Abnormal Attentions Toward The British Royal Family: Factors Associated With Approach and Escalation

Sample Selection

This study was undertaken as part of the Fixated Research Project, financed by the United Kingdom Home Office (interior ministry). It concerned a retrospective analysis of cases drawn from a pool of 5,702 files compiled by the Royalty Protection Division of the Metropolitan Police Service over a period of 15 years (1988–2003) on individuals who had made abnormal or inappropriate communications or approaches to members of the Royal Family. An abnormal or inappropriate communication was an attempt to communicate by letter, e-mail, telephone, or facsimile, when it went beyond normal interest. Included in this group were those whose communications stood out because of their content—the bizarre, incomprehensible, unrealistic, threatening, sexual, or persistent—and those communications that stood out because of their form, such as writing in blood or including odd enclosures. An approach was an inappropriate attempt to achieve proximity to members of the Royal Family, a royal residence, or royal event. Such approaches were qualitatively or quantitatively different from the actions of Royalty fans and included odd or threatening behavior in royal locations and attempts, successful or otherwise, to breach security cordons at royal residences or royal engagements.

The raw material from which the data were extracted comprised both police files in paper form and a computerized police intelligence record. The latter was organized on an incident rather than person basis. The format necessitated printing out 20,000 incident records and sorting them manually into material relating to specific individuals. The process of data extraction was conducted by two fulltime research workers (a psychiatrist and a clinical psychologist) and took two years. A preliminary pilot project was undertaken with 500 cases, to aid in the refinement of the methodology and the construction of the final dataset.

The 5,702 files were separated into groups according to the type of behavior in which the individuals had engaged. A stratified random sample was drawn from the file pool by using a random number generator until approximately 50 cases had been selected from each behavioral group. The study sample comprised 275 cases, which formed the basis for the consideration of behavior and mental state. Of these, 53 engaged in communications only, 53 engaged in communication and approach without breaching security, 58 engaged in approach without breaching and without having first communicated, 54 attempted to breach but were unsuccessful, and 57 were successful in breaching. In 26 of these 275 cases, insufficient information was available to reliably separate them into motivational groups. These cases were excluded from analyses involving motivational group, which were performed on a sample of 249 cases. For all the other variables examined in the tables in this article, the completeness of data ranged from 95.3 to 100 percent, unless otherwise stated.

Data Extraction

A 125-item data sheet was completed on each case and the data points entered into an SPSS database, with an automatic reader used to reduce the likelihood of transcription errors. The data gathered concerned sociodemographic information, and details of behavior, mental state, and motivation. A strict definition of serious illness was used. Its presence was recorded if any of the following were in evidence: obviously testable delusions; thought disorder, as illustrated in written material or recorded verbal output; clear evidence of abnormal perceptions (e.g., hearing voices), or evidence of passivity phenomena. Serious mental illness was also recorded as present when the files contained clear documentary evidence, taken from hospital records, of an ICD-10¹¹ diagnosis of F20 or F21-29. (The ICD classification is that used by the National Health Service in the United Kingdom for diagnostic returns on all patient contacts.) With regard to motivation, the cases were allocated to eight groups, produced through earlier study of the file pool.^1,12 The motivations of the subjects were determined through the content of their writings and the explanations they offered to police when stopped. Where more than one form of motivation appeared to be present, the most prominent was selected. The motivational categories are described in the results section below. Other terms are defined where they first occur.

Statistical Analysis

To determine differences between groups on categorical variables, analyses were performed by using Pearson's chi-square (χ²) where the appropriate assumptions were met. Where assumptions were violated, Fisher's exact tests were used. Odds ratios were also calculated. Independent t tests were used for comparing the ages of groups.

