BRAIN INJURY, 1997, VOL. 11, NO . 11, 791±799 Response bias in plaintiffs’ histories PAUL R. LEES-HALEY, CHRISTOPHER W. WILLIAMS, N A T H A N D . Z A S L E R ², S H E L D O N M A R G U I L I E S ², § ¶ LUE T. ENGLISH and KAY B. STEVENS Lees-Haley Psychological Corporation, Woodland Hills, CA, USA ² National Neurorehabilitation Consortium, Inc., Richmond, VA, USA ³ Independent Practice, Silver Springs, Maryland, USA §Health Education Services, Huntsville, Alabama, USA ¶ Department of Special Education and Rehabilitation Counseling, College of Education, University of Kentucky, Lexington, Kentucky, USA [Received 18 February 1997; accepted 10 April 1997] This study investigated response bias in self-reported history of factors relevant to the assessment of traumatic brain injury, toxic brain injury and related emotional distress. Response bias refers to systematic error in self-report data. A total of 446 subjects (comprising 131 litigating and 315 nonlitigating adults from five locations in the United States) completed a symptom questionnaire. Data were obtained from university faculty and students, from patients in clinics specializing in physiatry, neurology, and family medicine, and from plaintiffs undergoing forensic neuropsychological evaluations. Comparisons were made for litigant and non litigant ratings of their past and current cognitive and emotional functioning, including life in general, ability to concentrate, memory, depression, anxiety, alcohol, drugs, ability to work or attend school, irritability, headaches, confusion, self-esteem, and fatigue. Although there is no basis for hypothesizing plaintiffs to be healthier than the general population, plaintiffs rated their pre-injury functioning superior to non-plaintiffs. These findings suggest that response biases need to be taken into account by forensic examiners when relying on litigants’ self-reports of pre-injury status. Introduction Response biases are tendencies to respond in a systematically erroneous fashion, for example during medical, psychological, and neuropsychological evaluations. In forensic settings, in particular, pressures of litigation, potential financial rewards, tendencies to behave in a manner consistent with perceived injuries, and various social influences can induce some litigants to report symptoms of physical, cognitive and emotional distress casually disconnected from underlying physiological and psychological dysfunction. One way these biases can be manifested is in plaintiffs’ responses to questions assessing functioning before and after traumatic events such as traumatic brain injuries and toxic exposures [1]. Because financial compensation is based in part on differences in the quality of cognitive and emotional status from pre- to post-event levels, there are incentives for showing that post-injury functioning is significantly worse than pre-injury. Correspondence to: Paul R. Lees-Haley, 21331 Costanso Street, Woodland Hills, CA 91314, USA. 0269±9052/97 $12 ´ 00 Ñ 1997 Taylor & Francis Ltd. Paul R. Lees-Haley et al. 792 To document changes in status, forensic examiners often compare claimants’ self-reports of current to prior functioning. Insofar as self-reports accurately portray levels of pre-injury functioning, discrepancies between pre- and post-event functioning are potentially useful data when assessing loss of function. However, research indicates that claimants’ recall and evaluation of prior functioning is not always reliable [2±6]. To the extent claimants’ responses to questions probing preinjury functioning are distorted, estimates of discrepancies between pre- and postevent functioning are inaccurate and claims of injury relatively invalid. The purpose of this research was to examine one aspect of response bias and its implications for assessment of patients in litigation. Accuracy of recall When making claims of injury, plaintiffs need to demonstrate that a target event or situation caused diminution in neurobehavioural capacities. The wider the chasm between prior and current functioning the greater the extent of injury. The goal of forensic clinical evaluations is to determine the nature and extent of diminution of cognitive, emotional and behavioural functioning. To make this determination, forensic examiners use a variety of procedures including clinical interviews, physical examinations, neuropsychological tests, observation of the plaintiff, reviews of medical, employment and school records, and the like. Comparing current data on a particular