The Relevance of Victimization Experiences for Predicting the Development of Threat Symptoms During Adolescence and Early Adulthood

Abstract

The “threat” portion of the threat/control override symptom constellation is characterized by the belief by an individual that others seek to do them harm. There has yet to be an examination of the link between perceived threat symptomatology and the experience of victimization, despite the fact that both constructs are linked to the perpetration of violent behavior. Furthermore, there has yet to be research which examines the heterogeneity in developmental patterns of threat symptomatology. The present research utilized the Pathways to Desistance data to model the developmental heterogeneity in perceived threat symptomatology across adolescence and early adulthood using group-based trajectory modeling. A series of multinomial logistic regression models were then estimated to examine the relevance of victimization experiences for predicting trajectory group assignment. A five-group model of development during adolescence and early adulthood best fit the threat symptom data. Victimization experienced prior to age 16 predicted assignment to all groups in the model characterized by presentation of threat symptoms at some point during the study. Victimization experienced between ages 16 and 23 predicted assignment to the High Chronic group and marginally predicted assignment to the Accelerating group. These results indicate that victimization does indeed predict the presentation of perceived threat symptomatology. This indicates that presentation of threat symptoms may be indicative of a history of victimization. Proper screening of those presenting threat symptoms may help to better orient treatment plans for modalities centered on victimization experiences. Future research should investigate the possibility that threat symptoms mediate the relationship between victimization and violent behavior.

Keywords

violence exposure mental health and violence youth violence

Introduction

Threat/control override (TCO) symptoms represent a distinct set of psychotic symptoms that past research has linked to the perpetration of violent behavior (Imai, Hayashi, Shiina, Sakikawa, & Igarashi, 2014; Link, Stueve, & Phelan, 1998; Teasdale, Silver, & Monahan, 2006). Individuals suffering from these specific symptoms have higher likelihood of engaging in violent behavior (Nederlof, Muris, & Hovens, 2011; Swanson, Borum, Swartz, & Monahan, 1996). Often, this subclass of psychotic symptoms is further subdivided into “threat” symptomatology and “control override” symptomatology. Presentation of the “threat” symptomatology is associated with beliefs by an individual that others are seeking to do them harm. The “control override” symptomatology is associated with the feeling that the individual is not in control of his or her own thoughts in some way. Despite the relevance of these highly intrusive and disruptive symptoms, there has yet to be an investigation of how perceived threat symptomatology develops across time. This lack of interrogation has led to a dearth of knowledge related to whether or not presentation of this type of symptomatology is relatively stable across time or that it may be a dynamic developmental process. If the process is indeed characterized by change and heterogeneity in developmental patterns, there is little understanding of how social influences may affect this development. One social factor which may hold a great deal of relevance for this process is the experience of victimization. The present research sought to provide an examination of this development and examine the experience of victimization as a risk factor predicting this development. The identification of victimization experience as a risk factor for the arousal of threat symptoms may indicate the importance of screening for trauma when patients present these symptoms. While most individuals who experience victimization may not go on to present threat symptoms, this remains a highly relevant phenomenon which may be undeniably disruptive, as TCO symptoms have been linked to increased risk for perpetration of violence (Imai et al., 2014; Link et al., 1998; Teasdale et al., 2006).

Specific Aims

The present research sought to elucidate better understanding of the threat symptom cluster of the TCO constellation and examine the possible contribution that experiencing victimization may have for inciting an individual’s perception that others may seek to do him or her harm. There are several specific aims which the present study sought to accomplish:

The first goal of this research was to investigate the heterogeneity in general developmental patterns of perceived threat symptoms across adolescence and early adulthood. Despite extensive research on the topic, there has yet to be a systematic investigation of the differences which exist in general longitudinal patterns of change and continuity in symptom presentation across time. Using group-based trajectory modeling (GBTM; Nagin & Odgers, 2010), this study sought to provide better understanding of differences in how symptoms tend to develop across time.

The second phase of analyses sought to examine the relevance of victimization experienced for predicting general patterns of perceived threat symptom development. Time frames for victimization were delineated as occurring prior to baseline or during the study period. If individuals are experiencing or have experienced victimization at the hands of others, this may influence their perception that others do indeed seek to do them harm. The scale utilized to capture this concept in the Pathways to Desistance study is the Threat Control Override Symptoms Scale (Link et al., 1998). There has yet to be any research which examines how social influences like victimization may influence these perceptions of threat.

Perceived Threat Symptom Presentation and Victimization

The perceived threat symptom component of the TCO symptom constellation is characterized mainly by individuals’ feeling that others are out to do him or her harm (Link et al., 1998). There is reason to believe that the biological changes and learning processes facilitated by the experience of victimization may actually make such beliefs and the behaviors associated with them quite rational. Individuals who have been victimized have been found to have heightened levels of distrust toward others and paranoia regarding the actions of others (Li et al., 2014; Wilcox & Duffy, 2016; Young, Pruett, & Colvin, 2018). This is understandable given the violation that occurs when one is victimized by another individual. This understanding of paranoia regarding the actions of others following the infliction of trauma, like victimization, may lead to cognitive patterns characterized by constant hyperarousal and overreactive threat perception (Briere, Godbout, & Dias, 2015; Coyne, Barrett, & Duffy, 2000). These cognitive patterns bear a great deal of resemblance to those of the threat appraisal characteristic of TCO symptoms. Certainly, such appraisal may seem dysfunctional and delusional, as it is unlikely that the vast majority of others are seeking to do harm to an individual. However, one must also consider mechanisms of the learning process which individuals who have been victimized have undergone to provide the full context in which these interpretations of others’ actions exist.

