Realism of Confidence in Obsessive–Compulsive Checkers

Realism of Confidence in Obsessive–Compulsive Checkers
[Articles]
Dar, Reuven1,4; Rish, Sigalit1; Hermesh, Haggai2; Taub, Migdala3; Fux, Mendel3
1Department of Psychology, Tel Aviv University, Tel Aviv, Israel
2Geha Mental Health Center, Petah Tiqva, Israel
3Mental Health Center, Beer-Sheva, Israel
4Correspondence concerning this article should be addressed to Reuven Dar, Department of Psychology, Tel Aviv University, Tel Aviv 69978, Israel. Electronic mail may be sent to ruvidar@freud.tau.ac.il.
This study was supported by a grant from the Israel Foundation Trustees.
Received Date: February 17, 1999; Revised Date: March 23, 2000; Accepted Date: April 3, 2000
Abstract
The study examined whether obsessive–compulsive (OC) checkers have reduced confidence in their knowledge. OC checkers were compared with panic disorder (PD) patients and nonpatient controls using a calibration-of-knowledge procedure. Participants completed a general knowledge questionnaire, rated their confidence in each answer, and estimated the total number of correct answers. These responses were converted to 2 measures of confidence relative to performance—over/underconfidence and over/underestimation. OC checkers had lower scores than nonpatients did on both measures, whereas the PD patients did not differ from either group. For the OC checkers, relative confidence was inversely related to the severity of obsessions. The authors speculate that confidence may depend on a confirmation bias in testing hypotheses and that the reduced confidence in OC checkers may reflect a disconfirmation bias in this population.
________________________________________


A common clinical characteristic of obsessive–compulsive disorder (OCD) is malignant doubt, which is often followed by compulsive checking and typically concerns issues of contamination, aggression, or safety. Many OCD patients, especially those who have been dubbed obsessive–compulsive (OC) checkers (Rachman & Hodgson, 1980), appear to distrust their own memory and judgment. They make futile attempts to reconstruct events in their memory, engage in repeated checking rituals, and often demand reassurance from others that they have definitely cleaned their hands after using the toilet or that they have positively remembered to turn off the stove (Dar, 1991).
Recognition of these symptoms has led to an upsurge of studies examining possible memory or reality-monitoring deficits in OC checkers. These studies have been motivated by the rationale that memory deficits may account for the obsessive doubting and excessive need for checking in this population. Whereas a few studies have found nonverbal memory impairment in OCD patients (reviewed by Savage et al., 1999), most studies have not supported the hypothesis that OC checking is associated with a memory deficit. Rather, OCD patients and subclinical checkers consistently express less confidence in or feel less comfortable with their memory judgments, as compared with normal participants, despite their equally good objective performance (Brown, Kosslyn, Breitler, Baer, & Jenike, 1994; Constans, Foa, Franklin, & Mathews, 1995; Foa, Amir, Geshuny, Molnar, & Kozak, 1997; MacDonald, Antony, MacLeod, & Richter, 1997; McNally & Kohlbeck, 1993; Sher, Frost, & Otto, 1983).
This apparent lack of confidence in OC checkers may not necessarily be restricted to memory judgment. In fact, a striking phenomenon in these patients is their inability to feel certain even in the face of seemingly clear and unambiguous evidence. For example, an OC checker may turn the key in the lock over and over again without being able to convince himself or herself that the door has in fact been locked, even though he or she can plainly see that the key is in the proper position, hear it engaging, and feel the lock snapping. Previous theorists of OCD (Reed, 1985; Shapiro, 1965) have suggested that inability to experience conviction is a central trait in OC people and that this trait is not limited to particular obsessions or compulsions. These theorists have suggested that individuals with OCD are able to function well despite this deficit by using various compensation strategies such as adopting “objective” rules and norms to guide their behavior. In the present study, we follow this theoretical tradition by testing the hypothesis that uncertainty is a general feature of OC checkers, which affects not only memory but all areas of personal knowledge.
