Abstract
The present study, as part of a large-scale victim survey, examines the prevalence and nature of stalking in a representative German quota sample (N = 5779). Applying a broad definition of stalking, the lifetime prevalence added up to 15 percent, depending on respondents’ age, gender, and immigrant background, as well as household size and relationship status. Conditional inference trees revealed that gender, relationship status, and household size were key factors in identifying victims of stalking. Offenders mostly committed stalking against the opposite gender. This pertains especially to female (vs. male) victims. In most cases the offender and victim knew each other prior to the stalking. The results are compared with national and international findings. Implications and potential limitations are discussed.
Keywords
Introduction
Since the early 1990s, the phenomenon of stalking has increasingly been of interest in criminological, sociological, and psychological research. The term stalking first appeared in the late 1980s and was primarily used by the media to describe the persistent harassment of another person against his or her will (Mullen and Pathé, 2002). At first, cases with popular victims especially drew attention, but more and more incidents with average citizens as victims became known (Mullen and Pathé, 2002). The public debate led to the introduction of a first anti-stalking law in California in 1991; other US states followed suit. In Germany, anti-stalking legislation was implemented in 2007.
Large-scale epidemiological studies of stalking have mainly been conducted in the USA and the UK (for example, Basile et al., 2006; Black et al., 2011; Budd and Mattinson, 2000; Scottish Government, 2014; Tjaden and Thoennes, 1998; Walby and Allen, 2004; also see Spitzberg, 2002). But the epidemiology of stalking has also been investigated in European countries (for example, Dovelius et al., 2006; FRA, 2014; Matos et al., 2012; Stieger et al., 2008; Van der Aa and Kunst, 2009). For Germany, no nationally representative data on the prevalence of stalking exist so far: Dressing, Kuehner, and Gass (2005) reported first estimates of the prevalence of stalking in Germany. However, only a regional random sample of people living in the area of a mid-size German town was surveyed. More recently, Feltes et al. (2012) conducted an online survey on stalking experiences with female university students from Germany among others. Additionally, the European Union Agency for Fundamental Rights (FRA) collected data on violence against women (including stalking) from women of 28 countries, including Germany. In sum, either existing findings on the epidemiology of stalking in Germany focus on a certain region (Dressing et al., 2005) or the results are limited to a specific educational degree (Feltes et al., 2012) or gender of the participants (Feltes et al., 2012; FRA, 2014). In the present study, a nationally representative quota sample of male and female participants aged 16 to 40 years has been reached, allowing us to report valid prevalence estimates of stalking in Germany, to clarify which individuals are especially at risk of experiencing stalking and which constellations between victims and offenders are typical, as well as to describe key features of stalking in Germany.
Previous research
Characteristics and prevalence of stalking
According to Spitzberg (2002: 262), most legislatures define stalking as ‘an intentional pattern of repeated or ongoing unwanted pursuit that a reasonable person would find fearful or threatening’. Behaviours that can, accumulated, be referred to as stalking include repeated contacting, approaching, or harassing, which extend over a longer period, exceed implicit rules of social interaction, are directed at a specific person, are perceived by the other at least partially, and cannot, or only to a limited extent, be influenced by the other. The reported variance of stalking behaviours ranges from short-term single actions to episodes that extend over several months or years. Furthermore, stalking can stay stable over time in intensity and frequency or change quantitatively and/or qualitatively in the course of time. Other key features of stalking are that the behaviour continues, even though its undesirability has been communicated by the harassed person or a request to stop has been made, and that the victim feels threatened by the behaviour.
Regarding the specific criteria that classify a behavioural pattern as stalking, large disagreement persists in the scientific literature as well as in the various statutory offences of different countries. So, to date, no universally accepted definition of stalking exists (Langhinrichsen-Rohling, 2012). The threshold from socially acceptable behaviour to stalking is individually determined and the subjective perception of victimization does not always coincide with the definition of the relevant criminal offence (see Stadler, 2013). For example, a certain behaviour may not be defined as a crime, but is perceived by the affected person as such. Conversely, a person who might be classified as a victim of crime by law may not recognize the act as a crime and therefore may not perceive herself as a victim. In this vein, Tjaden, Thoennes, and Allison (2000) demonstrated that the prevalence rates for men tripled and for women increased by half if the self-definition as a stalking victim was taken into account. Thus, there indeed exist cases that are experienced as very harassing, even if they are not classified as stalking in the criminal sense. Hence, victim surveys on stalking with large-scale samples and rather broad conceptualizations of stalking are important in order to identify genuine prevalence rates of stalking.
Overall, differences in (scientific and legal) stalking definitions impede comparisons of stalking studies, since different criteria for categorizing stalking victims are used and definitions are based on different legislations. Moreover, differences in the empirical assessment of stalking behaviours should be considered when comparing prevalence estimates.
