the lives of women involved in prostitution and to better understand the factors associated with their initial involvement. Three sociological theories guide the analysis: general strain theory, control theory, and social learning theory. The major hypotheses predict that experiencing sexual abuse and exposure to delinquent peers increase the odds of prostitution whereas high levels of parental attachment and supervision decrease the odds of prostitution. These relationships are examined further using mediating and moderating variables, including running away from home, psychological distress, and drug use.

The data upon which this research is based come from The Ohio Lifecourse Study

(OLS), a multi-wave dataset of household and institutionalized respondents. Variables central to the analysis come from items that ask how often the respondent has been paid to have sex and a variety of other items that measure contentious family environment, sexual abuse experienced as a child, parental supervision and attachment levels, the influence of peers, their self-esteem and depression levels, and the use and abuse of drugs. Importantly, although the OLS is a highly delinquent sample, it is not a prostitute-biased sample. In addition, the OLS contains a variety of respondents, including those not involved in prostitution, prostitutes who were abused as adolescents, those who were also abused as adolescents but did are not prostitutes, and those of different races. Thus, although the sample is highly delinquent, there is much variation among iii

respondents on key variables including abuse, supervision, running away, and drug use/abuse to name a few. Furthermore, because the OLS contains both quantitative data and qualitative life history narratives, the latter serve as an important supplement to the former and provide rich and nuanced detail not obtainable from the quantitative analyses.

Binary logistic regression analyses show support for the hypothesis that higher levels of sexual abuse increase the odds of prostitution (strain theory), but this relationship is not mediated by running away as argued in previous research. On the other hand, there is little support for the hypothesis that higher levels of parental attachment decrease the odds of prostitution (social control theory); however, this could be a result of the sample being highly delinquent or the lack of variation among respondents on the parental attachment variables. In contrast, there is evidence that higher levels of supervision decrease the odds of prostitution (social control theory). In addition, there is support for the hypothesis that those with delinquent friends have higher odds of prostitution than those without delinquent friends (social learning theory).

Moreover, there is support for racial and job status differences. For example, blacks have consistently and significantly higher odds of prostitution than whites and those who are unemployed or employed part-time have consistently and significantly higher odds of prostitution than those with full time jobs. Finally, with the exception of parental caring and trust, most of the moderating variables are not found to be significant. iv

ACKNOWLEDGEMENTS

I would like to express my gratefulness and extreme appreciation to those who helped,

encouraged, and supported me throughout this long and stressful process. I am forever grateful to Stephen Cernkovich, my advisor, who guided me throughout this entire process, gave me honest advice and suggestions, and was always there to listen. I am also extremely appreciative of the suggestions and contributions from each of my committee members, Jorge M. Chavez,

Alfred DeMaris, Christopher S. Dunn, and Peggy C. Giordano. Last, but not least, I am forever grateful to my family, who gave me love, support, understanding, and encouragement not only throughout this whole process but throughout graduate school as well. Although it was an extremely tough year, without their strength and support, even during difficult times, I would never be where I am today. This research is dedicated to my family. v

TABLE OF CONTENTS

Page

CHAPTER 1. STATEMENT OF THE PROBLEM...... 1

CHAPTER 2. LITERATURE REVIEW ...... 5

Introduction…...... 6

Overview of Prostitution...... 6

Definitions of Prostitution…...... 8

Prostitution, Sexual Abuse, and Running Away...... 10

Prostitution, Family Environment, Attachment, and Supervision …...... 13

Encouragement from Family, Friends, and Other Prostitutes...... 15

Prostitution and Drugs …...... 17

Prostitution and Psychological Distress (Depression and Self-Esteem)...... 19

Why Research is Incomplete: Difficulty of Obtaining Willing Respondents...... 21

Why Research is Incomplete: Lack of Longitudinal Data …...... 22

Why Research is Incomplete: Different Types of Populations Used ...... 23

Conclusion …...... 25

CHAPTER 3. THEORETICAL ORIENTATION...... 27

General Strain Theory Overview...... 27

General Strain Theory and Gender Differences...... 28

General Strain Theory and Prostitution ...... 29

Control Theory: Overview...... 31

Control Theory and Attachment ...... 32

Control Theory and Supervision...... 33 vi

Control Theory, Gender, and Age...... 33

Control Theory and Prostitution ...... 34

Social Learning Theory: Overview...... 35

Social Learning Theory and Prostitution ...... 36

The Relationship of General Strain Theory, Control Theory,

and Social Learning Theory...... 37

Summary……...... 38

Hypotheses ……...... 40

Testing the Relationship Between the Strain Theory Measures

and Prostitution……...... 40

Testing the Relationship Between the Control Theory Measures

and Prostitution……...... 41

Testing the Relationship Between the Social Learning Theory Measures

and Prostitution……...... 41

CHAPTER 4. METHODS...... 43

Data…………...... 43

Measures……...... 45

Dependent Variable...... 45

Independent Variables...... 46

Sexual Abuse Experienced as an Adolescent ...... 46

Contentious Family Environment ...... 47

Conflict with Parents in 1982 ...... 47

Conflict with Parents in 1995 ...... 48 vii

Attachment to Parents: Parental Control and Supervision ...... 48

Attachment to Parents: Caring and Trust in 1982 ...... 49

Attachment to Parents: Caring and Trust in 1995...... 49

Attachment to Parents: Intimate Communication in 1982...... 50

Attachment to Parents: Intimate Communication in 1995 ...... 50

Friends Involved in Prostitution ...... 51

Peer Pressure in 1982 ...... 51

Peer Pressure in 1995 ...... 52

Drug Abuse in 1995 ...... 52

Drug Abuse in 2003...... 53

Used Drugs To Get High in 1982 ...... 53

Used Drugs To Get High in 1995 ...... 54

Used Drugs To Get High in 2003 ...... 54

Running Away From Home...... 54

Psychological Distress: Self-Esteem in 1995 ...... 55

Psychological Distress: Self-Esteem in 2003 ...... 55

Psychological Distress: Depression in 1995 ...... 55

Psychological Distress: Depression in 2003...... 56

Control Variables...... 56

Race ...... 56

Employment Status in 1995...... 57

Employment Status in 2003 ...... 58

Parental Social Class...... 58 viii

Age in 1982 ...... 59

Quantitative Analytic Strategy…………...... 60

Qualitative Analytic Strategy…………...... 62

CHAPTER 5. RESULTS...... 63

Descriptive Statistics and Bivariate Relationships…………...... 63

Comparison of Prostitutes and Non-prostitutes ...... 63

Comparison of Prostitutes and Non-prostitutes

Within and Between Races ...... 65

Logit Estimates of Becoming a Prostitute…………...... 67

Testing the Relationship Between the Strain Theory Measures

and Prostitution ...... 67

Testing the Relationship Between the Control Theory Measures

and Prostitution ...... 72

Testing the Relationship Between the Social Learning Theory

Measures and Prostitution ...... 77

CHAPTER 6. DISCUSSION AND CONCLUSION ...... 82

Summary…………...... 82

Limitations and Future Research…………...... 99

Conclusion…………...... 103

REFERENCES ...... 105

APPENDIX A. FIGURES ...... 116

APPENDIX B. DESCRIPTIVES AND BIVARIATE STATISTICS

FOR CORRELATES, CONTROL, AND ix

MEDIATING/MODERATING VARIABLES...... 119

APPENDIX C. REGRESSION MODELS...... 123

APPENDIX D. DESCRIPTIVES AND BIVARIATE STATISTICS

FOR THE CONSEQUENCES OF PROSTITUTION...... 127 x

LIST OF FIGURES

Figure Page

1a Relationship between experiencing strain and becoming a prostitute...... 116

1b Relationship between experiencing strain and becoming a prostitute...... 116

2a Relationship between social control variables and becoming a prostitute ...... 117

2b Relationship between social control variables and becoming a prostitute ...... 117

3 Relationship between having friends who are prostitutes and

becoming a prostitute...... 118 xi

LIST OF TABLES

Table Page

1 Means and Standard Deviations Comparing Female Prostitutes

and Female Non-Prostitutes...... 119

2 Means and Standard Deviations Comparing White Female Prostitutes

and White Female Non-Prostitutes ...... 120

3 Means and Standard Deviations Comparing Black Female Prostitutes

and Black Female Non-Prostitutes ...... 121

4 Means and Standard Deviations Comparing White Female Prostitutes

and Black Female Prostitutes...... 122

5 Logit Estimates for the Regression of Prostitution

on the Strain Theory Measures ...... 123

6 Logit Estimates for the Regression of Prostitution

on the Social Control Theory Measures...... 125

7 Logit Estimates for the Regression of Prostitution

on the Social Learning Theory Measures ...... 126

8 Means and Standard Deviations for the Consequences of Prostitution

Comparing Female Prostitutes and Female Non-Prostitutes ...... 127

9 Means and Standard Deviations for the Consequences of Prostitution

Comparing White Female Prostitutes and White Female Non-Prostitutes...... 128

10 Means and Standard Deviations for the Consequences of Prostitution

Comparing Black Female Prostitutes and Black Female Non-Prostitutes ...... 129

11 Means and Standard Deviations for the Consequences of Prostitution xii

Comparing White Female Prostitutes and Black Female Prostitutes ...... 130

1

CHAPTER 1: STATEMENT OF THE PROBLEM

Criminologists have long been interested in sex work, particularly prostitution. However, the research in this area has been very uneven and has produced conflicting results. For example, although some cite abuse experienced as a child, running away from home, or substance use as the major factors associated with prostitution (Dalla, 2001; Dilorio, Hartwell, &

Hansen, 2002; Kramer & Berg, 2003; Medrano, Hatch, Zule, & Desmond, 2003; Nadon,

Koverola, & Schludermann, 1998; Noll, Trickett, & Putnam, 2003; Ruggiero, McLeer, & Dixon,

2000; Seng, 1989; Silbert & Pines, 1983; Simons & Whitbeck, 1991; Widom & Kuhns, 1996), others disagree that such factors are determinant (Cates & Markley, 1992; Erickson, Butters,

McGillicuddy, & Hallgren, 2000; Lung, Lin, Lu, & Shu, 2004; Nuttbrock, Rosenblum, Magura,

Villano, & Wallace, 2004; Pleak & Meyer-Bahlburg, 1990; Potterat, Rothenberg, Muth, Darrow,

& Phillips-Plummer, 1998), so that there is little consensus among researchers about the factors associated with prostitution. Beyond this, and although considerable research has been conducted on prostitution, there are important gaps in the literature. Most studies are qualitative in nature, making it difficult to know if the factors associated with prostitution in these types of studies are statistically significant. In addition, those studies that are not qualitative are cross- sectional and consequently fail to follow prostitutes over time; this limits the amount of knowledge obtained about their experiences over the lifecourse as prostitutes, as well as the short and long-term consequences of being involved in prostitution (e.g. being in a domestic violence relationship, living in poverty, and being divorced or never married). Also, cross-sectional studies are more likely to have a lower number of prostitutes included in them because they are only based on one point in time, and thus fail to identify those who become prostitutes later in the life course. Moreover, many studies use prostitute-biased samples and are usually not

2

racially diverse, making it difficult to know if there are any differences between races with

respect to prostitution. Furthermore, few researchers have examined the role of a lack of

parental supervision in female involvement in prostitution, which is surprising since it has been

established that a lack of parental supervision is an important mechanism for adolescent

involvement in other deviant behaviors. Consequently, it is unknown what role parental supervision plays regarding prostitution.

The goal of the current research is to move forward in filling the gaps in our knowledge of the lives of women involved in prostitution. In order to achieve this goal, three sociological theories guide the analysis: general strain theory, control theory, and social learning theory.

General strain theory is employed in order to examine the relationship between sexual abuse and prostitution. It is hypothesized that higher levels of sexual abuse increase the odds of prostitution. Control theory, on the other hand, is utilized to determine the relationship between supervision, parental attachment as an adolescent, and prostitution. It is hypothesized that higher levels of attachment as well as higher levels of supervision decrease the odds of prostitution.

Lastly, social learning theory is used in order to determine if peers are an important factor in women becoming prostitutes. It is hypothesized that peers are an important influence in persuading women to become involved in prostitution, just as they are in other delinquent behaviors.

The data upon which this research is based come from The Ohio Lifecourse Study (OLS),

a multi-wave dataset of household and institutionalized respondents. Variables central to the

analysis come from items that ask how often the respondent has been paid to have sex and a

variety of other items that measure contentious family environment, sexual abuse experienced as

a child, parental supervision and attachment levels, the influence of peers, psychological distress

3 levels (self-esteem and depression), and the use and abuse of drugs. The OLS is also used because it is not a prostitute-biased sample and it contains a variety of respondents including those not involved in prostitution, those prostitutes who were abused as adolescents, those who were also abused as adolescents but did not become prostitutes, and those of different races.

Furthermore, because the OLS contains both quantitative data and qualitative life history narratives, the latter serve as a nice supplement to the former and provide rich and nuanced detail not obtainable from the quantitative analyses.

It is important to note that although does occur, only female prostitution is examined in this research because there are only 16 (22 percent) male prostitutes in the OLS, hardly enough to conduct meaningful analysis. Also, although the OLS contains household and institutionalized respondents, only the institutionalized respondents are used in this research because a Chow Test indicates that the two types of respondents are affected differently by the independent variables and thus need to be analyzed separately. In addition, only six neighborhood female respondents report being paid to have sex in any of the three waves, not nearly enough to conduct meaningful analysis; consequently, only the institutionalized sample is used. Therefore, the final analytic sample is comprised of 73 previously institutionalized female respondents interviewed in all three waves of the OLS. Even though the respondents that are used in this study are from a high-risk delinquent sample rather than a general population sample, there is considerable variation among them on key dimensions. For example, 36 (49 percent) respondents in this study are prostitutes and 37 (51 percent) respondents are non- prostitutes. In addition, 11 (15 percent) respondents report sexual abuse and no prostitution, 19

(26 percent) respondents report sexual abuse and prostitution, 17 (23 percent) respondents report no sexual abuse but do report prostitution, and 26 (36 percent) respondents report a lack of both

4 sexual abuse and prostitution. It is important to note that this sample is highly skewed toward the lower and working classes; consequently, these respondents are more likely to be street prostitutes than the high-class call girls.

5

CHAPTER 2: LITERATURE REVIEW

Introduction

The term “prostitution” typically calls forth a variety of negative stereotypes, with most people blaming the prostitute, “slut,” or “whore” for the “social problem,” not the customer who paid for the sexual services. It is not surprising that most citizens and politicians have considered prostitution a social problem for many years in the United States. In order to ameliorate the problem, many cities have tried to impose stricter penalties against prostitutes while others have tried to legalize prostitution. Although prostitution is illegal in the United

States (except for certain counties in Nevada), it is not illegal in other countries around the

world.

Although there are male prostitutes, females dramatically out number them, including in

the OLS (36 female prostitutes compared to 16 male prostitutes). For this reason and because

most people view the “prostitution problem” as one involving primarily women, this study

focuses only on female prostitution. According to Potterat, Woodhouse, Muth, and Muth (1990),

approximately 84,000 women worked as prostitutes in the United States during the 1980’s. In

2005, according to the Office of Juvenile Justice and Delinquency Prevention [OJJDP] (2007),

84,900 people were arrested for prostitution, a decrease from 2004 where 87,900 people were

arrested for prostitution. Although these statistics show a decrease in arrest for prostitution,

these statistics do not necessarily indicate that the number of individuals involved in prostitution

is decreasing.

Despite the large number of women and children involved in prostitution, as well as the

vast interest in prostitution among researchers, there are many discrepant findings in the

literature, including definitional issues, whether women are forced into prostitution or voluntarily

6

enter, and the various factors associated with prostitution (e.g., sexual abuse, drug use, running

away, low self-esteem, high levels of depression, contentious family environment, a lack of

parental supervision, low levels of parental attachment, and peer influence).

Overview of Prostitution

Although most people think of prostitutes being over the age of eighteen, many female

adolescents serve as prostitutes as well. Women enter prostitution at a variety of ages, ranging

from 13 to 30 years old (Potterat, Phillips, Rothenberg, & Darrow, 1985). In addition, although

prostitution has often been called the world’s oldest profession (Bullough & Bullough, 1996),

attitudes towards it have varied over time and across nations and cultures (Clinard & Meier,

2008). For example, in the Middle Ages, societies did not regard prostitution as a criminal

activity but rather viewed it as a “necessary evil” (Clinard & Meier, 2008, p. 324), whereas it is

seen as an immoral and criminal act today. It is seen so much as an immoral criminal act today

that politicians who have been with high-class call girls have been forced to resign from their

political offices.

Attitudes toward prostitution depend upon women’s roles in a society and whether

prostitutes provide services in addition to sexual ones, such as companionship or counseling.

Moreover, even though a majority of people and groups do not approve of prostitution, it appears

in all societies. Critics who oppose prostitution argue that it involves a great deal of promiscuity with strangers, it threatens public health by facilitating the spread of sexually transmitted diseases (e.g. AIDS), and that it reduces the general quality of law enforcement because police officers are busier arresting prostitutes rather than murderers and drug dealers (Clinard & Meier,

2008). In addition, Flexner (1914) suggests that prostitution is harmful to society because it causes economic waste, sexually transmitted diseases to be spread, personal demoralization, and

7

other forms of criminality. Although this definition is 94 years old, some still view prostitution

in this way. As a result of these views, the laws generally focus on the prostitutes to the virtual

exclusion of their customers.

Two main views explain why women become prostitutes. C.O.Y.O.T.E. (Call Off Your

Old Tired Ethics - a group composed of former and current prostitutes), as well as many scholars

argue that women enter prostitution willingly; in other words, prostitution is an occupational

choice that suits some women. From this perspective, prostitution offers a service to the

community, especially to those customers who visit prostitutes because they lack successful

social and sexual skills (Campbell, 1998; Xantidis and McCabe, 2000), because they want to

fulfill their desire for sexual relations or intimacy (Bullough, 1970; Campbell, 1998; Jordan,

1997), or because they desire sexual activities that their partner cannot or will not provide

(Campbell, 1998; Holzman & Pines, 1982; McKeganey & Barnard, 1996; Monto, 2001, Sullivan

& Simon, 1998). On the other hand, others argue that women are forced into prostitution

because of their need to survive life on the street (Bagley & Young, 1987), because they have

few legitimate employment opportunities (Abramovich, 2005; Bullough & Bullough, 1996), or

to support an existing drug habit (Graham & Wish, 1994; Miller, 1995). Thus, the decision to

become a prostitute is not often freely made; rather, events and personal experiences combined

with a belief that there are very few other opportunities available, especially in a male-dominated society force women into prostitution (Cole, 1987).

Most of the young females involved in prostitution in the United States enter the profession as part-time prostitutes, who are only hoping to earn a few extra dollars in order to survive life on the street, buy food, drugs, or shelter (James, 1980; Weisberg, 1985). This soon turns into a full-time position though because it pays better than the limited available

8

alternatives. Therefore, some women may become prostitutes because they perceive a lack of

other opportunities or there is in reality a lack of alternatives (Dalla, 2002; Dalla, 2001). For

instance, 44 percent of participants in Dalla’s (2002) research reported becoming prostitutes out

of economic necessity.

Definitions of Prostitution

There are a variety of definitions of prostitution. Typically, it is “defined in terms of

buying and selling, or more generally in terms of an asymmetrical relationship of exchange in

which the sex worker provides sexual services and the customer supplies recompense for those services, usually in the form of money, but sometimes also in the form of food, lodging, clothing,

or ‘luxuries’” (Overall, 1992, p. 705). Most definitions of prostitution assume that females provide sexual favors in exchange for drugs, money, clothes, or privileges (Schauer & Wheaton,

2006).

In addition, Flowers (1998) notes that the definitions of prostitution can be categorized as social or legal definitions of prostitution. “Social definitions of prostitution regard prostitution as sexual relations that include some form of monetary payment or barter and are characterized by promiscuity and/or emotional apathy” (Flowers, 1998, p. 6). Moreover, social definitions tend to influence the legal definitions of prostitution and consequently affect how prostitutes are handled by law enforcement agencies. Most legal definitions, on the other hand, focus on female prostitutes rather than the customer who is paying for the services; consequently, the law and criminal justice agencies focus their attention on arresting and charging the prostitutes more so than the customers. Therefore, the legal definitions of prostitution are biased. For example, prostitution is considered to be a misdemeanor, or an offense that results in a fine and/or up to a thirty-day jail term. Moreover, 9 states have no statutes punishing pimps and 25 states have no

9

statutes punishing the customers of prostitutes, whereas 47 states have statutes punishing

prostitutes and stating that prostitutes are committing a crime (Schauer & Wheaton, 2006).

Millett (1971) argues that prostitution is the only crime where two people are doing a thing mutually agreed upon and only one of the participants, the female partner, is subject to arrest.

For instance,

Females are more likely to be arrested, prosecuted, and incarcerated for prostitution-related activities than males. In 1995, there were 49,491 females arrested in the United States for prostitution and commercialized vice, compared with 31,573 males. One in three women in jail was arrested for prostitution; seven in ten women who were imprisoned for felonies were initially arrested for prostitution (Flowers, 1998, p. 8).

Official statistics also show that only two customers are arrested for every eight prostitutes

(Flowers, 1998, p. 8).

One explanation for the lack of attention paid to male clients of prostitutes, according to

Davis (1993), is because women are held responsible for male deviance, which is a double standard. Moreover, Prasad (1999) feels the lack of attention to male clients is due to the belief that the behavior of seeking prostitutes is natural for men, which is not “worthy” of an explanation. Such definitions create negative stereotypes of prostitutes and a consequent bias towards incarcerating them more than the customers who pay for their sexual services. Beyond this, it is important to remember that definitions of prostitution may differ cross-culturally because of diverse societal standards (Bullough & Bullough, 1996); consequently, the gap in arrests between customers and prostitutes may not be as vast in other cultures or countries as it is in the United States.

Despite the variety of definitions that are used to describe prostitution, in this research, any individual who reported ever having provided sexual favors for anything of value, such as

10

money, food, drugs, rent, clothing, and shelter, even once or twice a year is considered a

prostitute.

There are a variety of factors that are associated with prostitution; however, this study examines only those that are included in the OLS, such as sexual abuse, running away from home, lack of parental attachment and supervision, the influence of friends and family, drug abuse and use, and psychological distress, all of which are discussed below.

Prostitution, Sexual Abuse, and Running Away

Previous research estimates that 10 to 84 percent of prostitutes experience sexual abuse

(Bagley & Young, 1987; Dalla, 2001; Russell, 1988; Silbert & Pines, 1983) thus indicating that those who are sexual abused are placed at a higher risk for prostitution, especially for females

(Widom & Kuhns, 1996). Therefore, sexual abuse is frequently cited as an important correlate of prostitution (Dalla, 2001; Dilorio, Hartwell, & Hansen, 2002; Kramer & Berg, 2003;

Medrano, Hatch, Zule, & Desmond, 2003; Nadon, Koverola, & Schludermann, 1998; Noll,

Trickett, & Putnam, 2003; Ruggiero, McLeer, & Dixon, 2000; Silbert & Pines, 1983; Simons &

Whitbeck, 1991; Widom & Kuhns, 1996). For example, sexual abuse is associated with

maladaptive sexual behavior (Briere & Runtz, 1990). In addition, the effect of sexual abuse has

more of a negative effect on the sense of self than physical abuse, thus, increasing the risk of self-destructive behaviors (Reiker & Carmen, 1986), emotional problems, and dependency on substances including and drugs (Kunitz, Levy, McCloskey, & Gabriel, 1998). Moreover,

Noll et al. (2003) argue that the impact of child sexual abuse enhances possibilities for the development of sexual distortions more than other types of child abuse because of the sexual nature of their abuse. Thus, despite the fact that some prostitutes may have experienced physical abuse as a child more often than sexual abuse (Widom & Kuhns, 1996), the lasting and severe

11

effects of sexual abuse on the child, such as higher levels of depression, low self-esteem, more

permissive sexual attitudes (Browning & Laumann, 1997; Miller, Monson, & Norton, 1995), and sexual risk-taking behaviors (Brown, Lourie, Zlotnick, & Cohn, 2000) are much stronger and have a more substantial impact on children than the effects of physical abuse; therefore,

potentially leading to prostitution and other maladaptive or criminal behaviors, including running

away from home (Chesney-Lind, 1997).

Still others argue that experiencing sexual abuse at an early age teaches females to emotionally distance themselves from situations of abuse, a technique that some prostitutes use while engaging in sexual favors with their customers. For instance, Herman and Hirschman

(1977) speculate that those children who are sexually abused emotionally distance themselves as

a way to cope with the traumatic experience. Thus, as adults, they continue to emotionally

distance themselves from others and as a result, engage in sexual activities to counteract this. In

addition, other researchers suggest that some sexual abuse victims engage in habitual sex as a

way to control or avoid dealing with the emotional feelings associated with being sexually

abused (Westerlund, 1992). Therefore, experiencing sexual abuse at an early age results in a

separation between sexual activity and emotions (Herman & Hirschman, 1977; James &

Meyerding, 1977; Miller, 1986; Westerlund, 1992). Although many researchers argue that

experiencing sexual abuse during childhood poses a higher risk of prostitution (Simons &

Whitbeck, 1991), particularly if the abuse is by a family member (Kramer & Berg, 2003), some researchers have found no differences in the abuse experienced by prostitutes and non-prostitutes

(Cates & Markley, 1992; Nadon et al., 1998), suggesting that sex abuse is not a necessary correlate of prostitution. In this research, it is hypothesized that sexual abuse is positively associated with prostitution.

12

Despite the strongly held association between sexual abuse and prostitution, it is still unclear if abuse has a direct or indirect effect on prostitution or if this relationship is mediated

(e.g. running away) or moderated (e.g. attachment or psychological distress) by another factor.

For instance, some argue that if sexual abuse is experienced as a child and parental bonds are not intact, the adolescent is more likely to turn to prostitution (Brannigan & Van Brunschot, 1997), so the effect of sexual abuse is conditioned on the level of parental attachment. In addition,

Simons & Whitbeck (1991) found that the combination of early sexual abuse and destructive parenting generally lead to running-away behavior and increased participation in deviant activities, including prostitution.

Others argue, on the other hand, that those who are sexually abused are more likely to escape the abuse by running away, leading to prostitution in order to survive on the streets

(Nadon et al., 1998; Seng, 1989; Simons & Whitbeck, 1991); in this latter case it is more the

running away that causes the participation in prostitution rather than the sexual abuse per se. In

other words, the sexual abuse has an indirect effect on prostitution. Moreover, some argue that

running away, regardless of abuse history, has a direct effect on becoming a prostitute (Seng,

1989). For example, adolescents may run away from home for reasons such as rejection by

parents and disputes with family members; moreover, the child may feel rejected by the parent if

he/she tells them about the abuse and they do not believe them so they run away. Many

runaways and throwaways (those adolescents kicked out of their parents’ homes) have little

money or direction and consequently find themselves homeless without food, shelter, or

protection and must do something in order to survive; consequently, many of them turn to

prostitution. Therefore, selling their bodies almost becomes a necessity for getting food, shelter,

drugs, and the basic necessities of life. Few young prostitutes-often school dropouts with no

13 significant work experience-can realistically expect to find legitimate well-paying employment

(Flowers, 1998, p. 88). A prostitute in a study by Surratt, Inciardi, Kurtz, and Kiley (2004) supports this by stating, “When you need the cracks [crack ] and you need money for other things ‘cause your rent money went on the boards [was used to buy crack], you got to survive, and you know, to do that, the pussy works” (p. 55). Thus, running away from home as an adolescent is also an important factor associated with prostitution (Nadon et al., 1998;

Ruggiero et al., 2000; Seng, 1989; Silbert & Pines, 1983; Simons & Whitbeck, 1991) because many runaways turn to prostitution for survival on the street.

