Why Definitions Matter

Abstract

Although there is a growing understanding of stalking victimization, it remains difficult to define, and characterizations of the phenomenon vary within the literature. As such, research is needed to understand how variations in the definition of stalking may change who is defined as a victim and thereby limit the generalizability of findings across previous studies. The focus of this study is the inclusion or exclusion of subjective and reasonable measures of fear for 1,430 victims identified by the 2006 Supplemental Victimization Survey. Results suggest that the definition of stalking is important, and prior research has potentially excluded stalking victims due to restrictive operationalizations. Victims who report different types of fear appear to be similar to each other in some respects but differ in others, particularly with regard for gender representation, suggesting some definitions of stalking may be gendered and under-represent male stalking victims. Finally, using complex stratified survey weights, the impact of these varying operationalizations is examined. Using the same data but different definitions resulted in estimates of just over 1 to 5.3 million persons who are stalked in the United States each year.

Keywords

stalking victimization methods national victimization estimates NCVS

Introduction

It is generally accepted that stalking is a widespread problem with serious economic, social, medical, and psychiatric consequences (for reviews, see Blaauw, Winkel, Arensman, Sheridan, & Adriënne Freeve, 2002; Dressing, Kuehner, & Gass, 2006). Although research on stalking has increased in the past two decades, much is still unknown and the definition of stalking victimization remains disputed. Stalking is difficult to define, and there is little consensus among researchers, legislators, practitioners, and even the public on what constitutes stalking (Tjaden, 2003, 2009). This is because the legal definitions vary widely from state to state and within the stalking literature (Tjaden & Thoennes, 1998, Westrup & Fremouw, 1998). In addition, the emphases of different groups (criminologists, psychologists, psychiatrists, legal scholars, practitioners, and policy makers) have also contributed to the lack in uniformity in the definitions of stalking (Fox, Nobles, & Fisher, 2011).

The legal definition of stalking varies across states because anti-stalking laws are relatively new, are continuously being updated, and most frequently emerged from each state’s anti-harassment laws which also differ throughout the United States (Beatty, 2003; Mullen, Pathé, & Purcell, 2000; Nicol, 2006; Purcell, Pathé, & Mullen, 2004; Sheridan, Blaauw, & Davies, 2003; Tjaden, 2009). At the most basic level, stalking and harassment are similar insofar that stalking is a more serious type of repeated harassment. Unlike other types of victimizations, which only need to occur once, stalking is a series crime and must occur on more than one occasion. To complicate the issue, stalking can include otherwise legal behaviors. Many stalking behaviors, such as making phone calls, leaving messages, sending gifts, and showing up unannounced, are socially acceptable in some situations. These types of behaviors are neither illegal nor unwanted in certain contexts, such as intimate relationships or friendships (Fox et al., 2011).

Although stalking can be defined in a variety of ways, most legal definitions require three elements to occur in tandem: a pattern of conduct directed at a specific person, conduct intended to place that person in fear for his or her safety, and conduct that actually places that person in fear for his or her safety (Beatty, 2003). States, however, vary in terms of what behaviors constitute stalking, the level of threats or fear, and how often the events must occur. Based on the most recent National Survey of State Laws (Leiter, 2008), most states require fear and many require an actual threat be made (see also Fox et al., 2011), but there is considerable variation in how each state defines “fear.”¹ For example, a person could receive repeated, unwanted phone calls, which would be considered harassment. If these actions made a victim fearful or it was determined that the behavior would make a reasonable person fearful, this should be considered stalking; but not all people react in similar ways. What frightens or threatens one person might simply annoy or anger another. The highly subjective nature of stalking and harassing behaviors is one of the many reasons that creating a single, unified definition of stalking has been so difficult. Thus, crafting a definition is important to understand the prevalence, consequences, and offer legal protections to victims of stalking.

Much of the definitional debate centers on whether a fearful reaction by the victim is a necessary element for behaviors to constitute stalking (Melton, 2007). There are two sides to this argument. One group argues that fear is an essential part of stalking and “without fear, there is no crime of stalking” (Tjaden, 2009, p. 266). Fear is what makes otherwise legitimate behavior criminal (National Criminal Justice Association, 1993). In 1993, the National Criminal Justice Association developed a model anti-stalking code for states where stalking was a felony requiring a high level of fear of bodily injury or death. Behaviors that resulted in lower levels of fear, such as annoyance or emotional distress, would be considered harassment or trespassing (National Criminal Justice Association, 1993). California updated its stalking code to “reasonable fear” from “reasonable fear of death or great bodily harm” because it was difficult to prosecute offenders who did not make their victims fearful of actual death (Dietz & Martin, 2007).

Opponents of the fear element argue that requiring victims to be fearful may unintentionally alienate victims and is an unnecessary component of whether a crime occurred. There is no other crime that requires victims to be fearful in addition to some behavior(s) before their victimizations would be considered illegal (Dietz & Martin, 2007). In addition, requiring high levels of fear may exclude victims from the legal protections that stalking codes offer before the actions escalate into more serious bodily injury or death (National Center for Victims of Crime, 2007). While most state statutes on stalking require fear, many do not. Likewise, some researchers of stalking victimization require fear, although the majority of researchers do not (Fox et al., 2011).

Subjective Versus Reasonable Fear

In addition to whether to include fear as a necessary element of the definition of stalking, there is also debate on how to operationalize fear. In legal statutes, states have defined fear in two separate yet equally important ways: subjective and reasonable fear. To be subjectively fearful, victims must report they were emotionally frightened because of the offender’s repeated, harassing behaviors. Although originally victims had to be fearful of bodily harm or death, states like California have broadened this to include fear more generally. This is a subjective measure of fear and open to a victim’s interpretation of the offender’s actions. Second, reasonable fear draws upon a “reasonable person standard” where, in a court of law, a victim could show how another rational person would be similarly fearful. This measure of fear requires specific examples of behaviors or threats that would make a jury/judge feel fear was a justifiable emotional response. Because these two types of fear are used extensively throughout this project, the working definitions using in this project are outlined below:

Subjective fear refers to the victims’ emotional state. Respondents must disclose they were scared, fearful, frightened, afraid, terrified, and so on, because of the offender’s actions to have experienced subjective fear.

