BRIEF REPORTS A Comparison of Different Methods for Assessing the Features of Eating Disorders in Patients With Binge Eating Disorder Carlos M. Grilo and Robin M. Masheb Yale University School of Medicine G. Terence Wilson Rutgers-The State University of New Jersey The authors compared 3 methods for assessing the features of eating disorders in patients with binge eating disorder (BED). Participants were administered the Eating Disorder Examination (EDE) interview and completed the EDE Questionnaire (EDE-Q) at baseline. Participants prospectively self-monitored their eating behaviors daily for 4 weeks and then completed another EDE-Q. The EDE and the EDE-Q were significantly correlated on frequencies of objective bulimic episodes (binge eating) and on the Dietary Restraint, Eating Concern. Weight Concern, and Shape Concern subscales. Mean differences in the EDE and EDE-Q frequencies of objective bulimic episodes were not significant, but scores on the 4 subscales differed significantly, with the EDE-Q yielding higher scores. At 4 weeks, the EDE-Q retrospective 28-day assessment was significantly correlated with the prospective daily self-monitoring records for frequency of objective bulimic episodes, and the mean difference between methods was not significant. The EDE-Q and self-monitoring findings for subjective bulimic episodes and objective overeating differed significantly. Thus, in patients with BED, the 3 assessment methods showed some acceptable convergence, most notably for objective bulimic episodes. Binge eating disorder (BED) is a new eating disorder category included in the Diagnostic and Statistical Manual of Mental Dis- orders (4th ed. [DSM-IV); American Psychiatric Association [APA], 1994), in Appendix B, reflecting "criteria sets provided for further study." BED is characterized by recurrent binge eating without the compensatory weight control methods that distinguish bulimia nervosa. Binge eating is defined as the consumption of an unusually large amount of food with a subjective sense of loss of control over the eating (APA, 1994). Assessment of binge eating is challenging in several respects. The definition of binge eating and its operationalization are com- plex. Research has documented the variability in what is meant by the term binge eating (e.g., Beglin & Fairburn, 1992; Pratt, Niego, & Agras, 1998; Telch, Pratt, & Niego, 1998). Another challenge is that binge eating is generally secretive and solitary in nature (Grilo, Shiffman, & Carter-Campbell, 1994), thus precluding the use of informants to provide corroboration. There are no known Carlos M. Grilo and Robin M. Masheb, Department of Psychiatry, Yale University School of Medicine; G. Terence Wilson, Department of Psy- chology, Rutgers-The State University of New Jersey. This research was supported by Grant DK49587 from the National Institutes of Health. Earlier versions of this report were presented at the eighth international New York Conference on Eating Disorders, New York, April 1998, and at the World Congress of Behavioral and Cognitive Therapies, Acapulco, Mexico, July 1998. We acknowledge Elayne Daniels for her contribution to the assessment of participants. Correspondence concerning this article should be addressed to Carlos M. Grilo, Yale Psychiatric Research, Yale University School of Medicine. P.O. Box 208098. New Haven, Connecticut 06520. biological markers of binge eating. Hence, one must rely on patients' self-reports of these behaviors (Wilson, 1993). However, because binge eating frequently occurs in the context of negative emotions (Grilo et al., 1994) and may even represent a method of coping with or escaping from psychological distress (Heatherton & Baumeister, 1991), it may be difficult for patients to provide accurate retrospective reports of their binge eating. The Eating Disorder Examination (EDE; Cooper & Fairburn, 1987; Fairburn & Cooper, 1993), a semistructured investigator- based interview, is currently regarded as the gold standard in the assessment of binge eating and eating disorders (Grilo, 1998; Smith, Marcus, & Eldredge, 1994; Wilson, 1993). A number of factors (training requirements, cost, burden) may limit the use of the EDE and highlight the need for the continued development and evaluation of other assessment instrumentation. The EDE Ques- tionnaire (EDE-Q) is a self-report version of the EDE (Fairburn & Beglin, 1994). Studies of patients with bulimia nervosa have shown that, compared to the EDE, the EDE-Q yields higher estimates of objective bulimic