Prevalence rates
According to the first comprehensive study on the prevalence of stalking in the USA, 8 percent of the surveyed women and 2 percent of the men reported having been a victim of stalking in the past (Tjaden and Thoennes, 1998). These authors defined stalking victims as individuals who had repeatedly experienced at least one of eight specific harassing behaviours. At the same time they had to have experienced a considerable amount of fear to classify the behaviour as stalking. In a more recent US-study, Basile et al. (2006) reported a lifetime prevalence of stalking of 4.5 percent (women: 7 percent, men: 2 percent). Here, individuals were defined as a stalking victim if they had been followed at least once in their life and had perceived this experience at least as dangerous. Based on the legal definition of stalking in Australia (at least two unwanted intrusions causing fear), Purcell, Pathé, and Mullen (2002) reported lifetime prevalence rates of 32 percent for women and 13 percent for men.
The first European prevalence study on stalking stems from the UK (Budd and Mattinson, 2000). Here, 12 percent of the respondents (women: 16 percent, men: 7 percent) reported having already experienced stalking. Prevalence estimates were based on a relatively broad definition: ‘Since you were sixteen, have you EVER been the subject of … persistent and unwanted attention (from people you know or strangers)?’ (Budd and Mattinson, 2000: 111). In a subsequent British study, Walby and Allen (2004: 4) found stalking prevalence rates of 19 percent for female and 12 percent for male respondents based on a stricter definition (‘a course of conduct involving two or more events of harassment causing fear, alarm or distress, of three types: phone calls or letters; loitering outside home or work; damaged property’).
Dovelius et al. (2006) reported a lifetime prevalence of stalking in Sweden of 9 percent using the following definition of stalking: ‘when somebody has been followed or watched by the same person several times, or has had unwanted visits, telephone calls, letters, email, text messages, presents and the like from the same person on several occasions’ (Dovelius et al., 2006: 6). Applying experienced fear as an additional criterion to the definition of stalking led to an increased prevalence rate of 5.9 percent.
According to the first German stalking study (Dressing et al., 2005), the distribution of stalking in the area around Mannheim totalled 12 percent (women: 17 percent, men: 4 percent). This estimate was again based on a relatively broad definition of stalking: people were defined as a stalking victim when they had experienced at least 2 out of 18 fear-evoking harassing behaviours over a period of at least two weeks. Importantly, these prevalence rates are not based on a nationwide representative German sample. Deploying the same stalking definition as Dressing et al. (2005), Stieger et al. (2008) found a lifetime prevalence of stalking of 11 percent in a regional sample from Austria (women: 17 percent, men: 3 percent).
Van der Aa and Kunst (2009) reported a lifetime prevalence of stalking of 16.5 percent (women: 20.7 percent, men: 13.4 percent) based on a Dutch community-based sample from the area around Tilburg. These authors used a working definition of stalking: ‘Have you ever been the target of persistent unwanted attention from another person?’ (Van der Aa and Kunst, 2009: 42).
Overall, regardless of the definition applied, a higher risk of stalking is consistently reported for women compared with men. Accordingly, in a meta-analysis of 108 samples the average lifetime prevalence of stalking was higher for women (24 percent) than for men (11 percent) (Spitzberg, 2002; also see Spitzberg and Cupach, 2007).
Thus, some studies have focused on women as potential victims of stalking (for example, Feltes et al., 2012; FRA, 2014; Freidl et al., 2011): in a comprehensive study on stalking (N = 42,000), FRA compared prevalence rates for women aged 18 to 74 across the 28 member states of the European Union (EU). Overall, a lifetime prevalence (since the age of 15) of 18 percent was found for the EU, varying from 8 percent (Lithuania and Romania) to 33 percent (Sweden). For women in Germany, prevalence estimations of 24 percent were reported. In the FRA study, stalking victimization was defined as having repeatedly experienced at least one of eight harassing behaviours (for example, receiving offensive or threatening emails, text messages, instant messages, letters, cards, or phone calls).
Victim characteristics
In the scientific literature, certain characteristics besides victims’ gender have been described that may increase the risk of experiencing stalking. Budd and Mattinson (2000), for instance, found that students and singles had a particularly high victimization risk, whereas married people, for example, had a lower risk of becoming a victim of stalking. Similarly, in a study by Basile et al. (2006), unmarried, separated and divorced respondents reported higher victimization rates (also see Freidl et al., 2011; Van der Aa and Kunst, 2009). The authors of the FRA study found that single mothers (34 percent) were at an exceptionally high risk of stalking victimization and explained this conspicuousness with the high proportion of previous partners among the offenders. In general, marriage seems to have a protective function against experiencing stalking – probably owing to the considerable amount of stalking victimization by ex-partners (for example, FRA, 2014; Sheridan et al., 2003).
Whereas (female) university students seem to be at an especially high risk of being stalked (see Björklund et al., 2010; Feltes et al., 2012), no clear pattern of stalking victimization with respect to respondents’ educational degree could be detected in the FRA study (FRA, 2014): For women in the EU with basic education, the stalking prevalence was lower (12 percent) than that for ‘women with post-secondary (but not tertiary) education’ (23 percent) (FRA, 2014: 88).
