@Pergamon

Archives of Clinical Neuropsychology,Vol. 12, No. 7, pp. 635443, 1997Copyright@ 1997 National Academy of Neuropsychology

Printed inthe USA. AIl rights reserved0887-6177/97 $17.CO+.00

PIIS0887-6177(96)OO046-7

Accuracy of Self-Reported Educational

Attainment Among Diverse Patient

Populations: A Preliminary Investigation

Doug Johnson-Greene, Michael Dehring, Kenneth M. Adams, Todd Miller,Shalini Arora, Anna Beylin, and Rochelle Brandon

Departments of Psychiatry and Psycholog5 University of Michiganand Department 01”Veteran’s Affairs Medical Centers, Ann Arbo5 Ml

Despite speculation concerning t)le accuracy of self-repotied information, particularly frrrm certainpatient populations,many neumpq,chologistscontinueto estimateprernorbidintellectualfunctioning onthe basis of self-reported educati{malattainment. This study examined 116 individuals with diversediagnoses [i.e., alcoholism,posttrownatic stress disorder (PTSD), schizophreniaor schizoaffective,anddementia] to determine the accuracy of their selfreported high school educationalattainment. Resultssuggest that at least tuzlfof all participants were inaccurateas defined by discrepanciesbetween actualand estimated GPAgreaterthan .5 on a traditional4-pointgrading scale. Most patients were inaccuratein the directionof overestimatingtheir educationalattainment Patients diagnosedwith alcoholism andPTSD were signlfZcantlyless accurt~tein recallingtheir educationalhistorywhen comparedto a groupofnormal-controlsubjects. Several sltbjects, whose recordscoufd not be verified, werefound to have notattended high school as they had claimed. These results urrderscorzthe potential inaccuracythat existswhen estimatingpremorbid intelligenceusing self-reportedinformation. O 1997 National Academy ofNeuropsychology.Published by Ekevier Science Ltd

Clinical lore, as well as recent literature, suggest that some patients inaccurately reportfactual and personal information.. Information provided by these patients may range fromgross inaccuracies, such as denyi ~gprevious psychiatric hospitalizations or reporting collegedegrees that were not earned, to minimal distortions of current occupational performance orsocial well-being. Inaccurate information may greatly hinder the assessment of psychiatric,medical, and neuropsychologica. illnesses to the extent that these diagnoses presuppose acomplete, honest, and accurate patient self-report (Rogers, 1988).Also, it may lead clinicians

The authors wish to thank Dr. Loren Pankratz from the Oregon Health Sciences University and Dr. LisaJohnson-Greene from the University of ?rfichiganfor their helpfnl feedback on earlier versions of this manuscript.We also wish to thank Kevin Dehring, HollyJo Sparks and Melanie Topper for their assistance with data collection.This study was supported in part by NU,AA training grant T32-AA07477.

Address correspondence to: Doug Johmon-Greene, Ph.D., Department of Physical Medicine and Rehabilitation,Johns Hopkins University School of Mtdicine, Good Samaritan P.O.B. Suite 406, 5601 Loch Raven Boulevmd,Baltimore, MD 21239; E-mail: Johnsong@welchlink.welch.jhu.edu.

635

636 D. Johnson-Greene et al.

to form erroneous conclusions regarding diagnostic entities, adaptive functioning, prognosis,and performance on psychological tests. Inaccurate reports are particularly important topracticing neuropsychologists in that assessment of premorbid functioning is dependent uponaccurate self-reported information.

Investigators have hypothesized that some patient populations, particularly those withneurological impairments, may have a greater tendency than neurologically intact patients toinaccurately report personal and factual events (Johnson-Greene & Binder, 1995; Sierles,1984). For example, some patients with minor head injury apparently produce deliberatelyincorrect answers in a forced-choice testing of memory complaints (Binder, 1990).Alzheimer-type dementia patients have also been found to frequently confabulate, probablyas a result of faulty ml>mory(Mangone, Hier, Gorelick, Ganellen, Langenberg, Boarman, &Dollear, 1991). One study found that the frequency of confabulation exhibited by hydro-cephalic dementia patients is related to the stage of their illness (Berglund, Gustafson, &Hagberg, 1979).