Prostitution, Family Environment, Attachment, and Supervision

It has been argued that social bonds are important in order to function in a healthy manner (Hirschi, 1969). For example, social networks can influence one’s personal development and significantly affect personal adjustments (Dalla, 2001). Consequently, family attachments play a crucial role in one’s development throughout the lifecourse (Bronfenbrenner, 1989), although in adolescence and adulthood, peers and intimate partners are more influential than one’s family of origin (Brown, 1990; Steinberg & Silverberg, 1986). Therefore, those individuals who are less attached to their parents, a characteristic typical in a prostitute’s background (Earls, 1990; Nadon et al., 1998), are more likely to participate in delinquent activities, including prostitution.

Moreover, without strong social bonds, especially in early childhood, if a traumatic event is encountered, such as sexual abuse, coping with it can be even more difficult. As a result of this relationship, some researchers have examined the family environment that many prostitutes encountered as children and found that these households often contained a variety of forms of abuse, domestic violence, , drugs use, parental absence (Dalla, 2001), and a lack of

14 strong family ties (Earls, 1990; Nadon, et al., 1998). For example, Dalla (2001) found that the majority of the prostitutes interviewed felt abandoned by their parents as children, especially by their mothers, and felt they were unprotected when they told them about the abuse. Therefore, the emotional deprivation of juveniles leaves them with few coping mechanisms, considerable hopelessness, and the fear of abandonment (Newman & Caplan, 1981). Moreover, inadequate parental relationships, especially with their mothers, can cause a split in the world of self and object. As a result, the girls become depressed, feel helpless, and feel as if they are losing a love object (MacVicar & Dillon, 1980). Thus, positive evaluations from others and positive self- evaluations suffer because of the lack of familial support and contentious family environments

(Thoits, 1983), making it difficult to handle the stress of experiencing a traumatic event like sexual abuse. This, coupled with low self-esteem and a diminished sense of security can lead to prostitution because these girls may believe that they can gain the affection they long for from someone else, such as a pimp or a customer. “The man initially provides protection and affection for the girl, who may have lacked parental affection” (Ennew, 1986, p. 89). In this research, it is hypothesized that higher levels of parental attachment decrease the odds of prostitution.

A lack of supervision can also be a factor associated with prostitution because it can lead to substance use and other delinquent activities (Cernkovich & Giordano, 1987; Gove &

Crutchfield, 1982; Laub & Sampson, 1988; Loeber & Stouthamer-Loeber, 1986). For example,

Laub and Sampson (1988) found that family process variables, such as supervision, attachment, and discipline are directly related to serious and persistent delinquency, including prostitution.

In addition, Cernkovich and Giordano (1987) argue that in all family structures, supervision, as well as other family interaction variables, is related to delinquency. Thus, one of the most

15

powerful predictors of juvenile delinquency is a lack of parental supervision (Loeber &

Stouthamer-Loeber, 1986). Despite this strong association between a lack of supervision and

participation in delinquent activities, previous research has not examined the relationship

between prostitution and a lack of parental supervision (to the researcher’s knowledge).

Consequently, it is unknown whether or not this relationship holds for prostitution as well. In

this research, it is hypothesized that higher levels of supervision decrease the odds of prostitution

as it does with various delinquent activities.

Encouragement from Family, Friends, and Other Prostitutes

Friends and relatives can also influence a female to become a prostitute, especially if a

friend or relative is already involved in prostitution. This tactic is used to gain the woman’s trust and confidence and because it has been established that individuals learn to participate in

delinquent activities by watching and imitating their peers. Thus, there is a strong connection between participating in delinquent behaviors and peer involvement (Garnier & Stein, 2002;

Haynie, 2002). Consequently, those who have peers that encourage delinquent behavior are

more likely to become delinquent (Agnew & Brezina, 1997). Therefore, a woman’s first

experience with prostitution is likely influenced by her interpersonal network influences (Dalla,

2001)

Despite the established association between peer influences in the participation of

delinquent activities (Agnew & Brezina, 1997; Garnier & Stein, 2002; Haynie, 2002; Laub &

Sampson, 2003), there are only a small number of studies that examine the role of peers and

family members in influencing women to become prostitutes. Those few studies that have

examined familial or peer influences in turning to prostitution have found that friends, relatives,

and other prostitutes can be very influential in deciding whether or not an adolescent female

16

becomes a prostitute (James, 1980; Weisberg, 1985), especially if friends and family members

are part of the sex trade. For example, the Enablers study (1978) found that at the time the

respondents became involved in prostitution, 16 percent had friends who were already involved

in the business. Furthermore, James (1980) found that learning about prostitution from a relative

occurred among 23 percent of her sample and a family member was with the respondent at the

time of his/her first act of prostitution for four percent of the sample.

As a result of family members also being involved in prostitution, environments

involving the sex trade normalize the behavior and consequently, influence adolescents to feel

that sleeping with strangers for money or other things of value is “normal” (Tutty & Nixon,

2003). In addition, parental pressure can also be placed on the adolescent to become a prostitute

if the parent is involved in prostitution him/herself.

[Pimp telling the following story] I worked on Helen first, the one I was legally married to. Told her how much money we could make with her kid. Finally she agreed and all I had to do was keep her (Helen) drunk. We started out with nude pictures when she was eleven and I turned her out [started her in prostitution] a year later. The other mother-daughter act I met up north. I told her she could travel with me but she had to turn her kid out. She agreed and told her kid, ‘Do it or I’ll knock your damn brains out!’ (Campagna & Poffenberger, 1988, p. 101).

In a similar fashion of imitating peers, other prostitutes work as recruiters for pimps to influence some girls into prostitution. Pimps use this strategy because they feel it is “safer” and it allows current prostitutes to search for young girls who have the potential to be a successful, profitable prostitute. In addition, this also allows the promise of friendship from the current prostitutes to the new recruits. Moreover, the “rookie” prostitutes are able to learn the “tricks of the trade” from the “veteran” prostitutes in a manner similar to an apprenticeship. During this apprenticeship, the rookie prostitutes learn how to handle problem situations with customers, how to negotiate a price and collect fees, how to inspect the customer’s penis without him

17

knowing, what types of customers to avoid, and how to converse with customers (Clinard &

Meier, 2008). It is hypothesized in this research that delinquent peers have a similar relationship

to prostitution as they do with other criminal behaviors.

Prostitution and Drugs

It also has been suggested that individuals become involved in prostitution because of drugs (Cates & Markley, 1992; Erickson, Butters, McGillicuddy, & Hallgren, 2000; Lung, Lin,

Lu, & Shu, 2004; Nadon et al., 1998; Nuttbrock, Rosenblum, Magura, Villano, & Wallace, 2004;

Pleak & Meyer-Bahlburg, 1990; Potterat, Rothenberg, Muth, Darrow, & Phillips-Plummer,

1998; Simons & Whitbeck, 1991), particularly crack, whether it is to support an already existing habit or one that developed after entry into prostitution as a way to handle the stress of working the streets (Cates & Markley, 1992; Dalla, 2002; Erickson et al., 2000; Lung et al., 2004;

Nuttbrock et al., 2004; Potterat et al. 1998; Young, Boyd, & Hubbell, 2000). This suggests that prostitutes have higher levels of drug use than non-prostitutes. However, the rates of drug use among prostitutes are a far from settled issue. For instance, Goldstein (1979) suggests that approximately 40 to 85 percent of prostitutes are drug users whereas Dalla (2002) found that 95 percent of her respondents reported abusing drugs.

An additional unsettled issue is when drug use begins: before or after prostitution. For example, some researchers have found that prostitution generally follows rather than preceding it. James (1977), for instance, argues that prostitution followed addiction in 48 percent of her sample (compared to 38 percent that reported prostitution preceded addiction to drugs and 14 percent that reported prostitution and drug use happened simultaneously). In addition, Potterat et al. (1998) found that 66 percent of their sample reported using non-IV drugs before becoming a prostitute (compared to 17 percent that reported using drugs after becoming a

18 prostitute and 18 percent that reported entry into prostitution and drug use occurred simultaneously). The argument that prostitution follows drug addition is also supported by Dalla

(2002) who found that “37 percent [of her respondents] were drawn to prostitution to support an established drug habit and 19 percent reported that prostitution entry and drug abuse occurred simultaneously” (69).

However, these results may depend on whether she is a street-prostitute or a high-class . For example, Goldstein (1979) found that addiction to drugs before entering prostitution is more common among lower-class prostitutes, such as street-level prostitutes whereas the risk of an addiction to drugs after entering prostitution is high, particularly for high- class call girls (Goldstein, 1979). Thus, although one may not enter prostitution with a drug habit, the risk of addiction after entry increases the longer one is a prostitute. For example, 89 percent of Dalla’s (2002) sample who did not become prostitutes to support an already existing drug habit reported becoming regular drug users the longer they were involved in prostitution.

“’I think it’s possible to prostitute without drugs if you’ve never done them before. Once you get started on the drugs though, I think it’s impossible to go back to just prostituting. It would take a real strong person to prostitute without drugs, if they’ve done them before’” (Dalla, 2002, p. 69).

Thus, some women turn to prostitution to support an existing drug habit while others begin to use drugs after entering prostitution, usually as a means to psychologically and emotionally detach from the strains of being a prostitute (Erickson et al., 2000; Gossop, Powis, Griffiths, &

Strang, 1995). Therefore, they work in the area of prostitution to get money and/or drugs to support their own drug habit when few other sources of income may be available to them.

Eva is a sixteen year old patient at New York City’s Phoenix House drug rehabilitation center who got hooked on crack two years ago. The product of a troubled middle class family, she was already a heavy drinker and pot smoker

19

when she was introduced to coke by her older brother, a young dope pusher. ‘When you take the first toke on a crack pipe, you get on top of the world’ she says. She started stealing from family and friends to support her habit. She soon turned to prostitution and went through two abortions before she was sixteen. ‘I just wanted to get high. Fear of pregnancy didn’t even cross my mind when I hit the sack with someone for drugs’ (Flowers, 1998, p. 85).

Despite the differences in the temporal order of drug use/addiction and prostitution

(which is not the main focus of this research), most of the literature examining the substance use of prostitutes argues that the most commonly used drug among this population is

(Erickson et al., 2000; Maxwell & Maxwell, 2000; Nuttbrock et al., 2004; Weiner, 1996). For

instance, Maxwell and Maxwell (2000) found that no drug other than crack cocaine was

associated with the prevalence, frequency, or the age of onset into prostitution. Weiner (1996)

also argues that those women who use cocaine are more likely to become prostitutes than those

who do not. Thus, although some women became prostitutes to support a drug habit while others

began using drugs after they started prostitution, most prostitutes spend most of their money on

crack (Erickson et al., 2000). For example, a 25 year old prostitute states, “‘I’ve been doing it

(prostitution) since I was 16 years old, that’s like what I know best. I started prostitution to

support my habit for alcohol and marijuana…when you have so many people touching your

body, I did the drugs first. I got addicted (to crack) when I was 18 and I’ve never stopped since

that’” (Erickson et al., 2000, p.775). “To most, it is the solution-they continue to use [crack]

because it’s what makes their lives bearable or interesting. Selling sex is the means to this end”

(Erickson et al., 2000, 775).

Prostitution and Psychological Distress (Depression and Self-Esteem)

An additional factor associated with prostitution is psychological distress, particularly

high levels of depression and low levels of self-esteem. However, conflicting results have been

20 found (El-Bassel et al., 1997; Potterat et al., 1998; Potterat et al., 1985; Risser, Timpson,

McCurdy, Ross, & Williams, 2006). For example, Potterat et al. (1998) & Potterat et al. (1985) found that an equal number of prostitutes and women in the comparison group indicated that they had a low self-esteem; thus suggesting that low self-esteem did not explain why some women become prostitutes.

El-Bassel et al. (1997), on the other hand, found that prostitutes have significantly higher mean scores of psychological distress (measured by the General Severity Index and most of the

Brief Symptom Inventory subscales) than those who are non-prostitutes among a sample of active drug users; therefore, prostitutes are more likely to suffer psychological deficits, such as depression and low self-esteem, than those who are not prostitutes. In addition, Risser et al.

(2006) found that those who are current prostitutes reported lower levels of self-esteem and higher levels of depression compared to those who have never prostituted and those who used to be prostitutes, among an African American crack cocaine smoking population. Therefore, the significant relationship between prostitution and psychological deficits may be because of drug use, particularly crack cocaine, and may appear only if a sample of active drug users is utilized.

Thus, psychological deficits may be a moderator to the drug use and prostitution relationship rather than psychological distress directly influencing entry into prostitution.

On the other hand, others argue that low self-esteem or feelings of worthlessness coupled with experiencing a traumatic event, such as rape, sexual abuse, or physical abuse, makes some women more vulnerable to prostitution than others (Bess & Janus, 1976; Brown, 1979; Bullough,

1965; Jackman, O’Toole, & Geis, 1963; James, 1976; Maerov, 1965). For example, Finkelhor and Browne (1985) argue that the victim of abuse may stigmatize herself as a result of the abuse experienced. Consequently, victims of abuse will develop negative self-images. Thus, the

21

victim of abuse may have sex with others because of the negative self-images she has and because she feels as if she is a “damaged good” and unworthy of love (Davis & Petretic-Jackson,

2000; Finkelhor & Browne, 1985; James & Meyerding, 1978). Thus, the potent combination of

experiencing sexual abuse and having psychological distress increases the odds of prostitution.

Perhaps if women only had one component of this combination, the odds of prostitution would not be as high. Therefore, psychological distress may be a moderator to the sexual abuse and

prostitution relationship.

Why Research is Incomplete: Difficulty of Obtaining Willing Respondents

Previous research has produced conflicting results and there are several gaps and

unresolved issues in the literature as to why individuals become prostitutes. There are several

reasons for this. One is that obtaining information, particularly accurate information, about prostitutes is very difficult. For instance, some researchers have obtained information about this

population but find it extremely difficult to find prostitutes willing to participate in their studies.

In addition, the situations that would provide the best information might not always be the most

appropriate, such as on the street where a prostitute may be working. For example, if a

researcher attempted to interview a prostitute while she was trying to hook a trick, she may be

suspicious that the researcher is an undercover cop and consequently refuse to participate in the

study. Moreover, she may lose business and money by taking the time to participate in the study.

In addition, if the prostitute has a pimp, the pimp may become angry that “his girl” is not making

him money or may even become angry at the researcher for bothering his girl while she is

working, which could result in a violent altercation with the pimp for the prostitute or the

researcher. As a result of the difficulties in obtaining a willing population of respondents to

participate in research studies, researchers must use other means to study prostitutes. For

22

example, some researchers use patients in drug or sexually transmitted disease therapy in order

to obtain respondents for their studies on prostitutes. These means of obtaining respondents who

were or still are prostitutes can bias the findings due to the selective nature of the sample and can increase the likelihood of little variation among the respondents on key findings. In addition, the lack of willing participants limits the number of respondents that can be used in studies, thus, making the population size very small. Consequently, some of the variables that are associated with prostitution may not be significant predictors due to the small sample size.

Why Research is Incomplete: Lack of Longitudinal Data

Another reason why the research on the factors associated with prostitution is conflicting is that most of the research has been conducted with cross-sectional samples. As a result, it is difficult if not impossible to know if prostitution and the factors associated with it decrease over time. For instance, as stated above, many researchers argue that sexual abuse is an important factor in turning to prostitution (Dalla, 2001; Dilorio et al., 2002; Medrano et al., 2003; Nadon et al., 1998; Noll et al., 2003; Ruggiero et al., 2000; Silbert & Pines, 1983; Simons & Whitbeck,

1991; Widom & Kuhns, 1996); however, it is unknown if the effect of sexual abuse on becoming a prostitute decreases over time, if sexual abuse has an indirect effect on prostitution, such as through running away from home, or if it has a direct effect on prostitution. In addition, it is unknown what the consequences of being a prostitute are. These uncertainties are partially due to the lack of longitudinal research. “While there is correlational evidence that childhood victimization may be associated with increased risk of prostitution, promiscuity, and teenage pregnancy, current knowledge of the extent and nature of these relationships is ambiguous, at best, in the absence of longitudinal research” (Widom & Kuhns, 1996, p. 1607-8). Also, longitudinal studies supply more information about respondents and their behavior over the

23

lifecourse. Therefore, longitudinal studies can capture if respondents change their behavior over

time (e.g. if they were not a prostitute in Time 1 but they were in Time 2 and Time 3).

Consequently, longitudinal data can increase the total number of prostitutes in a sample. Cross-

sectional data only allows researchers to look at prostitution at one point in time, thereby missing

those who begin prostitution later in the life course. Therefore, longitudinal data allow

researchers to gather information about these respondents. Although this research uses a

longitudinal dataset, it does not take full advantage of it, mainly because the small sample size

restricts the types of analyses that can be undertaken.

Why Research is Incomplete: Different Types of Populations are used

The different types of populations used in previous research are another cause of the

mixed findings. For example, some studies only examine males or females whereas others only

examine whites or blacks. Most of the studies that are available focus on female samples

(Bagley & Young, 1987; Bullough & Bullough, 1996; Dalla, 2006; Maxwell & Maxwell, 2000;

Potterat et al., 1990; Potterat et al., 1985; Risser et al., 2006; Silbert & Pines, 1983; Simons &

Whitbeck, 1991; Young et al., 2000) and very few examine male samples (Cates & Markley,

1992; Caukins & Coombs, 1976; Pleak & Meyer-Bahlburg, 1990). In addition, even fewer utilize a mixed gender sample (Finkelhor & Ormrod, 2004; Medrano et al, 2003; Widom &

Kuhns, 1996). Therefore, very little is known about male prostitutes; thus, it is also unknown if the factors associated with prostitution are similar for males and females. However, the few studies that have used mixed gender samples are conflicting in their results about differences between males and females. For example, Widom and Kuhns (1996) found no differences between males and females in either the bivariate or multivariate levels. The examination of male prostitution is an important issue that needs to be examined further. This research is unable

24

to do so, however, because of the small number of male prostitutes in the OLS data. This is

explained in further detail below.

Furthermore, some studies examine racial differences (Finkelhor & Ormrod, 2004;

Kramer & Berg, 2003; Silbert & Pines, 1983; Potterat et al., 1998; Medrano et al., 2003; Simons

& Whitbeck, 1991; Weiner, 1996) while other studies examine either whites or blacks (Lung et

al., 2004; Nadon et al., 1998; Risser et al., 2006) limiting the ability to determine racial

differences in prostitution. The racial differences that have been examined in previous literature

have found some significant differences between whites and blacks that enter prostitution, but

these results are conflicting. For example, Kramer & Berg (2003) found that of those who were

prostitutes, white women were more likely than minority women to have experienced some type

of abuse during childhood, but that minorities entered prostitution at an age about two-and-a-half

years younger than whites. Furthermore, Finkelhor and Ormrod (2004) found that juvenile and

adult offenders of prostitution were predominately white.

In addition, Medrano et al. (2003) found that black women who experienced severe emotional abuse were more likely to engage in prostitution than whites and Hispanics but for white and Hispanic women, no form of abuse was significantly related to prostitution.

Moreover, Potterat et al. (1998) found that whites were more likely to be non-street rather than street prostitutes. In addition, these authors also found that in general, white prostitutes were more likely to report the use of intravenous drugs and to have experienced nonconsensual sex before puberty than African American or Hispanic prostitutes. Weiner (1996) also found that white women were more likely to use IV , whereas Hispanic women had higher levels of nasal heroin usage, and black women were more likely to drink alcohol or smoke crack. Potterat et al. (1998) also discovered that the only ethnicity variable to be a significantly consistent factor

25

associated with drugs before “sexual debut” was being white. Potterat et al. (1998) also found significant racial differences in the timing of the events that lead to becoming a prostitute but not in the order of the events. As can be seen, it is unknown whether or not the factors associated with prostitution are similar or different for various racial groups. The data used in this research is a highly disadvantaged sample of blacks and whites. Consequently, any results in this research indicating racial differences may contradict the above differences reported in previous literature.

Conclusion

Although there is sizable literature on the factors associated with prostitution, there is a lack of consensus on what they are. In addition, correlations between prostitution and life experiences, such as abuse and running away, have been established in the literature but it has also been found that other women experience these same incidents and do not become prostitutes

(Bullough & Bullough, 1996). Therefore, in previous research, “no single factor stands out as causal in a women becoming a prostitute” (Bullough & Bullough, 1996, p. 171). These limitations and contradictions notwithstanding, it is reasonable to assume that if different types of populations were utilized to study prostitution, such as using racially varied samples and variations among respondents on key variables, a better insight into the factors associated with prostitution would be achieved.

Moreover, if more longitudinal studies were available, not only would the factors associated with prostitution be better understood but so also would the consequences of being a prostitute over the life course be better known. In addition, the sample size for prostitutes may increase. Although it is difficult to obtain information about prostitutes from cross-sectional

26 studies let alone longitudinal studies, it is expected that the gains from longitudinal studies would outweigh the demands and financial costs of obtaining willing respondents.

27

CHAPTER 3: THEORETICAL ORIENTATION

As a framework for examining the factors associated with prostitution, three sociological

theories are utilized, including general strain theory, control theory, and social learning theory.

This is consistent with the assumption that there are a variety of factors that explain why females become prostitutes, such as experiencing sexual abuse (general strain theory), lack of parental attachment and supervision (control theory), and having delinquent friends, especially those who are prostitutes (social learning theory). It is important to note that although social learning theory may be able to better predict prostitution directly, general strain theory and control theory

need to utilize components from other theories (e.g. general strain theory needs to use measures

from control and social learning theories and control theory needs to use variables from general

strain theory and social learning theory).

General Strain Theory: Overview

Robert Agnew (1992), the major proponent of general strain theory, claims that

participation in delinquent and criminal activities is an adaptation to stress, including failure to

achieve positively valued goals (e.g. cannot obtain immediate or future goals due to blocked

opportunities or inadequate abilities or skills), removal of positively valued stimuli (e.g. the loss

of a significant other or loved one), and confrontation with negative stimuli (e.g. abuse,

contentious family environment, or parental conflict). Each type of strain experienced increases

the likelihood that individuals will undergo one or more of a range of negative emotions and

psychological distress, such as disappointment, depression, fear, and especially anger (Agnew,

1992). Therefore, general strain theory argues that experiencing strain can cause negative

emotions, which interfere with the conventional coping strategies of individuals, thereby leading

to participation in criminal activities. This theory is very careful not to say that strain causes

28

delinquency, but that each type of strain may create an emotional predisposition for delinquency

or function as a situational event that instigates a particular delinquent act. The key factors

affecting the disposition to delinquency are associations with delinquent peers and continued

experience with adversity (Agnew, 1992; Akers & Sellers, 2004). Therefore, whether or not an

individual responds to strain by participating in delinquent behaviors depends on internal and

external constraints (peer associations and self-efficacy). Thus, the combination of experiencing

a traumatic event (particularly sexual abuse) coupled with psychological distress can increase the

odds of participation in criminal activities because no other coping mechanisms are believed to

be available. Therefore, this combination has a synergistic effect.

Although the failure to achieve positively valued goals and the removal of positively

valued stimuli are important components of general strain theory, they are not the focus of this

research. Rather, this research concentrates on the exposure to negative stimuli, particularly

sexual abuse as an adolescent but other negative stimuli, although not the main focus in this

research, are also examined (contentious family environment and parental conflict).

General Strain Theory and Gender Differences

General strain theory can also explain differences in deviant behavior between males and females. For example, Broidy & Agnew (1997) found that male strains are more conducive to serious violent and property crimes, whereas female strains are more conducive to family violence, running away, and more self-directed forms of crime, such as drug use. As noted above, previous research has found running away from home as an adolescent (Nadon et al.,

1998; Ruggiero, McLeer, & Dixon, 2000; Seng, 1989; Silbert & Pines, 1983; Simons &

Whitbeck, 1991) and drug use (Cates & Markley, 1992; Erickson et al., 2000; Lung et al., 2004;

Nadon et al., 1998; Nuttbrock et al., 2004; Pleak & Meyer-Bahlburg, 1990; Potterat et al., 1998;

29

Simons & Whitbeck, 1991) to be important factors of prostitution. Moreover, females have been

found to respond to strain with depression and insecurity (Broidy, 2001). Psychological distress,

particularly depression and low self-esteem, have been found to be associated with prostitution

when they are coupled with experiencing a traumatic event, such as sexual abuse (Bess & Janus,

1976; Brown, 1979; Bullough, 1965; Jackman, O’Toole, & Geis, 1963; James, 1976; Maerov,

1965).

General Strain Theory and Prostitution

As a result of much previous research indicating that prostitutes are more likely to be

abused as children (Cates & Markley, 1992; Dalla, 2001; Dilorio et al., 2002; Erickson et al.,

2000; Lung et al., 2004; Medrano et al., 2003; Nadon et al., 1998; Noll et al., 2003; Nuttbrock et

al., 2004; Potterat et al., 1998; Ruggerio et al., 2000; Seng, 1989; Silbert & Pines, 1982; Simons

& Whitbeck, 1992; Widom & Kuhns, 1996), general strain theory is employed in this research in

order to examine the relationship between physical and/or sexual abuse (a source of strain) and

prostitution (a response to strain). It is important to note that prostitution is only one of several

responses to strain. Other deviant responses include drug use, theft, and assault; however, the

only responses to strain that are examined in this research are prostitution, drug use, and

psychological distress. Moreover, general strain theory alone cannot explain why some turn to

prostitution, rather other theories and variables from those theories must be utilized. Therefore, variables identified by other theories can moderate or mediate the effects of strain on delinquency (Akers & Sellers, 2004) and in this case, the delinquent behavior in response to strain is prostitution. For example, it has been demonstrated that participation in delinquent behaviors is more likely to result from strain if one has delinquent peers and weak social bonds

(Mazerolle, Burton, Cullen, Evans, & Payne, 2000). Thus, the components of social bonding

30 theory (lack of attachment to parents measures as moderating variables) and social learning theory (deviant peers as mediating variables) are needed to explain the relationship between experiencing strain (abuse) and prostitution (response to strain).

According to Agnew, Brezina, Wright, and Cullen (2002), delinquency is higher among those who experience family, school, and neighborhood strain. This may be a result of adolescents having fewer coping skills and legitimate resources to deal with strain than adults

(Agnew, 1997). Strain is most likely to lead to crime when individuals lack the skills and resources to cope with their strain, are low in conventional social support, and are low in social control (Agnew, 2001). In addition, if the strain is seen as unjust, crime is likely to result

(Agnew, 2001). Consequently, when an adolescent experiences sexual abuse, especially by an adult family member, the resulting strain is likely to be blamed on others and seen as unjust, consequently leading to participation in delinquent behavior. If this same individual is abused and has low self-esteem or high levels of depression then this combination is particularly potent.