domain, such as memory, to data existing prior to target events potentially can reveal differences between how claimants fared before target events and presently. Often however, objective data indicating how claimants were functioning pre-injury are incomplete, difficult to obtain, or nonexistent. When faced with an absence of independent, objective sources of data on preinjury functioning to compare to current status, examiners sometimes rely on claimants’ self-reports of their status pre-injury. Reliance on the recall of claimants, however, is fraught with uncertainty, particularly with regard to accuracy. Setting aside the causes of response bias (e.g. hysteria, self-deception, malingering) there is evidence that plaintiffs appear to overestimate their pre-injury functioning [1]. There is also evidence that persons reporting mild brain injury exhibit biases in self-reported history [7]. By comparison, then, current functioning appears especially impaired. A fundamental question therefore centres on the accuracy of litigants’ reporting of previous functioning. Research conducted by social and forensic psychologists provides compelling evidence that claimants’ recall of past events can be distorted. Such evidence has implications for the confidence forensic examiners place in the accuracy of litigants’ retrospective memories. Evidence of recall bias in non-forensic settings Social psychologists have long argued that recall of past events, including recall of personal attributes and experiences, is not based on a literal copy or recording of prior events stored in long-term memory [8±11]. Rather, personal memories are to a significant extent a construction of prior events based on current knowledge, feelings, beliefs, and expectancies [12]. Commenting on the nature of personal, episodic memory, Ross [13] makes the point that `. . . personal recall involves an Response bias 793 active, constructive process that is guided by people’s knowledge at the time of retrieval’ (Ross [13], p. 342). Studies show that people strive for consistency between their current self-conceptions and conceptions they had of themselves in the past [14±16]. And these strivings can involve unintentional altering for recollections to fit present circumstances [13]. For example, in one experiment demonstrating the power of current beliefs to alter recall research participants were asked their opinions before and after being exposed to persuasive arguments from a respected peer antithetical to their originally stated position [17]. The key finding was that when asked later to recall their original opinion, participants remembered an opinion more in line with their current than with their original position. In another study examining how memory can be distorted to fit current selfperceptions, participants were asked to categorize their personalities prior to measurements of their actual behaviour by the researchers [18]. Later participants were asked to measure their own behaviour. Findings showed that their self-reported estimates of their observable behaviour were biased towards consistency with their self-perceptions, and these biases tended to increase over time. In another study, patients with chronic pain were asked to keep diaries in which they made hourly ratings of pain intensity [19]. When coming to a clinic a week later, patients were asked to rate the intensity of pain they were experiencing currently and to recall amounts of pain they had experienced during the previous week. Because they were chronic pain patients there was little reason to assume that patients experienced substantial change in pain intensity in one week. Nevertheless, compared to what was written in their diaries earlier in the week, patients reporting high levels of current pain overestimated prior pain levels whereas patients reporting low levels of current pain underestimated earlier levels. In conclusion, these and other studies demonstrate the potency of current circumstances and perceptions to affect retrieval of self-related information from memory. In general, studies show that people strive for consistency between current and prior perceptions such that recall of the past is consonant with the present. In forensic settings we posit that the claimant needs to reconcile current claims of loss with perceptions of one’s past life. That is, to be consistent with the view that functioning was diminished by earlier traumatic events, they tend to see preinjury functioning as better in comparison to the