Reasonable fear refers to behaviors by the stalker that the victim experienced which could cause a reasonable person to become subjectively fearful. This measure of fear is objective as it does not require the victim to have actually felt fearful, only that a reasonable person could have been fearful because of the stalker’s actions or behaviors. Although the types of behaviors by the stalker that could result in reasonable fear can vary, they often include threats, attacks toward the victim or their loved ones, breaking and entering, or property damage.

National-Level Research on Stalking Victimization

Like states, researchers and agencies have defined stalking in numerous ways and relied on various types of sampling procedures. Due to convenience, a large number of studies on stalking have relied on samples of college students (Bjerregaard, 2000; Cupach & Spitzberg, 2000; Fremouw, Westrup, & Pennypacker, 1997; Logan, Leukefeld, & Walker, 2000; McCreedy & Dennis, 1996; Nobles, Fox, Piquero, & Piquero, 2009; Spitzberg, Nicastro, & Cousins, 1998)—often only female college students (Buhi, Clayton, & Surrency, 2009; Coleman, 1997; Fisher, Cullen, & Turner, 2000, 2002; Mustaine & Tewksbury, 1999). Unfortunately, the findings from university-based samples may not be generalizable to persons outside the college setting. For this reason, nationally representative data are preferred.

To date, there have been four nationally representative studies conducted that have addressed stalking victimization in the United States for persons 18 and above and each required a stalking victim report some level of fear (Basile, Swahn, Chen, & Saltzman, 2006; Baum, Catalano, Rand, & Rose, 2009; Black et al., 2011; Catalano, 2012; Tjaden & Thoennes, 1998). One might assume that several nationally representative samples conducted within a 15-year period (1995-2010) would produce similar results, but the estimates of stalking vary considerably across studies (see Table 1). Annual estimates suggest 1.4 to 5.6 million people are stalked each year, between 0.4% to 2% for men and 1% to 6.5% for women. Lifetime estimates are even higher, suggesting between 2% to 8% of men have been stalked and between 7% to 25% of women have been stalked. Why do these nationally representative samples yield such varied conclusions? Although sampling and the year the data were collected may influence this variation, it seems more likely that the differences in these estimates are due to the samples, how the researchers phrased the questions, and how stalking was operationalized.

Table 1.

Prior National Estimates of Stalking Victimization.

Study	NVAWS		NISVS		ICARIS-2		BJS-SVS
Authors (year)	Tjaden and Thoennes (1998)		Black et al. (2011)		Basile, Swahn, Chen, and Saltzman (2006)		Catalano (2012)
No. of interviews (n)	16,000		18,049		9,684		65,272
Year conducted	1995-1996		2010		2001-2003		2006
	Male	Female	Male	Female	Male	Female	Male	Female
Annual (%)	0.4	1	1.3	4.3	—	—	0.8	2.2
Lifetime (%)	2	8	5.2	16.2	2	7	—	—
Lowered fear standard
Annual (%)	1.5	6	2	6.5	—	—	—	—
Lifetime (%)	4	12	7.9	25	—	—	—	—

Note. NVAWS = National Violence Against Women Survey; NISVS = National Intimate Partner and Sexual Violence Survey; ICARIS-2 = Second Injury Control and Risk Survey; BJS = Bureau of Justice Statistics; SVS = Supplemental Victimization Survey.

The focus of this article is how stalking is operationalized. Two research questions guide this discussion:

Research Question 1: What is the relationship between fear and stalking? More specifically, do those who report fear from repeated-following/harassing differ from those who do not report fear?

Research Question 2: How does the inclusion or exclusion of fear affect national-level estimates of stalking victimization?

The Supplemental Victimization Survey (SVS), described in greater detail below, has several advantages that make it the most appropriate for addressing research questions outlined in this research. First, the SVS has the largest sample size. It is more than 3.5 times larger than the other studies (i.e., the Second Injury Control and Risk Survey [ICARIS-2]). This is important because victimization is a rare event, and a large sample is required when trying to understand the subtle nuances between definitions or groups. Second, the SVS is publicly available. Currently, the National Intimate Partner and Sexual Violence Survey (NISVS) is not available to the public. Third, the SVS is roughly 10 years newer than the National Violence Against Women Survey (NVAWS) and includes modern stalking manifestations such as cyberstalking. In addition, the SVS can be merged with the National Crime Victimization Survey (NCVS) to examine whether stalking victimization is similar to or different from other forms of victimization like violence or property crime. It provides a unique opportunity to examine whether traditional victimization surveys, which do not normally measure stalking, are also capturing stalking victimization that involves violence or property damage. Finally, the SVS is appropriate because of the rich detail and multiple measures of key variables to create multiple operationalizations of fear and stalking.

Data and Measures

To address these research questions, this project draws on two data sets: the NCVS and SVS (U.S. Department of Justice [USDJ], 2009a, 2009b). The NCVS began in 1973 and is an ongoing survey administered by the Bureau of Justice Statistics (BJS). The NCVS uses a rotating panel design that captures non-fatal victimization experiences from individuals every 6 months using a nationally representative sample of U.S. households. It is a hierarchical data set containing data at the household, person, and incident levels. All household respondents above the age of 12 are interviewed about their victimization experiences during the past 6 months. The NCVS measures various violent and property victimizations but does not normally measure stalking.

Between January and June 2006, the BJS collected a one-time supplement to the NCVS that measured stalking victimization: the SVS. After completing the NCVS interview, respondents 18 and above were asked to complete the SVS. To be consistent with this protocol, only adults 18 and above in the NCVS are examined in this study. Of the 78,741 eligible NCVS respondents, 65,272 were asked the SVS screener questions, which is a response rate of 83% (Baum et al., 2009). Of these, 1,686 people met the “screener” requirements and completed the SVS.

