Assessing Stalking Behaviors in a Control Balance Theory Framework

Abstract

Tittle’s control balance theory is a modern exemplar of theoretical integration premised on the ratio of perceived control exerted versus control subjected, which predicts the probability of deviant acts. Although the theory offers promise for contextualizing both perpetration and victimization, relatively few empirical tests have been conducted to date. This study describes stalking as a crime representing a unique manifestation of power dynamics and perceived control imbalances for both offenders and victims. Using a large college student sample, we assess the empirical associations between control balance and stalking perpetration and victimization outcomes, including gender-specific models, featuring several different operationalizations of the control balance ratio. Results suggest partial support for the theory, demonstrating that control surpluses are associated with stalking perpetration among men, while control deficits are associated with stalking perpetration and victimization among women, net of controls. Implications for control balance theory and stalking research are discussed.

Keywords

stalking control balance theory perpetration victimization gender differences

In its original formulation as well as subsequent revisions, Tittle’s (1995, 2004) control balance theory was conceived as a model for “synthetic” theoretical integration that incorporates elements of several independent theoretical perspectives into a general theory of behavior. Although criminologists have trumpeted the various benefits that accompany theory integration for years (Messner, Krohn, & Liska, 1989), control balance theory has received very limited empirical attention, with fewer than two dozen peer-reviewed articles published. To date, control balance theory successfully explains a number of offenses, including physical assault (Baron & Forde, 2007; Higgins, Lauterbach, & Tewksbury, 2005; Piquero & Hickman, 1999), serious theft (Baron & Forde, 2007), corporate crime (Piquero & Piquero, 2006), cheating on an exam (Curry, 2005), exploiting a fellow classmate for notes (Higgins & Lauterbach, 2004), engaging in devious sexual behavior (Piquero & Hickman, 1999), and reporting fellow police officers’ deviance (Hickman, Piquero, Lawton, & Greene, 2001). Control balance theory has also been extended to explain crime victimization, including family violence victimization (DeLisi & Hochstetler, 2002) and general and theft victimization (Piquero & Hickman, 2003). Collectively, the comparatively limited empirical attention shows much promise for the theory to explain a variety of deviance, offending, and victimization.

Separately, criminologists have only recently begun to explore the theoretical underpinnings of stalking, which has been explicitly defined in state statutes for around only 20 years. California implemented the first antistalking legislation in 1990 (and all 50 states quickly adopted similar legislation) in response to media coverage of celebrity stalking victims (Fox, Nobles, & Fisher, 2011). Although the language of antistalking laws varies somewhat by state, stalking is defined as a pattern of unwanted, repeated, frightening, and/or threatening pursuit behaviors (Tjaden & Thoennes, 1998). Since the criminalization of stalking, research on the effects of stalking reveals that victims often experience a number of emotional and physical consequences, including anxiety, paranoia, fear, and appetite and sleep disturbances (Hall, 1998; Pathé & Mullen, 1997). Experiencing stalking also results in victims altering their behavior for protection, such as changing usual daily activities, staying with friends or relatives, and taking time off from work (Nobles, Reyns, Fox, & Fisher, in press). The deleterious effects of stalking highlight the importance of understanding stalking victimization and perpetration, yet very few studies have empirically tested theories for understanding these phenomena. For example, attachment theory (Davis, Ace, & Andra, 2000; Dye & Davis, 2003; Kienlen, Birmingham, Solberg, O’Regan, & Meloy, 1997; Lewis, Fremouw, Del Ben, & Farr, 2001; Patton, Nobles, & Fox, 2010; Tonin, 2004) and social learning theory (Fox, Nobles, & Akers, 2011) have predicted stalking perpetration, whereas routine activity theory (Fisher, Cullen, & Turner, 2002; Mustaine & Tewksbury, 1999; Reyns, Henson, & Fisher, 2011), self-control theory (Fox, Gover, & Kaukinen, 2009), and social learning theory (Fox, Nobles, & Fisher, 2011) have explained stalking victimization. The current study represents the first empirical test of control balance theory’s ability to explain both stalking perpetration and victimization.

Despite the relatively brief bodies of literature on these topics, pairing control balance theory and stalking behaviors offers intuitive appeal for the validity of the constructs, the sophistication of the theoretical processes, and the complexity of the outcomes. A central theme of control balance theory is the push-and-pull relationship between control exerted and control subjected, while a key dynamic in stalking scenarios is the interplay between perceived control exerted and subjected on the part of the perpetrator and the victim. This study interprets stalking perpetration and victimization through the lens of control balance theory and offers a rare empirical test to extend knowledge in both domains.

An Overview of Control Balance Theory

At its most basic level, control balance theory is concerned with the amount of control one is capable of exerting versus the amount of control to which one is subjected. The dynamic between control exerted and subjected is a nonlinear parabolic relationship best understood as a continuum of control balance (see Figure 1). According to Tittle (1995), deviance, or “any behavior that the majority of a given group regards as unacceptable or that typically evokes a collective response of a negative type” (p. 124), is most likely to occur when the control ratio is unbalanced in either direction. In other words, deviance is a function of control deficits (i.e., an individual is subjected to more control than that individual exerts) or control surpluses (i.e., an individual exerts more control than the control to which he or she is subjected). Importantly, deviant behavior results from control imbalances in an effort to strategically “escape deficits and extend surpluses of control” (Tittle, 1995, p. 142). Alternatively, conformity is the product of a balanced control ratio, given that the equilibrium between control exerted and subjected does not require deviant behavior to “escape” or “extend” one’s level of control. Thus, when the ratio is expressed by the formula control exerted divided by control subjected (CE/CS), a control balance indicating conformity is represented by a ratio of 1:1 (i.e., control exerted equals control subjected), a control surplus is characterized by a ratio of more than 1 (i.e., control exerted exceeds control subjected), and a control deficit is illustrated by a ratio of less than 1 (i.e., control exerted is less than control subjected).

Figure 1:

Control Balance Ratio Continuum and Probability of Deviance

Tittle (1995) originally specified distinct types of deviance as expected outcomes from those with control surpluses versus control deficits. People with control deficits (left side of the figure) are more susceptible to repressive types of deviance, including predation (e.g., assault), defiance (e.g., vandalism), and submission (e.g., mindless obedience to commands). Alternatively, individuals with control surpluses (right side of the figure) are prone to autonomous deviance, consisting of exploitation (e.g., corporate price-fixing), plunder (e.g., pollution by industrial companies), and decadence (e.g., group sex with children). Based on compelling critiques of the theory (Braithwaite, 1997), Tittle (1997, 1999, 2004) reconceptualized the theory to consist of conformity, deviance, and submission.¹ Additionally, the theory now emphasizes the importance of self-control as a motivation for engaging in deviant behavior. Instead of suggesting that only low self-control leads to deviance, Tittle (2004) makes an important distinction between the level of self-control and the nature of deviant behavior. Given that individuals with low self-control have the inability to consider the long-term consequences of their actions yet they desire immediate gratification, Tittle (2004) argues that people with low self-control will be more likely to commit deviance that requires personal contact. Alternatively, individuals with high self-control will commit impersonal acts of deviance given that they are capable of resisting the urge for immediate action.

Control balance theory also accounts for the unique role that gender plays in terms of control ratios and deviance. Drawing from a feminist perspective (Daly & Chesney-Lind, 1988), Tittle (1995) acknowledges that “most females suffer serious deficits from control exercised by men collectively,” and therefore, women are “likely to commit submissive deviance” (p. 233). More specifically, Tittle (1995) argues that women will be more likely to commit deviant acts represented by submission (i.e., on the left extreme end of the spectrum) and that their involvement in the other types of deviance will be progressively less pronounced moving from the left side to the right side of the continuum (see Figure 2). Alternatively, the distribution of expected deviance among men is depicted as more evenly distributed on both sides of the control continuum, with the most deviance occurring among those who have the slightest imbalances (i.e., concentrated closest to the center of the figure) and gradually less deviance occurring as the outermost limits are approached on both extremes.

Figure 2:

Gendered Deviance Probabilities Versus Control Balance Continuum

Empirical Tests of Control Balance Theory

Perhaps due to the theory’s recent development and the need for original data collection efforts for appropriately testing its major tenets, relatively few empirical tests of the theory have been published. Nevertheless, several important trends have emerged from this new line of research. First, the majority of prior research employs the use of vignettes to measure the propensity toward deviance by presenting hypothetical scenarios and asking respondents to rate the likelihood that they would commit the deviant behavior featured (Baron & Forde, 2007; Curry, 2005; Hickman & Piquero, 2001; Hickman et al., 2001; Higgins & Lauterbach, 2004; Higgins et al., 2005; Piquero & Hickman, 1999; Piquero & Piquero, 2006). Although control balance theory originally predicted that control deficits would lead to repressive deviance whereas control surpluses would lead to autonomous deviance, a consistent finding emerging from much of the work using vignettes reveals that both deficits and surpluses are predictive of likelihood to commit physical assault (Baron & Forde, 2007; Higgins et al., 2005; Piquero & Hickman, 1999), serious theft (Baron & Forde, 2007), cheating on an exam (Curry, 2005), exploiting a fellow classmate for notes (Higgins & Lauterbach, 2004), and engaging in devious sexual behavior (Piquero & Hickman, 1999). However, consistent with theoretical predictions, Piquero and Piquero (2006) found that control surpluses (not deficits) were significantly predictive of exploitative corporate crime among a sample of adults. Additionally, among police officers, Hickman et al. (2001) found that control deficits (not surpluses) increased the likelihood of reporting fellow police officers’ defiance (e.g., a fellow officer driving under the influence of alcohol and using excessive violence on a suspected offender). Hickman et al. (2001) argue that rejecting the police cultural norm of covering up another officer’s deviance is intrinsically an act of defiance, and therefore, the finding that control deficits predicted both defiant scenarios is consistent with theoretical expectations.

