The psychometric properties of the Beck Depression Inventory and the Zung Self Rating Depression Scale were examined. Both tests were administered to 120 adults with mental retardation. Results were analyzed, correlated, and examined with principle component analysis. Findings suggest that the tests have good clinical utility with these persons and that depression may be more severe in people with mental retardation. Characteristics that can be employed in building other behavioral measures of depression for persons with mental retardation were also identified. Results show that there may be a tendency for persons with mental retardation to minimize distress report, further complicating the depression diagnostic process for these persons.
Numerous people with mental retardation suffer from depression (Matson, 1983; Sovner & Hurley, 1983a; Sovner & Pary, 1993), and many experts feel that they are at a greater risk for developing depression than persons of average intelligence (Borthwick-Duffy, 1994; Nezu, 1994; Sturmey, 1995) In a set of studies, Reiss and associates also pointed out that people with mental retardation were more likely to have severe social deficits and withdrawal (Reiss & Benson, 1985) and were more likely to engage in serious challenging behaviors, such as aggression (Reiss & Rojahn, 1993). This finding was replicated by other researchers (e.g., Meins, 1993, 1995) and extended to include problems such as temper outbursts and self-injury (Lowry, 1998; Sovner & Pary, 1993). Many persons with mental retardation have verbal difficulties that complicate the assessment process and make diagnosis of depression difficult (Matson & Sevin, 1994; Reiss, 1993). Special informant-based assessments have been developed to improve the accuracy of depression diagnosis in people with mental retardation (Aman, 1991), such as the Assessment for Dual Diagnosis and the Diagnostic Assessment for the Severely Handicapped II (Matson, 1995). Research validating these instruments is on-going, but these instruments may not capture the complete experience of depression (Benavidez & Matson, 1993).
Self-report instruments are available for assessing depression. Tests such as the Beck Depression Inventory—BDI (Beck & Steer, 1993) and the Zung Self-Report Depression Scale (Zung, 1965) have been extensively used with the general population; therefore, their psychometric characteristics with this population are well-known. A growing body of research shows that these tests may also have clinical utility for persons with mental retardation and can be used to develop depression assessment instruments for this population.
The BDI and the Zung Self Report Depression Scale (hereafter referred to as the Zung) were given to 10 males and 11 females with mild mental retardation (Prout & Schaeffer, 1985). The authors found correlation between scores on the two measures and that participants with mental retardation scored significantly higher on these measures than did similar groups in the general population. Beck, Gabrielle, Carlson, Russell, and Brownfield (1987) administered the BDI to 26 adolescents and found good correlation with psychiatric depression diagnosis.
Variants of the BDI and Zung have also been evaluated in persons with mental retardation. Kazdin, Matson, and Senatore (1983) compared a 13-item BDI and a simplified version of the Zung with the Hamilton Psychiatric Depression Scale (Hamilton, 1965), the depression scale of the MMPI, and self-report and informant-based versions of the Psychopathology Instrument for Mentally Retarded Adults—PIMRA (Matson, 1988) by testing 110 adults with mental retardation. Significant correlation was found with all instruments except for the informant's version of the PIMRA. Helsel and Matson (1988) administered a battery of tests to 99 adults with mild to severe mental retardation to examine social skills as a function of intellectual level. The tests included the BDI, Zung, Hamilton, the Peabody Picture Vocabulary Test—PPVT, and various social skills measures. Their goal was to determine whether there was a relationship between social skill deficits and depression in persons with mental retardation. These investigators found that depressed persons with mental retardation were more likely to have significant social skill deficits, but they also found a correlation between the BDI and the Zung. Reynolds and Baker (1988), in their extensive work validating the Self-Report Depression Questionnaire, an instrument used to assess depression in person's with mental retardation, emphasized that additional large samples need to be tested with the full versions of the BDI and Zung so the psychometric properties of these instruments can be better understood and assist those developing procedures for assessing people with mental retardation and depression.
This preliminary research shows that the BDI and the Zung, commonly used with the general population, may have a place in the assessment of depression for persons with mental retardation, but the psychometric properties of these instruments needs further examination.
