Intimate Partner Violence and Stalking Behavior: Exploration of Patterns and Correlates in a Sample of Acutely Battered Women

Abstract

The aims of this study were to provide descriptive data on stalking in a sample of acutely battered women and to assess the interrelationship between constructs of emotional abuse, physical violence, and stalking in battered women. We recruited a sample of 114 battered women from shelters, agencies, and from the community at large. Results support the growing consensus that violent and harassing stalking behaviors occur with alarming frequency among physically battered women, both while they are in the relationship and after they leave their abusive partners. Emotional and psychological abuse emerged as strong predictors of within- and postrelationship stalking, and contributed a unique variance to women’s fears of future serious harm or death, even after the effects of physical violence were controlled. The length of time a woman was out of the violent relationship was the strongest predictor of postseparation stalking, with increased stalking found with greater time out of the relationship. Results suggest the need to further study the heterogeneity of stalking and to clarify its relationship to constructs of emotional and physical abuse in diverse samples that include stalked but nonbattered women, as women exposed to emotional abuse, and dating violence.

Intimate partner violence has been deemed one of the most pressing public health concerns affecting women of all ethnic, racial, and socioeconomic backgrounds (Biden, 1993; Koss et al., 1994; Wilson & Daly, 1993). Results of a large nationally representative survey of 8,000 adult women and 8,000 adult men underscore the fact that most violence against women is committed by current or former intimate partners or dates (Tjaden & Theonnes, 1998a). Overall, nearly 25% of female survey respondents (vs. 8% of males) reported being raped and/or physically assaulted by a current or former intimate partner or date during their adult lives. Moreover, more than three-fourths of the women reporting incidents of rape and/or physical assault during their adult lives identified current or former intimate partners or dates as the perpetrators of these acts.

In contrast to other forms of violent victimization, intimate partner violence is remarkable for its serial and repetitive nature, with acts of actual or threatened violence often continuing after separation or divorce, at times ceasing only upon the death of one or both parties (Browne, 1987; Campbell, 1992; Ellis & DeKeseredy, 1997; Kurz, 1996; Wilson & Daly, 1992, 1993). Mahoney (1991) coined the term “separation assault” to highlight the issues of power and control underlying a batterer’s use of actual or threatened violence to keep his partner from physically or emotionally separating from him or to retaliate for her efforts to do so.¹ Additional support for this concept can be gleaned from research documenting increased rates of violence, particularly lethal violence upon perceived, attempted, or actual separation of women from their abusive partners (Bernard, Vera, Vera, & Newman, 1982; Campbell, 1992; Sev’er, 1997; Wilson & Daly, 1992; 1993); and from the more recent lines of research linking attachment theory to intimate partner abuse (Dutton, 1998; Dutton & Holtzworth-Monroe, 1997; Dutton, Saunders, Starzomiski, & Bartholomew, 1994; Holtzworth-Monroe, Stuart, & Hutchinson, 1997; Kesner, Julian, &McKenry, 1997; Murphy & Hoover, 1999) and stalking behavior (Kienlen, 1998).

Attachment theory provides an interesting conceptual framework for understanding the seemingly illogical persistence of pursuit/stalking behaviors by romantic partners in the face of repeated, clear-cut signs of resistance or rejection from their partners, ranging from verbal explanations to formal legal indicators of rejection (e.g., civil orders of protection, divorce decrees, or even remarriage). From an attachment perspective, the intense scrutiny, monitoring and harassing behavior engaged in by batterers can be conceptualized as proximity-seeking behavior designed to reestablish a secure base in the face of perceived or actual threats of separation (Bowlby, 1980). Histories of early attachment disruptions in childhood, as well as perceived losses or separation in the course of adult intimate relationships, have been identified as risk factors for engaging in stalking behavior (Kienlen, 1998), for committing intimate partner violence (Dutton, 1998; Dutton & Holtzworth-Monroe, 1997; Dutton et al., 1994; Holtzworth-Monroe et al., 1997), and for emotionally abusive behavior perpetrated in dating relationships (Murphy & Hoover, 1999). Murphy and Hoover (1999) found that anxious preoccupation with attachment-related issues was most strongly associated with a form of emotional abuse the authors labeled “restrictive engulfment,” which consisted of restrictive, isolating behaviors and acts of jealousy and possessiveness aimed at reducing perceived threats to the relationship.

The construct of emotional/psychological abuse has begun to receive increasing attention in the literature for its deleterious impact and for its relationship to physical aggression (see O’Leary, 1999, for a review). Despite some variation in how the construct of emotional abuse has been articulated across studies and measures, one theme that emerges is the consistent reference to coercive and controlling tactics that instill fear, as well as low-level surveillance behaviors that monitor and/or restrict a partner’s autonomy and freedom of movement. While recent investigations have begun to assess the many important relationships between psychological and physical aggression in female victims of intimate partner and dating abuse, stalking behavior has not been included in definitions of either construct. It is quite possible that when considered within the context of current or former romantic relationships, stalking behavior represents a severe form of emotional/psychological abuse.

In fact, recent research found that most victims of stalking are women (4 out of every 5) and that the majority of female stalking victims (59%) report being stalked by a current or former intimate partner (Tjaden & Theonnes, 1998b). Data from the only large national study of stalking (National Violence Against Women Study: NVAWS) identified a crucial link between stalking and intimate partner violence, finding that 81% of women who were stalked by a current or former (marital/cohabiting) partner also experienced physical assaults by those partners. A smaller number (31%) also reported being sexually assaulted by them. Moreover, women who were stalked by former intimate partners were significantly more likely to experience emotional abuse by those partners, compared to women who were not stalked by former partners. These findings led Tjaden and Theonnes (1998b) to conclude that there is compelling evidence of the link between stalking and controlling and emotionally abusive behavior in intimate relationships (p. 8).

