To our knowledge, this is the first study to examine whether discrepancies in self and informant-reported personality traits are associated with patient illness burden or cognitive status. Disease amounting to very severe burden in a single organ system was associated with informants over-reporting particular traits by approximately 0.3 SD relative to self-report. These discrepancies have substantial ramifications for individual patients, given that older adults commonly show non-trivial degrees of illness burden as a result of aging-relate chronic diseases.
Providing partial support for the hypotheses, illness burden was associated with self-informant discrepancies in Openness domain scores. Hypotheses about discrepancies in Extraversion and Conscientiousness domain scores were not supported, but there was an association between illness burden and an Extraversion facet, Positive Emotions. Illness burden was also associated with facet-level discrepancies for Openness to Aesthetics, Feelings, and Actions. These findings indicate that, in comparison to informants’ perceptions, older depressed persons with greater illness burden perceived themselves as experiencing higher levels of positive emotion, more interested in aesthetics and inner emotional experiences, and more behaviorally flexible. These findings are consistent with research showing that observers overestimate the potential impact of chronic illness and disability upon patient quality of life (Martire et al., 2006
). The lack of association between illness burden and Conscientiousness ratings suggests that informant beliefs about patient goal-directedness (achievement-striving, deliberation) or judgments of behaviors associated with Conscientiousness may be unaffected by patient illness burden.
As hypothesized, patient cognitive function was related to self-informant discrepancies for the Openness domain. Facet level analyses revealed that decreased cognitive function was associated with informants overreporting patient Openness to Actions and Values, relative to self-report. These traits involve behavioral and ideological flexibility, respectively (Costa & McCrae, 1992
). Whereas informants may overestimate the impact of illness burden upon patients’ internal emotional lives, they appear to underestimate the potential impact of cognitive decline upon patient behavioral and ideological flexibility.
Findings concerning the influence of patient illness burden and cognitive function upon observer ratings are consistent Funder’s (1995)
Realistic Accuracy Model, which indicates that the availability of trait information moderates self-informant agreement in personality ratings. Specifically, illness burden and deficits in cognitive function decrease the availability of trait information and were associated with more discrepant ratings. When adequate trait information is unavailable or uncertain, informants may rely upon stereotypes or personal assumptions about disease when making trait ratings, just as patients themselves overestimate the impact of disease when forecasting their future adjustment (Ubel, Loewenstein, Schwarz, & Smith, 2005
Whereas cognitive theories of depression suggest that self-perceptions are unduly negative (Beck, 1976
), the depressed older adults in our sample lacked a systematic negative bias in their personality ratings. Instead, they saw themselves as more open and more agreeable than they are perceived by informants. Furthermore, depression severity was not associated with self-informant agreement. Restriction of range may explain this lack of association because the sample was generally very depressed: the Mean (S.D) HDRS score was 28.2 (8.9). However, our findings are consistent with those of Ready and Clark (2002)
, who interpret the lack of association between depression severity and self-informant rating discrepancy in their study as a substantive rather than artifactual finding. Received wisdom that depressive symptoms distort self-reported personality traits or observer judgment may thus need to be re-examined.
With respect to the study’s clinical and public health implications, many public health and community-level interventions aimed at morbidity reduction assume that ordinary people can serve as gatekeepers or natural helpers by identifying at-risk individuals (Cowen, 1982
; Levine, 1994
; Pescosolido, 1993; Sarason, 1981
). The task of risk-identification ought to be informed by what is known about risk. Personality is powerfully associated with a variety of health and social problems (Borghans, Duckworth, Heckman, & ter Weel, 2008
; Chapman, Fiscella, Kawachi, & Duberstein, 2010
; Krueger, Caspi, & Moffitt, 2000
; Roberts, Kuncel, Shiner, Caspi, & Goldberg, 2007
) including in older adults (Chapman, Lyness, & Duberstein, 2007
; Crowe, Andel, Pedersen, Fratiglioni, & Gatz, 2006
; Duberstein et al., in press
; Duberstein, Pálsson, Waern, & Skoog, 2008
; Wilson, Bennett, Mendes de Leon, Bienias, Morris, & Evans, 2005
). Observer judgments about personality represents a natural capacity with evolutionary significance, as humans may have evolved to perceive broad variations in phenotypic behavior that have implications for group living and survival (McAdams & Pals, 2006
). It would be useful to see if this natural capacity to make judgments about personality could be exploited in public health initiatives. Strategies relying upon family and friends to identify at-risk individuals via personality judgments maybe more cost-effective if they focus on Neuroticism and Conscientiousness given their contribution to an array of public health threats and the apparent imperviousness of informant judgments in these two domains to external influences such as illness burden or cognitive function.
Clinically, it is not surprising that there are differences in perceived agreeableness between patients and their friends or family members, given the large literature on the interpersonal lives of depressed patients (Joiner & Coyne, 1999
). Clearly, there is a need for treatments that explore or help patients understand these discrepancies. Perhaps more interesting are the discrepancies in positive emotions and openness. Clinical research on the implications of these discrepancies for patient functioning or treatment outcomes would be useful. In research contexts where self-report data are unavailable or of questionable reliability, factors reducing the reliability of informant reports, including patient illness burden and cognitive function, should be assessed when feasible in order to account for their moderating role in informant judgments of personality.
Our findings must be qualified by study limitations. First, we document cross-sectional associations, and make no causal claims. Future longitudinal studies might determine whether intra-individual change in illness burden and cognitive status over time drives increasing divergence in self-informant ratings. Second, our small regional sample involved depressed adults, primarily Caucasian, as young as 50, and with complete dyadic data; generalization to national samples, non-depressed older adults, other races/ethnicities, the old-old, or incomplete dyads, are unknown. Third, we did not examine the impact of specific diseases or cognitive syndromes on personality rating agreement; our goal was to gauge the aggregate effect of disease and cognitive function.
These limitations are balanced by several strengths, including an assessment of illness burden based on physician ratings of medically documented data, and not solely on patient self-report. Personality ratings were made using an extensively validated measure, with comprehensive data collected on specific facet-level traits not available from shorter personality inventories. Moreover, we exercised rigorous control over Type I error throughout by testing only domains and facets where significant discrepancies existed, and through judicious application of the FDR.
In conclusion, our findings suggest that informants may overestimate the impact of overt health problems on patient dispositions linked to well-being and quality of life, while underestimating the degree to which deficits in cognitive function reduce behavioral and ideological flexibility. Findings underscore the potential utility of risk-detection strategies that rely on informant-reports of Neuroticism and Conscientiousness, and suggest the need for further research on how age-related changes in medical illness burden and cognitive function affect the ways in which observers (friends, family members, health care providers) use or misuse information about the patient’s condition when making decisions about older depressed patients.