Advancing the literature that operationalizes the likelihood of deviant behavior with hypothetical vignettes, Curry and Piquero (2003) were among the first to examine the effects of control balance on actual measures of deviance. Using a combined measure of drinking to the point of intoxication and using marijuana within the past 6 months, the findings revealed that both control deficits and surpluses predicted deviance, which is consistent with the body of published work featuring vignettes, yet only partially consistent with Tittle’s (1995) original expectations. DeLisi and Hochstetler (2002) also used actual measures of deviance in an effort to test control balance theory, although the authors used a wave of the National Youth Study; thus, direct measures of control were unavailable within the existing data. Using proxy measures of control, several indicators were significantly associated with defiance, predation, exploitation, and decadence (e.g., friendship stress imbalance). None of the proxy control measures predicted submission (and the data set did not include any available measures for plunder). Interestingly, DeLisi and Hochstetler (2002) also examined family violence perpetration and victimization, which were both related to some proxy control measures (e.g., partner emotional imbalance). These findings are suggestive of a generalization of control balance theory that indicates possible reciprocal effects of control imbalances, and they also provide a basis for examining the offender–victim relationship in greater depth with respect to interpersonal crimes that are inherently associated with expressions of domination and submission.

In addition to DeLisi and Hochstetler’s (2002) work, Piquero and Hickman (2003) are the only other scholars to date to extend control balance theory to account for victimization. Using a sample of college students, Piquero and Hickman (2003) examined a general measure of victimization within the past year and a measure of lifetime theft victimization. Consistent with the majority of the literature focused on perpetration, both control deficits and surpluses predicted general and theft victimization. This line of research is an important theoretical advancement, which provides an additional foundation for the current study. Although control balance theory, like most criminological theories, was originally crafted as an explanation for crime and deviant behavior, this recent extension suggests that the general theory may also be effective for understanding crime victimization. Piquero and Hickman (2003) develop a foundation for extending control balance theory to explain victimization separately for the control deficit and control surplus spectrum extremes. As articulated by Piquero and Hickman (2003), individuals with control deficits are considered weak, subordinate, and acutely vulnerable to victimization as they “do not have the confidence and/or skills to defend themselves against those who may victimize them. The sense of weakness and helplessness is detected by offenders who exploit such individuals for their own gain” (p. 286). Importantly, the authors acknowledge that much victimization is nonrandom and that many victims and offenders are acquainted with one another. Piquero and Hickman (2003) further explain that control surpluses may lead to feelings of invincibility that, in turn, place individuals in riskier situations that are likely to lead to victimization “because they are likely to seek out more frequent and increasingly exciting or risky actions, situations, and events in which to partake” (p. 286). As an extension of this contemporary line of research, the current study examines control balance theory’s ability to account for victimization.

Prior research using Grasmick, Tittle, Bursik, and Arneklev’s (1993) well-established scale suggests that the results appear to be mixed regarding the importance of self-control. Some research finds that self-control does not impact predation or defiance (Piquero & Hickman, 1999), nor autonomous and predation deviance among men or women (Hickman & Piquero, 2001). Baron and Forde (2007) found that low self-control was significantly associated with minor theft but not assault or serious theft. Alternatively, Curry’s (2005) results indicate that low self-control is related to a greater likelihood of cheating on an exam, greater deviant motivation, and less constraint (but not prevocational intensity). Other research has examined individual-level attributes that are closely related to self-control, such as impulsivity, sensation-seeking, and analogous behaviors (e.g., Gottfredson & Hirschi, 1990). Impulsivity has been shown to predict actual measures of defiant deviance (Curry & Piquero, 2003) as well as vignettes of exploitation (Higgins & Lauterbach, 2004), while sensation-seeking has been associated with predation (Higgins et al., 2005). Using analogous behaviors as a close proxy measure of self-control (i.e., smoke cigarettes, suspended from job, smoke marijuana, drink alcohol, car accidents, motor vehicle violations, exceed speed limit), Piquero and Piquero (2006) determined that self-control was not a significant predictor for explaining corporate crime. Overall, these results suggest that self-control may play a role in some types of deviance where control imbalances are also observed; yet the effects of self-control on victimization within a control balance framework remains unexplored. Both studies predicting victimization did not examine the effects of self-control (DeLisi & Hochstetler, 2002; Piquero & Hickman, 2003) given that they were published prior to Tittle’s (2004) theoretical reconceptualization, which emphasized the importance of self-control. However, drawing from the recent line of research that extends the general theory of crime to explain victimization (see Schreck’s 1999 landmark study) indicates support (Franklin, 2011; Holtfreter, Reisig, & Pratt, 2008; Reisig, Pratt, & Holtfreter, 2009; Schreck, 1999; Schreck, Stewart, & Fisher, 2006; Schreck, Wright, & Miller, 2002; Stewart, Elifson, & Sterk, 2004) or at least partial support for the relationship between low self-control and victimization (Fox et al., 2009; Kerley, Hochstetler, & Copes, 2009; Kerley, Xu, & Sirisunyaluck, 2008; Piquero, MacDonald, Dobrin, Daigle, & Cullen, 2005).

Given Tittle’s (1995) emphasis on gender differences, prior research has typically controlled for gendered effects by including gender as an independent or control variable. Although some studies indicate no significant gender differences in offending (Hickman & Piquero, 2001; Piquero & Hickman, 2003; Piquero & Piquero, 2006), others reveal that men are more likely to endorse deviant vignettes featuring assault (Baron & Forde, 2007; Higgins et al., 2005; Piquero & Hickman, 1999), serious theft (Baron & Forde, 2007), and exploiting a classmate for notes (Higgins & Lauterbach, 2004) even after accounting for the effects of control imbalances. However, Hickman and Piquero (2001) offer a unique examination of control balance on predation and autonomous deviance among men and women separately. Although no significant gender differences emerged in the full model, subsequent analyses revealed interesting differences between men and women. Among women, both control deficits and surpluses were significantly associated with unhealthy dieting (predation against oneself). Among men, both control deficits and surpluses were significantly predictive of exploiting a classmate for notes (autonomous deviance). Collectively, these findings, along with the supposition of gendered effects in Tittle’s (1995) original conceptualization of the theory, emphasize the importance of examining the influence of control balance ratios separately for men and women.

Control Balance Theory and Stalking

Theoretical Tests of Stalking

The stalking literature has been largely devoid of theoretical tests for either perpetration or victimization. The lack of theory tests may be due to the relative recency with which stalking has been termed criminal behavior (e.g., all states and the federal government implemented antistalking legislation between 1990 and 1995) or because of the methodological challenges in measuring and conducting research on stalking (Fox, Nobles, & Fisher, 2011). Among the limited literature that has tested theoretical explanations of stalking perpetration, six studies to date have focused on attachment theory featuring clinical and college student samples. Kienlen et al. (1997) laid the groundwork for applying attachment theory to stalking perpetration by providing indirect support for the theory. In her small clinical sample of stalkers, 63% experienced a change of primary caregiver during childhood, and many were socially withdrawn with behavioral problems, which Kienlen et al. (1997) interpreted as indicative of insecure attachment in adulthood. Tonin (2004) also found a link between stalking and insecure attachment given that higher levels of insecure attachment emerged among the ex-stalkers (n = 21) compared to the other offenders (n = 24) and nonoffenders (n = 33).

The results from these clinical samples have also been confirmed in larger samples of college students. Davis et al. (2000) and Dye and Davis (2003) examined over 300 undergraduates and determined that insecure-anxious attachment was indirectly associated with stalking perpetration. Similarly, Lewis et al. (2001) found that among the 240 college students sampled, stalking perpetration was associated with insecure attachment. Examining a large sample of college students (n = 2,783), Patton et al. (2010) confirmed that stalking perpetrators were significantly more likely to exhibit insecure-anxious attachment (not insecure-avoidant attachment).

The only published theoretical test of stalking perpetration that did not focus on attachment theory was conducted by Fox, Nobles, and Akers (2011) and consisted of a test of social learning theory among 2,766 university students. Stalking perpetrators were significantly more likely to have a friend who also committed stalking (e.g., differential peer association), more likely to report that they would react less negatively to friends who stalked (e.g., definitions), and more likely to believe that stalking was appropriate in some situations (e.g., neutralizing definitions).

