Originally published by: Wiley Online Library
Written by: Anissa Mike, Hannah King, Thomas F. Oltmanns, Joshua J. Jackson

Objective:

Obsessive-compulsive personality disorder (OCPD) is defined as being overly controlling, rigid, orderly, and perfectionistic. At a definitional level, OCPD would appear to be highly related to the trait of Conscientiousness. The current study attempts to disentangle this relationship by examining the relationship at a facet level using multiple forms of OCPD assessment and using multiple reports of OCPD and personality. In addition, the relationship between OCPD and each Big Five trait was examined.

Method:

The study relied on a sample of 1,630 adults who completed self-reports of personality and OCPD. Informants and interviewers also completed reports on the targets. Bifactor models were constructed in order to disentangle variance attributable to each facet and its general factors.

Results:

Across four sets of analyses, individuals who scored higher on OCPD tended to be more orderly and achievement striving, and more set in their ways, but less generally conscientious. OCPD was also related to select facets under each Big Five trait. Notably, findings indicated that OCPD has a strong interpersonal component and that OCPD tendencies may interfere with one’s relationships with others.

Conclusions:

Findings suggest that OCPD’s relationship with personality can be more precisely explained through its relationships with specific tendencies rather than general, higher-order traits.

1 | INTRODUCTION

There has been a long and sustained effort by personality disorder researchers to convert to a dimensional model of personality in order to conceptualize personality disorders (Morey et al., 2011). Although the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association, 2013) did not officially change the classification system for personality disorders from a categorical diagnosis, an alternative model was listed in Section III of the manual. This alternative model is often described as a hybrid model and includes categories of personality disorders as well as a dimensional perspective of personality traits and disorders (Yam & Simms, 2014). When conceptualizing personality disorders as extreme variants of personality traits, obsessive-compulsive personality disorder (OCPD) is closely aligned with the trait of Conscientiousness (Samuels & Costa, 2012).

Obsessive-compulsive personality disorder is characterized by preoccupation with detail, high levels of order and organization, perfectionism, workaholism, rigidity, and a reluctance to delegate (American Psychiatric Association, 2013). Conscientiousness describes individuals who are industrious, reliable, self-disciplined, and ordered (Jackson & Roberts, in press; Mike, Harris, Roberts, & Jackson, 2015). Due to the level of definitional overlap, high levels of Conscientiousness theoretically should appear in those individuals who meet criteria for OCPD. In particular, one would expect certain aspects of Conscientiousness, such as high levels of order and achievement striving, to relate highly to OCPD drives and tendencies. While the relationship between OCPD and Conscientiousness has been examined in the past, the strength and the direction of the relationship have varied depending on several factors, including the type of OCPD measure used, the type of personality measure used, the broadness of the Conscientiousness assessment, and whether OCPD was assessed by self-report or interview (Samuels & Costa, 2012; Saulsman & Page, 2004). The goal of the current article is to further test the nature of this relationship across two measures of OCPD, examining the relationship between OCPD and personality at the facet level utilizing self-, informant, and interviewer reports.

1.1 | The relationship between OCPD and Conscientiousness

Despite similarities in the definitions of OCPD and Conscientiousness, the relationship between the two constructs has been inconsistent across studies (Samuels & Costa, 2012). One factor that influences the magnitude of the relationship between OCPD and Conscientiousness is the source of information, that is, interview or self-report (Saulsman & Page, 2004). For example, a self-report measure of OCPD showed a large positive correlation with Conscientiousness, whereas interview-based measures of OCPD showed a nonsignificant relationship with Conscientiousness (Saulsman & Page, 2004). While the self has a greater range of access to one’s behaviors across situations (Vazire & Carlson, 2010, 2011), interviewers are trained to identify symptoms of clinical disorders (Oltmanns, Rodrigues, Weinstein, & Gleason, 2014). However, it is unclear how such reports converge or diverge, or whose reports are more reliable. These findings highlight the importance of including different information sources in the assessment of OCPD and personality traits.

Discrepant findings may also be due largely to OCPD and Conscientiousness being multifaceted constructs. OCPD is thought to unite fairly heterogeneous traits, such as perfectionism and difficulty expressing warmth or affection with others (Samuel, Riddell, Lynam, Miller, & Widiger, 2012). OCPD has been found to have low internal consistency (Grilo et al., 2001), and past studies have indicated that OCPD may have at least two distinct underlying drives—perfectionism and rigidity (Ansell et al., 2010; Grilo, 2004).

Since only four out of eight criteria are necessary for an OCPD diagnosis, it is possible that two individuals who are both diagnosed with OCPD may demonstrate symptoms entirely discrepant from one another (Samuel et al., 2012).

Similarly, Conscientiousness is a general, overarching construct that can be broken up into more specific tendencies. Conscientiousness is typically thought to subsume lower-order traits like industriousness, responsibility, order and organization, and self-control (Jackson et al., 2010; Mike et al., 2015). Individuals who are generally conscientious may be high on one facet of Conscientiousness, but low on another. For example, an individual could be very hardworking, but not particularly well organized or orderly (Jackson et al., 2010).