In the current study I looked at the psychometric properties of the BDI and the Zung with regard to adults with mental retardation. I also compared results with those in other research studies regarding these same parameters in persons without mental retardation.
The 120 participants in this study, 50 females (42%) and 70 males (58%), resided in the central portion of Illinois. Agencies serving people with mental retardation were contacted and allowed to review the proposed research. They then referred participants based on their perceived abilities to respond to the testing expectations. I saw all referred people, and those who successfully completed the screening described in the data-collection procedure were assessed with the BDI and Zung; only 6 referred people did not successfully complete the screening. Of the remaining participants, 65 (54%) lived in group homes, 29 (24%) resided in large state institutions, 14 (12%) lived in supervised apartments, and 12 (10%) resided on their own in houses or apartments. Ethnicity of participants was mixed: 109 (91%) were of White European ancestry, 10 (9%) were African American, and 1 (1%) was biracial (African American−Hispanic). Mean participant IQ was 53 (range = 22 to 75). The functioning level listed in participants' records included 43 (36%) in the mild range, 55 (46%) in the moderate range, and 14 (12%) in the severe range. No functioning level was listed for 8 (6%) of the participants.
Participants ranged in age from 19 to 74 years (average = 40.75, standard deviation [SD] = 11.79). Their sources of income varied, with 3 (2%) living on their own earnings and 7(6%) having no employment at all. The remainder received a mixture of income from employment and public assistance. Six (5%) were employed in the regular work force, 2 (2%) had jobs with work coaches, and 6 (5%) worked in specially supervised work crews at local colleges. Those remaining worked in sheltered workshops or prevocational activity centers.
The most common accompanying medical condition of these individuals was seizure disorder (n = 18, 19%). Cerebral palsy was listed for 7 participants (8%), and Down syndrome was diagnosed for 5 (6%); the remaining 90 participants (75%) had no other serious medical condition.
No medication was prescribed for 39 participants (40%) Some were taking multiple medications, so the totals do not add simply to the remainder. Twenty-four (24%) were taking anticonvulsant medication, 31 (32%) were on an antipsychotic medication, and 19 (21%) were taking some form of medication for mood disorder. Eight (9%) were taking antianxiety medication.
Many participants had a mental health or Axis I diagnosis, and several had more than one. No Axis I diagnosis was listed for 71 participants (59%) Thirty-three (28%) had some form of mood disorder. Psychosis or thought disorder was reported for 18 (20%), and a mental disorder due to a general medical condition was reported for 5 participants (6%). Anxiety disorders were reported for 6 (7%) and personality disorders were listed for 7 (8%). Impulse control disorders were recorded for 4 participants (5%)
The study was reviewed and approved by the author's Institutional Review Board (IRB) and IRBs of agencies that referred participants. Consent was then obtained from each participant or their guardian in accordance with the standards for protection for human subjects.
I conducted the participant interview and assessment in private areas at their homes or work sites. I talked briefly with each participant to explain the study, answer questions about the study, and build rapport. I told all of them that I wanted to know how they really felt about the questions that followed.
A brief screening exercise (6 questions) was conducted with each participant to ensure that they could comprehend the format of the testing. Three questions were likely to be agreed with and 3 would not usually be agreed with. An agreement statement was, for example, “I like to have fun.” An example of a statement that should generate disagreement was “I like to eat rotten food.” An examinee that failed to respond as expected was not tested further. This procedure is similar to that used in other research with this instrument and population (Kazdin et al., 1983) The BDI and Zung were administered at this point. Each selected participant was read the items and, if they were capable, allowed to read the items themselves. This is an approach that is recommended in the BDI manual and has been commonly employed in other studies employing these instruments with people who have mental retardation. Participants had difficult words explained to them, and a bar graph representation of Likert alternatives was shown to assist their understanding of semantic differential magnitude as used in the Zung. No further direction or assistance was given. Testing and interview took approximately 30 minutes for each person. A brief demographic questionnaire was completed by record review and interview of participants and case managers.
The BDI and the Zung were administered to 120 persons. Correlation of the two measures was assessed with the Pearson product-moment correlation coefficient, and the two measures were significantly correlated for this sample, r = .34, p = .000.