The evidence that OC checkers tend to doubt their own knowledge, at least in particular areas, stands in intriguing contrast to the robust finding that most people seem to be too certain in evaluating their knowledge. This finding has emerged in numerous studies of probability calibration (reviewed by Lichtenstein, Fischoff, & Phillips, 1982). In these studies, confidence judgments are elicited as assessments of the probability that a given statement is true. In the method adopted for the present study, participants are given a series of dual-choice questions. After choosing one of the answers, participants are asked to rate their confidence that the chosen answer is the correct one. The appropriateness, or realism, of their confidence judgments is measured by comparing these confidence ratings with the observed frequencies of correct responses. The pervasive finding in these studies has been that the subjective ratings of confidence are too high, reflecting participants' overconfidence in their answers. Overconfidence has been found in a wide range of tasks, including general knowledge questions, semantic or episodic memory, and predictions of future events, and in samples of laypersons and experts alike (Griffin & Tversky, 1992).
The methodology of calibration studies is particularly suitable to test the hypothesis that OC checkers' lack of confidence is a general characteristic of these patients and is not limited to memory or reality monitoring. The studies that found reduced confidence in OC checkers (cited previously) used a variety of scales to assess confidence in knowledge and did not employ any measures that quantify confidence in relation to actual performance. The methodology of calibration studies, in contrast, provides a well-tested procedure and standard measures for quantifying the relationships between knowledge and confidence. Two such measures of confidence relative to performance, over/underconfidence and over/underestimation, were used in the present study. 1 We describe the two measures in detail in the Method section.
Patients with OCD typically present high levels of anxiety and depression (Dar, 1991; Rassmusen & Eisen, 1992), which may account for any differences in performance between these patients and normal control participants, certainly when confidence ratings are the main dependent variable. However, none of the studies cited previously attempted to control for the potentially confounding effects of anxiety and depression. The present study counters this problem by including a control group of panic disorder (PD) patients with levels of anxiety and depression that are similar to those of the OCD patients.
We predict that OC checkers, as compared with normal control participants and patients with PD, show less overconfidence and overestimation of their knowledge. We also predict that the levels of overconfidence and overestimation are inversely related to the severity of OCD, specifically to the severity of obsessions in the OCD patients, but not to anxiety and depression.
Method
Participants
Three groups of participants were recruited for this study: (a) OC checkers, (b) PD patients without OCD, and (c) nonpatient controls. The OCD sample consisted of 20 patients, 14 men and 6 women. We selected patients who reported primary checking compulsions and doubting obsessions in their diagnostic interview and who confirmed this report by endorsing at least one of the checking compulsions on the Yale–Brown Obsessive–Compulsive Scale (Y-BOCS) (Goodman et al., 1989; see the Procedure section). The PD group consisted of 29 patients, 14 men and 15 women. All the patients were outpatients arriving for regular visits at two large urban mental health centers. The control group had 23 participants, 13 men and 10 women, who were recruited from the centers' staff and visitors. Of the OCD patients, 19 (95%) were receiving medication, as were 14 (70%) of the PD patients. None of the control participants were on medication, and none had a history of mental illness. Diagnoses were based on the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders (DSM–IV, 1994) and were based on a formal intake interview at the mental health center by a team of psychiatrists and clinical psychologists specializing in anxiety disorder. Prior to participation, all participants were asked to sign an informed consent, and patients agreed to grant the research team permission to collect data from their hospital records.
Instruments
All the instruments we use have been used in previous studies, and their psychometric properties are well established. The principal instrument for this study was a 100-item general knowledge, two-alternative questionnaire (e.g., “solar wind is (a) an arctic storm; (b) emission of atomic particles from the sun”). The items in this questionnaire were adapted from psychometric examinations provided by the National Institute for Testing and Evaluation in Israel.
The type and severity of OC symptoms were assessed with the Y-BOCS (Goodman et al., 1989). The Y-BOCS is a clinician-rated 10-item scale, with each item rated from 0 (no symptoms) to 4 (extreme symptoms), resulting in a total range of 0–40 with separate subtotals for severity of obsessions and compulsions. The Y-BOCS assesses the presence of various types of obsessions (e.g., aggressive, religious) and compulsions (e.g., cleaning/washing, checking).
Level of depression was measured by the Beck Depression Inventory (BDI; Beck, 1967). The BDI is a 21-item self-report scale that assesses the severity of depressive symptoms during the past week, with a total range of 0–63.