episodes and weight and shape concerns (Black & Wilson, 1996; Fairburn & Beglin, 1994). The EDE and EDE-Q have recently been compared in over- weight patients with BED (Wilfley, Schwartz, Spurrell, & Fair- burn, 1997) and obese patients seeking bariatric surgery (Kalar- chian, Wilson, Brolin, & Bradley, 2000). Wilfley et al. (1997), in their study of 52 overweight BED patients, reported that the EDE and EDE-Q produced frequencies of binge eating that were not significantly correlated and that differed significantly in magnitude (patients reported higher frequencies of binge eating days on the interview). The four EDE and EDE-Q subscales (Dietary Re- straint, Eating Concern, Weight Concern, and Shape Concern) 318 BRIEF REPORTS were significantly correlated (rs ranged from .63 to .69). Although significantly correlated. the EDE-Q subscale scores were signifi- cantly higher than the corresponding EDE scores. Wilfley et al. (1997) suggested that binge eating frequency may be difficult to assess because, unlike bulimia nervosa, binges are not sharply distinct from severe restraint or abruptly terminated by purging. and they are embedded within a more amorphous pattern of chaotic overeating. Kalarchian et al. (2000). in their study of 98 obese bariatric surgery candidates, found that the EDE and EDE-Q frequencies of binge eating were significantly correlated and were not significantly different in magnitude. The four scales of the EDE and EDE-Q were significantly correlated (rs ranged from .60 to .77), but the questionnaire produced significantly higher sever- ity scales. The discrepancies between the studies dictate the need for continued investigation of the assessment methodologies. Both the EDE and EDE-Q rely on retrospective recall of events. Recall of events can be influenced by many factors including, for example, subsequent or intervening (similar) events (Bradburn, Rips. & Shevell. 1987), subsequent (different) events that influ- ence mood or trigger reevaluation of previous events (Holmberg & Holmes, 1994), individuals' attempts to create meaning for the events (Ross, 1989), heuristic strategies (Bradburn et al., 1987), or mood states at the time of retrospective recall (Hodgins, el- Guebaly. & Armstrong. 1995). Moreover, retrospective recall of events and attributions can change over time (McKay, O'Farrell, Maisto, & Connors. 1989). One approach to providing additional data regarding ongoing behaviors is the use of self-monitoring (Wilson & Vitousek, 1999). Although not a panacea, self-monitoring can eliminate some bi- ases, including time and pure memory recall factors (Shiffman, Hufford. Hickcox. Paty, Gnys, & Kassel, 1997; Stone & Shiffman, 1994). Only one study (Loeb, Pike, Walsh, & Wilson, 1994) has compared the EDE with self-monitoring in outpatients with bu- limia nervosa. Seven days of prospective self-monitoring of binge eating and vomiting were significantly correlated with the frequen- cies generated by the EDE for the previous 28 days at baseline. At posttreatment. 28-day prospective self-monitoring of binge eating and vomiting frequencies were significantly correlated with EDE frequencies. In this study we examined different methods for assessing the features of eating disorders in patients with BED. The aim was twofold: first, to compare the EDE interview and the self-report EDE-Q. and second to compare the self-report EDE-Q with prospective self-monitoring of eating behaviors. Participants Method Participants were 82 adults who were consecutively evaluated for an outpatient randomized controlled clinical trial and met DSM-IV criteria for BED. Participants were aged 21 to 58 years (M = 42.1, SD = 9.8). 82% were female. 89% were Caucasian, and 60% were married. Mean current body mass index (BMI: weight [kg] divided by height [m] squared) was 34.7 (SD 8.9). Written informed consent was obtained from participants. DSM-IV diagnoses were derived by consensus and based on the inde- pendent and reliable (interrater reliability K = 1.0 for BED) administration of the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I; First. Spitzer. Gibbon. & Williams, 1996) and a clinical eating disorder interview conducted by trained and monitored doctoral-level clinicians. Three SCID-I interviewers in this study were evaluated as part of a larger interrater reliability study involving 12 interviewers (Zanarini et al., 2000). Interrater reliability (calculated using 84 pairs of raters for videotaped interviews) as reflected by coefficient kappa-for eating disorders diag- noses was .77 for all raters and was 1.0 for the three interviewers in this study. BED diagnoses were further confirmed by relevant portions of two instruments: the EDE (described below) and the Questionnaire on Eating and Weight Patterns-Revised (QEWP-R: Yanovski. 