Abstract
The Faceted Inventory of the Five-Factor Model (FI-FFM) is a comprehensive hierarchical measure of personality. The FI-FFM was created across five phases of scale development. It includes five facets apiece for neuroticism, extraversion, and conscientiousness; four facets within agreeableness; and three facets for openness. We present reliability and validity data obtained from three samples. The FI-FFM scales are internally consistent and highly stable over 2 weeks (retest rs ranged from .64 to .82, median r = .77). They show strong convergent and discriminant validity vis-à-vis the NEO, the Big Five Inventory, and the Personality Inventory for DSM-5. Moreover, self-ratings on the scales show moderate to strong agreement with corresponding ratings made by informants (rs ranged from .26 to .66, median r = .42). Finally, in joint analyses with the NEO Personality Inventory–3, the FI-FFM neuroticism facet scales display significant incremental validity in predicting indicators of internalizing psychopathology.
Keywords
Over the past several decades, personality researchers have made considerable progress toward the development of a consensual, comprehensive taxonomy of traits. One key development that facilitated this progress was the explicit recognition that personality traits are ordered hierarchically at different levels of breadth (e.g., Digman, 1997; Hogan & Hogan, 1992; John, Hampson, & Goldberg, 1991; Markon, Krueger, & Watson, 2005; Shiner & Caspi, 2003; Watson, Clark, & Harkness, 1994). For example, the broad higher order dimension of extraversion can be decomposed into several distinct yet empirically correlated traits (see Watson & Clark, 1997; Watson, Stasik, Ellickson-Larew, & Stanton, 2015b), such as assertiveness (i.e., extraverts are dominant and persuasive, and enjoy being the center of attention), sociability (i.e., extraverts seek out and enjoy the company of others), positive emotionality (i.e., extraverts are cheerful, energetic, and enthusiastic), and experience seeking (i.e., extraverts enjoy intense and exciting experiences).
The development of hierarchical models permitted the integration of general factor models—such as the prominent Big Three and Big Five schemes—with the multidimensional structures that were the focus of most older personality inventories (Watson et al., 1994). In fact, many widely used personality instruments—including the Multidimensional Personality Questionnaire (Patrick, Curtin, & Tellegen, 2002), the Hogan Personality Inventory (Hogan & Hogan, 1992), the Schedule for Nonadaptive and Adaptive Personality (Clark, 1993) and the Schedule for Nonadaptive and Adaptive Personality–2nd edition (Clark, Simms, Wu, & Casillas, 2014), and HEXACO Personality Inventory–Revised (HEXACO-PI-R; Lee & Ashton, 2016)—offer the hierarchical assessment of both higher order domains and lower order facets.
Our focus in this article is on the highly influential five-factor model (FFM) of personality. Several hierarchical FFM instruments already are available (for a recent review, see Soto & John, 2016). We highlight three of them here. First, the Revised NEO Personality Inventory (NEO PI-R; Costa & McCrae, 1992)—which subsequently was supplanted by the NEO Personality Inventory–3 (NEO-PI-3; McCrae, Costa, & Martin, 2005)—is a very popular measure of the FFM. Each of the FFM domains is assessed using six eight-item facet scales. For instance, the NEO Extraversion domain includes the facets of Warmth, Gregariousness, Assertiveness, Activity, Excitement-Seeking, and Positive Emotions. Similarly, the NEO Conscientiousness domain consists of Competence, Order, Dutifulness, Achievement Striving, Self-Discipline, and Deliberation.
Second, Soto and John (2016) recently introduced the Big Five Inventory–2 (BFI-2). The original BFI simply assessed the five higher order FFM domains, although post hoc facet scales subsequently were created from its items. In contrast, the BFI-2 is a fully hierarchical measure in which each domain is measured using three four-item facet scales. For example, BFI-2 Extraversion includes specific facet scales assessing Sociability (paralleling NEO-PI-3 Gregariousness), Assertiveness (similar to NEO-PI-3 Assertiveness), and Energy Level (an amalgam of NEO-PI-3 Activity and Positive Emotions), whereas BFI-2 Conscientiousness contains facets tapping Organization (similar to NEO-PI-3 Order), Productiveness (assessing content similar to NEO-PI-3 Self-Discipline), and Responsibility (corresponding to NEO-PI-3 Dutifulness).
Third, the Big Five Aspect Scales (BFAS; DeYoung, Quilty, & Peterson, 2007) is a 100-item measure that uses two 10-item scales to model each FFM domain. For instance, BFAS Extraversion contains aspect scales assessing Assertiveness (parallel to NEO-PI-3 and BFI-2 Assertiveness) and Enthusiasm (similar to BFI-2 Energy level, it combines content related to both NEO-PI-3 Positive Emotions and Activity), whereas BFAS Conscientiousness includes Orderliness (corresponding to NEO-PI-3 Order and BFI-2 Organization) and Industriousness (corresponding to NEO-PI-3 Self-Discipline and BFI-2 Productiveness).
Faceted Inventory of the Five-Factor Model (FI-FFM)
Overview
This article introduces another hierarchical FFM measure, the FI-FFM. Several previously published articles already have reported results on various sets of FI-FFM scales (Naragon-Gainey & Watson, 2016; Naragon-Gainey, Watson, & Markon, 2009; Watson, Stasik, Ellickson-Larew, & Stanton, 2015a, 2015b; Watson, Stasik, Ro, & Clark, 2013). Here, we describe the development of the FI-FFM and then present extensive reliability and validity data.
Distinctive Aspects of the FI-FFM
Multiple Stages of Scale Development
We present data establishing that the FI-FFM displays a strong level of convergent validity vis-à-vis other FFM measures. As we show, it converges particularly well with the NEO PI-R and NEO-PI-3. At the same time, however, it is important to point out distinctive aspects of the FI-FFM scale development process that distinguish it from other FFM measures. First, because of the very large number of items that were considered for inclusion in the instrument, the FI-FFM was constructed sequentially over five stages of data collection and analysis. As we discuss in greater detail subsequently, the first stage of data collection focused on the creation of extraversion facet scales. Stage 2 concentrated on the initial development of specific conscientiousness and agreeableness scales, whereas Stage 3 primarily was concerned with the creation of facets within neuroticism and openness. Stages 4 and 5 led to further refinements of already existing facets as well as the creation of some new scales.
Overinclusive, Bottom-up Approach
In developing the FI-FFM, we carefully examined the existing personality literature to create the initial target constructs. We then wrote multiple items to assess each of these target traits. However, we did not impose any a priori structure on the resulting scales, such as specifying that each domain should have the same number of facets. Rather we adopted a more exploratory approach, using the data we collected to explicate the nature and structure of each domain. In doing so, we adopted the classic scale construction principle of overinclusiveness (Clark & Watson, 1995; Loevinger, 1957). That is, we tested the boundaries of these domains by including content that was broader and more inclusive than our own theoretical conceptualization. Similarly, we did not specify in advance that each scale should have the same number of items; rather we varied the length of the scales to ensure that each of them displayed a reasonable level of internal consistency (our target here was a coefficient alpha ≥ .80).
Asymmetric Assessment
This exploratory, data-driven approach is responsible for an unusual feature of the FI-FFM that distinguishes it from the NEO-PI-3, BFI-2, and BFAS: Its assessment of the five domains is asymmetric. Specifically, the FI-FFM contains five neuroticism facets (Anxiety, Depression, Anger Proneness, Somatic Complaints, and Envy), five markers of extraversion (Positive Temperament, Sociability, Ascendance, Venturesomeness, and Frankness), five facets of conscientiousness (Self-Discipline, Dutifulness, Deliberation, Achievement Striving, and Order), four indicators of Agreeableness (Empathy, Trust, Straightforwardness, and Modesty), and three facets of Openness (Intellectance, Novel Experience Seeking, and Nontraditionalism). Overall, the FI-FFM contains a total of 22 facet scales that are used to define the five higher order domains.
Advantages of Creating a New Measure
Thus, the FI-FFM contains some distinctive features that make it an attractive option in trait-based research. Nevertheless, as was stated earlier, the FI-FFM converges strongly with other FFM measures and, in fact, assesses many of the same facets. Given that several hierarchical FFM measures already exist, why create another one?
We offer three considerations in answering this question. First, the FI-FFM contains several facets that differ from those included in existing hierarchical FFM measures. For instance, existing hierarchical FFM instruments do not contain clear counterparts to the FI-FFM neuroticism facets of Somatic Complaints and Envy. We subsequently examine the incremental validity of these two FI-FFM scales—in joint analyses with the neuroticism scales of the NEO-PI-3—in predicting indicators of internalizing psychopathology.
Second, the creation of the FI-FFM helps clarify the lower order level of the personality hierarchy. As we show, many FI-FFM scales clearly assess the same basic facet-level traits that are assessed in other hierarchical FFM inventories; in particular, we show that multiple scales show a clear convergent/discriminant pattern vis-à-vis their counterparts in the NEO PI-R/NEO-PI-3. The existence of these replicable facets across multiple instruments—which were created using different construction strategies—strengthens our confidence that they represent robust and distinctive components within their respective FFM domains.
Third, researchers increasingly are recognizing the value of studying personality at the specific facet level (for discussions of this issue, see Paunonen, 2003; Soto & John, 2016; Watson et al., 2015b). The creation of a novel hierarchical measure provides additional trait indicators that can facilitate a new generation of facet-based research. In particular, the availability of multiple hierarchical trait measures allows researchers to model consensually defined facets as latent factors. As one example of this approach, Watson et al. (2015b) conducted exploratory factor analyses using the extraversion facet scales from the NEO-PI-3, FI-FFM, and HEXACO-PI-R. These structural analyses revealed four underlying facet factors—Positive Emotionality, Sociability, Assertiveness, and Experience Seeking—that showed highly distinctive associations with psychopathology. These data illustrate the advantages of having access to multiple indicators of each target facet.