present. Evidence of recall bias in forensic settings Although there is little direct empirical evidence of biased recall of pre-injury functioning in forensic populations [1], some indirect evidence exists. For example, Mittenberg et al. [7] report data showing that persons with brain injury tend to underestimate the premorbid prevalence of symptoms. In their study, participants (non-patient controls) completed an affective, somatic, and memory checklist of symptoms associated with post-concussion syndrome. They were then instructed to imagine sustaining a brain injury in a motor vehicle accident and to endorse symptoms they expected to experience 6 months after the injury. A second group of participants (patients with brain injury) completed the same symptom checklist but were instructed to endorse symptoms as they would have experienced them before the accident. Findings showed that compared to controls, the patients with brain injury underestimated the premorbid frequency of 21 out of 30 specific symptoms. Paul R. Lees-Haley et al. 794 While studies suggest that forensic patients underestimate the frequency of preinjury symptoms, there is a need to replicate and extend earlier findings by exploring this phenomenon among forensic and non-litigating patients from a variety of sources with a diversity of complaints. The purpose of the current study was to compare perceptions of forensic patients to non-litigating controls regarding preinjury and current functioning. Using participants from several locations in the United States presenting with a variety of complaints, we compared recall of prior symptoms and judgments regarding current symptoms of forensic and nonforensic patients. We expected that compared to non-litigating controls, forensic patients would (1) perceive current functioning as inferior, and (2) exhibit response bias by recalling prior functioning as superior. Method Subjects A total of 446 participants (291 women, 151 men, 4 not indicating) consisting of different subpools of participants were recruited from five locations in the United States. Table 1 shows a breakdown of participants with regard to geographic location, gender, age, and involvement in litigation. The forensic sample consisted of 131 patients who retained attorneys in connection with their injuries. The remaining participants were recruited from faculty and students at a university in Kentucky, and patients from a neurology practice in Maryland, a physiatry practice in Virginia, and a family medical practice in Alabama. Materials and procedure Participants were given a questionnaire asking them to report the degree to which specific symptoms, behaviours and aspects of life were problematic in the past and at present. First, participants were asked to recall how problematic each item was at a particular time in the past. Non-plaintiff controls were instructed to make these ratings for 3 years ago, whereas plaintiffs were instructed to make these ratings for their functioning prior to injury. Then, all participants were asked to rate how much of a problem each area was currently. Participants rated how much of a problem was: life in general; ability to concentrate; memory; depression; anxiety; Table 1. Total number of participants, number of women, men, and NI, age, and number in litigation Location Total Women Men NI Mean age In litigation Alabama Southern California Kentucky Maryland Virginia Totals 233 74 43 40 56 446 161 40 37 24 29 291 70 34 4 16 29 151 2 0 2 0 0 4 39´ 9 ( 12´5) 39´ 5 ( 12´8) 31´1 ( 8´6) 39´ 4 ( 16´1) 38´ 5 ( 13´1) 38´ 8 ( 13´1) 0 74 1 24 32 131 Note NI = number of participants not indicating gender. Response bias 795 alcohol; drugs; ability to work or attend school; irritability; headaches; confusion; self-esteem; and fatigue. Ratings were made on 9-point scales anchored at the extremes (1 = very satisfied; this is/was not a problem area for me: 9 = very dissatisfied: this is/was a real problem area for me). Results and discussion Overview of analyses The hypotheses were that compared to non-forensic participants, forensic patients would perceive current functioning as inferior and prior functioning as superior. Data regarding current functioning were analysed first. Because these patients alleged diminished functioning due to their injury, forensic patients were expected to report decrements in current status. Data regarding prior functioning were then analysed. Because of response bias (e.g. biased retrieval or reporting of self-relevant information) forensic