7b
Martin Sellbom
Miller and Widiger (this volume) have provided an impressive analysis of the conceptualization and diagnostic operationalization of personality disorders (PDs) from two similar but distinct five-factor models of personality: the Five-Factor Model of personality (FFM) and the alternative DSM-5 model for personality disorders (AMPD; American Psychiatric Association [APA], 2013) trait model (Criterion B in the DSM-5 Section III). They make a persuasive argument that the field of PD is ready to adopt a trait model for formal diagnostic operationalization in light of the extensive evidence for both construct validity and clinical utility.
I do not take any significant issue with Miller and Widiger’s scholarly analysis and I agree with their general perspective and contention that the field of PD diagnosis is ready for a well-established trait model. Although I am not fully convinced that the FFM or the AMPD trait models must represent the actual trait operationalizations for PD diagnosis, I will not argue against their use here. Rather, in this commentary, I will discuss a few issues of importance as the field moves forward in this regard, most of which center on clinical application.
What Level of Abstraction Is Necessary?
The FFM and AMPD trait models are viewed as five-factor models owing to their higher order structures representing five broad domains. However, most of the research and data presented in Miller and Widiger (this volume) are not really focused on that level of abstraction. In a way, these can be considered 30 factor or 25 factor models, respectively, given that each of the PDs are considered within this level of abstraction. Although it is clear that a five-factor level of abstraction of personality and personality psychopathology has garnered impressive evidence as a higher order structure (e.g., Markon, Krueger, & Watson, 2005), the evidence is less clear at the facet level. Krueger and colleagues (Krueger, Derringer, Markon, Watson, & Skodol, 2012), for instance, started with 37 facet traits and eventually decided on 25 through a number of iterative factor analyses. Costa and McCrae (1992) decided on six rational facets underlying each revised NEO Personality Inventory (NEO PI-R) domain. Other extensive efforts to capture abnormal-range personality from a dimensional trait perspective with promising validity have yielded 15 (Schedule for Nonadaptive and Adaptive Personality, 2nd ed. [SNAP-2]; Clark, Simms, Wu, & Casillas, 2007), 18 (Dimensional Assessment of Personality Pathology – Basic Questionnaire [DAPP-BQ]; Livesley & Jackson, 2009), and 33 (Computer Adaptive Test for Personality Disorders [CAT-PD]; Simms et al., 2011) facets.
Psychometric properties of measurement aside, these various models are not distinct per se and can be identified in one another (e.g., Markon et al., 2005). Although some experts would likely argue that the selected trait model does not matter as long as they have sufficient validity and utility in capturing PDs, it would behoove the field to arrive at some degree of consensus about a level of abstraction optimal for PD. Also, the smaller the set of trait facets necessary to capture PD variance, the easier and less labor-intensive the process will become. As the field moves away from its attempt to retrofit fallacious DSM-IV/DSM-5 PD categories, for which facets are clearly needed for optimal distinction, perhaps description of personality even at the domain level might suffice? This is an empirical question that is still unanswered but, in my opinion, would serve to advance the field.
Are Trait Models Ready for Use in Clinical Assessment?
There are a number of issues that need to be resolved before trait models can be fully implemented in clinical practice for diagnostic purposes. I highlight only a few pertinent ones here.
What Constitutes the Presence of a Maladaptive Trait?