Development of the FI-FFM
Participants and Procedure
Overview
As was discussed earlier, the FI-FFM was created over five stages of scale development. All participants were undergraduate students who completed a pool of candidate FI-FFM items in partial fulfillment of a course research exposure requirement or for course credit. The participants responded to the items using a 5-point Likert-type format ranging from strongly disagree to strongly agree.
Stage 1. The participants in Stage 1 were 400 undergraduate students at the University of Iowa. They completed an initial pool of 107 extraversion items.
Stage 2. The participants in this stage were 344 undergraduate students at the University of Iowa. They were administered a reduced pool of 69 extraversion items, plus 96 conscientiousness and 90 agreeableness items that were newly written for this round.
Stage 3. The Stage 3 analyses were based on responses from 404 University of Iowa students. These participants completed a pool consisting of 47 extraversion items, 59 conscientiousness items, 109 agreeableness items, 98 neuroticism items, and 113 openness items.
Stage 4. The Stage 4 participants consisted of a new sample of 404 Iowa undergraduates. These students provided responses to 52 extraversion items, 69 conscientiousness items, 53 agreeableness items, 73 neuroticism items, and 109 openness items.
Stage 5. Analyses in this stage were based on the responses of 707 students from the University of Iowa and the University at Buffalo. They completed an overall pool of 296 items. This included 227 items from the 23 FI-FFM scales that had been created in Stages 2 through 4, plus 69 items that were written to examine four new candidate scales assessing envy, dependency, conventionality, and modesty.
Structural Analyses
In each stage, the participants’ responses were subjected to a series of principal factor analyses; the factors that emerged were rotated to orthogonal simple structure using varimax. The basic structural analyses were conducted separately within each domain (e.g., in Stage 2, we examined the 69 extraversion items separately from the 96 conscientiousness items and 90 agreeableness items), but we also examined the discriminant validity of items and scales in relation to the other domains. In the following section, we summarize the scale development process separately for each domain.
Scale Development Summary
Extraversion
The initial measurement model for extraversion drew heavily on the literature review and integrative trait structure presented by Watson and Clark (1997). It consisted of seven target traits (Simms, 2009; Simms, Wu, & Watson, 2001a): dominance, exhibitionism, positive affectivity, energy, sociability, venturesomeness, and frankness. However, principal factor analyses of the Stage 1 data indicated the presence of only five well-defined and interpretable factors. Three of these factors—Sociability, Venturesomeness, and Frankness—clearly represented their corresponding target traits. However, the dominance and exhibitionism items defined a single common dimension that we labeled Ascendance. Similarly, the energy and positive affectivity items combined to form a Positive Temperament factor.
This five-factor structure proved to be quite robust, as it replicated well in Stages 2 through 5. Accordingly, the FI-FFM Extraversion domain includes 41 items (31 positively keyed, 10 reverse keyed) that are organized into five facet scales: Positive Temperament (8 items; e.g., “I generally enjoy life”), Sociability (9 items; e.g., “I enjoy spending time with people”), Ascendance (8 items; e.g., “I usually take charge in a group of people”), Venturesomeness (8 items; e.g., “I enjoy intense, exciting experiences”), and Frankness (8 items; e.g., “I speak my mind”).
Conscientiousness
Based on a review of the literature, we identified 12 conceptual domains within conscientiousness (Wu, Simms, & Watson, 2002): achievement, caution, deliberation, dependability, dutifulness, goal orientation, hard work, orderliness, organization, responsibility, risk aversion, and self-discipline. These then were organized into five broader groupings to guide initial item creation.
Analyses of the Stage 2 data yielded five clear and well-defined factors that represented these initial target constructs; we labeled them Self-Discipline, Achievement Striving, Deliberation, Dutifulness, and Order. This five-factor structure was very robust and emerged again in Stages 3 through 5. In Stage 5, we also explored the possibility of a sixth facet—conventionality—within this domain. However, the conventionality items tended to be interstitial, displaying substantial associations with both conscientiousness and agreeableness. Moreover, a provisional Conventionality scale proved to be largely redundant with existing FI-FFM scales (Simms, 2009). Accordingly, the scale was dropped and not considered further. Consequently, the final FI-FFM Conscientiousness domain consists of 42 items (28 positively keyed, 14 reverse keyed) that are grouped into five facet scales: Self-Discipline (8 items; e.g., “It is easy for me to stay on task”), Achievement Striving (7 items; e.g., “I have many plans for the future”), Deliberation (9 items; e.g., “I take my time when making decisions”), Dutifulness (10 items; e.g., “I am very dependable”), and Order (8 items; e.g., “I am highly organized”).
Agreeableness
Our initial measurement model for agreeableness consisted of six target traits (Simms, 2009; Simms, Wu, & Watson, 2001b): altruism, cooperativeness/compliance, empathy, courtesy/politeness, straightforwardness, and trust (vs. cynicism). However, principal factor analyses of the Stage 2 data yielded only four interpretable factors: Empathy (which included content related to altruism, empathy, and courtesy), Straightforwardness, Trust, and Cynicism.
Analyses of a revised item pool in Stage 3 resulted in a five-factor structure. Four of these factors essentially replicated those seen in Stage 2; in addition, some Straightforwardness items split off to form a separate Hostility factor. Not surprisingly, a provisional Hostility scale correlated quite strongly (r = .76; see Simms, 2009) with the Anger Proneness facet of Neuroticism. It therefore was dropped from further consideration (with some of its items moving to Anger Proneness), leaving only four facets going into Stage 4.
This four facet structure largely replicated in Stage 4, except that—consistent with our original expectation—the trust and cynicism items now defined a single bipolar dimension. Accordingly, the markers of this factor were used to create a revised Trust (vs. Cynicism) scale.
The Agreeableness domain therefore was reduced to only three facets after Stage 4. However, we included items related to modesty/humility in the Stage 5 data collection. These items defined a clear and interpretable factor in the Stage 5 structural analyses and were used to create a corresponding scale. Thus, the final FI-FFM Agreeableness domain consists of four facet traits and a total of 42 items (20 positively keyed, 22 reverse keyed): Empathy (10 items; e.g., “I always try to consider the needs of others”), Trust (11 items; e.g., “I generally take people at their word”), Straightforwardness (11 items; e.g., “I am good at manipulating others” [reversed]), and Modesty (10 items; e.g., “I’m not one to boast or brag”).
Neuroticism
In Stage 3, items were written to assess eight proposed facets of neuroticism (Simms, 2009; Simms, Wu, & Watson, 2003): anxiety, depression, anger proneness, guilt/self-blame, emotional lability, oversensitivity, negativistic appraisal, and somatic complaints. Structural analyses, however, revealed only five factors, which we labeled Anxiety, Depression, Anger Proneness, Somatic Complaints, and Emotional Hardiness. Four of these factors clearly replicated in Stage 4, but the emotional hardiness items now split across multiple dimensions; the preliminary Hardiness scale therefore was dropped at this point.
The four remaining factors were robust and replicated again in Stage 5. In this stage, we also included items assessing two new candidate facets: envy and dependency. These items defined two clearly interpretable factors and were used to create corresponding scales. Structural analyses, however, established that Dependency was not a clear marker of the higher order Neuroticism factor; it therefore was eliminated as a facet scale. Accordingly, the final FI-FFM Neuroticism domain contains 48 items (32 positively keyed, 16 reverse keyed) that are grouped into five facets: Anxiety (10 items; e.g., “I find myself worrying a lot”), Depression (10 items; e.g., “Life looks pretty bleak”), Anger Proneness (10 items; e.g., “I am easily angered”), Somatic Complaints (8 items; e.g., “I am often bothered by an upset stomach”), and Envy (10 items; e.g., “I am prone to jealousy”).
Openness
We wrote items assessing eight target aspects of openness in Stage 3 (Simms, 2009): novel experience seeking, culture, intellectual curiosity, creativity, nontraditionalism, emotional resonance, unusual experiences, and eccentric beliefs. Principal factor analyses of the Stage 3 data revealed six well-defined factors. Five of these factors—Novel Experience Seeking, Nontraditionalism, Emotional Resonance, Unusual Experiences, and Eccentric Beliefs—clearly corresponded to our original target constructs. In contrast, the final factor represented a combination of the culture, intellectual curiosity, and creativity items; we labeled this dimension Intellectance.
These six factors replicated very well in Stages 4 and 5 and were used to create corresponding scales. However, only three of these scales ultimately were found to be clear markers of Openness in structural analyses. Thus, the final FI-FFM Openness domain includes 34 items (21 positively keyed, 13 reverse keyed) that are grouped into three component scales: Intellectance (11 items; e.g., “I enjoy learning for learning’s sake”), Novel Experience Seeking (11 items; e.g., “I’ll try anything once”), and Nontraditionalism (12 items; e.g., “I’m pretty old fashioned” [reversed]).
The FI-FFM
After these five stages of scale development, the final version of the FI-FFM contains 207 items (132 positively keyed, 75 reverse keyed) that are organized in 22 facet scales (five apiece for Neuroticism, Extraversion, and Conscientiousness; four for Agreeableness; three for Openness). The complete 207-item FI-FFM—along with scoring information—is provided in the Supplemental Appendix A (all supplemental materials are available online).