The severity of current anxiety was measured with the Hamilton Anxiety Rating Scale (HARS; Hamilton, 1959), as adapted by Gary, Alan, Reed, and Jacques (1994). The HARS is a clinician-administered scale that assesses the presence and severity of anxiety symptoms during the past week, with a total range of 0–56.
Procedure
The procedure for this study was adapted from previous calibration studies as described previously (e.g., Koriat, Lichtenstein, & Fischhoff, 1980). Participants completed the instruments we have described in individual sessions, beginning with the general knowledge questionnaire. For each question on this questionnaire, participants were asked to choose the correct alternative and then rate how confident they were that their choice was correct, using a scale of 5–100% with 5% intervals (confidence ratings). Following the first and second sets of 50 questions, participants were asked to estimate how many of the previous 50 questions they had answered correctly (estimated frequency). On completion of the general knowledge questionnaire, the two patient groups completed the BDI and the experimenter, a trained graduate student in clinical psychology, administered the HARS. The OCD patients were also interviewed with the Y-BOCS, including the checklist. Data related to the patients' treatment and medical history were collected from the patients' hospital records and from attending staff.
Realism of Confidence Measures
Over/underconfidence.
This measure indicates the extent to which a participant is overconfident or underconfident relative to his or her actual performance. Specifically, over/underconfidence = R - C, where R is the average confidence rating over the 100 items and C is the number of correct answers. The possible range of over/underconfidence is therefore -100 (maximally underconfident) to 100 (maximally overconfident).
Over/underestimation.
Over/underestimation can be measured by the equation over/underestimation = E - C. Here, E is the participant's estimate, following the test, of the number of questions he or she has answered correctly, and C is the participant's total number of correct answers. As for over/underconfidence, the range of underestimation in this study is -100 (maximal underestimation) to 100 (maximal overestimation). Over/underestimation has been shown by Gigerenzer, Hoffrage, and Kleinbolting (1991) to behave quite differently from the traditional measure of over/underconfidence. These authors demonstrated that under conditions in which people seem overconfident of their knowledge on the traditional measure, they appear quite accurate or realistic on this over/underestimation measure.
Results
Sample Matching
When we examined the results of the original sample, we found that the PD patients' performance was poor in comparison to that of the other two groups. The mean number of correct answers for the PD patients was 65.97 (SD = 13.50), as compared with 75.60 (SD = 14.23) for the OCD patients and 79.39 (SD = 15.45) for the nonpatient controls. These differences were statistically significant, t(69) = 2.31, p < .05, and t(69) = 3.35, p < .05, respectively. These unexpected differences in performance could produce a potential confound. Calibration measures are not linearly related to performance, and overconfidence, specifically, tends to be higher for difficult items. Consequently, differences in overconfidence between PD patients and the other two groups may be an artifact of the differences in performance. To overcome this problem, we eliminated the 9 participants with the poorest performance from the PD group and the 3 participants with the best performance from the nonpatient control group. The resulting sample had 20 participants in each group (14 men and 6 women in the OCD group, 11 men and 9 women in the PD group, and 12 men and 8 women in the nonpatient control group).
Table 1 presents participants' age and education levels and the scores on the BDI and HARS for the two patient groups. There were no statistically significant differences between the three groups on any of these variables. The mean Y-BOCS score of the OCD patients was 22.00 (range of 11–39), which is at the moderate severity level.

________________________________________

[Help with image viewing]
[Email Jumpstart To Image]
Table 1 Age, Education Level, and Scores on the BDI and HARS by Group
________________________________________
Realism of Confidence
We predicted that OC checkers would be less confident, relative to their performance, than the other two groups were. Table 2 displays the number of correct answers, average confidence rating, and estimated number of correct answers in the three participant groups. A clear pattern can be observed in these data: Although the OC checkers' actual performance on the test was no different from that of the nonpatient controls, the gap in confidence was statistically significant, F(1, 57) = 4.68, p < .05. The test performance of the PD patients, in turn, was slightly below that of the OC checkers, whereas their confidence ratings were essentially the same.