Given that OCPD and Conscientiousness are both multifaceted constructs, it is possible that certain drives underlying OCPD relate to specific aspects of Conscientiousness more strongly than others. Thus, past findings on the relationship between OCPD and Conscientiousness may vary due to different measures of OCPD and Conscientiousness differentially tapping or emphasizing certain aspects of each construct (Samuel & Widiger, 2010, 2011). For example, in a study comparing eight scales of OCPD, which differed in their conceptualizations of the disorder and level of fidelity to the current DSM-5 (also the same criteria in DSM-IV-TR), the magnitude of the relationship between OCPD and personality traits was highly dependent on the OCPD measure used (Samuel & Widiger, 2010). A similar study examined the variability of Conscientiousness measures (Samuel & Widiger, 2011): A comparison of multiple scales of Conscientiousness, including the Revised NEO Personality Inventory (NEO-PI-R, Costa & McCrae, 1992), HEXACO Personality Inventory–Revised (HEXACO PI-R, Ashton & Lee, 2010), and Multidimensional Personality Questionnaire (MPQ, Tellegen, 1982), found that the magnitude of the relationship between Conscientiousness and OCPD varied by the measure of Conscientiousness. In particular, the NEO-PI-R shared the smallest relationship with OCPD of all the measures of Conscientiousness (Samuel & Widiger, 2011).

One way to bypass limitations of disparate Conscientiousness measures is to look at more specific drives underlying Conscientiousness, that is, facets. Indeed, OCPD demonstrates clearer relationships with the specific facets of Conscientiousness than with the broader domain of Conscientiousness itself. A meta-analysis of findings on personality disorders and personality traits indicated that OCPD was positively related to all Conscientiousness facets (Samuel & Widiger, 2008). However, the size of the relationship varied across Conscientiousness facets. For example, the facets of order, dutifulness, achievement striving, and deliberation demonstrated positive relationships with OCPD across types of personality and OPCD measures, whereas the facets of competence and self-discipline shared positive, null, and even occasionally negative relationships with OCPD depending on the type of measures used (Samuel & Widiger, 2008). In general, these findings suggest a complex relationship between OCPD that necessitates looking at facets of Conscientiousness.

1.2 | Bifactor models for facet-level analyses

One obstacle when examining facets is that they are strongly related to one another. Previous studies investigating the importance of facets have often examined facets separately from their overall construct so as to limit this collinearity (Chen, Hayes, Carver, Laurenceau, & Zhang, 2012). However, a limitation of this procedure is that the examined facet still shares variance with the broad trait (e.g., Conscientiousness) as well as other facets. This results in difficulties of interpreting which facets are most important because some facets may be more closely aligned with the broad trait. For example, the facet of achievement striving might be strongly related to OCPD because it is a strong marker of Conscientiousness, not because behaviors, thoughts, and feelings associated with achievement striving are particularly related to OCPD. Thus, to better understand the association between OCPD and Conscientiousness, it would be beneficial to examine facet-specific variance, separated from overall trait variance.

Bifactor models provide a rigorous method to test facets independently from both their general trait and other facets by separating variance into common variance, shared across all facets and unique variance specific for each facet (Chen et al., 2012). To do so, several latent variables are constructed: The variable associated with the general factor captures what all items have in common, whereas the variables associated with the facets capture what is unique to each facet, after accounting for the general factor (see Figure 1). As a result, bifactor models are able to distinguish between the common variance associated with a general factor and the unique variance associated with more specific facets. Studies utilizing the bifactor model to study personality facets find evidence that facets are capable of predicting outcomes beyond their general factor, that facets meaningfully predict even when the general factor does not, and, importantly, that facets can predict in opposite directions of one another and their general factor (Chen et al., 2012; Jackson & Weston, 2015; McAbee, Oswald, & Connelly, 2014). We utilize bifactor models in the current study in order to examine whether Conscientiousness facets predict OCPD traits differentially from one another and whether they may be a better level of analysis compared to measuring Conscientiousness at a broad level.

1.3 | The current study

Overall, past studies indicate that the relationship between Conscientiousness and obsessive-compulsive personality disorder is more complex than originally expected based on conceptual arguments. Given the broad nature of the Big Five traits, we hypothesized that stronger predictions of out- comes can be gained by utilizing lower-order facets that better identify one’s inclination toward more specific behaviors. Existing research has demonstrated that facets offer important information regarding behaviors and outcomes (Ashton, 1998; Kern et al., 2013; Paunonen, 1998; Paunonen & Ashton, 2001; Paunonen, Haddock, Forsterling, & Keinonen, 2003), and thus an analysis of facets will clarify the association between Conscientiousness and OCPD. Furthermore, bifactor models will provide a more rigorous method to extend previous facet-level research that highlighted the utility of moving beyond broad measures of Conscientiousness (Samuel & Widiger, 2008, 2010, 2011). We expected that the facets of order, dutifulness, achievement striving, and deliberation would be positively related to OCPD, whereas the facets of competence and self-discipline would have a weaker or no relationship with OCPD.