In Tjaden and Thoennes’ (1998) study, stalking prevalence rates in the USA depended on the origin of the female respondents among other things: whereas female Native Americans were more frequently victims of stalking, female respondents with an Asian background were at lower risk. Similarly, Black et al. (2011) reported differences in the distribution of stalking as a function of ethnicity for female respondents but not for male respondents. Walby and Allen (2004) found a higher stalking risk for Asian compared with Caucasian women and a lower risk for African Americans in the UK. Again, in the male sample stalking prevalence varied little by origin. Overall, unique cultural characteristics in values and norms appear to be responsible for differences in victimization risks.
Victim–offender constellations
In particular, the victim’s and offender’s gender and a previous relationship between victim and offender have been discussed as relevant risk factors for stalking victimization. Regarding the offender’s gender in Spitzberg’s (2002) meta-analysis, men are responsible for stalking in the vast majority of cases (also see Spitzberg and Cupach, 2007). Tjaden and Thoennes (1998) reported that 94 percent of the female victims named a male offender, and only 60 percent of the male victims were stalked by other men (also see Budd and Mattinson, 2000; Purcell et al., 2002). In the German sample by Dressing et al. (2005), 91 percent of the female victims were harassed by men, while the proportion of same- and opposite-sex offenders was almost equally distributed in male victims. In the EU-wide stalking study, 71 percent of the female victims named a male offender (FRA, 2014; also see Feltes et al., 2012). Overall, stalking appears to be clearly dominated by male offenders, especially in cases with female victims.
Concerning the victim–offender relationship, Spitzberg (2002) summarized that a previous intimate relationship exists in almost half of the cases, whereas not even one-fifth of the victims are stalked by complete strangers (also see Spitzberg and Cupach, 2007). Dressing et al. (2005) found that three out of four victims knew their offender, in one-third of the cases the offender being an ex-partner. Strangers were responsible for almost a quarter of all stalking victimizations in that study. In the Swedish study, 34 percent of the offenders were strangers to the victims (or the victims did not know who the offender was), whereas 25 percent of the victims were stalked by a partner or ex-partner (Dovelius et al., 2006). According to the results of the Scottish Crime and Justice Survey, 42 percent of the Scottish stalking victims identified a (previous) partner as the offender, whereas 37 percent indicated that they were stalked by somebody they had never seen before or did not know (Scottish Government, 2014). Walby and Allen (2004) reported for the UK that women were considerably more frequently stalked by ex-partners than were men (37 percent vs. 8 percent). 1 With 7 percent for female and 30 percent for male victims, the percentages of strangers as offenders were accordingly lower. The US study by Tjaden and Thoennes (1998) also indicated that women are more frequently stalked by their ex-partners than are men and less frequently by an unknown person.
Duration and frequency
In their recent meta-analysis, Spitzberg and Cupach (2007) reported a mean duration of stalking victimization of 22 months, with a considerable range from four months to seven years. In Tjaden and Thoennes’ (1998) study from the USA, for example, 52 percent of the cases lasted for less than one year, 16 percent ended within one year, 23 percent ranged from two to five years, and 9 percent had a longer duration than five years, resulting in an average duration of nearly two years. For the UK, Budd and Mattinson (2000) reported durations of between up to a week (16 percent) and three or more years (7 percent), although most cases (26 percent) lasted for between one and three months (also see Walby and Allen, 2004). Purcell et al. (2002) found in their Australian study an average stalking duration of eight months. In one-third of the cases the stalking ended within a week; in 15 percent it stopped after one to three months. 13 percent of the victims indicated a duration of more than one year. Dressing et al. (2005) found a duration of less than one month in 17 percent of the German stalking cases; 24 percent of the cases lasted for one year or more. Considering the intensity of the harassment, 32 percent of the victims in the German study reported ‘some’ unwanted contact attempts. Repeated harassment occurred monthly in 8 percent of the cases and weekly in 35 % 9 percent of the victims had been harassed daily, and 16 percent several times a day.
In the EU-wide stalking study (FRA, 2014), the female victims were asked to indicate the duration of the most serious case of stalking since the age of 15. Most of the cases (45 percent) lasted up to three months; 11 percent of the serious cases had a duration of five years or more. Within their female Austrian community-based sample, Freidl et al. (2011) reported an average stalking duration of 16 months, ranging from 1 month to 54 years.
The present study
In the following, the results of the first nationwide German victim survey on stalking are presented. Hence, valid prevalence estimates for the occurrence of stalking victimization in Germany are finally available. According to the reviewed findings, the focus is on prevalence estimates as a function of selected demographic characteristics, identifying relevant risk factors, illustrating ‘typical’ victim–offender constellations and describing the duration and frequency of stalking in Germany. Further evaluations of the consequences of stalking are reported elsewhere (see, for example, Hellmann and Regler, 2014).