Inaccurate self-reported information is also associated with several psychiatric illnesses.For example, researchers have found that there is a high frequency of confabulation amongpatients diagnosed with schizophrenia (Sokolova, 1969; Wilson, Cockburn, & Baddeley,1985; Yen, 1983). Hyer, Woods, Harrison, Boudewyns, and O’Leary (1989) found thatposttraumatic stress disorder (PTSD) patients tended to overreport symptoms, which led theauthors to recommend inclusion of symptom exaggeration as part of the symptom cluster forthis disorder. Patients who abuse alcohol are known to understate their drinking history andmay respond inaccurately on inventories designed to assess the extent of their alcohol use(Wasyli, Haywood, G1ossman, & Cavanaugh, 1993). In addition, patients who abuse alcoholmay have high self-de:eption, particularly about their ability to control their drinking (Strom& Barone, 1993). Sclbell and Sobell (1990) suggest that clinicians cautiously interpretself-reports of patient:s with addictive behaviors. Patients with somatoform disorders inac-curately report symptoms and signs of their physical and mental illness, often with little orno known conscious motivation (Carney & Brown, 1983).

Neuropsychological test results are frequently presented as evidence in litigation cases(Leckliter & Matarazzo, 1989). However, neuropsychological test results may be interpretedinaccurately if patients perform below their optimal ability level. Several researchers havereported that patients’ dissimulation on neuropsychological tests may contaminate the accu-rate assessment of cognitive abilities of litigants involved in medico-legal cases (Cavanaugh& Rogers, 1984; Faust, Hart, & Guilmette, 1988a, 1988b). For example, symptomaticexaggeration by personal injury litigants was shown to interfere with the interpretation ofTrail Making Test performance (Lees-Haley & Fox, 1990).

Determination of premorbid intellectual ability is of paramount importance for theaccurate assessment cf a patient’s performance on neuropsychological tests, and is often acrucial piece of information in medico-legal cases. Conversely, interpretation of neuropsy-chological test performance may be inaccurate if patients provide inaccurate personal orhistorical information. Some neuropsychologists may turn to collateral sources to verifyself-reported credentials. However, collateral sources may also be inaccurate in their recall ofspecific information pertaining to patients (Platt, 1980; Rankin, 1990). Neuropsychologistshave developed regression equations that utilize demographic variables in order to estimatepremorbid IQ scores. However, these formulas have been found to predict the premorbid IQscores of normal psychiatric and brain-damaged patients with a low degree of success(Sweet, Moberg, & T~vian, 1990).

Education has been described as the strongest single predictor of premorbid intelligence(Matarazzo, 1972;Wilson,Rosenbaum,Brown, & Grisell, 1978).To the extent that educationhasbecome an important variable for determining premorbid intellectual capacity and for gauging

Accuracy of Self-Reported Education 637

TABLE 1Subject Characteristics

Schizophrenia and Normal UnverifiedDementia Scllizoaffective Alcoholism PTSD Controls Records Totals

Mean Age 57 (19) 39 (11) 49 (lo) 44 (5) 49 (8) 49 (12) 48 (13)Gender

Male 5 8 19 19 10 13 74Femate 13 11 1 0 13 4 42

RaceAfrican American o 3 3 1 0 4 11Caucasian 18 16 17 17 21 13 102Other o 0 0 1 2 0 3

OccupationUnemployed 11 11 6 9 5 6 48Unskilled o 3 0 2 0 3 8Semiskilled 1 2 6 4 3 6 22Skilled 2 0 5 1 3 1 12Managerial 1 2 3 1 4 1 12Professional 3 1 0 2 8 0 14

Years of education10-11 0 2 3 3 2 512 5

156 8 8 8 5 40

13–15 7 7 7 8 4 7 4016+ 6 4 2 0 9 0 21

Note. Vahres enclosed in parentheses re~resent standard deviations.

neuropsychologicalperformance,it is essential that clinicianshave accurate informationconcer-ningtheir patients’ educational history. A strong correlation has been found between formaleducation and performance on nwropsychological measures (llnlayson, Johnson, & Reitan,1977).Furthermore, a patient’s years of education has become part of the normative referencetables for the Halstead-Reitan Battery (Heaton, Grant, & Matthews, 1991).