To build upon the emergent literature bridging the gap between research on stalking behavior and intimate partner abuse and violence (Burgess et al., 1997; Coleman, 1997; National Institute of Justice, 1998; Tjaden & Theonnes, 1998a; Walker & Meloy, 1998), the goal of the present study was to provide a detailed picture of stalking behavior, patterns, and correlates in a sample of recently battered women. Specifically, we were interested in exploring the relationship between stalking, psychological/emotional abuse, and physical violence, in light of the considerable conceptual overlap in behaviors defined as part of the constructs of emotional/psychological abuse and stalking.

While some of the questions addressed in this paper were exploratory, we were able to offer several a priori hypotheses based on prior research and/or theory. First, given the findings of the NVAWS, we expected stalking behaviors to occur with a relatively high frequency within this sample of acutely battered women. Second, based on research highlighting the ongoing fear experienced by victims who are threatened with repeated violence and harassment (Dutton, 1992; Meloy, 1998), we expected to find that experiences of stalking would significantly contribute to battered women’s expectations of future violence and future harm from their partners. Expectations of future harm and future violence are important dimensions to study because they affect how a victim might appraise and respond to a climate of impending violence. Moreover, such perceptions are critical in self-defense cases of battered women who have severely or lethally injured their partners. Consequently, it is important to understand whether (and how) stalking contributes to ongoing fears that violence will recur and escalate to dangerous levels.

Third, we hypothesized that emotional abuse variables would predict the severity of stalking, even after controlling for the impact of physical violence in the relationship. The prediction was based on the notion that stalking represents an extreme form of emotional abuse that would contribute to stalking behavior independent of physical violence. Clinical and empirical definitions of emotional/psychological abuse include dimensions that closely resemble stalking behavior. Specifically, isolation and domination (Marshall, 1999; Murphy &Hoover, 1999;Sonkin, 1997;Tolman, 1999), restrictive or monopolizing behavior (Murphy & Hoover, 1990; Sonkin, 1997), pathological jealousy (Follingstad et al., 1990); surveillance and monitoring behavior (Marshall, 1999) constitute forms of emotional abuse that have been found to exist both with and without co-occurring physical violence. Finally, emotionally abusive and controlling behavior was reported to occur more often among women with ex-husbands who stalked them, compared to women whose ex-husbands did not stalk them (Tjaden & Theonnes, 1998). In his discussion of research on populations with co-occurring physical and psychological abuse, O’Leary (1999) commented that “more studies assessing the relative contribution of psychological and physical aggression are needed (p. 18).” Our analyses will address the relative impact of physical and emotional abuse variables.

Finally, in light of the research on separation violence (e.g., Campbell, 1992; Mahoney, 1991) and recent loss as a precipitant to stalking (Kienlen, 1998), and the description offered by Walker and Meloy (1998) that “stalking is the name given to the combination of activities that batterers do to keep the connection between themselves and their partners from being severed (p. 142),” we predicted that separation from an abusive partner be uniquely associated with escalation in postseparation stalking behavior, even after controlling for prior stalking and prior abuse.

METHOD

Participants

Data reported in this paper were collected as part of a larger study focusing on factors that influence recovery from intimate partner violence in a sample of acutely battered women. This paper will present data related to stalking and intimate partner violence from the first 114 battered women who participated in the study.

Recruitment/Screening Criteria

In order to obtain a heterogeneous sample of acutely battered women, a variety of recruitment strategies were employed. All local shelter and nonshelter agencies serving battered women were provided with information about the study and were asked to assist in recruitment. They were asked to inform their clients about our project and provide them with a brief verbal or written description of the study and a telephone number to contact a member of our staff for more information. Assistance was also sought from local police departments, hospital emergency rooms and other victim service agencies that might come into contact with battered women, but do not do so exclusively. These agencies were also provided with written material that could be handed out, mailed or described verbally to battered women.

To access non-help-seeking battered women, efforts were made to work with the media to produce stories or shows about domestic violence that could be paired with information about the study that might attract potential volunteers. To this end, a number of talk radio programs, newspaper articles, and television news stories were produced as a vehicle for recruiting study participants. Finally, staff members participated at many events geared toward women (e.g., working women’s survival show) or victims, (e.g., victims-rights-week rally) by staffing a booth and/or handing out informational flyers briefly describing the study.

All prospective participants, irrespective of the method of recruitment, were asked to contact our staff using a specially designated phone line to find out more about the study, and if interested, to participate in a brief screening interview over the telephone. Careful attention was paid to issues of confidentiality and safety planning at every level, from the greeting used on the telephone to the policy of leaving messages for prospective participants who phoned us.

To recruit a sample of battered women who experienced recent, serial, intimate partner violence, several screening criteria were employed. The following criteria were used to screen potential participants:

1. length of relationship;

2. recency of violence; and

3. severity of violence.