More precisely and in accordance with international findings, it is expected that the prevalence of stalking in Germany depends primarily on respondents’ gender and relationship status. Additionally, it is assumed that the ‘typical’ constellation is represented by a female victim and a male offender, with a considerable proportion of ex-partners, confirming results from comparable international studies. Regarding victimization characteristics such as duration and frequency, no divergence from common international studies is expected.
Method
Participants
With respect to the German population aged 16 to 40 years, the quota sample of N = 5779 individuals was representative of the non-institutionalized population of Germany as regards federal state, urban–rural distribution, age (16–40 years), gender, immigrant background (see below for further details), educational degree, and household size, according to national census statistics. With respect to the sampling procedure, it should be noted that combined quotas were used additionally. More specifically, the quota sample is representative (restricted to the aforementioned quota characteristics) not only of the German population in total, but also of the population of a specific federal state (with respect to the aforementioned quota characteristics), for example.
In addition to German people without an immigrant background, individuals with a Turkish or Russian immigrant background were surveyed, since these three groups represent the largest populations in Germany. Referring to the definition of the German Federal Statistical Office, respondents were defined as having an immigrant background if they had Turkish or Russian citizenship themselves or if this pertained to at least one parent at that parent’s birth. According to this definition, a total of 9.5 percent of the respondents had a Turkish immigrant background and 10.0 percent a Russian immigrant background. The proportions of respondents with an immigrant background were oversampled, in order to be able to compare the three groups. This oversampling was adjusted in analyses with the total sample by statistical weighting, so that overall representativeness in accordance with the quota sample characteristics was accomplished.
In addition, the number of respondents aged between 16 and 20 years was oversampled in order to gather differentiated insights into the most recent victimization experiences: It has repeatedly been shown that younger people are more often victims of stalking than are older individuals (for example, Budd and Mattinson, 2000; Purcell et al., 2002; Tjaden and Thoennes, 1998; Walby and Allen, 2004). When analyses of the total sample are reported, this over-representation was also adjusted by a statistical weighting.
The sample comprised 50.8 percent women and 49.2 percent men with a mean age of M = 26.80 years (SD = 7.60). In terms of age, 28.6 percent of the respondents were 16–20 years old, 38.0 percent were aged 21–30 years and 33.4 percent were aged 31–40 years. Most respondents were unmarried (71.2 percent); 23.4 percent were married and 5.4 percent were divorced, separated, or widowed at the time of the survey.
Regarding the level of education, 52.5 percent of the respondents indicated that they possessed either a Certificate of Secondary Education or a middle school degree. Approximately one-fifth (21.4 percent) of the participants had acquired the advanced technical certificate or the German high school diploma at the time of the survey, and about one-tenth held a university degree (10.5 percent); 13.5 percent of the respondents reported that they had not (yet) achieved an educational degree, which was presumably owing to the high proportion of younger respondents in this sample; 2.1 percent of the respondents claimed to hold other educational degrees. In sum, 35.6 percent of the respondents were classified with a low educational background, 32.5 percent with a medium educational level, and 31.9 percent with a high level of education.
With respect to respondents’ residential area, it can be summarized that nearly half of them lived in rather urban areas (49.0 percent), a third (34.3 percent) in rather rural areas, and a sixth in metropolitan areas (16.8 percent). As regards household size, 18.8 percent of the respondents were living alone at the time of data collection, 21.3 percent were living with one other person, 25.1 percent with two, and 23.0 percent with three other people; 11.7 percent of the respondents reported having four or more other people living with them in a household.
Materials
To assess stalking victimization, respondents were asked to select from 17 presented harassing behaviours that they had repeatedly experienced in the past by the same person (Table 1). Accordingly, in the present study individuals were defined as victims of stalking if they had experienced at least 1 of 17 harassing behaviours performed against them by the same person at least twice. This is a rather liberal definition of stalking, for cases may also be classified as stalking that do not meet the statutory criteria for the classification of stalking. However, since all respondents who met the defining criteria answered additional items (for example, on experienced fear owing to the stalking victimization), subsamples of stalking victims depending on stricter definitions can be selected (see Stadler, 2013).
Operationalization of experienced stalking victimization.
Note: Originally, the materials were presented in German.
In addition to the frequency of the incidents (‘fewer than 10 incidents’, ‘10–50 incidents’, and ‘more than 50 incidents’), their duration (‘less than four weeks’, ‘4–12 weeks’, ‘4–12 months’, and ‘more than one year’) was assessed, as was victims’ indications of the offender’s gender. Additionally, the victims were asked to indicate their previous relationship to the offender at the time of the harassment (for example, spouse, ex-spouse, or stranger).