Supplemental Scales
We developed four scales that were dropped from the final version of the FI-FFM. These are traits that we believed—based on our overinclusive review of the literature—potentially might represent facets within one of the FFM domains. Moreover, they did emerge as distinct factors in our scale development analyses. However, subsequent structural analyses established that they do not fit cleanly into the FFM. We have retained them as supplemental scales to provide assessment possibilities for these traits while emphasizing that our analyses indicate they do not actually fit well within the FFM. These scales are Dependency (10 items; e.g., “I like having other people take care of me”), Emotional Resonance (10 items; e.g., “I try to be open with my feelings”), Unusual Experiences (10 items; e.g., “Sometimes, things don’t seem real to me”), and Eccentric Beliefs (10 items; e.g., “I believe in the existence of ghosts or spirits”). These scales—along with scoring information—are presented in the Supplemental Appendix B. We present some basic psychometric data for these four scales in a series of supplemental tables.
Reliability and Validity of the FI-FFM Scales
Method
Participants and Procedure
Sample 1
We examine the reliability and validity of the FI-FFM using data from three samples. The first sample was the same one used in Stage 5 of the scale development process. As noted earlier, the participants were 707 students from the University of Iowa and the University at Buffalo. The sample consisted of 293 men and 413 women (the gender of one participant was unspecified) with a mean age of 19.4 years (range = 18-46 years). The sample was 64.3% White, 18.4% Asian/Pacific Islander, 7.4% Black/African American, 5.0% Hispanic/Latino, and 5.1% other/unspecified.
A subset of these students (N = 396) signed up for a retest study and agreed to complete questionnaires during two testing sessions that were scheduled 2 weeks apart. Of these, 312 participants (78.8%) returned for a second testing session 2 weeks later.
Another substantial portion of the sample enrolled in a self–other study and were instructed to “bring someone who knows you fairly well” to a single testing session in which they would be answering questions about themselves (i.e., the target) and their informant also would be answering questions about them (i.e., the target). Data were collected from 277 informants; this sample included 129 men and 148 women, with a mean age of 19.6 years (range = 16-52 years). A majority of these informants (N = 198, 71.5%) identified themselves as friends of the target, and another 25.3% (N = 70) were romantic partners. On average, the informants had known the participants an average of slightly more than a year (median = 14 months, range = 2-240 months). In addition, the informants were asked to rate how well they knew the target (1 = not at all, 5 = very well), how much they liked the target (1 = not at all, 5 = very much), and how close they felt to the target (1 = not close at all, 5 = very close). They indicated that they knew the targets well (M = 4.27), liked them (M = 4.65), and felt close to them (M = 4.37).
Finally, a small percentage of the students attended a single testing session without informants during which they simply answered questions about themselves. All of the primary participants received course credit, whereas the informants were compensated with a $10 gift certificate for an area shopping mall.
Sample 2
Adults living in the greater South Bend metropolitan area were recruited for two online data sessions that were completed an average of approximately 1 week apart (mean interval = 6.2 days). Participants were compensated $20 for each session (for more details regarding this sample, see Watson, Stasik, et al., 2013).
In the first session, 375 individuals completed both the FI-FFM and the Personality Inventory for DSM-5 (PID-5; Krueger, Derringer, Markon, Watson, & Skodol, 2012). During the second session, 335 of these participants (89.3%) were assessed on the original BFI (John & Srivastava, 1999). The overall sample consisted of 121 men and 254 women, with a mean age of 36.4 years (range = 20-83 years). The sample was 86.7% White, 6.4% Black/African American, and 6.9% other.
Sample 3
The participants were 439 adults from the greater South Bend metropolitan area (for more details regarding this sample, see Watson et al., 2015a, 2015b). Individuals who had provided their contact information from previous studies conducted at the Center for Advanced Measurement of Personality and Psychopathology (CAMPP) were recruited first; other adults were recruited though flyers posted in local mental health clinics and via word of mouth (participants could let other potentially eligible individuals know about the study). All potential participants were screened to ensure they met the following eligibility criteria: 18 years of age or older, able to read and write in English, and capable of providing consent to participate.
Participants were seen in three 3-hour sessions conducted at CAMPP; they were paid $60 for each session. They were assessed in small group sessions that typically involved 3 to 10 individuals. We report results here using data from the first two sessions. Session 1 consisted of an extensive battery of personality measures, plus a portion of a clinical interview. The FI-FFM and the NEO-PI-3 (McCrae et al., 2005) both were administered during this initial session.
Session 2 was held roughly 3 weeks later (mean interval = 20.3 days). It included an extensive battery of self-report psychopathology measures, plus the rest of the clinical interview. We present data from the PID-5, various self-report indicators of internalizing psychopathology, and the clinical interview, using responses from the 411 individuals who completed this second session (interview data are missing for four of these participants).
It should be noted that participants from previous CAMPP studies primarily were outpatients recruited from various sources, such as the local community mental health center. Consequently, this sample is characterized by a relatively high level of psychopathology. In fact, nearly half of the sample (N = 203, 46.2%) answered “yes” to one or more of these three questions: “Are you currently receiving psychological counseling/therapy for mental health issues?” “Have you received psychological counseling/therapy for mental health issues in the past?” “Are you currently taking medications to treat a mental illness?” Similarly, approximately half of the interviewed participants (198 of 407, or 48.6%) met criteria for at least one of the assessed DSM diagnoses.
The sample (mean age = 45.0 years, age range = 18-77) consisted of 140 men and 298 women (the gender of one participant was unspecified). It was 47.6% Black/African American, 44.2% White, and 8.2% multiracial or other (for additional demographics, see Watson et al., 2015a, 2015b).
Measures
FI-FFM
Participants in all three samples (combined N = 1,521) rated themselves on the final version of the FI-FFM. As noted previously, the FI-FFM uses a 5-point Likert-type format ranging from strongly disagree to strongly agree.
Two other-report versions (one for use with male targets, the other for female targets) were created for the Sample 1 informants. This essentially involved changing the FI-FFM items from the first person to the third person (e.g., “I sometimes blurt out what I am thinking” became “X sometimes blurts out what she is thinking,” where “X” refers to the target participant).
NEO
A subset of the students in Sample 1 (N = 464) completed the NEO PI-R (Costa & McCrae, 1992), whereas the Sample 3 adults (N = 433) were administered the more recent NEO-PI-3 (McCrae et al., 2005). These two versions of the NEO are very similar; the only change was that 38 NEO PI-R items were revised in the NEO-PI-3 to lower the reading level and to make the instrument more appropriate for younger examinees and adults with lower educational levels. Items are rated on a 5-point Likert-type scale ranging from strongly disagree to strongly agree.
As was discussed earlier, each NEO domain contains six eight-item facet scales. For example, the Neuroticism facets are Anxiety, Angry Hostility, Depression, Self-Consciousness, Impulsiveness, and Vulnerability. We report data on both the domain and facet scores. For ease of presentation, we report these results collapsed across Samples 1 and 3 (overall N = 897). To eliminate any mean-level differences across samples (due either to participant-based differences or to the use of slightly different versions of the NEO), we standardized the scale scores on a within-sample basis and then combined them to permit a single overall analysis. 1
BFI
Participants in Samples 1 (N = 694) and 2 (N = 335) completed the 44-item version of the original BFI (John & Srivastava, 1999). This instrument contains 8-item scales assessing Neuroticism and Extraversion, a 10-item Openness scale, and 9-item measures of Agreeableness and Conscientiousness. Respondents rated the extent to which each statement characterized them on a 5-point scale ranging from strongly disagree to strongly agree. As with the NEO, we report BFI results collapsed across the two samples (overall N = 1,029). Once again, we standardized the scores on a within-sample basis and then combined them in a single overall analysis. 2
PID-5
Participants in Samples 2 (N = 375) and 3 (N = 411) were assessed on the PID-5 (Krueger et al., 2012). The PID-5 is a 220-item self-report instrument that provides a comprehensive assessment of personality pathology as organized in Section III of DSM-5 (American Psychiatric Association, 2013). This DSM-5 model has strong and systematic associations with general traits of personality, such as those included in the FFM (see Watson, Stasik, et al., 2013).
The PID-5 includes 25 primary facet scales, varying in length from 4 to 14 items. The instrument consists of sentences that are rated on a 4-point scale ranging from 0 (very false or often false) to 3 (very true or often true). The Sample 3 participants completed the full, 220-item version of the PID-5; three suicidality items were dropped in Sample 2, however, yielding a reduced 11-item version of the Depressivity scale. Following the same approach used with the NEO and BFI, we report overall PID-5 results based on a standardized combined sample (overall N = 786). 3
Anxiety and depression diagnoses
The Sample 3 participants were assessed using the Mini-International Neuropsychiatric Interview (M.I.N.I.; Sheehan et al., 1998) during Sessions 1 and 2. The M.I.N.I. is a brief structured diagnostic interview that assesses symptoms of DSM-IV (American Psychiatric Association, 2000) and ICD-10 (World Health Organization, 1993) psychiatric disorders; we used an adapted version (with the authorization of the author) that incorporated diagnostic changes for DSM-5 (American Psychiatric Association, 2013). 4 We report data here on seven internalizing diagnoses. Five diagnoses—panic disorder, agoraphobia, posttraumatic stress disorder (PTSD), social anxiety disorder, and obsessive–compulsive disorder (OCD)—were assessed in Session 1. Major depressive disorder and generalized anxiety disorder (GAD) were assessed in Session 2. Watson et al. (2015a, 2015b) report prevalence rates for these diagnoses in this sample.