First, participants were required to have been in an intimate relationship, whether cohabiting or not, for a minimum of 3 months, effectively ruling out dating violence. Second, to improve accuracy of reporting, it was required that the most recent episode of violence occurred within the past 6 months. However, if the most recent episode occurred less than 2 weeks earlier, participants were scheduled so that there were at least 2 weeks between the most recent episode and the assessment. This was done to reduce potential inflation of scores on symptom measures as a consequence of assault recency. Finally, in order to obtain a sample of women who experienced more than an occasional episode of relationship violence, we required that participants experience at least four incidents of minor violence or two episodes of severe violence (or some combination of four incidents of minor and/or severe violence) within the past year. Minor violence items were: being pushed, shoved or grabbed; being slapped or hit; having things thrown at you that could hurt; having your arm twisted or your hair pulled. Severe violence items were: being hit or punched with a fist or with something that could hurt; anything that caused you to have physical injuries; being choked, slammed against a wall or thrown downstairs; being kicked or beaten up; threatened with a weapon; having a weapon used against you; being forced to have sex when you did not want to; and causing you to fear for your life or the lives of your family members.

Participants who were ruled out of the study based on their telephone screening were given support, thanked for their time, and were provided with information about appropriate resources in the community. Twenty-four women were screened out of the study for the following reasons: 2 women were with their partners for less than three months; 6 women reported fewer than the required number of episodes of physical violence; 15 women reported abuse that occurred more than 6 months ago (and for some women the abuse ended many years ago); and 1 woman declined to participate after hearing more about the study.

Instruments

The Stalking Behavior Checklist (SBC: Coleman, 1997)

The SBC is a 25-item inventory assessing a variety of unwanted harassing and pursuit-oriented behaviors. Each item is rated on a 6-point frequency scale, ranging from 1 (never) to 6 (once a day or more). On our version of the SBC, participants were asked to rate each item by focusing on unwanted contact during the past 6 months by their (most recent) abusive partner. The original version of the SBC inquired about any former dating partner’s use of these tactics following the breakup of a romantic relationship. The SBC was originally developed using a small sample of college students (N = 141), 13 of whom were classified as having been stalked, 38 of whom were classified as having been harassed, with 90 serving as control subjects on the basis of their responses to the author’s screening questions. The SBC was factor-analyzed resulting in two subscales, Violent Behavior (VB) with 12 items accounting for 34.7% of the variance, and Harassing Behavior (HB) with 13 items, accounting for 10.8% of the variance. The Violent Behavior subscale consists of items addressing overt acts of violence (e.g., broke into your home or car, violated a restraining order, etc.). The Harassing Behavior subscale consists of items reflecting nonviolent harassment, such as unwanted telephone calls, gifts or visits, being followed, and so forth.

Due to our modest sample size, we were unable to evaluate whether the factor structure could be replicated within our sample of battered women. Consequently, we decided to combine the subscales into a single measure of stalking for analyses using continuous scales. Because of the considerable conceptual and empirical overlap between 3 of the violence items on the SEC, and the items assessing physical violence (described later), 3 violence items were dropped from the SBC in all analyses using the SBC. The 3 deleted items were “threatened to cause you harm,” “attempted to harm you,” and “physically harmed you.” Coefficient alpha for the 23-item measure was .90.

The Standardized Battering Interview

This interview consists of a variety of questions assessing demographic and abusive relationship characteristics, including recent (past month) stalking behavior experienced by women who left their partner. Embedded within the interview are two questions focusing on a woman’s appraisal of future violence (Dutton, 1992). The first question asks the woman to rate her appraisal of the likelihood that violence by her partner will recur in the future (APV-FV). The second item asks the woman to estimate her likelihood of experiencing serious physical harm or death by her partner at some point in the future (APV-FH). Both appraisal questions are rated on a 7-point scale, anchored from 1 (not at all likely) to 7 (very likely). The interview also contains 10 questions assessing the frequency of recent (past month) stalking behaviors assessed only among women who have left the relationship. Postseparation stalking items include: threats to life, threats to children’s lives, threats of custody, threats to kidnap children, telephone and in-person harassment at work, harassment at home, stalking, physical assault, and sexual assault. Items are rated on a 5-point scale for the past month (0 = never; 1 = once or twice; 2 = once or twice a week; 3 = several times a week; 4 = daily or almost every day).

A summary of postseparation stalking was created by dichotomizing (and then summing) each of the seven non-child-related stalking items. The items were dichotomized before summing because of the restricted range on the individual items. The child-related items were left off because of missing data for the women without children living with them. Internal consistency as measured by coefficient alpha was .80 for the 7-item measure of postseparation stalking.

Psychological Maltreatment of Women Inventory—Abbreviated Version (PMWI)

The abbreviated 14-item version of the PMWI (Tolman, 1998, 1999) consists of 2 factor-derived subscales that measure dominance/isolation (DI) and emotional and verbal abuse (EV). Evidence of reliability and validity are presented in Tolman (1999). The scale is a self-report measure, and each item is rated on a 5-point frequency scale, ranging from never (1), to very frequently (5). Each subscale consists of 7 items. Coefficient alphas for both subscales in the present sample were .88.

Revised Conflict Tactics Scale-2 (CTS-2)

Three subscales of the revised CTS-2 (Straus, Hamby, Boney-McCoy, & Sugarman, 1995) were administered to assess the frequency and severity of physical assault (CTS-PA; 12 items), injury (CTS-I; 6 items), and psychological aggression (CTS-PSYCH; 8 items). Ratings are made in terms of frequency (0 = never 1 = one in past year, 2 = twice in past year; 3 = 3–5 times in past year; 4 = 6–10 times in past year; 5 = 11–20 times in past year; 6 = more than 20 times in past year). The authors of the CTS-2 suggest creating a severity index by adding the midpoint for each item and creating a summed score for each subscale. The midpoint equals the rating for ratings of 0, 1, and 2. Scores of 3 are recoded to 4, scores of 4 are recoded to 8 and scores of 5 are recoded to 15, and scores of 6 are recoded to 25. Coefficient alphas for the CTS-PA and CTS-PSYCH were .90 and .80, respectively. Coefficient alpha for the 6-item injury scale was .62, with one item having an item-total scale correlation of −.03 (“I had a broken bone from a fight with my partner”). This item was dropped, resulting in a 5-item scale with an alpha of .66.