Effect sizes were also calculated for each measure of association, because the group sizes in some of the analyses performed were uneven and some were relatively small, thus reducing power and increasing the probability of making Type II errors (i.e., failing to detect existing relationships). The use of effect sizes enabled interpretation of the data beyond, and independent of, the information provided by p values,¹³ providing further indications as to the strength of associations. The measure of effect size used was the phi coefficient (φ).¹⁴

Multivariate logistic regression was undertaken, with individual predictors selected for the regression equation if their univariate association was significant (significant odds ratio and chi-square significant at p < .25¹⁵). Receiver operating characteristic (ROC) curve analysis was used to ascertain the discriminatory power of the full logistic regression models and subsequent improper models. The area under the curve (AUC) in a ROC analysis is taken to indicate the probability that a randomly selected case from the group in question will have a greater number of predictors included in the model than a randomly selected case not in the group in question (i.e., the true positives as a function of the false positives, or sensitivity as a function of 1 − specificity). An improper model^16,17 is one in which the weights assigned to variables are obtained by a nonoptimal method. In the present study, all were made equal to 1, to overcome some of the traditional limitations of multiple regression. By setting the weights equal to 1 and retaining only the direction of the relationship to the criterion variable, the influence of sample-specific variance is reduced, and the model may have greater applicability outside the originating sample.^16,17

Multiple Testing

The study involved the analysis of a library of files, in which we examined parameters suggested by the literature. However, the purpose of the exercise was an exploration of possible associations, rather than the testing of hypotheses. As such, multiple testing was used. No corrections to significance values were incorporated to compensate for multiple testing. In consequence, conclusions drawn from p > .01 should be treated with caution.

Data Quality Assurance Checks

Formal testing was undertaken of the reliability of case behavioral type and classification of motivation by using Cohen's κ. Determination of case type showed 92 percent concordance (κ = 0.80) and motivation showed 81 percent concordance (κ = 0.77). The most frequent discordant pairing was between cases categorized as chaotic (described later) and those in which there was insufficient information to divine a motivation.¹²

Ethics

The project concerned retrospective consideration of police files, with anonymization of data. It did not involve any access to medical files or health service information systems. It did not involve any contact with the subjects, or any form of intervention. The research group operated within the relevant ethics frameworks determined by the Home Office and the Metropolitan Police Service, the sponsor and host of the study, respectively.

Approach

Those who approached were separated from the study sample (n = 275) and their characteristics established. They were then compared with those who did not approach. Of the approach cases (n = 222), 70 percent involved males. The mean age was 39.1 years (SD, 12.8). In the first of eight motivational groups were 31 percent who had delusions of royalty; these persons either believed that they were members of the Royal Family or that they themselves were the true sovereign. In the second group were the 11.7 percent who were intimacy seekers; these either harbored erotomanic delusions of a reciprocated relationship with a member of the Royal Family, or they were offering love or marriage with absolute confidence that they would succeed in their suit. The third group comprised the 18.8 percent who were amity seekers, subjects who made offers of friendship, which they expected to be accepted, apparently oblivious to the unrealistic nature of their aspiration. Group four were the 3.0 percent who believed that they were victims of organized persecution by the Royal Family. The fifth group comprised the 6.6 percent who were querulants, people who were pursuing highly personalized quests for justice and vindication. They wanted either to recruit royal assistance with their claims or to complain of royal indifference to their causes. The sixth group contained the 7.6 percent who were seeking help or sanctuary, usually by requesting royal protection from supposed persecutors. The seventh group were the 7.1 percent who wanted to act as counselors, a group of individuals who saw it as their role to offer advice and directions to the Royal Family as to how the Family should live their lives and who expected their advice to be taken. Finally, the eighth group comprised the 14.2 percent who were chaotic, individuals whose thought processes and behavior were so disturbed as to make a singularity of purpose unlikely. These eight groups represented the eight motivational types.¹¹