There were three sets of screener questions that assessed whether the respondent has experienced “stalking” and should complete the SVS. Respondents must have experienced at least one type of stalking behavior, experienced the behavior(s) repeatedly, and within the past 12 months.² Post-data-collection edits by the BJS suggested that 170 respondents were incorrectly screened into the SVS and therefore were removed (see Catalano, 2012).³ A person-level data file was constructed for the 1,516 people who were correctly screened into the SVS that included relevant incident and household-level variables. For simplicity throughout, the term “stalked” or “stalking” is used to define those who experienced repeated following or harassing behaviors and were screened into the SVS.

Dependent Variables

Stalking victimization

To complete the SVS, respondents must have indicated a positive response to three sets of screener questions to complete the full SVS survey, which provides detail about potential stalking incidents. The first set of screener questions to assess stalking behaviors asks,

Not including bill collectors, telephone solicitors, or other sales people, has anyone, male or female, EVER—frightened, concerned, angered or annoyed you by: 1) Making unwanted phone calls to you or leaving messages? 2) Sending unsolicited or unwanted letters, e-mails, or other forms of written correspondence or communication? 3) Following you or spying on you? 4) Waiting outside or inside places for you such as your home, school, workplace, or recreation place? 5) Showing up at places where you were even though he or she had no business being there? 6) Leaving unwanted items, presents, or flowers? 7) Posting information or spreading rumors about you on the Internet, in a public place, or by word of mouth? (USDJ, 2009a, SVS-1 p. 1).

In addition, the respondent must also have indicated that these actions occurred on more than one occasion. This question was intended to capture the repeated nature of stalking behaviors. Finally, the purpose of the SVS was to create stalking victimization estimates for a single year. The final screener question asks victims to report whether these behaviors occurred within the past 12 months to calculate annual estimates but not lifetime estimates of stalking. Therefore, all persons screened into the SVS experienced harassing behaviors, on multiple occasions, in the past 12 months. The BJS later classifies these people into one of two groups—stalking and harassment victims—based on whether they reported experiencing some type of fear.

Fear

Later in the survey, questions pertaining to fear were introduced. Fear was measured in the SVS in two separate ways: subjective and reasonable fear. Subjective fear is measured using two open-ended questions: How did the behavior make the victim “feel when it FIRST started” and “as the behavior progressed?” Respondents could give any number of feelings, but the interviewer was not to read the possible emotions to the respondents. Instead, the interviewers were given eight broad emotional response categories, as well as several synonyms to help the interviewer choose the most appropriate response.⁴ These eight emotional categories included anxious/concerned, annoyed/angry, frightened, depressed, helpless, sick, suicidal, and an “other” category. If victims stated they were frightened, either when the behaviors first started or as the behavior progressed, then they were coded as being subjectively fearful.

Reasonable fear is captured when respondents report they experienced some kind of event that would make a reasonable person fearful. There were a total of 23 possible reasonable fears and they include behaviors, threats, or crimes committed by the offender (see the appendix). If respondents said yes to any of these 23 events, they were coded using a dichotomous variable as experiencing reasonable fear.

Independent Variables

A series of individual, household, offender, event, and self-protecting factors are used to address if there are differences in who reports fear and those that do not. For a more detailed discussion, see the appendix. Individual-level variables include sex, age, race/ethnicity, employment, in-school status, and a control for unbounded interviews. Household-level variables included in this research are family structure, location, household income, household size, and mobility (measured in number of moves over the past five years and length of residence). Offender-level characteristics consist of: the number of offenders; the offender’s age, race, and gender; if the offender had a criminal record or was employed; and victim–offender relationship. Event-level characteristics include the number of following behaviors the respondent reported on the screener questionnaire, if the victim was physically followed, the duration, and frequency in the past year. Self-protecting behaviors include changes in day-to-day activities, changing personal information, asking for help, or moving (see the appendix). Due to multicollinearity, these measures are summed to create an over measure of the number of protective behaviors.

Analytic Approach for Research Questions

The first research question evaluates if those who report fear from repeated-following/harassing differ from those who do not report fear. Of the 1,516 persons who remained in the SVS, only 1,430 persons answered both the subjective and reasonable fear questions. These data are used to compare the four possible groups created by answering two separate measurements of fear. Respondents could fall into one of four groups: those who reported they were (a) unafraid, (b) subjectively fearful, (c) reasonably fearful, or (d) both subjectively and reasonably fearful. Table 2 presents a visual representation of the four possible groups. The BJS would later classify those who were unafraid as harassment victims and those reporting any indication of fear as stalking victims (Baum et al., 2009). There is little research, however, on whether there is agreement between these two measures of fear, if combining these types of fear produces a homogeneous group, or whether there are significant differences between respondents who do or do not report each type of fear. It is assumed that those who report fear are somehow different from those who do not. While one of the reasons for including fear in the definition of stalking is to prevent policy makers from criminalizing annoying behaviors, it is possible that this group is not significantly different from those reporting fear.

Table 2.

Types of Fear.

	Subjective Fear
Reasonable Fear	No	Yes
No	Harassment victim	Stalking victim
Yes	Stalking victim	Stalking victim

To assess the reliability and validity of the screener questions of the SVS to ensure it is consistently and accurately capturing a single latent construct, a secondary data set was created that contained the 65,261 people who fully answered the SVS screener questions. Being dichotomous variables, tetrachoric correlations are used to assess whether these variables load into a single factor.⁵ Next, using the respondents who answered both the subjective and reasonable fear questions, descriptive analyses are used to describe the mean or percentages of cases across various individual-, household-, offender-, and event-level characteristics. Chi-square tests or ANOVAs are used to conclude whether there are significant differences between groups.

