Aggression and Violent Behavior, Vol. 3, No. 3, pp.

255–274, 1998
Copyright  1998 Elsevier Science Ltd
Printed in the USA. All rights reserved
1359-1789/98 $19.00 1 .00

PII S1359-1789(97)00023-2

STALKING BEHAVIOR: A LITERATURE

REVIEW AND SUGGESTED FUNCTIONAL
ANALYTIC ASSESSMENT TECHNOLOGY
Darrah Westrup and William J. Fremouw
West Virginia University

ABSTRACT. This article provides a discussion of stalking behavior, a phenomenon very

much in the forefront of public, legal, and law enforcement attention. To rectify the lack of
a consistent and clear definition of stalking behavior in the literature, a behavioral definition
is suggested. A review and critique of the existing literature is also provided. The functional
analysis is proposed as a means to assist both intervention and research efforts. It is posited
that functional analyses will help illuminate the factors that maintain stalking behavior, sub-
sequently guiding intervention. It is also suggested that the information gained by consistent
and widespread use of functional analyses will provide points of departure for future re-
search.  1998 Elsevier Science Ltd
KEY WORDS. Stalking behavior, assessment

THE TOPIC OF THIS ARTICLE is the behavior popularly known as stalking. Despite its
notoriety, what exactly is meant by stalking is not easily defined. This difficulty is discussed
in detail in the following section. For now, stalking denotes the unwelcome, repetitive, and
intrusive harassing and/or threatening behavior directed toward a specific individual.
Stalking is currently receiving a great deal of attention from the media, law enforcement
communities, and the general public. It is not possible to determine whether the rate of stalk-
ing is increasing or decreasing (or is stable) due to the absence of prior information regarding
its prevalence. However, rising apprehension is reflected in the significant increase of main-
stream news articles on stalking over the last several years. In fact there are now more than
30 law review articles on the topic (Fein & Vossekuil, 1996). The most dramatic indicator
of the increased attention accorded this problem is the rapid development of antistalking
legislation. The first state to enact an antistalking law was California in 1990, propelled by
the widely mourned death of actress Rebecca Schaeffer (stalked and killed by Robert John
Bardo in 1989). Today, Maine is the only state without a stalking law; in this state an antiter-

Correspondence should be addressed to Darrah Westrup, Department of Psychology, P.O. Box 6040,
West Virginia University, Morgantown, WV 26506–6040.

255
256 D. Westrup and W. J. Fremouw

rorizing statute is used to prosecute stalking cases (Fein & Vossekuil, 1996; National Institute
of Justice Association, 1993).
Stalking affects the average individual, as opposed to just celebrities, more than previously
thought (Morin, 1993). Fremouw, Westrup, & Pennypacker (1996) administered a survey on
stalking to approximately 600 college undergraduates. Thirty percent of the women and 17%
of the men responded that they had been stalked as defined by West Virginia law (West
Virginia Code, 1995). Admittedly, this sample is not representative of the general population,
and the results of this study suggest that stalking is far from rare. The full gravity of this
social problem becomes more evident when this surprising prevalence rate is considered in
light of the impact that stalking can have on victims. Regardless of whether the victim endures
an actual assault, the insidious intrusiveness of being watched, followed, and otherwise ha-
rassed can be extremely distressing (De Becker, 1994; Mullen & Pathé, 1994a, 1994b). Addi-
tionally, the deleterious effects can be enduring. For example, it is not uncommon for a
stalking victim to significantly change his or her lifestyle in an attempt to cope (Fremouw
et al. 1996; Mullen & Pathé, 1994; National Institute of Justice Association, 1993). Com-
pounding the problem is the intractable nature of stalking. The behavior is repetitive by
definition, and many stalkers have an extensive history of similar conduct (Harmon,
Rosner, & Owens, 1995; Meloy, 1996).
Given the urgency of stalking, and its distressing and potentially harmful effects, the pau-
city of scientific research devoted to this behavior is surprising. This is unfortunate because
the psychological community in particular can offer much to ameliorate this significant prob-
lem. This article will further efforts in this area by addressing the following objectives:

1. Formulating a stalking definition that is both consistent with popular usage and condu-
cive to research and treatment efforts.
2. Providing a review, critique, and summary of relevant research. Studies will be exam-
ined in terms of how well they increase the ability to predict and control stalking.
3. Introducing a functional analytic model as an assessment technology that will assist
efforts to predict and control stalking.

DEFINITION OF STALKING
There is a glaring lack of agreement in the literature on what is meant by the term stalking.
Before commencing the literature review, it is important to clarify how various authors have
interpreted this term, as well as the alternate labels they have suggested. Included in this
section is a discussion of erotomania, as the misplaced emphasis on this delusional disorder
complicates the literature. Finally, as a means to reconcile both these difficulties, a behavior-
ally oriented definition of stalking will be provided.
Formal definitions of the verb stalking typically incorporate the concept of stealthy pursuit
or tracking (DeVinne et al., 1982; Neufeldt, 1988). These definitions emphasize the literal
act of following something, as well as stressing the furtive and predatory elements of this
behavior. As used by the general public, media, and law enforcement communities, however,
stalking loosely refers to a broader range of repeated behaviors (e.g., telephoning, letter writ-
ing, conducting surveillance) whose overall effect is to threaten and/or harass another individ-
ual. These acts may or may not be surreptitious and may or may not be maliciously intended;
some stalkers may actually be pursuing their target out of what they perceive to be love
(Harmon et al., 1995; Mullen & Pathé, 1994b; Zona, Sharma, & Lane, 1993). In addition,
one of the behaviors within this behavioral class is literally following someone. Stalking can
 Stalking Behavior 257

thus properly refer to the act of stealthily following someone, and it can also properly refer
to a broader class of behaviors. This entanglement may be the reason most researchers eschew
stalking as an overall descriptor of the behavior class. Unfortunately the alternate terms they
suggest engender their own interpretation difficulties.
Zona et al. (1993) and Harmon et al. (1995) use the term obsessional harassment to denote
stalking. Although harassment better signifies the class of behaviors, the term does not fully
represent stalking behavior. Evidence of this difference between stalking and harassment is
the fact that 28 states with current harassment statutes found it necessary to enact additional
antistalking legislation (National Institute of Justice Association, 1993). A second problem
with this terminology concerns the use of the word obsessional, which Zona et al. defined
as having ‘‘persistent ideas, thoughts, impulses, or images that result inevitably in some act
in relation to the victim’’ (p. 896). Traditionally, obsessions have been viewed as being
unwanted and intrusive (American Psychiatric Association, 1994). In recognition that many
of the persistent thoughts experienced by stalkers can actually be pleasurable or satisfying,
Zona et al. stated that their stalkers’ obsessions are not unwanted or intrusive, which is con-
trary to the traditional definition. Furthermore, this idiosyncratic interpretation does not take
into account those stalkers whose thoughts regarding the target are unwanted and distressing.
The definition of obsessions provided by the authors also implies that the obsessions, or
persistent thoughts, cause (i.e., inevitably result in) stalking behavior—an unproved hypothe-
sis at this point. The obsessional harassment label evokes an additional problem: Obsessions
so defined are thoughts that cannot be objectively identified and measured. To focus on these
thoughts as the cardinal aspect of stalking behavior may impede both scientific study and
intervention efforts (Barrett, Johnston, & Pennypacker, 1986; Ullman & Krasner, 1969).
Meloy and Gothard (1995) prefer obsessional following as a global term for stalking behav-
ior, reserving stalking for a discrete act within this repertoire (which they unfortunately do
not define). This solution is questionable because (a) following is no more effective than
stalking in differentiating between an entire class of behaviors and one specific action, and
(b) use of the word obsessional invites the incumbent previously described difficulties.
Finally, Mullen and Pathé (1994a, 1994b) labeled their sample of stalkers as those suffering
from ‘‘pathologies of love.’’ However, love is difficult to operationalize (philosophers have
been at it for years), and this term certainly cannot be applied to the stalker who is following
and making death threats to his or her previous employer. This nebulous term is an unfortu-
nate example of the disparate and ineffective nomenclature obscuring research in this area.
The lack of clarity and consistency in how stalking is defined in the literature is the first
roadblock in examining this behavior. A second problem is the entanglement between stalking
research and erotomania, a disorder wrought with its own controversies over definition of
meaning (Harmon et al., 1995; Meloy & Gothard, 1995; Mullen & Pathé, 1994a, 1994b;
Zona et al., 1993). The Diagnostic and Statistical Manual of Mental Disorders–IV currently
lists erotomania as a delusional disorder subtype and proposes its application ‘‘when the
central theme of the delusion is that another person is in love with the individual’’ (American
Psychiatric Association, 1994, p. 297). The interest in erotomania is not without reason, as
some of the more infamous stalking cases have been conducted by individuals fitting this
diagnosis (Anderson, 1993; Goldstein, 1987; Leong & Silva, 1992). There has been long-
standing controversy over whether erotomania exists as a separate diagnostic entity
(Goldstein, 1987; Segal, 1989) or whether it is a symptom of another condition, such as
schizophrenia (Ellis & Mellsop, 1985; Leong, 1993; Rudden, Sweeney, & Frances, 1990).
Most typically, primary erotomania is considered to be pure erotomania, meaning that it
exists without an accompanying disorder. Secondary erotomania is considered to be the case
when the delusions coexist with other psychiatric symptoms (Hollender & Callahan, 1975;
258 D. Westrup and W. J. Fremouw