The accuracy of patients’ self-reported personal information, including educational history,should be verified in view of the z’orementionedliterature,which suggests that many patients donot accurately report personal and historical information. However, neuropsychologistsdo notroutinely verify the accuracy of patients’ self-reportededucational attainment. Instead, neurop-sychologistsrely heavily upon pat[ents’unverifiedreports of educational attainment as an indexof premorbid intellectual functioning.As previouslydescribed, severalpatient groups are knownto frequently dissimulate factual information (Binder, 1990, 1992; Lorei, 1970; Sierles, 1984).Thus far, there have been no empirical investigationsexaminingeducationalexaggerationamongpatient populations shown previously to have an increased likelihood of distorting factualinformation.The goal of the present study was to assess the accuracyof self-reportededucationalattainment among specific patient populationsand to provide normative informationconcerningthe incidence of educational exaggeration.

METHOD

Subjects

Specific inclusion criteria for this study was at least 2 years of high school attendance,English as a native language, and a primary psychiatric diagnosis of dementia, alcohol abuse,schizophrenia spectrum illnesses (i.e., schizophrenia or schizoaffective disorder), or FTSD.From an initial sample of 124 subjects, 116 agreed to participate in this study. Among theparticipants were 18 diagnosed with mild dementia, 19 with schizophrenia or schizoaffective

———.

638 D. Johnson-Greene et al.

disorder, 20 with prolonged chronic alcoholism, 19 veterans with PTSD, and 23 communitydwelling normal-control volunteers. Academic records for 17 subjects could not be verifiedafter extensive telephone and written inquiries with state and local school boards. A total ofeight subjects who met the criteria listed above and who were solicited by the researchersrefused to participate irl the study (four with PTSD, two with schizophrenia or schizoaffectivedisorder, and two with dementia), usually because of scheduling conflicts with hospital wardactivities.

All patients with F’TSD and schizophrenia, and some patients with alcoholism, werehospitalized during their participation in the study. Hospitalized patients were studied at ornear the end of their inpatient care in order to increase the probability that their cognitivefunctioning was at or :nearbaseline at the time they participated in the study. All dementiapatients had a score of 18 or higher on the Mini Mental State Examination (MMSE; Folstein,Folstein, & McHugh, 1978) and were referred by a neurologist from a hospital-basedoutpatient clinic for evaluation of cognitive disorders. All patients met diagnostic criteria fortheir respective diagnclses according to DSM-ZV standards (American Psychiatric Associa-tion, 1993), and they were evaluated prior to their participation in the study by a boardcertified psychiatrist or neurologist as part of their medical work-up.

Procedure

Subjects read and signed and informed consent prior to participation. The consent clearlystated that the purpose of the study was to evaluate self-reported educational credentials, andinvolved signing a release allowing the investigators to obtain their high school transcript.

After a brief tutorial explaining the traditional 4-point grading scale and the associatedletter grades, subjects were asked to estimate their high school grade point average (GPA) tothe nearest tenth on a standard 4-point scale, as well as provide demographic informationsuch as age, gender, race, education, and occupation status (see Table 1). Subjects signed arelease of information form authorizing the researchers to obtain their academic high schooltranscripts, which wew then sent to the high school the subject reported attending. Highschools that did not respond to transcript release forms were contacted by phone to verify thesubject’s attendance, and in cases where the school was no longer in existence, duplicateletters were sent to the district school board.