Individuals who have been victimized may present a learned belief about the intentions of others as being malevolent. They have been taught by past experiences that others will indeed act to harm them. In this way, perceptions that others are seeking to do them harm and behaviors stemming from these feelings may actually be rational reactions to the world around them and, likely for many, adaptive survival mechanisms. Research indicates that those individuals who experience victimization are often revictimized, either because of the ways in which one act of victimization is likely to be part of a larger pattern of complex trauma or via other mechanisms (Ruback, Clark, & Warner, 2014; Walker, Freud, Ellis, Fraine, & Wilson, 2019). Perceived threat symptomatology then may actually be a highly rational means of not experiencing further victimization for some. If the individual learns that others will seek to do them harm, then the individual may presume that paranoia of the actions of others may be an adaptive survival mechanism protecting against further harm. These learned beliefs regarding others’ intentions and impulsive and reactive behaviors may then, theoretically, manifest in the violent behaviors to which TCO symptoms have been linked (Hodgins, Hiscoke, & Freese, 2003; Link et al., 1998). The proposed learning process involved in the presentation of threat symptomatology appear to complement our understanding of how threat symptomatology may become aroused by social influences and explain why some individuals who had been victimized may feel others seek to do them harm.

A final component of the proposed victimization–threat symptomatology relationship which bears further investigation relates to the temporal aspect of these processes. While general psychotic symptoms associated with schizophrenia often demonstrate a chronic course (Harrow, Jobe, & Faull, 2014; Jobe & Harrow, 2010), there has yet to be any research which examines the development of threat symptomatology specifically and examines the nature of temporality of victimization experiences for predicting symptom course. It may be that the experience of victimization may have a lasting impact on threat symptomatology, that is, victimization experienced early in life may lead to chronic symptomatology. Alternatively, the effects of victimization early in life may only have a transient effect, demonstrated by remission of symptoms following demonstration at earlier points in the life course. Furthermore, victimization experienced later in life may cause symptoms to surface when they did not exist in the absence of victimization earlier in life. The present study sought to provide better understanding of the relationship between victimization and threat symptomatology and the development of threat symptomatology. These goals were tackled via the use of longitudinal data to elucidate general developmental patterns of perceived threat symptomatology and examine the relevance of victimization experiences prior to the initiation of the elucidated developmental trajectory groups (prior to age 16) and during the course of symptom presentation measured by these trajectories (between ages 16 and 23).

Method

Data

The present research utilized the Pathways to Desistance data in analyses (Mulvey, 2000-2010). The Pathways to Desistance study captured the data of 1,354 juvenile offenders across the 84 months following their conviction of a serious criminal offense, resulting in 11 total data points for participants. These observation periods were spaced 6 months apart for the first 36 months of the study, at which point data collection occurred every 12 months for the remainder of the study. This purposive sample was recruited from collection sites in Maricopa County, AZ, and Philadelphia, PA. Recruitment took place between 2000 and 2003, with the full data collection period spanning 2000-2010. Serious offenses which qualified participants for inclusion in the study were mainly felonies, with some more serious misdemeanors (weapons charges, sexual assault, etc.) also qualifying participants for inclusion. Qualifying offenses had to have been committed when participants were between the ages of 14 and 17, resulting in a sample ranging between 14 and 18 years of age at baseline (one participant was 19 years of age at baseline). The proportion of male drug offenders included in the sample was capped at 15% to maintain heterogeneity in qualifying offense type. Twenty percent of potential participants declined to take part in the study when approached for recruitment by members of the research team. In all, 16.8% of participants were lost to attrition between baseline measurements and the final data collection point.

Computer assisted interview technology was utilized in the primary data collection efforts to obtain participant self-report data. Participants were provided with laptops by the research team to provide self-report data. Prompts were provided by members of the research team and participants responded to these prompts using the laptops provided. This procedure maximized privacy for participants as they provided data on a number of sensitive topics. Members of the research team met with participants in a number of locations to collect data, including participants’ homes, public places, and criminal justice facilities. Collateral data were also collected from several other sources to provide triangulation of participants’ self-report data and provide data pertaining to subjects of which participants may not be reliable reporters (e.g., attention deficit hyperactivity disorder [ADHD] symptomatology). Peer and parent report data were gathered for the first 36 months of the study, at which point collection was discontinued due to diminishing returns related to the relevance of such data. Official records provided a final form of collateral data. These records provided an objective measure of participants’ offending and treatment while under criminal justice supervision.

Measures

Perceived threat symptoms

TCO symptoms are delineated into categories of “threat” symptoms and “control override” symptoms. Analyses of this research focus on threat symptoms which concern perceptions of an individual centered on the belief that other people wish to do them harm. This construct was measured via the Threat Control Override Psychotic Symptoms Scale. Beliefs regarding threat appraisal were operationalized via an ordinal measure capturing how often in the past week that each participant felt that there were people who wished to do them harm (0 = never, 1 = almost never, 3 = sometimes, 4 = often, 5 = very often).¹ This variable was measured at every time point, allowing for the tracking of change and continuity across time in symptom presentation. The TCO scale and measure utilized in the present analyses have demonstrated criterion validity for predicting the perpetration of violence in past research (Link & Stueve, 1994; Link et al., 1998).