Seeman, 1978; Segal, 1989). A related debate is over the traditional understanding that a
diagnosis of erotomania can only be applied if the individual in question believes he or she
is loved in return. Most researchers in this area believe this component is the chief marker
for whether or not the diagnosis is made (Evans, Jeckel, & Slott, 1982; Goldstein, 1987),
whereas others feel it should not be the distinguishing feature (Mullen & Pathé, 1994; See-
man, 1978).
Stalking investigators have repeatedly attempted to create a nosology of stalkers that in-
cludes erotomania. Unfortunately, they have inherited the previously described conundrums.
For example, Meloy (1996a) revealed that three out of four individuals within the stalking
literature suffered from a comorbid Axis I disorder (they had secondary erotomania), creating
the possibility that their delusions (and subsequent stalking behaviors) arose from the primary
diagnosis. Finally, because (a) many stalkers are not erotomanic (Harmon et al., 1995;
Meloy & Gothard, 1995) and (b) few erotomanics actually engage in stalking (Dietz, Mat-
thews, Van Duyne, et al., 1991), erotomania’s conspicuous role in the stalking literature is
misleading. It is hoped that illumination of this point will assist interpretation of the literature
in the following review.
It is possible (and desirable) to define stalking in such a way that the previously described
difficulties are reconciled. Further, it is possible to arrive at a conceptualization of the behav-
ior more consistent with other professionals who actually contend with stalking situations.
These professionals (i.e., policy makers, law enforcers, and security consultants) increasingly
view stalking as a pattern of behavior that has the overall effect of significantly disturbing
the individual to whom it is directed (De Becker, 1994; Fein & Vossekuil, 1996; National
Institute of Justice Association, 1993). Viewed in this way, for a behavior to be considered
stalking it must be experienced by another as a disquieting and often threatening intrusion.
An odd implication of this definition is that, if a stalker is unnoticed by the victim, he or
she is not technically stalking! The pragmatic approach, and that taken by policy makers,
law enforcers, and other stalking experts, is that stalking is a problem behavior that exists
only when it serves to distress the targeted individual (Anderson, 1993; Morin, 1993; National
Institute of Justice Association, 1993). The proposed definition will incorporate this perspec-
tive and will acknowledge that this does not resolve the issue of unnoticed stalking.
We propose that the term stalking should be retained as a global descriptor of the behavior,
chiefly because of its persistent and pervasive use by the general public, and also because
the weaknesses of alternate labels precludes their use. However, adding the word behavior
helps signify that stalking typically refers to a group of behaviors. It is proposed then, that
stalking behavior be defined as one or more of a constellation of behaviors that (a) are repeat-
edly directed toward a specific individual (the ‘‘target’’), (b) are unwelcome and intrusive,
and (c), induce fear or concern in the target. Defined thusly, the behavior can be observed
and clearly identified, and the numerous behaviors that are relevant (ranging from conducting
surveillance to letter writing) are included. A stalker, then, is an individual who stalks or
engages in these behaviors.

LITERATURE REVIEW
This review chronologically examines the literature on stalking. Included herein is research
that focuses on at least one particular stalking behavior (i.e., letter writing) or on stalkers.
Excluded are case reports of erotomania in which stalking behavior is mentioned only in
passing. This review also provides two examples of antistalking efforts developed by the
government and private sector to address stalking behavior. Inclusion of these efforts is appro-
priate given the fact that the problem of stalking behavior affects a cross section of people,
 Stalking Behavior 259

agencies, and organizations, and effective intervention requires their joint cooperation (De
Becker, 1994; Fein & Vossekuil, 1996; National Institute of Justice Association, 1993).

Psychological Studies
Dietz, Matthews, Van Duyne, et al. (1991). This study focused on the association between
‘‘threatening and otherwise inappropriate’’ (p. 185) letters sent to Hollywood celebrities and
subsequent approach behaviors. This research is included because (a) the act of sending a
threatening or inappropriate letter to a targeted individual is a common stalking behavior
(Mullen & Pathé, 1994b; Zona et al., 1993); (b) the study is one of the few empirical
efforts in the literature; and (c) the authors’ pragmatic objective, to assist efforts to predict
the dangerousness of these letter writers, is of obvious benefit to those concerned with stalking
behavior.
To discern differentiating variables, the authors compared 107 letter writers who had at-
tempted to approach the object of their attentions with 107 letter writers who did not. In
some cases, these behaviors were but one of many repetitive, unwanted, and intrusive actions
the participants directed at their targets. The data were drawn from the archives of Gavin
De Becker, Inc., a security consulting firm that assesses and tracks individuals who send
threatening and inappropriate communications to celebrities. This sample is most likely repre-
sentative of only a small subset of possible letter writers, as it includes only those who write
to celebrities and whose letters are clearly threatening or disordered.
Although their research was archival in nature, Dietz and colleagues used rigorous methods
to enhance the validity of their findings. Both groups were randomly drawn and assigned to
coders blind to whether or not the participant had approached the celebrity. Coders were
carefully trained with clearly operationalized criteria, and interrater reliability (ranging from
64 to 100% on 904 variables) was calculated throughout. Corroborative information was
gathered from criminal records, media accounts, and public records, as well as from inter-
views with the participant and/or family and medical personnel.
The number of variables considered in this study (904) and the numerous reported findings
suggest that many analyses were made, raising the issue of whether the authors accounted
for the increased probability of a Type I error. Dietz, Matthews, Van Duyne, et al. (1991)
did acknowledge the problem by stating that, because of the large sample size (N 5 214)
and the number of comparisons made, the reader should exercise caution when interpreting
results with probabilities between .01 and .05. They did not, however, report the total number
of analyses made, and they did not clearly state whether they used a priori or post hoc correc-
tion procedures. For this reason, the following ‘‘significant’’ results should be interpreted
cautiously.
Participants who approached their targets sent more communications to their targets than
those who did not. The volume of letters did not predict approach behavior, as they were
sent not only prior to approaching the target but after as well. Writing for more than 1 year and
the combination of telephoning and letter writing were associated with approach behavior.
Participants who specifically stated they would like to meet the target, and whose letters
contained specific plans for doing so (i.e., a time and location were suggested), were more
likely to approach. Other features significantly associated with approach behavior were
(a) letters (specifically the first known communication) that had no return address; (b) letters
from more than one state, province, or country (i.e., that had more than one postmark);
(c) letters written on tablet paper, either lined or unlined; and (d) letters with enclosed pic-
tures. Letters that repeatedly mentioned public figures other than the target were also signifi-
cantly related to approach behavior.
Two analyses demonstrated an inverse relationship to approaching the target. Dietz, Mat-
260 D. Westrup and W. J. Fremouw

thews, Van Duyne, et al. (1991) assessed whether participants’ letters attempted to evoke a
particular emotion from the targets. The ‘‘intent to instill shame’’ (p. 202) was found to be
less likely in communications from those who approached their target. Likewise, participants
who indicated either a desire to marry or to have sexual relations or children with the target
were significantly less likely to seek an actual encounter. Surprisingly, the presence or ab-
sence of threats within the letters did not predict approach attempts.
This was an ambitious and difficult research effort. It is difficult to assess at this point its
practical value, as it described the relations between letter characteristics and approach behav-
ior, not letter characteristics and dangerousness (approaching was not assessed in terms of
subsequent harm to the target). However, those who must assess such letters for dangerous-
ness would no doubt be interested in knowing what features are associated with actually
approaching the target. The question remains as to why, or how, these characteristics are
related to approaching the target. This is a question of function, an aspect of stalking behavior
that is neglected by the literature.