________________________________________

[Help with image viewing]
[Email Jumpstart To Image]
Table 2 Number of Correct Answers, Average Confidence Rating, and Estimated Number of Correct Answers by Group
________________________________________
Table 3 displays the derived measures of realism of confidence described previously for the three groups. We used both over/underconfidence and over/underestimation as dependent measures for the statistical tests of our hypothesis. In examining the differences between the groups for each of the dependent measures, we followed Fisher's planned comparisons procedure, as described by Levin, Serlin, and Seaman (1994).

________________________________________

[Help with image viewing]
[Email Jumpstart To Image]
Table 3 Over/Underconfidence and Over/Underestimation by Group
________________________________________
Over/underconfidence.
The omnibus F test for group differences in over/underconfidence was statistically significant, F(2, 57) = 3.25, p < .05, permitting the test of the pairwise contrasts. As we predicted, the contrast between the OCD participants and the nonpatient controls was statistically significant, F(1, 57) = 6.49, p < .05. The contrasts between the OCD and PD patients, as well as between the PD patients and the nonpatient controls, were not statistically significant, F(1, 57) = 1.58 and F(1, 57) = 1.65, respectively. In OC checkers, overconfidence was negatively correlated with the severity of OCD on the Y-BOCS, r(19) = -.47, p < .05, so that greater severity of OCD was associated with less confidence relative to performance. When the two Y-BOCS subscales were examined separately, overconfidence correlated negatively with obsessionality, r(19) = -.52, p < .05, but its correlation with compulsivity was small and not statistically significant, r(19) = -.21, ns. However, the difference between the two correlation coefficients was not significant in this small sample (n = 20). The correlations of overconfidence with anxiety and depression (pooled within groups) were both small and nonsignificant, r(38) = .10 and r(38) = -.17, respectively. Similarly, over/underconfidence did not correlate significantly with either age, r(57) = .06, ns, or education, r(57) = -.21, ns. There was no effect of gender on over/underconfidence, F(1, 54) = 2.90, ns, and no interaction between gender and group, F(2, 54) = .62, ns.
Over/underestimation.
The pattern of group differences on over/underestimation replicated the pattern obtained for over/underconfidence (see Table 3). The omnibus test for group differences was significant, F(2, 57) = 5.33, p < .05. As predicted, the contrast between the OCD patients and the nonpatient control group was significant, F(1, 57) = 6.49, p < .05. Neither the contrast between the OCD and PD patients nor the contrast between the PD patients and the nonpatient controls was statistically significant, F(1, 57) = 1.99 and F(1, 57) = 3.40, respectively. Unlike for over/underconfidence, however, the correlations between over/underestimation and the severity of OCD on the Y-BOCS were small and did not reach statistical significance, r(19) = -.20, ns. This was also the case for the correlation of over/underestimation with the Y-BOCS subscale of obsessionality, r(19) = -.31, ns, and compulsivity, r(19) = .03, ns. As was the case for over/underconfidence, over/underestimation did not correlate significantly with anxiety, r(19) = .19, ns; depression, r(19) = -.04, ns; age, r(19) = .14 ns; or level of education, r(19) = -.003, ns. Similarly, there were no effects of gender on over/underestimation, F(1, 54) = .95, ns, and no interaction between gender and group, F(2, 54) = 1.40, ns.
An examination of the intercorrelations between the dependent measures within groups reveals interesting differences between the three participant groups. In nonpatient controls, the average confidence rating was extremely highly correlated with the estimated number of correct answers, r(19) = .91, p < .05, whereas in OC checkers, the same correlation was only moderate, r(19) = .52, p < .05. The difference between these correlation coefficients was statistically significant (p < .05, two-tailed). In the PD patients, the correlation coefficient between these variables was high, r(19) = .75, p < .05, and not statistically different from either of the other two. The correlation between over/underconfidence and over/underestimation was high and statistically significant in nonpatient controls, r(19) = .68, p < .05, as well as in PD patients, r(19) = .66, p < .05. In OC checkers, in contrast, the same correlation was very small (r = .13, ns). The difference between the correlation coefficients in the nonpatient and OCD groups was statistically significant (p < .05, two-tailed), whereas the corresponding difference between the OCD and PD groups failed to reach statistical significance (p = .058, two-tailed).