In addition to examining the relationship between facets of Conscientiousness and OCPD, we also analyzed facets of the other Big Five traits. Many personality disorder researchers have argued that in order to fully capture a dimensional model of a personality disorder, it is necessary to include representations of all traits (Samuels & Costa, 2012). Although the current definition of OCPD in the DSM-5 focuses on criteria related to Conscientiousness, previous conceptualizations of OCPD have also included interpersonal symptoms like restricted affect (Samuel & Widiger, 2010). Conceptually, on the level of the Big Five, we expected to find that OCPD is related to high Neuroticism, low Openness, low Agreeableness, and low Extraversion (Samuels & Costa, 2012). In terms of facets, experts on personality disorders have indicated, and initial research suggests, that individuals with OCPD are high on the Neuroticism facet of anxiousness, low on facets of Extraversion like excitement seeking, and low on facets of Openness such as actions and values (Lynam & Widiger, 2001; Samuel & Widiger, 2004, 2008).

Given that a substantial amount of previous research has highlighted the heterogeneity of current OCPD conceptualizations (Samuel et al., 2012), including the findings mentioned above that measures of OCPD only have moderate convergent validity (Samuel & Widiger, 2010), we also explored the relationship between personality and individual OCPD criteria. We expected to find that specific diagnostic criteria are more closely related to Conscientiousness than others. Specifically, perfectionism, workaholism, and preoccupation with details should be more closely related to facets of Conscientiousness (Samuel & Widiger, 2011), whereas rigidity and stubbornness should be associated more directly with low levels of Agreeableness and Openness (Samuels & Costa, 2012).

Finally, the current study also intends to address the discrepant findings due to different types of OCPD measurement and explore the role of the source of the information. Relying solely on self-reports can offer a restricted, or even misleading, view of personality, as self-reports may suffer from bias or blind spots in self-knowledge (Robins & John, 1997; Vazire & Carlson, 2010, 2011). A more complete picture of personality can be obtained by utilizing a variety of reports on personality. Thus, the current study examined self-reports of personality as well as reports from a close informant. Furthermore, multiple sources of information on OCPD symptoms were also obtained. Symptoms of OCPD were described by the self, an informant, and a trained diagnostic interviewer. Clinically assessing OCPD offers a more conservative, objective assessment than either an individual or informant may report, and examining interviewer-assessed OCPD and informant-reported personality allowed us to eliminate any biases individuals may have when reporting on themselves. Furthermore, using interviewer-assessed OCPD with self-reported personality allows for the elimination of measurement overlap between constructs. Thus, the current article is based on four sets of analyses: self-reported personality with self-reported OCPD, informant-reported personality with informant-reported OCPD, self-reported personality with interviewer-assessed OCPD, and informant reported personality with interviewer-assessed OCPD.

2 | METHOD

The current study utilized data from the St. Louis Personality and Aging Network (SPAN) study, which examines personality and clinical disorders in a community sample of later middle-aged adults (Oltmanns et al., 2014). The SPAN study recruited individuals between the ages of 55 and 64 ((Mₐ₉ₑ 5 59.53, SD5 2.70 years) and gathered a variety of information, including self- and informant reports on personality and self- and informant reports of personality disorder symptomatology from all participants.

2.1 | Participants

Participants were 1,630 individuals (55% female) living in the St. Louis area. The majority of participants received a high school diploma or higher (3% completed less than high school, 29% received a high school diploma or equivalent, 16% completed a 2-year college degree, 26% had a bachelor’s degree, and 26% had an advanced degree). Sixty-five percent of participants were Caucasian, 33% were African American, and 2% of participants reported being another ethnicity. Approximately 8% were diagnosed with at least one personality disorder; 2.9% qualified for a DSM-5 diagnosis of OCPD based on the Structured Interview for DSM-IV Personality.

2.2 | Informants

Participants nominated an informant who knew them well to provide additional information on their personality. Informant reports were collected for 91% of the participants. Approximately 50% of the informants were a spouse or partner, 25% were an adult child or other family member, and the remaining informants were close friends.

2.3 | Measures

2.3.1 | Personality

The Big Five traits (Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness to Experience) and their corresponding facets were assessed using the NEO-PI-R (Costa & McCrae, 1992). The NEO-PI-R includes 240 items and measures six facets for each Big Five trait: Scores for each facet are based on eight items, and each trait is an average of its total 48 items. Items were rated on a 5-point Likert scale ranging from 0 (strongly disagree) to 4 (strongly agree). Informants were asked to complete the full NEO-PI-R as it pertained to the participant. Self- and informant reports agreed moderately. Correlations ranged from .26 to .56 (average r = .41), suggesting self- and informant reports agreed in important ways but that each also provided unique information. The facets of the NEO-PI-R as well as personality means, standard deviations, reliabilities, and correlations between self- and informant reports can be found in Table 1.

