Predictors of varying levels of risks posed by fixated individuals to British public figures

Concerning approaches and communications to the Royal Family and other British public figures are relatively numerous. This paper examines over 2000 such cases logged over a three‐year period in the United Kingdom. Using police and health data, the paper conducts a series of bivariate and multivariate analyses to demonstrate the predictors of what types of risk are posed by an individual case (e.g., communicate only, approach, security breach). The results showed that (a) the rates of serious mental disorders are higher among this sample than the general population base rate, (b) approachers were significantly more likely than communicators to suffer from serious mental disorders, (c) approachers were significantly more likely than communicators to have a history of substance use and abuse problems, (d) approachers were significantly more likely than communicators to have a history of violent behavior against property and persons, and (e) the motivations of approachers and communicators significantly differ. The paper concludes with a consideration of the implications for threat assessment and management.


| INTRODUC TI ON
Over the last few years, a number of studies analyzing the socio-demographic characteristics and antecedent behaviors of a range of grievance-fueled violent actors have emerged. These include school shooters [1], mass murderers [2], lone-actor terrorists [3], attackers of politicians, public figures and Royal Family members [4,5], workplace shooters [6], and spree shooters [7]. While providing insight into the behavioral trajectory to such violence, their operational utility may be constrained by the fact that they sample on the dependent variable. Only those that successfully committed or Communicators are defined as individuals who made concerning communications to public figures and did not follow them with an approach. Approachers are individuals who made concerning approaches to their intended targets. This is not the first study of its type. For example, similar studies have been conducted on problematic approachers/communicators to the British Royal Family [5], the Dutch Royal Family [8], Canadian justice officials [9], and U.S.
Congress members and their staff [10][11][12]. However, this paper does offer some considerable improvements to the existing knowledge base. For example, the sample size here is far larger than is typical (The above-mentioned studies range in sample size from 86 to 326, except [11] which had a sample of 4387). This paper makes an additional theoretical contribution to the literature by incorporating insights from criminological studies concerning offender decision-making in the context of risk. We take a hypothesis-driven approach that is atypical of the wider threat management literature. The thinking here is that making approaches toward public figures entails considerable risk to the approacher, given the typically high levels of security surrounding prominent individuals. We posit that a number of offender-, motivation-, and situation-specific predictors linked to accepting risk may account for much of the variance between approachers and communicators.

| THEO RY
The emergent threat management literature offers four key variables that distinguish approachers from communicators. One focuses on the offender (mental health), and three focus on behavior (the nature, presence, and type of previous communications; previous criminal involvement; and motivational type). This paper suggests several additional variables concerned with the nature of the target, security, and risk, based on our understanding of criminal decision-making.

| Mental health
A major strand of threat management research consistently highlights the presence and role of mental disorders in aiding and abetting the offender's fixation and concerning behavior toward the public figure. "It has often been assumed that mentally ill assailants… have motives so irrational that they cannot be understood or have no motives other than their illness" [4, p.328]. This is clearly erroneous. Fein and Vossekuil [4] themselves found evidence of mentally ill individuals planning and executing attack-related behaviors as effectively as non-mentally ill actors. Corner and Gill highlight that lone actors diagnosed with mental illness frequently display rational motives and engage in rational and purposive pre-attack behaviors [13]. Borum notes numerous mentally ill lone actors who were capable of sophisticated attack planning [14]. Corner and Gill empirically compared a sample of mentally disordered lone-actor terrorists with a sample of non-mentally disordered lone-actor terrorists [13]. They found that those who were mentally disordered were just as likely (and in some cases more likely) to engage in a range of rational preattack behaviors as those who were not.
In comparative studies of communicators and approachers, the nature of the mental disorder tends to discriminate between groups.
For example, Adams et al. illustrated that approachers to Canadian politicians were significantly more likely to be psychotic compared with communicators [15]. James et al. demonstrated that approachers were significantly more likely than communicators to possess overt evidence of serious mental illness (e.g., psychosis) [5]. Others found similar results [10][11][12]. Additionally, Schoeneman et al. found that approachers were significantly more likely to have had previous substance use/abuse problems [12]. Given this evidence base, it is hypothesized that:

| Previous criminal involvement
In terms of previous criminal involvement discriminating between approachers and communicators, the evidence is mixed. For example, Scalora et al. found that approachers typically had more prior criminal charges than communicators and that these typically related to drugs/alcohol, theft/burglar, assault, and weapons offenses [11,16]. On the other hand, Eke et al. found approachers had engaged in significantly less offending [9]. Scalora's finding is perhaps more rigorous given the difference in sample size between the two studies (>4000 vs. 86). Therefore, it is hypothesized that:

| Previous communications
"Leakage" is a term used for the intentional or unintentional disclosure of planned violence against others [17]. Studies of various grievance-fuelled, violent offenders demonstrate remarkably high levels of leakage. These include studies of school shooters (81%) [18], mass murderers (67%) [19], attackers and approachers to public figures (63%) [4], and lone-actor terrorists (51%) [3]. However, the numbers who communicate direct threats to the target and later follow-up on them are proportionately quite small. For example, in a victim survey of Canadian politicians, Adams et al. found no relationship between threatening communications and subsequent physical approaches [15].
While a single communication may be a poor risk factor for physical approach, the threat management literature argues that the volume of previous communications may be more important. For example, approachers write more frequently than communicators to a variety of victim groups including celebrity stalkers [20], the British Royal Family [5], United States Congress members [10,11,21], and federal agencies [16]. Given the above evidence, we hypothesize that: There may also be a case to suggest that the level of effort involved may indicate a greater propensity for more worrying follow-up behaviors. We therefore might expect that correspondence which involves little effort and that can be conducted relatively spontaneously (e.g., email, social media communications) may be less "risky" than those communications that involve a number of steps (e.g., handwriting a letter, placing it in an envelope, and taking it to the post).

| Motivational types
Studies focused upon school shooters, mass murderers, lone-actor terrorists, and spree shooters typically tend to base their inclusion criteria on the underlying political, social, individual, or psychological motivation behind the violence. In comparison, most studies of public figure communicators and approachers define their inclusion criteria by these two particular behaviors. The study of motivation is either completely omitted or treated in aggregate terms (e.g., personal vs. political). Exceptions to this are James et al. [5,22] who first developed and later tested the presence of eight motivational groups.
The most common motivational group defined by James et al.
had "delusions of royal identity," believing either they were the true sovereign or related to the sovereign [5,22]. These individuals were also over-represented among those who attempted to breach security cordons. The second group were "amity seekers" who offered friendship and advice to the public figure and expected it to be embraced. Again, James et al. found this group were over-represented in terms of both communicating and approaching [5]. The third group were "intimacy seekers" who believed they were loved by, destined to be with or already married to the public figure, and had either erotomanic preoccupations or infatuations with the public figure. The fourth group were "sanctuary and help seekers" who sought assistance and/or protection from personal adversity and/or persecutors.
The fifth group were "royally persecuted," claiming to be victims of persecution conducted by a public figure. The sixth group were "counselors" who "saw it as their role to offer advice and opinions to the Royal Family on how they should live their lives and respond to political situations" [22. p1483]. This group are significantly more likely to communicate without approaching [5,22]. The seventh group were "querulants" who pursued highly personalized quests for justice and vindication. James et al. found this group disproportionately more successful at breaching security cordons [22]. The final group were the "chaotic," to whom "no clear motivation could be assigned because their writings and/or their statements to police were so difficult to follow or understand" [22, p 1493]. Given this, it is hypothesized that:

| Hypotheses
Given our outlines of the existent evidence base above, we hypothesize the following: H1: The rates of serious mental disorders will be higher in this sample than the general population base rate.
H2: Approachers will be significantly more likely than communicators to suffer from serious mental disorders.
H3: Approachers will be significantly more likely to have a history of substance use and abuse problems.
H4: Approachers will be significantly more likely to have a criminal history.
H5: Approachers will be significantly more likely to have a history of violent behavior against property and persons.
H6: Approachers will be significantly more likely to have a history of harassing behavior.
H7: Those individuals who both communicate and approach will engage in significantly more correspondence than those who only communicate.
H8: Different motivational types will display significantly different risks of communicating, approaching, and breaching.
Below, we outline the data and methods used to test these hypotheses.