Discussion
The results of this study support our hypothesis that OC checkers are less confident in their knowledge, relative to their actual performance, than nonpatient controls are. Our prediction that we would obtain the same result in comparison with PD patients, however, was not fully corroborated: Relative confidence in the PD patients was between that of the OC checkers and the normal controls and was not statistically different from either. As we predicted, relative confidence correlated with the severity of OCD symptoms in OC checkers. More specifically, it was related only to severity of obsessions, not to severity of compulsions. In both patient groups, relative confidence was not correlated with either anxiety or depression.
Before we continue our discussion, a caveat is necessary. The area of confidence judgment research is replete with controversy. Specifically, there is no agreement on whether over/underconfidence is really a bias or on which cognitive processes may contribute to it (e.g., Gigerenzer et al., 1991; Griffin & Tversky, 1992). This study does not take a stand on these controversies, and its results do not depend on the validity of any particular model. We simply find it convenient to use the methodology of confidence judgments to test our hypothesis of relative lack of confidence in OC checkers. This is also why we avoid the traditional “calibration curves” that are at the heart of the controversy on the relationship between frequency and single event probability. Similarly, we do not interpret over/underconfidence literally, as representing “too confident” or “not sufficiently confident”; we are simply using it as an index of confidence relative to performance, which allows between-group comparisons. At the same time, we should note that the pattern of results obtained in the normal control group replicates previous findings (e.g., Gigerenzer et al., 1991), with probability judgments showing overconfidence and frequency estimations being well calibrated in nonpatients.
We conclude from the obtained pattern of results that OC checkers indeed display a lack of confidence in their knowledge that is not limited to memories or to areas of obsessive concerns. This lack of confidence was significant in a sample of OCD patients who were treated with medications. It was related to level of obsessionality but was unrelated to anxiety or depression. The extent to which this deficit is specific to OCD, however, remains unclear in this study. PD patients displayed some confidence deficit as well, even though their performance was not statistically different from that of normal participants. The finding that both over/underconfidence and over/underestimation correlated with obsessionality and not with anxiety or depression suggests that obsessionality may be the key factor in explaining the reduced confidence in PD patients as well as in OCD patients. In other words, the lack of confidence may be related to having obsessions or obsessive doubts rather than to having OCD. Unfortunately, we did not assess obsessionality in the PD patients (the Y-BOCS is designed specifically for OCD), so this hypothesis remains to be explored in future studies.
In our nonpatient control sample, confidence ratings and estimated number of correct answers were extremely highly correlated, whereas in the OC checkers, the correlation was significantly smaller. The same pattern was reflected in the derived measures of over/underconfidence and over/underestimation, which were essentially uncorrelated in the OCD sample. In the PD group, the correlations fell between those of the OCD group and the normal group. Gigerenzer et al. (1991) made a distinction between judgments of single event probability (confidence) judgments and frequency judgments. They argued that these judgments rely on overlapping but not on identical sets of cues. Our results may suggest that in the normal sample, the extent of overlap between the two sets of cues is higher than in the OCD sample. Specifically, the OCD patients' estimates of their overall performance may be affected not only by their sense of how they performed the task but also by a general factor of doubt and uncertainty. This factor may bias their judgment, in retrospect, in a way that is not related to their performance on the specific task, thereby reducing the correlation between confidence ratings and the estimated number of correct answers.
Our finding that OC checkers have a relative deficit in feeling confident about their knowledge raises at least two major questions. The first has to do with the causal role of this deficit in this population. It is entirely unclear whether this deficit is a consequence of having OCD or whether it is a risk factor for developing the disorder. This issue can only be properly examined in prospective studies. The second question, not unrelated to the first one, has to do with possible underlying mechanisms for this relative lack of confidence.