2.3.2 | OCPD

OCPD was measured via self- and informant reports using the Multisource Assessment of Personality Pathology (MAPP; Oltmanns, Turkheimer, & Strauss, 1998) inventory and by interviewer assessment using the Structured Interview for DSM-IV Personality (SIDP). The MAPP measure of OCPD consists of eight items, each corresponding to a diagnostic feature of OCPD as listed in the DSM-IV. Items were created by translating the DSM-IV criteria into language that would be easily understood by the average person, eliminating jargon and psychopathological terms. Example items include“I am a perfectionist and my perfectionism gets in the way of getting things done” and“I need to do everything myself because no one else will do them right.” Items were rated by participants on a scale ranging from 0 (I am never like this) to 4 (I am always like this). Informants completed the items as they pertained to the participant. For the purposes of this article, an additional item was included in the construction of the OCPD variable “I become upset (irritated, angry, or anxious) at deviations from my routine.” Though not part of the original MAPP-OCPD measure, the content of the item fit well conceptually with the DSM-IV definition of OCPD and correlated highly with the other items from the MAPP. Cronbach’s alpha with the additional item was .69, compared to .65 with only the original items. Analyses run with and without the additional item demonstrated an equivalent pattern of results. Cronbach’s alpha for the informant version of the MAPP was .67 with the additional item and .61 without; thus, the additional item was also retained for informant analyses.

OCPD was also assessed using the SIDP. Trained interviewers questioned participants in a face-to-face, semistructured interview and evaluated them on eight criteria for OCPD. Responses were rated on a scale ranging from 0 (not present) to 3 (strongly present). Cronbach’s alpha for the SIDP-OCPD score was .61. The correlation between the SIDP and the MAPP was .35 for the self-report MAPP and .19 for the informant report MAPP, indicating the SIDP accesses information unique from the MAPP. The SIDP was found to be a more conservative measure of OCPD than either the self- or informant-rated MAPP (Oltmanns et al., 2014). Prevalence rates for the self-rated MAPP, informantrated MAPP, and SIDP are listed in Table 2.

2.4 | Analyses

In the current analyses, we examine the relationship between OCPD and the Big Five personality traits across self-reports of personality and self reports of OCPD, informant reports of personality and informant reports OCPD, self-reports of personality and interviewer reports of OCPD, and informant reports of personality and interviewer reports of OCPD. First, initial zero-order correlations were run for each Big Five trait and its underlying facets and the respective measure of OCPD. Analyses were done in R utilizing the psych package (Revelle, 2013). Then, in order to ensure that each trait and its facets were best represented, and in order to obtain the most accurate estimates for the relationship between OCPD and personality, bifactor models were constructed for each Big Five trait successively. Before constructing the bifactor models, personality items were parceled in groups of two, pairing a positively worded item with a negatively worded item where possible. Parceling reduces the number of parameters that have to be estimated (thus reducing the complexity of the model), increases reliability, and increases the likelihood that the error variances of the resulting parcels will be normally distributed (Hau & Marsh, 2004). Each Big Five latent factor was constructed using all 24 parcels associated with the trait, and the four appropriate parcels were allowed to load on their specific facet simultaneously. The latent factor and facets were constrained to be uncorrelated with one another so that the facets related to one another only through the general trait. A simplified, visual representation of the model can be seen in Figure 1.

Because there was concern that the bifactor facets may capture residual variance that is discrepant from the original, intended meaning of the facets, we conducted two sets of additional analyses. In the first set of analyses, we correlated the self-reported latent bifactor facets with the self-reported manifest facets to ensure that what was left in the facets after the domain variance was removed still corresponded with the original meaning of the facets and the overall domain. In the second set of analyses, we compared the self-reported bifactor facets with the corresponding informant-reported facets constructed latently, but not in a bifactor framework. For example, order from the self-reported bifactor model was correlated with the informant-reported, latently modeled order (where the overall domain and other facets were excluded from the model). Because the informant reported facets still contained domain information, this was another way to check whether the bifactor facets still captured essential information in line with the original constructs.

OCPD was examined in two different ways: as a composite and at the level of individual symptoms. First, OCPD was constructed latently in three separate models using all nine MAPP items (for both the self and the informant) or all eight SIDP items (for the interviewer). This resulted in four separate analyses for each trait: self-reported personality with self-reported OCPD, informant-reported personality with informant-reported OCPD, self-reported personality with interviewer-assessed OCPD, and informant-reported personality with interviewer assessed OCPD. In each model, OCPD was correlated with each Big Five trait and its facets simultaneously. Running the initial measurement models for personality revealed that some facets did not have significant variance after accounting for their general trait (consistent with previous bifactor models with the Big Five; e.g., Chen et al., 2012). The facets were retained in the model, but their relationships with OCPD were constrained to zero.

Second, in order to better examine the multidimensional nature of OCPD and disentangle whether certain aspects of OCPD differentially related to personality, analyses were conducted at the level of individual OCPD symptoms. We continued to model personality in a bifactor framework, but instead of correlating all OCPD items with all possible traits and facets, only the traits and facets that were found to be strongly related to OCPD in the composite analyses were selected. Thus, specific diagnostic symptoms of OCPD from self, informant, and interviewer were correlated against self- and informant reports of several personality traits.

Bifactor analyses were run in Mplus software 7.0 (Muthen & Muthen, 1998–2009) using maximum likelihood estimation with robust standard errors. Four goodness-of-fit indices were considered for the bifactor model: chi-square; RMSEA, which has a recommended cut-off point of .08; SRMR, which has a recommended cut-off of .08; and CFI, which has a recommended minimum of .95 but is highly sensitive to the number of indicators per factor, and a more liberal cut-off of .90 is sometimes used (Chen et al., 2012).

3 | RESULTS

3.1 | Zero-order correlations

We first examined the relationship between OCPD and personality traits and facets by correlating the constructs using traditional composites of each. This was so that we could compare results at the zero-order level to the results using the bifactor and latent traits.