Approachers (n = 222) were significantly more likely than nonapproachers (n = 53) to show overt evidence of serious mental illness (i.e., psychosis) (χ² = 9.169; p = .002; φ = 0.183; OR = 2.88; 95% CI, 1.422–5.813); to be deluded (χ² = 9.557; p = .002; φ = 0.186; OR = 2.571; 95% CI, 1.397–4.739); to be grandiose (χ² = 20.390; p = 0.000; φ = 0.272; OR = 4.329; 95% CI, 2.217–8.403); and to show delusions of royalty (χ² = 7.894; p = .005; φ = 0.178; OR = 3.436; 95% CI, 1.395–8.475). Approachers were less likely to be rambling or incoherent (χ² = 14.050; p = .000; φ = −0.226; OR = 0.315; 95% CI, 0.170–.587) and less likely to be counselors (χ² = 16.187; p = 0.000; φ = −0.255; OR = 0.206; 95% CI, 0.092–0.471).

Escalation

The sample for considering escalation excluded those cases in which there was an approach, but no communication. It comprised 53 cases in which communication was not followed by approach, 53 cases in which it was followed by approach that did not involve breaching activity, and 26 cases in which it involved breaching activity. The total sample included 132 subjects.

The statistical results from this section are detailed in Tables 1 and 2. Significant associations in the table are summarized in the text, giving only the odds ratio (OR), with more details set out in the tables.

Differences Between Approach and Communication Group and Approach Without Communication Group

Another portion of the sample approached without first communicating (n = 49). There were significant differences between this group and those who communicated as well as approached. Details of these are given here in the text, rather than in tabular form. Those who approached without communicating were more likely to behave in an intimidating manner on approach (χ² = 21.425; p = .000; φ = 0.311; OR = 3.817; 95% CI, 2.137–6.849) and more likely to engage in breach activity (χ² = 15.321; p = .000; φ = 0.263; OR = 3.096; 95% CI, 1.742–5.495). In terms of mental state, they were less likely to be deluded (16.5%; χ² = 12.058; p = .001; φ = −0.223; OR = 0.316; 95% CI, 0.163–0.617); less likely to feel persecuted (χ² = 5.711; p = .017; φ = −0.161; OR = 0.455; 95% CI, 0.236–0.875); less likely to be grandiose (χ² = 8.787; p = .003; φ= −0.199; OR = 0.410; 95% CI, 0.226–0.745); and less likely to be rambling or confused (χ² = 26.925; p = .000; φ = −0.348; OR = 0.217; 95% CI, 0.120–0.394). They were less likely to be fixated on a person (10%; χ² = 9.968; p = .002; φ = −0.212; OR = 0.283; 95% CI, 0.125–0.614) and less likely to be intimacy seekers (χ² = 12.514; p = .000; φ = −0.237; OR = 0.207; 95% CI, 0.081–0.529). They were more likely to number among the chaotic (χ² = 8.741; p = .003; φ = 0.211; OR = 4.672; 95% CI, 1.553–14.085) than were the approachers who communicated.

Approach and Escalation

It is evident from the results of this study that there are significant differences between those who simply communicate and those who communicate and then approach, and that these differences may be of use in predicting which groups of communicators are at increased risk of escalating their behavior to inappropriate approach. This analysis has potential importance in threat assessment, because preventing inappropriate approach equates with reducing the levels of threat to Royal Family members.

Three motivations together accounted for 71.8 percent of cases in which communicators went on to approach: those with delusions of royalty, amity seekers, and intimacy seekers. The first two groups were significantly more likely than the remainder of cases to escalate to approach. All three groups comprised persons with a particular sense of entitlement toward the royal individual, whether through perceived blood ties, love, or friendship, a quality described by Meloy et al.¹⁸ as entitled reciprocity: the person's belief that he is owed a debt of gratitude, or at least contact, by the public figure because of his devotion. In contrast, those who saw their role as offering counsel to the royal personage were significantly less likely to have a need to press their advice in person, and chaotic letter writers were also less likely to approach those to whom they had written. The other motivational groups accounted for only small proportions of letter writers, whether they approached or not, except for the help seekers. In that group, there is an indication that the type of approach may vary according to motivation,¹² as 13.5 percent of those who approached without breach activity were seeking help, but none of those who chose to breach security cordons or barriers. Most important, angry and hostile themes occurred in nearly five times as many of the communications in the group of those who engaged in breach activity than in the communications of the group of those approachers who did not.