Upon receipt of the academic transcript, a comparison was made between the subject’sself-reportedGPA and the actual GPAindicated on the academic transcript. In several instanceswhere subjects grades were not based upon a 4-point scale, rating scales were transformed to a4-point scale by multiplyingthe ratio of subjects’ actual and maximum school ratings by four. Inthis manner,a student with an averagerating of 6 in a schooldistrictwith and 8-pointscale wouldearn a 4-point scale rating of 3.0 (x=4 X (6/8).Differencescoreswere computedfor each subjectby subtracting self-reportedfrom actual GPAs. Difference scores were then separated into twogroups, including a high GPAdiscrepancygroupfor differences=.5 on a 4-point scale and a lowGPA discrepancy group consisting of differences <.5. Spearman rank order correlations werecomputedbetween subjects’self-reporteddemographicvariablesand their GPAdifferencescores.Student’s t tests were computed between the normal-control group and each patient group formean GPA discrepancy

The hypotheses for this investigation were threefold. First, it was expected that at least25% of the entire subject sample would show a high discrepancy (i.e., a >0.5 GPAdiscrepancy) between their self-reported and actual GPAs. Second, normal-control subjectswould more accurately recall their educational history than subjects from patient groups.Finally, patient groups previously shown in the literature to inaccurately report clinical and

Accuracy of Self-Reported Education 639

TABLE 2Group Means and Standard Deviations for Estimatsd,Actual, and DifferenceGPA

Estimated GPA Actual GPA Difference GPA

Patient Type M SD M SD M SD

Entire sample 2.8 0.7 2.3 0.8 0.5 0.6Dementia 3.1 0.7 2.6 0.9 0.5 0.6Schizophrenia 2.9 0.6 2.5 0.9 0.3Alcohol

0.62.5 0.8 1.8 0.8 0.7 0.9

PTSD 2.4 0.5 1.8 0.5 0.7 0.5Normal-Control 3.0 0.6 2.7 0.6 0.3 0.4

historical information (i.e., alcohol abuse and PTSD patients) would be less accurate inrecalling their high school GPA when compared to normal-control subjects.

RESULTS

We were able to obtain academic transcripts for 99 (85%) of our subjects. Academicrecords for 17 subjects could not be verified (5 with P’TSD, 6 with schizophrenia orschizoaffective disorder, 1 with dementia, and 5 with alcoholism). Through verification withrelatives and/or medical records, the researchers found that four subjects whose records couldnot be obtained reportedly never attended high school as they had initially claimed.

Review of the subjects’ transcripts revealed that 46Y0had a high discrepancy betweenactual and self-reported GPA (i.:., GPAdiscrepancy >.5). The remaining 54% of subjects are -in the low GPA discrepancy group. The high discrepancy group consisted of 8 dementiapatients, 6 patients with schizophrenia or schizoaffective disorder, 11 patients with severechronic alcoholism, 10 patients with PTSD, and 8 normal-control subjects. GPA discrepan-cies greater than 1.0 most often occurred for both alcohol and PTSD patients. A total of sevenpatients with alcoholism and six with PTSD had GPA discrepancies greater than 1.0,compared with only one patient each for the dementia, schizophrenia/schizoaffective, andnormal-control groups. Nearly ill instances of GPA discrepancies greater than 1.5 consistedof patients with alcoholism (se: Table 2).

Four t tests were computed, one for each patient group contrasted with the normal-controlgroup. Compared to the normal-control group, patients with alcoholism (t= 2.02, p < .05)and PTSD (t = 3.03, p < .004} showed a greater discrepancy between reported and actualGPA. In contrast, patients witk dementia (t = .14, p < .89), or schizophrenia or schizoaf-fective disorder (t = .96, p <: .34) did not differ significantly in terms of mean GPAdiscrepancy in comparison to normal-controls, though GPAdiscrepancy for these groups wasslightly higher. Approximately :.690of subjects (n= 16)underestimated their actual GPA, andonly 3 of 99 subjects underestimated their actual GPA by greater than .5 (2 patients withalcoholism and 1 with schizophrenia or schizoaffective disorder).

Spearrnan rank order correlations were computed for demographic variables and subjects’GPA difference score. An inverse relationship was found between self-reported years ofeducation and subjects’ GPA difference score, which suggests that subjects with more yearsof education more accurately reported their high school GPA (r~ = –.21, p < .04). Nosignificant correlations were found for other demographic variables of interest including age,education, sex, and race.