The second research question gauges the impact of fear on national-level estimates of stalking victimization. Five unique operationalizations of stalking are presented, each relying on how people responded to the two measurements of fear. The first definition requires both subjective and reasonable measures of fear, that is, they reported they were emotionally afraid and they report some behavior or action by the offender that would cause a reasonable person to be fearful. A second definition only requires reasonable fear; whether they reported subjective fear is inconsequential. The third definition focuses only on subjective fear and does not require a respondent to report reasonable fear. Next, a fourth definition of stalking is whether a person reported subjective or reasonable fear. This is similar to the definition originally used by the BJS. Finally, fear may not be required at all. This final operationalization only focuses on the repeated following or harassing behaviors that prompted a respondent to be screened into the SVS but does not require fear.

Survey weighted techniques are appropriate to control for the complex sample design utilized by the NCVS. The SVS has a separate incident-level survey weight that is separate from the incident-level weight used in the NCVS. To create national-level estimates, survey weights are used for each definition of stalking. These descriptive analyses demonstrate the possible variations in stalking prevalence based on the five ways a researcher or state statute could operationalize stalking using fear.

Results

Using tetrachoric correlations, the seven screener questions were highly correlated with the others, and factor analysis suggested only one factor should be retained.⁶ Reliability was assessed using Cronbach’s alpha and was found to be .82, suggesting good reliability. The deletion of any of the seven items did not improve the alpha level. This suggests that these seven indicators are measuring one conduct or behavior and measuring it consistently, thus providing the foundation to use the SVS to address the research questions posed here.

The first research question attempts to gain a better understanding of the relationship between fear and stalking. Specifically, do those who report fear from repeated-following/harassing differ from those who do not report fear? Table 3 displays all 1,430 persons who completed the SVS and answered both sets of questions pertaining to the two measurements of fear. Most legal statutes require subjective fear, some require reasonable fear, and some require both. Here, respondents are subjectively fearful if they answered that the following behaviors made them fearful/afraid either when the behaviors began or as they progressed. In the ideal situation in the courtroom, a victim would be able to express both feeling afraid (subjective fear) and be able to articulate the specific behaviors or actions by the stalker that could cause a reasonable person to feel fear (reasonable fear) and fall in the bottom right quadrant of Table 3; however, this is not the case.

Table 3.

Unweighted Frequencies of the Two Types of Fear.

	No Reasonable Fear	Reasonable Fear	Total
No subjective fear	621 (43.4%)	298 (20.8%)	919
Subjective fear	219 (15.3%)	292 (20.4%)	511
Total	840	590	1,430

As seen in Table 3, most respondents reported some measure of fear, but usually not both. The top left quadrant shows the 621 people who did not experience either form of fear. The BJS would later deem these people to be harassment and not stalking victims. Because the other three quadrants experienced some measure of fear, they are defined as stalking victims. Yet, this table demonstrates an almost equal distribution across the two types of fear. The lower left quadrant shows the 219 people who experienced subjective fear but did not experience any of the 23 threats or attacks that would cause reasonable fear. The top right quadrant shows the 298 people who experienced at least 1 of the 23 measures of reasonable fear but did not report feeling afraid. For these 298 people, the most common emotion they reported was anger/annoyance. Just over 84% reported anger/annoyance either when the behaviors first started or as the behaviors progressed. Finally, the bottom right quadrant shows that 292 or 20.4% of respondents fit both operationalizations of fear.

Further examination by gender suggests that men and women disproportionately fall into one group over another. Table 4 presents the same data as Table 3, but computes the percentages of each group that is female. A total of 69.2% of all the 1,430 respondents were female as shown in the very bottom right corner. If there were no gender differences by type of fear, we could assume that approximately 69.2% of each quadrant would be female; but as we see, women are overrepresented in the subjective fear categories.

Table 4.

Percent Female by Type of Fear.

	No Reasonable Fear (%)	Reasonable Fear (%)	Total (%)
No subjective fear	60.4	54.4	58.4
Subjective fear	90.0	87.3	88.5
Total	68.1	70.7	69.2

Looking at the subtotals around the perimeter of the table shows 88.5% of those reporting subjective fear (with or without reasonable fear) are female. If a researcher or state statute required victims to report fear, then it would appear that nearly 9 out of 10 stalking victims were female. It is, however, important to note that not all women reported subjective fear. More than 58% of those who did not experience subjective fear were women. On the contrary, whether someone experienced reasonable fear seems to be more gender-neutral. Of those reporting reasonable fear (with or without subjective fear), 70.7% were female and 68.1% of those who did not report reasonable fear were female. These percentages are closely allied with the overall gender representation of the sample, which is 69.2% female.

Gender differences are more apparent when looking inside Table 4. Of those reporting only subjective fear, 90% were female compared with just over 54% of those reporting only reasonable fear. This suggests that feeling and/or reporting subjective fear might be a gendered response. Additional analyses (not shown) found that out of all 1,430 respondents, 39% of males and 42% of females reported reasonable fear, whereas 13% of males and 48% of females reported subjective fear.

The BJS and others classified SVS respondents into one of two groups: harassment victims, who did not report either type of fear; and stalking victims, who reported either or both types of fear. What is not known is whether there are important similarities or differences that exist between the four groups. If, for example, harassment victims are found to closely resemble the other three groups or those reporting only subjective fear are inherently different from those reporting reasonable fear, it may call into question the usefulness or appropriateness of prior research’s definition of stalking and harassment.

Table 5 presents the results of comparing the 1,430 respondents by the type of fear they experienced: both types of fear (Group 1), reasonable fear only (Group 2), subjective fear only (Group 3), and neither type of fear (Group 4) across individual-, household-, offender-, and event-specific variables. The mean or average is presented for each group, and the appropriate chi-square or ANOVA test was conducted to see whether the groups are significantly different from one another.⁷ Of particular interest is whether the groups that the BJS defines as stalking victims (Groups 1-3) are different from one another or from those defined as harassment victims (Group 4).

Table 5.

Comparison of Four Fear Groups: Chi-Square and ANOVA Results.

