The Shedler–Westen Assessment Procedure (SWAP): Making Personality Diagnosis Clinically Meaningful

Abstract

There is a schism between science and practice in understanding and assessing personality. Approaches derived from the research laboratory often strike clinical practitioners as clinically naíve and of dubious clinical relevance. Approaches derived from clinical observation and theory often strike empirical researchers as fanciful speculation. In this article, we describe an approach to personality designed to bridge the science–practice divide. The Shedler–Westen Assessment Procedure (SWAP; Shedler & Westen, 2004a, 2004b; Westen & Shedler, 1999a, 1999b) is an empirically rigorous diagnostic method that preserves the richness and complexity of clinical case description. In this article, we describe its use in diagnosis, case conceptualization, and treatment planning. We review evidence for reliability, validity, and clinical utility. Finally, in the article, we present a system for personality diagnosis, as an alternative to Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; American Psychiatric Association, 2000) Axis II, that is empirically grounded, clinically relevant, and practical for routine use in both clinical and research contexts.

ACKNOWLEDGMENT

This article was adapted with permission from material previously published in J. Shedler & D. Westen, D., “Personality diagnosis with the Shedler–Westen Assessment Procedure (SWAP): Bridging the gulf between science and practice,” in Psychodynamic Diagnostic Manual (PDM), by Alliance Task Force (Ed.), 2006, Silver Spring, MD: Alliance of Psychoanalytic Organizations. Copyright 2006 by Alliance Task Force.

Notes

¹Poor test–retest reliability has led some researchers to suggest that PDs are less stable than previously believed. Such an interpretation of the data seems inconsistent with the observations of virtually all clinical theorists. A more viable hypothesis may be that the assessment instruments do not capture core features of personality that are salient to clinicians who treat patients with PDs and know them well. Specifically, the instruments may overemphasize transient behavioral symptoms (such as self-cutting and suicidality in borderline patients, which may emerge only when an attachment relationship is threatened) and underemphasize underlying personality processes that endure over time (such as affect dysregulation and feelings of emptiness and self-loathing in borderline patients).

²One way it does so is by ensuring that raters are “calibrated” with one another. Consider the situation with rating scales, in which raters can use any value as often as they wish. Inevitably, certain raters will tend toward extreme values (e.g., values of 0 and 7 on a 0–7 scale) and others will tend toward middle values (e.g., values of 4 and 5). Thus, the scores reflect not only the characteristics of the patients but also the calibration of the raters. The Q-sort method, with its fixed distribution, eliminates this kind of measurement error because all clinicians must assign each score the same number of times. If use of a standard item set gives clinicians a common vocabulary, use of a fixed distribution can be said to give them a “common grammar” (Block, 1961/1978).

³The material in this section is adapted from Lingiardi, Shedler, and Gazizllo (2006). Please see the original publication for a more complete description of the case, treatment methods, and findings.

⁴The relatively low thresholds reflect the fact that the reference sample consisted of patients with PD diagnoses. Thus, a T score of 50 indicates average functioning among patients with PD diagnoses, and a T score of 60 represents an elevation of 1 SD relative to other patients with PD diagnoses.

⁵The material presented here is adapted from Westen and Weinberger (2004) and updated.

⁶All coefficients are Pearson's r.

⁷The material presented here is adapted from Westen and Shedler (1999b) and updated.

⁸A discussion of person-centered versus variable-centered assessment is beyond the scope of this article and warrants a paper in its own right. We believe the distinction underlies much misunderstanding between clinicians and researchers because clinicians tend to think in person-centered terms and researchers tend to think in variable-centered terms. The choice of a person- or variable-centered approach, which may profoundly affect how we think about psychological issues, is often not even recognized as a choice. Instead, one or the other approach is accepted by convention and without consideration of what is at stake. (The fact that SPSS is designed to manipulate variables rather than cases may have shaped academic psychology in ways we can barely fathom). It is not that one approach is “right” and one is “wrong” but rather that they serve different purposes and draw our attention to different matters. Good assessment systems are like good maps, in that they must accurately depict the territory. However, sometimes one wants a roadmap, sometimes a map of elevations, and sometimes a political map. The motorist trying to navigate the interstate will have little interest in a map of elevations irrespective of the number of studies documenting its reliability and validity.