patients were expected to remember pre-injury functioning as less problematic than were non-forensic participants. Perception of current functioning To determine whether location resulted in systematic differences in responses of control samples a 4 (Location: Alabama, Kentucky, Maryland, Virginia) ´ 2 (Gender) multivariate analysis of variance (MANOVA) was performed on the dependent variables: life in general; concentration; memory; depression; anxiety; alcohol use; drug use; work-school; irritability; headache; confusion; self-esteem; fatigue; sex; marriage; children. Hotellings test found no main effect of location F ( 48, 595) = 1´32, p > 0´ 05, nor main effect of sex, F ( 16, 200) = 1´65, p > 0´05. The 2-way interaction was not reliable, F ( 48, 596) = 1´09, p > 0´ 05. To determine whether location affected responses of forensic patients, a 3 (Location; Southern California, Maryland, Virginia) ´ 2 (Gender) MANOVA was performed. Again, there were no reliable main effects of location, F ( 32, 84) = 1´11, p > 0´05, gender, F ( 15, 43) = 1´17, p > 0´05, or interaction, F ( 32, 84) = 1´ 01, p > 0´05. To test the hypothesis that forensic patients would perceive current functioning as more problematic than would non-litigants, separate ANOVAs were conducted for each dependent variable. Data were entered into 2 (Group: forensic vs nonforensic) ´ 2 (Gender) analysis of variance. As expected, ANOVA revealed the groups were reliably different on every variable. (Table 2 shows means and standard deviations). Compared to non-litigants, forensic patients perceived current functioning as inferior on the following variables: life in general, F ( 1, 432) = 131´64, p < 0´001; concentration, F ( 1, 432) = 151´25, p < 0´ 001; memory, ( F 1, 433) = 156´72, p < 0´001; depression, F ( 1, 433) = 93´87, p < 0´001; anxiety, F ( 1, 432) = 85´ 23, p < 0´ 001; alcohol, F ( 1, 411) = 5´64, p < 0´02; drugs, F ( 1, 409) = 8´99, p < 0´004; work or school, F ( 1, 429) = 220´81, p < 0´ 001; irritability, F ( 1, 430) = 63´ 78, p < 0´001; headaches, F ( 1, 429)64´68, p < 0´001; confusion, F ( 1, 429) = 139´25; self-esteem, F ( 1, 428) = 87´ 03, p < 0´ 001; fatigue, F ( 1, 428) = 69´14, p < 0´001; sex ( 1, 414) = 64´93, p < 0´001; marriage, F ( 1, 362) = 22´58, p < 0´001; and children, F ( 1, 326) = 20´ 33, p < 0´001. There was a main effect of Gender on only 2 variables. Men indicated being more satisfied currently with their self-esteem (M = 3´81) than did women Paul R. Lees-Haley et al. 796 Table 2. Means and standard deviations on perception of current functioning as a function of group Variables rated Life in general Concentration Memory Depression Anxiety Alcohol Drugs Work-school Irritability Headache Confusion Self-esteem Fatigue Sex Marriage Children Forensic group Control group M SD M SD 6´44 6´44 6´41 6´32 6´31 2´14 2´08 6´58 6´05 5´85 5´47 5´65 6´26 5´54 4´36 3´57 2´34 2´29 2´56 2´64 2´53 2´28 2´24 2´58 2´57 3´01 2´71 2´79 2´51 3´15 3´07 2´83 3´53 3´49 3´34 3´73 3´89 1´59 1´45 2´80 3´91 3´47 2´50 3´27 4´16 3´10 2´90 2´28 2´34 2´2 2´12 2´52 2´48 1´62 1´59 2´26 2´45 2´66 2´16 2´34 2´54 2´61 2´63 2´11 Note. Smaller numbers indicate greater satisfaction, from 1 = `Very satisfied: this is a problem area for me’ to 9 =`Very dissatisfied: this is a real problem for me’. (M = 4´ 05), F ( 1, 428) = 6´85, p < 0´01. Fatigue was perceived as more problematic by women (M = 4´ 89) than by men (M = 4´56). The only reliable 2-way interaction was for anxiety. In the control group men reported anxiety as less of a problem for them than did women, whereas in the forensic group women perceived it less problematic for them than did men. Perception of prior functioning To test for systematic differences in responses of control participants from different locations, a 3 (Location: Alabama; Kentucky; Virginia) ´ 2 (Gender) MANOVA was performed on the dependent variables. Hotellings test revealed no reliable main effect of location, F ( 32, 96) = 0´84, p = 0´72, nor was there a main effect of gender, F ( 16, 201) = 0´ 51, p = 0´94. The location by gender interaction also was not reliable, F ( 48, 599) = 0´51, p = 0´62. The same test was conducted for forensic participants from different locations. Hotellings test showed no reliable main effects of location, F ( 32, 92) = 0´ 82, p = 0´72, gender, F ( 16, 47) = 0´59, p = 0´87, nor was there a reliable interaction, F ( 32, 92) = 0´ 30, p = 0´98. Each of the dependent variables were analysed in a 2 (Group: forensic, nonforensic) ´ 2 (Gender) ANOVA. A