3.1.1 | OCPD and Conscientiousness

Across self-, informant, and informant/interviewer reports, there was no evidence for a relationship between OCPD and Conscientiousness at the general trait level (rs range from–.05 to .02; see Table 3). Meanwhile, in the self- and interviewer reports, Conscientiousness was slightly negatively related to OCPD (r = –.06).

3.1.2 | OCPD and the remaining Big Five traits

Across all reports, OCPD was positively related to Neuroticism (rs range from .10 to .34) and all Neuroticism facets (rs range from .05 to .38).

Similarly, OCPD was negatively related to Agreeableness (rs range from–.34 to–.09) and all Agreeableness facets (rs range from–.31 to–.05), with the exception of modesty and straightforwardness, whose estimates were all negative, but not significantly different from zero in select cases (straightforwardness rs range from–.20 to–.04; modesty rs range from–.20 to–.01).

3.2 | Bifactor models

Our next set of analyses examined the relationship between OCPD and personality traits and facets modeling OCPD as a latent construct and modeling personality using a bifactor model. This allowed for a more rigorous test of how specific variance attributable to the facets and general factor related to OCPD. Fit indices suggested adequate fit for the initial bifactor models for each Big Five trait. For self-reported personality, RMSEA values ranged from .04 to .05, and SRMR values ranged from .03 to .05; CFI values were .90 or higher, with the exception of Agreeableness, which had a CFI value of .88. For informant-reported personality, RMSEA values ranged from .05 to .06, SRMR values ranged from .04 to .06, and CFI values were above .90, with the exception of Extraversion, which had a CFI value of .88. Fit indices are displayed in Table 8.

In the initial models, a small number of facets did not demonstrate significant variance, suggesting that, after accounting for the general factor, there were no individual differences remaining in the facets. For Conscientiousness, these included self-discipline in self-reports and competence when using informant reports. Furthermore, depression (Neuroticism) failed to emerge as a facet when using self-reports, whereas aesthetics (Openness) failed to emerge for informant reports. Finally, altruism and tender-mindedness under Agreeableness did not emerge for either the self- or informant reports.

3.2.1 | Validating the bifactor facets

When comparing the facets constructed as manifest variables to the facets constructed in a latent, bifactor framework, we found the correlations to be quite high, ranging from .65 to .93, with an average correlation of .82. The average correlation between facets was .82 for Extraversion, .89 fo Agreeableness, .77 for Conscientiousness, .74 for Neuroticism, and .81 for openness. For a full list of correlations between facets, refer to Table 9. When examining the relationship between the self-reported bifactor facets and latent informant-reported facets, we again found significant relationships between all facet pairs, with the correlations ranging from .29 to .75. The average correlation of all facet pairs was .48, whereas the average correlation between facets was .59 for Extraversion, .38 for Agreeableness, .46 for Conscientiousness, .40 for Neuroticism, and .52 for Openness. The correlations compared favorably to the zero-order correlations between self- and informant-reported facets—in many cases, the correlations between the latent constructs were greater than the zero-order relationship. Please refer to Table 10 to see all correlations between facets. Together, these results bolstered our confidence that the bifactor facets adequately captured unique variance associated with each facet and retained the intended meaning of the facets.

3.2.2 | OCPD and Conscientiousness

3.2.3 | OCPD and the remaining Big Five traits

OCPD shared a robust, positive relationship with Neuroticism across reporting methods (std. bs range from .11 to .52). Bifactor Neuroticism results are displayed in Table 4. The facet of angry hostility was also positively related to OCPD across reports (std. bs range from .11 to .45), with the exception of informant–interviewer reports (std. b = .05). However, unlike the zero-order correlations, the relationships between OCPD and the remaining Neuroticism facets were insubstantial: OCPD was not consistently related to anxiety (std. bs range from .01 to .18), depression (std. bs range from–.01 to .01), self-consciousness (std. bs range from .01 to .12), impulsiveness (std. bs range from–.05 to .05), or vulnerability (std. bs range from–.06 to .09).

Similar to Neuroticism, relationships with Agreeableness became more specific in the bifactor framework: OCPD had a large, negative relationship with Agreeableness across reports (std. bs range from–.64 to–.09) and was also negatively related to trust (std. bs range from–.17 to–.09), with the exception of self-/interviewer reports (std. bs = –.02). OCPD was no longer found to be related to the remaining Agreeableness facets of straightforwardness (std. bs range from .03 to .09), compliance (std. bs range from–.13 to .03), or modesty (std. bs range from .02 to .09). Bifactor results for Agreeableness are displayed in Table 5.

3.3 | Analyses relating specific OCPD symptoms to personality traits

Our final set of analyses examined the relationship between personality and specific OCPD symptoms, which allowed us to more clearly understand which aspects of OCPD may be driving the relationships with specific personality traits and facets. To limit the number of tests in these exploratory analyses, we examined only the traits and facets that most highly related to OCPD: achievement striving, actions, Agreeableness, Neuroticism, gregariousness, order, and angry hostility. We correlated each trait with the individual symptoms of OCPD (nine items from each of the self- and informant reports of the MAPP and eight OCPD items from the interviewer SIDP reports). The results of these analyses can be seen in Table 11. Associations were mostly consistent across personality traits, though some symptoms demonstrated opposing relationships to facets under the same trait, indicating the need to look at OCPD in more precise terms.