Interviewers were graduate students and advanced undergraduate research assistants (RAs) who underwent extensive training on the M.I.N.I. Graduate students had prior training in clinical interviewing and the use of the M.I.N.I., and served as trainers for the undergraduate RAs. Training included in-depth review of DSM criteria for each disorder being assessed, didactics on clinical interviewing skills and administration of a semistructured interview, and a detailed overview of the administration of each item in the interview. Each RA was required to observe three administrations of the interview by a graduate student and subsequently be observed administering the interview on three separate occasions.
To assess interrater reliability, the interviews were audiotaped; a second rater independently scored 39 of the Session 1 interviews and 34 of the Session 2 interviews (due to audiotape problems, N = 38 and 33, respectively, for some disorders). Note, however, that a kappa could not be computed for agoraphobia because none of the rescored cases met criteria for this disorder. The kappas for the other diagnoses all were in the excellent range (Cicchetti, 1994), with values ranging from .77 to 1.00.
Anxiety and depression symptoms
As noted earlier, the Sample 3 participants completed an extensive battery of self-report psychopathology scales. We report results on seven self-report symptom scores that parallel the internalizing diagnoses assessed in the M.I.N.I. interview (for more details regarding these measures, see Watson et al., 2015a, 2015b). Wherever possible, we aggregated redundant, highly correlated scales into a series of symptom composites; in each case, the variables were standardized before being combined so that they would be equally weighted. First, the Generalized Anxiety Disorder Questionnaire–IV (Newman et al., 2002) provides comprehensive assessment of GAD symptoms. The Generalized Anxiety Disorder Questionnaire–IV was designed originally to provide an analogue diagnosis of GAD, and it therefore closely follows the diagnostic criteria for the disorder. However, the items also can be scored dimensionally, and this scoring is used here.
Second, we created a composite collapsing across two measures of depression: (a) the 9-item Patient Health Questionnaire (Kroenke, Spitzer, & Williams, 2001) and (b) the 20-item General Depression Scale from the Inventory of Depression and Anxiety Symptoms (IDAS; Watson et al., 2007). These scales correlated .83 with each other.
Third, we created a composite using three indicators of panic: (a) the 8-item Panic scale from the Expanded Version of the IDAS (IDAS-II; Watson et al., 2012) (b) a reduced, 9-item version of the Anxious Arousal scale of the Mood and Anxiety Symptom Questionnaire (Watson et al., 1995), and (c) an abbreviated, 6-item version of the Panic Attack Symptom Questionnaire (Watson, 2000). These scales had correlations ranging from .59 to .68 (mean r = .64).
Fourth, we combined two measures of PTSD symptoms: (a) the five intrusions items and two avoidance items from the PTSD Checklist–Civilian Version (Weathers, Litz, Herman, Huska, & Keane, 1993) and (b) an aggregate score based on the Traumatic Intrusions (4 items) and Traumatic Avoidance (4 items) scales of the IDAS-II. These indicators correlated .74 with one another.
Fifth, we combined four measures of social anxiety: (a) the 5-item Social Phobia scale from the Fear Questionnaire (FQ; Marks & Mathews, 1979), (b) the 10-item Social Phobia scale from the Albany Panic and Phobia Questionnaire (APPQ; Rapee, Craske, & Barlow, 1994/1995), (c) the 6-item IDAS-II Social Anxiety scale (Watson et al., 2012), and (d) a factor-analytically derived 10-item Social Anxiety scale from the Schizotypal Personality Questionnaire (Chmielewski & Watson, 2008). Correlations among these measures ranged from .54 to .69 (mean r = .64).
Sixth, we created a composite using (a) the 9-item APPQ Agoraphobia scale and (b) the 5-item FQ Agoraphobia scale. These measures correlated .67 with each other.
Finally, we aggregated three indicators of OCD: (a) the 18-item Obsessive–Compulsive Inventory–Revised (Foa et al., 2002); (b) a total score based on the Obsessive Checking (14 items), Obsessive Cleanliness (12 items), Compulsive Rituals (8 items), and Hoarding (5 items) scales from the Schedule of Compulsions, Obsessions, and Pathological Impulses (Watson & Wu, 2005); and (c) a combined score based on the IDAS-II Checking (3 items), Ordering (5 items), and Cleaning (7 items) scales. Correlations among these measures ranged from .66 to .77 (mean r = .71).
Health anxiety and hypochondriasis
The Sample 3 participants also completed three measures of hypochondriasis/health anxiety. First, the Health Anxiety Questionnaire (Lucock & Morely, 1996) is a 21-item instrument that was designed to measure cognitive and behavioral aspects of hypochondriasis; participants respond to each item using a 4-point scale ranging from not at all or rarely to most of the time. The Sample 3 participants completed two Health Anxiety Questionnaire scales: Health Worry and Preoccupation (8 items; e.g., “Do you ever find it difficult to keep worries about your health out of your mind?”) and Fear of Illness and Death (7 items; e.g., “Are you ever worried that you might get a serious illness in the future?”).
Second, the Multidimensional Inventory of Hypochondriacal Traits (MIHT; Longley, Watson, & Noyes, 2005) contains four-factor analytically derived subscales: the 7-item Affective subscale measures worry about one’s health (e.g., “When I experience pain, I fear I may be ill”), the 7-item Cognitive subscale measures conviction of illness despite contrary evidence (e.g., “Few people seem to take my health concerns as seriously as I do”), the 8-item Behavioral subscale taps reassurance seeking (e.g., “If my symptoms worry me, I appreciate sympathy from others”), and the 9-item Perceptual subscale assesses preoccupation with bodily sensations (e.g., “I keep close track of what is happening to me physically”). Participants respond to each item using a 5-point scale ranging from strongly disagree to strongly agree.
Third, the Whiteley Index (WI; Pilowsky, 1967) is a widely used measure of hypochondriasis; it includes 14 items that are rated on a 5-point scale ranging from strongly disagree to strongly agree. Longley et al. (2005) subjected the WI items to a structural analysis; they identified two factors representing cognitive symptoms of disease conviction and affective symptoms of illness worry. These same two factors were identified in the current sample; we therefore created corresponding Cognitive (7 items; e.g., “I have the symptoms of very serious illnesses”) and Affective (e.g., “I think that I worry about my health more than most people”) scales that are used in subsequent analyses.
Results and Discussion
Reliability
Internal Consistency
The first three columns of Table 1 report internal consistency data (coefficient alphas) for the FI-FFM scales in each sample. These alpha coefficients generally are strong and establish a substantial level of internal consistency. Looking first at the five domain scales, the alphas ranged from .86 to .93 (median = .90) in Sample 1; from .91 to .96 (median = .94) in Sample 2; and from .83 to .95 (median = .91) in Sample 3. These values are quite similar to those typically observed for the domain scales of the NEO PI-R and NEO-PI-3. For example, McCrae, Kurtz, Yamagata, and Terracciano (2011) present NEO PI-R reliability data in three samples (see their Table 2). They reported coefficient alphas for the NEO domain scores ranging from .86 to .92 (median = .89), from .89 to .93 (median = .92), and from .87 to .93 (median = .89).
Internal Consistencies and Test–Retest Reliabilities of the FI-FFM Domain and Facet Scales.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; Novelty Seeking = Novel Experience Seeking. N = 707 (Sample 1), 375 (Sample 2), 439 (Sample 3), 312 (Retest). Retest interval was 2 weeks.
Promax Factor Loadings of the FI-FFM Facet Scales.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; Novelty Seeking = Novel Experience Seeking.. N = 1,521. Loadings ≥ |.35| are bolded. The highest loading for each scale is indicated with an asterisk.
The FI-FFM facet scales also displayed good internal consistency. Coefficient alphas ranged from .75 to .86 (median = .79) in Sample 1; from .82 to .92 (median = .86) in Sample 2; and from .73 to .90 (median = .81) in Sample 3. These values are somewhat higher than those typically found for the NEO facet scales. For instance, McCrae et al. (2011) reported coefficient alphas for the NEO facets ranging from .56 to .81 (median = .71), from .50 to .87 (median = .77), and from .50 to .82 (median = .73). In contrast to these NEO data, no FI-FFM facet scale had an alpha below .70 in any sample.
Retest Reliability
The final column of Table 1 presents retest correlations computed across a 2-week time interval in a subset (N = 312) of the Sample 1 participants. Given this short time span, these correlations can be interpreted as dependability coefficients (Chmielewski & Watson, 2009; Gnambs, 2014; Watson, 2004). Dependability can be defined “as the correlation between two administrations of the same test when the lapse of time is insufficient for people themselves to change with respect to what is being measured” (Cattell, Eber, & Tatsuoka, 1970, p. 30). Thus, in contrast to longer term stability correlations—which are influenced both by measurement error and by true change—dependability coefficients provide a highly sensitive index of the effects of transient error (Gnambs, 2014).
Retest coefficients for the FI-FFM domain scales ranged from .78 to .80, with a median value of .79. These strong dependability correlations are quite similar to the meta-analytic results reported by Gnambs (2014), who obtained values ranging from .78 to .85 (median = .82) for trait measures of the Big Five.
Retest correlations for the FI-FFM facet scales also tended to be strong, but somewhat lower, ranging from .64 to .82 (median r = .75). By way of comparison, McCrae et al. (2011) reported 1-week retest correlations for the NEO facet scales that ranged from .70 to .91, with a median coefficient of .83. These higher retest correlations for the NEO scales may reflect, in part, the shorter time interval that was used (i.e., 1 week vs. 2 weeks), although Gnambs (2014) found that interval length had a relatively modest effect on the magnitude of dependability coefficients. These results obviously need to be replicated in different types of samples, but they tentatively suggest that some FI-FFM scales are somewhat more susceptible to transient error than are similar trait measures in other instruments.