For example, if one is sexually abused and is psychologically distressed, she may emotionally detach herself from others and thus engage in habitual sex to cope with the sexual abuse and psychological distress. She also may feel unworthy of love because of the abuse and psychological distress; as a result, she turns to prostitution. Consequently, the combination of experiencing sexual abuse and having psychological distress is particularly potent, and it is reasonable to believe that the effects of each separately, while substantial, is not as great as their joint effects. Thus, it is predicted that those respondents who experience sexual abuse coupled with having high levels of psychological distress have higher odds of turning to prostitution.

Therefore, the level of psychological distress conditions the level of coping with the sexual abuse.

31

However, the bond with family members may also be an important component in dealing

with the strain of abuse and may consequently impact participation in prostitution. In other words, a strong bond with a parent or family member may condition the negative consequences of abuse. On the other hand, if there is a weak bond between the abused child/adolescent and parent, the child/adolescent, especially if lacking legitimate coping skills, may run away from home in order to deal with the abuse and the negative emotions experienced because of the abuse. Moreover, as a result of running away and living on the streets, many adolescents are forced to participate in a variety of crimes in order to survive, including theft and prostitution. As a result, it is predicted that those who experience sexual abuse and have high levels of parental attachment are less likely to participate in prostitution than those who experienced sexual abuse and have low levels of parental attachment. It is also predicted that running away from home because of the sexual abuse experienced leads one to become friends with deviant peers, which can influence individuals to participate in criminal behaviors, particularly prostitution, in order to survive life on the street. In addition, it is predicted that one becomes involved with drugs because of running away in order to deal with the strains of living life on the street, which can in- turn lead to prostitution to support one’s drug habit. Please see Figures 1a and 1b (p. 116) documenting the relationship between the strain theory measures and prostitution.

Control Theory: Overview

Travis Hirschi (1969), the major proponent of control theory, claims that the social bond consists of four elements: attachment, commitment, involvement, and belief. Involvement refers to one’s participation in conventional activities, such as spending time with family, studying, and extracurricular activities at school (Hirschi, 1969). For example, if one is involved in conventional activities, time to participate in deviant behaviors is limited. Belief refers to the

32

acceptance of a conventional value system. If the belief in conventional values is weakened, the

chance of participating in delinquent activities is increased. Commitment is the rational element of the social bond and refers to the stake in conformity (e.g. one’s current personal, social, and future capital) one has that could potentially be lost by participation in law-violating activities

(Hirschi, 1969). In other words, commitment is the cost/benefit analysis of participation in nonconforming behavior. Consequently, if the benefits of participating in delinquent acts outweigh the costs, the adolescent will more than likely commit the delinquent behavior.

Although involvement, belief, and commitment are important elements of control theory, they are not the focus of this research. Rather, this research concentrates on attachment.

Control Theory and Attachment

The main assumption of control theory is that social bonds and attachments are a strong buffer against delinquency (Hirschi, 1969). This theory looks at the connection between the

individual and society by focusing on social institutions such as the family, school, and peers. In

addition, social bonding theory argues that when an individual’s bond to society is weak or

broken, delinquent acts are more likely to occur (Hirschi, 1969). In this research, family

bonding/attachment refers to one’s family relations/connectedness (interaction, attention,

supervision, communication, affectional identification with parents, and understanding). For

example, if a juvenile is not attached to a family member, the opinions and expectations of that

individual are of little importance to the juvenile. Consequently, the juvenile will be more likely

to participate in activities that are non-conforming than is a juvenile who is affectionally attached

to a family member or conventional peers. This is because the former cares little about the

family member’s opinion about his or her participation in delinquent acts and cares little about

the embarrassment caused to the family member due to the participation in delinquent acts. On

33

the other hand, if a juvenile is attached to a parental figure, that juvenile will be sensitive to the parent’s expectations for conformity. In addition, if the juvenile is strongly attached to a parental figure, the child is more likely to have an internalized set of conventional values and to feel as if

their parents are “psychologically present” when they are alone, thus governing the child’s

actions. Also, if one is strongly attached to parents, one will not be attached to delinquent peers;

rather one will be attached to conventional peers. Therefore, the more attached the juvenile is to

a family member, the less likely the juvenile is to engage in delinquent activities.

Control Theory and Supervision

According to control theory, individuals who are more attached to family members are

less likely to be delinquent (Hirschi, 1969). If an individual is more attached to family members,

he or she is more likely to spend time with his or her family. Consequently, the more time spent

with family members, the more supervision the individual receives. In turn, the more

supervision an individual has, the more attached that individual will be to a parental figure, thus,

decreasing the likelihood of participating in delinquent activities and associating with deviant

peers. Strong attachment also results in the psychological presence of the parent, a form of

indirect supervision. It should be noted, however, that too much supervision from the

perspective of the adolescent would likely reduce attachment to parental figures.

Control Theory, Gender, and Age

According to control theory (Hirschi, 1969), females are more likely to have stronger

attachments to family members because females are usually more sensitive to others’ opinions

and what others think of them, and because they are subjected to greater levels of supervision

than are males. Moreover, Cernkovich and Giordano (1987) found that the major predictors of

delinquency for females were arguments and low levels of communication with parents, parental

34

disapproval of peers, and tenuous parental support in regards to adolescent identity problems. In

addition, family bonds have a stronger impact on younger adolescents because the opinions of

family members are still important (Cernkovich and Giordano, 1987). From mid- to late

adolescence, the opinions and attachments to family members decrease as peer opinions and

values become more salient and influential. Consequently, bonds with friends are important

(Hirschi, 1969) in influencing participation in delinquent activities, including prostitution.

Control Theory and Prostitution

According to control theory, if an individual is strongly attached to family members or conventional peers, he or she will be less likely to participate in delinquent activities, such as prostitution. Moreover, control theory claims that if an individual is more attached to family members he or she will spend more time with those family members; thus, the more time spent with attached family members, the more supervision these individuals experience. The more

supervision experienced the less likely an individual will be to participate in delinquent

activities, such as prostitution. In this research, control theory is utilized to determine the

relationship between parental attachment as an adolescent and subsequent prostitution. The measures used in this research to measure attachment are intimate parental communication, parental caring and trust, and parental control and supervision. It is anticipated that those

respondents with higher levels of attachment and higher levels of supervision have lower odds of

prostitution throughout the life course because they will have less opportunities to become

attached to delinquent peers, which, in turn, decrease the likelihood of deviant behavior.

It is important to note, however, that lack of attachment and supervision alone cannot

account for the specific form of deviant behavior, prostitution, or otherwise. For example,

bonding variables have been found to be related to both minor and serious delinquency;

35

however, the relationship between the bonding variables and delinquency have been found to be mediated by variables associated with strain and social learning theories (Agnew, 1993).

Therefore, control theory by itself cannot explain why some turn to prostitution; rather, control

theory would predict a greater likelihood of deviance in general, not prostitution specifically.

Thus, components from other theories, such as strain theory and social learning theory, must be utilized by using mediating (social learning theory: deviant peers) or moderating variables (strain theory: sexual abuse) in order to fully understand and explain the relationship between control theory and prostitution. For example, if one has low levels of attachment and is sexually abused, one may turn to prostitution in order to feel loved or receive affection or attention from a stranger to fill the void left from the combination of the lack of parental attachment and sexual abuse. Also, the more one is abused, the level of parental attachment will decrease. On the other hand, weak parental attachment may create an environment conducive to sexual abuse.

Therefore, it is hypothesized that the level of sexual abuse conditions the level of parental attachment. Please see Figures 2a & 2b (p. 117) documenting the relationship between social control measures (e.g. parental supervision and attachment) and prostitution.

Social Learning Theory: Overview

Ronald Akers (1985), the major proponent of social learning theory, claims that criminal

behavior is learned through four processes: differential association, definitions, differential

reinforcement, and imitation. For example, the primary groups (particularly family and friends)

that individuals spend time with expose them to definitions of behavior, positively or negatively

reinforce a variety of behaviors, and are a venue within which certain behaviors are learned from

others or imitated (Akers & Sellers, 2004). Peer association variables have been found to be

36

strongly and significantly related to alcohol use and participation in delinquent behaviors (Warr,

2002).

Social Learning Theory and Prostitution

As noted above, social learning theory argues that individuals learn behavior from others, particularly family members and peers, through definitions, positive reinforcement, and imitation. It has been established that peers are an important mechanism in adolescents becoming involved in deviant behaviors but relatively little is known about the role of peers as a factor affecting entry into prostitution. It is anticipated that this research will document that peers play an important role in adolescents’ entry into prostitution for several reasons. First, friends and family members are the primary groups that an individual associates with

(differential association). Secondly, while friends and/or family members who are prostitutes may believe that it is morally wrong to murder someone, they may also believe that earning a living by sleeping with customers is not wrong; consequently the prostitution behavior is rationalized and it is believed that breaking this law is acceptable.

These definitions would more than likely be clearly known by all who socialize with individuals who are prostitutes. For example, if females see their friends with flashy cars, new clothes, or spending time with older men, these females may want to do the same exact things that their friends are in order to receive these valuable things without realizing the environments and conditions that they will be placing themselves in. In addition, if an adolescent sees her family members or friends with flashy clothes and money earned via prostitution, the behavior of sleeping with others for money is positively reinforced through these “positive” outcomes.

Moreover, if prostitutes have a pimp, the pimp may offer a “bonus” to any prostitute who encourages another female to become a prostitute. Consequently, even if the conditions of being

37 a prostitute are dismal, the prostitute may exaggerate the positive factors of prostitution in order to entice another female into this life for selfish gains. Also, if a family member or a peer is a prostitute, the adolescent may be more likely to see the techniques of catching the eye of a customer and understanding how to handle business with the customer. As a result, the adolescent will know the “tricks” of the prostitution business and how to act like a prostitute; thus, the act of sleeping with strangers for objects of value is normalized. Moreover, these individuals may trust their friends and/or family members’ opinions and see the “normalization” of sleeping with strangers for objects of value. Therefore, the likelihood of an adolescent becoming a prostitute is higher and these behaviors will be mimicked and/or learned.

Therefore, it is hypothesized that having deviant peers will increase the odds of prostitution. However, it is also hypothesized that having delinquent peers will influence one to use drugs (via peer pressure) which will in-turn lead to prostitution in order to get money for the new drug habit. Lastly, it is also predicted that attachment and supervision will condition the effect of deviant peers on prostitution because the more one is supervised and attached to parents, the less time there is available to be with delinquent peers. Please see Figure 3 (p. 118) documenting the relationship between having friends who are prostitutes and one becoming a prostitute herself.

The Relationship of General Strain Theory, Control Theory, & Social Learning Theory

Although three separate theories are utilized to examine factors that have been seen as important variables associated with prostitution, they are all interrelated. For example, if an adult (or older) family member or significant other of a parent perpetrates physical or sexual abuse (source of strain) on an adolescent victim, this presumes that appropriate adult supervision is lacking in the household (social bonding theory). Moreover, the bond between the adolescent

38

and a parent may be broken or weakened if the parent permits the abuse to continue or if the

parent does not believe the adolescent’s reports of abuse. Therefore, viewing the abuse experienced as a sense of strain can increase the feelings of anger and unjustness; consequently, if such adolescents lack legitimate coping mechanisms, they may resort to illegitimate coping mechanisms, such as running away from home. As a result of running away from home, many abused adolescents find themselves participating in delinquent activities in order to survive on the streets, such as drug use and prostitution, both of which can be imitated by viewing other adolescent runaways participating in these activities or becoming friends with other runaways who commit these behaviors. However, if the bond between an adolescent and a legitimate parental figure is strong, this bond may buffer against the negative impacts of abuse. Strong bonds with parental figures may also buffer against the influences by peers to participate in delinquent activities, including prostitution. Weak parental attachment though can create an environment conducive to abuse, which will lead to running away and consequently involvement in prostitution. Thus, it is important to note that components of each of these theories, used in conjunction with another via mediating or moderating variables, make the analysis more comprehensive and more likely represent the complexity of the factors associated with prostitution.

Summary

General strain theory argues that experiencing strain (failure to achieve positively valued goals, removal of positively valued stimuli, and confrontation with negative stimuli) increases the likelihood that individuals will feel some type of negative emotion, which increases the likelihood of participating in delinquent and criminal activities (which are seen as an adaptation to stress). Therefore, general strain theory argues that experiencing strain (e.g. sexual abuse,

39

contentious family environment, parental conflict) can cause negative emotions (e.g.

psychological distress), which interferes with the coping strategies of individuals, consequently

leading to participation in criminal activities, such as prostitution. General strain theory is

employed in this research in order to examine the relationship between sexual abuse and

prostitution. It is predicted that there is a strong positive relationship between sexual abuse and

prostitution. It is also predicted that the relationship between abuse and prostitution is mediated

by running away (which is also mediated by deviant peers and drug abuse/use) and moderated by

attachment and psychological distress.

Control theory, which claims that strong attachments decrease the likelihood of

participating in delinquent activities, is utilized in this research to determine the relationship

between parental attachment as an adolescent and subsequent prostitution. It is hypothesized that those respondents with higher levels of attachment and higher levels of supervision have lower odds of prostitution. In addition, it is predicted that deviant peers will mediate the relationship between parental attachment and supervision whereas sexual abuse will moderate the effect of attachment on prostitution. Lastly, it has been established that peers are an important mechanism in becoming delinquent but the role of peers has not been frequently examined for its role as a factor associated with prostitution. Consequently, social learning theory is used in this research in order to determine if peers are an important factor in females’ entry into prostitution. It is hypothesized that peers are an important influence in persuading women to become involved in prostitution, just as they are in other delinquent behaviors. It is also predicted that drug abuse/use will mediate the relationship between deviant peers and prostitution whereas parental attachment and supervision will moderate the effect of deviant peers on prostitution.

40

Hypotheses

The previous discussion of the prostitution literature and theoretical framework suggest

(1) there are correlations between prostitution, abuse, and running away, (2) strong positive

attachments to a parent and high levels of supervision may decrease the likelihood of deviant behavior, including prostitution, (3) peers are influential in their effect on deviant behavior,

including prostitution, and (4) there are positive correlations between drug abuse/use,

psychological distress, and prostitution. The following hypotheses are tested in this research.

Testing the Relationship Between the Strain Theory Measures and Prostitution

Previous research has shown that sexual abuse leads to prostitution. Consequently,

Hypothesis 1-- Higher levels of sexual abuse increase the odds of prostitution -- is assessed in

order to determine if this widely held assumption is true using a sample that is not prostitute

biased. This hypothesis is also examined in order to explore whether the experience of negative

stimuli, such as abuse, contentious family environment, and conflict with parents, leads to

prostitution, as proposed by strain theory. In addition, this relationship is also assessed using

components of social learning theory (deviant peers) as mediators and social control theory

(attachment) as moderators. Other factors thought to be associated with prostitution are also

used as mediators (running away and drug abuse/use) and moderators (depression and self-

esteem) in order to further assess the relationship between negative stimuli and prostitution (See

Figures 1a and 1b, on p. 116). It is hypothesized that psychological distress positively affects the

strength of the relationship between abuse and prostitution whereas attachment negatively affects

the strength of this same relationship. In addition, it is proposed that running away is a

significant mediator in the relationship between abuse and prostitution whereas, deviant peers

41

and drug abuse/use are significant mediators in the relationship between running away and prostitution.

Testing the Relationship Between the Control Theory Measures and Prostitution

It is argued that low levels of parental attachment and supervision (components of social control theory) during adolescence lead to participation in delinquent behaviors, in this case, prostitution. As a result, Hypothesis 2 tests the following relationship: Respondents with higher

levels of parental attachment have lower odds of being prostitutes than those respondents with

lower levels of parental attachment. In addition, Hypothesis 3 proposes that respondents who

receive higher levels of parental supervision have lower odds of being prostitutes than those who

receive lower levels of parental supervision. However, these relationships are further analyzed by utilizing components of other theories, such as strain theory (sexual abuse as a moderator) and social learning theory (delinquent peers as a mediator) in order to better explain the relationship between parental supervision and prostitution as well as between parental attachment and prostitution (See Figures 2a and 2b, p. 117). It is hypothesized that deviant peers are a

significant mediator in the relationship between parental supervision and prostitution as well as

in the relationship between attachment and prostitution. It is also conjectured that abuse

negatively affects the strength of the relationship between parental attachment and prostitution.

Testing the Relationship Between the Social Learning Theory Measures and Prostitution

A common argument in the criminology literature is that peers are an important

component of adolescents’ participation in delinquent activities. However, this relationship has

not frequently been analyzed specifically for prostitution. Consequently, Hypothesis 4--Those with delinquent friends have higher odds of becoming prostitutes than those who do not have delinquent friends--is examined in order to determine if this social learning theory relationship

42

holds true for prostitution. This relationship is further analyzed using components of social

control theory (attachment and supervision) as moderators and drug abuse and use as mediating variables to see if and how the relationship in Hypothesis 4 changes (See Figure 3, p. 118). It is hypothesized that higher levels of parental attachment and supervision conditions the effect of deviant peers on prostitution. It is also hypothesized that drugs mediate the relationship between

having deviant peers and prostitution.

43

CHAPTER 4: METHODS Data

The data used in this research comes from the Ohio Lifecourse Study (OLS). The OLS is a multi-wave dataset of household (in the Toledo, Ohio Metropolitan Area) as well as institutionalized (in several correctional facilities in Ohio) respondents who were originally interviewed in 1982. In 1982, there were a total of 942 neighborhood respondents, with an average age of 15, and 254 institutional respondents who averaged 16 years of age. The data were collected using face-to-face interviews and includes, for example, questions in the subject areas of family, peers, school, work history, and participation in delinquent behaviors.

These groups of individuals were interviewed again in 1992 (neighborhood respondents) and in 1995 (institutional respondents). The respondents were asked a variety of questions including their romantic relationships, participation in criminal activities, and abuse experienced as an adolescent. In this second-wave, there were a total of 721 neighborhood respondents and

209 institutional respondents. The neighborhood respondents were, on average, 25 years old at the time of the first follow-up survey whereas the institutional respondents were, on average, 29 years old at the first follow-up.

A third wave of data was collected in 2003 but only those respondents who were institutionalized in 1982 were re-interviewed. In this wave of data, there are a total of 152 respondents (76 males, 76 females, 57 blacks, and 90 whites) with an average age of 37 years.

Respondents were asked about a variety of content areas, including their current romantic relationships, work history, participation in criminal activities, and arrest history. Although all of the respondents from the institutionalized sample were in juvenile correctional facilities in

1982, many were out of the correctional system during the collection of Wave 3 data.

44

It should be noted that although there are male prostitutes, this research only examines

female prostitution because there are only 16 (22%) male prostitutes in this sample, hardly

enough to conduct meaningful analysis. Similar concerns arose regarding whether the two

samples (neighborhood and institutionalized) could be combined in this research. A Chow Test

was run by the researcher in order to determine if the independent variables affect the

2 institutionalized and neighborhood samples in similar or different ways. The Chow Test (χ (5) =

17.335, p < 0.05) concluded that these two groups are affected differently by the independent variables and thus must be analyzed separately. However, because there are only six neighborhood respondents who reported being paid to have sex in any of the three waves, not nearly enough to perform meaningful analysis, only the previously institutionalized sample is utilized in the analyses to follow.

The final analytic sample is comprised of 73 previously institutionalized female respondents from the OLS data that were interviewed in all three waves. Although these respondents are from a high-risk delinquent sample rather than a general population sample, there is considerable variation among them along key dimensions. For example, the sample contains non-prostitutes who were sexually abused as adolescents (N=11 or 15 percent), prostitutes who were sexually abused as adolescents (N=19 or 26 percent), prostitutes who were not sexually abused as adolescents (N=17 or 23 percent), and non-prostitutes who were not sexually abused as adolescents (N=26 or 36 percent). Moreover, because the sample is skewed toward the lower and working classes, these respondents are more likely to be street prostitutes rather than the higher class call girls.

The OLS data have several advantages for this study. First, this is one of very few longitudinal studies available that can be used to examine participation in prostitution and other

45

behaviors in more than one wave, which increased the number of respondents in the prostitute sample. Second, the OLS include questions about abuse experienced as a child, the usage of

drugs, the respondent’s psychological distress, the amount of parental control and supervision the

respondent experienced as an adolescent, parental attachment, peer involvement and influence,

and how often the respondent ran away from home as an adolescent, all of which are seen as

potential factors associated with prostitution. In addition, this is one of very few quantitative studies that can be used to study prostitution that is not a prostitute-biased sample. This is

important because most of the quantitative data that are currently available to study prostitution

are prostitute-biased samples with little variation among respondents on key factors. Lastly, the

OLS contains a variety of respondents, including those who have not been paid to have sex,

those from a variety of racial backgrounds, as well as those who were abused as adolescents but

are not prostitutes. These different types of respondents allow for a better comparison of differences between prostitutes and non-prostitutes and thus enable a better understanding of the factors associated with prostitution.

Measures

Dependent Variable

Being paid to have sex

Being paid for sex is the dependent variable in this research (“In the past 12 months, how often have you been paid for having sex with someone?”). This question is asked in all three waves of the OLS (1982, 1995, and 2003). Thirty-three percent of respondents in 1982, 29 percent of respondents in 1995, and 14 percent of respondents in 2003 report being paid to have sex with someone at least once or twice a year. In addition, three respondents report being paid

to have sex with someone in all three waves (4 percent), 21 respondents report being paid to have

46 sex with someone in only one of the waves (29 percent), 12 respondents report being paid to have sex with someone in two waves (16 percent), and 37 respondents report never being paid to have sex with someone in all three waves (51 percent).

Because their responses are skewed towards the lower frequencies, being paid for sex is coded as a binary variable. Those respondents who report being paid to have sex (“once or twice a year” to “once a day”) in any of the three waves are coded as a “prostitute” (coded 1). On the other hand, those respondents who report “never” being paid to have sex in all of the three waves are coded as a “non-prostitute” (coded 0). Of the 73 female respondents in this study, there are

36 prostitutes (49 percent) and 37 non-prostitutes (51 percent).

Independent Variables

Sexual Abuse Experienced as an Adolescent

This six-item scale measures sexual abuse experienced as a child or teenager and is based on the following questions asked retrospectively in 1995: “While you were growing up, did anyone do any of the following: ever make you do something sexual that you did not want to do; ever make you touch their breasts or genitals, or touch yours when you did not want them to; make you look at them naked, or look at you naked, or ever take sexual photographs of you, when you did not want them to; try, unsuccessfully to have sexual intercourse with you when you did not want to by either using force or threatening to use force; force you to have sexual intercourse either because someone used force or threatened you with physical violence; get you to have sexual intercourse when you did not want to because they had given you alcohol or drugs.” Each item is coded such that the response category of “yes” is coded as 1 and “no” is coded 0. The scores for each item are summed and then averaged in order to create a scale that empirically ranges from 0 to 6. Higher scores reflect higher levels of sexual abuse. For the one

47

missing case, the imputed mean of 2.0138889 is used. The standardized Chronbach alpha

coefficient for the six variables that create the sexual abuse scale is 0.878456.

Contentious Family Environment

This scale measures the level of negative family environment encountered as an adolescent. Five items, based on retrospective reports in 1995, comprise this scale: “How often would you say these kinds of things happened in your family: People getting on each others nerves; My parents getting on my case about something I didn't do right; My parents putting their own needs above ; Fighting and arguing; People just "going their own way."

Response categories range from “never” (coded 0) to “very often” (coded 4). This scale, with actual scores ranging from two to nineteen, is coded such that higher scores reflect higher levels of contentious family environment. The standardized Chronbach alpha coefficient for the five items that create the contentious family environment scale is 0.611834.

Conflict with Parents in 1982

This scale measures the level of conflict with parents in 1982. The two items that comprise this scale include “How often do you have disagreements or arguments with your parents” and “How often do you purposely not talk to your parents because you are mad at them?” The response categories for these two items range from “hardly ever or never” (coded 1) to “two or more times a week” (coded 5). This scale, with scores ranging from two to ten, is coded such that higher scores reflect higher levels of conflict with parents. For the four missing

cases, the imputed mean of 5.6376812 is used. The standardized Chronbach alpha coefficient for

the two items that create the conflict with parents at Time 1 scale is 0.620712.

48

Conflict with Parents in 1995

This measures the level of conflict with parents in 1995. The two items that create this

scale include “How often in the past 12 months have you had disagreements or arguments with

either of your parents?” and “How often in the past 12 months have you purposely not talked to

your parents because you were mad at them?” The response categories for these two items range

from “never or almost never” (coded 1) to “almost every day” (coded 6). Scale scores range

from two to twelve, with high scores reflecting high levels of conflict with parents. The

standardized Chronbach alpha coefficient for the two items that are used to create the conflict

with parents at Time 2 scale is 0.608232.

Attachment to Parents: Parental Control and Supervision

This three-item scale measures the level of parental control and supervision the

respondent had as an adolescent in 1982. The questions that create this scale include, “My

parents want to know who I am going out with when I go out with other girls,” “In my free time

away from home, my parents know who I'm with and where I am,” and “My parents want me to

tell them where I am if I don't come home right after school.” The response categories for these

questions range from “strongly disagree” (coded 1) to “strongly agree” (coded 5). Thus, scale

scores range from five to fifteen. This variable is coded so that higher values reflect a higher

level of parental control and supervision at Time 1. For the three missing cases, the imputed

mean of 10.8571429 is used. The Chronbach standardized alpha coefficient for the level of

parental control and supervision is 0.573082.

Previous research (Cernkovich & Giordano, 1987; Laub & Sampson, 1988) has found that family bonding/attachment is directly related to delinquency. The assumption in this research is that these variables should similarly effect prostitution, as one specific type of

49

delinquent/deviant behavior. As a result, four independent variables will measure family

bonding/attachment in this research.

Attachment to Parents: Caring and Trust in 1982

This six-item scale measures the level of parental caring and trust in 1982. The items that

create this scale include “My parents often ask about what I am doing in school,” “My parents

give me the right amount of affection,” “One of the worst things that could happen to me would be finding out that I let my parents down,” “My parents are usually proud of me when I've finished something I've worked hard at,” “My parents trust me,” and “I'm closer to my parents

than a lot of kids my age are.” The response categories for these items range from “strongly

disagree” (coded 1) to “strongly agree” (coded 5). Scale scores range from eight to thirty and are

coded so that higher scores reflect higher levels of parental caring and trust. For the three

missing cases, the imputed mean of 22.7142857 is used. The standardized Chronbach alpha coefficient for the six variables that are used to create the parental caring and trust scale at Time

1 is 0.840825.

Attachment to Parents: Caring and Trust in 1995

This six-item scale measures the level of parental caring and trust in 1995. The items that create this scale include “My parents often ask about what I am doing at work (or in college or during the day),” “My parents give me the right amount of affection,” “One of the worst things that

could happen to me would be finding out that I let my parents down,” “My parents are usually

proud of me when I've finished something I've worked hard at,” “My parents trust me,” and “I'm

closer to my parents than a lot of kids my age are to theirs.” The response categories for these

items range from “strongly agree” (coded 1) to “strongly disagree” (coded 5). Scale scores range

from nine to twenty-nine and are recoded such that higher scores reflect higher levels of parental

50

caring and trust. The standardized Chronbach alpha coefficient for the six variables that are used

to create the parental caring and trust scale at Time 2 is 0.713767.

Attachment to Parents: Intimate Communication with Parents in 1982

This three-item scale measures the level of intimate communication the respondent had

with parents in 1982. The questions that are used to comprise the scale include “How often do

you talk to your parents about the following things: The boy/girl whom you like very much;

Questions or problems about sex; Things you have done about which you feel guilty.” The

response categories for these items range from “never” (coded 0) to “very often” (coded 4).