The diagnostic feature of being rigid and stubborn was related to several personality traits, including lower levels of openness to actions (rs range from–.15 to–.25) and Agreeableness (rs range from–.18 to–.53) and higher levels of angry hostility (rs range from .13 to .41). Furthermore, rigidity was related to Neuroticism when using self- and informant reports of OCPD (rs = .15 and .22, respectively) and order when using interviewer reports of OCPD (rs = .14 and .07). Similarly, becoming upset at deviations from normal routine, which was only measured in the MAPP, was related to multiple aspects of personality, including being more orderly (rs = .16 and .21), more neurotic (rs = .29 and .37), more angrily hostile (rs = .23 and .25), and less agreeable (rs = –.16 and–.14) and lower on openness to actions (rs = –.28 and–.40).

4 DISCUSSION

Past findings regarding the relationships between OCPD, the broad domain of Conscientiousness, and specific Conscientiousness facets have been mixed. The current study attempted to clarify these relationships by examining the facet-level relationships between OCPD and personality Relationships were examined using zero-order and bifactor models, where bifactor models were used to differentiate common variance attributable to a general trait and unique variance attributable to each facet. Using the bifactor models, we found a more complex relationship between OCPD and Conscientiousness than was expected. Specifically, OCPD was negatively related to general Conscientiousness and only related to certain specific facets under Conscientiousness. Results largely replicated across measure and type of reporter, suggesting that the relationship between OCPD and personality is perhaps clearer and more observable than might be suggested by the previous literature on this topic.

4.1 | OCPD and Conscientiousness

Results across zero-order correlations and bifactor models presented somewhat differing patterns of association. Contrary to our expectations, OCPD and the broad trait of Conscientiousness were not related at the zero-order level, and, surprisingly, general Conscientiousness and OCPD were consistently negatively related in the bifactor analyses. Conscientiousness, at its core, refers to how self-controlled and competent individuals are highly conscientious individuals are good at handling situations and getting things done (Jackson & Roberts, in press; Mike et al., 2015). Yet, individuals with OCPD have perfectionistic tendencies and excessive need for order and regulation. While these drives align with Conscientiousness, they likely have the potential to interfere with other behaviors typically associated with Conscientiousness—chiefly one’s overall proficiency and ability to accomplish tasks and goals (Jackson et al., 2010). In other words, aspects of OCPD may hinder the ability to perform tasks and behaviors central to the definition of Conscientiousness.

Bifactor models allowed us to disentangle these contradictory aspects of OCPD and Conscientiousness at a facet level. Similar to the general trait, self discipline and competence had negative or null associations with OCPD. These facets may be closely associated with the common variance associated with the broad trait, making it difficult to determine the relationship between OCPD and these facets distinctly from general Conscientiousness. Indeed, in the bifactor models, competence and self-discipline were not distinguishable from general Conscientiousness by the self and informant, respectively. In contrast, the facets of order, dutifulness, and achievement striving demonstrated robust, positive relationships with OCPD. The opposing directions of the associations between the broad trait and some specific facets are similar to the findings of previous studies employing the bifactor model. For example, when predicting health and adaptability, the broad trait of Conscientiousness evidenced a positive relationship, but the facet of perfectionism was negatively associated with both health and adaptability (McAbee et al., 2014), suggesting that while general Conscientiousness leads to more positive health behaviors, being overly meticulous may have adverse consequences and may act counter to more general conscientious tendencies and behaviors.

4.2 | OCPD and the remaining Big Five traits

In the zero-order correlations, OCPD was related to Neuroticism and all Neuroticism facets (positively) and Agreeableness and all Agreeableness facets (negatively). OCPD retained its relationships with the general traits of Neuroticism and Agreeableness in the bifactor models, but it was only related to the facets of angry hostility and trust, respectively. The zero-order results likely indicate uniform relationships across facets because the zero-order versions of the facets overlap with their general traits, and OCPD may simply be relating to the general trait that is captured in each facet. These results are consistent with past research that finds most personality disorders are associated with high levels of trait Neuroticism, suggesting that there is a broad association between personality pathology and emotional distress captured by Neuroticism (Saulsman & Page, 2004). While one may expect OCPD to be more broadly related to Neuroticism and its facets, above and beyond general Neuroticism, individuals scoring highly on OCPD may not have more specific dispositions toward anxiety, depression, self-consciousness, impulsiveness, or vulnerability. Beyond general Neuroticism, OCPD was related to the facet of angry hostility, indicating an above-average impatience for others and intolerance and hostility toward those who disrupt order.

Similar to the findings on personality disorders and Neuroticism, most personality disorders are associated with low levels of the broad trait of Agreeableness, which is indicative of the interpersonal difficulties inherent to personality disorders (Saulsman & Page, 2004). Consistent with this relationship, we found that individuals scoring highly on OCPD were more generally disagreeable. In addition, they reported having more trouble relying on and trusting others, though OCPD was not related consistently to any other Agreeableness facets.