640 D. Johnson-Greene et al.

TABLE 3Number of Subjects with Progressively Higher GPA Discrepancy

Level of Discrepancy

Patient Tvoe >.5 >1.0 >1.5

Entire sample 43 16 7Dementia 8 1 1Schizophrenia 6 1 1Alcohol 11 7 5PTSD 10 6 0Normal-control 8 1 0

DISCUSSION

In general, subjects in this investigation do not appear to be reliable historians regardingtheir educational history. We found that nearly half of all subjects demonstrated pooraccuracy in their recollection of educational credentials as measured by the discrepancybetween their actual and self-reported GPA. This finding is significantly greater than the 25%discrepancy originally predicted by the authors. Subjects’ inaccurate self-report was almostalways in the direction of inflating their academic achievement. In contrast, level ofeducational attainment (i.e., years of education completed) was accurate for all subjects forwhom we could obtain transcripts. These findings are particularly noteworthy given that theinformed consent document read by each subject prior to their participation specified thattheir academic high school transcript would be requested. To the extent that some subjectsself-report might have been consciously exaggerated, inaccuracy may increase in situations

- when there is no expectation by the patient that their self-reported educational attainment willbe verified. These findings are generally consistent with a rather large body of literaturecriticizing the validity of self-reported information.

It was also hypothesizedthat normal-controlsubjectswould more accuratelyrecall their actualGPA when compared 1:0patient groups, which the data confirms. Statistically significant differ-ences were found between the normal-controlgroup and patients diagnosedwith alcoholism andPTSD. A trend was also found between dementia patients and normal controls, although, thisdiscrepancy was not statistically significant. It is well known that dementia patients exhibitprofound memory deficits. The results of this study suggest that patients with dementia, despitehavingpoorer memory and attendingschoolbefore most patients from other subjectsgroupswereborn, more accuratelyrecalled their educationalcredentialscompared to patients with alcoholismand PTSD. This finding is consistent with literature that suggests some patient groups are morelikely to provide inaccurate or misleading personal or factual information (Binder 1990, 1992;Johnson-Greene& Binder, 1995).

A significant correlation was found between educational attainment and accuracy ofself-reported educational history. No other demographic variables were significantly corre-lated with educational discrepancy, suggesting that age, gender, occupation and race are notexplanatory in describing educational discrepancies for these patient groups.

Transcripts for subjects whose records could not be verified were unavailable for a varietyof reasons. Records cnn be damaged or misplaced over time. At least one school that wascontacted reported that their records had been damaged many years ago in a fire. On the otherhand, it is probable that some patients in the study had unverifiable records because they didnot attend high schoo[. The last scenario was verified through independent corroboration forfour of the patients in the study. It is presumed that other subjects did not attend high schoolas they had reported either, though we were unable to obtain corroborating information to

Ac(:uracy of Self-Reported Education 641

definitively confirm this. For emrnple, five additional subjects reported attending schoolsthat could not be located and fcr which the school board for that state had no record of itsexistence, raising the possibility that the actual number of subjects inaccurately reporting thatthey had attended high school is much higher.

There are probably a variety of reasons that patients, and nonpatients, overestimate theireducational accomplishments. hnpaired memory, personality factors, psychiatric illness, andslight distortions of reality may ;dl lead to decreased accuracy. It is also possible that subjectswished to appear more accomplished to the researchers and to others around them and, as aresult, exaggerated their self-reported educational attainment. Nonetheless, the results of thisstudy suggest that some individuals may also purposefully exaggerate their educationalattainment, as at least four subjects had done in this study.