Supplemental Scales
Supplemental Table S1 presents parallel internal consistency and retest data for the four supplemental scales. These results are similar to those seen with the facet scales. Across the three samples, coefficient alphas ranged from .68 to .87, with a median value of .79. Dependability correlations ranged from .65 to .81, with a median value of .72.
Factor Structure
Factor Analysis of the Facet Scales
To clarify the hierarchical structure of the FI-FFM, we conducted a factor analysis of the 22 facet scales, using data collapsed across all three samples. Specifically, we conducted a principal factor analysis, using squared multiple correlations as the initial communality estimates. To eliminate mean-level differences across samples, we standardized the scale scores on a within-sample basis and then combined them to permit a single overall analysis. We extracted five factors and rotated them to oblique simple structure using promax (power = 3); the loadings from this five-factor solution are presented in Table 2. 5
The key finding in Table 2 is that the facet scales all loaded most strongly on the factor representing their target domain. That is, the five neuroticism facet scales had loadings ranging from .60 (Envy) to .90 (Anxiety) on Factor I; the five extraversion facets had loadings ranging from .52 (Frankness) to .68 (Venturesomeness) on Factor II; the five conscientiousness facets had loadings ranging from .57 (Dutifulness) to .73 (Self-Discipline) on Factor III; the four agreeableness facets had loadings ranging from .39 (Trust) to .76 (Empathy) on Factor IV; and the three openness facets had loadings ranging from .48 (Nontraditionalism) to .62 (Novel Experience Seeking) on Factor V.
It should be noted, however, that several scales had moderate secondary loadings on other factors. Most notably, the Trust scale had similar moderate loadings on both Factor IV (.39) and Factor I (−.34). In addition, Achievement Striving (.32), Novel Experience Seeking (.30), Deliberation (−.30), and Modesty (−.35) all had moderate secondary loadings on Factor II, whereas Sociability had a salient secondary loading (.35) on Factor IV.
Expanded Analysis Including the Supplemental Scales
Next, we repeated this analysis including the four supplemental scales as additional indicators. We initially examined whether additional factors could be identified using this expanded set of variables. However, no factors beyond the first five were well defined and clearly interpretable. For example, in the six-factor solution, the last factor had only a single marker (Dependency, with a loading of .57). Accordingly, we again extracted five factors and rotated them using promax (power = 3).
The loadings from this solution are presented in Supplemental Table S2. Three of the supplemental scales were not pure markers of any dimension: Unusual Experiences (loadings of .45 and .28, respectively) and Eccentric Beliefs (loadings of .19 and .32, respectively) both split between Factors I and V, whereas Dependency had modest loadings on both Factors III (−.39) and V (−.34). In contrast, Emotional Resonance—which was developed to be a component of Openness—actually was a strong marker of Factor IV (with a loading of .56), suggesting that it could be considered as an additional facet of agreeableness. This is an interesting issue that merits further investigation.
Convergent and Discriminant Validity
Domain Scales
We now present an extensive series of analyses to establish the convergent and discriminant validity of the FI-FFM scales. We begin by reporting data on the five domain scores. Table 3 presents three sets of validity analyses. The top portion of Table 3 displays correlations with the five corresponding domain scores from the NEO (N = 897). These results establish an impressive level of convergent validity: The convergent correlations range from .73 (Openness) to .82 (Conscientiousness), with a mean value of .78.
Convergent and Discriminant Correlations for the FI-FFM Domain Scores.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; Neur = Neuroticism; Extra = Extraversion; Con = Conscientiousness; Agree = Agreeableness; Open = Openness; NEO = NEO Personality Inventory; BFI = Big Five Inventory. N = 897 (NEO), 1,029 (BFI), 277 (informant ratings). The highest correlation in each row is in bold.
We now consider discriminant validity. A classic test of discriminant validity is that each of the convergent correlations should be higher than any of the other values in its row or column of the heteromethod block (Campbell & Fiske, 1959). Table 3 indicates that all of convergent coefficients easily met this criterion, as all of the discriminant correlations were less than |.45|. We further quantified these relations by conducting significance tests (using the Williams modification of the Hotelling test for two correlations involving a common variable; see Kenny, 1987) comparing these convergent correlations with each of the eight discriminant coefficients in the same row or column of the block; this yields a total of 40 tests of discriminant validity across the five traits. All 40 comparisons were significant (p < .01, one-tailed), which offers clear evidence of discriminant validity.
The middle portion of Table 3 presents correlations with the BFI scales (N = 1,029). We again see strong evidence of convergent validity; correlations ranged from .71 (Agreeableness) to .83 (Neuroticism), with a mean coefficient of .77. As before, the discriminant correlations were substantially lower (all coefficients were ≤ |.43|); replicating the NEO results, significance tests revealed that all 40 discriminant comparisons were significant (p < .01, one-tailed).
Finally, the bottom portion of Table 3 reports self-informant agreement correlations for the FI-FFM domain scales (N = 277). It is noteworthy that all five domain scores displayed moderate to strong levels of self-other agreement, with correlations ranging from .46 (Neuroticism and Extraversion) to .62 (Conscientiousness); the mean agreement correlation was .53 in these data. Once again, these convergent coefficients were substantially higher than the discriminant correlations (range = |.01| to |.30|, mean r = |.11|); significance tests established that all 40 discriminant comparisons were significant (p < .01, one-tailed).
Neuroticism Facets
We now present similar analyses for the five FI-FFM neuroticism facet scales. The top portion of Table 4 reports correlations with the six neuroticism facet scales from the NEO. It is noteworthy that three scale pairs show a very systematic convergent/discriminant pattern and clearly assess the same basic trait dimensions. That is, FI-FFM Anxiety correlated .74 with NEO Anxiety; FI-FFM Depression correlated .74 with NEO Depression; and FI-FFM Anger Proneness correlated .72 with NEO Angry Hostility. Follow-up significance tests indicated that each of these convergent correlations was significantly higher than all of the other correlations in its row or column of the heteromethod block (a total of 27 comparisons; all p < .01, two-tailed).
Convergent and Discriminant Correlations for the FI-FFM Neuroticism Facet Scales.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; NEO = NEO Personality Inventory; PID-5 = Personality Inventory for DSM-5. N = 897 (NEO), 786 (PID-5), 277 (informant ratings). The highest correlation in each row is in bold.
Next, the middle part of Table 4 presents correlations with selected PID-5 scales (N = 786). In choosing measures for inclusion in Table 4, we selected every PID-5 scale that correlated ≥ |.55| with at least one of the FI-FFM neuroticism facets. Paralleling the NEO results, three scale pairs show a systematic convergent/discriminant pattern and clearly assess the same basic trait: FI-FFM Anxiety correlated .75 with PID-5 Anxiousness; FI-FFM Depression correlated .78 with PID-5 Depressivity; and FI-FFM Anger Proneness correlated .76 with PID-5 Hostility. Follow-up significance tests indicated that each of these convergent coefficients was significantly higher than all of the other correlations in its row or column of the heteromethod block (a total of 33 comparisons; all p < .05, two-tailed).
The bottom portion of Table 4 reports self-informant agreement correlations for the FI-FFM neuroticism facets. Four scales—Anxiety, Depression, Anger Proneness, and Somatic Complaints—showed moderate self-informant agreement in these data, with convergent correlations ranging from .38 to .49 (mean r = .44). These scales also displayed very good discriminant validity. Follow-up tests revealed that 29 of the 32 discriminant comparisons (90.6%) involving these scales were significant (p < .05, one-tailed). The only exceptions were that the convergent correlation for Anxiety was not significantly higher than the discriminant coefficients between self-rated anxiety and informant-rated (a) Depression and (b) Somatic Complaints; and that the convergent coefficient for Anger Proneness was not greater than the discriminant correlation between self-rated Anger Proneness and informant-rated Depression.
In contrast, Envy produced a more modest agreement correlation (r = .26). Because of this, it showed unimpressive discriminant validity; in fact, only two of the eight comparisons involving this scale were significant.
Extraversion Facets
Table 5 presents parallel validity data for the FI-FFM extraversion facet scales. In these analyses, the middle portion of the table shows correlations with every PID-5 scale that correlated ≥ |.50| with at least one of the extraversion facets.
Convergent and Discriminant Correlations for the FI-FFM Extraversion Facet Scales.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; NEO = NEO Personality Inventory; PID-5 = Personality Inventory for DSM-5. N = 897 (NEO), 786 (PID-5), 277 (informant ratings). The highest correlation in each row is in bold.
In the NEO analyses, three scale pairs showed a highly systematic convergent/discriminant pattern and clearly assess the same basic trait dimensions: FI-FFM Ascendance correlated .72 with NEO Assertiveness; FI-FFM Sociability correlated .66 with NEO Gregariousness; FFM Venturesomeness correlated .64 with NEO Excitement-Seeking. Follow-up significance tests indicated that each of these convergent correlations was significantly higher than all of the other correlations in its row or column of the heteromethod block (a total of 27 comparisons; all p < .01, two-tailed). FI-FFM Positive Temperament showed a more diffuse and nonspecific pattern, correlating strongly with NEO Positive Emotions (r = .58), Activity (r = .55), and Warmth (r = .50; Warmth also correlated .50 with FI-FFM Sociability). Finally, in contrast to the other scales, Frankness was more moderately related to the NEO, with coefficients ranging from only .16 to .44.
