Learning from unreliability: the importance of inconsistency in coping dynamics Carolyn E. Schwartz a, b, *, Lawren H. Daltroy c, d a Frontier Science and Technology Research Foundation, 1244 Boylston Street, Suite 303, Chestnut Hill, MA 02467, USA b Department of Psychiatry, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA c Multipurpose Arthritis and Musculoskeletal Diseases Center, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA d Department of Health and Social Behavior, Harvard School of Public Health, Boston, MA, USA Abstract The role of response stability in the measurement of coping is examined with a focus on the unique information that can be gleaned from low test±retest reliability (`inconsistency'). Data from two studies are presented in which a card sort measure of coping ¯exibility was used on people with three dierent chronic diseases and the elderly (n = 219). We begin by testing the hypothesis that the low stability re¯ects unreliability due to measurement artifacts, such as random error, low ecological validity, long test±retest interval, surrogate assistance, or error due to completing the questionnaire in multiple sittings. Our ®ndings suggest that surrogate assistance in completing questionnaires was the only measurement artifact associated with low stability. We then tested the proposition that low stability re¯ects a genuine behavior pattern (i.e. inconsistency). Hierarchical modeling revealed that measurement artifact accounted for less than one percent of the variance in inconsistency in reported coping behavior and that an additional 21% of the variance could be explained by the behavioral factors, including neuropsychological problems (9%), psychological morbidity (4%), locus of control (3%) and eudaimonistic well-being (5%). Thus inconsistency in reported coping behavior was better explained by behavioral and psychosocial factors than by the tested measurement artifacts. We conclude that inconsistency in reported coping behavior does indeed re¯ect a meaningful behavior pattern, rather than simply measurement artifact. # 1999 Elsevier Science Ltd. All rights reserved. 1. Introduction Nothing endures but change Heraclitus, ca. 540±ca. 480 B.C. The more things change, the more they remain the same Ahphonse Karr, 1808±1890. Traditional psychometrics suggests that a good measurement instrument must be reliable. However, when one is trying to measure constructs which are dynamic in nature, measurement reliability may be attained at the cost of losing other valuable infor- mation and may actually re¯ect an overly simplistic model of the relationships of interest. Thus, measure- ment stability may be encouraging when one attempts to measure qualities which are unlikely to change (e.g. personality traits), but discouraging when one attempts to observe and describe dynamic processes (Meehl, 1973). Coping is such a process, as it requires a dynamic responsiveness to situational challenges. This paper attempts to further our understanding of the role of response stability in the measurement of coping by examining the unique information that can be gleaned from low test±retest reliability in a coping measure. Social Science & Medicine 48 (1999) 619±631 0277-9536/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0277-9536(98)00353-0 PERGAMON * Corresponding author. Fax: +1-617-632-2001; e-mail: schwartz@jimmy.harvard.edu.