Procedure

Participants who met study criteria and agreed to participate completed the study in two visits that typically occurred within several days of each other. On day 1, women first completed several symptom-based measures programmed onto a laptop computer in order to reduce the likelihood that symptom scores would be elevated as a consequence of discussing traumatic material. Next, they were interviewed by trained master’s- or doctoral-level clinicians with extensive experience dealing with traumatized populations. On day 2, participants completed a battery of nonsymptom-based self-report instruments that were programmed onto a laptop computer. Debriefings were conducted with participants following completion of all instruments.

RESULTS

Demographic and Relationship Characteristics of the Sample

Battered women in our sample averaged 35 years of age (SD = 7.9), ranging from 18 to 59 years old. In terms of race, 68% of the participants were African American; 25% were Caucasian; 2% were Latino; 3% were Native American; and 3% identified their ethnicity as “other.” On average, the women completed slightly more than a high school education (M = 12.9; SD = 2.0), with education ranging from 8 through 19 years. Annual household income was reported by a subgroup of participants (n = 111) and was as follows:

1. less that $5,000 (12%);

2. $5,000–$10,000 (20%);

3. $10,001–20,000 (17%);

4. $20,001–$30,000 (22%);

5. $30,001–$50,000 (14%); and

6. $50,000+ (14%).

The length of time women reported being in a relationship with their abusive partners ranged from a minimum of 6 months to a maximum of 27 years, with 7.3 years being the mean duration of the relationship (SD = 6.4). The duration of abuse averaged 5.37 years (SD = 5.79 years), and ranged from 2 months to more than 24 years. Three women identified their abusive partners as female, whereas the remainder reported male partners. Sixty percent of the battered women had one or more children under age 18 residing with them, with a mean of 1.2 (SD = 1.3) children per woman (range = 0–6).

At the time of the interview, 35% of the women were married to their abusive partners, 13% were separated/divorced; 40% were cohabiting; and 11 % described a non-cohabiting dating relationship. However, only 14% of the sample reported that they were currently living with their partners at the time of study participation, and a total of 17% indicated that they had not yet left the relationship. Of the 95 women who left their partners, the average length of time since leaving was 3 months (SD = 7.2 months), ranging from 4 days to 4 years. Women reported making many prior attempts to leave the abusive relationship, with 16% of the sample (n = 18) reporting that they left the relationship more times than they could count. Only 3 women (2.6%) reported making no attempts to leave their partners. Forty percent of the sample made 1 to 5 attempts to leave; almost one-quarter (23.2%) left 6–10 times, and 10% of the sample made 11–40 attempts to leave. Slightly more than half of the sample (52%) resided in a shelter for battered women at the time of participation in the study.

Stalking Behavior Frequencies

To obtain a detailed picture of how often each type of stalking behavior occurs, frequencies for each of the 12 items on the Violent Behavior subscale of the SBC are presented in Table 1. The time frame for these questions referred back to the past 6 months. To ascertain the relative frequency of violent behaviors, each item was recoded into a dichotomous score (never vs. at least once in the past 6 months). The stalking behaviors most commonly reported were: threatened harm (94%); physical harm (89%); attempts to harm (88%)²; and stole/read mail (61%). Less commonly experienced events were: violated orders (36%); threats of self-harm (33%); attempted and actual home break-ins (31%, 30%, respectively); attempted car break-ins (29%), physical acts of self-harm (28%); and damage to property of a new partner (11%). Rank ordering of these acts is presented in parentheses in Table 1.

TABLE 1.

Frequencies of Violent Behavior Items on the Stalking Behavior Checklist

Item	Frequency
	Never	<1	2–3	1–2	3–6	>1
		Month	Month	Week	Week	Day
(8) Broke into your home	71%	23%	5%	1%	1%	0%
(5) Violated a restraining order	64%	13%	13%	2%	5%	3%
(9) Attempted to break into your car	71%	24%	3%	1%	1%	0%
(1) Threatened to cause you harm	6%	25%	28%	17%	13%	11%
(11) Broke into you car	79%	17%	4%	0%	1%	0%
(3) Attempted to harm you	13%	28%	26%	15%	9%	10%
(2) Physically harmed you	11%	38%	27%	13%	5%	6%
(7) Attempted to break into your home	68%	20%	5%	2%	4%	2%
(10) Physically harmed himself	73%	18%	7%	1%	0%	2%
(4) Stole/read your mail	39%	22%	10%	6%	12%	11%
(12) Damaged the property of your new partner	89%	8%	1%	0%	2%	0%
(6) Threatened to harm himself	67%	21%	5%	4%	4%	0%

Numbers in parentheses refer to the rank ordering of the item; N = 112.

Frequencies for each of the 13 items on the Harassing Behavior subscale are listed in Table 2. Each harassing behavior item was also recoded to obtain the relative ranking of behaviors. Very commonly reported behaviors were: being watched (71%); receiving unwanted calls at home (66%); being followed (63%); unwanted visits at home (62%); hang-up calls (58%); unwanted messages (56%); and unwanted visits at work (42%). Less frequent forms of harassment were: unwanted gifts (29%); unwanted letters (27%); threats made to a new partner (18%); unwanted photos (8%); and harm to a new partner (6%). Rank ordering of these acts is presented in parentheses in Table 2.

TABLE 2.