Dietz, Matthews, Martell, et al. (1991). This research closely parallels the previous study,
although the participants had sent threatening or otherwise inappropriate letters (and also
cards, telegrams, and packages) to members of the U.S. Congress. The relations between
letter characteristics and approach behavior were examined following the same procedure
described before, except that data were drawn from the Intelligence Unit of the U.S. Capitol
Police rather than from the Gavin De Becker, Inc., archives. Forty-three approach-positive
letter writers were compared with 43 writers who did not pursue a face-to-face encounter
with the target. Some characteristics of these letters were found to be significantly associated
with approach behavior, but again, the authors did not report doing a correction procedure
so the validity of these results remains in question. Findings that supported those reported
in the previous study were as follows: (a) Participants who approached the target sent more
communications, (b) those who wrote and telephoned were more likely to approach, and (c)
those who expressed a desire for contact were more likely to pursue an actual encounter.
Interestingly, threatening communications from this sample were associated with not ap-
proaching the target—Participants whose letters were not threatening were significantly more
likely to initiate an approach. Moreover, unlike the previous study, no association was found
between how long these participants had been writing to their targets and whether an encoun-
ter was sought. The discrepancies between studies may be due in part to the different target
populations. If so, it is unlikely that the findings of both studies can be reasonably applied
to ‘‘regular’’ individuals (i.e., those who are not public figures).

Zona et al. (1993). In 1993, Zona, Sharma, and Lane conducted an archival study comparing
erotomanics with obsessional participants (defined as those who have ‘‘persistent ideas,
thoughts, impulses, or images that result inevitably in some act in relation to the victim,’’
p. 896) suffering from other mental disorders. Their research is included in this review be-
cause all of the individuals in this large sample were drawn from cases followed by the
Threat Management Unit (TMU) of the Los Angeles Police Department, a unit established
to deal expressly with stalking situations.
Information from 74 case files was used to create a data base from which the comparative
analyses were made. Participants were separated into three groups depending on how the
authors assessed the ‘‘quality’’ of each subject’s obsession (Zona et al., 1993, p. 896). Seven
participants were diagnosed as having primary erotomania. Thirty-two were labeled ‘‘love
obsessional,’’ the main distinction being that they suffered from a primary psychiatric diagno-
sis other than erotomania. (Unfortunately the manner in which psychiatric illness was deter-
mined was questionable; this will be discussed shortly.) Included in this group (even if they
 Stalking Behavior 261

did not have another disorder) were participants who, despite being obsessively in love with
their target, did not believe their love was reciprocated (thereby excluding a diagnosis of
erotomania). In addition, love obsessional participants ‘‘almost always’’ (p. 896) did not
personally know their target (the vagueness of this qualifier was not further explained). The
remaining 35 cases were classified as ‘‘simple obsessional.’’ These participants had experi-
enced a prior relationship with their target. By prior relationship, the authors considered not
just prior romantic involvement but also relationships such as that between neighbors or
between a professional and client. Once participants were divided accordingly, characteristics
such as telephoning, letter writing, threatening, and location visits were assessed and then
compared across groups.
One of the most striking findings was that duration of the obsession varied dramatically
between groups. Two considerations must be kept in mind when interpreting this finding.
Zona et al. (1993) did not report whether a statistical analysis had been performed on this
variable, so there is no way to know whether this difference was meaningful. Additionally,
the fact that these participants were drawn from a forensic population may mean that the
stalking behavior was interrupted, as opposed to being voluntarily ceased. Nonetheless, it is
interesting that the average erotomanic was obsessed with his or her target for as long as
125 months, including 19 months of what the authors called ‘‘actual contact’’ (p. 899), which
presumably refers to actual stalking behavior. The love obsessionals were obsessed with their
targets for an average of 146 months (9.7 months of contact), and the simple obsessionals
contacted their targets for an average of 5.1 months (the duration of their obsession with the
target was not assessed).
Additional findings were that the simple obsessionals were significantly less likely to write
their targets and significantly more likely to threaten. Of the two participants in the entire
sample who physically harmed their target, both were in the simple obsessional category.
Additionally, 14% of this group damaged their target’s property in some way. These results
suggest that stalking incidents involving former associates are potentially more dangerous
than those involving strangers, an interesting finding given the likelihood that the majority
of stalkers are those who stalk former romantic partners (Fremouw et al., 1996; Mullen &
Pathé, 1994b; National Institute of Justice Association, 1993). The applicability of this finding
to stalking situations between former intimates is limited by the fact that any type of prior
association between stalker and target (not just a prior romantic or sexual relationship) was
cause for inclusion in the simple obsessional group.
Zona et al. (1993) reported that 37% of the love obsessional and 40% of the simple obses-
sional participants evidenced some type of mental disorder (the erotomanic participants all
were delusional by definition). Presence of a major mental illness was determined by (a)
any indication in the subject’s file of an Axis I disorder, (b) evidence of prior psychiatric
hospitalization, and (c) clearly ‘‘psychotic thinking’’ (p. 897) demonstrated in perhaps a letter
or tape recording. This finding should be interpreted cautiously because important informa-
tion was not provided (e.g., what was in the participants’ case files, what criteria were used
to identify mental illness, how was psychotic thinking operationalized). The fact that some
cases apparently were reviewed by only one investigator further compromised the reliability
of this result.
The terminology and classification system used in this study were its most problematic
aspects, resulting in the interpretation problems discussed in the Definition of Stalking section
of this article. For example, when diagnosing erotomania, the authors retained the traditional
criterion that the participant must believe his or her love is returned by the target. Those
who fit the diagnosis in every other way were placed in the love obsessional group. However,
many erotomania researchers do not agree that this should be the cardinal criterion (Mullen &
Pathé, 1994b; Seeman, 1978). Zona et al. (1993) also attempted to group their participants
262 D. Westrup and W. J. Fremouw

based on the ‘‘quality’’ of their obsessions (p. 896), relying not only on an idiosyncratic
definition of the term (i.e., that obsessions are not unwanted or intrusive thoughts) but also
on a construct (i.e., obsessional quality) that does not lend itself to clear interpretation. For
example, the content of the obsessions could actually be quite similar from group to group
(e.g., a schizophrenic could have obsessions and engage in behavior very similar to an eroto-
manic’s), and participants within one such grouping could have very different persistent
thoughts. For instance, someone who was stalking her ex-husband’s divorce lawyer would
be in the same group as someone stalking his ex-wife, yet the content of their obsessions
could be quite dissimilar. Classification and definition difficulties delineated earlier highlight
the difficulties in relying on a construct, a diagnosis, or even topography alone to understand
a behavior.

Mullen and Pathé (1994b). In 1994, Mullen and Pathé compiled 14 case reports of patients
seen in their Australian forensic psychiatric practice, all who suffered from ‘‘pathologies of
love’’ (p. 469) and who had engaged in stalking behavior. A pathology of love (a term the
authors sometimes used interchangeably with erotomania) was defined according to the fol-
lowing criteria: (a) The participant believed he or she was loved by the target and/or had a
‘‘preoccupying infatuation’’ (this was not defined, p. 471); (b) the target had not encouraged
the participant and/or had rejected him or her; (c) the participant was preoccupied to the
point where everyday functioning was impaired; (d) the target’s actions were reinterpreted
to maintain the love or supposed relationship; and (e) the participant made ‘‘repeated attempts
to follow and approach the object of their affections, creating at the very least distress and
embarrassment’’ (p. 471). The authors reported that 9 of the 14 participants had preexisting
psychiatric diagnoses and that the remaining 5 had a ‘‘pure’’ pathology of love.
Despite the report’s obvious drawbacks (i.e., the small sample size, atypical terminology,
and the fact that it was merely descriptive rather than empirical), it did provide some useful
information. For example, the participants’ various stalking behaviors were assessed; it was
determined that 13 of 14 followed and/or loitered near the target and that 11 made an actual
approach. Approximately half phoned, and half wrote to their targets. Five participants physi-
cally assaulted their target.
From information ascertained during interviews with some of the involved targets, Mullen
and Pathé (1994b) described the negative effects of the stalking behavior, such as being afraid
to leave the home, needing to have the telephone disconnected, and moving out of state and
even out of the country. The authors reported that a target’s fear seemed to escalate as the
stalking behavior continued. The impact of stalking behavior from the targets’ perspective
has rarely been provided in the literature. Also provided were details regarding the stalkers’
professed motivations for acting violently. These were primarily feelings of jealousy, rage
at rejection, and sexual desire. Such attempts to describe the behavior’s function or to explain
why the stalker engaged in the behavior are lacking in the literature and are sorely needed
to understand this phenomenon fully.