Major mental illness and evidence of delusional beliefs were strongly associated with approach, being almost ubiquitous among approachers/communicators, although also present in most of the simple communicators. Grandiosity was a central factor in this type of behavior and appeared in the regression model predicting approach, while mental illness and delusional beliefs did not. Indeed, the one factor of whether grandiosity was present correctly predicted nearly three-quarters of the communicators who went on to approach and a similar proportion of those who did not. Such grandiosity is congruent with the motivations that accounted for most of the approacher cases. The findings concerning grandiosity provide some empirical support for the theory of Meloy^19,20 of narcissistic linking fantasies, conscious beliefs that one has a special and idealized relationship with another. Grandiosity, of course, is not self confidence, but rather the disparity between the facts of the subject's life and the sense of his own importance. In many cases, grandiose fantasies compensate for real-life failures in both work and love. Grandiosity is one facet of pathological narcissism, an aspect of personality that is quite apparent in most stalkers.^19,21 It complements entitlement in the sense that the subject's demands remain concordant with his expansive sense of self. Why would he not be entitled to see the Queen when his importance required it? It is also noteworthy that the grandiosity in most of our subjects passed the threshold of delusion, wherein consensual reality, even quirky consensual reality, is replaced by a private, idiosyncratic, and, in some cases, bizarre internal fantasy.

Dietz and Martell³ similarly found that any grandiose delusion appeared in 60 percent of all their subjects who inappropriately communicated to celebrities, and 44 percent demonstrated an excessive sense of self-importance or uniqueness. Most salient was their finding that those who approached celebrities were significantly more likely (χ² = 4.85, p < .03) to evidence an excessive sense of self-importance or uniqueness (52%) than those who did not approach (36%). Although this finding did not emerge in the Dietz et al.⁵ study for the U.S. Congress, if the subject took a special constituent role, it predicted an approach (46% vs. 16%, χ² = 7.77, p = .0053). Such self-identification also hints at pathological narcissism and perhaps grandiosity.

Nearly 80 percent of those who went on to approach had written more than once, and more than 25 percent had telephoned. More than 40 percent of those who engaged in breach activity had written to other prominent people as well as members of the Royal Family, a form of target dispersion.⁸ These significant findings reflect those problematic contacts with another group of prominent people in another country: members of the U.S. Congress, as reported by Dietz et al.⁵ and Scalora et al.^7,8 Approach behavior appears to be associated with more intrusive forms of attempted communication, multiple means of communication, and target dispersion, perhaps all measuring a more energized individual wherein the pathological fixation is fuelled by both intense affect and desire, usually psychotically driven.

The logistic regression model showed that the presence of three factors (grandiosity, multiple communications, and multiple means of communication) correctly predicted 77.5 percent of cases of escalation (sensitivity) and 69.8 percent of cases in which it did not occur (specificity). The tradeoff between sensitivity and specificity in the construction of the model is represented in the AUC, which at 0.82 indicates good discriminatory power. That the improper model differs little from the substantive model indicates that the model may have wider applicability than simply to the originating sample. In practical terms, the presence or otherwise of the three factors in the model provides an initial indication of the risk of escalation which will then be supplemented in the individual case by scrutiny for idiographic factors associated with increased or decreased risk. The power of the single-factor model comprising grandiosity alone suggests that this factor can be used as a rapid initial screen.