Beck Depression Inventory
Table 1 shows that 29.9% of the sample registered clinically significant depression (Beck & Steer, 1993). Mean BDI score was 12.38 (median score = 10, range = 0 to 43, SD = 10.18). Internal consistency was assessed with Cronbach's alpha, and the results for this sample was .86.
The mean BDI score for females was 13.36 (SD = 10.35). The males' average score was 11.69 (SD = 10.09). The females' score was higher than the males' score, but this difference was not significant, t(118) = .89. The comparison resulted in an effect size of .16, and the test had a power of .14 (Borenstein & Cohen, 1988).
Principle component analysis
Principle component analysis results are presented in Table 2. This exploratory factor analysis was selected for this study because the basic factor structure of the BDI and Zung has been found to vary depending on the population (Beck & Steer, 1993; Hedlund & Viewing, 1979). This exploratory study was conducted in hope of discovering characteristics that may be unique to the population or in common with people in the general population. Principle component analysis is the appropriate analysis in these situations (Stevens, 1996).
The principle component analysis, with Varimax rotation, resulted in a seven-factor solution that accounted for 65.7% of the sample's variance. Variable communality was high (range = .55 to .79), with the majority being in the mid to upper .60s. Table 3 presents the items and their loadings both before and after rotation. The table also has been arranged to show how items related in the rotated and unrotated solutions, suggesting possible item relationships.
The significance level was determined from tables found in Stevens (1996), which suggested that at a sample size of 120, a value greater than .47 would produce a two-tailed alpha significant at the .01 level. Item 5 (guilt) was the only item that did not load significantly either in the unrotated or rotated solution. Item 21 (loss of libido) produced a loading on the rotated solution that was not significant. All other loadings were significant at the .01 level.
The unrotated solution produced a large 17-item factor and 5 item-specific factors. Factor 1, 2, 3, and 5 in the rotated solution were all related in the large general factor of the unrotated solution. The large unrotated general factor then contained two items and one item from rotated Factors 4 and 6, respectively. Only the large unrotated Factor 1 was big enough to produce a reliable interpretation (Guadagnoli & Velicer, 1988), which is similar to the general depression factor found by Tanaka and Huba (1986).
Zung Self Rating Depression Scale
In this sample 30% of the participants reported moderate to severe depression on the Zung (see Table 4). The mean score on the Zung for the 120 participants was 42.78 (SD = 7.93, median score = 42). Internal consistency of the scale for this sample was .58, as assessed with Cronbach's alpha.
The 70 females in the sample averaged 43.14 on the Zung (SD = 7.07). Males had a slightly lower mean score of 42.51 and a larger dispersion (SD = 8.52). The difference was not significant. The effect size was .06, and the power of the test was .08 (Borenstein & Cohen, 1988).
Principle component analysis
This analysis of the Zung yielded a 7-factor solution, which accounted for 57% of the sample's variance (see Table 5). Communality of items ranged from .41 to .74.
Table 6 shows the items and their loadings both before and after rotation. The table, as with the BDI, has been arranged to show how items related in the rotated and unrotated solutions to enable item relationships to be seen.
Here, a hierarchical relationship between the unrotated and rotated solutions appeared, but it was not as well-defined as that seen with the BDI. The 7 items in the unrotated Factor 1 contained all the items in the rotated Factors 1 and 3. Rotated Factor 2 items were split between unrotated Factors 2, 4, and 6. Rotated Factor 4 was split with 2 items in unrotated Factor 2 and 1 item in unrotated Factor 5. Rotated Factor 5 was subsumed completely under rotated Factor 3, and rotated Factor 6 was split with 1 item in unrotated Factors 3 and 1. Rotated Factor 7 contained on item and was completely subsumed by unrotated Factor 4. None of the factors in either the rotated or unrotated solutions contained loadings large enough to produce reliable interpretations.
The Beck Depression Inventory and the Zung Self-Report Depression Scale were administered to adults with mental retardation, with items read to them. These test results were analyzed to determine the psychometric properties of the instruments with persons who have mental retardation so they may be used more comfortably by clinicians seeking to diagnose depression in members of this population.