The most striking aspect of the PID-5 data is that they are bidirectional, demonstrating both strongly positive and negative associations between extraversion and personality pathology. On the one hand, FI-FFM Positive Temperament correlated −.68 with PID-5 Anhedonia (e.g., nothing interests me) and FI-FFM Sociability correlated −.67 with PID-5 Withdrawal (e.g., I keep to myself). On the other hand, FI-FFM Venturesomeness correlated .57 with PID-5 Risk Taking (e.g., people say I am reckless), and FI-FFM Ascendance correlated .56 with PID-5 Attention Seeking (e.g., I like getting attention). These data are consistent with previous evidence establishing that extraversion has very complex bidirectional relations with psychopathology at the facet level (for a review, see Watson et al., 2015b). Moreover, the specificity of these relations is impressive; follow-up tests indicated that these four correlations were significantly greater than all of the other coefficients in their row or column of the heteromethod block (a total of 28 comparisons; all ps < .01, two-tailed).
All five facet scales displayed moderate self-other agreement, with coefficients ranging from .35 to .45 (mean r = .40). They also exhibited strong discriminant validity, as 36 of the 40 follow-up comparisons (90%) were significant (p < .05, one-tailed). The only exceptions were that the convergent correlation for Ascendance was not significantly higher than the discriminant coefficients between self-rated Ascendance and informant-rated (a) Positive Temperament and (b) Frankness; the convergent coefficient for Positive Temperament was not greater than the correlation between self-rated Positive Temperament and informant-rated Sociability (z = 1.64, p < .06); and the convergent correlation for Frankness did not exceed the coefficient between self-rated Frankness and informant-rated Ascendance.
Conscientiousness Facets
Table 6 presents corresponding results for the FI-FFM conscientiousness facet scales. As with extraversion, the middle portion of the table shows data for every PID-5 scale that correlated ≥ |.50| with at least one of the conscientiousness facets.
Convergent and Discriminant Correlations for the FI-FFM Conscientiousness Facet Scales.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; Dutiful = Dutifulness; NEO = NEO Personality Inventory; PID-5 = Personality Inventory for DSM-5. N = 897 (NEO), 786 (PID-5), 277 (informant ratings). The highest correlation in each row is highlighted.
In contrast to the other domains we have examined, each of the five FI-FFM conscientiousness scales has a clear counterpart in the NEO, with convergent correlations ranging from .64 to .75 (mean r = .70). Thus, the two Order scales correlated .75; the two Self-Discipline scales correlated .71; the two Deliberation scales correlated .71; the two Achievement Striving scales correlated .67; and the two Dutifulness scales correlated .64 with each other (it should be noted that FI-FFM Dutifulness also correlated .51 with NEO Competence). Follow-up tests established that these five convergent correlations were significantly greater (p < .05, two-tailed) than the discriminant coefficients in all 45 comparisons. These results establish that the two instruments assess conscientiousness in a highly similar manner.
In addition, three FI-FFM scales show a clear convergent/discriminant pattern with the PID-5: Self-Discipline correlated −.67 with Distractibility (e.g., can’t focus on things for long), Deliberation correlated −.70 with Impulsivity (e.g., do things on the spur of the moment), and Dutifulness correlated −.65 with Irresponsibility (e.g., make promises I don’t intend to keep). Follow-up tests established that these correlations were significantly greater (p < .01, two-tailed) than all 18 discriminant coefficients in their row or column of the heteromethod block.
Finally, these scales showed moderate to strong self–other agreement, with convergent correlations ranging from .38 to .66 (mean r = .48). Order and Self-Discipline—which had the two highest agreement correlations (r = .66 and .57, respectively)—also displayed excellent discriminant validity, as all 16 follow-up tests were significant (p < .05, one-tailed). In contrast, the discriminant validity of the three remaining facets—which had more moderate convergent correlations ranging from .38 to .42—was less impressive; in fact, only 14 of the 24 comparisons (58.3%) involving these scales were significant.
Agreeableness Facets
We report parallel findings for the agreeableness facets in Table 7. Once again, the middle portion of the table shows correlations with every PID-5 scale that correlated ≥ |.50| with at least one of these facets.
Convergent and Discriminant Correlations for the FI-FFM Agreeableness Facet Scales.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; Straightforward = Straightforwardness; NEO = NEO Personality Inventory; PID-5 = Personality Inventory for DSM-5. N = 897 (NEO), 786 (PID-5), 277 (informant ratings). The highest correlation in each row is in bold.
Three scale pairs showed a highly systematic convergent/discriminant pattern in the NEO analyses and clearly assess the same basic trait dimensions: FI-FFM Trust correlated .78 with NEO Trust; FI-FFM Straightforwardness correlated .63 with NEO Straightforwardness; and FI-FFM Empathy correlated .60 with NEO Altruism. Follow-up significance tests indicated that each of these correlations was significantly higher than all of the other coefficients in its row or column of the heteromethod block (a total of 24 comparisons; all ps < .01, two-tailed). Surprisingly, however, FI-FFM Modesty correlated only .47 with NEO Modesty. Moreover, this coefficient was not significantly greater than those between FI-FFM Modesty and (a) NEO Straightforwardness and (b) NEO Altruism. Consequently, despite the fact that they bear the same name, these two scales are not indicators of the same basic underlying trait. An inspection of their content indicates that both scales contain items reflecting humility and a reluctance to boast about one’s accomplishments. Unlike the FI-FFM scale, however, NEO Modesty also contains multiple reverse-keyed items tapping arrogance and inflated self-esteem (e.g., have a high opinion of self, am better than most people).
Two aspects of the PID-5 data are noteworthy. First, FI-FFM Trust shows a clear convergent/discriminant pattern with PID-5 Suspiciousness (e.g., people are out to get me); the correlation between these scales (−.66) was significantly greater than all nine discriminant comparisons (p < .01, two-tailed). Second, FI-FFM Straightforwardness has strong and nonspecific associations with multiple markers of antagonism. Straightforwardness had particularly strong associations with PID-5 Deceitfulness (r = −.70; e.g., lying comes easily to me), Manipulativeness (r = −.64; e.g., easy for me to take advantage of others), and Callousness (r = −.63; e.g., don’t care if I make others suffer).
All four facet scales displayed moderate self-other agreement, with coefficients ranging from .33 to .40 (mean r = .37). They also exhibited good discriminant validity, as 19 of the 24 follow-up comparisons (79.2%) were significant (p < .05, one-tailed). The only exceptions were that the convergent correlation for Trust was not significantly greater than the discriminant coefficients between self-rated Trust and informant-rated (a) Straightforwardness and (b) Empathy; the convergent coefficient for Straightforwardness did not exceed the correlations between informant-rated Straightforwardness and self-rated (c) Trust and (d) Modesty; and the convergent correlation for Modesty was not higher than the coefficient between self-rated Modesty and informant-rated Straightforwardness.
Openness Facets
Table 8 presents validity data for the three openness facet scales. The top portion of the table reports correlations with the six NEO openness facets; it also includes NEO Excitement-Seeking (a component of extraversion), which was the strongest correlate of FI-FFM Novel Experience Seeking in our data. The middle portion of Table 8 shows correlations with Risk Taking (e.g., people say I am reckless), which was the only PID-5 scale to correlate ≥ |.30| with any openness facet. Finally, the bottom portion of the table presents self-informant agreement correlations for these scales.
Convergent and Discriminant Correlations for the FI-FFM Openness Facet Scales.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; NEO = NEO Personality Inventory; PID-5 = Personality Inventory for DSM-5. N = 897 (NEO), 786 (PID-5), 277 (informant ratings). The highest correlation in each row is in bold.
Excitement-Seeking is a facet of extraversion; all other NEO scales are facets of openness.
The NEO analyses yielded relatively few strong correlations, indicating relatively modest overlap in this domain of personality. FI-FFM Intellectance correlated strongly with both NEO Aesthetics (r = .71) and Ideas (r = .58) and appears to be an amalgam of the content contained in these two scales. Novel Experience Seeking had substantial associations with both NEO Excitement-Seeking (r = .51) and Actions (r = .40). Finally, Nontraditionalism had moderate links to both NEO Values (r = .46) and Fantasy (r = .34).
As noted earlier, Risk Taking was the only PID-5 scale correlating ≥ |.30| with any openness facet. Specifically, it was moderately related to both Novel Experience Seeking (r = .43) and Nontraditionalism (r = .35).
Finally, the openness scales exhibited moderate to strong self-informant agreement, with convergent coefficients ranging from .35 to .59 (mean r = .48). These scales also demonstrated impressive discriminant validity, as 11 of the 12 follow-up comparisons (91.7%) were significant (p < .05, one-tailed). The only exception was that the convergent correlation for Novel Experience Seeking (r = .35) did not exceed the discriminant coefficient between self-rated Nontraditionalism and informant-related Novel Experience Seeking (r = .31).
Supplemental Scales
Supplemental Table S3 presents parallel data for the four supplemental scales. The top portion of the table reports results for all NEO scales correlating ≥ |.40| with any of these scales. Emotional Resonance was strongly related to NEO Feelings (r = .59), and more moderately linked to Warmth (r = .42) and Positive Emotions (r = .41). Dependency was moderately associated (r = .46) with NEO Vulnerability. Finally, neither Unusual Experiences nor Eccentric Beliefs was substantially correlated with any NEO scale.
The middle portion of Table S3 displays results for any PID-5 scale correlating ≥ |.40| with these scales. Unusual Experiences was substantially linked to multiple scales, displaying particularly strong associations with Cognitive and Perceptual Dysregulation (r = .71), Unusual Beliefs and Experiences (r = .64), and Eccentricity (r = .52); Eccentric Beliefs also correlated .52 with PID-5 Unusual Beliefs and Experiences. Emotional Resonance was moderately negatively related (r = −.45) to PID-5 Restricted Affectivity. Finally, Dependency was not substantially related to any PID-5 scale.