1993). The QEWP-R, used in the DSM-IV. assesses each criterion of BED. The QEWP-R assesses both the inclusion requirements (including the stipu- lated 6-month duration) and exclusion features (e.g., inappropriate com- pensatory practices). The QEWP-R has demonstrated adequate validity (Nangle, Johnson, Carr-Nangle, & Engler, 1994) and has been used widely (e.g.. Gladis, Wadden, Foster, Vogt, & Wingate, 1998). Measures The EDE interview (12th ed.; Fairburn & Cooper, 1993), a semistruc- tured investigator-based interview currently regarded as the gold standard in the assessment of eating disorders and their associated features (Grilo, 1998), was administered. The EDE focuses on the previous 28 days, except for the diagnostic items, which are rated for additional duration stipula- tions. For BED, following Wilfley et al. (1997), we modified the EDE to include questions designed to assess the DSM-IV requirement of a 6-month time frame. The EDE assesses the frequency of overeating broken down into objective bulimic episodes (i.e., large quantities of food coupled with subjective loss of control), subjective bulimic episodes (i.e., subjective loss of control while eating a quantity of food not judged to be large given the context), and objective overeating episodes (i.e., overeating without a subjective loss of control). The EDE also comprises four subscales: Dietary Restraint, Eating Concern, Weight Concern, and Shape Concern. The Dietary Restraint subscale reflects attempts to restrict food intake to influence weight and shape; the Eating Concern subscale reflects the degree of concern about eating; and the Weight Concern and Shape Concern subscales measure the degree of concern about weight and shape. respectively, and the degree to which these concerns unduly influence self-evaluation. Items are rated on 7-point forced-choice scales (0-6), with higher scores reflecting greater severity or frequency. We report here total scores for each subscale. Initial training of the doctoral-level interviewers and ongoing monitoring of the interviews was performed by Carlos M. Grilo, who had received training from one of the developers of the EDE (Christopher G. Fairburn). The training protocol involved review of the EDE manual, review of EDE tapes, practice administration, and review. To protect against drift, ongoing monitoring of the EDE interviews occurred (Grilo reviewed audiotapes of the first 20 assessments followed by ongoing random review of 20% of assessments) as did regular discussion of any ambiguous cases or items. G. Terence Wilson was contacted if problematic questions developed. The EDE-Q (Fairburn & Beglin, 1994) is the self-report version of the investigator-based EDE interview. The EDE-Q generates the same over- eating frequency data and the four subscales. Unlike Wilfley et al. (1997), we did not alter the EDE-Q questions to determine the frequency of binge days instead of binge episodes. The rationale is that-as in research in bulimia nervosa-determining how many binges occurred on a particular day is important, in addition to whether the patient binged on a particular day. Self-monitoring of overeating behaviors was done prospectively using daily record sheets. Participants were provided seven daily record sheets in a stapled packet each week. Each daily record inquired specifically whether participants had experienced any of the three types of overeating episodes (objective bulimic episodes, subjective bulimic episodes, and objective overeating) and, if so, how many episodes. For each form of overeating, the BRIEF REPORTS 319 daily record contained the definition (based on the EDE) that was reviewed coxin's matched pairs signed rank sum test were used to test for statistical with participants at the start of the prospective monitoring. significance in the associations and mean differences, respectively. Procedure At baseline, the 82 participants completed the EDE-Q prior to being administered the EDE interview. Participants were not told that they would be administered the two measures using different formats. We chose not to counterbalance the administration sequence and to administer the EDE-Q first followed by the EDE. This approach, which is consistent with the previous studies (Black & Wilson, 1996; Fairburn & Beglin, 1994; Kalar- chian et al., 2000; Wilfley et al., 1997), is unlikely to affect participants' understanding and definitions of items and therefore unlikely to influence their ratings. Thus, Comparison I involved an examination of the EDE-Q (administered first) and the EDE Sixty-six participants were involved in the second part of this study (Comparison 2), which compared 4 weeks of prospective daily self- monitoring with the EDE-Q. The 66 participants in Comparison 2 were all participants who were available at the second assessment point and thus completed the EDE-Q a second time, 4 weeks later. The remaining participants were excluded because they were not available for the 4-week assessment (i.e., they either chose not to participate and enroll in the trial or dropped out of the trial) rather than because of noncompliance. Participants were instructed to prospectively self-monitor the three types of overeating episodes (objective bulimic episodes, subjective bulimic episodes, and objective overeating) on a daily basis in the daily records provided (described earlier). Participants were told that in 4 weeks they would be given a battery of self-report questionnaires to complete. Weekly meetings were held by doctoral-level research clinicians with participants at which time self-monitoring records were collected and checked for completeness. Participants were reminded each time of the importance of doing the self-monitoring on an ongoing and daily basis. Analyses For normally distributed data (EDE and EDE-Q subscales), we used Pearson's coefficient to test strength of associations and the Student's paired test to test whether the means differed significantly. For non- normally distributed data (overeating behaviors), Kendall's tau-b and Wil Results Comparison 1: EDE Versus EDE-Q In Table 1 is a summary of the findings of the EDE and EDE-Q. administered at baseline. The EDE and EDE-Q frequencies of objective bulimic episodes reported for the previous 28 days were significantly correlated; the mean difference was not statistically significant. In contrast, the EDE and EDE-Q frequencies for subjective bulimic episodes and objective overeating episodes were not significantly correlated. The mean differences in over- eating frequency were significant, with the EDE producing a higher frequency than the EDE-Q. The four subscales of the EDE and EDE-Q were significantly correlated, with correlations rang- ing from .33 (Eating Concern) to .69 (Weight Concern). However, the mean differences in scores was significant for the four sub- scales, with a consistent pattern of higher scores on the EDE-Q than on the EDE. Comparison 2: Prospective Daily Self-Monitoring Versus EDE-Q In Table 2 is a summary of the findings of the 4 weeks of prospective daily self-monitoring and the EDE-Q readministration at 4 weeks. The prospective daily self-monitoring and retrospec- tive EDE-Q produced significantly correlated frequencies of ob- jective bulimic episodes that did not differ significantly. Although the two methods produced significantly correlated frequencies of subjective bulimic episodes, Wilcoxon's matched-pairs signed rank sum tests revealed that the self-monitoring produced signif- icantly higher frequencies. Frequencies of overeating generated by the two methods were not correlated; the self-monitoring produced significantly higher frequencies of objective overeating than did Table 1 Comparison of the EDE and EDE-Q on Overeating Behaviors and Subscales EDE EDE-Q Difference Behavior and subscale M SD M SD MSD Overeating behaviors Objective bulimic episodes Subjective bulimic episodes Objective overeating episodes 20.4 11.9 17.8 11.6 4.8 9.4 4.3 6.1 14.7 2.5 -2.7 12.3 8.1 -0.5 11.2 -.06 .29*** -1.71 -0.57 8.5 -3.5 15.8 -1 -2.10 Subscales Dietary Restraint 9.2 5.7 13.7 7.7 4.5 5.6 .69*** 7.27*** Eating Concern 13.1 18.7 19.5 6.3 6.5 17.6 .33** 3.33*** Weight Concern 16.8 6.3 21.1 29.0 II.I 39.5 5.4 4.3 4.8 .66*** 8.5 10.5 9.5 56*** 8.05*** 10.03*** Shape Concern Note. N = 82 pairs. EDE = Eating Disorder Examination (12th ed.) interview; EDE-Q = Eating Disorder Examination-Questionnaire self-report version. *EDE-Q rating EDE rating. Kendall's tau-b is reported for the three forms of overeating behaviors (non-normally distributed) and Pearson's r is reported for the four EDE/EDE-Q subscales. Wilcoxon's matched-pairs signed rank sum test (z scores) were used to compare the EDE and EDE-Q scores for the three overeating behaviors (non-normally