| SAMPLE
The source material comprised 2088 files compiled by the Fixated Threat Assessment Centre (FTAC) in the calendar years 2013-2015.
This covers the universe of cases dealt with by the unit during this time. FTAC is a joint police/mental health unit that has been in operation since 2006. It has a remit to assess and manage the risk to the British Royal Family, members of parliament and other prominent political figures from lone individuals who inappropriately communicate with or attempt to approach them. These data relate to cases of one or more incidents of inappropriate approach or communication toward figures within the remit of the unit. Inappropriate approaches included attempts at unauthorized physical contact, breaching security barriers, trespass in physical locations connected to prominent individuals, repeated loitering near such venues in a manner that raised concern, or unauthorized entry into events attained by deception.
Our dependent variable is split into a number of behavioral outcome categories. This is coded based upon the content of the referral made into FTAC and any accompanying documentation from the referrer (e.g., the threatening letter and witness statements). It is coded in real-time by FTAC personnel including trained and experienced mental health nurses, clinical psychologists, forensic psychiatrists, and police. Referrals came from over 100 locations, professionals, and agencies in the time-period covered by these data. The coded behavioral categories are as follows: 1. Communications. The individuals concerned communicated in an inappropriate manner, which brought them to the attention of FTAC. They had not, however, attempted to approach, nor breach security barriers. Motivation For the Behavior-This is coded by a small team of trained FTAC personnel with backgrounds in clinical psychology and forensic psychiatry. The motivational types were developed in previous FTAC research [5,22] and are outlined in-depth in the literature review section above.
Mental Health Diagnoses, Symptoms, and State of Mind-This is also coded by a small team of trained FTAC personnel with backgrounds in clinical psychology and forensic psychiatry, as well as mental health nurses. These data are coded from a consideration of mental health information they have access to including psychiatric reports, GP history, prior hospitalizations, and/or first-hand assessments.
Concern Levels-At the point of initial referral, a joint team of police, clinical psychologists/forensic psychiatrists, and mental health workers ascribe a risk level to the case based on their predominant concerns emanating from the referral.

| Statistical analysis
Associations between each variable of concern were sought with other items in the dataset. To determine differences/similarities between groups on categorical variables, analyses were performed using Pearson's chi-square (χ 2 ) where the appropriate assumptions were met. Where assumptions were violated, exact tests were used.
Multivariate analyses were run on continuous and dichotomous dependent variables.

| Descriptive results
Nearly three-quarters of the cases only conducted communications (73%), while most of the rest only conducted approaches (23.4%).
Very few (2.7%) conducted both communications and approaches.
Cases overwhelmingly involved a single individual (98.2%) as opposed to a dyad or group of individuals.
Communications tended to be directed to individuals within the Royal Family (46.3%) or individual politicians (26.6%). Many fewer communicated to arbitrary sites such as Royal Family residences (9.1%) or embassies (2.1%). The great majority communicated via either letter (70.9%) or email (20.2%). Of those who communicated, 55.3% did so once, 32% did so between two and nine times, 10% did so more than ten times, and 1.2% did so more than 100 times.
Within the communications, 6.5% made a direct threat and 6.3% made an indirect threat.
The language and content of the communications was also coded, 68.4% of the content being coded as incoherent, 13.8% as demanding, 21.3% as abusive, and 7.1% as containing "end of tether" language. The communications also showed evidence of sexualized (8.1%), amorous (7.2%), help-seeking (44.7%), and help-offering For those who conducted approaches, the majority did so to static sites such as Royal addresses, such as Buckingham Palace (66.1%), or to embassies (10.8%). Smaller numbers approached individual members of the Royal Family (8.5%) or politicians (4.9%).
Of those who did make an approach, 23.1% did so more than once. 56% of approaches were deemed concerning, 41.1% problematic, and 2.7% violent. Only 2.3% of approachers possessed a weapon.
Two hundred and thirty-two approaches concerned an attempt and failure to conduct a breach, while 17 successfully managed to breach security.
In terms of the initial concern level, FTAC assessed 61.7% of cases as low risk, 34.4% as moderate, and 3.9% as high. Following a full investigation and/or intervention, 96.9% were deemed low risk, 1.5% moderate, and only one case was still considered high. At the time of data collection, 1.6% of cases were still open.
In terms of previous criminal behaviors, many individuals had a history of police cautions or criminal convictions (45.8%, with 2.2% having a history of serious violence and/or sexual offending), 24.6% had a UK police record, and 3.7% had a foreign police record. These included a history of violence (22.3%), harassment (10.6%), making threats (6%), and weapon (9.4%), and drug use (4.7%).
In terms of previous correspondence, 6.6% had a previous record were as rambling or incoherent, 21.4% were as persecuted, 65.3% were as deluded, and 12.4% were as hostile. These categories were not mutually exclusive.