Victimization experience

Directly experiencing victimization has been identified as one of the most robust predictors of psychopathology (Klomek, Sourander, & Elonheimo, 2015; Silberg et al., 2016). The Pathways to Desistance study measured victimization experience at each time point via a measure of the number of types of victimization experienced by participants during each observation period using the My Exposure to Violence Interview (Selner-O’Hagan, Kindlon, Buka, Raudenbush, & Earls, 1998). The original version of this scale was found to have test–retest reliability and high internal consistency (Selner et al., 1998). The present research utilizes two recoded operationalizations of this original variable as predictors of threat symptom development. Both variables were dummy coded versions of the original victimization variable delineating individuals who had experienced victimization during a given observation period from those who had not (0 = no, 1 = yes). Because this research focused on the relevance of victimization experience for predicting the development of threat symptoms across time, these two variables separately captured victimization experience prior to the commencement of the trajectories (prior to age 16) and victimization experienced during the developmental period described (ages 16 to 23).

Control variables

To best reduce bias in estimation, a number of additional variables were included in analyses to best control for possible confounding effects. The first of these variables was participant gender. Past research has demonstrated that the prevalence of psychotic symptom presentation varies by gender (American Psychiatric Association [APA], 2013). Participants’ gender was captured as a dummy coded baseline measure which delineated male and female participants (0 = males, 1 = females).

Another variable included in analyses to mitigate bias in estimations was participants’ race. Racial differences in psychotic symptom prevalence have been detected by past research (APA, 2013). Participant race was measured at baseline via self-report of self-identification (1 = White, 2 = Black, 3 = Hispanic, 4 = Other). A series of dummy variables were computed to estimate the effects of race on threat symptom presentation. The dummy variables which delineated Black, Hispanic, and Other categories from all other racial categories were included in analyses. This resulted in effects interpretable as each of these racial identities in relation to the omitted reference category (White).

Socioeconomic status (SES) is another construct which past research has linked to mental health (Bøe et al., 2014). Because of this, a baseline indicator of participants’ SES was included in analyses to reduce bias in estimations. An SES score was assigned to each present parental caregiver for each participant. This score was a weighted average of educational attainment and occupational prestige. In cases of participants who had both parents present, the average of both parents’ individual SES scores was computed to provide a single SES measure for each participant to be used in analyses.

A final set of control variables included in analyses were “control override” symptom indicators. Because TCO is composed of both “threat” symptoms and “control override” symptom clusters, and both symptom types comprise the TCO symptom constellation, it is necessary to control for possible confounding effects by including these indicators as control variables in analyses. Two “control override” symptom indicators were available in the Pathways to Desistance data, both measured ordinally (0 = never, 1 = almost never, 2 = sometimes, 3 = often, 4 = very often).² The first measured how often in the past week that participants felt that thoughts in their head were not their own. The other captured how often in the past week that participants felt that their mind was dominated by forces beyond their control. Baseline measure of both of these constructs were included in analyses to control for possible confounding effects related to the correlation between these TCO symptoms and the developmental trajectories of threat symptomatology.

Analytic Strategy

The analyses of the present research occurred in two phases. The first phase of analyses utilized GBTM to elucidate general developmental patterns from the heterogeneity in individual threat appraisal response data. This method entails an iterative process of fitting varying numbers of groups and the shapes of those groups to the response data to identify the model which best fits the data based on a number of fit criteria. Each participant is assigned a probability of assignment to each group based on their own individual trajectory. Participants are assigned membership to the group to which they have the highest probability of assignment. Following the elucidation of the threat symptom trajectory model, the second phase of analyses utilized a series of multinomial logistic regression models to examine the relevance of risk factors for predicting group assignment. Coefficients in the models indicated the risk of assignment to trajectory groups based on covariates, relative to an omitted reference category.

Hypothesis 1: Assignment to threat symptom trajectory groups will be predicted by having experienced victimization.

The GBTM method involves an inductive process of elucidating a trajectory model that may have any number of groups and these groups may follow any path within the range of response options for the outcome variable of interest. Because of this, and because there does not exist any past research which has focused on the development of TCO symptoms across time, a priori specification of predicted effects on assignment to specific groups is not possible. However, given the characteristics of threat symptoms and the robust relationship between victimization and psychopathology (Tsai, Weiser, Dilworth, Shumway, & Riley, 2015; Turner, Finkelhor, & Ormrod, 2006), there are some predictions that may be made. It is hypothesized that prior victimization history (prior to age 16) and victimization which occurred during the period of the life course covered by the trajectory model (ages 16-23) will predict assignment to trajectory groups characterized by elevated levels of threat symptomatology. Significant effects would indicate that those individuals presenting elevated levels of threat symptomatology also were more likely to experience victimization. Differences in the effects of these victimization variables by trajectory group are expected to be related to the temporality of change and continuity in threat symptom presentation. Victimization experienced prior to age 16 should be related to increased risk of assignment to groups characterized by high baseline levels of threat symptoms and possibly chronic course; whereas victimization experienced between ages 16 and 23 would be expected to be related to elevations in threat symptom presentation during this period of time, rather than having elevated levels of symptom presentation at baseline.