Meloy and Gothard (1995). Recognizing the lack of clinical information on stalkers, Meloy
and Gothard compared ‘‘obsessional followers’’ (i.e., those with persistent thoughts or im-
pulses that result in a stalking behavior, p. 258) with other criminal offenders diagnosed with
mental disorders in order to compare certain clinical and demographic variables. A sample
of 20 obsessional followers was drawn from the Forensic Evaluation Unit in San Diego. The
comparison group was randomly drawn from individuals who were also court referred to the
clinic, who had received the same type of evaluation (e.g., presentence evaluations, civil
commitment) at approximately the same time as had the obsessional followers.
Before conducting analyses between the two groups, Meloy and Gothard (1995) reported
 Stalking Behavior 263

demographic data on the obsessional followers that had been obtained via forensic evaluations
and review of records. Examination revealed that 90% were men with an average age of 35
and that most were White followed by Blacks and Hispanics (35%, 25%, and 15%, respec-
tively). More than 75% of them were single (half had either never been married or were now
divorced), and most were unemployed or had ‘‘unstable’’ (not defined) positions. Some 60%
of the obsessional followers had previous psychiatric treatment, and 85% had a diagnosable
Axis I disorder (i.e., schizophrenia, mood disorder, substance abuse or dependence, adjust-
ment disorder, erotomania, paraphilia, or other) at the time of the forensic evaluation. The
authors also reported that 85% met criteria for an Axis II personality disorder at the time of
evaluation.
Some descriptive information regarding the obsessional followers’ stalking behavior was
provided. Eleven of the 20 obsessional followers had targeted former romantic partners, and
9 had pursued a stranger. Twelve went to the target’s home, 8 had telephoned, and 5 wrote
letters. The authors also stated that 9 of the obsessional followers had ‘‘stalked’’ (p. 260)
their victims (meaning that they had been criminally charged with stalking as defined by the
California Criminal Code; J. R. Meloy, personal communication, February, 1997). When
possible, the total duration of the obsessional following was assessed as well, with most
participants (7) engaging in 10 or more incidents over a period of less than a year. Fourteen
of the obsessional followers verbally threatened their target; however, the 2 that did physically
assault their targets had not made verbal threats previously. This counterintuitive finding was
consistent with Dietz, Matthews, Martell, et al. (1991) and with Dietz, Matthews, Van Duyne,
et al. (1991), who found that threats were not positively associated with approaching the
target.
When comparative analyses were made between the obsessional followers and the other
group of offenders, the obsessional followers were found to be significantly older, better
educated, and more intelligent (according to IQ scores). No differences in the prevalence of
Axis I disorders were found between the two groups, although schizophrenia was determined
to be significantly less common in the obsessional follower group; one obsessional follower
fit this diagnosis compared with 10 of the other offenders. Two of the obsessional followers
were diagnosed with antisocial personality, compared with eight of the other offenders; this
difference was statistically significant.
This effort by Meloy and Gothard (1995) confirmed previous impressions that stalkers in
offender populations suffer from psychiatric illness. However, no one diagnosis was shown
to be particular to those who engage in stalking behavior. Both a schizophrenic and an eroto-
manic may engage in similar stalking behavior to fulfill a particular delusion—so may an
irate ex-husband. Stalking behaviors are multivarious even among those with the same mental
disorder. Knowing that a stalker has bipolar disorder or suffers from a personality disorder
does not lead to a better understanding of his or her stalking behavior (which is, after all,
the immediate problem as far as the target is concerned). Again, results of this study demon-
strate the need to focus on illuminating the causal relationships (rather than associated charac-
teristics) involved in stalking behavior.

Harmon et al. (1995). In this retrospective archival study, Harmon, Rosner, and Owens
examined 48 ‘‘obsessional harassers’’ (p. 188). To explain how their sample was selected,
the authors stated only that they ‘‘exhibited the type of repetitive behavior patterns with
which this study is concerned’’ (p. 188). Although in some cases behaviors were described
(e.g., phone calls, letter writing), in others the authors reported simply that the participant
‘‘harassed’’ the target (p. 190). All 48 participants had been court referred to their forensic
clinic between January 1987 and January 1994.
Recognizing the diversity of stalking behavior, Harmon, Rosner, and Owens separated
264 D. Westrup and W. J. Fremouw

their 48 participants according to two axes. They referred to the first of these as the ‘‘type
of attachment’’ (p. 189) between stalker and target. They considered two attachment types:
(a) affectionate/amorous and (b) persecutory/angry. This distinction is an indirect but impor-
tant first attempt to study the function of a stalker’s behavior empirically, a concept discussed
in the last section of this article.
The second axis was based on whether or not a prior relationship existed between the
stalker and target. Six subgroups were created: (a) personal, meaning that a romantic or
‘‘other personal attachment’’ (p. 190) had existed between the participant and target;
(b) professional, indicating that the participant had at one time used the professional services
of the target; (c) employment, in which the participant was the employee or the employer
of the target; (d) media, meaning that the target was a well-known public figure; (e) acquain-
tance, a situation where the participant and target knew each other only casually; and (f)
none, in which no connection between the participant and target could be made.
Unfortunately, the distinctions between the various just-listed relationships were unclear.
For example, it was not clear how a professional relationship between the participant and
target differed from the ‘‘acquaintance’’ category. The authors did not provide a rationale
for the just-noted classification system. Therefore, the generalizability of their findings to
other stalking situations (their ‘‘external validity’’) is not readily apparent.
Analyses were made between the two groups based on the first axis (i.e., attachment type).
No significant differences between the affectionate/amorous and the persecutory/angry par-
ticipants were noted in sex, age, criminal charges, or education level. In fact, the only statisti-
cally significant difference was that there were more single participants in the affectionate/
amorous group. The analyses may have been confounded by the fact that the emotions stalkers
entertained toward their targets often changed over time. It was not uncommon for a pursuit
that began in affection to become vindictive as the stalker was repeatedly rebuffed (De
Becker, 1994; Mullen & Pathé, 1994b; National Institute of Justice Association, 1993). Har-
mon et al. (1995) reported that no statistical differences based on the second axis (i.e., the
six classes of prior relationships) were found.
The group of 48 obsessional followers was also compared with all the other cases (N 5
915) that had been referred to the Forensic Psychiatry Clinic during 1993. Although no statis-
tical analyses were performed, the authors reported that obsessional followers were older
than the clinic population (mean ages were 40 and 31 years, respectively) and were better
educated, with 80% (vs. 30%) having graduated from high school. Forty percent held college
degrees (compared with just 6% of the nonstalking cohort). There was also a higher propor-
tion of female participants in the stalking group than in the other clinic population, and more
of the stalkers were White (two thirds were White compared with 12% of the clinic popula-
tion). Whether or not these observed differences were significant enough to be meaningful
(as opposed to being merely due to chance) is not clear due to the absence of statistical
procedures.