In their model of overconfidence, Koriat et al. (1980) suggested that one of its major causes is a confirmation bias—that is, the tendency to generate and overweight evidence supporting one's favored hypothesis while failing to generate and underweighting evidence that is inconsistent with the favored hypothesis. Confirmation bias has been widely studied and appears to be ubiquitous in many contexts of hypothesis testing (e.g., Klayman & Ha, 1987; Trope & Bassok, 1982). We believe that the relative lack of confidence evidenced by OC checkers may reflect an opposite bias toward attending to and searching for disconfirming evidence. The process of doubt, according to this view, can be conceptualized as reflecting a disconfirmation bias, which generates a search for evidence that might undermine OC checkers' confidence in their hypotheses. In the example we used earlier in this article, when an OC checker attempts to convince himself or herself that he or she has indeed locked the door properly, this disconfirmation bias is reflected in the process of generating events or consequences that could prevent the door from having been locked or staying locked. This process of disconfirmation, or active doubting, ends up reducing the patient's confidence in the hypothesis that the door is in fact locked.
We are currently conducting a series of studies examining the hypothesis that OCD is associated with a disconfirmation bias. If evidence can be found to support this idea, it will be very much in line with current thinking about the adaptive value of judgment biases (e.g., Arkes, 1991; Barkow, Cosmides, & Tooby, 1992) and specifically of the confirmation bias (e.g., Friedrich, 1993; Lewicka, 1998). We suggest that confidence may depend on actively sidestepping disconfirming instances as a means of preventing doubt from creeping in and stalling functioning.
Finally, we note that an interesting parallel to this idea might be found in the history of science and its rational reconstruction. Following the decline of logical positivism, refutation was proclaimed the only rational strategy for scientific progress. In this view, OC checkers may seem to be ideal scientists, constantly questioning their own hypotheses. But as Lakatos (1970) has shown, strict refutation of theories is not only impossible (for reasons that cannot be discussed within the scope of this article) but also irrational. He provided numerous examples of celebrated research programs (such as Newton's) that grew in the face of apparent anomalies that were resolved only many years later. Had such anomalies been allowed to refute these research programs, they never would have developed, and the progress of science would have been severely crippled. This reconstruction of science, then, may indicate that a healthy dose of confirmation bias is essential to scientific as well as to personal growth.
References
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. [Context Link]
Arkes, H. R. (1991). Costs and benefits of judgment errors: Implications for debiasing. Psychological Bulletin, 110, 486–498. Ovid Full Text [Context Link]
Barkow, J. H., Cosmides, L., & Tooby, J. (Eds.) (1992). The adapted mind: Evolutionary psychology and the generation of culture. New York: Oxford University Press. [Context Link]
Beck, A. T. (1967). Depression—Clinical, experimental, and theoretical aspects. New York: Harper. [Context Link]
Brown, H. D., Kosslyn, S. M., Breitler, H. C., Baer, L., & Jenike, M. A. (1994). Can patients with obsessive–compulsive disorder discriminate between percepts and mental images? A signal detection analysis. Journal of Abnormal Psychology, 103, 445–454. Ovid Full Text Bibliographic Links [Context Link]
Constans, J. I., Foa, E. B., Franklin, M. E., & Mathews, A. (1995). Memory for actual and imagined events in OC checkers. Behaviour Research and Therapy, 33, 665–671. [Context Link]
Dar, R. (1991). Obsessive–compulsive disorder: An update. International Review of Psychiatry, 1, 235–241. [Context Link]
Foa, E. B., Amir, N., Geshuny, B., Molnar, C., & Kozak, M. (1997). Implicit and explicit memory in obsessive-compulsive disorder. Journal of Anxiety Disorders, 11, 119–129. Bibliographic Links [Context Link]