Results

The iterative and inductive GBTM process resulted in the elucidation of a five-group model which best fit the threat symptomatology response data. Model selection was made based on a number of criteria. The first major criterion for model selection was the comparison of Bayesian Information Criteria (BIC) statistics obtained for each model. The five-group model provided better fit to the data than did two, three, or four group models. A six group model fit the data better based on this criterion, but the addition of a sixth group did not provide any additional nuance. Because of this, the more parsimonious five-group model was chosen. The shapes of each group also needed to be elucidated. Linear, quadratic, or cubic polynomial functions could be utilized to represent the developmental track of each group. For the chosen model, four groups were found to be best described using the linear polynomial function and one group was described by a quadratic function.³ Nagin (2005) prescribes model selection based on this rationale. Beyond model comparison based on obtained BIC statistics, there are a number of other criteria which a model should meet if it is to ultimately be chosen as the best fitting model. Nagin (2005) notes that posterior probabilities of assignment, average odds of correct classification, and 95% confidence intervals for each should also be considered in model selection. Posterior probabilities refer to the probabilities that each participant has for assignment to each trajectory group in the model. The average probability of assignment for participants assigned to each group should exceed .7. This eschews the possibility of high numbers of participants with similar probabilities of assignment to multiple groups being assigned to a given group, thus reducing the meaning of group assignment. Posterior probabilities of assignment for each group are described in Table 1. Average odds of correct classification refer to the improvement of the odds correctly identifying the group to which a given participant was actually assigned based on their posterior probability of assignment, relative to random chance. This improvement should exceed 5, or a 500% improvement in odds. Finally, 95% confidence intervals for each group in the model should be relatively tightly bound around each group. The chosen five-group model meets all of these additional criteria.

Table 1.

Posterior Probabilities of Assignment to Delusional Threat Symptomatology Trajectory Groups.

	Low Decelerating	Nonpathological	Accelerating	High Chronic	High Decelerating
Low Decelerating	.792	.047	.083	.004	.075
Nonpathological	.095	.876	.025	.001	.003
Accelerating	.075	.015	.837	.049	.025
High Chronic	.005	<.001	.050	.899	.046
High Decelerating	.109	.001	.033	.066	.791

Note. Italicized font indicates matched probabilities.

The chosen five-group model is described in Figure 1. The first group in the model was characterized by low levels of symptom presentation at the age 16 intercept, followed by complete remission around age 20. This group is described as the “Low Decelerating” group. This group was represented by a linear polynomial function and 24.8% of participants were assigned to this group. The second trajectory group described in Figure 1 was characterized by a complete lack of any threat symptom presentation at any point during the study period. Because of this consistent lack of symptoms, this group is described as the “No Symptoms” group. This group was characterized by a linear path and 40% of the sample was assigned to this group. Group 3 was characterized by slight and steady acceleration in symptom presentation across the study period following a low intercept at age 16. This group is described as the “Accelerating” trajectory group. A linear polynomial function best described the general developmental pattern of this group and 15.3% of the sample was assigned to this trajectory group. The fourth group in the model was characterized by consistently high symptom presentation relative to other groups in the model, with slight deceleration in presentation across the study period. Because of this developmental track, this group is described as the “High Chronic” group. This trajectory group was represented by a linear polynomial function and 9.4% of the sample was assigned to this group. The final group described in Figure 1 was characterized by a relatively high intercept at age 16, followed by a slight increase in symptom presentation before decelerating and full remission around age 21. This group is described as the “High Decelerating” group and was represented by a quadratic polynomial function, and 10.5% of the sample was assigned to this group.

Figure 1.

Delusional threat symptomatology development among juvenile offenders.

The second phase of analyses entailed the estimation of a series of multinomial logistic regression models to examine the relevance of victimization and other covariates for predicting assignment to the elucidated trajectory groups. Model 1 examined the relevance of the two victimization indicators as the sole predictors of group assignment. Model 1 results are described in Table 2. Model 2 also not only examined the effects of the two victimization indicators of interest but also included relevant covariates to control for bias in estimations. Model 2 results are described in Table 3. Results from Model 1 indicate that both prior victimization history and concurrent victimization are relevant for predicting trajectory group assignment. Having a history of victimization prior to age 16 was associated with significantly elevated risk of assignment to the Low Decelerating (2.781), Accelerating (1.891), High Chronic (4.837), and High Decelerating (3.886) trajectory groups, relative to assignment to the No Symptoms group. Experiencing victimization between the ages of 17 and 23 as associated with significantly elevated risk of assignment to the Accelerating (2.202) and High Chronic (3.611) groups, relative to assignment to the No Symptoms group.

Table 2.

Relative Risk of Assignment to Delusional Threat Symptomatology Trajectory Groups Based on Covariates: Model 1 (Reference = Nonpathological).

	Relative Risk Ratio	p Value	95% Confidence Interval
Low decelerating
Victimization prior to age 16 (0 = no, 1 = yes)	2.781	<.001	[1.954, 3.959]
Victimization ages 16-23 (0 = no, 1 = yes)	1.311	.337	[0.754, 2.277]
Constant	0.325	<.001	[0.254, 0.415]
Nonpathological	— (omitted)	—	—
Accelerating
Victimization prior to age 16 (0 = no, 1 = yes)	1.891	.003	[1.243, 2.877]
Victimization ages 16-23 (0 = no, 1 = yes)	2.202	.007	[1.241, 3.906]
Constant	0.220	<.001	[0.166, 0.293]
High chronic
Victimization prior to age 16 (0 = no, 1 = yes)	4.837	<.001	[2.571, 9.102]
Victimization ages 16-23 (0 = no, 1 = yes)	3.611	<.001	[1.777, 7.338]
Constant	0.049	<.001	[0.023, 0.083]
High decelerating
Victimization prior to age 16 (0 = no, 1 = yes)	3.886	<.001	[2.152, 7.017]
Victimization ages 16-23 (0 = no, 1 = yes)	1.533	.317	[0.664, 3.539]
Constant	0.072	<.001	[0.045, 0.114]

Table 3.