	Group 1	Group 2	Group 3	Group 4
	Both Fears	Reasonable Fear Only	Subjective Fear Only	Neither Fears	Significance
Article I. Individual
Male	12.7%^a	45.6%^b	10.0%^a	39.6%^b	***
Age	36.8^a	36.9^a	41.7^b	41.1^b	***
Education level	13.0^a,b	12.7^a	13.3^b	13.3^b	***
Race/ethnicity
White non-Hispanic	75.7%	76.2%	77.6%	76.0%
Black non-Hispanic	9.2%	9.4%	10.5%	10.5%
Hispanic	9.6%	10.1%	7.3%	7.2%
Asian/Other non-Hispanic	5.5%	4.4%	4.6%	6.3%
In-school status	16.9%	11.1%	11.4%	13.8%
Employed	74.1%	76.5%	75.0%	77.3%
Bounding effect	56.8%	56.4%	54.3%	51.0%
Article II. Household
Family Structure
Married	31.7%^a	36.4%^a	38.8%^a,b	46.4%^b	***
Single parent	28.3%^a	25.6%^a,b	18.7%^b,c	16.2%^c	***
Single person	40.0%	38.0%	42.5%	37.4%
Household location
Urban	33.9%	34.6%	33.8%	30.9%
Suburban	35.6%^a	39.6%^a,b	46.6%^b,c	46.9%^c	**
Rural	30.5%^a	25.8%^a,b	19.6%^b	22.2%^b	*
Moves in last 5 years	1.9^a	1.8^a	1.3^b	1.2^b	***
Years in current home	6.0^a	6.3^a	8.3^a,b	8.4^b	***
Household income	8.7^a	9.3^a	9.8^b	10.2^b	***
Household size 12+	2.2	2.2	2.1	2.3
Article III. Offender
Missing offender info	3.8%^a	5.7%^a	11.5%^b	12.3%^b	***
Number of offenders when known
Single offender	62.4%^a	63.8%^a,b	76.3%^c	69.6%^b,c	**
Two offenders	20.2%	22.6%	16.7%	19.7%
Three+ offenders	17.4%^a	13.6%^a,b	7.1%^c	10.7%^b,c	**
Offender male when known	76.4%^a	50.2%^b	83.0%^a	58.0%^c	***
Offender’s age when known
Below 21	8.0%	13.9%	8.6%	11.7%
21-29	20.4%^a,b	27.4%^b	16.6%^a	20.8%^a	*
30-39	26.5%	25.2%	23.4%	25.8%
40-49	25.5%	19.7%	25.1%	21.4%
50+	19.6%^a,b	13.9%^b	26.3%^a	20.3%^a	*
Offender’s race when known
White	79.0%	77.2%	75.0%	78.1%
Black	11.2%	14.1%	14.0%	11.7%
Other	9.8%	8.7%	11.0%	10.2%
Offender employment when known
Employed	57.3%^a	51.6%^a	56.4%^a	66.0%^b	**
Unemployed	31.7%^a	40.7%^b	34.6%^a,b	28.6%^a	*
Sometimes employed	11.0%	7.7%	9.0%	5.4%
Offender had criminal record when known	67.1%^a	51.4%^b	58.3%^a,b	37.2%^c	***
Victim-off relationship when known
Intimate	39.3%^a	32.9%^a	23.5%^b	19.7%^b	***
Known	55.7%	61.7%	56.3%	64.1%
Stranger	5.0%^a	5.4%^a	20.2%^b	16.2%^b	***
Event
No. of following behaviors	3.2^a	2.6^b	2.3^b	1.8^c	***
No. of physically following	1.6^a	1.1^b	1.0^b	0.6^c	***
Duration in days	828	628	876	636
Duration capped at 5 years	573^a	491^a	492^a	439^b	*
Lasted more than 6 months	64.9%^a	59.9%^a,b	54.7%^b,c	49.7%^c	***
Frequency in past year	70.5^a	57.2^a	40.2^a,b	39.9^b	***
Self-protecting behaviors
Changed daily activities	1.9^a	0.9^b	0.7^b	0.3^c	***
Changed personal information	0.8^a	0.5^b	0.4^b	0.2^c	***
Asked for help	2.8^a	1.7^b	1.4^b	0.7^c	***
Moved	25.4%^a	12.8%^b	7.8%^b	2.4%^c	***
No. of self-protective behaviors	5.9^a	3.2^b	2.6^b	1.3^c	***
Financial cost to victim	US$1,796^a	US$1,214^a,b	US$466^b,c	US$179^c	***
Was it stalking?	64.5%^a	49.1%^b	54.0%^b	24.7%^c	***

Note. Each superscript letter (a, b, c) denotes that the column proportions across Groups 1 to 4 do not differ significantly from each other at the .05 level (i.e., similar letters represent similar estimates/proportions).

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

First, Table 5 shows the variation across the four groups. Although there was no statistically significant difference across groups in terms of race or ethnicity, Table 5 suggests that there is significant variation across groups in terms of gender, age, and education. Unsurprisingly, individuals reporting subjective fear only (Group 3) are much less likely to be male compared with groups that did not report subjective fear (Groups 2 and 4). Other variables including race/ethnicity, in-school status, employment, and first time appearance in the sample (bounding effect) do not vary across the four groups.⁸

In terms of household-level characteristics, there is no difference between single person households, urban location, or household size 12 and above across the four groups. However, those reporting both fears or reasonable fear only are significantly less likely to be married and more likely to be single parents compared with those not reporting fear. The results also suggest that those reporting both fears are more likely to reside in rural areas, report more mobility, and have lower levels of income especially in comparison with those not reporting fear.