DSM-5 Section III calls for the presence of maladaptive traits (for Criterion B) in the diagnosis of personality disorder in the AMPD. They describe these traits as dimensional entities but provide for no clarity as to how presence of such traits would be indicated. In general, while extreme levels of traits are necessary for the maladaptive range, I do not believe the field has arrived at a consensus on how such elevations or extremities would be determined (Al-Dajani, Gralnick, & Bagby, 2016; Samuel, Hopwood, Krueger, Thomas, & Ruggero, 2013). In the assessment of Intellectual Disability (ID), for instance, the presence of extremely low intelligence coupled with associated impairment in functioning is required. And indeed, for ID, there is clear operational guidance in terms of what is considered extremely low intelligence (typically two standard deviations below the mean, but these thresholds have loosened somewhat [APA, 2013]). Of course, the same principles can (and, in my opinion, should) be applied to personality disorder diagnosis as well, but unlike for ID, the optimal threshold for maladaptivity still needs to be determined. Many common personality assessment instruments (e.g., Minnesota Multiphasic Personality Inventory – 2 – Restructured Form [MMPI-2-RF; Ben-Porath & Tellegen, 2008], Personality Assessment Inventory [PAI; Morey, 2007]) use standardized scores of 1.5 to 2 SDs above a normative mean to indicate a clinical elevation, but these levels have rarely been directly examined for diagnostic decision-making. Clinical judgment is an alternative (i.e., a clinician decides on whether certain maladaptive traits are present, which is how PD diagnosis is typically currently assigned), but it is imprecise. Such judgment alone would also be viewed as unacceptable in ID assessment, so why should psychopathology assessment broadly be held to a lower standard? After all, dimensional trait models are tailored for a more quantitative approach.
Is There an Assessment Device Already Available?
There is one five-factor assessment device currently available for measuring the FFM that meets most agreed upon Standards for Educational and Psychological Testing (see American Psychological Association, American Educational Research Association, & National Council on Measurement in Education [APA/AERA/NCME], 2014): the NEO PI-3 (Costa & McCrae, 2010).1 However, Miller and Widiger have seemingly moved away from the NEO instruments as acceptable for this purpose, as their scale scores do not have a sufficiently maladaptive range. Instead, Miller, Widiger, Lynam and their colleagues have developed eight promising FFM tools for the majority of PDs that do have a greater range of maladaptivity (see Miller & Widiger’s [this volume] review). In my view, these instruments do indeed have promising psychometric properties, but far more work is necessary across a range of settings before they can be applied clinically. Furthermore, the AMPD trait model is directly associated with the Personality Inventory of DSM-5 (PID-5; Krueger et al., 2012) – a self-report inventory which has amassed an extensive research base (e.g., Al-Dajani et al., 2016).2 Even so, the PID-5 is also not a current alternative for clinical practice, because it does not meet current standards for psychological testing (APA/AERA/NCME, 2014). More specifically, it lacks a sufficient normative sample, has no test manual guiding its use, and does not have formal measures of response bias (see Al-Dajani et al., 2016, for a detailed review), though the latter situation is currently being rectified (see later section). Any viable clinical alternative should meet these recommendations prior to formal clinical use.
Need for test manuals. The Standards for Educational and Psychological Testing (APA/AERA/NCME, 2014) state that psychological tests should have a manual that provides clear instructions and articulated rationale for test administration, scoring, and interpretation. Such manuals are not currently available for most five-factor PD instruments that assess dimensional personality traits, but are needed for widespread clinical application. For instance, one of the factors that are evaluated in court expert testimony is whether techniques upon which experts rely to inform their opinions have formal guides for their administration and use (Daubert v. Merrell Dow Pharmaceuticals, Inc., 1993).
Normative referencing. Dimensional constructs lend themselves well to normative referencing, but the field needs to determine the most appropriate reference group (usually a representative community sample) to which a test taker’s scores should be compared. Most common clinical personality inventories (e.g., MMPI-2-RF, PAI) and standard intelligence tests rely on normative referencing, which allows for the calculation of standardized scores. The FFM-PD measures and PID-5, for instance, do not have formal normative samples. It will therefore behoove dimensional PD assessment researchers to either select instruments with a large representative normative sample or generate such samples for existing or new measures before a psychometric approach to determining maladaptivity in traits can be achieved. If other types of norming are to be preferred (e.g., criterion-based referencing), then such need to be articulated as well, along with sound empirical justification.
Measures of non-credible responding. Self-report inventories are susceptible to response bias, which can have tremendous effects on both the observed scores and their psychometric validity if left unmeasured (e.g., Dhillon, Bagby, Kushner, & Burchett, 2017). Response bias is particularly common in forensic settings (e.g., Ardoff, Denney, & Houston, 2007) and it is probably not a preposterous suggestion that individuals with personality pathology might be apt to mischaracterize themselves either intentionally or unintentionally (e.g., poor insight). The most common clinical assessment inventories (e.g., MMPI-2-RF, PAI) have established validity scales to assess for non-credible responding, with a range of validation studies to support their use.