distributed), and the Student's paired r test (df = 81) were used for the normally distributed subscale scores. *p 320 Table 2 Comparison of the EDE-Q and Self-Monitoring Overeating behavior Objective bulimic episode Subjective bulimic episode Objective overeating episode BRIEF REPORTS EDE-Q Self-monitoring Difference Kendall's Wilcoxon M SD M SD M SD test 9.1 10.4 10.0 11.9 -0.9 6.9 .60*** -1.16 6.2 9.3 13.2 17.0 -6.9 II.1 49*** -5.40*** 0.6 1.3 5.0 6.5 -4.4 6.5 .16 -5.27*** Note. N 66. EDE-Q Eating Disorder Examination-Questionnaire self-report version. "EDE-Q rating self-monitoring rating. Kendall's 7b is reported (instead of Pearson's r) for EDE-Q scores because of non-normality in the distributions. Wilcoxon's matched-pairs signed rank sum test (score) was used to compare the subscale scores of the EDE-Q and self-monitoring. the EDE-Q. Whereas 31.8% (n = 21) of participants reported no overeating episodes on the self-monitoring forms (direct question about overeating). 78.8% of participants (n=52) reported no overeating episodes on the EDE-Q (i.e.. all overeating episodes reported were classified as binge episodes because of associated loss of control). Discussion In this study we compared different methods for assessing the features of eating disorders in patients with BED. An investigator- based semistructured interview (EDE), self-report questionnaire (EDE-Q), and prospective daily self-monitoring of different forms of overeating showed some areas of acceptable convergence, most notably with regard to objective bulimic episodes (binge eating). The EDE-Q appears to have utility for assessing objective bulimic episodes and, to a lesser extent, the eating disorder attitudinal psychopathology (EDE scales), but not for assessing subjective bulimic episodes or objective overeating in patients with BED. Our findings also suggest the potential importance of self-monitoring for assessing overeating behaviors. The EDE and EDE-Q were significantly correlated on frequen- cies of objective bulimic episodes reported for the previous 28 days at the first assessment, and the magnitude of the difference was not significant. The EDE-Q was significantly correlated (with no significant mean difference) with prospective self-monitoring at the second assessment 4 weeks later. This consistency across methods and time periods suggests that the EDE-Q may have utility for assessing binge eating (objective bulimic episodes) in patients with BED. These findings are consistent with those of Kalarchian et al. (2000) but inconsistent with the results of Wilfley et al. (1997), who observed poor agreement between the EDE and EDE-Q on the frequency of days with objective bulimic episodes. Wilfley et al. relied on the DSM-IV criterion for BED, counting days on which binges occurred, whereas our approach assessed the fre- quency of binge episodes. Except for this one methodological difference, our study was quite similar to that of Wilfley et al-that is, DSM-IV BED patients were rigorously assessed with semistructured diagnostic interviews, similar recruitment ap- proaches were used (i.e.. for controlled clinical trials at university centers), and participants had similar demography and BMI and comparable levels of eating-disordered symptomatology. The DSM-IV research criterion for binge eating involves days, not episodes, in accordance with clinical lore that it is difficult for patients to remember episodes. Our data suggest that this needs to be considered an open question for empirical work. Moreover, this example points to the need for researchers to assess BED psycho- pathology beyond the working formative definition in the DSM- IV. The respective merits of days versus episodes as the unit of measure warrants continued investigation. In contrast to the findings for objective bulimic episodes, the EDE-Q does not appear to have utility for assessing other forms of overeating. The EDE and EDE-Q frequencies for subjective bu- limic episodes and objective overeating episodes were not signif- icantly correlated. Moreover, the magnitude of the difference in overeating frequency was significant, with the EDE producing a higher frequency than the EDE-Q. The difference in findings is particularly striking for objective overeating episodes: Participants rarely reported on the EDE-Q overeating episodes without loss of control. Similarly, the EDE-Q was not correlated with prospective self-monitoring for either subjective bulimic episodes or for ob- jective overeating episodes. These findings suggest that