Fremouw et al. (1996). To examine prevalence of stalkers and victims of stalking behavior
among college students, Fremouw et al. (1996) administered a survey to 294 college under-
graduates (165 women, 129 men). The survey was made up of stalking behaviors described
in the existing literature (e.g., following, threatening, conducting surveillance) and asked the
students whether they had ever been either the recipient or enactor of such behaviors. If they
had been the targets of stalking behavior, the students were asked to identify the strategies
they used to deter the stalker.
When asked the literal question of whether they had ever been stalked (and whether they
had stalked someone), defined according to West Virginia law as ‘‘having someone know-
ingly and repeatedly following, harassing, or threatening you’’ (West Virginia Code, 1995)
 Stalking Behavior 265

a surprising 27% of the women and 19% of the men indicated they had been stalked. Only
three students admitted to having stalked someone. To replicate these results, the survey was
readministered to 299 additional participants, of whom 153 were women and 146 were men.
The findings were strikingly similar, with 18% of the men and 35% of the women reporting
being victims of stalking. Overall, 30% of the total female sample and 17% of the total male
sample indicated they had been stalked.
Responding to additional survey questions, the participants indicated that approximately
80% of them knew their stalker. In fact, 52% of the women and 37% of the men who had
been stalked had seriously dated the individual. Only one participant was stalked by someone
of the same sex. The finding that most of these participants were stalked by former intimates
supports previous assertions that most stalking incidents involve this dynamic (Mullen &
Pathé, 1994b; National Institute of Justice Association, 1993). Results also illustrate the social
stigma attached to stalking behavior. Given that approximately 30% of the participants re-
ported being stalked by someone with whom they had been romantically involved (someone
who most likely is a cohort), and yet only 1% admitted to stalking someone else, the undesir-
ability of having engaged in this behavior seems clear.
The participants dealt with being stalked in numerous ways (i.e., changing their phone
number, changing their schedule, moving). Women were most likely to ignore the stalker,
whereas men were more likely to attempt a confrontation.
An obvious limitation of this study was its sample, as college undergraduates are not
representative of the community at large. This shortcoming is perhaps partially offset by the
fact that this is the only nonforensic sample in the literature. The information was gained
via self-report, leading to concerns over exactly how the legal definition of stalking behavior
was being interpreted. The survey did not examine what ‘‘repeated following, harassing or
threatening behavior’’ meant to each participant who responded positively. However, the
participants’ overall responses to the survey (i.e., whether they confirmed or denied experi-
encing specific stalking behaviors) as well as additional written comments volunteered on
their response sheets suggest that this population does indeed experience stalking behavior.
The provocative results of this exploratory effort indicate the need for additional research
on stalking behavior within this student population as well as within the general community.

Meloy (1996a). Meloy provided a review of 10 studies of ‘‘obsessional following’’ published

from 1978 to 1995. Two criteria were used to select studies for review: (a) The participants
were obsessional followers (defined as someone who ‘‘engages in an abnormal or long term
pattern of threat or harassment directed toward a specific individual,’’ p. 2), and (b) the
participants were criminally charged due to their obsessional following. Six studies that
Meloy included were excluded from the present review because they entailed small sample
size case reports (i.e., ranging from 1 to 7 participants) that focused on erotomania (rather
than on stalking behavior per se).
Meloy grouped and summarized the obessional followers (n 5 180) in an effort to unearth
common demographic and psychiatric characteristics. Although the sample was heteroge-
neous, the data indicated that (a) a stalker is likely to be male, single, and in his mid-30s;
(b) he likely suffers from both Axis I and Axis II disorders; (c) he will have repeatedly
attempted to contact his victim over a period of several years; and (d) despite having threat-
ened her, he will probably not attempt a physical assault.
Although Meloy (1996) distilled much from the available data, grouping the existing re-
search in this way does not reconcile the fundamental sampling, analytical, and definitional
difficulties discussed in this review. Interpretation of the reported findings is restricted by
these difficulties. Although Meloy’s review was thorough and his intention admirable, who
stalks and who is likely to be stalked (and why) remains in question.
266 D. Westrup and W. J. Fremouw

Romans, Hays, and White (1996). To assess the incidence of counselors who have experi-
enced stalking behavior by their current or former clients, Romans et al. (1996) sent a stalking
survey to 41 university counseling centers in the United States. Stalking was defined as ‘‘will-
fully, maliciously, and repeatedly following or harassing another person and making a credi-
ble threat,’’ and harassing was defined as a ‘‘willful course of conduct directed at a specific
person which seriously alarms or annoys the person’’ (p. 596). Out of 178 counseling staff
members who responded to the survey, 10 (5.6%) reported that they had been stalked by a
current or former client. Five of the stalkers were female, and three were male (the gender
of two was not identified). The age range was from 17 to 40 years. Six stalkers were described
as having Axis II personality disorder symptoms. All 10 reported negative effects from being
stalked, including lost time from work, financial loss, and stress or worry.
These results cannot be fully assessed without knowing what the authors meant by the
‘‘credible threat’’ component included in their definition of stalking (p. 596). Stalking statutes
differ from state to state in terms of how threatening behavior is defined. For example, some
states require that the stalker make a verbal threat against the target. Others require a verbal
threat and behavior that furthers the threat. Most states (33) require only that a ‘‘reasonable’’
person feels threatened, even if no actual verbal threat has been made (National Institute of
Justice Association, 1993, p. 13). In the present study, a large number (111, or 63%) of
counseling staff members indicated they had experienced ‘‘stalking-related events’’ (p. 597).
It is important to know on what basis (i.e., absence of a credible threat, isolated vs. repeated
incidents) these individuals were not considered to have experienced stalking in order to
interpret these results fully. In addition, the authors stated that 1 of the 10 stalking targets
had been stalked three times, without clarifying whether this meant stalked by three different
clients or that there were three series of stalking behavior (because stalking is a repetitive
behavior by definition, it is assumed that the authors did not mean three events or incidents).
Nonetheless, this survey did explore the incidence of stalking behavior in a population (i.e.,
university counselors) that, given the established comorbidity of stalking behavior and other
clinical disorders (Meloy, 1996a, 1996b; Meloy & Gothard, 1995), is likely to be exposed
to stalkers. An important point that was highlighted by this survey was that, although 63%
of the counseling staff members had experienced some form of stalking-like behaviors, 60%
had received no formal training as to how to address such situations. As pointed out by the
authors, such training is necessary and attainable.

Antistalking Efforts by Other Organizations

National Institute of Justice Association (1993). The National Institute of Justice created
the ‘‘Model Anti-Stalking Code for States’’ in response to a 1993 directive by Congress to
develop ‘‘constitutional and enforceable’’ antistalking legislation (p. 43). A comprehensive
survey on existing stalking policies was distributed to over 400 police chiefs (70% of which
responded) from the largest local, county, and state police agencies. The data gathered were
extensive and will not be reported in full. The respondents reported that most stalkers were
former intimates and that the total population was diverse, including coworkers, neighbors,
and strangers. The respondents indicated that the most frequent stalking behaviors were fol-
lowing, verbal threats, letter writing, vandalism, appearing at the workplace, phone calls and,
physical assaults.
This document included suggestions for further development and study, stressing the need
for a multidisciplinary effort between law enforcement agencies and the mental health com-
munity. The authors stated the need for more specific information on actual stalking behav-
iors, including stalkers’ behavioral histories and the environmental factors that contribute to
 Stalking Behavior 267

their actions. It was also suggested that stalking statutes require psychiatric evaluation and
counseling as part of sentencing, as untreated individuals with psychiatric disorders often
worsen while incarcerated (Although it was not mentioned, it is reasonable to propose that
incarcerated stalkers without a psychiatric diagnosis would also be unlikely to improve with-
out treatment of their stalking behavior.)
Along with providing comprehensive information on the status of antistalking policy, this
carefully constructed and thoughtful document represents the concerns and issues encoun-
tered by police and professionals charged with resolving stalking situations. Stalking re-
searchers would do well to study it carefully in order to understand fully the nature of the
stalking problem and to help determine where their contributions are most needed.

Gavin De Becker, Inc. This private security consulting firm headed by Gavin De Becker
provides guidance and intervention strategies for public figures who are being ‘‘pursued’’
(De Becker, 1994, p. 1), and it assists organizations such as the TMU of the Los Angeles
Police Department, the Secret Service, and several Fortune 500 companies with threat assess-
ment. Since its inception in 1979, the agency has gathered more than 350,000 pieces of data
on 18,000 individuals who pursue others.
During the course of his work with stalkers and those they target, De Becker developed
Mosaic, a computer software program for use by organizations (e.g., the Los Angeles TMU)
that encounter numerous stalking incidents and need to determine whether a particular situa-
tion requires further attention. The user enters information regarding the current situation
(e.g., Does the participant own a weapon? or Is he or she violent?), thus creating a template
that is electronically compared with other cases from a vast data base of previous stalking
incidents. The end product is an overall evaluation of how similar a particular stalking case
is to others that have escalated.
The Mosaic system provides consistency in how stalking cases are handled. Regardless
of their level of expertise, users are able to assess stalking situations thoroughly, thereby
facilitating communication with other professionals and concerned parties. Unfortunately,
because no research has been conducted regarding its validity (i.e., concurrent, discriminant,
predictive), it cannot actually predict whether a case will or will not escalate. The ultimate
decision whether or not to intervene in a given case remains a judgment call.