Comparison with Stalker Samples

Few stalker studies or reviews have touched on the topic of approach,^22,23 and there has been little attention paid to escalation. The exception in this respect is the Melbourne group, who examined a cohort of 211 stalkers for associations of approach and excalation (McEwan T, MacKenzie RD, James DV, et al: Approach and escalation in stalking: a study of associations. Manuscript submitted.).The cases were divided into two groups according to whether the stalker was an ex-intimate or not. This was in part to allow comparison with public figure samples. In the sample of 140 non ex-intimate cases, approachers were significantly more likely than nonapproachers to be psychotic (p < .01, OR = 3.85; 28.6% of non ex-intimate approachers were psychotic and 87.5% of the psychotic approached) and to be intimacy seekers in the Mullen et al.²¹ typology (p = .03, OR = 3.94). In this sample of non ex-intimates, factors associated with escalation from communication to approach were psychosis (p < .01, OR = 5.31) and being an intimacy seeker (p < .01, OR = 9.76).

The finding in this general stalker sample concerning psychosis is entirely consistent with the findings concerning the Royal Family sample, in the analyses of approach and of escalation. The findings concerning intimacy seekers are also very similar. The definition of intimacy seeker in the Melbourne studies is wider than that employed in the current study, in which it would be likely to include most amity seekers and people with delusions of royalty. Delusions of royalty were significantly associated with approach, and both amity seekers and those with delusions of royalty were significantly associated with escalation. Overall, in both samples, approach and escalation are associated with psychosis and a particular sense of entitlement to closeness with the individual who is the focus of interest. The associations of approach are the same as those for escalation, but the associations are more significant with escalation.

Strengths and Limitations

The strengths of this study are that it is the first of its kind in the United Kingdom and that a random stratified sampling method was used, which allowed the separation of approaches into different types. Its weaknesses include small samples in some of the subanalyses. Other weaknesses are, for the most part, shared with other studies in this area: retrospective methodology, use of police files, and the determination of the case selection of the base sample by the choice of cases referred to and taken on by the police. This selection bias could have resulted in an overrepresentation of mental illness, if such cases more easily attracted attention. Conversely, it is likely to have resulted in an under-recognition of troubling letters when the form and content were not obviously disturbed. This drawback applies in particular to the querulant, of whom Glueck said: … in his writings … [he] frequently succeeds in conveying the idea of being quite normal. Each isolated fact looks normal enough to the casual observer… . It is only after a close study of the entire life history, of the many fine shades of deviation from the normal that the man exhibits, that one discovers that his mind is very seriously affected indeed, and that because of his plausibility he belongs to a rather dangerous type of mentally diseased individuals [Ref. 24, p 129].

The data with regard to approaches are likely to be more reliable and valid, in that observable behaviors, such as breaching security cordons, are open to more accurate recording. This likelihood is suggested by the comparison of those who approached without writing with those who wrote and approached. The latter were significantly more likely to be deluded, feel persecuted, and be grandiose and confused. They were more likely to be fixated on a person (reflecting specific motivational groups), less likely to behave in an intimidating manner on approach, and less likely to engage in breach activity.

Certain of the less common motivational groups were insufficiently represented in the study sample to allow conclusions to be drawn, and the manner in which the police file archive was amassed may have excluded communications from those in some motivational groups whose outward expression of their concerns appeared lucid and, at least superficially, rational. We single out the querulant, a group that has been found to be of particular concern in other studies conducted by our research group,^11,25,26 when violence or proxies for violence were the focus. The fact that those in this sample who approached, but were not recorded as having communicated beforehand, differed in significant aspects from those who communicated as well, may speak to the characteristics and apparent rationality of unrecorded communications, rather than reflect real differences in whether individuals have previously communicated or not. Study of this troubling group who make inappropriate approaches without first communicating is likely to become more practicable with advances in information technology and the more widespread adoption by correspondence offices of computerized logs of all communications, whether or not these evoked concern. The matter is also complicated by the fact that troubling correspondence may be written to agencies or persons different from the one who is eventually approached. It is necessary to consider prominent people (public figures) as one entity, and it is important that intelligence and threat assessment agencies avoid fragmentation into groups with too specific a focus. Future studies of abnormal attention, whether to prominent persons, celebrities, or the general population, should oversample from the minority of cases concerning fixation on a cause, as opposed to a person, to help elucidate the differences engendered by this fundamental difference in motivation.