sole reliance on the EDE-Q may produce substantial underestimates of objective overeating. We can only speculate that without more specific prompting that it may be difficult for patients to readily recall such overeating episodes that are not as salient as the objective bulimic episodes. The current format of the EDE-Q inquires about overeating epi- sodes first and only then addresses loss of control. Our findings suggest the need to more concretely or specifically ask about the two types of episodes more directly, as in the EDE. The EDE and EDE-Q were significantly correlated on the four subscales, but the EDE-Q resulted in significantly higher scores for each of the subscales. These findings replicate those previously reported by Wilfley et al. (1997) for obese persons with BED and Kalarchian et al.'s (2000) results for obese persons seeking surgi- cal treatment. This finding is consistent with that found in the assessment literature for BED (e.g., Gladis et al., 1998; Greeno, Marcus, & Wing, 1995) as well as for many other domains of psychopathology (i.e., self-report inventories tend to generate higher frequencies than semistructured interviews; see Zimmer- man, 1994). It is worth noting that such variability is not limited to self-reports versus investigator interviews. For example, variabil- ity has been found between findings generated by two different independently administered semistructured diagnostic interviews for personality disorders even when administered by experienced research clinicians (e.g., Oldham et al., 1992). BRIEF REPORTS It is not possible to determine which method generates the most valid estimates of BED features, because there is no objective measure. The eating disorder field has increasingly moved in the direction of viewing the EDE as one gold standard (Grilo, 1998; Wilson, 1993). Previous psychometric work reported good internal consistency, discriminant validity, and concurrent validity (Fair- burn & Cooper, 1993; Rosen, Vara, Wendt, & Leitenberg. 1990; Wilson, Nonas, & Rosenblum, 1993). It must be kept in mind that the EDE, although allowing for probing and clarification of dif- ferent types of overeating episodes using specific guidelines (i.e.. to minimize between-rater variability) and using well-established timeline calendar follow-back procedures to obtain frequency es- timates, does nonetheless rely ultimately on respondents' retro- spective recall and self-report of events. Different methods have potential advantages and disadvantages. Self-report may eliminate some degree of interpersonal discomfort when disclosing sensitive and embarrassing material but may result in inflated scores. Interviews administered by experienced evaluators can make use of anchoring, clarification, and question- ing to establish frequency of events at desired thresholds. Inter- views, however, require greater cost and greater investigator and participant burden than self-reports. It is possible that some people answer more truthfully on self-reports than in direct interviews (or vice versa). Self-monitoring might eliminate some biases, including time and pure memory recall factors (Shiffman et al.. 1997; Stone & Shiffman, 1994). If the self-monitoring is done in real time, the reports are likely to be less subject to memory biases. Given these potential advantages and our findings regarding the potential for capturing data regarding other forms of overeating in addition to objective bulimic episodes, self-monitoring should receive consideration when determining comprehensive assessment. We assessed a consecutive series of adults with DSM-IV- defined BED who presented at a university-based outpatient eating disorder program for a randomized controlled clinical trial. The investigator-based assessments (SCID-I, EDE) were administered by carefully trained and closely monitored doctoral-level clinicians who were part of a research evaluation team of an eating disorder program. Our study represents a replication of Wilfley et al.'s (1997) methodology, and the analysis of the 4-week prospective self-monitoring protocol represents an incremental contribution to this literature. Several potential limitations should be considered. Our findings are based on a recruitment procedure for individuals with BED seeking treatment at a university-based eating disorder program and willing to enroll in a randomized controlled clinical trial. Thus, our findings may not be generalizable to general outpatient clinics (Wilson, 1998) or community populations (Fairburn, Welch, Nor- man, O'Connor, & Doll, 1996). Findings from previous studies that have compared the EDE with the EDE-Q provide some indication that the EDE-Q may be less accurate in community samples than in eating-disordered clinical samples (Fairburn & Beglin, 1994). Self-monitoring may also produce reactivity in the target behavior. Thus, it is possible that the self-monitoring pro- cedures used in this study enhanced remembering and thereby inflated the apparent degree of convergence with the EDE-Q. We do note, however, that this is unlikely, because no such conver- gence was observed with regard to objective overeating. It would also have been desirable to have administered the EDE again after 4 weeks to assess (a) its correspondence with the EDE-Q at 321 a second, different point in time and (b) its correspondence with self-monitoring. Only a single study (Loeb et al., 1994) has done the latter, and this should be a priority in future research. To summarize, investigator-based interview (EDE), self-report questionnaire (EDE-Q), and prospective daily self-monitoring of BED features showed some areas of acceptable convergence, most notably with regard to objective bulimic episodes (binge eating). The EDE-Q appears to have utility for assessing objective bulimic episodes and, to a lesser extent, the eating-disordered attitudinal symptomatology (EDE scales) but not for assessing subjective bulimic episodes or objective overeating in patients with BED. Our findings suggest the potential utility of self-monitoring for the comprehensive assessment of overeating behaviors in BED. References American Psychiatric Association. (1994). Diagnostic and statistical man- ual of mental disorders (4th ed.). Washington, DC: Author. Beglin. S. J., & Fairburn, C. G. (1992). What is meant by the term "binge"? American Journal of Psychiatry, 149, 123-124. Black, C. M. D., & Wilson, G. T. (1996). Assessment of eating disorders: Interview versus questionnaire. International Journal of Eating Disor ders, 20, 43-50. Bradburn, N. M., Rips, L. J., & Shevell. S. K. (1987). Answering auto- biographical questions: The impact of memory and inference on surveys. Science, 236, 157-161. Cooper. Z., & Fairburn, C. (1987). The Eating Disorder Examination: A semistructured interview for the assessment of the specific psychopa- thology of eating disorders. International Journal of Eating Disor ders, 6, 1-8. Fairburn, C. G., & Beglin, S. J. (1994). Assessment of eating disorders: Interview or self-report questionnaire? International Journal of Eating Disorders, 16, 363-370. Fairburn, C. G., & Cooper. Z. (1993). The Eating Disorder Examination (12th ed.). In C. G. Fairburn & G. T. Wilson (Eds.). Binge eating: Nature, assessment, and treatment (pp. 317-360). New York: Guilford Press Fairburn, C. G., Welch. S. L... Norman, P. A., O'Connor. M. E., & Doll. H. A. (1996). Bias and bulimia nervosa: How typical are clinic cases? American Journal of Psychiatry, 153, 386-391. First, M. B., Spitzer. R. L., Gibbon. M., & Williams, J. B. W. (1996). Structured Clinical Interview for DSM-IV Axis I Disorders--Patient Version (SCID-I, Version 2.0). New York: New York State Psychiatric Institute. Gladis, M. M., Wadden. T. A.. Foster, G. D., Vogt. R. A., & Wingate, B. J. (1998). A comparison of two approaches to the assessment of binge eating in obesity. International Journal of Eating Disorders, 23, 17-26. Greeno, C. G., Marcus, M. D., & Wing. R. R. (1995). Diagnosis of binge eating disorder: Discrepancies between a questionnaire and clinical interview. International Journal of Eating Disorders. 17. 153160. Grilo, C. M. (1998). The assessment and treatment of binge eating disorder. Journal of Practical Psychiatry and Behavioral Health, 4, 191-201. Grilo, C. M., Shiffman. S., & Carter-Campbell, J. (1994). Binge eating antecedents in normal weight-weight non-purging females: Is there consistency? International Journal of Eating Disorders, 16, 239-249. Heatherton. T. F., & Baumeister, R. F. (1991). Binge eating as an escape from self-awareness. Psychological Bulletin. 110, 86-108. Hodgins, D. C., el-Guebaly, N., & Armstrong, S. (1995). Prospective and retrospective reports of mood states before relapse to substance use. Journal of Consulting and Clinical Psychology, 63, 400-407. Holmberg. D., & Holmes, J. G. (1994). Reconstruction of relationship memories: A mental models approach. In N. Schwarz & S. Sudman (Eds.). Autobiographical memory and the validity of retrospective re- ports (pp. 267-288). New York: Springer-Verlag.