RESEARCH SUMMARY
To summarize the studies included in the previous review, three were purely descriptive of
certain stalkers and stalking behavior (Fremouw et al. 1996; Mullen & Pathé, 1994; Romans
et al., 1996). Five described the relationship between stalkers and/or stalking behavior and
some other variable such as the presence of a psychiatric illness or approach behavior (Har-
mon et al., 1995; Meloy & Gothard, 1995; Zona et al., 1993). Two studies were correlational
(Dietz, Matthews, Martell et al., 1991; Dietz, Matthews, Van Duyne et al., 1991), and they
examined the relationship between certain letter characteristics and approach behavior. Meloy
(1996a) provided a review of 10 studies of stalking behavior and summarized the aggregate
data. As examples of serious nonclinical work, the National Institute of Justice Association
(1993) gathered extensive information on stalking policy, and Gavin De Becker, Inc., com-
piled a large data base and developed a between-cases comparison system. These efforts
have been reviewed with particular attention to how well they contribute to the prediction
and control of stalking behavior—A brief summation of the progress made in these areas is
presented next.
268 D. Westrup and W. J. Fremouw

Predicting Stalking Behavior

Accurate prediction of stalking behavior entails knowing who is going to stalk and when.
Attempts to identify distinctive characteristics of stalkers have instead illustrated that the
population is far from homogenous. The following features of stalkers (in offender popula-
tions) have been suggested:

1. Stalkers are likely to have either an Axis I or an Axis II mental disorder (or both) and/
or a prior history of mental illness (Harmon et al., 1995; Meloy & Gothard, 1995; Zona
et al., 1993). Erotomania is rarely present in its pure form, leaving open to interpretation
whether the delusional thinking (and subsequent stalking behavior) is due to this subtype
of delusional disorder or to a more primary diagnosis (Harmon et al., 1995; Meloy &
Gothard, 1995; Mullen & Pathé, 1994b; Zona et al., 1993).
2. Although many women engage in stalking behavior, most stalkers are men (Harmon
et al., 1995; Meloy & Gothard, 1995; Mullen & Pathé, 1994; Zona at al., 1993).
3. Most stalkers are single, and many have never been married (Harmon et al., 1995;
Meloy & Gothard, 1995; Zona et al., 1993).
4. Stalkers in offender populations are typically older than other offenders; their average
age ranges from 35 (Meloy & Gothard, 1995; Zona et al., 1993) to 40 years (Harmon
et al., 1995; Mullen & Pathé, 1994b).
5. Stalkers are better educated (Harmon et al., 1995) and more intelligent (Meloy & Goth-
ard, 1995) than other offenders.

No causal factors of stalking behavior have been empirically investigated, which precludes
understanding the behavior at this time.

Controlling Stalking Behavior

To those who must deal with an actual stalking situation, the compelling question is how to
make the behavior stop. To date, however, no systematic efforts to assess intervention out-
come (either in terms of treatment or legal sanctions) have been attempted. Policy makers
and the law enforcement community acknowledge that legal sanctions are often ineffective
and unenforceable (Morin, 1993; National Institute of Justice Association, 1993; Perez,
1993). Other experts have observed that legal action can actually exacerbate the situation (De
Becker, 1994). Although there is some reported success in treating stalkers with antipsychotic
medication and cognitive therapy (Mullen & Pathé, 1994a, 1994b; Taylor, Mahendra, &
Gunn, 1983), most clinicians report poor prognoses. Given the apprehension over stalking
behavior and its harmful effects, the current state of affairs is unacceptable, providing all the
more reason to consider a new approach to the problem.

Difficulties in Researching Stalking

Inconsistencies in terminology and various classification difficulties appeared consistently
throughout the literature. All of the studies but two (Fremouw et al., 1996; Romans et al.,
1996) involved forensic populations, which limits generalization to the majority of stalking
situations. None of the studies focused specifically on stalkers who target former romantic
partners. This represents an unfortunate omission, as this population is thought to make up
the largest number of stalkers (Fremouw et al., 1996; Mullen & Pathé, 1994b; National Insti-
tute of Justice Association, 1993). None of the studies focused on stalking behavior (as op-
posed to stalkers) in adequate detail.
It must be acknowledged, however, that it is comparatively easy to critique the efforts of
 Stalking Behavior 269

others, having the benefit of hindsight and being armed with guidelines, texts, and diatribes
as to what ideally constitutes good research. Those who make the first attempts to shed light
on a previously murky area have the far more difficult task. Stalking behavior is complex,
often secretive in nature, and is currently so denigrated that stalkers are understandably reluc-
tant to present themselves for treatment or study (Fremouw et al., 1996). In fact, it is the
appreciation of the daunting nature of the task that fuels this proposal for the following
assessment strategy.

A Functional Analytic Approach to Stalking Behavior

This section proposes utilizing functional analyses when assessing stalking behavior. As the
term implies, a functional analytic approach emphasizes the function served by a given behav-
ior. In contrast, the existing literature on stalking behavior has focused on describing various
characteristics of stalkers and some stalking behaviors, not the purpose(s) served by the be-
havior. Our comprehension of stalking behavior has not been appreciably increased from
these efforts. It is proposed that a functional analytic approach will generate different informa-
tion that in turn may improve our understanding of this problem behavior.
A functional analysis views behavior in terms of the context in which it occurs. In this
sense, behavior is assumed to be purposeful because it is affected both by preceding environ-
mental events and by environmental consequences that follow. Behaviors tend to repeat and
escalate when they are successful (i.e., they are rewarded in some way) and when they occur
in contexts similar to those where the behavior was successful in the past (Skinner, 1953).
Likewise, stalking behavior is purposeful or has a function (no matter how illogical), and it
occurs or escalates under some conditions and not under others. This functional analytic
account is simply another way to state what has already been found by experts who have
contended with stalking behavior—that the relevance of the situation in which it occurs is
inescapable (De Becker, 1994; Fein & Vossekuil, 1996; Fein, Vossekuil, & Holden, 1995;
National Institute of Justice Association, 1993).
A fundamental tenet of the functional analytic approach is that when a behavior’s function
is determined, so are the factors that maintain the behavior (Nelson & Hayes, 1986). Addition-
ally, identification of these factors is instructive as to where intervention should be directed
(Carr, 1994; Horner, 1994; Mace, 1994). Functional analyses have been used with success
in assessing and treating a variety of problematic behaviors from self-injurious behavior to
depression (Carr, 1994; Ferster, 1973), and stalking behavior can benefit from this technology
as well.
It is therefore posited that applying the technology of functional analysis to stalking behav-
ior will help (a) identify what conditions promote and maintain an individual’s stalking behav-
ior (i.e., the behavior’s controlling variables) and (b) determine effective intervention. It is
also proposed that consistent and widespread use of functional analyses when assessing stalk-
ing behavior will provide direction for future research by better elucidating the controlling
factors in stalking behavior. In other words, applying functional analyses to stalking behavior
will help gather the information necessary for its prediction and control.