Three of the scales—Emotional Resonance, Dependency, and Unusual Experiences—displayed moderate self-informant agreement, with convergent correlations ranging from .33 to .47 (mean r = .39). In contrast, the agreement correlation for Eccentric Beliefs was only .25.
Incremental Validity
Overview
The data reported in Tables 3 through 8 have established a strong level of convergence between the FI-FFM and the NEO, and it is clear that they assess many of the same lower order traits. Moreover, all of the FI-FFM facets correlate at least moderately with one or more NEO scales. These findings raise potential incremental validity concerns. That is, they suggest that the FI-FFM is redundant with existing instruments such as the NEO and, therefore, might not contribute any important new information beyond that already obtainable from current measures. We address these concerns in two series of analyses, taking advantage of the extensive battery of psychopathology measures that was collected in Sample 3. Given that neuroticism shows the strongest and broadest associations with internalizing psychopathology (Kotov, Gamez, Schmidt, & Watson, 2010; Watson & Naragon-Gainey, 2014), we focus here on the incremental validity of the FI-FFM neuroticism facet scales vis-à-vis their counterparts on the NEO.
Table 4 established that three NEO/FI-FFM scale pairs show a very systematic convergent/discriminant pattern and clearly assess the same basic facet traits: FI-FFM Anxiety and NEO Anxiety (r = .74), FI-FFM Depression and NEO Depression (r = .74), and FI-FFM Anger Proneness and NEO Angry Hostility (r = .72). To lessen collinearity concerns and to reduce the number of predictors in these analyses, we standardized these scales and then combined them to form composites. Thus, these analyses examined the individual and joint contributions of eight facet measures: the Anxiety composite, the Depression composite, the Anger composite, FI-FFM Somatic Complaints, FI-FFM Envy, NEO Self-Consciousness, NEO Impulsiveness, and NEO Vulnerability.
In examining associations with self-reported symptoms, we report Pearson correlations (bivariate analyses) and standardized β weights from multiple regression analyses (multivariate analyses). In presenting results with dichotomous diagnoses, we report polyserial correlations (bivariate analyses) and odds ratios from logistic regression analyses (multivariate analyses). Polyserial correlations estimate the linear association between two normally distributed latent continuous variables when one of the observed variables is ordinal and the other is continuous (Flora & Curran, 2004; Olsson, Drasgow, & Dorans, 1982). The disorders were scored as 0 = absent, 1 = present, so that positive correlations indicate that higher scale scores were associated with an increased likelihood of receiving that diagnosis.
Our a priori expectation was that neuroticism facets should be positively related to internalizing psychopathology. In several cases, however, our multivariate analyses revealed evidence of suppressor effects (Gaylord-Harden, Cunningham, Holmbeck, & Grant, 2010; Watson, Clark, Chmielewski, & Kotov, 2013), wherein positive bivariate relations were transformed into significant negative associations (i.e., negative βweights, odds ratios significantly less than 1.00) in the regression results. Given that these suppressor effects were not predicted and are challenging to interpret, we ignore them in interpreting our results and focus exclusively on positive multivariate associations.
Anxiety and Depression
The first series of analyses examined relations with self-report and interview-based indicators of seven depressive and anxiety disorders: major depression, GAD, PTSD, panic disorder, agoraphobia, social phobia, and OCD. Bivariate correlations between the eight neuroticism facet scores and these 14 criteria are presented in Supplemental Table S4. Consistent with previous research, virtually all of these associations were significant, demonstrating again that neuroticism is broadly linked to internalizing psychopathology. At the same time, however, the magnitude of the correlations varied substantially across facets and disorders, ranging from .17 (FI-FFM Envy with diagnoses of agoraphobia) to .74 (Depression composite with self-rated depression symptoms).
Table 9 presents the crucial incremental validity results from the multivariate analyses. The most noteworthy aspect of these data is that the FI-FFM Somatic Complaints scale was the strongest single predictor of anxiety and depression, making a significant incremental contribution in 9 of 14 analyses (64.3%). Furthermore, it showed robust associations with three disorders—PTSD, agoraphobia, and social anxiety/social phobia—that replicated across both self-report symptoms and interview-based diagnoses. In addition, the Depression composite made a significant incremental contribution in seven analyses; the Anxiety composite added significantly in five analyses; the Anger composite and NEO Self-Consciousness both contributed in three analyses; and NEO Impulsiveness produced a significant effect in one analysis. Finally, neither FI-FFM Envy nor NEO Vulnerability demonstrated any evidence of incremental validity in these analyses.
Incremental Validity Analyses: Multivariate Associations Between Neuroticism Facets and Indicators of Depression and Anxiety.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; NEO-PI-3 = NEO Personality Inventory–3; Anx = Anxiety; Dep = Depression; Som = Somatic Complaints; Self = Self-Consciousness; Impul = Impulsiveness; Vuln = Vulnerability; GADQ-IV = Generalized Anxiety Disorder Questionnaire–IV; PTSD = Posttraumatic stress disorder; OCD = obsessive–compulsive disorder; GAD = Generalized anxiety disorder. N = 405 (self-rated symptoms), 402 (diagnoses).
p < .05. **p < .01.
Health Anxiety and Hypochondriasis
The second series of analyses examined the incremental validity of these neuroticism facets in relation to self-rated health anxiety and hypochondriasis. Given the nature of its content, one would expect that FI-FFM Somatic Complaints (e.g., “I have a lot of aches and pains,” “It seems like I get sick more than other people,” “I have a lot of headaches”) would be substantially related to these criteria.
Supplemental Table S5 presents the bivariate associations between the neuroticism facets and eight indictors of health anxiety and hypochondriasis. As in the previous analyses, these associations varied substantially in magnitude, ranging from .05 (NEO Vulnerability with MIHT Perceptual) to .49 (FI-FFM Somatic Complaints with WI Cognitive).
Table 10 presents the incremental validity results from the multiple regression analyses. For our purposes, the most important finding is that FI-FFM Somatic Complaints (four significant effects) and Envy (three significant effects) both displayed substantial incremental validity in these analyses. The Anxiety composite was the strongest individual predictor in these data, contributing significantly in six of eight cases. In addition, the Anger composite added significant variance in four analyses, whereas the Depression composite, NEO Self-Consciousness, and NEO Impulsiveness each showed two significant effects. Finally, NEO Vulnerability again failed to demonstrate any incremental validity in these analyses.
Incremental Validity Analyses: Multivariate Associations (Standardized β Weights) Between Neuroticism Facets and Self-Report Measures of Hypochondriasis/Health Anxiety.
Note. FI-FFM = Faceted Inventory of the Five-Factor Model; NEO-PI-3 = NEO Personality Inventory–3; Anx = Anxiety; Dep = Depression; Som = Somatic Complaints; Self = Self-Consciousness; Impul = Impulsiveness; Vuln = Vulnerability; WI = Whiteley Index; HAQ = Health Anxiety Questionnaire; MIHT = Multidimensional Inventory of Hypochondriacal Traits. N = 406.
p < .05. **p < .01.
General Discussion
Basic Properties of the FI-FFM Scales
We have presented data establishing three basic properties of the FI-FFM scales. First, consistent with their factor analytic origin, the scales are internally consistent. Across the three samples, the domain scales had coefficient alphas ranging from .83 to .96, with a median value of .91; corresponding values for the facet scales ranged from .73 to .92, with a median coefficient of .82. Second, the FI-FFM scales generally displayed strong retest reliability over a 2-week interval. The five domain scales had dependability coefficients ranging from .78 to .80 (median r = .79); retest correlations for the 22 facet scales ranged from .64 to .82 (median r = .75). Third, self-reports on the scales showed substantial convergent and discriminant validity in relation to corresponding ratings made by well-acquainted informants. Convergent correlations for the domain scales ranged from .46 to .62, with a median value of .53; moreover, all 40 tests of discriminant validity were significant. Convergent coefficients for the facet scales ranged from .26 to .66, with a median coefficient of .40. Furthermore, 127 of 156 tests of discriminant validity (81.4%) were significant. Overall, these data provide strong evidence of reliability and validity.
Convergence With Other Measures
Relations With the BFI and NEO-PI-3
The FI-FFM domain scales also exhibited impressive convergent and discriminant validity vis-à-vis corresponding scores on the BFI and the NEO. Convergent correlations with the BFI scales ranged from .71 to .83 (mean r = .77); corresponding values with the NEO domain scales ranged from .73 to .82 (mean r = .78). Moreover, all 80 tests of discriminant validity were significant across the two sets of analyses.
Furthermore, our analyses established that many FI-FFM facets assess the same basic traits as corresponding indicators within the NEO. Specifically, we identified 14 corresponding facet pairs: three within neuroticism (FI-FFM Anxiety vs. NEO Anxiety, FI-FFM Depression vs. NEO Depression, FI-FFM Anger Proneness vs. NEO Angry Hostility); three in extraversion (FI-FFM Ascendance vs. NEO Assertiveness, FI-FFM Sociability vs. NEO Gregariousness, FI-FFM Venturesomeness vs. NEO Excitement-Seeking); five within conscientiousness (the Self-Discipline, Achievement Striving, Order, Dutifulness, and Deliberation scales from both inventories); and three in Agreeableness (FI-FFM Trust vs. NEO Trust, FI-FFM Straightforwardness vs. NEO Straightforwardness, FI-FFM Empathy vs. NEO Altruism). In addition, two FI-FFM scales appear to represent blends of multiple NEO traits. FI-FFM Positive Temperament correlated .58 with NEO Positive Emotions, .55 with NEO Activity, and .50 with NEO Warmth. Similarly, FI-FFM Intellectance appears to be an amalgam of NEO Aesthetics (r = .71) and Ideas (r = .58).