Frequencies of Harassing Behavior Items on the Stalking Behavior Checklist

Item	Frequency
		<1	2–3	1–2	3–6	>1
	Never	Month	Month	Week	Week	Day
(2) Made calls to you at home when you didn’t want him to	34%	15%	11%	9%	14%	17%
(4) Came to your home when you didn’t want him to	38%	16%	10%	15%	11%	10%
(3) Followed you	38%	21%	14%	10%	10%	8%
(5) Made hang-up telephone calls	43%	13%	11%	10%	5%	18%
(9) Sent your unwanted gifts	71%	16%	4%	5%	3%	2%
(7) Made calls to you at work/school when you didn’t want him to	49%	16%	7%	10%	9%	9%
(1) Watched you	29%	18%	12%	12%	16%	14%
(8) Came to your workplace/school when you didn’t want him to	58%	13%	11%	6%	5%	7%
(6) Left unwanted messages on answering machine, voice mail or email	44%	9%	11%	4%	16%	17%
(12) Sent photographs when you didn’t want him to	92%	5%	2%	0%	2%	0%
(11) Made threats to your new partner	82%	7%	2%	4%	4%	2%
(10) Sent letters to you when you didn’t want him to	73%	13%	6%	1%	3%	4%
(13) Harmed your new partner	94%	4%	1%	1%	1%	0%

Numbers in parentheses refer to the rank ordering of the item; N = 112.

Dichotomized SBC items were summed to create an index reflecting the total number of different stalking and harassing behaviors experienced by a woman over the past 6 months. With 25 items, the possible range of scores was from 0–25. On average, women reported experiencing 11 (SD = 5.4) different acts of stalking and harassing behaviors during the past 6 months, with scores ranging from 0–23. Finally, frequencies of stalking behavior experienced within the last month were assessed for all women who left the relationship. These data are presented in Table 2.

In addition to the stalking items from the SBC that were measured for the past 6 months, separated women (n = 94) were also asked about the frequency of stalking over the course of the past month (see Table 3). They were most likely to report experiencing threats to their lives (37%) and being harassed at home on the telephone (34%) within the past month. For 9% of the women, telephone harassment at home occurred every day or nearly every day. When asked directly whether they were “stalked” within the past month, slightly more than one-quarter of the separated women (29%) endorsed that label. Fifteen percent reported one or two incidents of stalking, 5% reported being stalked one to two times per week, 5% indicated several times per week, and 3% reported being stalked on a daily or near daily basis.

TABLE 3.

Frequencies of Stalking Behavior for Separated Women in Past Month

Item n = 94	Frequency
	Never	1–2*	1–2/wk	Several/wk	Daily
Threatened your life/well-being	63%	23%	3%	6%	4%
Threatened the life or well-being of your children	90%	8%	0%	2%	0%
Threatened to gain custody of your children	70%	15%	3%	5%	7%
Threatened to kidnap your children	83%	12%	0%	3%	2%
Harassed you at work by phone	80%	9%	1%	7%	3%
Harassed you at work in person	86%	9%	4%	1%	0%
Harassed you at home by phone	66%	12%	7%	7%	9%
Stalked you	71%	15%	5%	5%	3%
Physically assaulted you	84%	14%	1%	1%	0%
Sexually assaulted you	94%	4%	2%	0%	0%

Several analyses were conducted to determine whether demographic characteristics and shelter status were associated with stalking rates. There were no differences in the rates of stalking as a function of shelter status, t(110) ≤ 1.0, ns There were also no differences in stalking for African American and Caucasian women, t(102) ≤ 1.0, ns Similarly, rates of stalking were comparable among dating, married, cohabiting, and separated or divorced battered women, F(3, 111) = 1.5, ns Stalking did not differ as a function of economic status, F(5, 107) < 1.0, ns.

Data Analysis Plan

First, we will present descriptive findings on the continuous measures, followed by correlations between the variables used in regression models. Next, the findings from a series of regression analyses will be presented to test hypothesized relationships between stalking and its effect on expectations of future violence and future harm:

1. more extensive stalking will significantly predict women’s expectations of future violence;

2. more extensive stalking will predict battered women’s expectations of future harm from their partners.

Next, we evaluated the hypothesis that emotional abuse variables would predict the severity of stalking, even after controlling for the impact of physical violence in the relationship. Finally, we examined the prediction that separation from an abusive partner be uniquely associated with escalation in postseparation stalking behavior, even after controlling for prior stalking and prior abuse. Hierarchical multiple regression analyses were chosen to examine and isolate

1. the unique role of stalking in contributing to the ongoing fear of future violence and serious harm, and

2. the unique predictors of past and recent stalking behavior.

Means and standard deviations for all measures used in the study are presented in Table 4. Modest correlations were obtained between the emotional abuse and stalking variables. Stalking was most strongly correlated with the dominance/isolation subscale of the PMWI (r = .47). As expected, the highest correlation was found between the physical assault and injury variables (r = .79). Correlations among the predictor variables used in the regression analyses are presented in Table 5.

TABLE 4.

Means, Standard Deviations for Continuous Measures

Scale	M	SD	Min.	Max.
SBC	42.62	16.80	22	97
PMWI-EV	28.71	6.04	12	35
PMWI-DI	26.46	7.35	10	35
CTS-I	27.16	22.34	5	87
CTS-PA	82.84	65.49	12	256
CTS-PSYCH	106.90	48.49	11	200
APV-FV	4.45	2.47	1	7
APV-FH	5.16	2.26	1	7
Postseparation stalking	1.49	1.80	0	6

Note. SBC = Stalking Behavior Checklist; PMWI-EV = Psychological Maltreatment of Women Inventory-Emotional Violence Subscale; PMWI-DI = Psychological Maltreatment of Women Inventory-Dominance/Isolation Subscale; CTS-I = Conflict Tactics Scale 2-Injury Subscale; CTS-PA = Conflict Tactics Scale 2-Physical Assault Subscale; CTS-PSYCH = Conflict Tactics Scale 2-Psychological Aggression Subscale; APV-FV = Appraisal of Violence-Future Violence Item; APV-FH = Appraisal of Violence-Future Serious Harm/Death.