Functional Analyses and the Identification of Controlling Variables

A functional analysis hinges on gathering detailed information about the given behavior and
the context in which it occurs. A common heuristic for collecting this data is provided by
the acronym, ABC. That is to say, the antecedents to the behavior, the behavior itself, and
the consequences must be carefully specified. Because the presenting difficulty in a stalking
situation is the action directed toward the target, specification of the actual stalking behavior
270 D. Westrup and W. J. Fremouw

FIGURE 1. Illustration of functional relations among antecedent stimuli, responses,

and maintaining consequences as demonstrated by a functional analysis. From Ad-
vanced Abnormal Psychology (p. 498), by K. T. Larkin and J. L. Edens, 1994, New
York: Plenum Press. Copyright 1994 by Plenum Press. Adapted with permission.

is a logical place to begin. The following aspects should be considered: its topography (i.e.,
how the action is performed, what it looks like), frequency, duration, and intensity. It is also
important to learn the history of the behavior (e.g., when it first occurred, whether it has
increased or decreased).
A functional analysis also determines antecedents or environmental conditions that imme-
diately precede a given stalking behavior. The consequences that follow the behavior deter-
mine whether or not a particular antecedent will become a controlling variable. If an anteced-
ent stimulus precedes a behavior that is rewarded, that antecedent then signifies an occasion
where future behavior may be reinforced (for an example of a functional relation, see Skinner,
1953, p. 26). The future presence of this antecedent thus increases the likelihood that the
behavior will be repeated.
Figure 1 illustrates the functional relations described earlier. Imagine that a functional
analysis of a stalker who repeatedly telephoned his target determined that he typically tele-
phoned from 5:00 to 7:00 in the evening. It was also found that the act of picking up the
telephone receiver was precipitated by wondering what the target was doing at that point in
time. Further, the functional analysis determined that such thoughts arose while the stalker
prepared his evening meal and that once the thoughts began he experienced increasing unease
and excitement, feelings and sensations alleviated only by the sound of the target’s voice on
the other end of the line. In this example, a particular setting, fixing dinner, sets the occasion
for thoughts and feelings that then serve as antecedents to the telephoning (i.e., fixing a meal,
thoughts about the target, uncomfortable physical sensations and feelings). What the stalker
gains from calling (also revealed by the functional analysis) apparently reinforces the behav-
ior, so that when he again experiences these antecedents he is likely to call again.
To complete the functional analysis, the function of the stalking behavior is determined
by identifying its consequences. It is useful to keep in mind that the function of any given
behavior can be reduced to two basic processes: It allows one to escape and avoid something
 Stalking Behavior 271

or to gain something (O’Neil, Horner, Albin, Storey, & Sprague, 1990; Skinner, 1953). For
example, engaging in stalking behavior may allow the stalker to escape or avoid uncomfort-
able feelings associated with loss or loneliness (i.e., it is negatively reinforced), or the target’s
response may evoke a sense of power in the stalker or provide a source of attention (i.e., it
is positively reinforced). It is possible that both functions are served by a given stalking
behavior; that is to say, the stalking behavior is being multiply maintained (Carr, 1994; Day,
Horner, & O’Neill, 1994; Iwata, Dorsey, Slifer, Bauman, & Richman, 1982). Because stalk-
ing behavior refers to a class of behaviors, a stalker may also cease one stalking behavior
only to resume another that is functionally equivalent (i.e., serves the same purpose). Hence,
functional analysis must be ongoing for each problematical behavior, even as interventions
are underway.
There are many ways to gather the previously described information. It is likely that most
of the information regarding stalking behavior will come from interviews with targets of
the behavior, witnesses of the behavior (e.g., law enforcers, significant others), and stalkers
themselves. Even though individuals who engage in stalking behavior are unlikely to present
themselves voluntarily for treatment of this behavior (Fremouw et al., 1996), it is likely,
given the high rate of comorbid diagnoses reported in the stalking literature, that stalkers are
seen clinically but for other, more familiar disorders (i.e., schizophrenia). Additionally, those
who stalk former intimates may actually be in treatment for relationship or marriage difficul-
ties. It is possible, therefore, that clinical access to stalking behavior is more common than
presently believed. It is important that clinicians be aware of this possibility, because the
identification and subsequent treatment of the behavior rest on the ability to recognize it.

Functional Analysis and Intervention

Once a behavior’s controlling variables are identified, they can then be targeted for effective
treatment (Nelson & Hayes, 1986). Given the complexity of stalking behavior, it is unlikely
that any one factor will stand out as the only contributor (Ferster, 1973; Haynes, 1986).
Deciding what controlling variable(s) to target is thus an important part of the intervention
process and is greatly aided by functional analysis. Hawkins (1986) provided a thorough
discussion on how to best determine the appropriate focal point for intervention. Consider-
ations such as the degree to which a particular variable affects the stalking behavior, how
amenable the variable is to change, and its potential position in a chain of related behaviors
must be considered. A particular variable can influence a group of related behaviors; this
‘‘keystone’’ behavior may accordingly merit attention (Nelson & Hayes, 1986, p. 32). For
example, the delusional belief that a relationship exists between the stalker and target may
be the pivoting point for an entire constellation of behaviors, in which case intervention may
be best directed there.
Intervention may involve clinically treating the stalker, or it may be directed toward the
target of the behavior. De Becker (1994) noted that altering the target’s behavior is sometimes
more effective than attempting to change the stalker’s. In such instances, he has determined
what responses from the target reinforce the stalker and has altered them so that they no
longer do so. A functional analysis of a particular stalking situation could determine that
effective intervention would best be directed toward the target’s responses that are rewarding
the stalking behavior. Many unreported stalking incidents are most likely handled effectively
by targets who perceive what responses encourage the stalker and change their behavior
accordingly. A finding that lends support to this hypothesis is that many incidents of stalking
behavior that did not escalate to legal action were handled by targets who ignored their
stalker (Fremouw et al., 1996). Another possibility is that naturally occurring environmental
variables, such as the stalker meeting someone else or acquiring a new job, affect the entire
272 D. Westrup and W. J. Fremouw

context; the stalker is now provided with other means of reinforcement, or perhaps more
appropriate behavior is being reinforced.
Intervention with stalking behavior is at a fledgling state. Consequently, very little is known
regarding what techniques are effective in what circumstances. Performing a functional analy-
sis provides an immediate avenue for intervention in the individual case. The functional
analytic approach can also contribute to efforts to understand the behavior from a larger
perspective.

Functional Analysis and Research

Stalking research thus far has followed a course typical of the educative process described
by Scotti, Morris, McNeil, and Hawkins (in press). A problem was recognized and labeled
as stalking after observations of certain unwelcome behaviors (e.g., telephoning, following)
with common features (e.g., repetitive, threatening) accumulated. Descriptions of these obser-
vations followed, as seen in case reports and the group descriptive studies cited in the previous
literature review. Next (or simultaneously), potential classification schemes were proposed
that attempted to sort out and organize stalkers based on some criterion. Ideally, the created
taxonomy promotes the development of theory and testable hypotheses. Because of the het-
erogeneous nature of the stalking population, however, distinct features have not arisen for
the most part.
If there are typical conditions that elicit stalking behaviors, the consistent and detailed
information accumulated through the widespread use of functional analyses should eventually
illuminate patterns of purposeful behavior. (If the behavior is random, or nonpurposeful, then
no amount of study will benefit us!) These patterns should in turn broaden the field of scien-
tific inquiry and, it is hoped, lead to increased understanding of this phenomenon. In fact,
functional analyses may lend support to some of the theories regarding stalking behavior
that have already been suggested. The supposition that stalkers engage in stalking behavior
due to social isolation is such an example. It has been suggested that, as a type of self-
medication, a delusional stalker may irrationally believe he or she is in a relationship with
the target in order to rectify unbearable loneliness (Meloy, 1996b; Meloy & Gothard, 1995).
Another possibility is that the stalking behavior is one manifestation of a maladaptive social
behavior repertoire. Such individuals are unlikely to enjoy a mature relationship. Both possi-
bilities emphasize the function the stalking behavior is serving—a question best addressed
with functional analysis.

CONCLUSION
This article has reviewed the literature on stalking behavior from the pragmatic perspective
of whether the ability to predict and control this problem behavior has been increased. The
previous efforts have fallen short of this goal, due in part to the nature of the task as well
as to the approaches taken. Given the concern generated by this problem and the limited
state of current intervention efforts, it is arguably wise at this point to postpone nomothetic
approaches in favor of gathering badly needed detailed and functional information on stalking
behaviors.
The functional analysis has been proposed as a means to gather information that will guide
intervention with not only stalkers but with targets as well. The clinician would be provided
with details as to when and why an individual is stalking, subsequently learning what might
help him or her to stop. Information would also be gained as to how the targeted individual
may best respond to being stalked, which would be of interest to anyone who must contend
with stalking situations. It is possible that patterns of behavior will eventually emerge that
 Stalking Behavior 273

lead to productive research. It may be found, for example, that individuals who respond to
relationship dissolution in a particular manner may be more likely to stalk their former part-
ner. Discovery of such a behavioral pattern would guide subsequent research, even leading
to preventive treatment. This example is only one hypothesized outcome of the information
potentially garnered by functional analyses of stalking behavior; the real benefits of this
assessment technology remain to be seen. It is hoped that clinicians and other involved profes-
sionals will utilize this powerful analytic tool when assessing stalking behaviors.