| Comparing communicators and approachers
First, we compared those who made communications with those who made approaches (Table 1). If an individual conducted both behaviors, they were treated as an approacher for the purpose of this analysis. Approachers were significantly more likely to have a history of contact with the police (including cautions and convictions). When examining specific criminal behaviors, approachers were significantly more likely to have a history of violence, drugs, and weapon concealment.
There was no significant difference between the two behavioral categories in terms of likelihood of previous serious violence and sexual offenses, a history of harassment or threats, firearms' possession history or access, militancy experience, arson, or having a foreign police record.
Those who solely communicate are significantly more likely to have done so previously (see Table 2). The inverse is also true. Those who have made previous approaches are significantly more likely to repeat such a behavior. The contents of these communications and/ or approaches are unknown and may have been relatively benign or benevolent. The malleable nature of the focus is also reflected in the high degree to which both communicators and approachers change the targets of their fixation. This is significantly more likely within the communicator sub-sample.
Next, we compared mental health backgrounds in both subsamples ( Table 3). The data suggest more serious mental disorders occur within the approacher sub-set. They are significantly more likely to have display symptoms of an overt mental disorder at assessment and have been previously admitted to a hospital. The type of diagnosis is also important to consider here with approachers significantly F I G U R E 1 Comparing prevalence rates of disorders between sample and general population [Colour figure can be viewed at wileyonlinelibrary.com] more likely to have previously been diagnosed with bipolar disorder and delusional disorder. They are also significantly more likely to have a substance abuse problem. There were no significant differences in terms of schizophrenia psychosis, personality disorder, or depression.
In terms of mental state at the time of assessment (Table 4), approachers were significantly more likely to be deemed by the psychologists and/or psychiatrists to be grandiose and organized.
Communicators were significantly more likely to be coded as  In terms of motivations (Table 5), those who only communicated were significantly more likely to be categorized as holding a resentful agenda or seeking help. Those who followed their communication with an approach were significantly more likely to be seeking friendship or be chaotic. There were no significant differences between communicators and approachers in terms of resentful, persecuted, erotic, incompetent suitor, or attention-seeking motivations.

TA B L E 3 A bivariate comparison of illness history of communicators and approachers
A binary logistic regression analysis (

| Comparing communicators and communicator-approachers
Earlier, it was demonstrated that communicator-approachers comprised a very small sub-sample of the whole dataset (2.7%).
We compared those who only communicated (communicatoronly) with those who communicated and later approached (communicator-approachers; Table 6). The latter group were significantly more likely to have a foreign police record, send packages, express their intention to approach, have previously approached, and present as organized. Motivational types also proved a consistent differentiator between the two sub-types.
Communicator-approachers were significantly more likely depicted as seeking friendship, being an incompetent suitor, and seeking attention. There was no difference in terms of the volume of communications (Table 7).
A binary logistic regression analysis was conducted on the variables with significance values of p < 0.01 and p < 0.001. The results indicated that previous approaches were significantly associated with communicator-approachers, χ 2 (6) =65.752, p < 0.001. The model correctly predicted 95.3% of the cases (Table 8). approaches, and to be categorized as an attention-seeking type.

| Comparing approacher-only and attempted breachers
They were significantly less likely to have a history of cautions or convictions, a UK police record, a history of threats, to be diagnosed with bipolar or personality disorder, or to be characterized as chaotic (Table 9).

| Comparing successful and unsuccessful breachers
The next analyses focused on comparing successful and unsuccessful breachers (Table 10) They were significantly less likely to be help-seeking. Third, it may be due to effective interventions by agencies such as FTAC subsequent to the initial communication. As for approachers, many skip the communication segment entirely. Table 11 outlines the results of our hypothesis testing.