Relative Risk of Assignment to Delusional Threat Symptomatology Trajectory Groups Based on Covariates: Model 2 (Reference = Nonpathological).

	Relative Risk Ratio	p Value	95% Confidence Interval
Low decelerating
Victimization prior to age 16 (0 = no, 1 = yes)	2.557	<.001	[1.705, 3.835]
Victimization ages 16-23 (0 = no, 1 = yes)	1.530	.178	[0.824, 2.842]
Gender (0 = males, 1 = females)	0.606	.090	[0.340, 1.080]
Race (ref = White)
Black	0.983	.951	[0.575, 1.681]
Hispanic	0.840	.551	[0.473, 1.490]
Other	1.059	.915	[0.367, 3.056]
SES	0.986	.114	[0.969, 1.003]
TCO Dom. Thoughts	1.677	.011	[1.128, 2.491]
TCO thoughts not own	1.462	.037	[1.023, 2.091]
Constant	1.113	.858	[0.344, 3.603]
Nonpathological	— (omitted)	—	—
Accelerating
Victimization prior to age 16 (0 = no, 1 = yes)	1.761	.024	[1.076, 2.882]
Victimization ages 16-23 (0 = no, 1 = yes)	1.996	.051	[0.998, 3.992]
Gender (0 = males, 1 = females)	0.966	.916	[0.509, 1.835]
Race (ref = White)
Black	1.532	.220	[0.775, 3.028]
Hispanic	1.050	.897	[0.502, 2.194]
Other	0.718	.689	[0.142, 3.639]
SES	1.001	.954	[0.980, 1.022]
TCO Dominated thoughts	1.255	.384	[0.753, 2.092]
TCO thoughts not own	1.344	.187	[0.866, 2.085]
Constant	0.164	.015	[0.038, 0.709]
High chronic
Victimization prior to age 16 (0 = no, 1 = yes)	4.017	.001	[1.827, 8.830]
Victimization ages 16-23 (0 = no, 1 = yes)	4.344	.001	[1.813, 10.408]
Gender (0 = males, 1 = females)	0.250	.052	[0.062, 1.011]
Race (ref = White)
Black	2.056	.190	[0.699, 6.048]
Hispanic	0.960	.946	[0.296, 3.115]
Other	1.857	.485	[0.327, 10.542]
SES	0.980	.222	[0.949, 1.012]
TCO Dom. thoughts	2.933	<.001	[1.776, 4.844]
TCO thoughts not own	1.212	.487	[0.704, 2.087]
Constant	0.292	.313	[0.023, 3.180]
High decelerating
Victimization prior to age 16 (0 = no, 1 = yes)	3.175	.002	[1.525, 6.610]
Victimization ages 16-23 (0 = no, 1 = yes)	1.711	.294	[0.627, 4.664]
Gender (0 = males, 1 = females)	0.330	.085	[0.094, 1.163]
Race (ref = White)
Black	.514	.147	[0.209, 1.263]
Hispanic	.456	.105	[0.177, 1.179]
Other	0.356	.355	[0.040, 3.175]
SES	0.986	.353	[0.956, 1.016]
TCO Dom. thoughts	1.767	.040	[1.025, 3.047]
TCO thoughts not own	1.746	.031	[1.051, 2.900]
Constant	0.631	.674	[0.073, 5.416]

Note. SES = socioeconomic status; TCO = threat/control override.

Model 2 examined the effects of victimization history prior to age 16 and victimization experienced between ages 16 and 23 on threat symptom trajectory group assignment while controlling for confounding effects via the inclusion of relevant covariates in estimation. Results from Model 2 indicate that both victimization variables remain relevant for predicting group assignment even after the inclusion of control covariates. Victimization experienced prior to age 16 was associated with significantly increased risk of assignment to the Low Decelerating (2.557), Accelerating (1.761), High Chronic (4.017), and High Decelerating (3.175) trajectory groups, relative to assignment to the No Symptoms group. After the inclusion of covariates, victimization experienced between ages 17 and 23 was only significantly associated with assignment to the High Chronic (4.344) group, relative to assignment to the No Symptoms group. However, this association was also essentially significant for the Accelerating group as well (p = .051). Few significant effects were observed for control covariates as they related to risk of assignment to trajectory groups. Only control override symptoms demonstrated significant effects on the risk of group assignment. Increased degrees to which participants reported feeling as though that their thoughts were being dominated by forces beyond their control was associated with elevated risk of assignment to the Low Decelerating, High Chronic, and High Decelerating trajectory groups, relative to assignment to the No Symptoms group. Increased degrees to which participants reported feeling as though the thoughts inside their heads were not their own were associated with increased risk of assignment to the Low Decelerating and High Decelerating trajectory groups, relative to assignment to the No Symptoms group. No significant effects of race, gender, or SES on risk of assignment were observed.

Comparative model fit of the estimated multinomial logistic regression models was assessed via the use of BIC statistics. Obtained BIC statistics indicated that Model 2 provided better fit to the data than did Model 1. The obtained BIC statistics for each model can be found in Tables 2 and 3 with coefficients related to the corresponding model.