Friedrich, J. (1993). Primary error detection and minimization (PEDMIN) strategies in social cognition: A reinterpretation of confirmation bias phenomena. Psychological Review, 100, 298–319. Ovid Full Text [Context Link]
Gary, S. B., Alan, M. G., Reed, D. G., & Jacques, P. B. (1994). Hamilton Anxiety Rating Scale Interview Guide: Joint interview and test–retest methods for interrater reliability. Psychiatry Research, 53, 191–202. [Context Link]
Gigerenzer, G., Hoffrage, U., & Klienbolting, H. (1991). Probabilistic mental models: A Brunswickian theory of confidence. Psychological Review, 98, 506–528. Ovid Full Text [Context Link]
Goodman, W. K., Price, L., Rasmussen, S. A., Masure, C., Fleischmann, R. L., Hill, C. L., Heninger, G. R., & Charney, D. S. (1989). The Yale–Brown Obsessive–Compulsive Scale: I. Development, use, and reliability. II. Validity. Archives of General Psychiatry, 46, 1006–1016. [Context Link]
Griffin, D., & Tversky, A. (1992). The weighing of evidence and the determinants of confidence. Cognitive Psychology, 24, 411–435. [Context Link]
Hamilton, M. (1959). The assessment of anxiety states by rating. British Journal of Medical Psychology, 32, 50–55. [Context Link]
Klayman, J., & Ha, Y. W. (1987). Confirmation, disconfirmation, and information in hypothesis-testing. Psychological Review, 94, 211–228. [Context Link]
Koriat, A., Lichtenstein, S., & Fischoff, B. (1980). Reasons for confidence. Journal of Experimental Psychology: Human Learning and Memory, 6, 107–118. [Context Link]
Lakatos, I. (1970). Falsification and the methodology of scientific research programmes. In I. Lakatos & A. Musgrave (Eds.), Criticism and the growth of knowledge (pp. 91–196). Cambridge, England: Cambridge University Press. [Context Link]
Levin, J. R., Serlin, R. C., & Seaman, M. A. (1994). A controlled, powerful multiple-comparison strategy for several situations. Psychological Bulletin, 115, 153–159. [Context Link]
Lewicka, M. (1998). Confirmation bias: Cognitive error or adaptive strategy of action control? In M. Kofta, G. Weary, & G. Sedek (Eds.), Personal control in action: Cognitive and motivational mechanisms (pp. 233–258). New York: Plenum. [Context Link]
Lichtenstein, S., Fischoff, B., & Phillips, L. D. (1982). Calibration of probabilities: The state of the art to 1980. In D. Kahneman, P. Slovic, & A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases (pp. 306–334). Cambridge, England: Cambridge University Press. [Context Link]
MacDonald, P. A., Antony, M. M., MacLeod, C. M., & Richter, M. A. (1997). Memory and confidence in memory judgments among individuals with obsessive compulsive disorder and non-clinical controls. Behaviour Research and Therapy, 35, 497–505. [Context Link]
McNally, R. J., & Kohlbeck, P. A. (1993). Reality monitoring in obsessive–compulsive disorder. Behaviour Research and Therapy, 31, 249–253. [Context Link]
Rachman, S., & Hodgson, R. L. (1980). Obsessions and compulsions. Englewood Cliffs, NJ: Prentice Hall. [Context Link]
Rassmusen, S. A., & Eisen, J. L. (1992). The epidemiology and clinical features of obsessive compulsive disorder. Psychiatric Clinics of North America, 15, 743–758. [Context Link]
Reed, G. F. (1985). Obsessional experience and compulsive behaviour: A cognitive–structural approach. Orlando, FL: Academic Press. [Context Link]
Savage, C. R., Baer, L., Keuthen, N. J., Brown, H. D., Rauch, S. L., & Jenike, M. A. (1999). Organizational strategies mediate nonverbal memory impairment in obsessive–compulsive disorder. Biological Psychiatry, 45, 905–916. Bibliographic Links [Context Link]
Shapiro, D. (1965). Neurotic styles. New York: Basic Books. [Context Link]
Sher, K. J., Frost, R. O., & Otto, R. (1983). Cognitive deficits in compulsive checkers: An exploratory study. Behaviour Research and Therapy, 21, 357–363. [Context Link]
Trope, Y., & Bassok, M. (1982). Confirmatory and diagnosing strategies in social information gathering. Journal of Personality and Social Psychology, 43, 22–34. [Context Link]
1We also computed two nondirectional measures of sensitivity to performance, calibration and resolution. We omit both from this article, as they were not pertinent to our hypotheses and did not reveal any group differences. [Context Link]
________________________________________

Accession Number: 00004468-200011000-00012
Copyright (c) 2000-2007 Ovid Technologies, Inc.
Version: rel10.5.8, SourceID 1.13281.2.32.1.0.2.197.1.4.1.5