REFERENCES
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washing-
ton, DC: Author.
Anderson, S. (1993). Anti-stalking laws: Will they curb the erotomanic’s obsessive pursuit? Law and Psychology
Review, 17, 171–191.
Barrett, B. H., Johnston, J. M., & Pennypacker, H. S. (1986). Behavior: Its units, dimensions, and measurement.
In R. O. Nelson & S. C. Hayes (Eds.), Conceptual foundations of behavioral assessment (pp. 1–41). New York:
Guilford.
Carr, E. G. (1994). Emerging themes in the functional analysis of problem behavior. Journal of Applied Behavior
Analysis, 27, 393–399.
Day, H. M., Horner, R. H., & O’Neill, R. E. (1994). Multiple functions of problem behaviors: Assessment and
intervention. Journal of Applied Behavior Analysis, 27, 279–289.
De Becker, G. (1994, March). Intervention decisions: The value of flexibility. Paper presented at the CIA threat
management conference, Washington, DC.
DeVinne, P. B., et al. (Eds.). (1982). The American heritage dictionary. Boston: Houghton-Mifflin.
Dietz, P. E., Matthews, D. B., Martell, D. A., Stewart, T. M., Hrouda, D. R., & Warren, J. (1991). Threatening and
otherwise inappropriate letters to members of the United States Congress. Journal of Forensic Sciences, 36, 1145–
1468.
Dietz, P. E., Matthews, D. B., Van Duyne, C., Martell, D. A., Parry, C. D., Stewart, T., Warren, J., & Crowder,
J. D. (1991). Threatening and otherwise inappropriate letters to Hollywood celebrities. Journal of Forensic Sci-
ences, 36, 185–209.
Ellis, P., & Mellsop, G. (1985). De Clérambault’s syndrome—A nosological entity? British Journal of Psychiatry,
146, 90–95.
Evans, D. L., Jeckel, L., & Slott, N. E. (1982). Erotomania: A variant of pathological mourning. Bulletin of the
Menninger Clinic, 46, 507–520.
Fein, R. A., & Vossekuil, B. (1996). Stalking behaviors: An introduction to what is known. Unpublished manuscript.
Fein, R. A., Vossekuil, B., & Holden, G. A. (1995, September). Threat assessment: An approach to prevent targeted
violence. [National Institute of Justice: Research in Action, 1–7]. (Available from the U.S. Department of Justice,
Office of Justice Programs, National Association of Justice, Washington, DC 20531).
Ferster, C. B. (1973). A functional analysis of depression. American Psychologist, 28, 857–867.
Fremouw, B., Westrup, D., & Pennypacker, J. (1996). Stalking on campus: The prevalence and strategies for coping
with stalking. Journal of Forensic Sciences, 42, 666–669.
Goldstein, R. L. (1987). More forensic romances: De Clérambault’s syndrome in men. Bulletin of the American
Academy of Psychiatry and Law, 15, 267–274.
Harmon, R. B., Rosner, R., & Owens, H. (1995). Obsessional harassment and erotomania in a criminal court popula-
tion. Journal of Forensic Sciences, 40, 188–196.
Hawkins, R. P. (1986). Selection of target behaviors. In R. O. Nelson & S. C. Hayes (Eds.), Conceptual foundations
of behavioral assessment (pp. 1–41). New York: Guilford.
Haynes, S. N. (1986). The design of intervention programs. In R. O. Nelson & S. C. Hayes (Eds.), Conceptual
foundations of behavioral assessment (pp. 1–41). New York: Guilford.
Hollender, M. H., & Callahan, A. S. (1975). Erotomania or Clérambault syndrome. Archives of General Psychiatry,
32, 1574–1576.
Horner, R. H. (1994). Functional assessment: Contributions and future directions. Journal of Applied Behavior
Analysis, 27, 401–404.
Iwata, B. A., Dorsey, M. F., Slifer, K. J., Bauman, K. E., & Richman, G. S. (1982). Toward a functional analysis
of self-injury. Analysis and Intervention in Developmental Disabilities, 2, 197–209.
Larkin, K. T., & Edens, J. L. (1994). Behavior therapy. In V. B. Van Hasselt & M. Hersen (Eds.), Advanced abnormal
psychology (pp. 497–512). New York: Plenum.
274 D. Westrup and W. J. Fremouw

Leong, G. B. (1993). De Clérambault syndrome (erotomania) in the criminal justice system: Another look at this
recurring problem. Journal of Forensic Sciences, 39, 378–385.
Leong, G. B., & Silva, J. A. (1992). The physician as erotomanic object. Western Journal of Medicine, 156, 77–
78.
Mace, F. C. (1994). The significance and future of functional assessment methodologies. Journal of Applied Behavior
Analysis, 27, 385–392.
Meloy, J. R. (1996a). Stalking (obsessional following): A review of some preliminary studies. Aggression and
Violent Behavior: A Review Journal, 1, 147–162.
Meloy, J. R. (1996b). Unrequited love and the wish to kill. Bulletin of the Menninger Clinic, 53, 477–492.
Meloy, J. R., & Gothard, S. (1995). Demographic and clinical comparison of obsessional followers and offenders
with mental disorders. American Journal of Psychiatry, 152, 258–263.
Morin, K. (1993). The phenomenon of stalking: Do existing state statutes provide adequate protection? San Diego
Justice Journal, 1, 123–162.
Mullen, P. E., & Pathé, M. (1994a). The pathological extensions of love. British Journal of Psychiatry, 165, 614–
623.
Mullen, P. E., & Pathé, M. (1994b). Stalking and the pathologies of love. Australian and New Zealand Journal of
Psychiatry, 28, 469–477.
National Institute of Justice Association. (1993). Project to develop a model anti-stalking code for states. Washing-
ton, DC: Author.
Nelson, R. O., & Hayes, S. C. (1986). The nature of behavioral assessment. In R. O. Nelson & S. C. Hayes (Eds.),
Conceptual foundations of behavioral assessment (pp. 1–41). New York: Guilford.
Neufeldt, V. et al. (Eds.). (1988). Webster’s new world dictionary. New York: Simon & Schuster.
O’Neil, R. E., Horner, R. H., Albin, R. W., Storey, K., & Sprague, J. R. (1990). Functional analysis of problem
behavior: A practical assessment guide. Sycamore, IL: Sycamore Publishing.
Perez, C. (1993). Stalking: When does obsession become a crime? American Journal of Criminal Law, 20, 263–
280.
Romans, J. S. C., Hays, J. R., & White, T. K. (1996). Stalking and related behaviors experienced by counseling
center staff members from current or former clients. Professional Psychology: Research and Practice, 27, 595–
599.
Rudden, M., Sweeney, J., & Frances, A. (1990). Diagnosis and clinical course of erotomanic and other delusional
patients. American Journal of Psychiatry, 147, 625–628.
Scotti, J. R., Morris, T. L., McNeil, C. B., & Hawkins, R. P. (in press). DSM-IV and disorders of childhood and
adolescence: Can structural criteria be functional? Journal of Consulting and Clinical Psychology.
Seeman, M. V. (1978). Delusional loving. Archives of General Psychiatry, 35, 1265–1267.
Segal, J. H. (1989). Erotomania revisited: From Kraepelin to DSM-III-R. American Journal of Psychiatry, 146,
1261–1266.
Skinner, B. F. (1953). Science and human behavior. New York: The Free Press.
Taylor, P., Mahendra, B., & Gunn, J. (1983). Erotomania in males. Psychological Medicine, 13, 645–650.
Ullman, L. P., & Krasner, L. (1969). A psychological approach to abnormal behavior. Englewood Cliffs, NJ:
Prentice-Hall.
West Virginia Code. (1995). 62-2-9.
Zona, M. A., Sharma, K. K., & Lane, J. (1993). A comparative study of erotomanic and obsessional subjects in a
forensic sample. Journal of Forensic Sciences, 38, 894–903.