Theoretical tests of stalking victimization are also scarce and have been limited to examining routine activity theory, self-control theory, and social learning theory among samples of college students. Studies examining the effects of routine activity theory have found support for its ability to explain stalking victimization. Mustaine and Tewksbury (1999) were the first to test the impact of a series of lifestyle and routine activities behaviors on stalking victimization among 861 university women. Among other noteworthy findings, these scholars found that many important lifestyle variables enhance women’s stalking victimization risk, including public behaviors (e.g., shopping), alcohol and drug use (e.g., being drunk in public, using marijuana, and other drug use), and self-protective behaviors (e.g., owning a gun, carrying mace, and carrying a pocket knife). In a rare national sample of 4,446 college women, Fisher et al. (2002) conducted an empirical test of routine activity theory of stalking victimization. Although proximity to motivated offenders was not a significant predictor for victimization, target attractiveness (e.g., class standing and victims engaged in some dating or in committed relationships) and lack of guardianship (e.g., living alone) were significant risk factors for being stalked (Fisher et al., 2002). Extending routine activities to explain cyberstalking victimization (e.g., stalking via cyberspace tactics instead of face-to-face interactions), Reyns et al. (2011) find support for the theory given that routine activities (online risk, online proximity to motivated offenders, lack of online guardians, online target attractiveness, and online deviance) predicted cyberstalking victimization. Among a sample of 1,490 college students, Fox, Gover, and Kaukinen (2009) found that women with low self-control were significantly more likely to be stalking victims, although self-control did not emerge as a significant predictor for victimization among men. Similar to their findings focused on stalking perpetration, Fox, Nobles, and Fisher (2011) also found support for social learning theory to explain stalking victimization. Stalking victims were more likely to have a friend who had been victimized by stalking (i.e., differential association/modeling), to believe that their friends would react with less sympathy (i.e., differential peer reinforcement), and to believe that stalking was sometimes acceptable (balance of definitions favorable vs. unfavorable).

Although a handful of theoretical explanations have garnered preliminary support for explaining stalking perpetration (e.g., attachment theory and social learning theory) and stalking victimization (e.g., routine activities theory, self-control theory, and social learning theory), very little scholarly attention has been devoted to testing other theoretical explanations for stalking. The following section describes the importance for the first empirical test for extending control balance theory to explain stalking perpetration and victimization.

Understanding Stalking in a Control Balance Framework

Although control balance theory’s capacity to explain stalking has not yet been empirically examined, the nature of stalking naturally lends itself to this investigation. While stalking legislation varies somewhat across jurisdictions, stalking generally involves unwanted, repeated, frightening, and/or threatening pursuit behaviors (Tjaden & Thoennes, 1998), many of which are intended by the perpetrator to preserve or extend control over the victim. Measuring stalking continues to be a challenging task for researchers due to, in part, the variety of behaviors that can be classified as stalking. For example, many legal behaviors can be considered stalking under certain conditions, including phone calling, showing up at places uninvited, gift giving, and leaving notes or messages (Fox, Nobles, & Fisher, 2011). Arguably, a series of behaviors crosses the conceptual and legal thresholds from merely irritating to “officially” constituting stalking at the point when the victim (or a “reasonable person” according to many state statutes) perceives danger or another ulterior, harmful motive. In this way, stalking can be viewed as a nonrandom event that carries important connotations relevant to the expression of control balance ratios.

Proximity in interpersonal relationships plays a critical role in understanding stalking, and this dynamic is especially important to consider in light of the control balance framework. Among a national sample of college women, four of five victims knew their stalker (Fisher et al., 2002), and among a national sample of adults, only 10% identified their stalker as a stranger (Baum, Catalano, Rand, & Rose, 2009). Not only are most stalking victims and perpetrators known to each other, but much stalking also occurs within the context of intimate relationships (Bjerregaard, 2000; Fisher et al., 2002; Fremouw, Westrup, & Pennypacker, 1997). Importantly, a small number of studies have begun to identify theoretical similarities between stalking and other interpersonal crimes such as intimate partner violence and sexual assault (Finney, 2006; Fisher, Cullen, & Turner, 2000; Walby & Allen, 2004), which have been characterized primarily as crimes motivated by control (Tjaden & Thoennes, 2000). Interestingly, when stalking victims were asked why they were targeted, nearly 33% reported that the perpetrator wanted to control them and nearly 37% indicated the perpetrator stalked due to retaliation, anger, or spite (Baum et al., 2009).

Drawing from this line of research, we contend that stalking is characterized primarily by an element of control. Advancing this concept further to explicitly incorporate tenets of control balance theory, we also anticipate that imbalances in control are likely to be associated with stalking perpetration and victimization. These imbalances may be manifested globally, as Tittle (1995) suggests, or they may be localized to perceptions of control subjected and control exerted in the context of an intimate relationship. Considering the nature of stalking perpetration, it is plausible that individuals with a perceived control deficit (e.g., less control exerted than subjected) may stalk in an effort to regain control over any area of their lives, even without an explicit motive related to an actual or perceived romantic attachment. Separately, it is also plausible that individuals with a perceived control surplus (e.g., greater control exerted than subjected) may stalk in order to maintain or extend dominance, especially if that individual uses stalking to compensate for real or imagined inadequacies across other domains. Both conceptualizations of the control ratio with respect to stalking are inherently consistent with the key tenets of control balance theory but would likely require different operationalizations and empirical tests in order to establish support.

Although stalking is not limited to intimate relationships, a maladaptive desire to initiate or maintain an intimate relationship provides one relatively common condition under which stalking may occur. Furthermore, the unique context of a potential or existing intimate relationship may further magnify perceived control imbalances because of the intensity and emotional impact of those interactions as well as the stakes involved (e.g., potential financial, property, social, and other consequences). An imbalance in either the deficit or surplus direction could result in stalking. For example, an individual with a control surplus may stalk a victim because of a perceived threat to existing control, or an individual who perceives a control deficit after being spurned in an unsuccessful intimate partnership could escalate to stalking to attempt to gain leverage over that relationship. Individuals with control surpluses might also be predisposed to stalking nonintimates in order to maintain or extend influence in other environments (e.g., social networks, the workplace).

Using Piquero and Hickman’s (2003) landmark study as a backdrop for explaining how control balance theory can explain victimization, we also argue that the theory can appropriately be extended to explain stalking victimization. According to Piquero and Hickman (2003), people with control deficits lack the skills, confidence, and self-protection necessary for avoiding victimization; alternatively, those with control surpluses exhibit feelings of invincibility, and they are more likely to place themselves in increasingly riskier situations that may lead to victimization. Along these lines, someone who conveys weakness, submissiveness, and vulnerability (i.e., control deficit) may inadvertently appear attractive to a stalking perpetrator, whereas someone who exhibits an aura of control and superiority may inadvertently embolden a stalking perpetrator to target them in an effort to surpass their level of exerted control.

Current Study

This study uniquely advances the growing line of research on control balance theory, providing one of the few available tests featuring behavioral measures within a large and representative data set that was sampled from a diverse college population. Although a college student sample may be viewed as less than ideal in some ways, Tittle (1995) reaffirms that data from any population can be used, including college students (C. R. Tittle, personal communication, November 4, 2011). Moreover, the majority of published tests of control balance theory feature student data representing vignette responses or behavioral measures. Finally, Payne and Chappell (2008, p. 186) extensively discuss the applicability of college students in criminological research, concluding that “theory tests, tests of scales, and assessments of student attitudes and behaviors” are particularly appropriate for student samples.

Using a modification of the traditional control ratio employed by Piquero and Hickman (1999) and others, we advocate for methodological changes in control balance tests featuring a recentered ratio that uses zero as its balance point, thus simplifying the interpretation of ratio values and obviating the analytical confusion over whether balanced individuals should be included in the deficit versus surplus groups. This study is also unique in its focus on stalking, in that very few theoretical tests with mainstream criminological theories are available in the peer-reviewed stalking literature. It is one of the few published studies to evaluate control balance theory with respect to victimization, and among the first to conduct comparative analyses of gender effects, both of which have roots in Tittle’s (1995) original theoretical formulation.

Given the current states of knowledge on stalking as well as control balance theory, we offer three theoretically driven research questions for the present study. Consistent with Tittle’s (2004) reconceptualization that collapses the original categories of repressive and autonomous deviance, we anticipate that control imbalances (surplus or deficit) will explain stalking behaviors. More specifically, we hypothesize the following: (1) stalking perpetration will be empirically associated with control balance imbalances, net of controls; (2) stalking victimization will be empirically associated with control imbalances, net of controls; and (3) the nature of the associations between control balance and stalking outcomes will differ by gender and operationalizations of the control balance ratio.

Method

Data were collected using a random sample of 15,000 college students from the population of 50,701 students attending a large southeastern university. Individuals were e-mailed an invitation to participate in a web-based survey regarding crime and violence and were also e-mailed a reminder notice approximately 2 weeks later, similar to the procedure for follow-up contact recommended by Dillman (2000). A total of 2,783 students elected to participate, resulting in a response rate of approximately 19%. Although the response rate was relatively low, the percentage is consistent with other Internet-based studies, particularly those that do not incorporate respondent incentives or compensation (Couper, 2000; Kaplowitz, Hadlock, & Levine, 2004; Ngo & Paternoster, 2011; Sheehan & McMillan, 1999). Moreover, several recent meta-analytic studies have identified typically lower response rates in surveys that are administered via the web, although explanations for this trend are not yet evident (Couper, 2011). More importantly, according to Dillman (1991) and others, low survey response rate is not necessarily indicative of measurement error when nonrespondents resemble respondents on variables of interest. Despite the lower response rate, the sample demographics were similar to the overall population. For example, the sample and population reflected approximately the same distributional properties for gender (58% of survey respondents were female compared with 53% of the population) and race (76% of the sample were White compared with 65% of the population). Furthermore, regardless of the survey response rate, the external validity of the findings from this study is supported by similarities in the estimates of stalking observed in this sample compared to other published stalking research based on college samples (Fremouw et al., 1997; Haugaard & Seri, 2003).