Scale scores range from zero to twelve and are coded so that higher values reflect higher levels

of intimate communication with parents at Time 1. For the three missing cases, the imputed

mean of 5.9714286 is used. The standardized Chronbach alpha coefficient for the three variables

that are used to comprise the intimate communication with parents scale at Time 1 is 0.797587.

Attachment to Parents: Intimate Communication with Parents in 1995

This scale measures the level of intimate communication the respondent had with parents

in 1995. The two questions that comprise the scale are “About how often do you talk to one or

both of your parents (or closest living relative if parents deceased) about these things: Your

marriage (or boyfriends/girlfriends); Things you have done about which you feel guilty.” The

response categories for these items range from “never” (coded 0) to “very often” (coded 4).

Scale scores range from zero to eight and are coded such that higher values reflect higher levels

of intimate communication with parents at Time 2. The standardized Chronbach alpha

coefficient for the two variables that comprise the intimate communication with parents scale at

Time 2 is 0.368197. This low alpha may be a consequence of the nature of the sample, which is

51

comprised entirely of previously institutionalized respondents, who generally report low levels of

parental attachment.

Friends Involved in Prostitution in 1995

This independent binary variable measures whether the respondent had friends who were

paid to have sex with someone in 1995. Those who report “none of them” are coded “0.” On the

other hand, the response categories of “a few of them,” “some of them,” and “most of them” are

coded “1” to reflect that the respondent had at least a few friends who were paid to have sex with

someone. These latter categories are collapsed into one because the distribution is skewed

toward the lower response values. For the ten missing on this variable, the imputed mean of

0.6349206 is used. As a result of the mean being greater than zero, these ten respondents are

placed in the category of having at least a few friends who were paid to have sex with someone

(coded 1). This is also done because there is a good possibility that respondents not answering

this question actually had friends who were paid to have sex with someone in 1995, but did not

feel comfortable reporting this.

Peer Pressure in 1982

This six-item scale measures the level of peer pressure the respondent experienced in

1982. The items that comprise the scale include “I sometimes do things because my close friends are doing them,” “I sometimes do things because that's what the popular kids in school are into,” “I sometimes do things because my friends give me a hard time or hassle me until I do them,” “I don't like being different or sticking out in a crowd so I sometimes go along with things for that reason,” “I sometimes do things not because my friends pressure me but just because I think it will impress them,” “I sometimes do things because I don't want to lose the respect of my friends.” The response categories for these items range from “strongly disagree”

52

(coded 1) to “strongly agree” (coded 5). Consequently, scale scores range from six to twenty-

eight and are coded so that higher values reflect higher levels of peer pressure. For the one

missing case, the imputed mean of 17.1805556 is used. The standardized Chronbach alpha

coefficient for the six variables that are used to create the Time 1 peer pressure scale is 0.800432.

Peer Pressure in 1995

This six-item scale measures the level of peer pressure the respondent experienced in

1995. The items that comprise this scale include “I sometimes do things because my close

friends are doing them,” “I sometimes do things because that's what successful people are into,”

“I sometimes do things because my friends give me a hard time or hassle me until I do them,” “I

don't like being different or sticking out in a crowd so I sometimes go along with things for that

reason,” “I sometimes do things not because my friends pressure me but just because I think it

will impress them,” and “I sometimes do things because I don't want to lose the respect of my

friends.” The response categories for these items range from “strongly agree” (coded 1) to

“strongly disagree” (coded 5). Scale scores range from six to twenty-seven and are recoded so

that higher values reflect higher levels of peer pressure. For the two missing cases, the imputed

mean of 14.3802817 is used. The standardized Chronbach alpha coefficient for the six variables

that are used to create the Time 2 peer pressure scale is 0.850451.

Drug Abuse in 1995

This five-item scale measures the respondent’s level of drug abuse in 1995. The items

that create this scale include “How often in the past twelve months have you: Not felt so good the next day because of using drugs; Felt unable to do my best job at work or school because of using drugs; Gotten into trouble with my relatives or friends while using drugs; Hit one of my family members because of my using drugs; Gotten into fights with others because of my using

53

drugs; Stolen money or other things in order to get cash to buy drugs.” The response categories for these items range from “never” (coded 0) to “2-3 times a week or more” (includes the categories of “2-3 times a week,” “once a day,” and “more than once a day”), which is coded

“6.” Scale scores range from zero to thirty and are coded such that higher values reflect higher

levels of drug abuse at Time 2. For the one missing case, the imputed mean of 4.0833333 is

used. The standardized Chronbach alpha coefficient for the five variables that are used to create

the drug abuse in 1995 scale is 0.903674.

Drug Abuse in 2003

This five-item scale measures the respondent’s level of drug abuse in 2003. The items

that comprise this scale include “How often in the past twelve months have you: Not felt so good

the next day because of using drugs; Gotten into trouble with my relatives or friends while using

drugs; Hit one of my family members because of my using drugs; Gotten into fights with others

because of my using drugs; Stolen money or other things in order to get cash to buy drugs.” The

response categories for these items range from “never,” (coded 0) to “2-3 times a week or more”

(includes the categories of “2-3 times a week,” “once a day,” and “more than once a day”),

which is coded “6.” Scale scores range from zero to thirty and are coded so that higher values

reflect higher levels of drug abuse at Time 3. The standardized Chronbach alpha coefficient for

the five variables that are used to create the drug abuse in 2003 scale is 0.939222.

Used Drugs To Get High in 1982

This independent variable measures the respondent’s use of drugs to get high in 1982 (“In

the past 12 months, how often have you used drugs to get high (not because you were sick)”).

The response categories for this variable range from “never” (coded 0) to “2-3 times a week or

more” (includes the categories of “2-3 times a week,” “once a day,” and “more than once a

54

day”), which is coded “6.” Scores for this continuous variable range from zero to six and are coded so that higher values to reflect higher levels of using drugs to get high at Time 1.

Used Drugs To Get High in 1995

This independent continuous variable measures the respondent’s use of drugs to get high in 1995 (“In the past 12 months, how often have you used drugs to get high (not because you were sick)”). The response categories for this variable range from “never” (coded 0) to “2-3 times a week or more” (includes the categories of “2-3 times a week,” “once a day,” and “more than once a day”), which is coded “6.” Scores for this variable range from zero to six and are coded such that higher values reflect higher levels of using drugs to get high at Time 2.

Used Drugs To Get High in 2003

This variable measures the respondent’s use of drugs to get high in 2003 (“In the past 12

months, how often have you used drugs to get high (not because you were sick)”). The response

categories for this variable range from “never” (coded 0) to “2-3 times a week or more” (includes

the categories of “2-3 times a week,” “once a day,” and “more than once a day”), which is coded

“6.” Scores on this continuous variable, which are coded so that higher values reflect higher

levels of drug use at Time 3, range from zero to six.

Running Away From Home

This continuous variable measures how often the respondent ran away from home in

1982 (“In the past 12 months, how often have you run away from home?”). The response

categories for this variable range from “never” (coded 0) to “2-3 times a week or more” (includes

the categories of “2-3 times a week,” “once a day,” and “more than once a day”), which is coded

“6.” For the one missing case, the imputed mean of 1.9305556 is used. This variable is coded so

that higher values reflect a higher frequency of running away and ranges from zero to six.

55

Psychological Distress: Self-Esteem in 1995

This six-item scale measures the respondent’s self-reported level of self-esteem in 1995,

which is based on six questions including, “What is your level of agreement with the following

questions: I can do things as well as others; I have a number of good qualities; I do not have much to be proud of; I am no good at all; I am a person of worth; I have a positive attitude

towards myself.” The response categories for self-esteem in 1995 are recoded in order to range

from “strongly disagree” (coded 1) to “strongly disagree” (coded 5). Therefore, self-esteem at

Time 2 is coded so that higher values reflect higher levels of self-esteem and range from fifteen

to twenty-eight. The standardized Chronbach alpha coefficient for the six variables that are used

to create self-esteem at Time 2 is 0.694363.

Psychological Distress: Self-Esteem in 2003

This six-item scale measures the respondent’s self-reported level of self-esteem in 2003.

At Time 3, self-esteem is based on the respondent’s agreement to the following questions: “I can

do things as well as others; I have a number of good qualities; I do not have much to be proud of;

I am no good at all; I am a person of worth; I have a positive attitude towards myself.” The

response categories for self-esteem in 2003 range from “strongly disagree” (coded 1) to

“strongly agree” (coded 5). Scores on this continuous variable are coded so that higher values

reflect higher levels of self-esteem and range from six to thirty. The standardized Chronbach

alpha coefficient for the six variables that are used to create the self-esteem scale at Time 3 is

0.789598.

Psychological Distress: Depression in 1995

This eight-item scale measures the respondent’s self-reported level of depression in 1995,

which is based on the following questions: “In the past twelve months, how often have you:

56

wondered if anything was worthwhile; been in low spirits; had trouble sleeping; had periods of

time where you could not get going; felt things never turn out right; had trouble remembering things; felt irritable, fidgety, or tense; felt restless?” Depression is coded along an 8-point scale

ranging from 0 (“never”) to 7 (“everyday”). Thus, depression at Time 2 is coded so that higher

values reflect higher levels of depression and ranges from zero to forty. The standardized

Chronbach alpha coefficient for the eight variables that are used to create depression at Time 2 is

0.882083.

Psychological Distress: Depression in 2003

This seven-item scale measures the respondent’s self-reported level of depression in

2003, which is based on the respondent’s answers to the following questions: “In the past seven days, how often have you: been unable to get going, could not shake the blues, had trouble keeping your mind on things, felt lonely, felt sad, had trouble sleeping, and felt everything was an effort?” Depression level ranges from 0 (“never”) to 7 (“everyday”). Thus, depression at

Time 3 is coded such that higher values reflect higher levels of depression and ranges from zero to forty-five. The standardized Chronbach alpha coefficient for the seven variables that are used to create depression at Time 3 is 0.910515.

Control Variables

Prior research has shown that certain demographic variables predict participation in delinquent activities. These variables are controlled in the analyses.

Race

This variable measures the race of the respondent as reported in 1995. The 1995 variable is utilized because the respondent’s race is not asked at Time 3; consequently, the respondent’s reported race in 1995 is the most currently available documentation. Although some respondents

57 chose “Hispanic” or “Other Race,” there are only three respondents who are considered “another race” (Hispanics and other). As a result, these respondents are excluded from the analyses.

Therefore, race is measured by two dummy variables including “white” and “black.” The race variables are coded so that the response category of “yes” is coded “1” and the response category of “no” is coded "0.” In this research, there are 38 whites (52 percent) and 35 blacks (48 percent) with whites being the reference category.

Race is a control variable in this study for various reasons. First, the sample utilized in this research is very small (N = 73). As a result, if the sample is broken down by race, the sample size would get even smaller, potentially making variables that may be significant if a larger sample is used non-significant. In addition, previous research has also not used race as an independent variable due to small population sizes and the difficulty of obtaining willing respondents to participate in studies on prostitution.

Employment Status in 1995

This variable measures the respondent’s employment status in 1995 and is based on the following question “[Please think back to your present or last job as you answer these next questions] Is (was) this job full-time or part-time?” Three dummy variables are initially utilized including “full-time,” “part-time,” and “unemployed.” These variables are coded so that the response category of “yes” is coded “1” and “no” is coded “0.” Forty-four respondents (61 percent) report having a full-time job, twenty-five (34 percent) report a part-time position, and four (5 percent) report being unemployed in 1995. As a result of the small number of respondent’s being “unemployed” in 1995 and a quasi-complete separation error message1

1 This error message is received because all of the prostitutes report unemployment in 1995 while none of the non- prostitutes report unemployment in 1995; as a result of all the unemployed respondents being characterized as “unemployed,” legitimate comparisons could not be made.

58

received while running the analyses, the categories of “part-time” and “unemployed” are

combined to create an “other employment” category. As a result, there a total of 29 (40 percent)

respondents that report having “other employment” status in 1995. Full-time employment status

in 1995 is the reference category.

Employment Status in 2003

This control variable measures the respondent’s employment status in 2003 initially using

three dummy variables, including “full-time,” “part-time,” and “unemployed.” These variables are coded so that the response category of “yes” is coded “1” and “no” is coded “0” to the

following question: “Please think about you present or last job as you answer these next

questions. Is [was] this job full-time or part-time?” In this research, 45 respondents (62 percent) report having a full-time job, 15 (21 percent) report a part-time position, and 13 (18 percent) report being unemployed in 2003. However, in order to be consistent with the employment variables at Time 2, the categories of “part-time” and “unemployed” are also combined for Time

3. Consequently, there are a total of 28 (38 percent) respondents with “other employment” at

Time 3 in this research. Full-time employment status in 2003 is the reference category.

Parental Social Class

Although parental education and occupational status are not major interests in this research, they are included because of the well-known relationship between parental social class and participation in delinquent behaviors. This continuous variable measures the parents’ social class in 1982. First, the level of education for both mothers/stepmothers and fathers/stepfathers is created using the following two questions, “How much education does your mother/stepmother have?” and “How much education does your father/stepfather have?” The response categories range from “some high school or less” (coded 1) to “more than 4 years of

59

college” (coded 5). As a result, parental education is coded such that higher response values

reflect higher levels of parental education. The response values for each of these questions are

then added together in order to obtain a combined parental education value. If a value is missing

on one of these questions, the response for the one question without missing data is then used as

the combined parental education score. For the eight respondents that have both questions missing, the imputed mean of 3.3846154 is used.

In order to determine the occupational status of both parents, the following questions are utilized “What kind of work does your mother/stepmother usually do?” and “What kind of work does your father/stepfather usually do?” The response categories range from “Handlers, equipment cleaners, helpers, and laborers” (coded 1) to “Executive, administrative, managerial”

(coded 7). As a result, parental occupational status is coded such that higher response values

reflect higher levels of parental occupational status. If a value is missing on one of these

questions, the response for the one question without missing data is then used as the combined

parental occupational score. For the eight respondents missing on both the mother and father

occupational status variables, the imputed mean of 4.20 is used.

The scores for parental education and parental occupational status are then standardized

and added together in order to obtain a standardized combined parental social class score, which

empirically ranges from -1.3144821 to 2.6869027.

Age in 1982

The well-known age-crime curve indicates that age is a powerful predictor of antisocial

behavior. This continuous variable measures the respondent’s age at Time 1 and is calculated by

subtracting the respondent’s year of birth from 1982 (time of the Wave 1 survey). Age at Time 1

ranges from thirteen to nineteen and has an average of 16.5 years old.

60

Quantitative Analytic Strategy

The first quantitative analytic section in this research includes bivariate statistics for the

OLS sample. These statistics include all of the independent, control, and mediating/moderating variables. These statistics are used in order to determine if the significance levels are similar at the bivariate and regression levels. Moreover, statistical differences are examined not only between prostitutes and non-prostitutes but also between white prostitutes and white non- prostitutes, black prostitutes and black non-prostitutes, as well as white prostitutes and black prostitutes.

The second quantitative analytic technique used in this research is regression using a binary response variable and logit regression estimates in order to examine the factors associated with prostitution (also called binary logistic regression). This technique is used for several reasons. First, due to the skewed nature of the dependent variable (being paid to have sex), regression using a binary response variable is the most accurate analytic technique. Moreover, because the total sample size is only seventy-three in this research (36 prostitutes and 37 non- prostitutes), using multinomial logistic regression is not possible. Although it would have been interesting to create categories of “persister,” (report being paid to have sex in all three waves)

“desister,” (report never being paid to have sex in all three waves) and “unstable” (report being paid to have sex in some waves and not others), these categories are not able to be utilized due to the small sample size overall, as well as the small numbers that would have been in each category. In addition, if these categories had been used, variables that may have been significant if a larger sample size had been used would probably not be significant. Furthermore, although logit or probit estimates could be used for this research, logit estimates are used due to the easy interpretability of the odds ratios.

61

Moreover, this analysis technique (regression using logit estimates and a binary response

dependent variable) is important because most of the data that is available in the literature does

not include much variation among the respondents and this sample of respondents enables the

researchers to do this. In this study, of those who were sexually abused as adolescents, 19 (63

percent) are prostitutes and 11 (37 percent) are not prostitutes. Moreover, there are a total of 22

(30 percent) respondents that report experiencing both sexual and physical abuse as adolescents,

8 (11 percent) who report experiencing sexual abuse only as adolescents, 13 (18 percent) who report encountering physical abuse only as adolescents, and 30 (41 percent) respondents who report no experience with any type of abuse as adolescents. Although physical abuse is not a

focus in this research, these statistics are reported to emphasize the variation among the

respondents in this study. However, it is important to keep in mind that this sample is highly

delinquent and disadvantaged and is not a general population sample; therefore, the results are

not generalizable to other populations.

Three sets of regression models are used in this research. The first set of models

investigates the effects of the strain theory measures (sexual abuse, contentious family

environment, and conflict between parents) and their association with prostitution. These models

also utilize components from social control theory (attachment as moderating variables) and

social learning theory (delinquent peers as mediating variables) as well as drug abuse/use

(mediating variable) and psychological distress (moderating variables) in order to understand the complex relationship between the strain theory measures and prostitution.

The second set of regression models examines the effects of the control theory measures

(parental control and supervision, parental caring and trust, and parental intimate communication) and their association with prostitution. Moreover, elements of strain theory

62

(sexual abuse as a moderator) and social learning theory (delinquent peers as mediators) are used in order to better explain the relationship between parental attachment and prostitution as well as parental supervision and prostitution.

Lastly, the third set of regression models analyzes the effects of the social learning theory measures (having friends that were paid to have sex and peer pressure) on prostitution. In addition, these models also use components from social control theory (attachment and supervision as moderators) as well as drug abuse and use as mediating variables to determine if these variables can help better explain the relationship between having delinquent friends and becoming a prostitute.

Qualitative Analytic Strategy

In addition to the two quantitative analytic strategies, this research also uses the qualitative components of the OLS, a very unique and valuable aspect of these data. Due to the large number of narratives that are available in the OLS data, this research only uses a small portion of them because more extensive analysis is beyond the scope of this research. The narratives that are used are chosen because of the detailed and distinctive insight they offer regarding why some of these women became prostitutes. Consequently, these detailed life- history narratives are used in this research to support findings from the quantitative components as well as to explore elements of becoming a prostitute that could not be examined or supported using the quantitative components in this research.

63

CHAPTER 5: RESULTS

This chapter first presents the descriptive statistics for the dependent, independent, and control variables used in the analysis.

Descriptive Statistics and Bivariate Relationships

Comparison of Prostitutes and Non-prostitutes

The data in Table 1 (p. 119) demonstrate that prostitutes and non-prostitutes are significantly different on most of the factors associated with prostitution, control variables, and mediating/moderating variables included in this research. This is interesting because both groups (prostitutes and non-prostitutes) are both highly delinquent and disadvantaged.

Therefore, the prostitutes are even more disadvantaged than the non-prostitutes. Consequently, there must be something about prostitution that makes them (the prostitutes) even more disadvantaged than their highly delinquent and disadvantaged counterparts. However, it is important to note that most of the variables associated with strain theory (contentious family environment and conflict between parents) as well as all of the variables associated with social control theory (parental control and supervision, parental caring and trust, and parental intimate communication) are not significant, indicating that prostitutes and non-prostitutes are not significantly different from each other on any of these variables, on average. This is interesting considering that it has been argued that a lack of parental attachment, especially parental supervision, leads to participation in delinquent activities. Perhaps the parental attachment argument does not apply to all delinquent activities, such as prostitution, at least at the bivariate level. On the other hand, this could be a result of the nature of the sample used in this research such that very few of the respondents in the institutionalized sample have very stable or positive

64 family backgrounds; thus, supporting the argument that a lack of parental attachment leads to participation in delinquent activities.

On the other hand, prostitutes and non-prostitutes are significantly different on a variety of other variables. For example, prostitutes are found to have experienced higher mean levels of sexual (2.75 versus 1.30, p<0.01) and physical abuse (3.03 versus 2.27, p<0.10) than non- prostitutes, and report a higher percentage of friends who had been paid to have sex in 1995 (56 percent versus 22 percent respectively, p<0.01). In addition, prostitutes report higher mean levels of peer pressure in 1995 than non-prostitutes (15.98 versus 12.82, p<0.01). Moreover, prostitutes are less likely to have had full-time jobs in 1995 (47 percent versus 73 percent, p<0.05), more likely to have had other types of employment in 1995 (53 percent versus 27 percent, p<0.05), and more likely to be unemployed in 1995 (11 percent versus 0 percent, p<0.05) and 2003 (28 percent versus 8 percent, p<0.05) than non-prostitutes. These results are consistent with conventional assumptions about prostitutes, especially street-prostitutes being less likely to have fulltime jobs and more likely to be unemployed or have part-time employment.

Table 1 (p. 119) also shows that prostitutes have higher mean levels of drug problems

(7.45 compared to 0.81, p<0.001; 8.61 compared to 1.54, p<0.001) and drug use (4.81 compared to 3.70, p<0.05; 3.33 compared to 1.19, p<0.001; 3.22 compared to 1.05, p<0.001), as well as higher mean levels of running away from home in 1982 (2.44 versus 1.43, p<0.05) compared to non-prostitutes. This also supports previous literature that argues that drug use as well as running away are both important factors associated with prostitution. Moreover, prostitutes have lower mean levels of self-esteem (21.86 compared to 23.59, p<0.05; 21.08 compared to 23.46, p<0.05) and higher mean levels of depression (24.58 compared to 16.24, p<0.001; 24.69

65 compared to 12.89, p<0.001) than non-prostitutes. This also supports previous research that suggests that psychological distress, particularly self-esteem and depression, may be associated with prostitution.

Lastly, as a result of the race variables being significantly different for prostitutes (31 percent white and 69 percent black) and non-prostitutes (73 percent white and 27 percent black), the descriptive statistics are re-run in order to determine not only if there are significant differences between whites and blacks but also if there are differences within each race with respect to prostitutes and non-prostitutes. In conclusion, the results from Table 1 indicate that even though both groups (prostitutes and non-prostitutes) are highly disadvantaged and delinquent, prostitutes are worse off on several variables (sexual and physical abuse, friends paid for sex, employment status in 1995, unemployment in 2003, drug problems/use, running away, and psychological distress) than their non-prostitute counterparts; these findings suggest that these women become prostitutes because of these poor conditions and/or that prostitution has a variety of negative consequences on individuals’ lives.

Comparison of Prostitutes and Non-prostitutes Within and Between Races

When the descriptive statistics are evaluated for each race (white prostitutes versus white non-prostitutes as well as black prostitutes versus black non-prostitutes), the data in Table 2 (p.

120 indicate that there are significant differences between white prostitutes and white non- prostitutes on some of the factors associated with prostitution. For instance, these results show that white prostitutes report less parental supervision than white non-prostitutes (18 percent versus 63 percent respectively, p<0.01), thus supporting social control theory. Moreover, although white prostitutes have significantly higher mean levels of drug abuse (4.73 compared to

0.74, p<0.10; 6.09 compared to 1.00, p<0.10) and drug use (5.45 compared to 3.85, p<0.10; 3.64

66 compared to 1.04, p<0.001; 3.55 compared to 1.19, p<0.01) than white non-prostitutes. In addition, white prostitutes have significantly higher mean levels of depression (22.36 compared to 16.56, p<0.10; 26.09 compared to 12.93, p<0.05) than white non-prostitutes in both 1995 and

2003.

An examination of the differences between black prostitutes and black non-prostitutes

(Table 3, 121) shows that black prostitutes report higher mean levels of sexual abuse (3.20 compared to 1.10, p<0.01) and lower levels of parental caring and trust in 1982 (56 percent compared to 90 percent respectively, p<0.10) than the black non-prostitutes. Moreover, black prostitutes are significantly less likely to have had a full-time job in 1995 (48 percent versus 80 percent, p<0.10) and more likely to have had other types of employment in 1995 (52 percent compared to 20 percent, p<0.10) than black non-prostitutes. In addition, black prostitutes and black non-prostitutes are only significantly different from each other on two out of the five

“drug” variables, with prostitutes having higher mean levels of drug problems in 1995 (8.64 compared to 1.00, p<0.01) and drug use in 2003 (3.08 compared to 0.70, p<0.05). Lastly, black prostitutes, as compared to black non-prostitutes, have significantly lower mean levels of self- esteem (21.96 compared to 24.40, p<0.05; 21.00 compared to 25.20, p<0.01) and higher mean levels of depression (25.56 compared to 15.40, p<0.01; 24.08 compared to 12.80, p<0.05).

Black prostitutes also have higher mean scores of running away from home in 1982 (2.44 versus

0.90 respectively, p<0.10).

The final comparisons between white prostitutes and black prostitutes (see Table 4, p.

122) reveal that these two groups are significantly different on only four of the key variables.

White prostitutes report lower mean scores with respect to sexual abuse (1.73 compared to 3.20, p<0.10) and lower levels of conflict with parents in 1995 than black prostitutes (3.00 compared

67

to 4.64, p<0.01). In addition, 18 percent of white prostitutes report receiving above average

supervision levels compared to 60 percent of black prostitutes (p<0.05). Finally, white

prostitutes are less likely to report being paid for sex in 1995 than black prostitutes (0.27 versus

0.72 respectively, p<0.01).

Overall, there are significant differences between prostitutes and non-prostitutes, which

are important and impressive with such a small sample size. In addition, although both groups

(prostitutes and non-prostitutes) are both highly delinquent and disadvantaged, these results

indicate that prostitutes are more disadvantaged than non-prostitutes. Therefore, it appears that

there is something about prostitution that makes this group more disadvantaged than their non-

prostitute counterparts. Moreover, it is important to note that there are significant race

differences in this research, which would likely be even more distinct and widespread if a larger

sample size was used.

Logit Estimates of Becoming a Prostitute

Testing the Relationship Between the Strain Theory Measures and Prostitution

Previous research has documented that sexual abuse is often associated with prostitution.

Consequently, the following hypothesis –Higher levels of sexual abuse increase the odds of prostitution– is assessed in order to determine if this widely held assumption is true among a sample that is not characterized by the selection bias typical of these other studies.

First, zero order models are run (results not shown) in order to determine the main effect of each strain theory measure. These results indicate that sexual abuse is the only strain theory measure that is significantly associated with prostitution (p<0.01). Next, all of the strain theory measures (sexual abuse, contentious family environment, and conflict between parents) are included in a model (Table 5, Model 1, pgs. 123-124) with all of the control variables

68

(employment, age, parental social class, and race), but sexual abuse is the only strain theory measure that is significant; specifically for each unit increase in sexual abuse, the odds of prostitution increase by 36 percent, controlling for other variables (p<0.05). Consequently, the model is re-run with only sexual abuse and the control variables (Model 2). Sexual abuse remains positively and significantly related to prostitution, indicating that for each unit increase in sexual abuse, the odds of prostitution increase by 39 percent, controlling for other covariates

(p<0.05). Therefore, hypothesis one is supported. This hypothesis also finds support in the life- history narratives from the OLS qualitative data. For example, one respondent from the OLS reports the following experience:

…this janitor guy he was having sex with me for money to give me weed, liquor, whatever the case may be…I fell into that situation just because of earlier molestation issues and whatever…You know…it’s like everybody must know you’re this vulnerable, weak person…when they say that predators sit out and they stalk and they study they prey. They do. It’s like they can look at you. They can talk to you. They can look at your demeanor, how you carry yourself and tell if you’re easy prey. And somehow he singled me out. He became very friendly. Uh…you know it was do you need a ride home? Uh…do you want you know a couple of dollars for lunch? It starts out a very slow, gradual thing…They just get to know a little bit more and a little bit more and little bit more until they kind of figure you know I wonder if she’ll go for this then.