Next, offender-level variables are compared across the four groups. Before discussing the results, it is important to note that many respondents did not answer the offender questions in the SVS. If victims did not know how many people had followed/harassed them or whether they were followed/harassed by multiple offenders and were unable to identify a single person as most responsible, they were not asked any offender characteristic questions such as age, gender, or race.⁹

Those only reporting subjective fear have the highest percentage of stalking by a single person and the lowest percentage of stalking by three or more offenders while those reporting both fears were the opposite. Groups 1 and 3, both fears and subjective only fear, were significantly more likely to be stalked by male offenders (76% and 83%, respectively) compared with Groups 2 and 4, reasonable fear or neither fears (50% and 58%, respectively). Given that these groups had a higher percentage of female victims, this finding should not be surprising. Respondents reporting only reasonable fear had the highest percentage of male victims (45%) and the lowest percentage of male offenders (50%). There were few differences by offender age and employment, but race of the offender did not appear to differ across groups. A greater proportion of those reporting both fears knew the offender had a criminal record. Although this is an interesting finding, perhaps knowing or not knowing about a person’s criminal record reflects a familiarity with the offender. While the known category did not differ, Groups 1 and 2, both fears and reasonable only fear, were significantly more likely to report that the offender was an intimate compared with Groups 3 and 4, subjective only fear or unafraid, who were more likely than the other groups to report being stalked by a stranger.

Finally, Table 5 presents the final cross-group comparisons by event-specific characteristics. Overall, this table suggests that those displaying both types of fear, Group 1, report more serious stalking incidents than any other group, and those who are unafraid, Group 4, report slightly less serious stalking incidents compared with the other groups. For example, Group 1 reports more following and physically following behaviors as measured in the SVS screener. Duration is reported in three separate ways due to the wide range of responses. In its raw format, there are not significant differences across groups but some differences emerge in the capped or dichotomous measures. Group 1 had higher numbers for each type of self-protective behavior and estimated financial cost compared with those reporting only one type of fear, but these results were also significantly higher than those in Group 4. Finally, the SVS did not use the term “stalking” until the very last question on the SVS. When asked whether the respondents felt their experiences were stalking, those facing both types of fear, Group 1, were significantly more likely to believe they were stalked compared with all other groups at 64.5%. Those experiencing only one type of fear (Groups 2 and 3 at 49% and 54%) were more likely to believe it was stalking than those who did not report fear (Group 4 at 25%). Taken together, these findings reflect important distinctions between reasonable and subjective fear groups but suggest many do not consider these behaviors to be stalking. On the contrary, of those who did not report fear (Group 4) experience similar, albeit less severe, behaviors and nearly a quarter of them consider these behaviors to be stalking.

The second research question examines how relying on different operationalizations of fear can have a dramatic impact on who is defined as a stalking victim and consequently national-level estimates of stalking victimization. Drawing from the two ways fear was measured in the SVS, respondents could fall into different groups depending on their response to a measure of subjective and reasonable fear. Depending on the researcher’s definition or a state’s statute, there are five possible ways stalking victimization could have been defined: (a) require both measures of fear, (b) require reasonable fear but reporting subjective fear is inconsequential, (c) require subjective fear but reasonable fear is inconsequential, (d) require either type of fear, or (e) not require fear.

After weighting the data, Table 6 shows the variation in the prevalence of stalking victimization depending on the utilization of fear and gender representation. Using the most stringent criteria, requiring both types of fear, just over 1 million people were stalked during a 12-month period; 86% of them were female. If, however, it was decided that threatening behaviors are vital and reasonable fear is a better indicator of determining whether stalking occurred, then almost 2.1 million people would report being stalked. Of these, 68.2% were female. Alternatively, if emotions are paramount, then nearly 1.8 million people were stalked because they experienced subjective fear and, of these, 87.4% were female. If, like the BJS’s definition, any type of fear is appropriate, then nearly 2.9 million people were stalked over a 12 month period and 73.8% were female.¹⁰ If fear is inconsequential or biased the findings in some way, then relying on a definition that does not use fear may be the most appropriate. Using this operationalization means anyone screened into the SVS would be considered a stalking victim, whether they report fear. Approximately 5.3 million people 18 or older were victims of stalking using this definition and nearly 67% were women.

Table 6.

National Estimates of Stalking Using Different Operationalizations of Fear.

	Victimizations	% Female	% Male
Both fears	1,007,820	86.2	13.8
Reasonable fear	2,088,270	68.2	31.8
Subjective fear	1,785,680	87.4	12.6
Any type of fear	2,866,130	73.8	26.2
Fear not required	5,317,100	66.9	33.1

Conclusion

In the many definitions of stalking, fear has played a central but controversial role, yet there has been little consistency in its operationalization. Research Question 1 assessed the inclusion or exclusion of the two types of fear measured by the SVS and frequently required in state statutes. Four groups were created to demonstrate the ways victims could report fear: subjective fear, measured as an emotional responses to the behaviors, and reasonable fear, measured as experiencing a fear-inducing event or threat that would cause a reasonable person to be fearful. Each person could report both subjective and reasonable fear, reasonable fear only, subjective fear only, or report neither type of fear. In operationalizing stalking, the BJS included the first three groups as stalking victims while excluding the fourth group, those reporting neither fear, and called them harassment victims. Although separating harassment from stalking is difficult and reporting any type of fear is a commonly used technique, the goal of this question was to ascertain whether there was enough evidence to suggest that the first three groups were different enough from the fourth group to justify removing them from being defined as stalking victims.

Throughout Table 5, bivariate analyses highlighted the differences across the four groups. Victims reporting both types of fear reported more serious stalking situations. For example, they reported more following behaviors, more self-protecting behaviors, more financial costs, and were more likely than any other group to label their experiences as stalking. Even using the most stringent definition of stalking, 64.5% called their experiences “stalking,” whereas over one third of respondents did not self-identify as stalking victims. Alternatively, almost one quarter of people who did not report either type of fear labeled the events they experienced as stalking.

The overall pattern and conclusion is not entirely clear. In many cases, the groups are similar in terms of race/ethnicity, student status, employment, whether this was their first time in the sample, single person households, urban locations, household size, offender’s race, a known (nonintimate partner) victim–offender relationship, and duration measured in days. While those experiencing both types of fear (Group 1) did report more serious stalking situations, the difficulty is distinguishing the similarities and differences between the other three groups. Those reporting only reasonable fear (Group 2) and those reporting only subjective fear (Group 3) were frequently indistinguishable from one another. Other times, Group 2 was similar to Group 1, whereas Group 3 resembled Group 4, those who did not report either measure of fear.