Procedure
Stalking experiences were assessed as part of an anonymous questionnaire on diverse victimization experiences. Between January and May 2011, potential participants were visited at home by employees of an institute specialized in recruiting research participants (GfK). The questionnaire was preceded by a short face-to-face interview in which respondents’ demographics (for example, gender, postal code, federal state, age, educational degree) were assessed, among other things. Next, the interviewer left the participant with the questionnaire on diverse victimization experiences, which was collected at an agreed upon time. 2 The aim of leaving the questionnaire with the respondents was to improve their openness to reporting experienced victimization and to reduce social desirability concerns. The interviewers chose to approach potential participants according to the aforementioned quota characteristics. For a detailed description of the entire victim survey, see Hellmann (2014).
Missing data and statistical procedures
Missing data (missing values per variable: 0.1–2.2 percent) were treated by applying chained equation modelling (White et al., 2011) and including the following variables: gender, age, educational degree, relationship status, origin, residential area, and household size. For individuals who were classified as stalking victims, missing values were estimated using these variables, as well as the available responses regarding the victimization experiences. We used the R package MICE (Multivariate imputation by chained equations; Van Buuren and Groothuis-Oudshoorn, 2011) for this procedure.
Non-parametric conditional inference trees (C-Trees; Hothorn et al., 2006; Strobl et al., 2009) based on the principle of recursive partitioning were applied to analyse associations between socio-demographic characteristics and respondents’ risk of experiencing stalking (for example, Kröger et al., 2014; Rehbein et al., 2015). C-Trees are non-parametric modelling techniques with the capability to deal with large numbers of predictor variables and complex interactions simultaneously. The C-Tree algorithm tests the global null hypothesis of independence between any input variable (gender, age, educational degree, relationship status, origin, residential area, household size) and the response variable (prevalence of stalking) using a permutation test framework. If this hypothesis is rejected, the input variable with the strongest association to the response variable is chosen and a binary split of this variable is implemented. Steps are recursively repeated until the hypothesis is rejected. The stop criterion is based on univariate p < .001. Permutation tests derive p-values from sample-specific permutation distributions of the test statistics.
Results
Prevalence rates
According to the above criteria, 15.2 percent (95% CI [14.3, 16.13) of the 5779 respondents were classified as stalking victims in the present study. 3 The prevalence of stalking depended on respondents’ age, χ²(2, N = 5792) = 19.68, p < .001, Cramér’s V = .058. 4 Bonferroni-adjusted comparisons revealed that respondents aged 16–20 years reported significantly lower victimization rates (11.5 percent) than older respondents (21–30 years: 16.2 % 31–40 years: 15.8 percent), p < .050. 5
In addition, the stalking prevalence was related to respondents’ gender, χ²(1, N = 5769) = 72.16, p < .001, φ = .111, since women (19.4 percent) indicated a significantly higher stalking prevalence than men (11.4 percent). 3 In other words, nearly two-thirds of the victims (64.6 percent) were female. Finally, the risk of experiencing stalking depended on the respondent’s origin, χ²(2, N = 5766) = 18.79, p < .001, Cramér’s V = .057. Respondents with a Turkish immigrant background were less often victims of stalking (9.3 percent) than were German respondents without an immigrant background (15.5 percent) or respondents with a Russian immigrant background (18.1 percent). 6
Simultaneously taking into account respondents’ gender and origin revealed that the gender of respondents with a Turkish immigrant background was not related to their victimization risk, χ²(1, N = 541) < 1. 6 For both men and women of Turkish origin, the lifetime prevalence of stalking was particularly low (Table 2). Overall, stalking prevalence rates did not differ between male respondents with respect to their origin, χ²(2, N = 2999) = 3.32, p = .190. However, the variance in stalking prevalence depended on the female respondents’ background, χ²(2, N = 2767) = 22.16, p < .001, Cramér’s V = .089. The risk of stalking victimization for women with a Turkish immigrant background (10.2 percent) was particularly low compared with German women without immigrant background (19.8 percent) and women with a Russian immigrant background (25.6 percent).
Stalking prevalence rates depending on respondents’ gender and origin (N = 5766) (percent).
Note: Results are weighted for respondents’ age; different indices imply significant Bonferroni-adjusted differences at p < .050.
Respondents who were separated, divorced, or widowed (25.7 percent) were at a significantly higher stalking risk than unmarried (15.6 percent) and married respondents (11.5 percent), χ²(2, N = 5769) = 44.26, p < .001, Cramér’s V = .088. 3 According to Bonferroni-adjusted comparisons, each group’s stalking prevalence differed significantly from the other, p < .050. 3
Furthermore, participants’ household size was associated with their stalking risk: people living alone had a significantly higher risk of becoming stalking victims (20.2 percent) compared with those living with one (14.3 percent), two (15.1 percent), three (11.9 percent), or more than three people (12.7 percent) in a household, χ²(4, N = 5769) = 38.45, p < .001, Cramér’s V = .082.