Measures

Dependent Variables

Consistent with prior research (Fox, Nobles, & Fisher, 2011; Tjaden & Thoennes, 1998), stalking was defined as harassing, intrusive, frightening or threatening, and unwanted behavior that occurred on more than one occasion (excluding communication from bill collectors, solicitors, etc.). Survey participants responded to a series of questions measuring stalking victimization and then responded to an identical set of items assessing stalking perpetration. Given that many stalking victims may not consider their experiences to constitute stalking (Nobles et al., in press), we employed precautions to avoid respondents’ subjective interpretations of stalking. First, respondents were specifically informed that stalking behavior is defined as “harassing (not including bill collectors), frightening and threatening, intrusive and unwanted, happening on MORE THAN ONE occasion, and involving a single perpetrator.” Next, the instructions informed respondents: “The following questions ask you about stalking behavior. You may have experienced stalking from multiple people. Please respond to the following questions about ONLY ONE stalker. You will have the opportunity to respond to the same questions about different stalkers on the next page.” Therefore, the measure of stalking required that the behavior occur multiple times by the same perpetrator. Although respondents were given the opportunity to report multiple stalking victimization or perpetration episodes, the current study was concerned with whether any stalking had occurred and did not specifically address repeat victimization/offending.

Nine of the 11 items were adapted from the National Violence Against Women Survey (Tjaden & Thoennes, 1998), and two supplemental items were included to measure cyber-stalking (e.g., unwanted electronic messages and unwanted messages/pictures posted to Internet websites). Respondents were asked to respond to the 11 items in order to assess stalking victimization (“Has the SAME PERSON ever…”) and stalking perpetration (“Have you ever done any of the following to the SAME PERSON…”): (1) followed, watched, or spied on; (2) stood outside home, school, or workplace; (3) showed up at a place uninvited; (4) vandalized property or destroyed something; (5) sent unsolicited letters, written correspondence, or unwanted e-mails; (6) made unwanted phone calls; (7) left unwanted messages; (8) left unwanted items; (9) tried to communicate in other ways; (10) sent unwanted messages electronically; and (11) posted unwanted messages/pictures to Internet websites. Response options for each of the items (for perpetration and victimization) were never (0), once (1), and more than once (2). Responses from the 11 items were summed and dichotomized separately for perpetration and victimization such that a summed score of two or more indicated stalking (meaning that one behavior occurred more than once or at least two behaviors occurred), which is consistent with prior literature suggesting that stalking behaviors must occur on two or more occasions (see Fox, Nobles, & Fisher, 2011, for a review). Cronbach’s alphas indicated high reliability for both stalking victimization (0.893) and perpetration (0.861).

Independent Variables

Control balance ratio

Consistent with several published tests, control was assessed on a multi-item scale covering different domains.² Survey respondents indicated their level of control exerted (e.g., “ How much control do you think YOU are able to exercise over each of the following items?”) for 13 areas: friendships (overall), specific friendships, relationships with significant others, performance in school, job/employment, familial relationships, pets, recreational activities, society as a whole, children, the physical environment, the respondent’s own physical body, and “other people” including neighbors or acquaintances not captured elsewhere. Consistent with much of the control balance literature (Hickman et al., 2001; Higgins & Lauterbach, 2004; Piquero, MacIntosh, & Hickman, 2001; Piquero & Piquero, 2006), response options included 1 (“no control”), 2 (“some control”), 3 (“lots of control”), and 4 (“complete control”). These items were then summed to represent global control exerted. The process was repeated with similar items for control to which an individual was subjected (e.g., “How much control do you think each of the following items are able to exercise OVER you?”). These items were then summed to represent global control subjected. Exploratory factor analysis for both subscales supported a single-factor solution in both cases, and Cronbach’s alpha values were satisfactory for both subscales (CE = 0.817; CS = 0.797). Control exerted was divided by control subjected (CE/CS) to create a global control balance ratio, where values of 1 represented balance, values less than 1 indicated control deficit, and values greater than 1 represented control surplus.

A second control balance ratio was constructed to represent relationship-specific control balance. For this measure, a single item assessing control exerted in “relationships with significant others” was divided by a similar item assessing control subjected in the same domain to create a relationship control balance ratio. As before, values of 1 represented balance, values less than 1 indicated control deficit, and values greater than 1 represented control surplus. Because the relationship-specific control balance ratio was premised on a single domain, factor structure and alpha consistency values were superfluous. Additionally, the global and relationship control balance ratios were only moderately correlated (r = 0.525), suggesting that separate analyses may be fruitful since the two measures may be tapping different underlying constructs.

Both of the control balance ratio operationalizations were assessed for distributional properties after their initial construction. Modifying the ratio to recenter balance at zero offers advantages for model estimation (e.g., fitted values better approximated Tittle’s [1995] conception of a curvilinear model around balance) as well as intuitive appeal in expressing control deficits as negative values. Recentering was accomplished by subtracting 1 from the original calculated balance ratios. This procedure recentered balanced individuals at zero on the scale and resulted in control deficits being expressed as negative fractional values. Also, ratio values on the deficit side were transformed by taking their multiplicative inverse (1/x) in order to provide symmetry in the curvilinear function that Tittle (1995) describes for the continuum of control (see Figure 1). Thus, an extreme minimum control imbalance, where an individual reported “no control” exerted (a value of 1 on the scale item) and “maximum control” subjected (a value of 4 on the original scale item) originally calculated as ¼ = 0.25 would be recentered to -4 in the modified balance ratio. This value would conveniently mirror the maximum possible surplus (4/1 = 4). The functional effect of the negative inverse for control deficits is to create equivalence of scale when evaluating segmented nonlinear regression models.

Self-control

Consistent with published tests to date involving control balance theory (Baron & Forde, 2007; Curry, 2005; Hickman & Piquero, 2001; Piquero & Hickman, 1999), the 23-item scale developed by Grasmick et al. (1993) was used to capture low self-control. Respondents indicated their agreement with the statements using a 4-point Likert-type scale, and the items were combined into a self-control scale where lower values on the scale indicated lower self-control.

Morals

Most published tests of control balance theory feature at least one item that asked respondents to indicate whether the deviant acts featured in the vignettes were morally wrong (Baron & Forde, 2007; Hickman & Piquero, 2001; Higgins et al., 2005; Higgins & Lauterbach, 2004; Piquero & Hickman, 1999; Piquero & Piquero, 2006). This study used a single item capturing moral beliefs about stalking by asking respondents to rate their agreement as to the statement, “There are some situations in which stalking someone is okay,” where higher values indicated stronger moral disagreement (1 = strongly agree; 4 = strongly disagree).

Peers

Consistent with most published tests to date involving control balance theory that asked respondents how many friends had engaged in the deviant acts featured in hypothetical vignettes (Baron & Forde, 2007; Hickman & Piquero, 2001; Higgins et al., 2005; Higgins & Lauterbach, 2004; Piquero & Hickman, 1999), peer deviance was assessed with a single item that asked whether any friend had experience with stalking. Two versions of the question were alternately tailored to perpetration (“As far as you know, do any of your friends engage in stalking behaviors?”) and victimization (“As far as you know, are any of your friends victims of stalking?”), where higher values indicated greater peer influence (1 = none; 4 = most or all), and the appropriate version of the question was used to model stalking perpetration versus victimization outcomes.³

Control variables

Given that Tittle (1995, p. 150) considers age (continuous), gender (0 = male, 1 = female), race (0 = White, 1 = non-White), and ethnicity (0 = non-Hispanic, 1 = Hispanic) to be “crucial demographic variables” necessary for testing control balance theory, we include all of these variables in our analysis. Gender was alternately featured as a control variable in initial multivariate models featuring the full sample to establish overall gender differences; then, the models were split by gender to facilitate male versus female comparisons.

Analysis

Consistent with prior research, this analysis features segmented nonlinear regression to model the effects of control balance on stalking perpetration and victimization (Baron & Forde, 2007; Curry & Piquero, 2003; Hickman & Piquero, 2001; Hickman et al., 2001; Higgins & Lauterbach, 2004; Higgins et al., 2005; Piquero & Hickman, 1999, 2003; Piquero & Piquero, 2006). See Neter, Wasserman, and Kutner (1985) for a detailed discussion of segmented regression. The general form of the models was as follows:

Y = gender * µ_{1} + age * µ_{2} + race * µ_{3} + ethnicity * µ_{4} + morals * µ_{5} + peers * µ_{6} + self - control * µ_{7} + (CBR deficit) * µ_{8} + (CBR surplus) * µ_{9} + constant,

where Y alternately represents the two dependent variables for stalking perpetration or victimization, gender indicates respondent gender, age indicates respondent age in years, race indicates respondent race, ethnicity indicates respondent ethnicity, morals indicates respondent moral belief about stalking, peers indicates respondent peer stalking experience specific to perpetration or victimization, and self-control indicates respondent mean score on the Grasmick et al. (1993) scale. Estimation of the control balance deficit and surplus segments was conditional upon the respondent’s recentered control balance ratio (CBR), such that deficit = CBR < 0 and surplus = CBR > 0; see above for explication of ratio recentering). We first estimated models using the full sample. Given Tittle’s (1995) predictions about gender differences outlined earlier, in addition to the significant gender differences that emerged in each of the following full models (discussed below), the models were then split by gender to assess comparative effects for men and women; in the gender split models, the gender term was subsequently removed.⁴ Cases with missing values were removed from the analyses automatically via listwise deletion.