Findings in this study suggest that the BDI has clinical validity with this population, but the Zung may have problems. Participants' scores on the two instruments correlated. The indices were not as large as those found in the general population (Senra, 1995), but if other research with persons who have mental retardation on the simplified versions is considered, the results were comparable (Kazdin et al., 1983). Looking specifically at the BDI findings shows the sample's internal consistency measure was at the mean found for this index from other research (Beck, Steer, & Garbin, 1988). Scoring conventions for the BDI also hold that scores higher than 40 suggest either an exaggeration of symptoms or can be found for those with borderline or histrionic personality disorders. Three participants scored in this range. One had been recently hospitalized for depression, and the other 2 had diagnosis of these personality disorders.
The factor structure found for the BDI varies slightly between tested populations, but the definitive analysis conducted by Tanaka and Huba (1986) produced a hierarchy structure, where specific first order factors are subsumed into a second order depression factor (Beck & Steer, 1993) This study's factor analysis produced the same sort of factor structure, suggesting the instrument's validity with people who have mental retardation.
Results for this sample using the Zung were equivocal. There was a significant correlation with the BDI, and the amount of variance accounted for by the obtained factor structure was high in relationship to other factor analytic studies done with this instrument (Hedlund & Viewing, 1979) The sample's internal consistency measures, however, were low, and factor analysis did not yield factors having reliably interpretable loadings. This problem for the Zung may not be unique to its use with persons who have mental retardation. The Zung's factor (Blumenthal, 1975) and normative (Gabrys & Peters, 1985; Holmes, Fouty, Wurtz, & Burdick, 1988) structure varies with the tested population. Also, demonstrated differences, such as those between gender scores, have not been consistently found from study to study (deJonghe & Baneke, 1989). Clearly, additional research into the properties of this instrument needs to be done before it can be used confidently by practitioners in all areas of practice and not just with persons with mental retardation. The current study did show that the test would correlate with the BDI when used with persons who have mental retardation, but specific results may need to be interpreted with caution and, at best, it may need to be used as a screening test in a battery (Blumenthal, 1975).
The sample's BDI data further suggest a strong tendency to minimize the self-report of distress in this population. Others in the field have also noted that persons with mental retardation may try to give socially acceptable answers and, thereby, minimize the report of distress (e.g., Sovner & Hurley, 1983b; Sovner & Pary, 1993; Szymanski, 1988). This tendency for people with mental retardation to minimize reported distress would be important for clinicians seeking to evaluate depression in this population. Additional research is needed to make sure that this tendency, however, was not just unique to this sample.
The diagnostic criteria for depression, as defined in the Diagnostic and Statistical Manual-Fourth Edition—DSM-IV (American Psychiatric Association, 1994), has been questioned because depression may be manifested differently in people who have mental retardation (Aman, 1991; Einfeld, 1992). People with depression and mental retardation may withdraw socially, have increased somatic complaints, and engage in aggression (Reiss & Rojahn, 1993; Sovner & Hurley, 1983a). These characteristics are not mentioned in the current DSM-IV diagnostic criteria. The sample's data also illuminated some relationships between subjective experiences in depression and observable behaviors. Verbalizations about failure or not wanting to attempt new things, for example, may be as related to depression as is crying or tearfulness. Further research is necessary to make sure that this was not a finding unique to this sample, but practitioners may want to keep this in mind when developing behavioral systems to monitor and assess depression in persons with developmental disabilities.
In the current study I found that the Beck Depression Inventory and the Zung Self-Report Depression Scale could be readily administered to persons with mental retardation. Further, results suggest that the BDI may be a better instrument for this assessment because it shows similar psychometric properties in both the general population and with people who have mental retardation. Some characteristics unique to persons with mental retardation may have been identified, such as the tendency to minimize distress, and these findings also suggest some areas for continued research into the use of these instruments, both with persons who have mental retardation and with the general population.
Author: Richard Powell, PhD, Regional Psychology Consultant, Department of Finance and Administration, Division of Mental Retardation Services, Western Tennessee Regional Office, 8383 Wolf Lake Dr., Bartlett, TN 38133 and The Columbus Organization. Richard.firstname.lastname@example.org