Next, associations between gender, age, educational degree, relationship status, origin, residential area, and household size as potential explanatory factors and the risk of stalking victimization were tested using a conditional inference tree plot, thus considering interactions between the input variables. Participants’ gender showed the highest association with the risk of experiencing stalking. 7 The right side of Figure 1 shows the risk factors of stalking victimization for female participants. Being female alone was associated with being a victim of stalking in 19.0 percent of the cases (95% CI [17.6, 20.4]). If female participants were also separated, divorced, or widowed, the risk of experiencing stalking in their life increased to 32.1 percent (95% CI [25.8, 38.4]). On the other hand, being female, but not separated, divorced, or widowed, led to a comparably lower lifetime prevalence of stalking victimization of 18.0 percent (95% CI [16.6, 19.4]). For female participants who were not separated, divorced or widowed, the risk of stalking victimization significantly increased if they lived alone to 25.4 percent (95% CI [21.2, 30.0]) compared with female participants who were not separated, divorced or widowed but lived with at least one other person in a household (16.7 % 95% CI [15.2, 18.2]).
Conditional inference tree plot predicting lifetime prevalence of stalking by gender, age, educational degree, relationship status, origin, residential area, and household size (N = 5779).
The left side of Figure 1 shows the risk factors of stalking victimization for male participants. Being male was associated with a stalking prevalence of 10.8 percent (95% CI [9.7, 11.9]). However, male participants who lived on their own had an increased risk of experiencing stalking of 16.8 percent (95% CI [13.9, 19.7]). In contrast, male participants who lived with at least one other person in a household had a considerably lower risk of stalking victimization: 9.1 percent (95% CI [7.9, 10.3]) of all male participants who did not live on their own were identified as victims of stalking. In the multivariate C-Tree analyses, age, educational degree, origin, and residential area were not significantly related to experiencing stalking for either male or for female participants.
Victim–offender constellations
The victims indicated 72.1 percent male and 27.9 percent female offenders. As expected, female victims (93.2 percent) were significantly more likely to be stalked by male offenders than were male victims (33.7 percent), whereas male victims were more frequently stalked by female offenders (66.3 percent) than were female victims (6.8 percent), χ²(1, N = 864) = 348.20, p < .001, φ = .635. Thus, in the majority of the cases, stalking is committed by an offender of the opposite sex, although this pattern especially pertains to female victims. Taken together, the combination of female victim and male offender seems to be the ‘typical’ victim–offender constellation, since it accounted for more than half of all cases (Table 3). Least likely was the combination of female victim and female offender with 4.4 percent of all stalking cases.
Offender’s gender depending on victim’s gender (N = 864) (percent).
With respect to the victims’ responses regarding their relationship to the offender at the time of the victimization, three large groups stood out, which accounted for more than three-quarters of the offenders: ex-partners and former dates (39.6 percent), friends and neighbours (22.3 percent), and strangers (14.2 percent). In 8.5 percent of the cases the offenders were (marriage) partners, followed by 8.1 percent offenders from the professional context; 2.4 percent of the offenders were from the victim’s family (parents, step-parents, or other family members) and 5.0 percent fell into the category ‘other people’. As expected, in the majority of the stalking cases a previous history between offender and victim existed.
Duration and frequency of the stalking victimizations
The duration as well as the frequency of the stalking victimizations varied considerably (Table 4). More than half of the victims reported having experienced fewer than 10 stalking incidents (56.6 percent), over a third stated 10–50 incidents (37.7 percent), and 5.7 percent of the victims had experienced more than 50 incidents. Regarding the duration of stalking, it can be noted that most episodes lasted for less than a year (80.8 percent); in almost a fifth of cases the stalking lasted for longer than a year (19.2 percent). On average the stalking persisted approximately four to six months.
Frequency and duration of stalking victimizations (N = 864) (percent).
Although no relationship could be detected between the stalking duration and the victim’s gender, χ²(3, N = 864) = 1.65, p = .648, the frequency of incidents is clearly related to the victim’s gender, χ²(2, N = 864) = 8.60, p = .014, Cramér’s V = .100. Overall, female victims had to endure more incidents (< 10 incidents: 53.0 % 10–50 incidents: 41.2 % > 50 incidents: 5.7 percent) than male victims (< 10 incidents: 63.1 % 10–50 incidents: 31.4 % > 50 incidents: 5.6 percent). These differences may be due to the fact that male offenders tended to behave more persistently (< 10 incidents: 54.4 % 10–50 incidents: 40.1 % > 50 incidents: 5.5 percent) than female offenders (< 10 incidents: 62.2 % 10–50 incidents: 31.5 % > 50 incidents: 6.2 percent), χ²(2, N = 864) = 5.46, p = .065, Cramér’s V = .080. However, Bonferroni-adjusted comparisons revealed significant differences between female and male offenders only in the two categories with lower frequencies, p < .050, but not in the highest frequency category (> 50 incidents).