TABLE 5.

Correlations Among Abuse and Stalking Predictor Variables

Scale	1	2	3	4	5
1. SBC
2. PMWI-EV	.36**
3. PMWI-DI	.47**	.63**
4. CTS-I	.32**	.41**	.46**
5. CTS-PA	.35**	.38**	44**	.79**
6. CTS-PSYCH	.38**	.67**	.40**	.48**	.59**

Note. SBC = Stalking Behavior Checklist; PMWI-EV = Psychological Maltreatment of Women Inventory-Emotional Violence subscale; PMWI-DI = Psychological Maltreatment of Women Inventory-Dominance/Isolation Subscale; CTS-I = Conflict Tactics Scale 2-Injury Subscale; CTS-PA = Conflict Tactics Scale 2-Physical Assault Subscale; CTS-PSYCH = Conflict Tactics Scale 2-Psychological Aggression Subscale.

^**p<.01.

Prediction of Expectations of Future Violence and Future Serious Harm/Death

To examine the relative contributions of stalking, emotional abuse and physical violence to expectations of future violence (APV-FV) and to expectations of future harm (APV-FH), two hierarchical multiple regression analyses were conducted on each dependent measure. First, physical assault (CTS-PA) and injury (CTS-I) were entered, followed by the addition of psychological abuse variables on the second step (DI, EV, CTS-PSYCH). Finally, stalking was added (SBC) once the effects of prior emotional abuse and violence were controlled.

Then, the order of entry was reversed to test for unique variance associated with the addition of each set of variables.

Expectations of Future Violence

In the first analysis assessing expectations of future violence, physical assault and injury were added into the model first, followed by the emotional abuse variables and finally, the stalking measure. Entered first, physical assault and injury did not significantly contribute to women’s expectations that future violence will recur, F(2, 106) = 1.3, ns; R² = .02. Consistent with predictions, the addition of the emotional abuse variables significantly increased the amount of variance explained, F(3,106) = 4.0, p = .01, R² change = .10; cumulative R² = .127. The addition of the SBC on the final step did not account for a significant increase in explained variance, F(1,106) = 2.8, p = .10, R² change = .02; cumulative R² = .15.

Reversing the order of entry, when entered first, stalking explained 10% of the variance in expected future violence, F(1,106) = 11.8, p = .000, R² = .10. The addition of the set of emotional abuse variables contributed a nonsignificant increase of 4% of explained variance, F(3,106) = 2.8, ns, R² change = .04, cumulative R² = .15. The final entry of the physical abuse and injury variables did not contribute additional variance to the predictive model, once the effects of stalking and emotional abuse were accounted for, F(2,106) ≤ 1, ns, R² change = .00, cumulative R² = .15. None of the variables entered into the model were significant individual predictors, although stalking evidenced a trend in that direction (t = 1.7, p = .09, β = .19).

Expectations of Future Harm/Death

In the second analysis assessing expectations of serious harm or death, physical assault and injury were added into the model first, followed by the emotional abuse variables and finally, the stalking measure. Entered on the first step, the physical assault and injury variables made significant contributions to explained variance in women’s expectations that they will experience future serious harm or death at the hands of their abusive partners, F(2,106) = 5.4, p = .006, R² = .09. As expected, the emotional abuse variables contributed unique variance to the predictions of expectations of future harm, even after controlling for the effects of physical violence, F(3,106) = 6.6, p= .000, R² change = .15; cumulative R² = .24. Finally, the addition of stalking contributed a nonsignificant increase of 3% of explained variance in expected future harm, F(1,106) = 3.8, p = .055, R² change = .03, cumulative R² = .27.

When the order of entry was reversed and stalking was entered into the model first, stalking explained 14% of the variance in expected future harm, F(1,106) = 16.4, p = .000, R² = .14. An additional 12% of variance was accounted for by the addition of the emotional abuse variables, F(3,106) = 5.7, p = .001; R² change = .12; cumulative R² = .26. No significant improvement in the prediction of expected future harm was made when the physical violence variables were added after controlling for the effects of stalking and emotional abuse, F(2,106) ≤ 1, ns; R² change = .01; cumulative R² = .27. The stalking (SBC), t(106) = 1.9; p = .055, β = .20 and PMWI-emotional/verbal abuse scales were significant independent predictors of expected future harm/death, t(106) = 2.9; p = .00; β = .42.

Prediction of Stalking

A hierarchical multiple regression was conducted on the Stalking Behavior Checklist with the physical assault and injury variables entered as predictors in the first step, followed by the emotional abuse variables on the second step. These physical violence variables accounted for 14% of the variance, F(2,106) = 8.7, p = .000, R² = .14, in SBC scores. Consistent with expectations, the addition of the three emotional abuse variables into the equation contributed additional unique variance to the prediction of stalking, F(3,106) = 9.8, p = .000, R² change = .19, cumulative R² = .34.