Results

Univariate properties and bivariate relationships were examined first, with particular emphasis on gender differences. Similar to the university population, there were more women (n = 1,618, 58.50%) than men (n = 1,148, 41.50%) in the sample. The overall prevalence of stalking perpetration in the sample was 5.79% (n = 161), while the overall stalking victimization rate was 25.87% (n = 720). Despite the differences in the prevalence rates (e.g., 5.79% vs. 25.87%), these rates are consistent with prior research on stalking victimization (Fremouw et al., 1997) and perpetration (Haugaard & Seri, 2003) among college samples. The discrepancy between victimization and perpetration prevalence rates may be due to the possibility that college students are victimized by noncollege student stalkers, college student stalkers are targeting noncollege students, or that one stalker could have multiple victims. Women in the sample were significantly more likely than men to be stalking perpetrators (χ² = 4.206, df = 1, p < .05) and stalking victims (χ² = 69.922, df = 1, p < .001). Men in the sample had significantly higher mean scores on the global control balance ratio (t = 2.683, df = 2,003, p < 0.01) and had higher mean age than women (t = 4.016, df = 2,717, p < .001). Women had significantly higher scores on the Grasmick et al. scale representing self-control (t = 5.673, df = 1,884, p < .001) and were more likely to be of Hispanic ethnicity compared to men in the sample (χ² = 10.748, df = 1, p < .001). There were also significant gender differences for the item representing moral beliefs about stalking, with observed counts lower than expected for men and higher than expected for women in the “strongly disagree” category, suggesting polarization of attitudes across gender (χ² = 36.922, df = 3, p < .001). There were no statistically significant differences between men and women in the sample with respect to the relationship-specific control balance ratio, peer stalking perpetration or victimization, or race. Descriptive statistics are presented in Table 1.

Table 1:

Descriptive Statistics for Full Sample and Split by Gender

	Full Sample (n = 2,783)			Men (n = 1,148)			Women (n = 1,618)
Variable (Sig. Gender Difference)	M/% (n)	SD	Range	M/% (n)	SD	Range	M/% (n)	SD	Range
Dependent variables
Stalking perpetration*	5.79% (161)	0.234	0 to 1	4.70% (54)	0.212	0 to 1	6.55% (106)	0.248	0 to 1
Stalking victimization***	25.87% (720)	0.438	0 to 1	17.68% (203)	0.382	0 to 1	31.83% (515)	0.466	0 to 1
Independent and control variables
Overall CB ratio**	0.153	0.300	-1.462 to 3.00	0.175	0.313	-1.00 to 3.00	0.139	0.290	-1.462 to 3.00
Relationship CB ratio	0.030	0.549	-3.00 to 3.00	0.017	0.577	-3.00 to 3.00	0.039	0.531	-3.00 to 3.00
Self-Control Scale***	3.014	0.325	1.478 to 4.00	2.959	0.342	1.478 to 3.957	3.046	0.310	1.870 to 4.00
Morals***	3.424	0.733	1 to 4	3.294	0.771	1 to 4	3.500	0.699	1 to 4
Peer stalking perpetration	1.122	0.375	1 to 4	1.124	0.407	1 to 4	1.121	0.356	1 to 4
Peer stalking victimization	1.355	0.504	1 to 4	1.359	0.526	1 to 4	1.354	0.492	1 to 3
Demographic variables
Gender (0 = male)	58.50% (1618)	0.493	0 to 1	—	—	—	—	—	—
Age***	22.834	5.996	18 to 62	23.385	6.145	18 to 61	22.450	5.869	18 to 62
Race (0 = White)	23.53% (646)	0.424	0 to 1	23.48% (267)	0.424	0 to 1	23.58% (378)	0.425	0 to 1
Ethnicity (0 = non-Hispanic)***	12.63% (337)	0.332	0 to 1	10.14% (112)	0.302	0 to 1	14.43% (225)	0.352	0 to 1

Note. CB = Control Balance.

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

Multivariate analysis permits closer examination of the role of control balance as well as the hypothesized gender differences. Table 2 contains parameter estimates for segmented nonlinear regression models with the overall control balance ratio predicting stalking perpetration for the full sample, men only, and women only. Gender was positive and significant in the full model (b = 0.041, p < .001), indicating that stalking perpetrators were more likely to be women, after controlling for other factors. Of the included demographics, only ethnicity was significantly associated with stalking perpetration across all models, indicating that Hispanics were more likely to stalk. Moral beliefs about stalking and peer stalking perpetration were consistently significant across the three models. Self-control was negatively and significantly associated with stalking perpetration for the full sample and for men, meaning that those with low self-control were more likely to stalk compared to those with high self-control. Critically, although the control balance ratio terms were not significantly related to stalking perpetration in the full sample, the surplus term was positively and significantly associated with stalking perpetration for men (b = 0.099, p < 0.01), indicating that men with more severe control imbalances in the surplus direction were significantly more likely to be stalking perpetrators compared to women. Explained variance across the three models was relatively consistent, suggesting that this model was appropriately specified and that the independent terms function similarly for the full and gender split samples.

Table 2:

Segmented Nonlinear Regression Models With Overall Control Balance Ratio Predicting Stalking Perpetration

	Full Sample (n = 1,631)	Men (n = 599)	Women (n = 1,032)
Variable	b (SE)	b (SE)	b (SE)
Gender	0.041 (0.014)***	—	—
Age	0.001 (0.001)	0.001 (0.002)	0.001 (0.001)
Race	-0.015 (0.016)	-0.034 (0.026)	-0.001 (0.021)
Ethnicity	0.053 (0.020)**	0.056 (0.032)*	0.050 (0.025)*
Morals	-0.085 (0.009)***	-0.068 (0.014)***	-0.095 (0.013)***
Peer stalking perpetration	0.075 (0.018)***	0.051 (0.026)*	0.090 (0.024)***
Self-Control Scale	-0.064 (0.021)**	-0.078 (0.031)**	-0.052 (0.028)
Overall CB ratio (deficit)	-0.118 (0.070)	-0.107 (0.121)	-0.127 (0.087)
Overall CB ratio (surplus)	0.034 (0.024)	0.099 (0.035)**	-0.014 (0.033)
Constant	0.367 (0.073)***	0.419 (0.103)***	0.433 (0.099)***

Model residual SS	106.612	33.831	72.019
Model R²	0.096	0.094	0.105

Note. CB = Control Balance; SS = Sum of Squares.

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

Table 3 details parameter estimates for segmented nonlinear regression models with the overall control balance ratio predicting stalking victimization for the full sample, men only, and women only. Gender was positive and significant in the full model (b = 0.174, p < .001), indicating that stalking victims were more likely to be women, after controlling for other factors. Of the included demographics, only age was significantly related to stalking victimization for the full sample and for men, such that older respondents were more likely to be stalking victims than younger respondents. Moral beliefs about stalking were significantly predictive of stalking victimization for the full sample and for women. Peer stalking victimization was consistently significant across the three models. Self-control was negatively and significantly associated with stalking victimization for men only, meaning that those with low self-control were more likely to be victims compared to those with high self-control. In contrast with the analogous models featuring stalking perpetration, neither of the control balance ratio terms were significantly associated with stalking victimization for any of the three models. Model explained variance was consistent for the full sample and for women, but was slightly lower for men.

Table 3:

Segmented Nonlinear Regression Models With Overall Control Balance Ratio Predicting Stalking Victimization

	Full Sample (n = 1,630)	Men (n = 598)	Women (n = 1,032)
Variable	b (SE)	b (SE)	b (SE)
Gender	0.174 (0.023)***	—	—
Age	0.006 (0.002)***	0.009 (0.003)***	0.003 (0.002)
Race	-0.044 (0.027)	-0.048 (0.042)	-0.045 (0.036)
Ethnicity	0.027 (0.033)	0.044 (0.053)	0.027 (0.042)
Morals	-0.043 (0.015)**	-0.023 (0.022)	-0.058 (0.021)**
Peer stalking victimization	0.239 (0.022)***	0.133 (0.031)***	0.310 (0.029)***
Self-Control Scale	-0.065 (0.035)	-0.125 (0.051)**	-0.027 (0.048)
Overall CB ratio (deficit)	-0.189 (0.118)	-0.369 (0.196)	-0.107 (0.148)
Overall CB ratio (surplus)	0.009 (0.041)	0.101 (0.057)	-0.058 (0.057)
Constant	-0.099 (0.119)	0.238 (0.166)	0.160 (0.160)

Model residual SS	306.227	91.400	209.914
Model R²	0.111	0.070	0.116

Note. CB = Control Balance; SS = Sum of Squares

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

Similar models were also estimated using the relationship-specific control balance ratio instead of the global ratio in order to assess the potential utility of a localized operationalization of this theoretical predictor. Parameter estimates are presented in Table 4. Results were similar to the analogous models featuring global control balance with respect to the control variables. As with the previous models, gender was positive and significant for the full sample model (b = 0.038, p < .01), indicating that stalking perpetrators were more likely to be women, after controlling for other factors. Of the included demographics, ethnicity was significantly associated with stalking perpetration for the full sample and for women, such that Hispanics were more likely to stalk than non-Hispanics. Moral beliefs about stalking, peer stalking perpetration, and low self-control were consistently significant across the three models. The key differences in results are for the relationship control balance deficit and surplus terms. Relationship balance deficit was negatively and significantly related to stalking perpetration in the full sample (b = -0.067, p < 0.001) and for women (b = -0.093, p < .001), indicating that individuals who experience more severe control deficits were more likely to be stalking perpetrators. For men, the relationship balance surplus term was positively and significantly associated with stalking perpetration (b = 0.059, p < .05), indicating that men with more severe control imbalances in the surplus direction were significantly more likely to be stalking perpetrators. Collectively, these two main findings support the hypothesis that control imbalances in either direction will be associated with stalking perpetration, although the precise mechanism appears to differ across gender. Model explained variance was highest for women, lower for the full sample, and slightly lower still for men.