Discussion
Although a relatively broad definition of stalking was applied in the present study, the prevalence rates added up to 15 percent in sum and still came under the prevalence of 25 percent reported in Spitzberg and Cupach’s (2007) meta-analysis. Dressing et al. (2005) found a slightly lower prevalence for the German area of Mannheim, with a somewhat narrower stalking definition (minimum duration of two weeks, experienced fear as an additional criterion). 8 Also, in comparison with studies that applied even stricter defining criteria for stalking (for example, Basile et al., 2006; Tjaden and Thoennes, 1998), the prevalence reported in the present study was considerably higher, as might be expected. In sum, the results from this study are in line with previous evidence that prevalence estimates vary largely depending on the underlying definition of stalking (see Dovelius et al., 2006).
In accordance with national and international findings on stalking, women in the present study also showed a significantly greater risk of becoming the victim of stalking (see Black et al., 2011; Dressing et al., 2005; Spitzberg and Cupach, 2007; Walby and Allen, 2004). A higher proportion of female victims can be observed for other violent crimes as well, especially when it comes to violence in the close social environment (for example, Black et al., 2011; FRA, 2014; Hellmann, 2014; Hellmann and Blauert, 2014; Walby and Allen, 2004). Applying C-Tree analyses confirmed the variable gender as the main risk factor for experiencing stalking. Additionally and in line with the bivariate analyses, people living on their own (both female and male) were at an increased risk of being stalked. Replicating previous research (see Budd and Mattinson, 2000; Walby and Allen, 2004), women’s relationship status had even more predictive value for experiencing stalking since nearly 3 out of 10 women who were separated, divorced, or widowed had experienced stalking in their lives. It can be assumed that the increased stalking prevalence for these individuals is also due to a relatively larger number of ex-partners. However, it should be noted with respect to the present study that the question regarding household size directly related to the time of the survey, whereas stalking experiences were assessed in relation to respondents’ lifetime.
According to the findings of the present as well as similar studies, the majority of stalking victimizations are committed by ex-partners or other known people (Black et al., 2011; Spitzberg and Cupach, 2007; also see Sheridan et al., 2003). This especially pertains to female victims (see FRA, 2014). Consequently, stalking can be regarded as a relationship offence. According to Meloy and Boyd (2003), for example, ex-partner stalking is associated with an increased risk of violence. Against the background of these considerations, the findings of this study provide further evidence that women’s close social environment carries the greatest victimization risk for them.
Another characteristic feature of stalking is the constellation between female victims and male offenders. On the one hand, this result corresponds to the prevailing notion that stalking is mainly committed by men against women (Langhinrichsen-Rohling, 2012). On the other hand, it is consistent with outcomes of other stalking studies (for example, Dressing et al., 2005; FRA, 2014) and with results of other studies on violence in the close social environment (for example, Black et al., 2011; Hellmann, 2014; Hellmann and Blauert, 2014; Walby and Allen, 2004). This finding may also be due to the particularly high proportion of ex-partners among the offenders who are stalking female victims. Overall, in the majority of the cases the stalking was committed by an offender of the opposite gender and this pattern was especially pronounced for female victims.
Regarding respondents’ origin, women with a Turkish immigrant background had a particularly low risk of experiencing stalking, although this result did not occur applying multivariate analyses. However, relationships between stalking and ethnicity have been reported elsewhere (Black et al., 2011; Tjaden and Thoennes, 1998; Walby and Allen, 2004). Noticeably, male respondents’ origin does not seem to play any role in whether they become a victim of stalking. Some authors explained these differences among other things with the fact that the perception as well as the disclosure of stalking are affected by cultural distinctiveness in values and norms (for example, Dietz and Martin, 2007; Jagessar and Sheridan, 2004). Thus, with regard to the present study one could argue that female respondents with a Turkish immigrant background might have a greater inhibition threshold against specifying such experiences in a questionnaire. However, the present study was part of a large-scale survey on different forms of victimization (see Hellmann, 2014). Within this research project, participants responded to questions regarding experienced physical violence by partners and other family members among others. In this regard, women with a Turkish immigrant background reported by far the highest victimization rates in comparison with female respondents with a Russian immigrant background and German women without an immigrant background (Hellmann, 2014). Thus, there is no reason to assume that women with a Turkish immigrant background simply did not mention stalking victimizations, since they readily reported intra-familial violence.
The lower stalking prevalence rates for women with a Turkish immigrant background in the bivariate analyses may be due to relationship status and household size. According to their own statements, these respondents live alone less often and marry earlier than women with a Russian immigrant background and women without an immigrant background. Considering solely the group of female respondents with a Turkish immigrant background, the risk of stalking was nearly 10 times lower for the married women compared with women living on their own. Importantly, according to the results of the multivariate analyses, female participants’ relationship status and their household size were key factors in explaining the risk of experiencing stalking.