Across zero-order and bifactor models, OCPD did not have a consistent relationship with the broad domain of Openness, but the facets of actions and values evidenced negative associations across reporting methods and type of analyses. The actions facet captures being set in one’s ways versus enjoying and trying new things and includes items such as“I prefer to spend my time in familiar surroundings” and“Once I find the right way to do something, I stick to it,” suggesting that the higher an individual is on OCPD criteria, the more he or she will dislike change and new experiences Openness to values is a reflection of the extent to which a person measures how conservative one’s beliefs and attitudes are. Moreover, similar to the facet of angry hostility, openness to values may capture an intolerance toward the values of others, including items like“I consider myself broad-minded and tolerant of other people’s lifestyles.” Thus, the relationship between OCPD and values may capture the low tolerance for other people’s lifestyles that can be reported by individuals with OCPD.

Finally, OCPD retained its negative relationship with Extraversion in the bifactor models and was also consistently negatively related to gregariousness and positive emotions, indicating that individuals who scored more highly on OCPD were less sociable and also experienced less positive emotions than individuals with fewer OCPD tendencies. Results relying on self-reports of personality also indicated a connection between OCPD and the facets of openness to activity and excitement seeking. Describing oneself as high on excitement seeking seems at odds with someone who would also describe oneself as rigid and stubborn. It is important to note this relationship was only found in analyses relying on self-reports of personality and OCPD—informants and interviewers did not see those high on OCPD as higher on excitement seeking. Thus, the relationship betwee OCPD and excitement seeking may be due to the way individuals answer items—perhaps those high on OCPD respond more strongly to strongly worded items. Alternatively, OCPD may be relating to a specific component of excitement seeking—such as energy or preference for keeping busy—rather than general excitement-seeking tendencies. Combined with the positive relationship between OCPD and self- reported activity, these findings suggest that individuals high on OCPD tendencies may see themselves as more energetic than their close others see them.

4.3 | Analyses relating specific OCPD symptoms to personality traits

Analyses conducted at the level of specific symptoms of OCPD helped further clarify the association between OCPD and personality traits. Being rigid and stubborn and becoming upset at deviations were both broadly related to multiple traits, suggesting each item captured multiple symptoms of maladaptive personality that have traditionally been used to define OCPD. Meanwhile, other symptoms were more narrowly related to only one or two aspects of the personality traits examined. Interestingly, OCPD items could show opposing relationships with the same personality traits. For instance, being preoccupied with details was positively related to order, whereas keeping old or worthless objects and being excessively devoted to work were both negatively related to order. Together, these results suggest that OCPD unites several specific symptoms or features of personality pathology and also highlight the multidimensionality of OCPD as a construct.

4.4 | The role of type of reporter

Past research indicates that while self- and informant report measures agree in important ways, each has unique perspectives and can provide important information about normal and maladaptive personality features (Jackson, Connolly, Garrison, Leveille, & Connolly, 2015; Oltmanns & Turkheimer, 2006; Robins & John, 1997; Vazire & Carlson, 2010). The current study examined reports across the self, a close other, and an interviewer in order to maximize available information despite differences in trait observability, potential blind spots, and potential biases across reporters. Across type of reporter, the results regarding OCPD and personality were fairly consistent; even the magnitudes of the effect sizes were similar across self- and informant reports of personality and OCPD. Such agreement across reports indicates that the relationship between OCPD tendencies and personality traits is observable regardless of type of reporter, and that there is little information, in this specific instance, that the self or the informants are unaware of.

4.5 | Bifactor models as a way of examining facet-level personality

It is important to keep in mind that the bifactor facets represent what is left over after their general factor has been taken into account—that is, what order might be beyond its relationship to Conscientiousness. The meaning of the facets after the general factor is removed could be discrepant from the original meaning of the facet. The current article attempted to validate the meaning of the facets in two ways: by correlating the self-reported bifactor facets with their self- reported, manifest counterparts and by correlating self- reported bifactor facets with their latently modeled, informant reported counterparts. The self-reported bifactor facets correlated strongly with the manifest facets, with correlations ranging from .65 to .93, demonstrating that, even with the domain information removed, the bifactor facets still captured information essential to the facets, and not just error. Similarly, analyses comparing the bifactor facets from the self-reports to the latent facets from the informant reports indicated that the two were related, and they often demonstrated stronger relationships than the zero-order ones, which again led us to believe we were indeed capturing meaningful variance in line with the original, intended meaning of the facets. Together, we believe that these analyses provide a strong case for using the bifactor model when examining personality at the facet level.

5 | LIMITATIONS AND FUTURE DIRECTIONS

However, there is still a possibility that the bifactor facets captured variance that is not in line with the original meaning of the facet. For instance, it is possible that the facets are capturing extreme variance that is left over after parceling out the general trait. For example, general Conscientiousness encompasses tendencies toward orderliness and organization; after accounting for general levels of Conscientiousness, unique variance in the order facet may then represent a preference for extreme order and cleanliness, above and beyond what one would expect given a person’s standing on general Conscientiousness. Some have speculated that inconsistent results between OCPD and Conscientiousness may be the result of personality measures failing to capture more extreme ends of personality (Haigler & Widiger, 2001). One study modified the NEO-PI-R in order to include more maladaptive items and found that the correlation between Conscientiousness and OCPD increased significantly (Haigler & Widiger, 2001). Thus, if the bifactor facets of the current model are capturing extreme variants of personality, this may help clarify the relationship between personality and pathology. More research is needed to determine whether bifactor models of personality facets are simply distinguishing facets from their general factor or whether the bifactor facets represent extreme variants of their general factor. Furthermore, future research may wish to further explore ways to capture maladaptive or extreme ends of personality.