Discussion

The results of this study provide insight into the development of threat symptomatology and provide evidence that experiencing victimization is predictive of symptom presentation. Experiencing victimization prior to age 16 significantly increased the risk of assignment to all groups in the chosen trajectory model which demonstrated some level of threat symptomatology at some point between ages 17 and 23. Furthermore, experiencing victimization between the ages of 17 and 23 was associated with assignment to the High Chronic trajectory group and essentially predicted assignment to the Accelerating group as well. Interestingly, experiencing victimization during this time did not predict assignment to the High Decelerating or Low Decelerating groups. The High Chronic and Accelerating trajectory groups were the only two groups in the model characterized by consistent presentation of some degree of threat symptomatology across the entire study period. These results indicate that individuals who demonstrate some level of perception of others wishing to harm them are likely to have actually been harmed by someone in the past. Considering the wealth of research on victimization and revictimization (Moreira et al., 2015; Ruback et al., 2014; Walker et al., 2019), there is reason to believe that the individuals’ perceptions that others may want to do harm to them may actually be quite rational for some. This is indicative of the importance of this social influence on this psychological construct. Considering that both TCO symptoms and victimization are robust predictors of violent behavior (Imai et al., 2014; Sadeh, Binder, & McNiel, 2014; Teasdale et al., 2006), there exists the possibility that the three constructs exist along a causal pathway. It may be that the victimization–violence relationship is mediated in part by the threat cluster of the TCO symptoms. The belief that others may seek to do them harm (aroused by the experience of victimization) may lead to hypervigilance and hyperarousal that could lead to the choice of violence as a response to perceived provocation. Furthermore, recent research has indicated that other psychological constructs play mediational roles in this same relationship (Baskin-Sommers & Baskin, 2016), so threat symptomatology playing a similar role seems possible. Future research should examine the possibility that these symptoms mediate this relationship and also the role that the control override symptom cluster and the full TCO constellation may play for mediating the victimization–violence relationship.

The elucidated trajectory model presents several attributes of the development of threat symptomatology that are of interest. This is the first study to examine the development of threat symptom presentation across the life course utilizing the GBTM method. This in itself is a step forward, as a much clearer picture of the general developmental heterogeneity of threat presentation during adolescence and early adulthood is now available. While psychotic disorders, like schizophrenia, generally commence in late adolescence and early adulthood (APA, 2013), the elucidated trajectory groups demonstrate that a sizable portion of participants were demonstrating threat symptomatology prior to age 16. Furthermore, despite an often chronic course for schizophrenia (Harrow et al., 2014; Jobe & Harrow, 2010), a great many participants were assigned to trajectory groups characterized by deceleration and remission of threat symptomatology following presentation of symptoms at age 16. These general developmental trends provide rationale for researchers to delineate psychotic symptomatology by symptom cluster or constellation. Doing so may provide better understanding of which aspects of schizophrenia and/or psychoticism are generally characterized by chronicity and which symptom types may remit over time. Further investigation of this may allow for more accurate treatment of symptomatology in the future. It should be made clear that this does not suggest that all participants who displayed any level of symptomatology were suffering from psychoticism. This is actually quite unlikely considering the low prevalence of psychotic disorders (APA, 2013). Rather, these results simply provide impetus for further investigation of whether or not some individuals who demonstrate remission of threat symptoms may also present psychoticism which also remits. Beyond the elucidation of general developmental trends in threat symptomatology, this research also demonstrated the relevance of victimization experiences and the importance of the temporal nature of such experiences for predicting this development.

The findings of the present research indicate that individuals presenting threat symptoms are likely to have been victimized at some point during their lives. The proposition of this research is that this observed relationship is due to the learned belief that others are indeed seeking to do individuals who present these symptoms harm that is precipitated by the experience of harm at the hands of others. There also appeared to be a temporal aspect to this relationship. Relative to the No Symptoms group, only victimization experienced prior to age 16 significantly increased the risk of assignment to the Low Decelerating and High Decelerating trajectory groups. Victimization prior to age 16 also predicted assignment to the Accelerating and High Chronic groups, but the consistent presentation of symptoms across the entire study period that was characteristic of these groups was also predicted by victimization experienced during the disorder course. This provides some evidence that the effects of victimization may remit over time and that only the continued experience of victimization will exert an effect that buoys the belief that others seek to do individuals presenting threat symptomatology harm. This is demonstrative of the need for further investigation of the effects of treatment which takes a trauma informed care (TIC) perspective on threat symptomatology. TIC is a treatment philosophy focused on recognizing the effects that victimization has on the biology and behavior of individuals that has a major orienting aim of reducing revictimization during the course of treatment at all costs. TIC involves the creation of a treatment environment characterized by safety, the facilitation of relationship creation and maintenance, and the construction of emotional self-regulation (Bath, 2008). This entails the education of staff about the effects of victimization, training staff regarding how to implement the philosophy in treatment and practice, and the use of restraining mechanisms or treatment techniques that may result in experiencing further trauma only as a last resort (Barton, Johnson, & Price, 2009; Guarino, 2014). Because TIC is more of a guiding orientation to treatment, rather than a specific treatment modality, demonstrating effectiveness of TIC for reducing mental health symptomatology has been limited in the sense of traditional evaluative assessment. However, there is some preliminary evidence that TIC facilities may reduce the use of restraints which may serve to further traumatize patients (Azeem, Aujla, Rammerth, Binsfeld, & Jones, 2011; Muskett, 2014). The results of this research would seem to lend credence to the belief that this may have an impact on reducing threat symptomatology. These results indicate that proper screening of individuals presenting threat symptoms may be paramount to provide the services and treatment that best accommodates the needs of victims, as those presenting threat symptoms appear to be at elevated risk for having experienced victimization.