Table 4:

Segmented Nonlinear Regression Models With Relationship Control Balance Ratio Predicting Stalking Perpetration

	Full Sample (n = 1,732)	Men (n = 644)	Women (n = 1,088)
Variable	b (SE)	b (SE)	b (SE)
Gender	0.038 (0.013)**	—	—
Age	0.001 (0.001)	0.000 (0.002)	0.001 (0.001)
Race	-0.009 (0.016)	-0.011 (0.025)	-0.005 (0.020)
Ethnicity	0.050 (0.019)**	0.037 (0.031)	0.055 (0.024)*
Morals	-0.083 (0.009)***	-0.063 (0.014)***	-0.094 (0.012)***
Peer stalking perpetration	0.088 (0.017)***	0.067 (0.026)**	0.103 (0.023)***
Self-Control Scale	-0.060 (0.020)**	-0.062 (0.030)*	-0.057 (0.027)*
Relationship CB ratio (deficit)	-0.067 (0.018)***	-0.031 (0.027)	-0.093 (0.024)***
Relationship CB ratio (surplus)	0.027 (0.017)	0.059 (0.026)*	0.011 (0.022)
Constant	0.340 (0.071)***	0.367 (0.102)***	0.419 (0.095)***

Model residual SS	113.625	38.328	74.525
Model R²	0.105	0.084	0.124

Note. CB = Control Balance; SS = Sum of Squares.

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

Table 5 contains parameter estimates for segmented nonlinear regression models with the relationship control balance ratio predicting stalking victimization for the full sample, men only, and women only. As before, substantive results with respect to the included controls were identical to the models presented in Table 3 (e.g., gender, age, moral beliefs, peer stalking victimization, and low self-control were all significantly related to stalking victimization). However, key differences emerged with the relationship-specific control deficit term. Specifically, relationship control deficit was negatively and significantly related to stalking victimization for women (b = -0.082, p < .05), indicating that more severe relationship control deficits were significantly associated with stalking victimization and that these influences were unique to women. Model explained variance was approximately equivalent for the full sample and for women and relatively lower for men.

Table 5:

Segmented Nonlinear Regression Models With Relationship Control Balance Ratio Predicting Stalking Victimization

	Full Sample (n = 1,731)	Men (n = 643)	Women (n = 1,088)
Variable	b (SE)	b (SE)	b (SE)
Gender	0.176 (0.022)***	—	—
Age	0.006 (0.002)***	0.007 (0.002)**	0.005 (0.002)*
Race	-0.028 (0.026)	-0.014 (0.040)	-0.033 (0.034)
Ethnicity	0.015 (0.032)	0.026 (0.051)	0.017 (0.042)
Morals	-0.043 (0.015)**	-0.025 (0.021)	-0.056 (0.020)**
Peer stalking victimization	0.245 (0.021)***	0.147 (0.030)***	0.310 (0.028)***
Self-Control Scale	-0.065 (0.034)	-0.112 (0.048)*	-0.034 (0.047)
Relationship CB ratio (deficit)	-0.051 (0.030)	-0.004 (0.043)	-0.082 (0.041)*
Relationship CB ratio (surplus)	0.001 (0.028)	0.022 (0.041)	-0.016 (0.037)
Constant	-0.118 (0.115)	0.241 (0.160)	0.127 (0.154)

Model residual SS	322.635	98.612	220.400
Model R²	0.114	0.060	0.118

Note. CB = Control Balance; SS = Sum of Squares

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

Discussion

Theoretical and Methodological Contributions

This study was concerned with two primary objectives: first, to provide an empirical test and methodological extension to the limited body of literature on control balance theory; and second, to apply a theoretical framework and corresponding empirical test to the limited body of literature on stalking. To the extent that both of these areas are understudied in criminology, our study offers several specific contributions for control balance theory as well as substantive knowledge on stalking.

Results from this study suggest partial support for Tittle’s (1995, 2004) control balance theory after controlling for factors such as morals, peer influence, self-control, and demographics. Using properly specified segmented regression models, we demonstrated that perceived control imbalance, represented by the individual-level control balance ratio, is a consistent predictor of stalking perpetration for men. Additionally, a relationship-specific operationalization of the control balance ratio was associated with stalking perpetration and victimization for women. This finding is especially notable given that only two published studies to date have examined the relationship between control balance and victimization outcomes (DeLisi & Hochstetler, 2002; Piquero & Hickman, 2003). Although Tittle (1995) suggests that women’s involvement in deviant acts will be disproportionately predicted by control imbalance (see Figure 2), other published work illustrates that female involvement in stalking represents a special case of gender inequity in which women are overrepresented as both victims and perpetrators (see Nobles, Fox, Piquero, & Piquero, 2009, for discussion).

The traditional measures used to operationalize control imbalance, the ratio itself, has been adjusted here to enhance its interpretability and applications. Using a simple mathematical transformation, we recentered the ratio to use zero as the balance point. This approach offers several benefits, including the elimination of conceptual confusion over whether “balanced” individuals should be categorized with those perceiving deficits versus surpluses or whether they should be excluded from the analysis entirely. Since empirical testing of the key tenet of control balance theory requires a contrast between individuals on either side of the balance point, recentering the ratio at zero mathematically nullifies the predictive influence of balanced individuals relative to other cases that contribute to model estimation. Additionally, the recentered control ratio was transformed to extend the range of possible values on the control deficit side of the equation, resulting in symmetry that more closely approximates the parabolic function that Tittle (1995) and others have described for the control continuum. Graphing fitted values of the dependent variables against the original, and recentered control balance ratios also results in more symmetrical relationships, with minimum values for stalking at or near the balance point in most cases.⁵

A second methodological contribution from this study was the exploration of context-specific operationalizations of the control balance ratio. Tittle (1995) suggests that control balance theory is a general theory in the sense that it can be used to explain deviant as well as nondeviant behavior, and various scholars have built upon this basic premise to measure perceived control using multi-item scales assessing control exerted and control subjected over many domains. Our approach to measurement is consistent with prior research, although we compared the full ratio with a subset of the perceived control scale used to a calculate control balance for a single domain only. The principal rationale for this choice is that stalking, in contrast to many other dependent variables capturing deviance in other published tests, is arguably more targeted to specific interpersonal relationship dynamics. When considering the applicability of control balance theory to broad motivations for deviance such as the acquisition of money or property, many factors could be reasonably hypothesized to influence crimes such as larceny, fraud, or embezzlement. Stalking, however, has been referred to as an interpersonal crime of power and control that is frequently presumed to be linked to maladaptive or dysfunctional adult attachment, socialization, or other mechanisms.

The role of gender has received some attention in the stalking literature but relatively little in terms of empirical tests of control balance theory. Many published studies simply rely upon gender as a control variable, although there are theoretical reasons to explore this relationship in greater detail. Beginning with Tittle’s (1995) discussion of gendered control imbalances, control has been postulated to influence men and women differently, and this study supports the argument that there are gender differences in the ways in which control imbalances relate to stalking outcomes for men and women. Specifically, there is a relationship between control surpluses and stalking perpetration for men, using both operationalizations of the control ratio. There is also a relationship between control deficits and both stalking perpetration and victimization for women, using the relationship-specific control ratio only. Importantly, these findings would have been obscured if gender were relegated to the status of a single control variable in the full-sample models. In general, this may indicate the relative strength of interpersonal relationships in influencing social contexts, decision making, or behavior for women compared to men. It also supports a more nuanced and less general reading of control balance theory, to the extent that perceptions capturing control imbalances over all domains may be less appropriate when empirically testing control balance for certain crime types.

More generally, findings regarding our demographic variables (e.g., ethnic and gender differences) warrant discussion. This study found that Hispanics were significantly more likely than non-Hispanics to perpetrate stalking. It is difficult to put this finding into context given that prior research on stalking largely overlooks the potential importance of ethnicity on stalking victimization and perpetration. In fact, race and ethnicity are not even included as control variables in many studies of stalking. Many factors, most of which remain unexplored, could account for these differences, including variations in social support and willingness to report stalking across racial and ethnic groups. However, at present these potential explanations are largely speculative. Therefore, we encourage future research to advance our understanding of how ethnicity (and race) impact stalking.