The BJS used an operationalization of stalking that required some measure of fear, essentially combining Groups 1, 2, and 3, and defined those who did not report fear, Group 4, as harassment victims (Baum et al., 2009; Catalano, 2012). These analyses suggest this particular grouping, at times, excludes a group that is remarkably similar to the other three, particularly Group 3 (those experiencing only subjective fear). Even though the event characteristics variables suggest these behaviors exist on a seriousness continuum where Group 1 reports more serious behaviors and negative effects compared with Group 4, Group 4 is still reporting serious behaviors and effects. For example, they report being physically following for long periods of time, used self-protecting behaviors, were financially burdened, and nearly 25% believed what they had experienced was stalking.

Is Group 4 different enough from Groups 1, 2, and 3 that it should be called something else, like harassment? Given the findings above, perhaps not. These analyses suggest that the current categorizations may oversimplify a complex phenomenon and a bifurcate group of people who share many common characteristics and experiences. By excluding those who do not report fear, there is the possibility that we are underestimating the number of people affected by stalking, thereby reducing our understanding of stalking victimization, its victims, and consequences. We also saw that men are predominately captured in Groups 2 and 4 and by excluding Group 4, as the BJS does, we are dropping a large number of men from being defined as victims.

This article suggests there is not a clear-cut distinction between what BJS calls stalking and harassment. Grouping all four categories would reduce the overall appearance of the severity of the behaviors and consequence. To some extent, this is already being done by combining Groups 2 and 3 with Group 1. Those experiencing both fears, Group 1, reported actions and consequences that were frequently more severe than those reported by victims reporting only reasonable or subjective fear (i.e., Groups 2 and 3). A compromise is to publish multiple estimates, similar to the four groups examined here, that highlight the complexities of defining stalking and better demonstrate that not all people experience the same type of fear.

Research Question 2 examined the impact of using different fear requirements on national-level estimates of stalking victimization. The results showed the wide variation that could result depending on how stalking was defined using the SVS data: (a) require both measures of fear, (b) require reasonable fear but reporting subjective fear is inconsequential, (c) require subjective fear but reasonable fear is inconsequential, (d) require either type of fear, or (e) not require fear. Estimates ranged from just over 1 to 5.3 million persons stalked annually. It is also important to note the variation in the percentage of male and female who are defined as victims. For example, the percentage of male victims varied across the different definitions ranging from 13.8% to 33.1%, suggesting some operationalizations of stalking are more restrictive of male stalking victimization than others. These first two research questions begin to provide a rich description of the variation, particularly by gender, inherent in stalking definitions in different surveys, data analyses, and across state statues, which depend largely on requiring fear.

Researchers and policy makers need to develop a definition for stalking that captures the variety of behaviors and reactions that stalking victims experience. This study assessed the validity and reliability of the multiple-item indicators of stalking behaviors to determine whether each of the seven behaviors used by the SVS (which are similar to other stalking surveys) provides a reliable measure of stalking victimization as some researchers have recommended (Fox et al., 2011). Factor analysis assessed whether each of these items load on the same factor, indicating a single latent construct, or whether these seven behaviors are multidimensional, representing separate constructs. These analyses found that the seven-item scale used by the SVS loads on a single construct that is both valid and reliable. This finding lends support to prior research that utilized the SVS screener questions, but researchers should also address whether other stalking surveys have valid and reliable instruments.

Beyond using a multi-item indicator that is similar to past studies, this project uses multiple operationalizations of stalking. Stalking is not the only area of research that has struggled to define its dependent variable. Defining “desistance” or what constitutes a “gang” has challenged researchers, policy makers, and law enforcement professionals (Bushway, Piquero, Broidy, Cauffman, & Mazerolle, 2001; Esbensen, Winfree, He, & Taylor, 2001). Although defining what constitutes a gang is not fully resolved, researchers suggest that asking youth to self-identify as gang members may currently be the best solution.

Fox et al. (2011) suggest that using the term stalking (i.e., “Have you even been stalked?”) may introduce bias compared with the use of several behavioral questions. Drawing from research on sexual victimization, the authors suggest that using the term “stalking” like the term “rape” may discourage reporting. Using behavioral indicators, however, helps respondents recall experiences, avoid stigmatizing language like “stalker,” and reduce confusion on what types of behaviors the interviewers are asking for. The SVS used a series of dichotomous stalking activities in the screener and did not use the term stalking until the last question which should reduce complications from this issue.

A notable limitation of this approach is that it may undercount victims who did not meet the screener requirements. That is, some victims may believe they were stalked but not meet the repeated behaviors outlined in the SVS screener questions. The types of following behaviors outlined in the SVS may not be exhaustive and some victims may not have been prompted to complete the survey, even though they would have indicated they were a stalking victim. As many of the people who entered the SVS did not label their experiences as stalking, it is very likely that there would also be a group of people who would initially label their experiences as stalking even though they would not meet the screener requirements. Perhaps random assignment to one of these two possible screener questions would provide a better understanding of these two possible approaches to estimate stalking victimization.

Future research should attempt to address these limitations and better understand how and why these four groups differ. For example, using qualitative interviews, it would be particularly useful to understand why those who experience reasonable fear do not constantly report subjective fear or why individuals who report fear do not always consider these events to be stalking. Likewise, interviews with those who did not report either type of fear, Group 4, but reported their experiences to be stalking would help researchers better understand how the public defines stalking, the types of individuals who report their crimes to police, and, most importantly, those who refrain from seeking help. Perhaps with a better understanding of the phenomena, more targeted laws and legal protections can be enacted to protect victims from stalking.

Footnotes

Appendix

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Jennifer Gatewood Owens is an assistant professor in the Department of Criminal Justice and Criminology at the University of Missouri–Kansas City. Her work focuses on victimization, gender, substance use, and methodological issues.