Despite the strengths of this study, there are a number of limitations to this research which temper the obtained results described here. The first of these limitations pertains to the GBTM method utilized to elucidate general developmental patterns of threat symptomatology. The elucidated trajectory groups exist only as general approximations of development, rather than concrete representations of paths which individuals assigned to groups actually follow across time. These groups are better understood as a means of providing visual representation of otherwise intractable heterogeneity in response data. Despite this acknowledgment of the method’s limitation, GBTM does indeed provide perhaps the most powerful analytical tool available for describing heterogeneity in development across time currently available. Another limitation of the present research is the unknown generalizability of results based on the indicated sample utilized. The Pathways to Desistance sample solely consists of juvenile offenders. Mental disorder exists at high rates among this population of adolescents (Underwood & Washington, 2016); thus, the results described in the trajectory model may not be generalizable to the general population. Despite this, this research provides a first step for better understanding the development and predictors of development of threat symptomatology in adolescence and early adulthood. Future research should seek to investigate the validity of these results by conducting similar research among a general population sample of individuals of the same ages. Another limitation of the present study relates to the victimization measure used in analyses. The original measure of victimization captures the variety of victimization types which an individual experienced in each observation period. While there is merit to using this measure as is, the publicly available Pathways to Desistance data do not allow delineation of individual items, nor is full information available regarding what each measure included in the scale describes. These issues make the use of the scale in its original form problematic; therefore, the decision was ultimately made to dichotomize the variable. A final limitation of this research pertains to the measure utilized to measure victimization which occurred between ages 16 and 23. This is obviously a very rough measure of victimization experienced during a wide range of time. The measure, as it is currently operationalized, does not allow for investigation of specific effects of victimization at particular ages or examine how these specific effects may be associated with trends in threat symptomatology development for each group described in the trajectory model. This lack of temporal specification also limits claims to causality for this measure. However, the prior victimization measure examining the effects of victimization prior to age 16 does not suffer from this final limitation, as temporality of these victimization experiences precedes the commencement of trajectory groups. Also, this research again should simply be viewed as a first step in the process of parsing out the association of victimization with threat symptomatology. Future research should seek to further examine the effects of victimization on threat symptomatology at more specific points in the life course.

The limitations described above demonstrate the great deal of research left to be conducted on victimization and threat symptomatology. However, the present research provides a strong first step in better understanding these phenomena and provides empirical support for the notion of this social influence on this psychological construct. The extant literature focused on the effects of victimization provide evidence that a mind-set characterized by distrust and paranoia regarding others’ actions may indeed be a relatively normative response to victimization trauma, particularly complex trauma. The present study indicates that for many, these symptoms, to some degree, may be aroused by the experience of victimization and may remit over time in its absence. This indicates the importance of understanding threat symptomatology as a dynamic process and this was the first study to approach this construct from this perspective. Future research should seek to examine other elements of the TCO symptom constellation from a similar perspective to investigate how these processes of development may be similar or dissimilar to the threat symptom cluster. Furthermore, research should focus on investigating the possibility that threat symptoms, or TCO symptomatology more broadly, mediate the relationship between victimization and violent behavior.

Footnotes

Acknowledgements

I would like to thank the faculty at the University of Florida who helped me better understand the methodologies employed by this research project.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Thomas W. Wojciechowski is currently finishing his PhD in Sociology at the Universtiy of Florida while also working as the Juvenile Competency Program Coordinator for the Michigan Department of Health and Humans Services. He completed his BA in Criminal Justice at Central Michigan University and his MA Sociology at the University of Florida. His research focuses on life-course processes invoolved with mental health, substance use, and violence.

References

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.

Azeem

M. W.

Aujla

Rammerth

Binsfeld

Jones

R. B.

(2011). Effectiveness of six core strategies based on trauma informed care in reducing seclusions and restraints at a child and adolescent psychiatric hospital. Journal of Child & Adolescent Psychiatric Nursing, 24, 11-15.

Barton

S. A.

Johnson

M. R.

Price

L. V.

(2009). Achieving restraint-free on an inpatient behavioral health unit. Journal of Psychosocial Nursing and Mental Health Services, 47, 34-40.

Baskin-Sommers

A. R.

Baskin

(2016). Psychopathic traits mediate the relationship between exposure to violence and violent juvenile offending. Journal of Psychopathology and Behavioral Assessment, 38, 341-349.

Bath

(2008). The three pillars of trauma-informed care. Reclaiming Children and Youth, 17, 17-21.

Bøe

Sivertsen

Heiervang

Goodman

Lundervold

A. J.

Hysing

(2014). Socioeconomic status and child mental health: The role of parental emotional well-being and parenting practices. Journal of Abnormal Child Psychology, 42, 705-715.

Briere

Godbout

Dias

(2015). Cumulative trauma, hyperarousal, and suicidality in the general population: A path analysis. Journal of Trauma & Dissociation, 16, 153-169.

Coyne

J. J.

Barrett

P. M.

Duffy

A. L.

(2000). Threat vigilance in child witnesses of domestic violence: A pilot study utilizing the ambiguous situations paradigm. Journal of Child and Family Studies, 9, 377-388.

Guarino

K. M.

(2014). Trauma-informed care for families experiencing homelessness. In Haskett

Perlman

Cowan

(Eds.), Supporting families experiencing homelessness (pp. 121-143). New York, NY: Springer.

10.

Harrow

Jobe

T. H.

Faull

R. N.

(2014). Does treatment of schizophrenia with antipsychotic medications eliminate or reduce psychosis? A 20-year multi-follow-up study. Psychological Medicine, 44, 3007-3016.

11.

Hodgins

Hiscoke

U. L.

Freese

(2003). The antecedents of aggressive behavior among men with schizophrenia: A prospective investigation of patients in community treatment. Behavioral Sciences & the Law, 21, 523-546.