It is not surprising that younger individuals in this study had an overall lower lifetime prevalence of stalking, when seen against the background that the likelihood of becoming a victim of crime increases with time. However, the stalking literature mainly consists of reports that younger people are more likely to experience stalking than older individuals (Budd and Mattinson, 2000; Dressing et al., 2005; Scottish Government, 2014; Van der Aa and Kunst, 2009). As an explanation for this finding, Walby and Allen (2004: 84), for example, stated: ‘The association of age with inter-personal violence may be due to the greater vulnerability of younger women, or it may be due to the greater propensity of younger rather than older men to use violence coupled with the tendency of people to associate with people of the same age group. This age-related finding is not unique to inter-personal violence.’ Against this background, the lower prevalence of stalking among the younger respondents in the present study may be due to the applied broad definition of stalking. Possibly, the reversed age effect described elsewhere would occur, too, if the criterion of experienced violence was additionally applied or stricter criteria were used in defining stalking.
In the present study, the average stalking duration was considerably lower than reported by Spitzberg (2002) for instance (also see Purcell et al., 2002; Tjaden and Thoennes, 1998). This difference might be owing to stricter definitions of stalking applied elsewhere. Supporting this notion, in most of the cases reported in this study the stalking lasted for less than one year (see Budd and Mattinson, 2000; Walby and Allen, 2004). In line with the findings of other authors, the stalking duration varied largely between the victims of the present study (for example, Budd and Mattinson, 2000; Spitzberg and Cupach 2007; Tjaden and Thoennes, 1998). Although the estimated average stalking duration was comparably low in this study, almost 2 out of 10 cases were reported as lasting for more than a year (also see Dressing et al., 2005; Walby and Allen, 2004). Thus, albeit a rather liberal definition of stalking has been applied in the present study, a considerable proportion of stalking victimizations had a comparably long duration.
Similar variations in the frequency of stalking behaviours occurred between the victims, depending on their gender. The fact that the majority of victims reported fewer than 10 incidents might again be explained by the liberal definition of stalking applied in this study. That female victims are stalked in higher frequency might be due to the higher proportion of male offenders in this group. Consistent with this assumption, offenders of the same gender tended to show more frequent stalking behaviours if the victim was female. Considering the victims who reported more than 50 incidents in sum, it can be concluded that the stalking increased depending on its duration. Nearly 3 percent of victims experienced more than 50 incidents for more than one year. Presumably, these individuals suffer the most from the stalking victimization (see Hellmann and Regler, 2014).
Limitations
Even though a representative quota sample of respondents was recruited to take part in this victim survey, there are several factors that should be considered when generalising the reported prevalence estimates. First, one should take into account that the sample was restricted to the German population aged 16–40 years. Several studies have found that the risk of stalking victimization is age dependent (for example, FRA, 2014; Freidl et al., 2011; Scottish Government, 2014). However, since younger people in particular have been reported to be at a higher risk of experiencing stalking, the main potential risk group of stalking victims was covered in the present study.
Second, this survey does not capture populations living in institutions (for instance, prisons, nursing homes), or those who may be living in shelters or are homeless (see Black et al., 2011). Similarly, people with an illegal residence status have presumably not taken part in this study. Those people in turn might be at rather high risk of experiencing stalking victimization. Therefore, the prevalence rates reported in this study are more likely to be underestimations than overestimations. However, taking into account that these groups of individuals represent relatively small minorities, the present data are likely to depict the stalking reality in Germany to a reliable extent.
Potential respondents who are currently stalked may have been too afraid to take part in the victim survey or to fully disclose their experiences for a number of reasons. Additionally, respondents who have experienced severe stalking victimization may in particular experience such a survey as intrusive and therefore refuse to respond, potentially leading to an underestimation of prevalence rates (Purcell et al., 2002). In contrast, other scholars suspected an overestimation of prevalence rates in victim surveys because victims are more motivated than non-victims to participate (Sheridan et al., 2003).
Beyond that, some respondents may not be aware that what they experienced was stalking or did not feel comfortable reporting their experiences because of the social stigma associated with being a victim. Additionally, respondents might have forgotten about less severe stalking experiences or simply cannot remember the details of incidents that may lie several years in the past. However, regarding the more severe and thus more salient experiences, this is less likely (Henry et al., 1994). Similarly, victims may be reluctant to discuss stalking experiences, which in turn could influence estimated prevalence rates in unpredictable ways. Additionally, this aversion and its consequences may be confounded with certain unknown characteristics of the victims.
Conclusion
Even with respect to the aforementioned limitations, the present research as the first nationwide study on stalking in Germany yields important insights in providing estimates of the prevalence of stalking in Germany for both men and women aged 16–40 years. The results are comparable to those of other large-scale surveys from different countries. Representative victim surveys remain an important source in understanding the prevalence and nature of stalking. This holds especially true if the victimization is not considered a crime by the victims and (maybe consequently) is not likely to come to the attention of the prosecuting authorities. Therefore, such surveys can serve as bases for the development of specific prevention measures as well as means for protecting individuals against stalking victimization.
Footnotes
Funding
This research was supported by a grant from the German Federal Ministry of Education and Research (01SR1002).