References

Basile

K. C.

Swahn

M. H.

Chen

Saltzman

L. E.

(2006). Stalking in the United States: Recent national prevalence estimates. American Journal of Preventive Medicine, 31, 172-175.

Baum

Catalano

Rand

Rose

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

Beatty

(2003). Stalking legislation in the United States. In Brewster

(Ed.), Stalking: Psychology, risk factors, interventions, and law. (pp. 21-55). Kingston, NJ: Civic Research Institute.

Biderman

A. D.

Cantor

(1984). A longitudinal analysis of bounding respondent conditioning and mobility as sources of panel bias in the National Crime Survey. In Proceedings of the Survey Methods Research Section. Washington, DC: American Statistical Association.

Bjerregaard

(2000). An empirical study of stalking victimization. Violence and Victims, 15, 389-406.

Blaauw

Winkel

F. W.

Arensman

Sheridan

Adriënne Freeve

(2002). The toll of stalking: The relationship between features of stalking and psychopathology of victims. Journal of Interpersonal Violence, 17, 50-63.

Black

Basile

Breiding

Smith

Walters

Merrick

. . .Stevens

(2011). The National Intimate Partner and Sexual Violence Survey (NISVS): 2010 summary report. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention.

Buhi

Clayton

Surrency

H. H.

(2009). Stalking victimization among college women and subsequent help-seeking behaviors. Journal of American College Health, 57, 419-425.

Bushway

Piquero

Broidy

Cauffman

Mazerolle

(2001). An empirical framework for studying desistance as a process. Criminology, 39, 491-515.

10.

Catalano

(2012). Stalking victims in the United States–Revised. Washington, DC: U.S. Department of Justice.

11.

Coleman

F. L.

(1997). Stalking behavior and the cycle of domestic violence. Journal of Interpersonal Violence, 12, 420-432.

12.

Cupach

W. R.

Spitzberg

B. H.

(2000). Obsessive relational intrusion: Incidence, perceived severity, and coping. Violence and Victims, 15, 357-372.

13.

Dietz

N. A.

Martin

P. Y.

(2007). Women who are stalked: Questioning the fear standard. Violence Against Women, 13, 750-776.

14.

Dressing

Kuehner

Gass

(2006). The epidemiology and characteristics of stalking. Current Opinion in Psychiatry, 19, 395-399.

15.

Esbensen

F. A.

Winfree

L. T.

Jr. He

Taylor

T. J.

(2001). Youth gangs and definitional issues: When is a gang a gang, and why does it matter? Crime & Delinquency, 47, 105-130.

16.

Fisher

B. S.

Cullen

F. T.

Turner

M. G.

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

17.

Fisher

B. S.

Cullen

F. T.

Turner

M. G.

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

18.

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.

19.

Fremouw

Westrup

Pennypacker

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

20.

Leiter

R. A.

(2008). National Survey of State Laws (6th ed.). Detroit, MI: Thomson Gale.

21.

Logan

T. K.

Leukefeld

Bob Walker

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

22.

Lauritsen

J. L.

(2001). The social ecology of violent victimization: Individual and contextual effects in the NCVS. Journal of Quantitative Criminology, 17, 3-32.

23.

McCreedy

K. R.

Dennis

B. G.

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

24.

Melton

H. C.

(2007). Closing in: Stalking in the context of intimate partner abuse. Sociology Compass, 1, 520-535.

25.

Missouri State Statutes § 455.01, et seq.

26.

Mullen

P. E.

Pathé

Purcell

(2000). Stalkers and their victims. Cambridge, UK: Cambridge University Press.

27.

Mustaine

E. E.

Tewksbury

R. A.

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

28.

National Center for Victims of Crime. (2007). Model stalking code revisited: Responding to the new realities of stalking. Washington, DC: Stalking Resource Center.

29.

National Criminal Justice Association. (1993). Project to develop a model anti-stalking code for states. Washington, DC: U.S. Department of Justice, National Institute of Justice.

30.

Nicol

(2006). Stalking. London, England: Reaktion Books.

31.

Nobles

M. R.

Fox

K. A.

Piquero

A. R.

(2009). Career dimensions of stalking victimization and perpetration. Justice Quarterly, 2, 476-503.

32.

North Dakota Century Code § 12.1-17-07.1.

33.

Purcell

Pathé

Mullen

P. E.

(2004). Stalking: Defining and prosecuting a new category of offending. International Journal of Law and Psychiatry, 27, 157-169.

34.

Sheridan

L. P.

Blaauw

Davies

G. M.

(2003). Stalking: Knowns and unknowns. Trauma Violence Abuse, 4, 148-162.

35.

Spitzberg

B. H.

Nicastro

A. M.

Cousins

A. V.

(1998). Exploring the interactional phenomenon of stalking and obsessive relational intrusion. Communication Reports, 11, 33-48.

36.

Tjaden

(2003). Prevalence and characteristics of stalking. In Brewster

(Ed.), Stalking: Psychology, risk factors, interventions, and law. (pp 1-19). Kingston, NJ: Civic Research Institute.

37.

Tjaden

(2009). Stalking policies and research in the United States: A twenty year retrospective. European Journal of Criminal Policy Research, 15, 261-278.

38.

Tjaden

Thoennes

(1998). Stalking in America: Findings from the National Violence Against Women Survey (NCJ 169592). Washington, DC: National Institute of Justice and Centers for Disease Control and Prevention.

39.

U.S. Department of Justice. (2009a). Bureau of Justice Statistics. National Crime Victimization Survey: Stalking Victimization Supplement, 2006 [Computer file]. ICPSR 20080-v1. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2009-01-16. doi: 10.3886/ICPSR20080.v1

40.

U.S. Department of Justice. (2009b). Bureau of Justice Statistics. National Crime Victimization Survey, 2006 [Record-Type Files]. ICPSR22560-v3. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2009-08-25. doi:10.3886/ICPSR22560.v3

41.

Westrup

Fremouw

W. J.

(1998). Stalking behavior: A literature review and suggested functional analytic assessment technology. Aggression and Violent Behavior, 3, 255-274.