12.

Imai

Hayashi

Shiina

Sakikawa

Igarashi

(2014). Factors associated with violence among Japanese patients with schizophrenia prior to psychiatric emergency hospitalization: A case-controlled study. Schizophrenia Research, 160, 27-32.

13.

Jobe

T. H.

Harrow

(2010). Schizophrenia course, long-term outcome, recovery, and prognosis. Current Directions in Psychological Science, 19, 220-225.

14.

Klomek

A. B.

Sourander

Elonheimo

(2015). Bullying by peers in childhood and effects on psychopathology, suicidality, and criminality in adulthood. The Lancet Psychiatry, 2, 930-941.

15.

Leung

C. N. W.

Cotier

F. A.

Mark

Chen

van Os

Toulopoulou

(2014). The roles of victimization experiences, paranoia and salience misattribution in predicting psychosis proneness: The Twinsscan China Study. Schizophrenia Research. doi: 10.1016/S0920-9964(14)70957-0

16.

Link

B. G.

Stueve

(1994). Psychotic symptoms and the violent/illegal behavior of mental patients compared to community controls. In Monahan

Steadman

H.J.

(Eds.), Violence and mental disorder: Developments in risk assessment (pp. 137-159). Chicago, IL: University of Chicago Press.

17.

Link

B. G.

Stueve

Phelan

(1998). Psychotic symptoms and violent behaviors: Probing the components of “threat/control-override” symptoms. Social Psychiatry & Psychiatric Epidemiology, 33(13), S55-S60.

18.

Moreira

F. G.

Quintana

M. I.

Ribeiro

Bressan

R. A.

Mello

M. F.

Mari

J. J.

Andreoli

S. B.

(2015). Revictimization of violence suffered by those diagnosed with alcohol dependence in the general population. Biomed Research International, 2015(2015), Article 805424.

19.

Mulvey

E. P.

(2000-2010). Research on pathways to desistance [Maricopa County, AZ and Philadelphia County, PA]: Subject measures, 2000-2010 (ICPSR29961-v2). Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor]. doi:10.3886/ICPSR29961.v2

20.

Muskett

(2014). Trauma-informed care in inpatient mental health settings: A review of the literature. International journal of mental health nursing, 23, 51-59.

21.

Nagin

(2005). Group-based modeling of development. Cambridge MA: Harvard University Press.

22.

Nagin

Odgers

C. L.

(2010). Group-based trajectory modeling in clinical research. Annual Review of Clinical Psychology, 6, 109-138.

23.

Nederlof

A. F.

Muris

Hovens

J. E.

(2011). Threat/control-override symptoms and emotional reactions to positive symptoms as correlates of aggressive behavior in psychotic patients. The Journal of Nervous and Mental Disease, 199, 342-347.

24.

Ruback

R. B.

Clark

V. A.

Warner

(2014). Why are crime victims at risk of being victimized again? Substance use, depression, and offending as mediators of the victimization-revictimization link. Journal of Interpersonal Violence, 29, 157-185.

25.

Sadeh

Binder

R. L.

McNiel

D. E.

(2014). Recent victimization increases risk for violence in justice-involved persons with mental illness. Law and Human Behavior, 38, 119-125.

26.

Selner-O’Hagan

M. B.

Kindlon

D. J.

Buka

S. L.

Raudenbush

S. W.

Earls

F. J.

(1998). Assessing exposure to violence in urban youth. Journal of Child Psychology and Psychiatry, 39, 215-224.

27.

Silberg

J. L.

Copeland

Linker

Moore

A. A.

Roberson-Nay

York

T. P.

(2016). Psychiatric outcomes of bullying victimization: A study of discordant monozygotic twins. Psychological Medicine, 46, 1875-1883.

28.

Swanson

J. W.

Borum

Swartz

M. S.

Monahan

(1996). Psychotic symptoms and disorders and the risk of violent behaviour in the community. Criminal Behaviour and Mental Health, 6, 309-329.

29.

Teasdale

Silver

Monahan

(2006). Gender, threat/control-override delusions and violence. Law and Human Behavior, 30, 649-658.

30.

Tsai

A. C.

Weiser

S. D.

Dilworth

S. E.

Shumway

Riley

E. D.

(2015). Violent victimization, mental health, and service utilization outcomes in a cohort of homeless and unstably housed women living with or at risk of becoming infected with HIV. American Journal of Epidemiology, 181, 817-826.

31.

Turner

H. A.

Finkelhor

Ormrod

(2006). The effect of lifetime victimization on the mental health of children and adolescents. Social Science & Medicine, 62, 13-27.

32.

Underwood

L. A.

Washington

(2016). Mental illness and juvenile offenders. International Journal of Environmental Research and Public Health, 13, Article 228.

33.

Walker

H. E.

Freud

J. S.

Ellis

R. A.

Fraine

S. M.

Wilson

L. C.

(2019). The prevalence of sexual revictimization: A meta-analytic review. Trauma, Violence, & Abuse, 20(1), 67-80.

34.

Wilcox

J. A.

Duffy

P. R.

(2016). “They’re out to get me!”: Evaluating rational fears and bizarre delusions in paranoia: Paranoid delusions can reveal a number of psychiatric disorders; building trust is key to treatment. Current Psychiatry, 15(10), 28-33.

35.

Young

S. M.

Pruett

J. A.

Colvin

M. L.

(2018). Comparing help-seeking behavior of male and female survivors of sexual assault: A content analysis of a hotline. Sexual Abuse: A Journal of Research and Treatment, 30(4), 454-474.