Next, Model 2 is re-estimated with running away added in order to determine if it mediates the relationship between sexual abuse and prostitution (see Model 3). Despite the significance of running away in this model (for each unit increase in running away, the odds of prostitution increase by 48 percent, controlling for other variables, p<0.05), sexual abuse remains significant and the coefficient does not change (Table 5, Model 3, p. 123-4); consequently, it cannot be concluded that running away from home mediates the relationship between sexual abuse and prostitution. This is interesting considering that it contradicts previous research, as

69

well as some data in the qualitative life-history narratives. As one prostitute in the OLS data says,

Run away child. Just run away. Because I wouldn’t stay at home because I was getting sexually abused. By my older brother. And I just wouldn’t stay at home. So, my father didn’t want to believe me that he was, my brother was sexually abusing me. So, I just left home. And I’ve been on my own ever since.

Another respondent from the OLS supports this by stating, “’I was raped by just about every man my mother ever had…well, hell, if I gotta stay home and get fucked by all her men, I might as well be out on the street and get fucked and get paid for it.’”

Next, models are run in order to determine if deviant peers (social learning theory) and/or

drug use mediates the relationship between running away from home and prostitution (See

Figure 1a, p. 116). As before, the only strain theory measure to be significantly associated with prostitution is sexual abuse (Model 6); consequently, the models are re-run with sexual abuse as the only strain theory measure (Model 7). According to the data in Table 5 (p. 123-4), neither deviant peers nor drug use mediate the relationship between running away and prostitution

(Models 4-7). Despite their lack of mediation, peer pressure in 1982 (for each unit increase in peer pressure in 1982, the odds of prostitution increase by 58 percent, controlling for other variables, p<0.10) and drug use in 2003 (for each unit increase in drug use in 2003, the odds of prostitution increase by 70 percent, controlling for other variables, p<0.10) are positively and significantly associated with prostitution (Model 7). In addition, Model 7 indicates that sexual abuse is also positively and significantly associated with prostitution (for each unit increase in sexual abuse the odds of prostitution increase by 118 percent, controlling for other covariates, p<0.05). It is important to note, however, that the significance of running away from home disappears but its magnitude increases when delinquent friends and drugs are both added to the

70

model (See Model 7). These results indicate a possible collinearity problem, but the variance

inflation factors are all less than 10, thus indicating that collinearity is not a problem.

In addition, when drug use and deviant peers are both added to the models, the magnitude

of sexual abuse increases (Models 6 and 7). Despite the positive and significant relationship

between drugs and prostitution in these findings, it is still unknown what comes first: drug use or

prostitution. For example, some women become prostitutes to buy drugs as reported by this

respondent in the OLS qualitative data “’Up until a year ago I was prostituting, I was…you name

it. You know, anything to…To get drugs…For the money. I do it for the money and then I’d go

buy drugs.’” Others, however, report they use drugs or other substances to handle the stress of

sleeping with strangers for money; “’I wanted to do it [prostitute] again…and again…and

again…because the money was so easy but I would have to have a drink…I couldn’t go out there

sober’” (OLS Data). Overall, both the quantitative and qualitative analyses indicate that drugs

are an important element in prostitution. Though the temporal order of the relationship is

uncertain; more likely than not, this relationship is reciprocal.

Models are then run with psychological distress (depression and self-esteem) and attachment as moderating variables (see Figure 1b, p. 116). However, when these models are run, none of the interaction terms are significant (results not shown). Consequently, psychological distress (depression and self-esteem) and attachment do not affect the strength of the relationship between abuse and prostitution. Because these variables are not significant moderators, the researcher speculated that they might act as mediators, even though these relationships are not predicted (See Figure 1b, p. 116). As a result, the models are re-estimated using them as mediators for exploratory purposes. Results (not shown) indicate that psychological distress (depression and self-esteem) and attachment do not mediate the

71

relationship between abuse and prostitution. Despite their lack of mediation, depression in 1995

and parental caring and trust in 1995 are positively and significantly associated with prostitution.

This is interesting because it suggests that having higher levels of parental caring and trust in

1995 increases the odds of prostitution, which is counterintuitive to social control theory.

However, this could be a result of the sample respondents generally lacking high parental attachment levels.

Continuing the exploratory analysis, psychological distress (depression and self-esteem) and attachment are all placed in a model (not shown) with deviant peers, running away, abuse, and drugs, but an error message is received stating that the model of fit is questionable and complete separation exists2. Consequently, the model is run with different combinations of the

variables in order to obtain a model that does not have a questionable fit, this model, Model 8,

contains sexual abuse, delinquent friends, parental intimate communication, drug use, running

away from home, self-esteem, and depression. This model contains only two significant

variables, drug use in 2003 (for each unit increase in drug use in 2003 the odds of prostitution

increase by 330 percent, controlling for other factors, p<0.10) and depression in 1995 (for each

unit increase in depression in 1995 the odds of prostitution increase by 65 percent, controlling for

other covariates, p<0.10). The Hosmer and Lemeshow Goodness of Fit Test indicates that

2 Model 8 is not empirically consistent (χ (6)=17.5470, p=0.0075), which may be due to the

omission of some important effects that are not measured in this research. As a result of Model 8

containing only two significant variables and the lack of it being an empirically consistent

model, a trimmed model is run. Model 9 indicates that sexual abuse, having a drug problem in

2 A complete separation occurs when one or more predictors completely separate the outcome groups or “discriminate between the categories of response” (DeMaris, 2004, p. 268).

72

1995, using drugs in 2003, running away from home, having another type of employment status

in 1995, and being black are all positively and significantly associated with prostitution.

According to the Hosmer and Lemeshow Goodness of Fit Test, this model (Model 9) is

2 empirically consistent (χ (8)=7.707, p=0.4626)

Overall, sexual abuse is found to be an important factor associated with prostitution;

therefore, Hypothesis 1 is supported. On the other hand, running away from home does not

mediate the relationship between sexual abuse and prostitution, which contradicts previous

research. In addition, drugs and having delinquent friends do not mediate the relationship between running away from home and prostitution despite their positive and significant effect on prostitution. Lastly, psychological distress (depression and self-esteem) and attachment do not

moderate the effect of sexual abuse on prostitution as indicated in Figure 1b (p. 116). In short,

abuse and running away from home are robust factors associated with prostitution.

Testing the Relationship Between the Control Theory Measures and Prostitution

It is argued that low levels of parental attachment and parental supervision (components

of control theory) during adolescence both lead to participation in delinquent behaviors. As a result, Hypothesis 2—Respondents with higher levels of parental attachment have lower odds of being prostitutes than those respondents with lower levels of parental attachment —and

Hypothesis 3—Respondents who receive high levels of parental supervision have lower odds of being prostitutes than those who receive lower levels of parental supervision—are tested in order to see if this is also true with prostitution.

First, zero-order models are run (results not shown) in order to determine the main effect of each control theory measure. These results show that none of the social control theory measures predict prostitution. Despite their lack of significance, these measures are then all

73

placed into a model along with the control variables (employment, age in 1982, parental social

class, and race). In addition, supervision is also placed into a model with only the control

variables and the same is done for the attachment measures. None of the attachment variables

are significant (Models 10 and 12); consequently, these results do not support the second hypothesis insofar as the other attachment variables (parental caring and trust and parental

intimate communication) are not significant. This could be a result of the sample having low

levels of parental attachment and/or the lack of variation among prostitutes and non-prostitutes

on these measures. However, Model 11 does indicate that for each unit increase in parental

supervision, the odds of prostitution decrease by 77 percent (p<0.05), thus supporting the third

hypothesis (respondents who receive higher levels of parental supervision have lower odds of

being a prostitute than those respondents who receive lower levels of parental supervision).

The results from Model 11 support the belief that supervision is important in inhibiting

delinquent activities, specifically prostitution, even among a highly delinquent sample which is

also supported in some of the OLS life-history narratives. “I’m not putting my mom down or

anything and I love my sisters that did raise me but there just wasn’t no supervision. I mean, like

I said, I had a tab at the bar when I was 15. I started going in bars, I started going in the bars

when I was 13 years old and drank” (OLS Data). On the other hand, these results contradict some of the OLS respondents’ reasons for becoming prostitutes. For example, some state that

they did not feel loved by their family and wanted to know somebody cared about them; implying that they turned to prostitution because their customers made them feel cared for and

gave them things they wanted or needed. As one respondent says,

I ran the streets by myself…my mother didn’t care about me. I needed to know that somebody cared…and that being loved or hugged or told that you’re

74

loved…no one gave me that…they didn’t say ‘I love you’ or hug me or let me know that they cared about me so I figured that they didn’t give a shit.

As another respondent in the OLS data explains,

Well hell, people at home are having sex with me and I’m not really getting anything from them. Well, now this…this guy is wanting to have sex with me and he treats me nice at least and he’s giving me things I want.

Another OLS respondent also supports this by stating, “’Wes was paying the mortgage, helping

me out…I would have sex with Wes for the mortgage…He didn’t say nothing today but usually

sometimes he’ll say something like, ‘You want to make some extra money?’” Therefore,

according to the qualitative but not the quantitative results, some OLS respondents felt they

became involved in antisocial activities because their parents did not care about them (i.e. lack of parental attachment).

[I was in trouble as a kid] ‘cause my mom let me do whatever I wanted. She just didn’t care. I mean she didn’t want to take the mental anguish and everything and be chasing us down and be yelling and screaming at us. Just go do it.

These relationships (attachment, supervision, and prostitution) are further analyzed by using components of social learning theory (delinquent peers) as a mediating variable. As shown in Figure 2a (p. 117), it is proposed that deviant peers mediate the relationship between parental supervision and prostitution as well as that between attachment and prostitution. Using Figure

2a as a guide (p. 117), models are run to determine if deviant peers mediate the relationship between attachment and prostitution. The model is first run with all of the attachment variables in order to see which of them is significant with the deviant peers’ variables in the model (Model

13). Because the attachment variables are not significant correlates, the model is re-run with only supervision, the social learning variables, and the control variables (Model 14). This model indicates that delinquent peers do not mediate the relationship between supervision and

75 prostitution. This model also suggests that for each unit increase in supervision the odds of prostitution decrease by about 78 percent (p<0.05), controlling for other covariates. Moreover,

Model 14 indicates that having friends who are prostitutes increase the odds of prostitution by

235 percent (p<0.10) and for each unit increase in peer pressure in 1982, the odds of prostitution increase by 17 percent (p<0.05), controlling for other covariates. In conclusion, deviant peers do not mediate the relationship between parental supervision and prostitution, but deviant peers do slightly mediate the relationship between parental intimate communication in 1995 and prostitution even though this attachment measure is not a significant correlate (Model 15).

Also of interest is the effect of sexual abuse on the relationship between attachment and prostitution, specifically if sexually abuse negatively affects the strength of the relationship between attachment and prostitution (See Figure 2b, p. 117). Therefore, Model 13 is re-run and includes abuse as a moderator to attachment. However, none of the interaction terms are significant (results not shown) for the attachment measures (parental caring and trust and parental intimate communication). The possibility that sexual abuse might mediate the effect of attachment on prostitution is then explored (although not hypothesized in Figures 2a and 2b, p.

117) with a model that includes supervision, attachment measures (parental caring and trust and parental intimate communication), delinquent friends, and sexual abuse (see Model 16). These results indicate that sexual abuse slightly mediates the relationship between parental caring and trust in 1982 and prostitution, but this attachment variable is not a significant correlate (see

Model 16). It is interesting to note that most of the attachment variables are positively (although not significantly) associated with prostitution (except for parental caring and trust in 1982) in some of the models (Models 10, 12, 13, and 15), which is counterintuitive to social control theory.

76

However, because the attachment variables are not significant correlates in Model 16, the model is re-run with these variables removed from the equation. The results from Model 17 indicate that sexual abuse does slightly mediate the relationship between supervision and prostitution though this relationship is not shown in Figure 2b (p. 117). This model also suggests that supervision is negatively and significantly associated with prostitution (for each unit increase in supervision, the odds of prostitution decrease by 73 percent, p<0.10), but peer pressure in 1982 (for each unit increase in peer pressure in 1982, the odds of prostitution increase by about 20 percent, p<0.05) and sexual abuse (for each unit increase in sexual abuse, the odds of prostitution increase by 40 percent, p<0.05) are both positively and significantly associated with prostitution, controlling for other covariates. Lastly, having friends who are prostitutes in

1995 becomes non-significant and decreases in magnitude. In addition, the Hosmer and

Lemeshow Goodness-of-Fit test indicates that this model (Model 17) is empirically consistent

2 (χ (8) =5.661, p=0.6851).

In conclusion, Hypothesis 2 is not supported since the attachment variables failed to be significant correlates of prostitution. Perhaps some respondents become prostitutes because they are closely attached to their families while others do not, causing the positive and negative effects of attachment to not be significant in the analyses. “’I’m not a toy and if you get any ass from me you gonna pay for it…that’s how I viewed sex all my life…that’s the way I was raised.

My momma told me you don’t lay down and get no wet ass for nothin’” (Respondent from the

OLS Data). On the other hand, these findings could be a result of the lack of high attachment levels and the lack of variation on the attachment variables among this highly delinquent sample.

Moreover, the relationship between supervision and prostitution is not mediated by having delinquent friends. In addition, supervision is consistently significant in all of the models, thus

77 supporting Hypothesis 3. Sexual abuse, on the other hand, does not moderate the relationship between supervision and prostitution but it does slightly mediate this relationship. Sexual abuse does not mediate or moderate the relationship between attachment and abuse, however. Blacks are also consistently found to have higher odds of prostitution (ranging from 439 percent to 995 percent) than whites, and those with other types of employment in 1995 (ranging from 436 percent to 671 percent) are more likely to be prostitutes than those that have had a fulltime position in 1995. Consequently, it can be argued that those without fulltime positions in 1995 turned to prostitution to survive. As one prostitute in the OLS explains, “Umm, it wasn’t never okay. It’s what I had to do to survive.”

Testing the Relationship Between the Social Learning Theory Measures and Prostitution

It is commonly argued that peers are an important component of adolescents’ involvement in antisocial behavior but this relationship is not frequently associated with participation in prostitution. As a result, the fourth hypothesis in this research proposes that those with delinquent friends have higher odds of prostitution than those who do not have delinquent friends.

Zero-order models are run (results not shown) and indicate that having friends who were paid for sex in 1995 and experiencing peer pressure in 1995 are positively and significantly related to prostitution. However, when all three social learning theory measures (friends paid to have sex in 1995, peer pressure in 1982, and peer pressure in 1995) are entered into a model

(Table 7, Model 18, p. 126) along with the control variables (employment, age, parental social class, and race), peer pressure in 1995 becomes non-significant, though peer pressure in 1982 becomes significant and having friends who were paid for sex remains significant (p<0.10).

78

Model 18 (Table 7, p. 126) supports Hypothesis 4 in that having delinquent friends

increase the odds of prostitution, particularly if one has friends who have been paid to have sex

in 1995 (the odds of prostitution increase by about 269 percent for those with friends involved in

prostitution compared to those who do not have friends involved in prostitution, controlling for other covariates, p<0.10). Moreover, Model 18 indicates that for each unit increase in peer

pressure in 1982, the odds of prostitution increase by approximately 14 percent, controlling for

other variables (p<0.10). These results support those found in the qualitative data that peers are

an important behavioral influence. “’They [older friends] would, sneak into a show…play with

the guys…play hooky from school. They was doing it…so I went right with them’” (Respondent from the OLS Data). Another respondent from the OLS data also supports this idea by saying,

“’I was just influenced in that if somebody had an idea of something to do and it sounded adventurous and exciting, it really didn’t matter to me if it was a bad thing to do or not, because I was out for the excitement.’” It is interesting to note that blacks, according to Model 18, have about 329 percent higher odds of prostitution than whites (p<0.05), controlling for other covariates. This robust statistic is further discussed in the Discussion/Conclusion Chapter found below.

Next, the relationship between deviant peers and prostitution is examined using drug use, a factor frequently associated with prostitution, as a mediator (See Figure 3, p. 118). Model 19 indicates that a potential collinearity problem may exist with regard to having friends who are prostitutes because the magnitude of this variable increases and it becomes non-significant; however, the variance inflation factors are all less than 10, thus indicating that collinearity is not

a problem. This model also suggests that drug use slightly suppresses the relationship between

peer pressure in 1982 and prostitution and drug use mediates the relationship between peer

79

pressure in 1995 (although not a significant variable) and prostitution. Perhaps those

respondents that receive higher levels of peer pressure are more likely to use drugs because of

that peer pressure, which can in-turn lead to prostitution. Therefore, peer pressure may be a

factor better associated with drug use rather than prostitution. In addition, Model 19 indicates

that peer pressure in 1982 (for each unit increase in peer pressure in 1982, the odds of

prostitution increase by 47 percent, controlling for other factors, p<0.05), drug problems in 1995

(for each unit increase in drug problems in 1995, the odds of prostitution increase by 28 percent

controlling for other covariates, p<0.05), and using drugs in 2003 (for each unit increase in drug

use in 2003, the odds of prostitution increase by 69 percent, controlling for other factors, p<0.05)

are all positively and significantly associated with prostitution. The importance of drugs and

“friends” supports one respondent’s reasoning for why she turned to prostitution. “’Some pimp

turned me on the crack. [Began soliciting because]…I was hanging out around with this older

girl…and that’s what she did…I wanted to be like her…so I followed her’” (Respondent in the

OLS Data).

As indicated in Figure 3 (p. 118), it is theorized that attachment and supervision affect the strength of the relationship between deviant peers and prostitution. When the models are re- estimated using supervision as a moderating variable, the interaction terms are not significant, indicating that supervision does not affect the strength of the relationship between delinquent peers and prostitution. When the attachment variables are used as moderating variables, only one of them is found to moderate the effect of having delinquent friends on prostitution: parental caring and trust in 1982 moderates the effect of peer pressure in 1982 (Model 20). Therefore, the level of parental caring and trust in 1982 conditions the effect of having delinquent friends on prostitution insofar as for those who are a standard deviation or 0.1892 higher in peer pressure in

80

1982, the effect of parental caring and trust in 1982 increases by 0.00000000002, which is very small.

As a result of supervision failing to be a moderator and most of the attachment measures failing to act as moderators, the possibility of these measures being mediators is examined for exploratory purposes (not shown in Figure 3, p. 118). The full model (Model 21), shows that peer pressure in 1982 (for each unit increase in peer pressure in 1982, the odds of prostitution increase by about 74 percent, p<0.05), having a drug problem in 1995 (for each unit increase in the 1995 drug problem scale, the odds of prostitution increase by about 60 percent, p<0.05), and using drugs in 2003 (for each unit increase in using drugs in 2003, the odds of prostitution increase by approximately 124 percent, p<0.05) are all positively and significantly associated with prostitution. Parental supervision (for each unit increase in parental supervision, the odds of prostitution decrease by 92 percent, p<0.10) and parental caring and trust in 1982 (for each unit increase in parental caring and trust in 1982, the odds of prostitution decrease by 86 percent, p<0.10), on the other hand, are negatively and significantly associated with prostitution. In addition, Model 21 indicates that supervision and attachment slightly mediate the relationship between having friends that were paid for sex and prostitution. The Hosmer and Lemeshow

2 Goodness-of-Fit test indicates that this model is empirically consistent (χ (8) =4.272, p=0.8318).

In conclusion, Hypothesis 4--having delinquent friends increases the odds of prostitution-

-is supported. Supervision and attachment slightly mediate the relationship between having friends that are paid for sex and prostitution. Although this is not the usual temporal order assumed in most theories, which argue that supervision leads to conventional peers, which in- turn leads to conformity, it is an interesting finding, which is discussed further in the next chapter. In addition, parental caring and trust in 1982 does moderate or condition the effect of

81

peer pressure in 1982 on prostitution but this effect is very small. Moreover, two of the drug variables (drug problem in 1995 and drug use in 2003) are consistently positively and significantly associated with prostitution. Lastly, blacks have significantly higher odds of prostitution in all of the social learning models than whites (with a minimum of 329 percent

higher odds of prostitution than whites).

82

CHAPTER 6: DISCUSSION AND CONCLUSION

Summary

Discovering and understanding the factors associated with prostitution have been of

significant interest among criminologists for many years. Despite this interest, previous research

has produced conflicting findings, resulting in confusion as to why some women enter

prostitution and others do not. For example, some researchers have cited sexual abuse as an

important factor associated with prostitution because sexual abuse produces negative effects on

the self, which increases the risk of self-destructive behaviors, including prostitution (Reiker &

Carmen, 1986) and teaches victims to emotionally distance themselves from abuse, a technique

also used by prostitutes while performing sex acts (Herman & Hirschman, 1977; James &

Meyerding, 1977; Miller, 1986; Westerlund, 1992). However, there is little consensus as to

whether sexual abuse has a direct or indirect effect (via running away or psychological distress,

for example) on prostitution (Simons & Whitbeck, 1991). For instance, some argue that if one is

sexually abused in the home, they are more likely to escape the abuse by running away and must

subsequently turn to prostitution in order to survive economically. Furthermore, others have argued that abuse is more likely to lead to prostitution if there is a lack of parental attachment

(Brannigan & Van Brunschot, 1997). On the other hand, some researchers have found that sexual abuse is not an important factor associated with prostitution (Cates & Markley, 1992;

Nadon et al, 1998). These conflicting results make it difficult to draw firm conclusions about the

relationship between sexual abuse and prostitution.

Other researchers have argued that some women become prostitutes not because of abuse

but because of a lack of parental attachment (Dalla, 2001). For instance, those that feel they are

not loved or supported by their family may feel that they can gain the attention and affection they

83

desperately desire from someone else, such as customers or pimps. The logic of these positions

suggests that strong parental bonds can decrease the likelihood of prostitution.

Although a lack of supervision has been found to increase the participation in criminal

behaviors, including substance use (Cernkovich & Giordano, 1987; Gove & Crutchfield, 1982;

Laub & Sampson, 1988; Loeber & Stouhamer-Loeber, 1986), the relationship between

prostitution and supervision has not been investigated (at least to the researcher’s knowledge).

Therefore, it is unknown whether there is a relationship between supervision and prostitution.

However, it seems reasonable to assume that the relationship between supervision and

prostitution is similar to the relationship of supervision and other delinquent activities.

An additional factor associated with prostitution is the influence of deviant family

members and peers, just as they are in other criminal behaviors. The strong association between these deviant associates and prostitution may be due to the fact that peers and family members often are part of the sex trade themselves, and newcomers learn this behavior through interaction with them. Similarly, family and friends may define sleeping with a stranger for money, food,

shelter, clothing, or drugs as a “normal” behavior (Tutty & Nixon, 2003), consequently giving

the future prostitute the impression that there is nothing wrong with this behavior. In short,

friends and family members can be an important influence in individuals turning to prostitution,

especially if family and friends are part of the sex trade themselves (James, 1980; Weisberg,

1985).

Drug use, particularly crack, has also been found to be an important factor associated

with prostitution. Although many researchers agree that there is a strong relationship between

drug use and prostitution, it is not known whether drug addiction develops before, after, or

simultaneously with prostitution. Most argue that drug addiction precedes prostitution (James,

84

1977; Potterat et al., 1998; Dalla, 2002) but those that do not have a drug habit before becoming

a prostitute often subsequently develop one as a way to deal with the strains of the life of a

prostitute (Erickson et al., 2000; Gossop et al., 1995). Although the data used in this research is

longitudinal, due to the small sample size, the time order of the drug-prostitution relationship was unable to be determined.

In addition, prostitutes have been found to have higher levels of psychological distress

(El-Bassel et al., 1997; Risser et al., 2006), and it has been argued that there is a relationship between psychological distress and prostitution. However, some argue that psychological distress coupled with a traumatic event, such as sexual abuse makes some women more vulnerable to prostitution because they develop negative self-images (Bess & Janus, 1976;

Brown, 1979; Bullough, 1965; Jackman et al., 1963; James; 1976; Maerov, 1965), thus increasing the risk of destructive behaviors and habitual sex. Conversely, others argue that psychological distress leads to drug use (in order to cope with low self-esteem and high levels of depression), which in turn leads to prostitution (El-Bassel et al., 1997; Risser et al., 2006).

Therefore, it is unknown if psychological distress has a direct or indirect effect on prostitution.

As it can be seen, although many researchers have examined prostitution, firm conclusions are difficult to reach. A major obstacle in the prior research is the difficulty of securing willing participants for prostitution studies. The situations in which the researcher is most likely to find research subjects are often the most dangerous, such as at 2:00 a.m. on the city streets or at truck stops late at night, where the researchers run the risk of being robbed or

attacked. On the other hand, if the prostitute has a pimp, the researcher may get into an

altercation with her pimp for bothering her while she is working, causing the pimp to lose

money. In addition, prostitutes may be reluctant to answer the researcher’s questions because

85 they may feel the researcher is an undercover cop. Because it is so difficult to obtain willing participants, many researchers have found respondents in drug habilitation centers or those in therapy for sexually transmitted diseases. Such samples are usually biased towards prostitutes in that there are very few non-prostitute respondents in such samples to serve as useful comparison groups.

Previous research is also incomplete because most studies have been qualitative in nature and those that are not are usually cross-sectional studies, making it difficult to determine whether the effects of certain variables diminish over time. Moreover, a lack of longitudinal data makes it impossible to understand the consequences of prostitution, something which has not been examined in previous research.

An additional limitation of previous research is the dearth of samples comprised of diverse racial groups; those studies that are based on racially diverse samples have produced mixed results. For example, some have found that prostitutes are more likely to be white than black (Finkelhor & Ormrod, 2004; Potterat et al., 1998) but if blacks experience abuse, they are more likely to be prostitutes than whites and Hispanics (Medrano et al., 2003). As a result, it is unknown if the factors associated with prostitution are similar or different for whites and blacks or if there are any racial differences in prostitution.

As a result of the incomplete data and conflicting results found in previous research, the goal of the current research was to move forward in filling the gaps in our knowledge of the lives of women involved in prostitution and to better understand the factors that are associated with their involvement in prostitution. In order to do this, three sociological theories guided the analysis: general strain theory, control theory, and social learning theory, all of which provide tenable explanations of prostitution. For example, strain theory argues that experiencing a

86

traumatic event increases the likelihood of participating in criminal or destructive behaviors. In

the case of prostitution, the experience of sexual abuse (a traumatic event) is likely to increase the odds of prostitution. Control theory argues that low levels of parental attachment and supervision lead to participation in criminal behaviors. In this research, it was hypothesized that

high levels of parental attachment decrease the odds of prostitution because respondents who are

highly attached to their parent(s), are more likely to care what the parents think about her and

will not want to ruin her own reputation or that of her family’s by becoming a prostitute. It was

also hypothesized that high levels of parental supervision decreased the odds of prostitution

because individuals who are supervised more have fewer opportunities to participate in

prostitution. The last theory used in this research was social learning theory, which argues that

individuals learn behaviors from family and friends via definitions, positive reinforcement, and

imitation. Thus, it was hypothesized that individuals learn prostitution as they learn other

behaviors, via their delinquent peers, especially if their peers are involved in prostitution.