Reversing the order of entry into the equation, we first forced in the three emotional abuse variables, which accounted for 33% of the variance in stalking behavior, F(3,106) = 17.24, p = .000. Next, the physical assault and injury variables were forced into the equation, but they did not account for a significant increase in explained variance after controlling for the effects of emotional abuse, F(2,106) ≤ 1.0, n.s., R² change = .003, cumulative R² = .34. The dominance/isolation subscale was the only one of the predictor variables that emerged as an independent predictor of SBC scores, (t = 4.5; p = .00, β =.51).

Separation and Postseparation Stalking

To look at predictors of postseparation stalking, we conducted a hierarchical regression analysis with the women who had left their partners a minimum of 2 weeks ago (n = 75). This cutoff was chosen with the idea that there would be insufficient time to measure postseparation stalking in women who had been out of the relationship less than 2 weeks, and consequently that there would be restricted variance on the range of possible scores for the recently separated women. We tested the effects of prior stalking, length of time out of the relationship, emotional abuse, and physical violence and injury on the prediction of reported stalking over the past month. The predictor variables were entered in the model in the order stated above, and then reversed to assess for unique variance. Due to considerable positive skew, the amount of time since the woman left the relationship was transformed using a log transformation (Tabachnick & Fidell, 1996).

When entered into the equation on the first step, prior stalking accounted for 7% of the variance in postseparation stalking, F(1, 68) = 4.8, p = .03, R² = .069. The log of time since leaving was added next, and it contributed an additional 44% of explained variance, F(1, 68) = 59.5; p = .000, R² change = .44; cumulative R² = 51. The emotional abuse variables made no significant addition to the variance explained, F(3, 68) = 1.3, ns, R² change = .03, cumulative R² = .54. The final addition of physical violence and injury contributed 6% of explained variance, F(2, 68) = 4.6, p = .01; R² change =.06, cumulative R² = .60.

Reversing the entry order, the physical violence and injury variables accounted for only 4% of the variance, F(2,68) = 1.3, ns, cumulative R² = .04. The emotional abuse variables contributed unique variance to the prediction of postseparation stalking, F(3, 68) = 3.4 p = .02; R² change = .13; cumulative R² = .17. The log of the amount of time since the woman left her partner added 41% to the explained variance, F(1, 68) = 62.2, p = .000, R² change =.41; cumulative R² = .59. Prior stalking made no additional contribution to the explained variance once the other predictors were entered into the model, F(1, 68) = 2.1, ns, R² change = .01, cumulative R² = .60. Significant individual predictors of postseparation stalking were CTS-PSYCH, t(67) = 2.9; β = .37) and (log) length of time out of the relationship t(67) = 7.7; β = .65).

DISCUSSION

The aims of this paper were to describe the frequencies and patterns of stalking behaviors experienced in a heterogeneous sample of acutely battered women and to examine the relationship of stalking to emotional abuse and physical violence in battered women. The battered women who participated in this research were in fairly long-term relationships with their abusive partners, though most were not currently living with, or romantically involved with their partners at the time of study participation. Because of the stringency of screening criteria, participants can be described as “battered women” based on the number and severity of abusive incidents they experienced. Moreover, unlike many samples of battered women, only about half of our sample were shelter residents at the time they participated in this research.

Examination of the individual violent and harassing stalking behavior items revealed the staggering number, type, variety, and frequency of stalking behaviors to which battered women are subjected. These behaviors range from the typical types of behaviors expected in a sample of acutely battered women (i.e., threatened, attempted, and actual physical harm), to the more typical pursuit-oriented behaviors generally conceptualized as stalking (e.g., being followed, watched, or receiving unwanted contact). With the number and range of stalking behaviors reported by these women, it is not surprising that many battered women report feeling terrorized. It was distressing to find that nearly half of the battered women reported that their partner violated an order of protection within the past 6 months. This finding is consistent with the 69% of women who reported having an order of protection violated by their male stalker in the NVAWS. The absolute differences in rates may be a function of the shorter interval (6 months) used in our study, compared to the life time criterion used in the NVAWS.

In contrast to the findings from the National Violence Against Women Study (Tjaden & Theonnes, 1998b), in which less than half of the female victims reported receiving overt threats of violence, nearly all of the battered women in our study (94%) reported being threatened by their partners. It is possible that this difference is due to the fact that our participants rated stalking behaviors occurring over the past 6 months, which may have included a period of time in which they were still involved with their partner. However, this is not a likely explanation because the NVAWS reported only 43% of women stalked by intimate partners experienced stalking only after leaving the relationship. The remainder reported being stalked only before the relationship ended (21%) or both before and after it ended (36%). It is more likely that these Findings are a consequence of the differences in samples between the studies. Nonetheless, both sets of findings underscore the very high rates of stalking in intimate relationships that occur across the life span of an abusive relationship.

Physical Abuse, Emotional Abuse, and Stalking: Topographical and Functional Overlap

Topographical similarities between these three constructs include the fact that battering, emotional abuse, and stalking tend to be serial and ongoing and can occur during and after the termination of the romantic relationship. Additionally, aspects of violent stalking (e.g., physical harm, threatened harm) have been historically characterized as battering, while aspects of harassing behavior (e.g., unwanted calls at work, unwanted visits at home, following you) have been historically characterized as emotional abuse. Functional similarities include findings that battering, stalking, and some aspects of emotional abuse appear to be motivated by attempts to control and intimidate the victim and may increase in frequency and/or severity in the context of actual or perceived threats to the security of the attachment. With the increasing acknowledgment of the pervasiveness of stalking within battering relationships, future research will need to examine the heterogeneity of the stalking phenomena and grapple with conceptual/definitional issues and delimit lines of demarcation between these forms of abuse.