reliable main effect of Group was found for 15 out of 16 variables. Table 3 shows means for each variable. Forensic patients recalled functioning as superior or less problematic than did non-forensic patients on the following variables: life in general, F ( 1, 433) = 26´77, p < 0´001; concentration, F ( 1, 432) = 36´18, p < 0´ 001; memory, F ( 1, 433) = 15´66, p < 0´ 001; depression, F ( 1, 432) = 23´77, p < 0´ 001; anxiety, F ( 1, 433) = 29´07, p < 0´001; alcohol, F ( 1, 414) = 559, p < 0´ 02; work-school, F ( 1, 432) = 6´ 92, p < 0´01; irritability, F ( 1, 429) = 29´94, p < 0´001; headache, F ( 1, 430) = 18´17, p < 0´001; confusion, F ( 1, 429) = 11´61, p < 0´002; self-esteem, Response bias Table 3. 797 Means and standard deviations on recall of prior functioning as a function of group Variables rated Life in general Concentration Memory Depression Anxiety Alcohol Drugs Work-school Irritability Headache Confusion Self-esteem Fatigue Sex Marriage Children Forensic group Control group M SD M SD 2´25 1´88 2´00 2´01 2´18 1´58 1´32 1´84 2´12 1´95 1´43 2´18 1´93 1´85 2´04 1´73 1´88 1´71 1´68 1´78 1´83 1´43 1´35 1´86 1´63 1´78 1´27 1´88 1´73 1´80 2´15 1´68 3´ 53 3´ 35 2´ 90 3´ 32 3´ 61 2´ 09 1´ 61 2´ 58 3´ 40 3´ 11 2´ 23 3´ 35 3´ 47 2´ 60 2´ 93 2´ 37 2´33 2´28 2´13 2´51 2´55 2´15 1´84 2´34 2´19 2´55 2´09 2´38 2´42 2´37 2´56 2´38 Note. Smaller numbers indicate greater satisfaction, from 1 = `Very satisfied; this was not a problem area for me’ to 9 = `Very dissatisfied: this was a real problem for me.’ F ( 1, 429) = 21´58, p < 0´001; fatigue, F ( 1, 429) = 39´ 54, p < 0´001; sex, F ( 1, 414) = 5´99, p < 0´02; marriage, F ( 1, 363) = 7´47, p < 0´ 008; children, F ( 1, 326) = 4´26, p < 0´ 05. Only drugs failed to reach significance ( p > 0´05) . There was a mean effect of Gender for six of the variables. Work or school was seen as less problematic by men (M = 1´ 89) than by women (M = 2´59), F ( 1, 432) = 6´92, p < 0´01. Women recalled depression as more of a problem (M = 3´ 14) than did men (M = 2´46), F ( 1, 432) = 4´11, p < 0´05. Men reported self-esteem as less of a problem (M = 2´45) than did women (M = 3´24), F ( 1, 429) = 7´21, p < 0´01. Prior sexual functioning was reported as more satisfying by men (M = 1´90) than by women (M = 2´ 63), F ( 1, 414) = 7´05, p < 0´01. Men reported that irritability was less of a problem (M = 2´58) than did women (M = 3´ 25), F ( 1, 429) = 4´78, p < 0´ 03. and men recalled fewer difficulties with concentration (M = 2´41) than did women (M = 3´ 15), F ( 1, 432) = 5´62, p < 0´02. The only 2-way interaction was for confusion, with women but not men reporting more difficulties in the non-forensic condition whereas no gender differences emerged under the forensic condition, F ( 1, 429) = 22´13, p < 0´ 02. General discussion Findings from the present study replicate and extend earlier findings on differences in reporting of functioning between litigating and non-litigating patients. As expected, data show that current functioning is characterized suboptimally by litigants compared to non-litigants. Among research participants in various locations across the United States, patients in litigation perceived earlier functioning as more satisfactory and trouble-free than did non-litigating individuals. This suggests that response biases need to be taken into account by forensic examiners when considering litigants’ self-reports of pre-injury status. The important finding in this Paul R. Lees-Haley et al. 798 study is that litigants characterize themselves as hypernormal pre-injury Ð superior to other comparable persons. Because there is no evidence that cognitively and emotionally high functioning persons are injured more frequently than others (there is some evidence to the contrary), we believe these inflated self-assessments reflect response bias. When assessing discrepancies between pre- and post-injury status, forensic examiners commonly consider plaintiff’s retrospective assessments of prior functioning. Extent of injury is determined in part by how quality of current functioning pales in comparison to prior functioning. Insofar as plaintiff recollections are biased, the extent of injury may be estimated inaccurately. Such biases could mislead rather than inform those attempting to render objective measurements of the patients’ functional losses. Although the response biases discovered in these data potentially would work to the advantage of plaintiffs, these findings do not resolve questions of the role of plaintiffs’ consciousness or