TA B L E 7 A bivariate comparison of communicator-only and communicator-approachers
The results do not, of course, enable precise estimation of the risk in a single case. Such a hope is common among some academics and politicians not well versed in the limitations of risk assessment.
In practice, each individual case should be assessed on its own merit based on the clustering of different behaviors and motives, rather Key: *p <.05; **p <.01; ***p <.001

Accepted/ Rejected
1 The rates of serious mental disorders will be higher in this sample than the general population base rate.
Accepted 2 Approachers will be significantly more likely than communicators to suffer from serious mental disorders.
Partially Accepted 3 Approachers will be significantly more likely to have a history of substance use and abuse problems.
Accepted 4 Approachers will be significantly more likely to have a criminal history.
Accepted 5 Approachers will be significantly more likely to have a history of violent behavior against property and persons.

6
Approachers will be significantly more likely to have a history of harassing behavior.

7
Those individuals who both communicate and approach will engage in significantly more correspondence than those who only communicate.  and nervousness [28] Just under one-fifth of Kang and Lee's sample admitted to substance use prior to a robbery in order to "fortify themselves" [29]. Gill found a negative correlation between degrees of planning and whether substances had been consumed prior to the offense [30]. These points lead us to assume that approachers may include a sizeable proportion of almost spontaneous non-planners who act opportunistically and may therefore be less likely to engage in correspondence beforehand.
Previous criminal engagement inconsistently predicts next behavior across the sequences depicted in Figure 2. Whereas it intuitively delineates approachers from communicators, and successful from unsuccessful breachers, it also unexpectedly predicts a breach will not be attempted once the approach has been conducted. The reasons behind the first two results may be found in the wider criminological literature. There, prior criminal experience has been linked with "more elaborate target schemes" [31, p147], increased confidence that helps monitor and control behavior and nerves [32], dehumanization of the victim [33], decreased concern about being detected and caught [34][35], perceptual and procedural skill superiority and self-belief [36], highly developed offense conduct scripts to fall back on [37], and decreased worries concerned with sanctioning by family members [35]. individual behavior appears to be a disinhibitor for further risky behavior and at others appears to be an inhibitor. For example, one of the key differentiators between communicators and approachers on a bivariate level is whether the individual has the same target focus. Where the same focus was present, the individual is significantly more likely to communicate. However, in those smaller numbers with a same focus that do approach, there is then a significantly more likely chance of an attempted breach. In other words, "same focus" predicts a less risky behavior in the first instance (e.g., to communicate). Should the riskier initial behavior be chosen, an even riskier behavior (e.g., to attempt breach) is more likely. The same pattern is also true for homicidal ideation being apparent in the individual's mental state. Homicidal ideation predicts communications rather than approaches. When present during an attempted breach, however, it also predicts success. and many more, within the data. Sequencing and pathway analyses, however, have the ability to provide a sense of the prevalence of such interdependencies within the data. It may be irrelevant for prediction or assessment, but could be crucial for management particularly in the case of attempting to formulate what the individual of concern gained from such worrying or threatening behaviors [38].
It must also be remembered that this is a correlational design.
We may be able to infer causation via formulation on a case-to-case basis by gaining a mapping as to how all the key variables crystallized for the individual experiencing them. It would be a struggle to do so based on such an aggregate analysis as this. Horgan correctly notes that, although studies of lone actors, for example, often find a high preponderance of mental health problems within the sample, "detailed research would be needed to further clarify the precise nature and role (if any) of mental health problems in the development of their violent activity" [39]. This debate is ongoing within the wider study of crime also. On the one hand, a strand of research assumes a consistent causal link between psychiatric symptoms (where they are found to be present) and criminal behavior. On the other hand, a more nuanced strand of research argues there are "a (small) group of offenders whose symptoms relate directly to crime and a (larger) group whose symptoms and crimes are not directly related" [40, p1].
For example, various studies illustrate that the offender (across a range of crimes) was experiencing the symptoms of their symptoms at the time of their (often violent) crime between 4% and 18% of the time [23]. The same may also be true for factors such as motivation, previous behaviors, and mental states. Only deep case-study-driven approaches will provide such nuanced understanding of the functional role that different risk factors have for the outcomes being considered.