Although the components of each theory were examined separately in order to understand

their relationship with prostitution, integrated models were also proposed. For example, the

relationship between sexual abuse (strain theory) and prostitution (see Figure 1a and 1b, p. 116)

was examined using components of control theory (attachment as moderating variables) and

social learning theory (deviant peers as mediating variables). In addition, the relationships

between the control theory measures (attachment and supervision) and prostitution (see Figures

2a and 2b, p. 117) were further examined using components of strain theory (sexual abuse as a

moderating variable) and social learning theory (deviant peers as mediating variables). Lastly,

the relationship between delinquent peers (social learning theory) and prostitution (see Figure 3,

87

p. 118) was examined using components of control theory (attachment and supervision as moderating variables).

The dataset that was used in this research is the Ohio Lifecourse Study (OLS), which is a multi-wave dataset of household and institutionalized respondents. For a variety of

methodological reasons, only the institutionalized respondents (incarcerated in several correctional facilities in Ohio in 1982) were used in this research. These respondents were

interviewed in 1982 (N = 254), 1995 (N = 209), and 2003 (N = 152) and had an average age of

16 in 1982, 29 in 1995, and 37 in 2003. Using face-to-face interviews, respondents were asked a

variety of questions at each time period in the subject areas of family, school, criminal behaviors,

and peers.

The OLS data was used in this research because it is not a prostitute-biased sample,

which allows for variation among respondents on key variables. Although the respondents used

in this research are from a high-risk delinquent sample, there is much variation among them. For

example, there are respondents who were abused and did not become prostitutes (N = 11 or 15

percent), respondents who were abused and did become prostitutes (N = 19 or 26 percent),

respondents who were not abused and became prostitutes (N = 17 or 23 percent), and

respondents who were not abused and did not become prostitutes (N = 26 or 26 percent). There

is also significant racial variation in the sample (38 white females and 35 black females).

Moreover, the OLS is useful because it is a longitudinal dataset, thus enabling an examination of

prostitutes at multiple points in time, thereby increasing the total number of prostitutes in this

research. Lastly, the OLS not only contains extensive quantitative data, but a qualitative

component as well, which is very useful supplementary data helping to explain some of the

88

findings in the quantitative data as well shedding light on issues not apparent in the quantitative

findings.

The dependent variable in this research is being paid to have sex and the independent

variables include exposure to negative stimuli (sexual abuse, contentious family environment, and conflict with parents), attachment to parents (supervision, caring and trust, and intimate

communication), influence of delinquent peers (friends involved in prostitution and peer

pressure), drug use and abuse, running away from home, and psychological distress (low levels

of self-esteem and high levels of depression). Race, employment status, parental social class,

and age served as control variables in all of the analyses.

The final analytic sample in this research consisted of 73 previously institutionalized

female respondents that were interviewed at all three time periods. Although male prostitution

does occur, this research only examined female prostitution because prostitution is seen

primarily as a female social problem and because of the small number of male prostitutes in the

OLS (N = 16). In addition, this research only used the institutionalized sample due to the lack of

prostitutes in the neighborhood sample (N = 6) and a Chow Test indicated that the two samples

2 were affected differently by the independent variables ( (5) = 17.335, p<0.05). Moreover, even

though there are different types of prostitutes, the respondents in this research are more likely to

be street-prostitutes than high-class call girls because the OLS is skewed toward the lower and

working classes.

Three hypotheses were supported in this research (Hypotheses 1, 3, and 4). The

empirical validation of Hypothesis 1-- Higher levels of sexual abuse increase the odds of prostitution --supported the long-held argument that sexual abuse is an important factor associated with prostitution. This finding also indicates that strain theory may be sufficient to

89 explain why some women become prostitutes and other women do not insofar as aversive stimuli, such as abuse, an important factor associated with antisocial behavior. However, it is still unknown if the effect of abuse diminishes over time. If one is abused as a child but does not become a prostitute until adulthood, perhaps the reason for becoming a prostitute is different for this individual compared to the individual that is abused as a child and becomes a prostitute as an adolescent. Thus, abuse may not have as strong of an effect on prostitution if a significant amount of time has passed between the abuse and becoming a prostitute.

Contrary to popular belief, running away from home did not mediate the relationship between sexual abuse and prostitution. Thus, sexual abuse had a direct effect on prostitution rather than an indirect effect, which contradicted previous research (Nadon et al., 1998; Simons and Whitbeck, 1991). This finding also contradicts strain and feminist accounts that view running away as a mediating variable in the relationship between sexual abuse and prostitution.

Perhaps sexual abuse had a direct effect on prostitution rather than an indirect effect through running away because those who were sexually abused learned to emotionally distance themselves from situations of abuse (which is also a tactic used by prostitutes) rather than running away in order to cope with the abuse; consequently, sexual abuse victims may engage in habitual sexual activities such as prostitution as a means to avoid dealing with the abuse

(Herman & Hirschman, 1977; Westerlund, 1992) rather than running away. On the other hand, others may feel that if they are getting sexually abused, they might as well go onto the streets and get paid for it as suggested in the OLS qualitative data discussed above (“I was raped by just about every man my mother ever had…well, hell, if I gotta stay home and get fucked by all her men, I might as well be out on the street and get fucked and get paid for it”).

90

Moreover, using drugs and having delinquent friends did not mediate the relationship

between running away from home and prostitution despite their positive and significant effect on

prostitution; therefore running away has a direct effect on prostitution, which supported previous

research (Seng, 1989). Perhaps these respondents ran away from home because they felt as if

their parents did not love them or felt rejected by their parents. As stated earlier, one prostitute

from the OLS says,

[I was in trouble as a kid] ‘cause my mom let me do whatever I wanted. She just didn’t care. I mean she didn’t want to take the mental anguish and everything and be chasing us down and be yelling and screaming at us. Just go do it.

As a result of having little money, direction, or employment opportunities while living on the

streets, they therefore may have found themselves homeless without food or shelter and needed

to do something to survive; thus, turning to prostitution to survive life on the streets. As

discussed above, “Umm, it wasn’t never okay. It’s what I had to do to survive” (Respondent in

the OLS Data).

In addition, neither psychological distress nor attachment were found to affect the

strength of the relationship between abuse and prostitution, either as mediating or moderating

variables. These results contradict previous research that suggest that psychological distress (e.g.

feelings of worthlessness or low self-esteem) coupled with experiencing a traumatic event make

some women more vulnerable to prostitution (Bess & Janus, 1976; Brown, 1979; Bullough,

1965; Davis & Petretic-Jackson, 2000; Finkelhor & Browne, 1985; Jackman, O’Toole, & Geis,

1963; James, 1976; James & Meyerding, 1978; Maerov, 1965) and those that argue if sexual

abuse is experienced as a child and parental bonds are not intact, the adolescent is more likely to

turn to prostitution (Brannigan & Van Brunschot, 1997). These non-significant results may be due to the nature of the sample because the majority of the respondents have low levels of

91 parental attachment and high levels of psychological distress, thus resulting in little variation among them on these variables. In addition, these results may have been significant if a larger sample had been available for this research. Overall, most of the predictions for the strain models were not supported with the exception of the impact of abuse (experiencing a traumatic event) on prostitution. Therefore, it cannot be concluded that strain theory alone is a completely adequate theory to explain why some women become prostitutes and others do not.

Hypothesis 3--Respondents who receive high levels of parental supervision have lower odds of being prostitutes than those who receive lower levels of parental supervision -was also supported in this research. Supervision was found to be consistently negatively and significantly related to prostitution. Consequently, it can be concluded that higher levels of parental supervision decrease the odds of prostitution. This result supports previous research that one of the most powerful predictors of delinquent behavior is a lack of parental supervision

(Cernkovich & Giordano, 1987; Laub & Sampson, 1988; Loeber & Stouthamer-Loeber, 1986).

Perhaps the non-prostitutes received not just more supervision but more effective supervision than the prostitutes. For example, maybe the parents of the non-prostitutes checked to make sure the respondents were not lying about where or whom they were with after school. In addition, perhaps the parents of the non-prostitutes punished them if they found out that the respondents were not telling the truth whereas the parents of the prostitutes did not check on them as often or at all or even punished them for lying or being defiant. The significant relationship between supervision and prostitution also supports control theory, thus suggesting that control theory may be a useful theory for explaining why some women become prostitutes and other women do not, at least with respect to supervision.

92

In addition, deviant peers were not found to mediate the relationship between attachment

and prostitution nor between supervision and prostitution, which contradicts theoretical arguments that suggest if one experiences less attachment and less supervision, this will increase

the likelihood that the individual will become friends with deviant peers, which, in-turn will lead

to participation in deviant behaviors. Perhaps some of the parents to whom the respondents are

attached are encouraging the respondents to spend time with deviant friends, either to get the

respondents to be more outgoing or “less square,” to make the parent money (particularly if the

parent is involved in criminal behaviors such as prostitution or drug dealing), or the parents may

not know that the peers their daughter are spending time with are deviant. On the other hand,

other parents may be encouraging their daughters to spend time with conventional peers. As a result of these different parenting styles, their effects may be canceling one another out resulting

in a non-significant effect of deviant peers. However, it is important to note that the majority of the respondents in the sample used in this research are not highly attached to their parents. As a result, the respondents may not be concerned with their parents’ feelings or the parents may not

place much importance on the group of friends with which their daughter spends time. On the

other hand, a larger sample may have produced results showing that deviant peers do mediate the relationship between attachment and prostitution and the relationship between supervision and

prostitution.

Although abuse did not affect the strength of the relationship between attachment and

prostitution as hypothesized, abuse was found to mediate the relationship between supervision

and prostitution. These results indicate that part of the impact of supervision on prostitution is

due to the influence of abuse. Therefore, as the level of parental supervision increases, the odds

of abuse decrease, thereby decreasing the odds of prostitution. The opposite is also true: as the

93 level of parental supervision decreases, the odds of abuse increases, thus increasing the odds of prostitution. Therefore, abuse is less likely to occur when the parent is in the home supervising the respondent and abuse is more likely to occur when the parent is not home. An explanation for this finding may be that the perpetrators of abuse are family members (an older sibling or a parent’s significant other) and knows when the respondent will be without parental supervision.

Results from the qualitative data suggest that these explanations may have some merit (“…I was getting sexually abused. By my older brother” and “I was raped by just about every man my mother ever had”). This result also indicates that an integrated theory using components of strain theory and social control theory may be a dependable theory to explain why some women become prostitutes and others do not.

This research also examined whether delinquent friends are an important factor associated with prostitution as they are for other delinquent activities. This hypothesis was supported and is consistent with previous research showing that peers are an important influence in participation of delinquent activities (Agnew & Brezina, 1997; Garnier & Stein, 2002; Haynie,

2002; Laub & Sampson, 2003), especially if one’s fellow peers are deviant themselves (Dalla,

2001; Enablers, 1978). In addition, this result gives merit to pimps’ tactic to use other prostitutes as recruiters (befriend and gain the trust of the potential prostitute) and to teach “rookies” the tricks of the trade.

Drug use was found to mediate the relationship between peer pressure in 1995 but suppress the relationship between peer pressure in 1982 and prostitution. This is a surprising finding, along with the fact that there was not a significant direct effect of peer pressure in 1995 on prostitution. But it is important to note that the effect of peer pressure in 1995 may have been significant had a larger sample been used. On the other hand, perhaps peer pressure is more

94

influential in a younger adolescent’s life (the respondents were, on average, 16 years old in 1982

and 29 years old in 1995) than in an adult’s life. Overall, these results suggest that peers

encourage drug use, which increases the odds of addiction, which in turn makes prostitution a

viable way to get money for the drug habit or to get drugs as direct payment for their sexual

services.

Supervision was not found to affect the strength of the relationship between deviant peers

and prostitution. Perhaps some respondents have delinquent peers and spend time with them

under parental supervision but their parents do not know that their friends are delinquent. For

example, the respondent and the delinquent peers do criminal behaviors when the parent is

unaware of what is going on (e.g. smoke marijuana in a locked bedroom when the parent is in the kitchen). On the other hand, perhaps other OLS parents made sure they utilized more effective ways to supervise their daughters and would not allow them to have delinquent friends.

Moreover, some OLS parents may have encouraged their daughters to spend time with delinquent peers to get their daughters involved in prostitution. As a result of these varieties of parental supervision and their divergent consequences, the net effect of parental supervision is that it is not a significant moderator of the relationship between delinquent peers and prostitution. An additional potential explanation could be that supervision would significantly affect the strength of the effect of deviant peers on prostitution if a larger sample size had been used.

Furthermore, supervision and attachment did mediate the relationship between friends who were paid to have sex and prostitution. These results could indicate that having friends who are prostitutes causes individuals to de-emphasize what their parents think about them or the rules set by their parents, consequently leading to prostitution because they want to be like their

95

friends. Another potential explanation for this result could be that the “friends” of the

respondents could be family members and the respondents want to be like their family members

and receive praise from them by participating in prostitution as they do, thus influencing them to

become prostitutes themselves.

Additionally, parental caring and trust in 1982 was found to affect the strength of the relationship between deviant peers and prostitution but the attachment measures were not found to mediate this same relationship. Although, the impact of deviant peers on prostitution is conditioned on the level of parental caring and trust in 1982 this effect is relatively small, so that much of the impact of deviant peers on prostitution is direct, further supporting the belief that having delinquent peers influences individuals to participate in criminal behaviors, even prostitution.

Moreover, drug problems in 1995 and drug use in 2003 were found to be consistently positively and significantly related to prostitution. These results support the long held belief that there is a strong positive connection between drug use and prostitution (Cates & Markley, 1992;

Erickson, Butters, McGillicuddy, & Hallgren, 2000; Lung, Lin, Lu, & Shu, 2004; Nadon et al.,

1998; Nuttbrock, Rosenblum, Magura, Villano, & Wallace, 2004; Pleak & Meyer-Bahlburg,

1990; Potterat, Rothenberg, Muth, Darrow, & Phillips-Plummer, 1998; Simons & Whitbeck,

1991). However, it is still unknown which occurs first in time: drug use or prostitution. Perhaps the mixed results found in previous research are due to individual differences. For example, perhaps some become involved with drugs before prostitution and others become involved with prostitution before drugs. Others, however, may turn to drugs and prostitution simultaneously.

Consequently, these differing entries may be a result of personal characteristics (e.g. race, socio- economic status, gender) or situations (e.g. loss of a loved one or stress) encountered rather than

96

a clear time-order pattern between drugs and prostitution. Compared to previous research, this

study is unique in that it examines drug use and drug problems, both of which were found to be

positively associated with prostitution, particularly higher levels of drug problems in 1995 and

drug usage in 2003.

Hypothesis 2-- Respondents with higher levels of parental attachment have lower odds of

being prostitutes than those respondents with lower levels of parental attachment --was not supported in this research. This result suggests that being strongly attached to a parent does not decrease the odds of prostitution. Therefore, the attachment aspect of control theory was not supported in this research, although control theory was found to be an adequate theory to explain prostitution with respect to supervision. It is interesting that most of the attachment variables

(parental caring and trust and parental intimate communication) were not significant factors in most of the models estimated in this research. An explanation for this finding could be a result of the nature of the sample; very few of the institutionalized respondents have positive family

backgrounds, therefore contributing to the non-significant results for the attachment variables.

Therefore, the prostitutes in this sample are not more uniquely disadvantaged in parental

attachment than their non-prostitute counterparts.

Despite the small number of respondents that constitute the analytic sample in this

research, the strong robust result of blacks having significantly higher odds of becoming

prostitutes than whites is impressive. For example, blacks were found to have at least 329

percent higher odds of prostitution than whites, controlling for other covariates. These results are surprising and impressive because the sample in this research only contains 73 respondents, of which 36 are prostitutes. These findings could be a result of the higher number of black prostitutes (25 respondents or 69 percent of prostitutes are black) compared to white prostitutes

97

(11 respondents or 31 percent of prostitutes are white) in this study, which contradicts Finkelhor and Ormrod (2004) who found that whites are more likely to be prostitutes.

On the other hand, these findings may be a result of blacks becoming prostitutes earlier than whites as argued by Kramer and Berg (2003) but this possibility cannot be examined in this study. An additional explanation for the robust result of blacks having higher odds of prostitution than whites could be because black prostitutes had significantly higher levels of sexual abuse then white prostitutes at the bivariate level and sexual abuse was consistently found to have a strong direct effect on prostitution in this sample. Thus, it can be concluded that among this highly delinquent and disadvantaged sample, black prostitutes are more disadvantaged than their white prostitute counterparts. This consistent result indicates that future studies need to include both blacks and whites in their sample in order to determine if the factors associated with prostitution are similar or different across racial groups, especially since it contradicts previous research that finds prostitutes are more likely to be white (Finkelhor &

Ormrod, 2004; Potterat et al., 1998). Differences were found in the bivariate results but due to the small size of the groups in this research, the models could not be estimated separately for whites and blacks.

Robust differences were also found among the 1995 employment variables; specifically, unemployed respondents and those who had part-time employment in 1995 were found to have at least 298 percent higher odds of prostitution than those who had fulltime jobs in 1995, controlling for other variables. These results indicate that many women may become involved in prostitution because there are no other employment opportunities for them, whether this is real or perceived. Therefore, they feel that prostitution is the easiest or quickest way for them to make money, whether it is to pay the rent, buy drugs, or food. On the other hand, these respondents

98 may not work at any other job because they may think they are making quick cash, they are providing a service to the community, or they may enjoy working whatever hours they please

(e.g. can work around their children’s schedules). These results also suggest that this sample is more likely to be comprised of street-prostitutes rather than the high-class call girls since most of the respondents in this research are working class and many of the prostitutes in this study were unemployed or had part-time jobs. These results indicate a need to include employment status in future research when examining prostitution in order to determine if job status is an important factor associated with prostitution in the general population as it was for this highly delinquent and disadvantaged population.

The data analyses conducted in this research were guided by strain theory, control theory, and social learning theory. Overall, social learning theory was found to be the most useful theory in explaining prostitution among the highly delinquent and disadvantaged sample of respondents. Although the predictions made by each of the three theories were evaluated separately, this did little to develop our knowledge of the prostitution process. As a result, an integrated theory, using components of strain, control, and social learning theory is proposed as a more useful tool because such a model more accurately captures the complex nature of prostitution. For example, low levels of supervision can increase the odds of abuse which in turn leads to prostitution. In addition, low levels of supervision can lead to having delinquent peers, which can influence one to use drugs and drug use has been found to be an important factor associated with prostitution. In addition, the findings indicated that measures of drug use and running away can complicate the reasons for prostitution and must also be incorporated into the integrated theory.

99

Overall, several findings from this research contribute to the currently available literature

on prostitution. First, this research found that sexual abuse had a direct effect on prostitution, not an indirect effect as suggested in previous research. Secondly, it was suggested in the analyses

examining important factors associated with prostitution at each wave that the effect of sexual

abuse may decrease over time, which indicated that adolescent females might become prostitutes

for different reasons than adult female prostitutes, especially if the time between the abuse and

prostitution is great. In addition, this is one of the few studies that examined the relationship

between supervision and prostitution. Supervision was not only found to have a direct effect on

prostitution but it also mediated the relationship between friends who were paid for sex and

prostitution. This research also found that blacks have higher odds of prostitution than whites,

which contradicts previous research. However, it is important to note that the sample used in this

research is highly delinquent and disadvantaged; thus, the contradicting results could be due to

the nature of the sample This result may be due to the significantly higher level of sexual abuse

found among black prostitutes in this sample. Lastly, this research found that social learning

theory by itself was the most suitable theory to explain prostitution when compared to strain and

control theory. However, an integrated theoretical approach, using components of strain,

control, and social learning theory as well as measures of drug use and abuse, and running away

was suggested as the most efficient and coherent manner to understand the complex nature of

prostitution.

Limitations and Future Research

This study has several limitations. First, the sample used in this research is highly

delinquent and disadvantaged; therefore, it is not a general population sample. In addition,

because the neighborhood respondents were not asked the question, “how often have you been

100

paid to have sex in the past 12 months,” (the dependent variable in this research) during the second wave data collection, the neighborhood sample could not be used in this study. As a result, it is unknown if the factors associated with prostitution are similar for the institutional and

the neighborhood respondents (general population sample). Consequently, the results that were

found to be associated with prostitution in this research cannot be applied to the general

population.

Although the dataset utilized in this research provides information over a 22-year period,

the three waves of data were not collected annually. Rather, the first follow-up survey was

collected 14 years after the initial interview and the second follow-up survey was collected eight

years after the second wave. As a result of the large time spans between each follow-up survey,

little information is known about the respondents in these time periods, which limits the capacity

to draw conclusions about prostitution over the lifecourse. Even though the OLS is a very rich

dataset, future research should include more frequent follow-up surveys to fully explore the

extent and nature of prostitution over the lifecourse.

The small sample size utilized in this research is also a significant limitation. If a larger

number of respondents, especially those who were paid to have sex, would have been available,

this research may have produced different results. In addition, the small sample size that was

used in this research did not allow for the examination of race differences outside of the bivariate

level. As a result, race was unable to be used as an independent variable and was forced to be

used as a control variable. Consequently, it is unknown if the reasons for becoming a prostitute

are similar or different for whites and blacks. Therefore, it is important for future research to

obtain larger samples of prostitutes as well as more diverse populations. It is also important that

these future samples not be prostitute-biased samples so that comparisons to quasi-control groups

101 can be made. However, the researcher acknowledges that it is very difficult to obtain willing participants in research studies on prostitution.

An additional shortcoming in this research is that given the nature of the dependent variable (being paid to have sex in any of the three time periods) it is difficult to know, for example, if the effect of having deviant peers in 1982 or 1995 and drug problem in 1995 or 2003 affect prostitution in 1982, 1995, or 2003 or at all three time periods. However, models were run

(results not shown because of the small sub-sample sizes) with each factor and the control variables for 1982 prostitution, 1995 prostitution, and 2003 prostitution separately. Results indicate that sexual abuse and using drugs in all three waves were positively and significantly associated with prostitution in 1982 (p<0.10). Supervision and parental caring and trust in 1982, on the other hand, were negatively and significantly associated only with prostitution in 1982

(p<0.10). For prostitution in 1995, sexual abuse, having friends who were paid for sex in 1995, peer pressure in 1995, higher levels of drug problems in 1995, using drugs in 1995 and 2003, and depression in 1995 and 2003 were all positively and significantly associated with prostitution in

1995 (p<0.10). Conversely, self-esteem in 1995 and 2003 were negatively and significantly associated with prostitution in 1995 (p<0.10). Lastly, drug problems in 1995 and 2003, depression in 1995 and 2003, using drugs in 1982 and 2003, and parental intimate communication in 1995 were all positively and significantly associated with prostitution in 2003

(p<0.10). These results indicate that drug use was an important factor of prostitution and that the strong effect of sexual abuse on prostitution may diminish over time. These results suggest that theories may need to be revisited in order to take into account that the effects of key variables can increase or decrease over time. These results also indicate that integrated theories may be more useful in explaining a complex issue as prostitution.

102

A final limitation of this and previous research is that the long-term consequences of

being a prostitute are unknown. Although the OLS is a longitudinal dataset, the researcher was

not able to take advantage of this unique characteristic in this research due to the small sample

size utilized in this study. Another reason for the lack of focus on the consequences of being a

prostitute is because most research done on prostitutes is qualitative and focuses on why women

become prostitutes. Future research needs to examine the consequences of being a prostitute in

order to better understand these women’s lives. Some potential consequences of being a

prostitute, particularly a street prostitute, include being in a domestic violence relationship, living

in poverty, and receiving public assistance. Some of these consequences were examined at the bivariate level in this research (See Tables 8-11). For example, prostitutes reported higher mean

levels of perpetrating domestic violence and being victims of domestic violence than non-

prostitutes (Table 8, p. 127). A significantly higher percentage of prostitutes reported receiving

public assistance in 2003 (97 percent versus 65 percent, p<0.001), being currently divorced in

2003 (36 percent versus 14 percent, p<0.05), never being married in 1995 (56 percent versus 32

percent, p<0.05), and living in poverty in 2003 (64 percent verses 24 percent, p<0.001) than non- prostitutes. Moreover, when white prostitutes and black prostitutes were compared, white prostitutes were found to report lower mean scores perpetrating domestic violence (5.56 versus

6.88, p<0.05) and living in worse neighborhoods (12.83 versus 13.89 respectively, p<0.05). In addition, 50 percent of white prostitutes were married more than one time compared to 36 percent of black prostitutes (p<0.01). As it can be seen, prostitutes are significantly less well off than their non-prostitute counterparts even though all respondents are drawn from an economically disadvantaged sample.

103

Conclusion

Factors associated with prostitution is an important topic in criminology, but there is little consensus among criminologists. There are still gaps in the literature and the mixed results that are evident need to be reconciled. Although sexual abuse and having delinquent friends were found to be important factors associated with prostitution in the current research, it is still unclear what the temporal ordering of these factors are with respect to prostitution. Therefore, it is crucial that longitudinal data be available in order to fully understand this issue, especially to differentiate if most women become prostitutes before, after, or simultaneously with drug use. In addition, longitudinal data would be able to provide insight into if the effect of sexual abuse diminishes over time, which could potentially lead to understanding differences between why adolescent females become prostitutes compared to adult women. This information could also give better insight into how to prevent women from becoming prostitutes, either by providing better support networks (either via friends, family, or non-profit agencies) to deal with abuse or the addiction to drugs better than currently are available.

Moreover, although employment status was found to be significantly associated with prostitution in this study, it is still unknown if women become prostitutes because they perceive that there no other reliable job options available or if they stay unemployed in order to prostitute full-time because they can make easy money or they enjoy the hours of working the streets.

Future research should obtain more details about the employment history of respondents. In addition, longitudinal data could show if women lose their job during certain time periods whether they only engage in prostitution in order to make ends meet during that difficult time or if they continue to prostitute while they have another job (may prostitute to get drugs rather than money) or if some other patterns are occurring. Lastly, racial differences need to be further

104 examined in order to determine if it there are cultural differences for in prostitution rates (e.g. blacks are more likely to come from single-parent households which can decrease the amount of supervision received) or if it is more because of socio-economic status that blacks were more likely to be prostitutes as they were in this research. Longitudinal research can also examine if it is because certain racial groups experience key factors associated with prostitution earlier than others (e.g. sexual abuse), thus facilitating earlier entry into prostitution.

Despite all of these interesting findings this research using a quantitative dataset that is not prostitute biased to examine prostitution, applying these results to a general population is cautioned. This sample is small and consists of highly delinquent and economically disadvantaged respondents that are not highly attached to their parents. However, the variation among these respondents along a number of dimensions is substantial and indicates that prostitutes are uniquely disadvantaged even in an already disadvantaged sample. Thus, the researcher believes these results are an important contribution to the existing literature.

105

Works Cited

Abramovich, E. (2005). Childhood sexual abuse as a risk facto for subsequent involvement in

sex work: A review of empirical findings. Journal of Psychology & Human Sexuality,

17, 131-146.

Agnew, R., Brezina, T., Wright, J.P., & Cullen, F.T. (2002). Strain, personality traits, and

delinquency: Extending general strain theory. Criminology, 40, 43-72.

Agnew, R. (2001). Building on the foundation of general strain theory: Specifying the types of

strain most likely to lead to crime and delinquency. Journal of Research in Crime and

Delinquency, 38, 319-362.