Consistent with prior research, we found that women were significantly more likely than men to be stalking victims (McCreedy & Dennis, 1996), and contrary to expectations, women were also more likely than men to admit perpetrating stalking. While some prior research indicates men are more likely to be stalking perpetrators (Bjerregaard, 2000; Logan, Leukefeld, & Walker, 2000), other research finds no significant differences among men and women (Dye & Davis, 2003; Langhinrichsen-Rohling, Palarea, Cohen, & Rohling, 2000). Although our data do not allow for a more thorough investigation into the causes of these gender differences, it is possible that our unique research design (using a web-based survey) may have made women feel more comfortable admitting to stalking perpetration. At its core, stalking is pursuit behavior that has elements of aggression, persistence, and control, and these characteristics are consistent with traditional masculinity. Despite the many advances in terms of women’s rights and equality, traditional gender roles are still pervasive and these gendered expectations may create social acceptability for men to admit committing such “masculine” pursuit behaviors. However, women who admit engaging in pursuit (i.e., stalking) behaviors may be perceived as violating acceptable gender norms. In other words, it is possible that the anonymity of a web-based survey removed women’s worry about the social desirability of their responses compared to the context of answering an in-person survey (e.g., classmates seated near the respondent may easily view survey responses). The use of web-based surveys is a relatively new and underused method for collecting data on stalking (and control balance), and future research that examines differences between online and in-person responses will be important contributions for future research (see Wells, Cavanaugh, Bouffard, & Nobles, 2012).

Limitations and Implications For Future Research

Although the current study has important theoretical implications, a discussion of the limitations warrants consideration. First, the findings with respect to control ratios and stalking outcomes are correlational in nature, given the cross-sectional nature of the data. A key question for stalking researchers (and a challenge for studying many types of interpersonal crimes, especially with victimization outcomes) is whether perceived imbalances in control occurred before or after the onset of the first stalking episode. It is plausible that control deficits and stalking victimization, for example, could be reversed in time-order such that victimization results in less perceived control exerted, more perceived control subjected, or both. However, the opposite scenario is also plausible, in which an individual with less perceived control for whatever reason might be at heightened risk for subsequent exploitation. While this is certainly a limitation, it is important to note that all other empirical tests of control balance theory also rely upon cross-sectional data. Future research that employs the use of longitudinal data testing these relationships would enhance our theoretical understanding; however, in these early stages of testing control balance theory, it is equally important to examine these complex relationships using a variety of types of data. Finally, the comparatively low response rate in this study must be acknowledged as a potential limitation, possibly even a severe one. Although the study sample resembled the university population from which it was drawn in terms of available demographic properties, and the observed stalking prevalence rates were similar to other published studies, the generalizability of results must be undertaken with great caution.

The literature on stalking has overwhelmingly focused on traditional stalking behaviors that are characterized by contact between victim and offender, including many of the behaviors captured in the current study (i.e., following, watching, and showing up uninvited). However, given society’s increasing reliance on cyber-communication, scholarly attention has recently begun focusing on cyberstalking, which broadly refers to the “repeat pursuit of an individual using electronic or Internet-capable devices” (Reyns et al., 2011, p. 1153). Recent research indicates that although there are many similarities between stalking and cyberstalking victims (e.g., financial cost of victimization, fear at onset of victimization, and whether victims considered the behavior to constitute stalking), there are also significant differences between the groups (e.g., fear overtime and whether victims were threatened or attacked) (Nobles et al., in press). Although the current study included two cyberstalking items (unwanted electronic messages and posting unwanted pictures/messages to Internet websites), we are limited in our ability to comprehensively examine cyberstalking. Given the differences between stalking and cyberstalking victims (Nobles et al., in press) and society’s increasing reliance on technology to communicate, it will become increasingly important for future research to examine stalking and cyberstalking victims (and offenders) separately.

A key limitation in all of the published tests of control balance theory to date is the lack of attention to a key tenant of the theory, which Tittle (1995) refers to as control balance desirability. According to the theory, observed control imbalances are likely to result in a process of rational calculus on the part of a potential offender, whereby that individual selects from a range of possible deviant acts that is perceived to be most opportune, most effective, and involves the smallest probability of countercontrol. Stalking, therefore, is one possible outcome from a range of possible deviant acts that potential offenders could attempt as a result of a control imbalance. Although data are not available to formally establish the empirical relationships implied by control balance desirability, stalking would presumably occupy a place on the hypothetical desirability scale closer in nature to interpersonal crimes like intimate partner violence and sexual assault rather than other broad classifications of property or white-collar crimes. In fact, given the current state of knowledge on stalking perpetration and victimization, this type of crime is arguably the prototypical deviant response for a predisposed potential offender in a college student population, due to the influence exerted by dynamic social situations, alcohol consumption, technology/social media, and the likelihood for “provocation,” which in the case of stalking may involve rejection from a possible or former relationship/sexual partner. Thus, results presented here should be treated with some caution and subjected to future study.

Consistent with Tittle’s (1995) conceptualization of control, this study represents a conservative test of the theory especially if perceived control imbalances are only likely to predict some deviant outcome on a continuum of possible behaviors, of which stalking is one specific manifestation. The realization of any specific deviant behavior is a function of several interrelated factors, including opportunity and the individual’s assessment of which behaviors are most likely to result in resolution for the perceived control imbalance. To this end, the present study has illustrated that control imbalances are associated with one category of possible deviant outcomes (stalking). It is plausible that individuals perceiving control imbalances, particularly severe imbalances in the interpersonal relationship domain, could also select from several other behavioral alternatives according to individual factors and context. This element of control balance theory is ripe for future investigation.

Overall, the current study advances our understanding of control balance theory and stalking. Taken together, the results suggest that control balance theory explains stalking perpetration and that the theory can effectively be extended to explain stalking victimization. The findings also emphasize the importance of testing control balance theory separately among men and women, given the significant differences between these groups. Furthermore, the current study lends support for future research to measure control balance as an overall ratio (comprised of a variety of measures) in addition to a paired-down relationship ratio, especially for predicting outcomes that are interpersonal in nature, like stalking. The next logical step based on the current study may be to examine the extent to which control balance theory can explain other interpersonal crimes, such as intimate partner violence, harassment, bullying, and sexual assault. In closing, the current study is offered as the first test of control balance theory’s ability to explain stalking perpetration and victimization, with the hope that future research will continue to advance this line of research.

Footnotes

Acknowledgements

The authors gratefully acknowledge Charles R. Tittle, several anonymous reviewers, and the editor for feedback on previous drafts of this manuscript. Additional thanks go to Alex R. Piquero and Nicole Leeper Piquero for their assistance with scale development.

Notes

Matt R. Nobles is an assistant professor in the College of Criminal Justice at Sam Houston State University. His research interests include violent and interpersonal crimes, criminological theory, and quantitative methods. His recent work has appeared in Justice Quarterly, Crime & Delinquency, Criminal Justice & Behavior, Aggression and Violent Behavior, Journal of Quantitative Criminology, Journal of Criminal Justice, Journal of Interpersonal Violence, and the American Journal of Public Health.

Kathleen A. Fox is an assistant professor in the School of Criminology & Criminal Justice at Arizona State University. She earned her PhD in criminology, law and society from the University of Florida. Her research interests focus on the effects of crime victimization and her work examines theories of victimization, the victim-offender overlap, the gang-victimization link, and the ways in which crime changes victims’ behavior. Her work has recently appeared in Criminal Justice & Behavior, Justice Quarterly, Crime & Delinquency, Journal of Interpersonal Violence, and Aggression and Violent Behavior.

References

Baron

S. W.

Forde

D. R.

(2007). Street youth crime: A test of control balance theory. Justice Quarterly, 24, 335-355.

Baum

Catalano

Rand

Rose

(2009). Stalking victimization in the United States. Washington, DC: U.S. Department of Justice.

Bjerregaard

(2000). An empirical study of stalking victimization. Violence and Victims, 15, 389-405.

Braithwaite

(1997). Charles Tittle’s control balance and criminological theory. Theoretical Criminology, 1, 77-97.

Couper

M. P.

(2000). Web surveys: A review of issues and approaches. Public Opinion Quarterly, 64, 464-494.

Couper

M. P.

(2011). The future modes of data collection. Public Opinion Quarterly, 75, 889-908.

Curry

T. R.

(2005). Integrating motivating and constraining forces in deviance causation: A test of causal chain hypotheses in control balance theory. Deviant Behavior, 26, 571-599.

Curry

T. R.

Piquero

A. R.

(2003). Control ratios and defiant acts of deviance: Assessing additive and conditional effects with constraints and impulsivity. Sociological Perspectives, 46, 397-415.

Daly

Chesney-Lind

(1988). Feminism and criminology. Justice Quarterly, 5, 497-538.

10.

Davis

K. E.

Ace

Andra

(2000). Stalking perpetrators and psychological maltreatment of partners: Anger-jealousy, attachment insecurity, need for control, and break-up context. Violence and Victims, 15, 407-425.

11.

DeLisi

Hochstetler

A. L.

(2002). An exploratory assessment of Tittle’s control balance theory: Results from the National Youth Survey. Justice Professional, 15, 261.

12.

Dillman

D. A.

(1991). The design and administration of mail surveys. Annual Review of Sociology, 17, 225-249.

13.

Dillman

D. A.

(2000). Mail and Internet surveys: The tailored design method. New York: Wiley.

14.

Dye

M. L.

Davis

K. E.

(2003). Stalking and psychological abuse: Common factors and relationship-specific characteristics. Violence and Victims, 18, 163-180.

15.

Finney

(2006). Domestic violence, sexual assault and stalking: Findings from the 2004/2005 British Crime Survey. London: Home Office Research, Development and Statistics Directorate.

16.

Fisher

B. S.

Cullen

F. T.

Turner

M. G.

(2000). The sexual victimization of college women. Washington, DC: U.S. Department of Justice.

17.

Fisher

B. S.

Cullen

F. T.

Turner

M. G.

(2002). Being pursued: Stalking victimization in a national study of college women. Criminology & Public Policy, 1, 257-308.

18.