Some efforts to assess response bias are underway. Some FFM-PD measures (e.g., Elemental Psychopathy Assessment; Lynam et al., 2011) have validity scales. The PID-5 does not have formal validity scales, though experimental versions have been published that assess both inconsistent responding (Keeley, Webb, Peterson, Roussin, & Flanagan, 2016) and over-reporting (Sellbom, Dhillon, & Bagby, 2018) with promising utility. I am encouraged that some PD assessment scholars consider this an important issue, but any validity scales for newer measures need to be extensively validated before widespread use. The MMPI-2-RF Validity Scales, for instance, have over 70 studies supporting their use in a variety of contexts (e.g., Sellbom, 2019).
Superiority over existing clinical measures. Finally, although multi-scale clinical assessment instruments like the MMPI-2-RF and PAI are not directly designed to measure the trait facets as articulated in contemporary models, such as the FFM or AMPD, they nevertheless capture the relevant variance (e.g., Anderson et al., 2015). The MMPI-2-RF, in particular, has the Personality Psychopathology Five (PSY-5) scales, which are both conceptual and empirical cognates of the PID-5 domain scales (Anderson et al., 2013, 2015). Given how well-established such instruments are in clinical practice, and that they meet the recommendations for standard psychological tests (APA/AERA/NCME, 2014), it is important that newer measures for clinical use also demonstrate some superiority over these well-known alternatives in the assessment of PD diagnosis (Al-Dajani et al., 2016). Furthermore, other measures such as the DAPP-BQ and the SNAP-2, which might not have garnered the same clinical attention as the MMPI-2-RF and PAI, still assess established dimensional trait models, are further developed in terms of Standards for Educational and Psychological Testing (APA/AERA/NCME, 2014) guidelines (especially DAPP-BQ), and should therefore also at least be considered as viable alternatives (but see Krueger & Markon [2014] for why these trait model measures might not be sufficient conceptually).
Conclusions
Miller and Widiger (this volume) have authored a very illuminating and persuasive chapter with impressive evidence to support their main argument for considering five-factor models in PD diagnosis. I agree with them that this direction is important for the field. But, as I articulated, there remain important areas of scientific inquiry and applied assessment developments before we can fully realize these models’ clinical use. I believe that such scholarship is in progress and I look forward to the field’s development in this regard.
References
Al-Dajani, N., Gralnick, T. M., & Bagby, R. M. (2016). A psychometric review of the Personality Inventory for DSM–5 (PID–5): Current status and future directions. Journal of Personality Assessment, 98, 62–81.
American Educational Research Association, American Psychological Association, & National Council on Measurement in Education. (2014). Standards for Educational and Psychological Testing. Washington, DC: American Educational Research Association.
American Psychiatric Association. (2013). Diagnostic and Statistical Manual of Mental Disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.
Anderson, J. L., Sellbom, M., Ayearst, L., Quilty, L. C., Chmielewski, M., & Bagby, R. M. (2015). Associations between DSM-5 Section III personality traits and the Minnesota Multiphasic Personality Inventory-2-Restructured Form (MMPI-2-RF) scales in a psychiatric patient sample. Psychological Assessment, 27, 801–815.
Anderson, J. L., Sellbom, M., Bagby, R. M., Quilty, L. C., Veltri, C. O. C., Markon, K. E., & Krueger, R. F. (2013). On the convergence between PSY-5 domains and PID-5 domains and facets: Implications for assessment of DSM-5 personality traits. Assessment, 20, 286–294.
Ardoff, B. R., Denney, R. L., & Houston, C. M. (2007). Base rates of negative response bias and malingered neurocognitive dysfunction among criminal defendants referred for neuropsychological evaluation. Clinical Neuropsychologist, 21, 899–916.
Ben-Porath, Y., & Tellegen, A. (2008/2011). Minnesota Multiphasic Personality Inventory-2 Restructured Form: Manual for Administration, Scoring, and Interpretation. Minneapolis: University of Minnesota Press.