intention in overestimating pre-injury functioning. Biases in reporting may be a by-product of human information processing. Although conscious deception is a possibility, recalling prior functioning as particularly satisfactory could be a function of remembering the past in the context of the present. That is, plaintiffs remember their past standing on an attribute, such as memory, by noting their present status on the attribute in question. As Ross [13] stated, `The present serves as a bench-mark because it is generally more salient and available than a person’s earlier standing on an attribute. As a result, construction of the past may consist, in large part, of characterizing the past as different from or the same as the present’ (Ross [13], p. 343). Thus plaintiffs may note how dissatisfied they are in some domain and perceive, by comparison, that the past was especially satisfactory. They may also feel compelled to report losses from their pre-injury status due to social pressures, litigation incentives, cognitive dissonance, or for other reasons. Although we agree that the standard of practice is to consider the possibility of malingering in forensic evaluations, we do not believe malingering fully accounts for biases that emerge in forensic data. Many psychological diagnoses rely heavily on subjective self reports of plaintiffs, and much of the information affecting experts’ reasoning about causation is derived from plaintiff self reports. Consequently, research on response biases affecting the validity of plaintiff self reports is critically important for clinicians evaluating consequences of injuries to persons in litigation. References 1. LEES-HALEY, P. R., WILLIAMS, C. W. and ENGLISH, L. T.: Response bias in self-reported history of plaintiffs compared with nonlitigating patients. Psychological Reports, 79, 811±818, 199. 2. FEINSTEIN, A. R.: Methodological problems and standards in case-control research. Journal of Chronic Diseases, 18, 35±41, 1979. 3. FEINSTEIN, A. R. and HOROWITZ, R. I.: Double standards, scientific methods, and epidemiologic research. New England Journal of Medicine, 307, 1611±1617, 1982. 4. GEHLBACH, S. H.: Interpreting the Medical Literature, 3rd edn (McGraw-Hill, New York), 1983. 5. ROTHMAN , K. J.: Modern Epidemiology (Little Brown, Boston, MA), 1986. 6. TURK, D. C. and SALOVEY, P. (editors): Reasoning Inference, and Judgment in Clinical Psychology (Free Press, New York), 1988. 7. MITTENBER, W., DIGIULIO , D. V., PERRIN, S. et al.: Symptoms following mild head injury: expectation as aetiology. Journal of Neurology, Neurosurgery, and Psychiatry, 55, 200±204, 1992. Response bias 799 8. BARTLETT , F. C.: Remembering: A study in experimental and social psychology (Cambridge University Press, Cambridge), 1932. 9. BRUNER, J. S.: On perceptual readiness. Psychological Review, 64, 123±152, 1957. 10. MARKUS, H. and ZAJONC, R. B.: The cognitive perspective in social psychology. In G. Lindzey and E. Aronson (editors) The Handbook of Social Psychology, 3rd edn (Random House, New York), pp. 137±230, 1985. 11. TAYLOR , S. E. and CROCKER, J.: Schematic bases of information processing. In E. T. Higgins, C. P. Herman and M. P. Zanna (editors) Social Cognition: The Ontario Symposium (Erlbaum, Hillsdale, NJ), pp. 89±134, 1981. 12. FISKE, S. T. and TAYLOR, S. E.: Social Cognition (McGraw-Hill, New York), 1991. 13. ROSS, M.: The relation of implicit theories to the construction of personal histories. Psychological Review, 96, 341±357, 1989. 14. SWANN, W. B., JR: Self-verification: bringing social reality into harmony with the self. In J. Sulls and A. G. Greenwald (editors) Psychological Perspectives on the Self (Erlbaum, Hillsdale, NJ), pp. 33± 66, 1984. 15. SWANN, W. B. JR and READ , S. J.: Acquiring self-knowledge: the search for feedback that fits. Journal of Personality and Social Psychology, 41, 1119±1128, 1981. 16. SWANN, W. B. JR and READ, S. J.: Self-verification processes: how we sustain our self-conceptions. Journal of Experimental Social Psychology, 17, 351±370, 1981. 17. GOETHALS, G. R. and RECKMAN, R. F.: The perception of consistency in attitudes. Journal of Experimental Social Psychology, 9, 491±501, 1973. 18. KULIK , J. A. and MAHLER, H. I. M.: Self-confirmatory attribution, egocentrism, and the perpetuation of self-beliefs. Journal of Personality and Social Psychology, 50, 587±593, 1986. 19. EICH, E., REEVES, J. L., JAEGER, B., et al.: Memory for pain: relation between past and present pain intensity. Pain, 23, 375±380, 1985.