Fox

K. A.

Gover

A. R.

Kaukinen

(2009). The effects of low self-control and childhood maltreatment on stalking victimization among men and women. American Journal of Criminal Justice, 34, 181-197.

19.

Fox

K. A.

Nobles

M. R.

Akers

R. L.

(2011). Is stalking a learned phenomenon? An empirical test of social learning theory. Journal of Criminal Justice, 39, 39-47.

20.

Fox

K. A.

Nobles

M. R.

Fisher

B. S.

(2011). Method behind the madness: An examination of stalking measurements. Aggression and Violent Behavior, 16, 74-84.

21.

Franklin

C. A.

(2011). An investigation of the relationship between self-control and alcohol-induced sexual assault victimization. Criminal Justice and Behavior, 38, 263-285.

22.

Fremouw

W. J.

Westrup

Pennypacker

(1997). Stalking on campus: The prevalence and strategies for coping with stalking. Journal of Forensic Sciences, 42, 666-669.

23.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

24.

Grasmick

H. G.

Tittle

C. R.

Bursik

R. J.

Arneklev

B. J.

(1993). Testing the core empirical implications of Gottfredson and Hirschi’s general theory of crime. Journal of Research in Crime and Delinquency, 30, 5-29.

25.

Hall

D. M.

(1998). The victims of stalking. In Meloy

J. R.

(Ed.), The psychology of stalking: Clinical and forensic perspectives (pp. 113-137). San Diego, CA: Academic Press.

26.

Haugaard

J. L.

Seri

L. G.

(2003). Stalking and other forms of intrusive contact after the dissolution of adolescent dating or romantic relationships. Violence and Victims, 18, 279-297.

27.

Hickman

Piquero

A. R.

(2001). Exploring the relationships between gender, control balance, and deviance. Deviant Behavior, 22, 323-351.

28.

Hickman

M. J.

Piquero

A. R.

Lawton

B. A.

Greene

J. R.

(2001). Applying Tittle’s control balance theory to police deviance. Policing: An International Journal of Police Strategies and Management, 24, 497-519.

29.

Higgins

G. E.

Lauterbach

(2004). Control balance theory and exploitation: An examination of contingencies. Criminal Justice Studies, 17, 291-310.

30.

Higgins

G. E.

Lauterbach

Tewksbury

(2005). Control balance theory and violence: An examination of contingencies. Sociological Focus, 38, 241-260.

31.

Holtfreter

Reisig

M. D.

Pratt

T. C.

(2008). Low self-control, routine activities, and fraud victimization. Criminology, 46, 189-220.

32.

Kaplowitz

M. D.

Hadlock

T. D.

Levine

(2004). A comparison of web and mail survey response rates. Public Opinion Quarterly, 68, 94-101.

33.

Kerley

K. R.

Hochstetler

Copes

(2009). Self-control, prison victimization, and prison infractions. Criminal Justice Review, 34, 553-568.

34.

Kerley

K. R.

Sirisunyaluck

(2008). Self-control, intimate partner abuse, and intimate partner victimization: Testing the general theory of crime in Thailand. Deviant Behavior, 29, 503-532.

35.

Kienlen

K. K.

Birmingham

D. L.

Solberg

K. B.

O’Regan

J. T.

Meloy

J. R.

(1997). A comparative study of psychotic and nonpsychotic stalking. Journal of American Academic Psychiatry & Law, 25, 317-334.

36.

Langhinrichsen-Rohling

Palarea

R. E.

Cohen

Rohling

M. L.

(2000). Breaking up is hard to do: Unwanted pursuit behaviors following the dissolution of a romantic relationship. Violence and Victims, 15, 73-90.

37.

Lewis

S. F.

Fremouw

W. J.

Del Ben

Farr

(2001). An investigation of the psychological characteristics of stalkers: Empathy, problem-solving, attachment and borderline personality features. Journal of Forensic Science, 46, 80-84.

38.

Logan

T. K.

Leukefeld

Walker

(2000). Stalking as a variant of intimate violence: Implications from a young adult sample. Violence and Victims, 15, 91-111.

39.

McCreedy

Dennis

(1996). Sex related offenses and fear of crime on campus. Journal of Contemporary Criminal Justice, 12, 69-80.

40.

Messner

S. F.

Krohn

M. D.

Liska

A. E.

(1989). Theoretical integration in the study of deviance and crime: Problems and prospects. Albany: State University of New York Press.

41.

Mustaine

E. E.

Tewksbury

(1999). A routine activity theory explanation for women’s stalking victimizations. Violence Against Women, 5, 43-62.

42.

Neter

Wasserman

Kutner

M. H.

(1985). Applied linear statistical models: Regression, analysis of variance, and experimental design. Homewood, IL: Richard D. Irwin, Inc.

43.

Ngo

F. T.

Paternoster

(2011). Cybercrime victimization: An examination of individual and situational level factors. International Journal of Cyber Criminology, 5, 773-793.

44.

Nobles

M. R.

Fox

K. A.

Piquero

N. L.

Piquero

A. R.

(2009). Career dimensions of stalking victimization and perpetration. Justice Quarterly, 26, 476-503.

45.

Nobles

M. R.

Reyns

B. W.

Fox

K. A.

Fisher

B. S.

(in press). Protection against pursuit: A conceptual and empirical comparison of cyberstalking and stalking victimization among a national sample. Justice Quarterly.

46.

Pathè

Mullen

P. E.

(1997). The impact of stalkers on their victims. British Journal of Psychiatry, 170, 12-17.

47.

Patton

C. L.

Nobles

M. R.

Fox

K. A.

(2010). Look who’s stalking: Obsessive pursuit and attachment theory. Journal of Criminal Justice, 38, 282-290.

48.

Payne

B. K.

Chappell

(2008). Using student samples in criminological research. Journal of Criminal Justice Education, 19, 175-192.

49.

Piquero

A. R.

Hickman

(1999). An empirical test of Tittle’s control balance theory. Criminology, 44, 319-341.

50.

Piquero

A. R.

Hickman

(2003). Extending Tittle’s control balance theory to account for victimization. Criminal Justice and Behavior, 30, 282-301.

51.

Piquero

A. R.

MacDonald

Dobrin

Daigle

L. E.

Cullen

F. T.

(2005). Self-control, violent offending, and homicide victimization: Assessing the general theory of crime. Journal of Quantitative Criminology, 21, 55-71.

52.

Piquero

A. R.

MacIntosh

Hickman

M. J.

(2001). Applying Rasch modeling to the validity of a control balance scale. Journal of Criminal Justice, 29, 493-505.

53.

Piquero

N. L.

Piquero

A. R.

(2006). Control balance and exploitative corporate crime. Criminology, 44, 397-430.

54.

Reisig

M. D.

Pratt

T. C.

Holtfreter

(2009). Perceived risk of internet theft victimization: Examining the effects of vulnerability and financial impulsivity. Criminal Justice and Behavior, 36, 369-384.

55.

Reyns

B. W.

Henson

Fisher

B. S.

(2011). Being pursued online: Applying cyberlifestyle-routine activities theory to cyberstalking victimization. Criminal Justice and Behavior, 38, 1149-1169.

56.

Schreck

C. J.

(1999). Criminal victimization and low self-control: An extension and test of a general theory of crime. Justice Quarterly, 16, 633-654.

57.

Schreck

C. J.

Stewart

E. A.

Fisher

B. S.

(2006). Self-control, victimization, and their influence on risky lifestyles: A longitudinal analysis using panel data. Journal of Quantitative Criminology, 22, 319-340.

58.

Schreck

C. J.

Wright

R. A.

Miller

J. M.

(2002). A study of individual and situational antecedents of violent victimization. Justice Quarterly, 19, 159-180.

59.

Sheehan

K. B.

McMillan

S. J.

(1999). Response variation in e-mail surveys: An exploration. Journal of Advertising Research, 39, 45-54.

60.

Stewart

E. A.

Elifson

K. W.

Sterk

C. E.

(2004). Integrating the general theory of crime into an explanation of violent victimization among female offenders. Justice Quarterly, 21, 159-181.

61.

Tittle

C. R.

(1995). Control balance: Toward a general theory of deviance. Boulder, CO: Westview Press.

62.

Tittle

C. R.

(1997). Thoughts stimulated by Braithwaite’s analysis of control balance theory. Theoretical Criminology, 1, 99-110.

63.

Tittle

C. R.

(1999). Continuing the discussion of control balance. Theoretical Criminology, 3, 344-352.

64.

Tittle

C. R.

(2004). Refining control balance theory. Theoretical Criminology, 8, 395-428.

65.

Tjaden

Thoennes

(1998). Stalking in America: Findings from the National Violence Against Women Survey. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics.

66.

Tjaden

Thoennes

(2000). Extent, nature, and consequences of intimate partner violence: Findings from the national violence against women survey (NCJ 181867). Washington, DC: U.S. Department of Justice.

67.

Tonin

(2004). The attachment styles of stalkers. Journal of Forensic Psychiatry & Psychology, 15, 584-590.

68.

Walby

Allen

(2004). Domestic violence, sexual assault and stalking: Findings from the British crime survey. London: Home Office Research, Development and Statistics Directorate

69.

Wells

Cavanaugh

M. R.

Bouffard

J. A.

Nobles

M. R.

(2012). Non-response bias with a web-based survey of college students: Differences from a classroom survey about carrying concealed handguns. Journal of Quantitative Criminology, 28, 455-476.