Clark, L. A., Simms, L. J., Wu, K. D., & Casillas, A. (2007). Manual for the Schedule for Nonadaptive and Adaptive Personality–2nd Edition (SNAP-2). South Bend, IN: Author.
Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO PI-R) and NEO Five-Factor Inventory (NEO-FFI) Professional Manual. Odessa, FL: Psychological Assessment Resources.
Costa, P. T., Jr., & McCrae, R. R. (2010). NEO Personality Inventory-3 (NEO PI-3) and NEO Five-Factor Inventory-3 (NEO-FFI-3) Professional Manual. Odessa, FL: Psychological Assessment Resources.
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993).
Dhillon, S., Bagby, R. M., Kushner, S. C., & Burchett, D. (2017). The impact of underreporting and overreporting on the validity of the Personality Inventory for DSM-5 (PID-5): A simulation analog design investigation. Psychological Assessment, 29, 473.
First, M. B., Skodol, A. E., Bender, D. S., & Oldham, J. M. (2014). Structured Clinical Interview for the DSM-5 Alternative Model for Personality Disorders (SCID–AMPD). New York: New York State Psychiatric Institute.
Keeley, J. W., Webb, C., Peterson, D., Roussin, L., & Flanagan, E. H. (2016). Development of a response inconsistency scale for the Personality Inventory for DSM–5. Journal of Personality Assessment, 98, 351–359.
Krueger, R. F., Derringer, J., Markon, K. E., Watson, D., & Skodol, A. V. (2012). Initial construction of a maladaptive personality trait model and inventory for DSM-5. Psychological Medicine, 42, 1879–1890.
Krueger, R. F., & Markon, K. E. (2014). The role of the DSM-5 personality trait model in moving toward a quantitative and empirically based approach to classifying personality and psychopathology. Annual Review of Clinical Psychology, 10, 477–501.
Livesley, W. J., & Jackson, D. N. (2009). DAPP–BQ: Dimensional Assessment of Personality Pathology–Basic Questionnaire. Port Huron, MI: Sigma Press.
Lynam, D. R., Gaughan, E. T., Miller, J. D., Miller, D. J., Mullins-Sweatt, S., & Widiger, T. A. (2011). Assessing the basic traits associated with psychopathy: Development and validation of the Elemental Psychopathy Assessment. Psychological Assessment, 23, 108–124.
Markon, K. E., Krueger, R. F., & Watson, D. (2005). Delineating the structure of normal and abnormal personality: An integrative hierarchical approach. Journal of Personality and Social Psychology, 88, 139–157.
Morey, L. C. (2007). Personality Assessment Inventory Professional Manual (2nd ed.). Odessa, FL: Psychological Assessment Resources.
Samuel, D. B., Hopwood, C. J., Krueger, R. F., Thomas, K. M., & Ruggero, C. J. (2013). Comparing methods for scoring personality disorder types using maladaptive traits in DSM-5. Assessment, 20, 353–361.
Sellbom, M. (2019). The MMPI-2 Restructured Form (MMPI-2-RF): Assessment of personality and psychopathology in the 21st century. Annual Review of Clinical Psychology, 15, 149–177.
Sellbom, M., Dhillon, S., & Bagby, R. M. (2018). Development and validation of an Overreporting Scale for the Personality Inventory for DSM-5 (PID-5). Psychological Assessment, 30(5), 582–593.
Simms, L. J., Goldberg, L. R., Roberts, J. E., Watson, D., Welte, J., & Rotterman, J. H. (2011). Computerized adaptive assessment of personality disorder: Introducing the CAT-PD project. Journal of Personality Assessment, 93, 380–389.
1In this section, I emphasize instruments developed for the five-factor models Miller and Widiger discussed. There are other dimensional trait measures with a lower order structure (e.g., DAPP-BQ and SNAP-2) that are likely further along with respect to meeting the criteria I highlight, but might not be optimal for other reasons (e.g., Krueger & Markon, 2014).
2There is also a structured clinical interview available for the AMPD (First, Skodol, Bender, & Oldham, 2014), but I cannot find a single study in which it has been used, and therefore do not view it as a viable alternative at this time.