It is also possible that there is not much variance to be accounted for beyond the general factor and that some facets represented in the bifactor models actually capture nonmeaningful variance. Indeed, the current bifactor models did not find support for several facets, indicating that either the facets did not have enough distinct items to define each trait sufficiently or that the facets were not conceptually distinct from their general trait. This raises the larger question of whether narrow facets are useful and informative, or whether the broader-bandwidth Big Five trait domains are sufficient to explain individual differences. Addressing the validity of facets may be difficult to address with existing scales of the Big Five traits, which may not adequately capture the entire range of facets for each trait (John, Naumann, & Soto, 2008; Roberts, Bogg, Walton, Chernyshenko, & Stark, 2004; Roberts, Chernyshenko, Stark, & Goldberg, 2005; Saucier & Ostendorf, 1999). Many measures do not agree on the number of facets underlying each trait, nor do they agree on which facets underlie which traits. In either case, our study indicates that facets deserve a more thorough examination, as they were related to OCPD above and beyond common trait variance. Researchers should investigate whether facets of the Big Five reliably predict beyond their general factor for a variety of outcomes. Furthermore, future studies may work to better identify and measure the facets of the Big Five.

Past research has indicated that findings regarding the relationship between OCPD and Conscientiousness may vary by type of personality measure used. Unfortunately, the current article was unable to address this issue by testing multiple measures of personality. Future research may wish to us the current methods to further test the relationship between OCPD and other measures of Conscientiousness. Examining other measures of Conscientiousness will also allow for an examination of other facets that are not included in the NEO-PI-R, which may provide elucidating information on how OCPD and Conscientiousness are related. For instance, the HEXACO measures perfectionism under Conscientiousness, which conceptually aligns highly with OCPD.

Finally, the current sample included only later middle-aged adults. It is worth noting that personality develops with age (Lucas & Donnellan, 2011; Roberts & Mroczek, 2008), and there is also evidence that different age groups may endorse OCPD criteria at different rates (Balsis, Woods, Gleason, & Oltmanns, 2007). Future studies may examine the function of age in the relationship between OCPD and personality.

6 | CONCLUSIONS

Overall, findings from the bifactor models present more specific relationships with OCPD than findings from zero-order correlations. Bifactor findings were largely consistent across reporter and, when taken together, paint the picture of an individual with high standards, but whose perfectionistic tendencies interfere with the ability to accomplish goals. Furthermore, individuals high on OCPD report experiencing more negative emotions, experiencing less positive emotions, and being more set in their ways and resistant to change. Finally, findings indicate that OCPD has a strong interpersonal component and that OCPD tendencies may interfere with one’s relationships with others. Individuals high on OCPD preferred to be alone, had trouble trusting others, and reported being intolerant of others’ lifestyles and choices. Interestingly, OCPD’s strongest relationships were not specific to just Conscientiousness or Conscientiousness facets: OCPD was most strongly related to achievement striving (Conscientiousness), actions (Openness), Agreeableness, and Neuroticism.

Personality disorders are still currently classified using a categorical diagnosis (American Psychiatric Association, 2013), yet there has been a push to convert to a dimensional model that will unite research on personality traits and personality disorders (Morey et al., 2011). Our present results are important not only to clarify previous contradictory findings, but also to provide hints as to how OCPD should be represented in a personality space. Currently, the proposal in Section III of the DSM-5 defines OCPD as being characterized by rigid perfectionism, perseveration, intimacy avoidance, and restricted affectivity (American Psychiatric Association, 2013). The results of this study are mostly consistent with this new proposal for OCPD, particularly in regard to rigidity (OCPD was related to preferring familiar routines and high levels of order), perfectionism (high achievement striving), and intimacy avoidance (low trust combined with low Extraversion and gregariousness). In addition, our findings may support perseveration as a component of OCPD. Perseveration is defined as“persistence. . .after the behavior has ceased to be functional or effective” (American Psychiatric Association, 2013, p. 779). Our results indicated that OCPD and Conscientiousness were negatively related, suggesting that individuals high on OCPD tendencies are ineffective at making decisions and accomplishing tasks.

Our results differ from the current definition of OCPD in that individuals higher on OCPD tendencies were also higher on several emotional traits; namely, they had higher levels of general Neuroticism (i.e., negative affect; Widiger, 2009) and higher levels of angry hostility, as opposed to having restricted affectivity as hypothesized in the DSM. However, our results indicated that OCPD was related to experiencing less positive emotions. Finally, OCPD shared a relationship with several interpersonal traits, namely, angry hostility and trust. Our results suggest that it may be useful to further examine the relationships between OCPD and emotionality and interpersonal traits in order to further inform a dimensional model of OCPD.

References available in original article on Wiley Online Library.