Relations With the PID-5
The FI-FFM scales also showed strong and systematic associations with the PID-5, which assesses the alternative model of personality disorder in Section III of DSM-5 (American Psychiatric Association, 2013). Similar to the NEO, our analyses revealed that 11 FI-FFM facets assess the same basic traits (in most cases, defining the opposite end of the dimension) as corresponding scales within the PID-5: three within neuroticism (FI-FFM Anxiety vs. PID-5 Anxiousness, FI-FFM Depression vs. PID-5 Depressivity, FI-FFM Anger Proneness vs. PID-5 Hostility); four in extraversion (FI-FFM Ascendance vs. PID-5 Attention Seeking, FI-FFM Sociability vs. PID-5 Withdrawal, FI-FFM Venturesomeness vs. PID-5 Risk Taking, FI-FFM Positive Temperament vs. PID-5 Anhedonia); three within conscientiousness (FI-FFM Self-Discipline vs. PID-5 Distractibility, FI-FFM Deliberation vs. PID-5 Impulsivity, FI-FFM Dutifulness vs. PID-5 Irresponsibility); and one in Agreeableness (FI-FFM Trust vs. PID-5 Suspiciousness). In addition, FI-FFM Straightforwardness was substantially correlated with multiple markers of PID-5 Antagonism, displaying particularly strong associations with Deceitfulness (r = −.70), Manipulativeness (r = −.64), and Callousness (r = −.63).
Distinctive Properties of the FI-FFM
Although the FI-FFM clearly overlaps substantially with other hierarchical measures of the FFM, it also contains some distinctive traits that are not assessed in existing instruments. In this regard, we conducted incremental validity analyses that established that the FI-FFM Somatic Complaints and Envy scales contain important information not already obtainable from the NEO-PI-3. We conducted a total of 22 tests of incremental validity across two sets of analyses. It is noteworthy that FI-FFM Somatic Complaints emerged as the strongest single predictor in these analyses, making a significant incremental contribution in 14 cases (63.6%); the Anxiety composite was next, adding significantly in 11 instances (50%). The FI-FFM Envy scale was less impressive, but still produced three significant effects (13.6%) in these analyses.
Based on the data we have presented, we believe the FI-FFM represents a useful complement to existing hierarchical FFM measures. It should be noted that these hierarchical measures vary substantially both in terms of their length and in their assessment of the lower order facets. For instance, the BFAS (100 items) includes two component traits per domain; the BFI-2 (60 items) contains three facets per domain; and the NEO-PI-3 (240 items) has six facets per domain. The FI-FFM (207 items) contains 3 to 5 facets per domain, and thus would be particularly appealing to those seeking more differentiated assessment than is provided by instruments such as the BFI-2 and BFAS. Given the number of items it contains, the FI-FFM also takes longer to complete than measures such as the BFI-2 and BFAS. However, the FI-FFM is modular in form, such that investigators facing significant time constraints can select and assess only those scales that are most relevant to their research. For instance, researchers who primarily are interested in a differentiated assessment of neuroticism can use only the scales from that domain.
Clarifying the Lower Level of the Hierarchy
As we discussed in the Introduction, researchers increasingly are recognizing the value of studying personality at the specific facet level (e.g., Soto & John, 2016; Watson et al., 2015b). Interested researchers now have access to multiple hierarchical instruments that assess FFM domains and facets. The existence of these multiple instruments provides both rich research opportunities and important challenges. We use extraversion—which currently has the best-understood facet structure of any FFM domain—to illustrate these points.
As noted earlier, Watson et al. (2015b) conducted exploratory factor analyses using the extraversion facet scales from the NEO-PI-3, FI-FFM, and HEXACO-PI-R. Replicating previous results (e.g., Naragon-Gainey et al., 2009), these structural analyses revealed four underlying facet factors, which Watson et al. (2015b) labeled Positive Emotionality, Sociability, Assertiveness, and Experience Seeking. Table 11 provides a schematic placement of the individual trait scales in this four facet structure. Note that NEO-PI-3 Warmth split between the Positive Emotionality and Sociability factors, so it is listed as a marker of both dimensions.
Scales Defining Extraversion Facet Factors.
Note. NEO-PI-3 = NEO Personality–3; FI-FFM = Faceted Inventory of the Five-Factor Model; HEXACO-PI-R = HEXACO Personality Inventory–Revised; BFI-2 = Big Five Inventory–2; BFAS = Big Five Aspect Scales.
In addition, based on data presented by Soto and John (2016; see especially their Table 1) and an inspection of item content, Table 11 also places the lower order BFI-2 and BFAS extraversion scales within this four-factor scheme. Note that BFAS Enthusiasm contains content relevant to both Positive Emotionality (e.g., have a lot of fun, laugh a lot) and Sociability (e.g., make friends easily, warm up quickly to others) and so has been listed on both factors.
Table 11 establishes that there are important similarities in the assessment of extraversion across these five instruments. Most notably, they all include at least some specific content relevant to positive emotionality, sociability, and assertiveness. This, in turn, facilitates future facet-level research. For instance, following the approach used by Watson et al. (2015b), researchers can use multiple indicators to define facets at the latent factor level. Thus, indicators such as NEO-PI-3 Gregariousness, FI-FFM Sociability, HEXACO-PI-R Sociability, and BFI-2 Gregariousness can be used to define a latent Sociability factor; similarly, scales such as NEO-PI-3 Assertiveness, FI-FFM Ascendance, HEXACO-PI-R Social Boldness, BFI-2 Assertiveness, and BFAS Assertiveness can be assessed as markers of an underlying Assertiveness factor.
At the same time, however, Table 11 highlights the fact that these instruments all carve up the domain somewhat differently. For instance, the NEO-PI-3, HEXACO-PI-R, and BFI-2 all appear to emphasize the communal aspects of the domain, whereas the FI-FFM provides more differentiated assessment of its agentic component (for a discussion of this distinction, see Ansell & Pincus, 2004). More important, only two instruments—the NEO-PI-3 and FI-FFM—provide any assessment of experience seeking.
Depending on one’s purpose, these assessment differences may or may not be important. However, it cannot be assumed in advance that they are trivial in nature. After all, the growing interest in lower level traits reflects the fact that facets often show highly distinctive correlates. For example, Watson et al. (2015b) demonstrated that these four extraversion factors had very different associations with psychopathology. Positive Emotionality had strong and specific negative associations with indicators of depression. Sociability also was negatively related to psychopathology, showing particularly strong associations with indicators of social dysfunction and negative symptoms of schizotypy. Assertiveness generally had weak associations with psychopathology, but was negatively related to social anxiety and was positively correlated with some forms of externalizing. Finally, Experience Seeking had substantial positive associations with a broad range of indicators related to externalizing and bipolar disorder. Consequently, researchers can expect to obtain somewhat different results depending on exactly how they model personality at the lower order level.
In light of these nontrivial differences, researchers need to consider carefully how best to assess personality at the lower order level. More fundamentally, future research should explicate the lower order level of the personality hierarchy to clarify the optimal assessment of each FFM domain.
Limitations and Future Directions
We have presented a broad range of reliability and validity data, reporting results based on self-reports, informant ratings, and interview-based diagnoses. These findings are very encouraging and indicate that the FI-FFM scales provide reliable and valid assessment of both higher order and lower order personality traits. Validity is a complex and ongoing process, however, so that additional research is needed to explicate the construct validity of the instrument more fully. We already have noted, for example, that further data are needed to clarify the short-term dependability of the FI-FFM scales. Of course, it also will be important to examine the stability of these scales over much longer time spans.
In addition, the FI-FFM scales need to be validated against a broader array of variables. For example, our diagnostic assessment of psychopathology was rather limited, as we did not have interview measures of the DSM personality disorders. Personality disorder data would be extremely helpful in establishing the criterion and incremental validity of several FI-FFM scales. For instance, the DSM-5 diagnosis for narcissistic personality disorder contains the criteria “Has a grandiose sense of self-importance (e.g., exaggerates achievements and talents, expects to be recognized as superior without commensurate achievements” and “Is often envious of others or believes others are envious of him or her.” Consequently, narcissistic personality disorder data would be very informative in validating the FI-FFM Modesty and Envy scales.
In spite of these limitations, we believe the FI-FFM is a very promising instrument that provides comprehensive assessment of FFM traits at both the higher order and lower level. We hope that its availability stimulates further work on the FFM traits, particularly at the specific facet level.
Footnotes
Appendix A
The Faceted Inventory of the Five-Factor Model (FI-FFM)
Copyright © 2017 by E. Nus, D. Watson, & K. D. Wu
Appendix B
FI-FFM Supplemental Scales
Copyright © 2017 by E. Nus, D. Watson, & K. D. Wu
Acknowledgements
We thank Lee Anna Clark, Patrick Cruitt, Stephanie Ellickson-Larew, Mark Godding, Haley Heibel, Brittany Katz, Ana Kent, Katie Kraemer, Mallory Meter, Eunyoe Ro, John Souter, Kasey Stanton, Sara Stasik-O-Brien, Nadia Suzuki, and Elizabeth Yahiro for their help in the preparation of this article.
Authors’ Note
Ericka Nus formerly was Ericka Simms.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
Supplemental Material
Supplementary material for this article is available online.
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.