Agnew, R. (1997). Stability and change in crime over the lifecourse: A strain theory

explanation. In Terence P. Thornberry (Ed.), Developmental theories of crime and

delinquency, advances in criminological theory (Vol. 7., pp. 101-132). New Brunswick,

NJ: Transaction Publishers.

Agnew, R. (1993). Why do they do it? An examination of the intervening mechanisms between

‘social control’ variables and delinquency. Journal of Research in Crime and

Delinquency, 30, 245-366.

Agnew, R. (1992). Foundation for a general strain theory of crime and delinquency.

Criminology, 30, 47-88.

Agnew, R., & Brezina, T. (1997). Relational problems with peers, gender, and delinquency.

Youth and Society, 29, 84-111.

Akers, R.L., & Sellers, C.S. (2004). Criminological theories: Introduction, evaluation, and

application. Los Angeles, CA: Roxbury Publishing Company.

106

Akers, R.L. (1985). Deviant behavior: A social learning approach. Belmont, CA: Wadsworth.

Bagley, C., & Young, L. (1987). Juvenile prostitution and child sexual abuse: A controlled

study. Canadian Journal of Community Mental Health, 6, 5-26.

Bess, B.E., & Janus, S.S. (1976). Prostitution. In B.J. Sadock, H.I. Kaplan, & A.M. Freedman

(Eds.), The sexual experience (pp. 564-610). Baltimore: Williams & Wilkens.

Brannigan, A., & Van Brunschot, E.G. (1997). Youthful prostitution and child sexual trauma.

International Journal of Law and Psychiatry, 20, 337-354.

Briere, J., & Runtz, M. (1990). Differential adult symptomatology associated with three types of

child abuse histories. Child Abuse and Neglect, 14, 357-364.

Broidy, L. (2001). A test of general strain theory. Criminology, 39, 9-36.

Broidy, L., & Agnew, R. (1997). Gender and crime: A general strain theory perspective.

Journal of Research in Crime and Delinquency, 34, 275-306.

Bronfenbrenner, U. (1989). Ecological systems theory. In R. Vista (Ed.), Six theories of child

development: Revised formulations and current issues (pp. 187-249). Philadelphia:

Jessica Kingsley Publishers.

Brown, B. (1990). Peer groups. In S. Feldman & G. Elliott (Eds.), At the threshold: The

developing adolescent (pp 171-196). Cambridge, MA: Harvard University Press.

Brown, L.K., Lourie, K.J., Zlotnick, C., & Cohn, J. (2000). Impact of sexual abuse on the HIV-

risk-related behavior of adolescents in intensive psychiatric treatment. American Journal

of Psychiatry, 157, 1413-1415.

Brown, M.E. (1979). Teenage prostitution. Adolescence, 14, 665-680.

Browning, C.R., & Laumann, E.O. (1997). Sexual contact between children and adults: A life

course perspective. American Sociological Review, 62, 540-560.

107

Bullough, V.L. (1970). Why do married men visit prostitutes? Medical Aspects of Human

Sexuality, 4, 93-94.

Bullough, V.L. (1965). Problems and methods for research in prostitution and the behavioral

sciences. Journal of History of Behavioral Sciences, 1, 244-251.

Bullough, B., & Bullough, V.L. (1996). Female prostitution: Current research and changing

interpretations. Annual Review of Sex Research, 7, 158-181.

Campbell, R. (1998). Invisible men: Making visible male clients of female prostitutes in

Merseyside. In J.E. Elias, V.L. Bullough, V. Elias, & G. Brewer (Eds.) Prostitution: On

whores, hustlers, and johns (pgs. 155-171). Amherst, NY: Prometheus Books.

Campagna, D.S. & Poffenberger, D.L. (1988). The sexual trafficking in children: An

investigation of the child sex trade. Dover, MA: Auburn House Publishing Company.

Cates, J.A., & Markley, J. (1992). Demographic, clinical, and personality variables associated

with male prostitution by choice. Adolescence, 27, 695-707.

Caukins, S.E., & Coombs, N.R. (1976). The psychodynamics of male prostitution. American

Journal of Psychotherapy, 30, 441-451.

Cernkovich, S.A., & Giordano, P.C. (1987). Family relationships and delinquency.

Criminology, 25, 295-321.

Chesney-Lind, M. (1997). The female offender: Girls, women, and crime. Thousand Oaks, CA:

Sage.

Cole, S.G. (1987). Sexual politics: Contradictions and explosions. In L. Bell (Ed.), Good

girls/bad girls: Sex trade workers and feminists face to face (pp. 33-36). Toronto,

Canada: Women’s Press.

108

Clinard, M.B., & Meier, R.F. (2008). Sociology of Deviant Behavior (13th Edition). Belmont,

CA: Thomson Higher Education.

Dalla, R. (2006). “You can’t hustle all your life”: An exploratory investigation of the exit

process among street-level prostituted women. Psychology of Women Quarterly, 30,

276-290.

Dalla, R. (2002). Night moves: A qualitative investigation of street-level sex work. Psychology

of Women Quarterly, 26, 63-73.

Dalla, R. (2001). “Et Tu Brute? A Qualitative Analysis of Streetwalking Prostitutes’

Interpersonal Support Networks.” Journal of Family Issues, 22, 1066-1085.

Davis, J.L., & Petretic-Jackson, P.A. (2000). The impact of child sexual abuse on adult

interpersonal functioning: A review and synthesis of the empirical literature. Aggression

and Violent Behavior, 5, 291-328.

Davis, N. (1993). Prostitution: An international handbook on trends, problems, and policies.

London: Greenwood Press.

DeMaris, A. (2004). Regression with social data: Modeling continuous and limited response

variables. Hoboken, NJ: John Wiley and Sons, Inc.

Dilorio, C., Hartwell, T, & Hansen, N. (2002). Childhood sexual abuse and risk behaviors

among men at risk for HIV infection. American Journal of Public Health, 92, 214-219.

Earls, C.M. (1990). “Early Family and Sexual Experiences of Male and Female Prostitutes.”

Canada’s Mental Health, 38, 7-11.

El-Bassel, N., Schilling, R.F., Irwin, K.L., Faruque, S., Gilbert, L., Von Bargen, J., Serrano, Y.,

& Edlin, B.R. (1997). Sex trading and psychological distress among women recruited

from the streets of Harlem. American Journal of Public Health, 87, 66-70.

109

Enablers, The. (1978). Juvenile prostitution in Minnesota: The report of a research project. St.

Paul: The Enablers.

Ennew, J. (1986). The sexual exploitation of children. New York: St. Martin’s Press.

Erickson, P.G., Butters, J., McGillicuddy, P., & Hallgren, A. (2000). Crack and prostitution:

Gender, myths, and experiences. Journal of Drug Issues, 30, 767-788.

Finkelhor, D., & Browne, A. (1985). The traumatic impact of child sexual abuse: A

conceptualization. American Journal of Orthopsychiatry, 55, 530-541.

Finkelhor, D. & Ormrod, R. (2004). Prostitution of juveniles: Patterns from NIBRS. Juvenile

Justice Bulletin, June, 1-12.

Flexner, A. (1914). . New York: Century.

Flowers, B.R. (1998). The prostitution of women and girls. Jefferson, NC: McFarland &

Company.

Giordano, P., & Cernkovich, C. (2003). The Ohio Lifecourse Study (OLS), Wave I (1982);

Wave II (1995); Wave III, (2003) [machine readable data file and documentation].

Bowling Green, OH: Center For Family and Demographic Research, Bowling Green

State University.

Goldstein, P.J. (1979). Prostitution and drugs. Lexington, MA: Lexington Books.

Gossop, M., Powis, B., Griffiths, P., & Strang, J. (1995). Female prostitutes in South London:

Use of heroin, cocaine, and alcohol and their relation to health risk behaviors. AIDS

Care, 7, 253-260.

Gove, W.R., & Crutchfield, R.D. (1982). The family and juvenile delinquency. Sociological

Quarterly, 23, 301-319.

110

Graham, N., & Wish, E.D. (1994). Drug use among female arrestees: Onset, patterns, and

relationship to prostitution. Journal of Drug Issues, 24, 315-329.

Herman, J. & Hirschman, L. (1977). Father-daughter incest. Signs: Journal of Women in Culture

and Society, 2, 735-756.

Hirschi, T. (1969). Causes of delinquency. Berkeley: University of California Press.

Holzman, H.R., & Pines, S. (1982). Buying sex: The phenomenology of being a john. Deviant

Behavior, 4, 89-116.

Jackman, N.R., O’Toole, R., & Geis, G. (1963). The self-image of the prostitute. Sociological

Quarterly, 4, 150-161.

James, J. (1980). Entrance into juvenile prostitution. Washington, D.C.: National Institute of

Mental Health.

James, J. (1977). Ethnography and social problems. In R.S. Weppner (Ed.) Street ethnography:

Selected studies of crime and drug use in natural settings (Vol. 1, pp. 179-200). Beverly

Hills, CA: Sage Publications.

James, J. (1976). Motivations for entrance into prostitution. In L. Crites (Ed.) The female

offender (pp. 177-205). Lexington, MA: Lexington.

James, J., & Meyerding, J. (1978). Early sexual experience as a factor in prostitution. Archives

of Sexual Behavior, 7, 31-42.

James, J., & Meyerding, J. (1977). Early sexual experience and prostitution. American Journal

of Psychiatry, 134, 1381-1385.

Jordan, J. (1997). User buys: Why men buy sex. Australian and New Zealand Journal of

Criminology, 30, 55-71.

111

Kramer, L.A., & Berg, E.C. (2003). A survival analysis of timing of entry into prostitution: The

differential impact of race, educational level, and childhood/adolescent risk factors.

Sociological Inquiry, 73, 511-528.

Kunitz, S., Levy, J., McCloskey, J., & Gabriel, R. (1998). Alcohol dependence and domestic

violence as sequelae of abuse and conduct disorder in childhood. Child Abuse and

Neglect, 22, 1079-91.

Laub, J. & Sampson, R. 2003. Shared beginnings, divergent lives: Delinquent boys to age 70.

Cambridge, MA: Harvard University Press.

Laub, J.H. & Sampson, R. J. (1988). Unraveling families and delinquency: A reanalysis of the

Gluecks’ data. Criminology, 26, 355-380.

Loeber, R., & Stouthamer-Loeber, M. (1986). Family factors as correlates and predictors of

juvenile conduct problems and delinquency. Crime and Justice, 7, 29-149. Chicago:

University of Chicago Press

Lung, F., Lin, T., Lu, Y., & Shu, B. (2004). Personal characteristics of adolescent prostitutes and

rearing attitudes of their parents: A structural equation model. Psychiatry Research, 125,

285-291.

Maerov, A.S. (1965). Prostitution: A survey and review of 20 cases. Psychiatric Quarterly, 39,

675-701.

Maxwell, S.R., & Maxwell, R.D. (2000). Examining the “criminal careers” of prostitutes within

the nexus of drug use, drug selling, and other illicit activities. Criminology, 38, 787-809.

Mazerolle, P., Burton, V.S., Cullen, F.T., Evans, T.D., & Payne, G.L. (2000). Strain, anger, and

delinquent adaptations: Specifying general strain theory. Journal of Criminal Justice, 28,

89-101.

112

McKeganey, N., & M. Barnard. (1996). Sex work on the streets: Prostitutes and their clients.

Philadelphia: Open University Press.

Medrano, M.A., Hatch, J.P., Zule, W.A., & Desmond, D.P. (2003). Childhood trauma and adult

prostitution behavior in a multiethnic heterosexual drug-using population. The American

Journal of Drug and alcohol Abuse, 29, 463-486.

Miller, B.C., Monson, B.H., & Norton, M.C. (1995). The effects of forced sexual intercourse on

White female adolescents. Child Abuse and Neglect, 19, 1289-1301.

Miller, E.M. (1986). Street women. Philadelphia: Temple University Press.

Miller, J. (1995). Gender and power on the streets: in the era of crack cocaine.

Journal of Contemporary Ethnography, 23, 427-452.

Millett, K. (1971). Prostitution: A quartet for female voices. In V. Gornick and B.K. Moran

(Eds.), Women in a sexist society. New York: New American Library.

Monto, M. (2001). Prostitution and fellatio. Journal of Sex Research, 38, 140-145.

Nadon, S.M., C. Koverola, E.H. Schludermann. (1998). Antecedents to prostitution: Childhood

victimization. Journal of Interpersonal Violence, 13, 206-221.

Noll, J.G., Trickett, P.K., & Putnam, F.W. (2003). A prospective investigation of the impact of

childhood sexual abuse on the development of sexuality. Journal of Consulting and

Clinical Psychology, 71, 575-586.

Nuttbrock, L.A., Rosenblum, A., Magura, S., Villano, C., Wallace, J. (2004). Linking female

sex workers with substance abuse treatment. Journal of Substance Abuse Treatment, 27,

233-239.

113

Office of Juvenile Justice and Delinquency Prevention [OJJDP] (2007). Easy access to the FBI’s

supplementary homicide reports: 1980-2005. Retrieved April 17, 2008, from

http://ojjdp.ncjrs.gov/ojstatbb/ezashr/.

Overall, C. (1992). What’s wrong with prostitution? Evaluating sex work. Signs: Journal of

Women in Culture and Society, 17, 705-724.

Pleak, R.R., & Meyer-Bahlburg, H.F. (1990). Sexual behavior and AIDS knowledge of young

male prostitutes in Manhattan. The Journal of Sex Research, 27, 557-587.

Potterat, J.J., Rothenberg, R.B., Muth, S.Q., Darrow, W.W., & Phillips-Plummer, L. (1998).

Pathways to prostitution: The chronology of sexual and drug abuse milestones. The

Journal of Sex Research, 35, 333-340.

Potterat, J.J., Woodhouse, D.E., Muth, J.B., & Muth, S.Q. (1990). Estimating the prevalence and

career longevity of prostitute women. The Journal of Sex Research, 27, 233-243.

Potterat, J.J., L. Phillips, R.B. Rothenberg, W.W., Darrow. (1985). On becoming a Prostitute:

An explanatory case-comparison study.” Journal of Sex Research, 20-21, 329-326.

Prasad, M. (1999). The morality of market exchange: Love, money, and contractual

justice. Sociological Perspectives, 42, 181-214.

Reiker, P.P., & Carmen, E. (1986). The victim to patient process: The disconfirmation and

transformation of abuse. American Journal of Orthopsychiatry, 56, 360-370.

Risser, J.M., Timpson, S.C., McCurdy, S.A., Ross, M.W., & Williams, M.L. (2006).

Psychological correlates of trading sex for money among African American crack

cocaine smokers. The American Journal of Drug and Alcohol Abuse, 32, 645-653.

Ruggiero, K.J., McLeer, S.V., & Dixon, J.F. (2000). Sexual abuse characteristics associated with

survivor psychopathology. Child Abuse and Neglect, 24, 951-964.

114

Schauer, E.J. & Wheaton, E.M. (2006). Sex trafficking into the United States: A literature

review. Criminal Justice Review, 31, 146-169.

Seng, M.J. (1989). Child sexual abuse and adolescent prostitution: A comparative analysis.

Adolescence, 24, 665-675.

Silbert, M., & Pines, A. (1983). Early sexual exploitation as an influence in prostitution. Social

Work, 28, 285-289.

Simons, R.L., & Whitbeck, L.B. (1991). Sexual abuse as a precursor to prostitution and

victimization among adolescent and adult homeless women. Journal of Family Issues,

12, 361-379.

Steinberg, L., & Silverberg, S. (1986). The vicissitudes of autonomy in early adolescence. Child

Development, 57, 841-851.

Sullivan, E., & Simon, W. (1998). The client: A social, psychological and behavioral look at the

unseen patron of prostitution. In Elias, J.E., Bullough, V.L., Elias, V. & Brewer, G.

(Eds.), Prostitution: On whores, hustlers, and johns (pp. 134-154). Amherst, NY:

Prometheus.

Surratt, H.L., Inciardi, J.A., Kurtz, S.P., & Kiley, M.C. (2004). Sex work and drug use in a

subculture of violence. Crime and Delinquency, 50, 43-59.

Thoits, P. (1983). Social support and psychological well-being: Theoretical possibilities. In J.G.

Sarason & B.R. Sarason (Eds.), Social support: Theory research and applications (pp.

51-72). Boston: Dordrecht.

Tutty, L., & Nixon, K. (2003). “Selling sex? It’s really like selling your soul”: Vulnerability to

and the experience of exploitation through . In K. Gorkoff & J. Runner

115

(Eds.), Being heard: The experiences of young women in prostitution (pp. 29-45).

Canada: Fernwood Publishing.

Weiner, A. (1996). Understanding the social needs of streetwalking prostitutes. Social Work,

41, 97-105.

Weisberg, D. Kelly. (1985). Children of the night: A study of adolescent prostitution.

Lexington: Lexington Books.

Westerlund, E. (1992). Women’s sexuality after childhood incest. New York: W.W. Norton and

Company.

Widom, C.S. & Kuhns, J.B. (1996). Childhood victimization and subsequent risk for

promiscuity, prostitution, and teenage pregnancy: A prospective study. American

Journal of Public Health, 86, 1607-1612.

Xantidis, L., & McCabe, M. (2000). Personality characteristics of male clients of female

commercial sex workers in Australia. Archives of Sexual Behavior, 29, 165-176.

Young, A.M., Boyd, C., & Hubbell, A. (2000). Prostitution, drug use, and coping with

psychological distress. Journal of Drug Issues, 30, 789-800.

116

APPENDIX A: FIGURES: DIRECTIONS OF ASSOCIATION

Figure 1a: Relationship between experiencing strain and becoming a prostitute

DEVIANT PEERS

(+) (+)

(+) (+) SEXUAL ABUSE RUNNING AWAY PROSTITUTION

(+) (+)

DRUGS

Figure 1b: Relationship between experiencing strain and becoming a prostitute

DEVIANT PEERS

(+) (+)

(+) (+) SEXUAL ABUSE RUNNING AWAY PROSTITUTION

(+) (+)

DRUGS

ATTACHMENT (-) PSYCHOLOGICAL DISTRESS (+)

117

Figure 2a: Relationship between social control variables and becoming a prostitute

(-) PARENTAL PROSTITUTION SUPERVISION (-) (+)

DEVIANT PEERS

(-) (-) ATTACHMENT —————————————————————————

Figure 2b: Relationship between social control variables and becoming a prostitute

(-) PARENTAL PROSTITUTION SUPERVISION (-) (+)

DEVIANT PEERS

(-) (-) ATTACHMENT —————————————————————————

SEXUAL ABUSE (-)

118

Figure 3: Relationship between having friends who are prostitutes and becoming a prostitute

DRUG USE

(+) (+)

(+) DEVIANT PEERS PROSTITUTION

ATTACHMENT (-) SUPERVISION (-)

119

APPENDIX B: DESCRIPTIVES AND BIVARIATE STATISTICS FOR CORRELATES, CONTROL, AND MEDIATING/MODERATING VARIABLES

Table 1: Means and Standard Deviations Comparing Female Prostitutes and Female Non-prostitutes

Prostitutes Non-prostitutes (N = 36) (N = 37)

CORRELATES MEAN S.D. MEAN S.D. Sexual Abuse Scale 2.75 ** 2.33 1.30 1.79 Physical Abuse Scale 3.03 † 2.02 2.27 1.63 Contentious Family Environment in 1995 Scale 12.53 3.37 11.81 3.86 Conflict between Parents in 1982 Scale 5.77 2.77 5.51 2.68 Conflict between Parents in 1995 Scale 4.14 2.34 3.70 1.88 Parental Control & Supervision in 1982 Scale 10.38 2.46 11.32 2.66 Parental Control & Supervision in 1982A 0.47 0.51 0.62 0.49 Parental Caring & Trust in 1982 Scale 22.17 5.40 23.24 5.49 Parental Caring & Trust in 1982A 0.56 0.50 0.62 0.49 Parental Caring & Trust in 1995 Scale 20.08 4.40 19.86 4.84 Parental Caring & Trust in 1995A 0.58 0.50 0.62 0.49 Parental Intimate Communication in 1982 Scale 5.94 3.02 6.00 3.50 Parental Intimate Communication in 1982A 0.64 0.49 0.54 0.50 Parental Intimate Communication in 1995 Scale 3.89 2.15 3.38 1.62 Parental Intimate Communication in 1995A 0.64 0.49 0.49 0.51 Friends Paid for Sex in 1995A 0.56 ** 0.50 0.22 0.42 Peer Pressure in 1982 Scale 18.01 4.29 16.38 5.46 Peer Pressure in 1995 Scale 15.98 ** 5.19 12.82 3.40 A Paid for Sex in 1982 0.67 0.48 —— A Paid for Sex in 1995 0.58 0.50 —— A Paid for Sex in 2003 0.28 0.45 ——

CONTROL VARIABLES Has/Had Full-time job in 1995A 0.47 * 0.51 0.73 0.45 A Has/Had Part-time job in 1995 0.42 0.50 0.27 0.45 Unemployed in 1995A 0.11 * 0.32 0.00 0.00 Other Types of Employment in 1995A 0.53 * 0.51 0.27 0.45 Has/Had Full-time job in 2003A 0.53 0.51 0.70 0.46 Has/Had Part-time job in 2003A 0.19 0.40 0.22 0.42 Unemployed in 2003A 0.28 * 0.45 0.08 0.28 Other Types of Employment in 2003A 0.47 0.51 0.30 0.46 Age in 1982 16.61 1.23 16.54 1.04 Age in 1995 29.61 1.23 29.54 1.04 Age in 2003 37.34 1.38 37.23 1.47 Parental Social Class 0.08 0.93 -0.08 1.07 WhiteA 0.31 *** 0.47 0.73 0.45 BlackA 0.69 *** 0.47 0.27 0.45

MEDIATING/MODERATING VARIABLES Drug Problem in 1995 7.45 *** 9.17 0.81 2.40 Drug Problem in 2003 8.61 *** 10.91 1.54 5.27 Drug Usage in 1982 4.81 * 2.40 3.70 2.42 Drug Usage in 1995 3.33 *** 2.62 1.19 2.01 Drug Usage in 2003 3.22 *** 2.85 1.05 2.04 Running Away from Home in 1982 2.44 * 2.38 1.43 1.68 Self-esteem in 1995 21.86 * 3.40 23.59 2.86 Self-esteem in 2003 21.08 * 3.97 23.46 4.70 Depression in 1995 24.58 *** 8.41 16.24 8.38 Depression in 2003 24.69 *** 15.29 12.89 12.74 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 73 S.D. = STANDARD DEVIATION

†p<0.10, *p<0.05, **p<0.01, ***p<0.001 A = Dichotomous Variable

120

Table 2: Means and Standard Deviations Comparing White Female Prostitutes and White Female Non-prostitutes

White White Prostitutes Non-prostitutes (N = 11) (N = 27)

CORRELATES MEAN S.D. MEAN S.D. Sexual Abuse Scale 1.73 2.28 1.37 1.78 Physical Abuse Scale 2.64 2.29 2.11 1.63 Contentious Family Environment in 1995 Scale 11.45 4.08 11.81 3.73 Conflict between Parents in 1982 Scale 5.24 2.89 5.63 2.65 Conflict between Parents in 1995 Scale 3.00 0.77 3.67 1.94 Parental Control & Supervision in 1982 Scale 9.73 † 2.15 11.33 2.53 A Parental Control & Supervision in 1982 0.18 ** 0.40 0.63 0.49 Parental Caring & Trust in 1982 Scale 21.91 5.43 22.44 5.61 Parental Caring & Trust in 1982A 0.55 0.52 0.52 0.51 Parental Caring & Trust in 1995 Scale 20.55 4.91 20.18 4.99 A Parental Caring & Trust in 1995 0.73 0.47 0.63 0.49 Parental Intimate Communication in 1982 Scale 5.55 2.84 5.96 3.40 Parental Intimate Communication in 1982A 0.45 0.52 0.52 0.51 Parental Intimate Communication in 1995 Scale 4.09 1.76 3.41 1.31 Parental Intimate Communication in 1995A 0.73 0.47 0.52 0.51 Friends Paid for Sex in 1995A 0.36 0.50 0.15 0.36 Peer Pressure in 1982 Scale 18.91 3.94 17.04 5.08 Peer Pressure in 1995 Scale 14.27 5.42 12.33 2.99 A Paid for Sex in 1982 0.73 0.47 —— A Paid for Sex in 1995 0.27 0.47 —— A Paid for Sex in 2003 0.18 0.40 ——

CONTROL VARIABLES Has/Had Full-time job in 1995A 0.45 0.52 0.70 0.47 A Has/Had Part-time job in 1995 0.27 0.47 0.30 0.47 A Unemployed in 1995 0.27 † 0.47 0.00 0.00 A Other Types of Employment in 1995 0.55 0.52 0.30 0.47 Has/Had Full-time job in 2003A 0.64 0.50 0.74 0.45 Has/Had Part-time job in 2003A 0.18 0.40 0.22 0.42 Unemployed in 2003A 0.18 0.44 0.04 0.19 Other Types of Employment in 2003A 0.36 0.50 0.26 0.45 Age in 1982 16.73 1.01 16.48 1.09 Age in 1995 29.73 1.01 29.48 1.09 Age in 2003 37.39 1.02 37.12 1.53 Parental Social Class -0.04 1.04 -0.23 0.82 WhiteA 1.00 0.00 1.00 0.00 BlackA —— ——

MEDIATING/MODERATING VARIABLES Drug Problem in 1995 4.73 † 6.87 0.74 2.12 Drug Problem in 2003 6.09 † 9.35 1.00 2.48 Drug Usage in 1982 5.45 † 1.81 3.85 2.54 Drug Usage in 1995 3.64 *** 2.16 1.04 1.99 Drug Usage in 2003 3.55 ** 2.84 1.19 2.11 Running Away from Home in 1982 2.45 2.47 1.63 1.60 Self-esteem in 1995 21.64 4.01 23.30 2.93 Self-esteem in 2003 21.27 5.48 22.81 4.86 Depression in 1995 22.36 † 7.55 16.56 8.51 Depression in 2003 26.09 * 17.63 12.93 13.86 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 38 S.D. = STANDARD DEVIATION

†p<0.10, *p<0.05, **p<0.01, ***p<0.001 A = Dichotomous Variable

121

Table 3: Means and Standard Deviations Comparing Black Female Prostitutes and Black Female Non-prostitutes

Black Black Prostitutes Non-prostitutes (N = 25 ) (N = 10 )

CORRELATES MEAN S.D. MEAN S.D. Sexual Abuse Scale 3.20 ** 2.25 1.10 1.91 Physical Abuse Scale 3.20 1.91 2.70 1.64 Contentious Family Environment in 1995 Scale 13.00 2.97 11.80 4.39 Conflict between Parents in 1982 Scale 6.00 2.75 5.20 2.90 Conflict between Parents in 1995 Scale 4.64 2.63 3.80 1.81 Parental Control & Supervision in 1982 Scale 10.66 2.58 11.30 3.13 Parental Control & Supervision in 1982A 0.60 0.50 0.60 0.52 Parental Caring & Trust in 1982 Scale 22.29 5.50 25.40 4.72 A Parental Caring & Trust in 1982 0.56 † 0.51 0.90 0.32 Parental Caring & Trust in 1995 Scale 19.88 4.25 19.00 4.57 Parental Caring & Trust in 1995A 0.52 0.51 0.60 0.52 Parental Intimate Communication in 1982 Scale 6.12 3.14 6.10 3.90 Parental Intimate Communication in 1982A 0.72 0.46 0.60 0.52 Parental Intimate Communication in 1995 Scale 3.80 2.33 3.30 2.36 Parental Intimate Communication in 1995A 0.60 0.50 0.40 0.52 Friends Paid for Sex in 1995A 0.64 0.49 0.40 0.52 Peer Pressure in 1982 Scale 17.61 4.45 14.60 6.33 Peer Pressure in 1995 Scale 16.73 5.01 14.14 4.23 A Paid for Sex in 1982 0.64 0.49 —— A Paid for Sex in 1995 0.72 0.46 —— A Paid for Sex in 2003 0.32 0.48 ——