By convention, or perhaps as a result of limited resources or time constraints, it is theauthors’ impression that some ]leuropsychologists simply rely upon patients’ self-reportedinformation concerning educational achievements for estimating premorbid intelligence.Kareken and Williams (1994) have recently shown that clinicians overestimate the predictivevalidity of demographic variablas when estimating premorbid intelligence. Inaccurate edu-cational information, if used to estimate premorbid intelligence as part of a demographic-based regression equation or by itself, may result in inappropriate interpretation of neurop-sychologicai tests. Obviously, there are important implications of these findings forneuropsychologists who practica within the medico-legal arena. The authors are aware ofseveral practicing neuropsychok)gists who, while testifying in court, found themselves in theembarrassing situation of reinterpreting their initial conclusions on the witness stand afterdiscovering that the patient did lot have the educational background they initially reported.In many respects this study is less relevant to medico-legal cases because most practicingneuropsychologists exercise extreme caution in dealing with legal cases and are moreinclined to obtain academic tra~scripts. The findings of this study are more applicable toclinical and research situations in which neuropsychologists do not routinely verify premor-bid educational history and then subsequently use this information to assess premorbidintelligence. There is at least sonle reason to believe that the predictive validity of regression-based equations that have been developed using self-reported educational information havebeen subjected to the same inaccurate self-reports that were found in this study. Empiricalstudies of premorbid intellectual estimation using verified educational history are needed andmay yield regression-based forrmlas with greater predictive validity.

The authors encourage neuropsychologists to exercise caution when utilizing self-reportededucational information providei by patients, especially with the aforementioned diagnoses.Information that corroborates a ~atient’s self-reported educational history should be obtained,when possible, from educationid records, since inaccurate information is also sometimesprovided by corroborative sources (Platt, 1980; Rankin, 1990).

There are several potential limitations of the present study. One limitation is that the studyutilized patients diagnosed with FTSD from a VAhospital whose symptoms were proximallyrelated to combat-related experiences, which may represent a biased cohort. Divergentfindings might be expected if the PTSD group was more heterogeneous by including PTSDpatients with noncombat-related experiences. A few subjects were not graded on a traditional4-point scale. As a result these subjects may have been prone to inaccurately recall theireducational accomplishments, Ihough every attempt was made to orient patients to the4-point grading scale. Finally, tle generalizability of these findings to other neurologicallyimpaired patients should be considered tentative.

It takes considerably more eifort and time to obtain educational records, but this consci-entious approach is more likely to result in an accurate interpretation of neuropsychologicaltest results. Since our data would seem to indicate that patients have relatively poor recall of

642 D. Johnson-Greene et al.

their educational accomplishments, particularly patients with PTSD or alcoholism, it wouldbe beneficial for neuropsychologists to verify the educational background provided bypatients.

REFERENCES

American Psychiatric Association. (1993). Diagnostic and statistical manual (4th cd.). Washington, DC: Author.Binder, L. M. (1990). Malingering following minor head trauma. The Clinical Neuropsychologist, 4, 25-36.Binder, L.M. (1992). Malingering and Deception. In A. Puente & R. McCaffrey (Eds.), Handbook of neuropsycho-

logical assessment: A biojmychosocialperspective (pp. 353–374). New York: Plenum.Berglund, M. Gustafson, L., & Hagberg, B. (1979). Amnestic-confabulatory syndrome in hydrocephalic dementia

and Korsakoff’s psychosis in alcoholism. Acta Psychiatric ,Scandinavica,60, 323–333.Carney, M. W. P., & Brown, J. P. (1983). Clinical features and motives among 42 artifactual illness patients. British

Journal of Medical Psychology, 56, 57-66.Cavanaugh, J. L., & Rogers, R. (1984). Mafingering and deception. Behavioral Sciences and the Law, 2, 3.Faust, D., Hart, K., & Guilmette, T. J. (1988a). Neuropsychologists’ capacity to detect adolescent malingerers.

Professional Psychology: Research and Practice, 19, 508–515.Faust, D., Hart, K., & Guilmctte, T. J. (1988b). Pediatric malingering: The capacity of children to fake believable

deficits on neuropsychological testing. Journal of Consulting and Clinical Psychology, 56, 578-582.Finlayson, M. A. J., Johnson, K. A., & Reitan, R. M. (1977). Relationship of level of education to neuropsycho-

logical measures on brain-damaged and non brain damaged adults. Journal of Consulting and Clinical Behavior,45, 536-542.

Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1978). “Mini-Mental State” A practical method for grading thecognitive state of patients for the clinician. Journal of Psychiatric Research, 12, 189–198.

Heaton, R. K., Grant, I., & Matthews, C. G. (1991). Comprehensivenormsforan expandedHalstead-Reitan Battery.Odessa, FL: Psychological Assessment Resources, Inc.

Hyer, L., Woods, M., Harrison, W. R., Boudewyns, P., & O’Leary, W. C. (1989). MMPI F-K Index amonghospitalized Vietnam veterans. Journal of Clinical Psychology, 45, 25&254.

Johnson-Greene, D., & Binder, L. (1995). Evaluation of an efficient method for verifying higher educationalcredentials. Archives of Clinical Neuropsychology, 10, 251–253.

Kareken, D. A., & Williams, J. M. (1994). Human judgment and estimation of premorbid intellectual function.Psychological Assessment, 6(2), 83–91.

Leckfiter, I. N., & Matarazzo, J. (1989). The influence of age, education, IQ, gender, and alcohol abuse onHalstead-Reitan Neuropsychological Test Battery Performance. Journal of Clinical Psychology, 45,484-512.

Lees-Haley, P. R., & Fox, D. D. (1990). Neuropsychological false positives in litigation: Trail Making Test findings.Perceptual & Motor Skill,r,70, 1379–1382.

Lorei, T. W. (1970). Staff ratings of the consequences of release from or retention in a psychiatric hospital. Journalof Consulting and Clinical Psychology, 34, 4655.

Mangone C. A., Hier, D. B. Z., Gorelick, P. B., Ganellen, R. J., Langenberg, P., Boarman, R., & Dollear, W. C.(1991). Impaired insight in Alzheimer’s disease. Journal of Geriatric psychiatv and Neurology, 4, 189-193.

Matarazzo, J. D. (1972). Wechsler’s Measurement and Appraisal of Adult Intelligence (5tJrand enlarged cd.). NewYork: Oxford University Press.

Platt, S. (1980). On establishing the validity of “objective data”: Can we rely on cross-interview agreement?Psychological Medicine, .10,573–581.

Rankin, H. (1990). Validity of self-reports in clinical settings. Behavior Assessments, 12, 107-116.Rogers, R. (1988). Clinical assessment of malingering and deception. New York: Guilford Press.Sierles, F. S. (1984). Correlal.esof malingering. Behavioral Sciences and the Law, 2, 113-118.Sobell, L., & Sobell, M.(199D).Self-reports across addictive behaviors: Issues and future directions in clinical and

research settings. Behavioral Assessment, 12, 14.Sokolova, B. V. (1969). On acute confabualtion in schizophrenia. Zhurnal Nevropatologii i Psikhiatrii, 69,261-267.Strom, J., & Barone, D. (1993). Self-deception, self-esteem, and control over drinking at different stages of alcohol

involvement. Journal of Drug Issues, 23, 705–714.Sweet, J. J., Moberg, P. J., & Tovain, S. M. (1990). Evaluation of Wechsler Adult Intelligence Scale-Revised:

Premorbid IQ formulas in clinical populations. Psychological Assessment, 2,41-44.Wasyli, O., Haywood, T., Grossman, L., & Cavanaugh, J. (1993). The psychometric assessment of alcoholism in

forensic groups: The MacAndrew Scale and response bias. Journal of Personality Assessment, 60, 252-266.

Accuracy of Self-Reported Education 643

Wilson, B. A., Cockburn, J., & Baddel:y, A. (1985). The Rivermead Behavioral Memory Test. Reading, England:Thames Valley Test Company: Gaylord, MI: National Rehabilitation Services.

Wilson, R. S., Rosenbaum, G., Brown, G., & Gnsell, J. (1978). An index of premorbid intelligence. Journal ofConsulting and Clinical Psychology;),46, 15561555.

Yen,Y. (1983). Deviant verbalization in the Rorshach Test as indices of pathological thinking of schizophrenia.ActaPsychologia Taiwanica, 25, 13–23.