Correlations between stalking and the emotional and physical abuse variables tended to be modest, with the strongest relationship occurring between the stalking and the dominance/isolation subscale of the PMWI. In light of the considerable conceptual overlap between these two constructs, this is not a surprising result. Perhaps, stalking represents an extreme form of dominance and control when it occurs in the context of physically violent relationships. Results of our analyses provide support for the idea that stalking is more closely linked with emotional/psychological abuse than it is with physical violence, at least when studied within a sample of acutely battered women. First, the measures of physical violence and injury were unable to predict stalking, once we controlled for the effects of emotional and psychological abuse. Moreover, the dominance/isolation scale emerged as the only significant individual level predictor of stalking. Second, stalking did not uniquely predict fear of future violence and expectations of future serious harm or death, once the effects of emotional abuse were controlled. These results highlight the considerable crossover between stalking and emotional abuse and support the notion that stalking might be conceptualized under the rubric of emotional and psychological abuse. Taken together, the experiences of stalking and emotional abuse create a climate of unrelenting fear that haunts battered women even after they separate from their abusive partners. Undoubtedly, the severe physical violence perpetrated against many of these women by their partners serves to legitimize their fears that stalking behavior may escalate into more serious, life-threatening violence.

Quite possibly, these findings might not be replicated in samples of women who experienced emotional abuse without physical violence, or in samples of less severely battered women, or women exposed to courtship violence. Future research needs to address the topographical overlap between stalking, emotional abuse, and physical violence in samples that are constructed more broadly than ours. Nonetheless, we believe that me results of this study provide compelling support for the close ties between emotional abuse and stalking behavior.

Our results are consistent with the emergent literature focusing on the pivotal role of emotional and psychological abuse in the lives of physically battered women. Sackett and Saunders (1999) found that psychological abuse was a stronger predictor of battered women’s fear of abuse than was physical violence. Marshall (1999) also measured contributors to battered women’s fear of death/serious harm and found that both subtle and overt forms of psychological abuse made independent contributions to fears of future harm. In a study of emotional abuse in dating relationships, Murphy and Hoover (1999) examined a multifactorial model of emotional abuse, comprised of four factors (dominance/intimidation; denigration; hostile withdrawal; and restrictive engulfment) and found that while all four forms of emotional abuse were associated with coercive control in interpersonal relationships, the dominance/isolation and denigration scales were the most strongly related to physical aggression.

Descriptions of nonintimate partner stalking are characterized solely by a multiplicity of unwanted pursuit behavior experienced as (nonphysical contact) harassment. In fact, at least one literature review has defined a subset of stalking as “obsessional following” with relatively low incidence of actual physical violence (Meloy, 1996). Future research needs to examine the possibility that there are subtypes of stalking that appear topographically similar, but that may be functionally different. This may be particularly true when comparing cases of stranger stalking with stalking that takes place in the context of current or former romantic and intimate partnerships.

Our findings, in concert with the developing literature on stalking and emotional abuse, underscore the importance of developing and testing comprehensive theoretical models that incorporate multiple dimensions of coercive and controlling behavior that are studied prospectively in a variety of populations exposed to romantic and intimate partner coercion, abuse, and violence.

Although the impact of battering has often included the effects on physical, emotional, and psychological functioning, little research has addressed the occupational impact. One notable finding in this study was that battered women reported frequent harassment by phone or in person in their places of employment or where they attend school. Historically, most workplaces have been relatively unresponsive to and unaware of the dynamics and impact of intimate partner violence, resulting in battered women losing their jobs as a result of workplace interference by their partners (Friedman, Tucker, Neville, & Imperial, 1996). Losing a job further hinders a battered woman’s efforts to achieve the financial independence necessary to establish a life without her partner. Loss of employment as well as missed workdays were also reported by stalked victims in the NVAWS. The need to provide education and training on stalking and intimate partner violence in employment settings may be an important next step.

Stalking and Separation

In support of the literature highlighting separation as a risk factor for battered women, we too found that stalking behavior escalated among women who left their abusive partners. Even more notable was the finding that the length of time a woman was out of the abusive relationship was the single strongest predictor of postseparation stalking, even after accounting for previous levels of stalking in the relationship. Perhaps, once a woman has left the relationship, it is more difficult (due to proximity) to commit acts of violence against her, but acts of stalking, such as harassment via phone and e-mail are achieved rather easily. Among the women who left their abusive relationships, many continued to experience recent harassing, threatening and stalking behaviors, and these women reported significant fears that their partners would commit lethal or sublethal acts of violence against them. Thus, in spite of the oft heard “Why doesn’t she just leave?” these data support the growing consensus that leaving does not end the violence, and in some cases escalates it. In her discussion of separation violence, Mahoney (1991) redirects us from attending solely to battered women’s efforts to separate from abusive partners. She argues that we need to reframe the question from why doesn’t she just leave, to “why doesn’t he let her go?” Future research and intervention strategies may need to focus on understanding and ameliorating the attachment disruptions in perpetrators of relationship abuse, as it appears that one function of pursuit-oriented behaviors, of which stalking is a particularly virulent form, is to regulate attachment and proximity seeking via coercive control strategies.

Unfortunately, we were unable to document the type, number, and frequency of stalking behaviors that might have emerged as women engaged in the process of extricating themselves from relationships with their abusive partners. Future research might address this by using qualitative and quantitative methods to study in more detail the experience of women in the course of negotiating exits from battering relationships. Findings from this study suggest that stalking of acutely battered women has been a neglected dimension in most studies of battered women and needs to be considered as a pivotal component of intervention and prevention efforts with batterer and victim populations.