CONTROL VARIABLES A Has/Had Full-time job in 1995 0.48 † 0.51 0.80 0.42 A Has/Had Part-time job in 1995 0.48 0.51 0.20 0.42 Unemployed in 1995A 0.04 0.20 0.00 0.00 A Other Types of Employment in 1995 0.52 † 0.51 0.20 0.42 Has/Had Full-time job in 2003A 0.48 0.51 0.60 0.52 Has/Had Part-time job in 2003A 0.20 0.41 0.20 0.42 Unemployed in 2003A 0.32 0.48 0.20 0.42 Other Types of Employment in 2003A 0.52 0.51 0.40 0.52 Age in 1982 16.56 1.33 16.70 0.95 Age in 1995 29.56 1.33 29.70 0.95 Age in 2003 37.32 1.53 37.53 1.35 Parental Social Class 0.14 0.89 0.31 1.56 WhiteA —— —— BlackA 1.00 0.00 1.00 0.00

MEDIATING/MODERATING VARIABLES Drug Problem in 1995 8.64 ** 9.90 1.00 3.16 Drug Problem in 2003 9.72 11.53 3.00 9.49 Drug Usage in 1982 4.52 2.60 3.30 2.16 Drug Usage in 1995 3.20 2.83 1.60 2.12 Drug Usage in 2003 3.08 * 2.90 0.70 1.89 Running Away from Home in 1982 2.44 † 2.40 0.90 1.85 Self-esteem in 1995 21.96 * 3.18 24.40 2.63 Self-esteem in 2003 21.00 ** 3.23 25.20 3.94 Depression in 1995 25.56 ** 8.73 15.40 8.40 Depression in 2003 24.08 * 14.49 12.80 9.74 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 35 S.D. = STANDARD DEVIATION

†p<0.10, *p<0.05, **p<0.01, ***p<0.001 A = Dichotomous Variable

122

Table 4: Means and Standard Deviations Comparing White Female Prostitutes and Black Female Prostitutes

White Black Prostitutes Prostitutes (N = 11) (N = 25)

CORRELATES MEAN S.D. MEAN S.D. Sexual Abuse Scale 1.73 † 2.28 3.20 2.25 Physical Abuse Scale 2.64 2.29 3.20 1.91 Contentious Family Environment in 1995 Scale 11.45 4.08 13.00 2.97 Conflict between Parents in 1982 Scale 5.24 2.89 6.00 2.75 Conflict between Parents in 1995 Scale 3.00 ** 0.77 4.64 2.63 Parental Control & Supervision in 1982 Scale 9.73 2.15 10.66 2.58 Parental Control & Supervision in 1982A 0.18 * 0.40 0.60 0.50 Parental Caring & Trust in 1982 Scale 21.91 5.43 22.29 5.50 Parental Caring & Trust in 1982A 0.55 0.52 0.56 0.51 Parental Caring & Trust in 1995 Scale 20.55 4.91 19.88 4.25 Parental Caring & Trust in 1995A 0.73 0.47 0.52 0.51 Parental Intimate Communication in 1982 Scale 5.55 2.84 6.12 3.14 Parental Intimate Communication in 1982A 0.45 0.52 0.72 0.46 Parental Intimate Communication in 1995 Scale 4.09 1.76 3.80 2.33 Parental Intimate Communication in 1995A 0.73 0.47 0.60 0.50 A Friends Paid for Sex in 1995 0.67 1.41 0.64 0.49 Peer Pressure in 1982 Scale 18.00 3.74 17.61 4.45 Peer Pressure in 1995 Scale 13.67 5.05 16.73 5.01 A Paid for Sex in 1982 0.73 0.47 0.64 0.49 Paid for Sex in 1995A 0.27 ** 0.47 0.72 0.46 Paid for Sex in 2003A 0.18 0.40 0.32 0.48

CONTROL VARIABLES Has/Had Full-time job in 1995A 0.45 0.52 0.48 0.51 A Has/Had Part-time job in 1995 0.27 0.47 0.48 0.51 Unemployed in 1995A 0.27 0.47 0.04 0.20 Other Types of Employment in 1995A 0.55 0.52 0.52 0.51 Has/Had Full-time job in 2003A 0.64 0.50 0.48 0.51 Has/Had Part-time job in 2003A 0.18 0.40 0.20 0.41 Unemployed in 2003A 0.18 0.44 0.32 0.48 Other Types of Employment in 2003A 0.36 0.50 0.52 0.51 Age in 1982 16.73 1.01 16.56 1.33 Age in 1995 29.73 1.01 29.56 1.33 Age in 2003 37.39 1.02 37.32 1.53 Parental Social Class -0.04 1.04 0.14 0.89 A White 1.00 0.00 —— A Black —— 1.00 0.00

MEDIATING/MODERATING VARIABLES Drug Problem in 1995 4.73 6.87 8.64 9.90 Drug Problem in 2003 6.09 9.35 9.72 11.53 Drug Usage in 1982 5.45 1.81 4.52 2.60 Drug Usage in 1995 3.64 2.16 3.20 2.83 Drug Usage in 2003 3.55 2.84 3.08 2.90 Running Away from Home in 1982 2.45 2.47 2.44 2.40 Self-esteem in 1995 21.64 4.01 21.96 3.18 Self-esteem in 2003 21.27 5.48 21.00 3.23 Depression in 1995 22.36 7.55 25.56 8.73 Depression in 2003 26.09 17.63 24.08 14.49 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 36 S.D. = STANDARD DEVIATION

†p<0.10, *p<0.05, **p<0.01, ***p<0.001 A = Dichotomous Variable

123

APPENDIX C: REGRESSION MODELS

Table 5: Logit Estimates for the Regression of Prostitution on the Strain Theory Measures

CORRELATES Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Intercept -7.871 -4.720 -8.105 † -16.920 * -15.893 ** -29.372 † -20.031 † -73.355 † -23.299 * (5.681) (4.323) (4.608) (8.185) (6.451) (17.609) (10.901) (41.327) (11.285)

Sexual Abuse Scale 0.311 * 0.333 * 0.3321 * 0.6378 ** 0.3735 * 1.214 † 0.780 * 1.999 0.721 * (0.145) (0.142) (0.146) (0.261) (0.164) (0.689) (0.366) (1.548) (0.339)

Contentious Family Environment in 1995 Scale 0.091 0.334 (0.107) (0.416)

Conflict between Parents in 1982 Scale 0.009 -0.908 (0.117) (0.722)

A Conflict between Parents in 1995 Scale -0.001 -0.359 1.197 0.032 0.374 -4.367 (0.150) (0.386) (0.759) (2.857) (1.821) (5.285)

Friends Paid for Sex in 1995Peer Pressure in 1982 Scale 0.1376 † 0.623 † 0.461 † 1.867 (0.079) (0.374) (0.271) (1.385)

Peer Pressure in 1995 Scale 0.0996 -0.211 -0.063 0.337 (0.086) (0.329) (0.177) (0.579)

Parental Intimate Communication in 1982 -1.207 (3.366)

Parental Intimate Communication in 1995 1.714 (2.298)

Drug Problem in 1995 0.318 * 0.689 0.596 2.588 0.302 † (0.158) (0.560) (0.394) (1.976) (0.171)

Drug Problem in 2003 0.094 0.217 0.068 -0.141 0.125 (0.068) (0.163) (0.080) (0.187) (0.085)

Drug Usage in 1982 0.215 0.114 0.193 0.377 0.313 (0.211) (0.335) (0.267) (0.617) (0.242)

Drug Usage in 1995 -0.260 -0.208 -0.235 -1.038 -0.342 (Continued on Next(0.255) Page) (0.398) (0.285) (0.667) (0.298) Drug Usage in 2003 0.402 † 0.961 † 0.530 † 1.460 † 0.397 † (0.215) (0.517) (0.289) (0.833) (0.232)

124

Table 5: Logit Estimates for the Regression of the Strain Theory Measures on Prostitution (Contd.)

Model 1B Model 2B Model 3B Model 4B Model 5B Model 6B Model 7B Model 8B Model 9B Running Away from Home in 1982 0.395 * 0.586 * 0.410 * 1.270 0.717 3.923 0.603 * (0.172) (0.252) (0.212) (0.846) (0.491) (2.566) (0.269)

Self-esteem in 1995 0.093 (0.719)

Self-esteem in 2003 0.419 (0.488)

Depression in 1995 0.502 † 0.098 (0.269) (0.063)

Depression in 2003 -0.007 -0.025 (0.158) (0.046) Had/Have Full-time Job in 1995 - Ref. Had/Have Other Types of Employment in 1995A 1.601 ** 1.382 * 1.447 * 2.314 * 1.586 * 5.069 * 3.433 * 9.453 2.819 * (0.660) (0.598) (0.625) (0.980) (0.695) (2.588) (1.569) (6.230) (1.244) Had/Have Full-time Job in 2003 - Ref. Had/Have Other Types of Employment in 2003A 0.564 0.495 0.827 0.782 1.059 2.684 2.628 18.501 0.845 (0.607) (0.591) (0.660) (1.089) (0.777) (3.891) (2.674) (15.450) (1.186)

Age in 1982 0.267 0.151 0.2971 0.606 0.508 0.541 0.139 -1.308 0.852 (0.292) (0.255) (0.266) (0.467) (0.336) (1.070) (0.775) (2.407) (0.587)

Parental Social Class -0.035 0.000 -0.242 -0.240 0.034 0.934 0.363 -1.160 -0.452 (0.283) (0.277) (0.307) (0.474) (0.372) (0.910) (0.644) (0.946) (0.478) White - Ref. BlackA 1.589 ** 1.614 ** 1.823 ** 2.076 * 1.427 * 5.059 3.959 † 18.446 2.318 * (0.604) (0.587) (0.641) (1.030) (0.720) (3.247) (2.336) (14.011) (1.195)

R2 0.304 0.295 0.352 0.560 0.430 0.635 0.607 0.664 0.576 Model Χ2 26.425 ** 25.486 *** 31.681 *** 59.887 *** 41.016 *** 73.594 *** 68.250 *** 79.519 *** 62.638 *** Hosmer & Lemeshow Goodness-of-Fit Test (χ2) 13.019 4.957 7.526 4.283 8.416 4.584 11.076 17.547 *** 7.707 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 73 Ref. = Reference Category A = Dichotomous Variable p†<0.10, *p<0.05, **p<0.01, ***p<0.001 Additive Coefficients are in Italics Standard Errors are in Parentheses B = Continued

125

Table 6: Logit Estimates for the Regression of Prostitution on the Social Control Theory Measures

CORRELATES Model 10 Model 11 Model 12 Model 13 Model 14 Model 15 Model 16 Model 17 Intercept -2.187 -1.632 -1.676 -8.592 -7.358 -8.431 -16.755 * -10.890 † (4.644) (4.294) (4.355) (5.897) (5.362) (5.589) (7.795) (5.891)

Parental Control & Supervision in 1982 -1.773 * -1.484 * -1.713 * -1.506 * -1.733 † -1.306 † (0.777) (0.640) (0.840) (0.731) (0.997) (0.775)

Parental Caring & Trust in 1982 -0.388 -0.778 -0.688 -0.994 -0.067 (0.671) (0.630) (0.754) (0.718) (0.922)

Parental Caring & Trust in 1995 -0.177 -0.173 0.449 0.545 1.345 (0.637) (0.614) (0.734) (0.710) (0.965)

Parental Intimate Communication in 1982A 0.765 0.219 0.767 0.3271.248 † 0.961 1.209 † 1.381 † 1.316 † 1.112 (0.704) (0.624) (0.736) (0.663)(0.762) (0.840) (0.720) (0.742) (0.792) (0.739)

Parental Intimate Communication Peerin 1995 Pressure in 19821.003 Scale 0.914 0.924 0.8750.172 * 0.933 0.157 * 0.143 † 0.187 * 0.181 * (0.632) (0.595) (0.684) (0.649)(0.082) (0.749) (0.078) (0.075) (0.084) (0.082)

Friends Paid for Sex in 1995 Peer Pressure in 1995 Scale 0.091 0.065 0.112 0.132 0.066 (0.089) (0.080) (0.084) (0.107) (0.083)

Sexual Abuse Scale A 1.784 ** 1.680 ** 1.234 * 2.043 ** 1.925 ** 1.501 * 1.960 * 1.9530.560 *** 0.336 * (0.681) (0.633) (0.583) (0.788) (0.731) (0.676) (0.828) (0.758)(0.249) (0.162) Had/Have Full-timeHad/Have Job Full-timein 2003 - Ref.Job in 1995A - Ref.0.277 0.330 0.504 0.397 0.489 0.530 0.718 0.704 Had/Have OtherHad/Have Types of Other Employment Types of in Employment 2003 (0.619) in 1995 (0.594) (0.591) (0.707) (0.675) (0.669) (0.764) (0.707)

Age in 1982 0.029 0.033 -0.006 0.113 0.128 0.105 0.440 0.270 (0.277) (0.259) (0.261) (0.326) (0.304) (0.309) (0.380) (0.316)

Parental Social Class -0.119 -0.147 -0.085 0.185 0.149 0.228 0.464 0.266 A (0.296)2.348 *** (0.279)2.208 *** (0.274)2.055 *** (0.330)2.394 ** (0.321)2.048 ** (0.321)1.903 ** (0.371)1.684 † (0.338)1.764 * White - Ref. (0.694) (0.632) (0.624) (0.855) (0.742) (0.760) (0.880) (0.763) 2 Black 0.334 0.295 0.276 0.416 0.380 0.377 0.471 0.419 2 RModel Χ 29.634 *** 25.533 *** 23.591 ** 39.230 *** 34.877 *** 34.492 *** 46.411 *** 39.606 *** Hosmer & Lemeshow Goodness-of-Fit Test (χ2) 2.906 2.264 6.237 5.721 9.847 15.278 * 7.474 5.661 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 73 Ref. = Reference Category A = Dichotomous Variable

†p<0.10, *<0.05, **p<0.01, ***p<0.001 Additive Coefficients are in Italics Standard Errors are in Parentheses

126

Table 7: Logit Estimates for the Regression of Prostitution on the Social Learning Theory Measures

CORRELATES Model 18 Model 19 Model 20 Model 21 Model 22 Model 23 Intercept -8.195 -15.608 † -10.207 -9.060 -14.849 10.285 (5.236) (8.745) (11.666) (10.173) (10.190) (9.289)

Friends Paid for Sex in 1995A 1.304 † 1.771 3.509 0.759 1.327 1.370 (0.698) (1.172) (2.178) (1.399) (1.267) (1.324)

Peer Pressure in 1982 Scale 0.131 † 0.384 * 0.000 0.552 * 0.472 * 0.364 * (0.072) (0.170) (0.189) (0.267) (0.216) (0.171)

Peer Pressure in 1995 Scale 0.082 -0.157 -0.548 -0.193 -0.177 -0.120 (0.076) (0.123) (0.346) (0.183) (0.139) (0.130)

Parental Control & Supervision in 1982 -2.553 † -1.811 (1.570) (1.134)

Parental Caring & Trust in 1982 -25.033 * -1.961 † -1.756 (11.245) (1.213) (1.086)

Parental Caring & Trust in 1995 -0.654 -0.287 (1.587) (1.303)

Parental Intimate Communication in 1982 -0.874 -0.910 (1.225) (1.086)

Parental Intimate Communication in 1995 1.945 1.316 (1.534) (1.304)

Friends Paid for SexA X Care & Trust in 1982 -1.989 (2.597)

Peer Press. 1982 X Care & Trust in 1982 0.850 * (0.400)

Peer Press. 1995 X Care & Trust in 1982 0.509 (0.391)

Drug Problem in 1995 0.250 * 0.566 * 0.464 * 0.342 * 0.291 * (0.119) (0.239) (0.234) (0.168) (0.146)

Drug Problem in 2003 0.058 0.080 -0.006 0.046 0.026 (0.058) (0.089) (0.079) (0.062) (0.071)

Drug Usage in 1982 0.305 0.475 0.324 0.201 0.420 (0.223) (0.358) (0.298) (0.237) (0.275)

Drug Usage in 1995 0.046 -0.664 † -0.188 -0.047 -0.075 (0.187) (0.388) (0.256) (0.213) (0.243)

Drug Usage in 2003 0.523 * 1.217 ** 0.806 * 0.562 * 0.624 * (0.241) (0.501) (0.390) (0.279) (0.279) Had/Have Full-time Job in 1995 - Ref. Had/Have Other Types of Employment in 1995A 1.402 * 3.540 ** 5.798 ** 5.287 ** 4.289 ** 3.662 ** (0.634) (1.324) (2.119) (2.096) (1.627) (1.391) Had/Have Full-time Job in 2003 - Ref. Had/Have Other Types of Employment in 2003A 0.488 0.102 0.482 1.669 0.103 1.163 (0.648) (1.026) (1.349) (1.795) (1.173) (1.531)

Age in 1982 0.171 0.228 0.490 -0.369 0.140 -0.116 (0.293) (0.454) (0.658) (0.675) (0.540) (0.574)

Parental Social Class 0.220 0.424 -0.684 -0.246 0.462 -1.526 (0.325) (0.469) (0.750) (0.599) (0.498) (0.566) White - Ref. BlackA 1.457 * 2.912 * 5.458 ** 6.619 * 3.934 * 4.161 * (0.634) (1.234) (2.179) (3.025) (1.684) (1.690)

R2 0.338 0.539 0.607 0.585 0.557 0.566 Model Χ2 30.109 *** 56.488 *** 68.164 *** 64.266 *** 59.508 *** 60.878 *** Hosmer & Lemeshow Goodness-of-Fit Test (χ2) 6.140 7.479 3.587 4.272 2.319 3.579 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 73 Ref. = Reference Category A = Dichotomous Variable †p<0.10, *<0.05, **p<0.01, ***p<0.001 Additive Coefficients are in Italics Standard Errors are in Parentheses

127

APPENDIX D: DESCRIPTIVES AND BIVARIATE STATISTICS FOR THE CONSEQUENCES OF PROSTITUTION

Table 8: Means and Standard Deviations for the Consequences of Prostitution Comparing Female Prostitutes and Female Non-prostitutes

Prostitutes Non-prostitutes (N = 36) (N = 37)

1995 CONSEQUENCES OF PROSTITUTION MEAN S.D. MEAN S.D. Perpetrated Domestic Violence 5.56 * 8.35 2.57 3.51 A Perpetrated Domestic Violence 0.64 0.49 0.54 0.51 Victim of Domestic Violence 3.73 † 6.26 1.79 3.12 Victim of Domestic ViolenceA 0.58 0.50 0.54 0.51 Neighborhood Type Scale 12.17 4.31 11.44 4.19 MarriedA 0.25 † 0.44 0.43 0.50 Never MarriedA 0.56 * 0.50 0.32 0.47 DivorcedA 0.11 0.32 0.14 0.35 SeparatedA 0.03 0.17 0.08 0.28 WidowedA 0.06 0.23 0.03 0.16 Education Level 1.39 0.84 1.59 0.69

2003 CONSEQUENCES OF PROSTITUTION Number of Times MarriedA 0.50 * 0.51 0.76 0.43 Received Public Assistance in 2003A 0.97 *** 0.17 0.65 0.48 Currently MarriedA 0.17 *** 0.38 0.59 0.50 Currently DivorcedA 0.36 * 0.49 0.14 0.35 Currently CohabitingA 0.19 † 0.40 0.05 0.23 Never MarriedA 0.25 0.44 0.16 0.37 Currently Not In a RelationshipA 0.31 0.47 0.22 0.42 In a Happy RelationshipA 0.36 † 0.49 0.57 0.50 Lack Money for Items Scale 10.61 ** 4.76 7.49 4.39 Lack Money for Items ScaleA 0.58 ** 0.50 0.30 0.46 Need Items Scale 4.69 ** 2.48 3.16 2.01 Need Items ScaleA 0.61 * 0.49 0.38 0.49 People in Neighborhood Scale 12.77 * 12.77 14.40 3.11 Neighborhood Type Scale 12.83 * 4.25 10.44 4.41 A Living in Poverty 0.64 *** 0.49 0.24 0.43 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 73 S.D. = STANDARD DEVIATION

†p<0.10, *p<0.05, **p<0.01, ***p<0.001 A = Dichotomous Variable

128

Table 9: Means and Standard Deviations for the Consequences of Prostitution Comparing White Female Prostitutes and White Female Non-prostitutes

White White Prostitutes Non-prostitutes (N = 11) (N = 27)

1995 CONSEQUENCES OF PROSTITUTION MEAN S.D. MEAN S.D. Perpetrated Domestic Violence 2.55 2.77 2.56 3.76 Perpetrated Domestic ViolenceA 0.73 0.47 0.48 0.51 Victim of Domestic Violence 3.82 5.81 1.80 3.52 Victim of Domestic ViolenceA 0.64 0.50 0.52 0.51 Neighborhood Type Scale 10.96 3.54 10.32 3.57 A Married 0.36 0.50 0.48 0.51 A Never Married 0.36 0.50 0.19 0.40 DivorcedA 0.09 0.30 0.19 0.40 SeparatedA 0.00 † 0.00 0.11 0.32 WidowedA 0.18 0.40 0.04 0.19 Education Level 1.73 1.01 1.70 0.72

2003 CONSEQUENCES OF PROSTITUTION Number of Times MarriedA 0.82 0.40 0.85 0.85 Received Public Assistance in 2003A 0.91 ** 0.30 0.52 0.51 Currently MarriedA 0.36 * 0.50 0.70 0.47 Currently DivorcedA 0.18 0.40 0.07 0.27 Currently CohabitingA 0.27 0.47 0.04 0.19 A Never Married 0.09 0.30 0.11 0.32 Currently Not In a RelationshipA 0.27 0.47 0.19 0.40 In a Happy RelationshipA 0.45 0.52 0.67 0.48 Lack Money for Items Scale 10.55 † 4.06 7.81 3.87 Lack Money for Items ScaleA 0.64 † 0.50 0.30 0.47 Need Items Scale 4.18 2.27 3.11 1.85 A Need Items Scale 0.45 0.52 0.41 0.50 People in Neighborhood Scale 12.33 * 3.46 14.86 3.14 Neighborhood Type Scale 10.42 2.96 9.13 3.25 A Living in Poverty 0.45 † 0.52 0.19 0.40 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 38 S.D. = STANDARD DEVIATION

†p<0.10, *p<0.05, **p<0.01, ***p<0.001 A = Dichotomous Variable

129

Table 10: Means and Standard Deviations for the Consequences of Prostitution Comparing Black Female Prostitutes and Black Female Non-prostitutes

Black Black Prostitutes Non-prostitutes (N = 25 ) (N = 10 )

1995 CONSEQUENCES OF PROSTITUTION MEAN S.D. MEAN S.D. Perpetrated Domestic Violence 6.88 * 9.61 2.60 2.92 Perpetrated Domestic ViolenceA 0.60 0.50 0.70 0.48 Victim of Domestic Violence 3.69 6.56 1.77 1.73 Victim of Domestic ViolenceA 0.56 0.51 0.60 0.52 Neighborhood Type Scale 12.69 4.57 14.46 4.43 MarriedA 0.20 0.41 0.30 0.48 Never MarriedA 0.64 0.49 0.70 0.48 DivorcedA 0.12 † 0.33 0.00 0.00 A Separated 0.04 0.20 0.00 0.00 WidowedA 0.00 0.00 0.00 0.00 Education Level 1.24 0.72 1.30 0.48

2003 CONSEQUENCES OF PROSTITUTION Number of Times MarriedA 0.36 0.49 0.50 0.53 Received Public Assistance in 2003A 1.00 0.00 1.00 0.00 Currently MarriedA 0.08 0.28 0.30 0.48 Currently DivorcedA 0.44 0.51 0.30 0.48 Currently CohabitingA 0.16 0.37 0.10 0.32 Never MarriedA 0.32 0.48 0.30 0.48 Currently Not In a RelationshipA 0.32 0.48 0.30 0.48 In a Happy RelationshipA 0.32 0.48 0.30 0.48 Lack Money for Items Scale 10.64 * 5.11 6.60 5.70 Lack Money for Items ScaleA 0.56 0.51 0.30 0.48 Need Items Scale 4.92 † 2.58 3.30 2.50 Need Items ScaleA 0.68 * 0.48 0.30 0.48 People in Neighborhood Scale 12.97 2.39 13.16 2.81 Neighborhood Type Scale 13.89 4.34 13.96 5.31 A Living in Poverty 0.72 † 0.46 0.40 0.52 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 35 S.D. = STANDARD DEVIATION

†p<0.10, *p<0.05, **p<0.01, ***p<0.001 A = Dichotomous Variable

130

Table 11: Means and Standard Deviations for the Consequences of Prostitution Comparing White Female Prostitutes and Black Female Prostitutes

White Black Prostitutes Prostitutes (N = 11) (N = 25)

1995 CONSEQUENCES OF PROSTITUTION MEAN S.D. MEAN S.D. Perpetrated Domestic Violence 5.56 * 8.35 6.88 9.61 Perpetrated Domestic ViolenceA 0.64 0.49 0.60 0.50 Victim of Domestic Violence 3.73 6.26 3.69 6.56 Victim of Domestic ViolenceA 0.58 0.50 0.56 0.51 Neighborhood Type Scale 12.17 4.31 12.69 4.57 MarriedA 0.25 0.44 0.20 0.41 Never MarriedA 0.56 0.50 0.64 0.49 DivorcedA 0.11 0.32 0.12 0.33 SeparatedA 0.03 0.17 0.04 0.20 WidowedA 0.06 0.23 0.00 0.00 Education Level 1.39 0.84 1.24 0.72

2003 CONSEQUENCES OF PROSTITUTION Number of Times MarriedA 0.50 ** 0.51 0.36 0.49 Received Public Assistance in 2003A 0.97 0.17 1.00 0.00 Currently MarriedA 0.17 † 0.38 0.08 0.28 Currently DivorcedA 0.36 0.49 0.44 0.51 Currently CohabitingA 0.19 0.40 0.16 0.37 A Never Married 0.25 0.44 0.32 0.48 Currently Not In a RelationshipA 0.31 0.47 0.32 0.48 In a Happy RelationshipA 0.45 0.52 0.32 0.48 Lack Money for Items Scale 10.61 4.76 10.64 5.11 Lack Money for Items ScaleA 0.58 0.50 0.56 0.51 Need Items Scale 4.69 2.48 4.92 2.58 A Need Items Scale 0.61 0.49 0.68 0.48 People in Neighborhood Scale 12.77 12.77 12.97 2.39 Neighborhood Type Scale 12.83 * 4.25 13.89 4.34 A Living in Poverty 0.64 0.49 0.72 0.46 Source: Ohio Lifecourse Study (OLS), 1982, 1995, & 2003 Total N = 36 S.D. = STANDARD DEVIATION

†p<0.10, *p<0.05, **p<0.01, ***p<0.001 A = Dichotomous Variable