Abstract
Test Description
The Children’s Depression Inventory 2 (CDI 2) was published by Multi-Health Systems (MHS) to assess depressive symptoms in 7- to 17-year-old children and adolescents. Evolving from the original CDI developed by Kovacs in 1977 and formally published by MHS in 1992, the CDI 2 was designed for youth to respond easily to the scales with three choices per item and items written at a low reading level. The CDI 2 contains two self-rated scales of a full-length (CDI 2: SR) and a short version (CDI 2: SR[S]) and two adult-rated scales of parent report (CDI 2: P) and teacher report (CDI 2: T). The time interval for rating each form of CDI 2 is the past 2 weeks. The CDI 2 tools can be used for both individual and group administration. The individual assessment of youth can offer clinicians auxiliary information for clinical diagnosis and therapy along with other sources. Group-based assessment can be employed for screening to identify what proportion of students need subsequent evaluation and support.
Given the importance of early diagnosis and treatment (Kovacs & Devlin, 1998), the CDI 2 can assist professionals to pinpoint critical depressive symptoms a child may experience. Parent and teacher forms facilitate inclusive assessment of children’s affective disorders and dysfunctions through multiple viewpoints.
Specific Description and Scoring
CDI 2 tools are accessible through either paper-based or computer-based versions. The CDI 2: SR[S] was designed particularly for screening. CDI 2: SR, CDI 2: P, and CDI 2: T include two subscales: Emotional Problems and Functional Problems. In addition, the scales on Emotional Problems of CDI 2 SR are subdivided into Negative Mood/Physical Symptoms and Negative Self-esteem, and the scales on Functional Problems are subdivided into Ineffectiveness and Interpersonal Problems. The scales on Emotional Problems indicate distressed feelings, such as sadness and guilt, and abnormalities in interests, sleep, appetite, and energy. The scales on Functional Problems reflect dysfunctions in social relationships and performance such as decrease in school grades and troubles in peer and/or family relationships resulting from irritability.
As with the original CDI, respondents aged 7 to 17 years answer each item of CDI 2: SR on the three-choice scales from 0 (none) to 2 (definite), and parents and teachers reply to CDI 2: P and CDI 2: T, respectively, on a 4-point Likert-type scale from 0 (not at all) to 3 (much or most of the time). Therefore, the higher score on each CDI 2 form indicates greater possibility for the child to experience depressive symptoms.
The CDI 2 has three scoring options: hand-scoring (Quick Score), scoring software, or MHS Online. Raw scores resulting from the sum of values on the answered items are standardized into T-scores with a mean of 50 and standard deviation of 10 for total and subscales. Also, the norms for T-score conversions are based on age and gender. Specifically, two age groups of younger (7- to 12- year-olds) and older respondents (13- to 17-year-olds) independently for boys and girls are employed for CDI 2 norms. In addition, empirical percentiles computed by using the practical frequency counts in the standardized sample (Kovacs, 2011) are available in CDI 2 Scoring Software, the MHS Online Assessment Center, or appendix D of the CDI 2 Manual. The T-scores and percentile ranks are categorized into five classifications: Very Elevated (T-score 70+, Percentile Rank 98+), Elevated (T-score 65-69, Percentile Rank 93-97), High Average (T-score 60-64, Percentile Rank 84-92), Average (T-score 40-59, Percentile Rank 16-83), and Low (T-score <40, Percentile Rank <16). Both 90% and 95% confidence intervals are available for all scales.
Test Materials and Stimuli
The manual consists of six chapters specifically describing the background and history of the scales, psychometric characteristics, and guidelines for administration, scoring, and interpretation. The information including the various appendixes, tables, and figures is well organized and detailed. Moreover, specific examples help users and reviewers to easily understand the contents of the CDI 2.
Technical Adequacy
Test Construction
The original self-report CDI with 27 items was adapted into the CDI 2: SR with 28 items throughout two phases. First, considering a change in clinical thinking about hypersomnia and hyperphasia in youth and the age-relevant depressive symptom of dysfunction in thinking and concentration, new items were added and prior ones were deleted. Second, some items were revised due to the discrepancy between respondents’ interpretation and initial intent or eliminated to maintain a reasonable length. The short version of the CDI 2 self-report was extracted from the full-length form.
Both the parent- and teacher- rated CDI 2 forms were derived from the CDI 2 self-report, focusing on common and observable depressive symptoms. Those adult forms have no changes from the previous version. The CDI: P includes 17 items and the CDI: T includes 12 items. The original CDI has formal versions in 43 different languages, but the CDI 2 has had only Spanish translation so far.
Standardization Sample
The sample for the CDI 2: SR and CDI 2: SR[S] comprised 1,100 youngsters aged 7 to 17 years inhabiting 28 states throughout all four major geographic regions (Northeast, Midwest, West, and South) of the United States. The collected data were categorized based on children’s sex, age, race/ethnicity, and geographic location. The ethnic groups include Asian, African, Hispanic, White, and Multiracial/Other. The sample was constructed by balancing the number of boys and girls and stratifying the racial distribution based on U.S. Bureau of the Census in 2000, and across four major geographic regions of the United States. The age- and sex-based norm groups were derived from the symptomatic differences associated with those categories in youth which were analyzed for all scales of all CDI 2 forms through ANOVAs and MANOVAs.
In addition, the standardization samples of CDI 2: P and CDI 2: T comprised 800 and 600 adults, respectively, with a balanced number between girls and boys of 7 to 17 years old. The interpretability of the adult forms of CDI 2 was improved through more representativeness of the U.S. population based on U.S. Bureau of the Census in 2000.
In all, 319 children aged 7 through 17 years comprised the clinical sample, which were categorized based on sex, race/ethnicity, and the Diagnostic and Statistical Manual-IV (DSM-IV) symptoms. Most psychometric analyses were conducted independently between the standardization and clinical sample, with the exception of the confirmatory factor analysis (CFA), item selection for CDI 2: SR[S], and internal consistency analyses.
Reliability
All CDI 2 forms showed high or acceptable levels of internal consistency with the Cronbach’s alpha values from .67 to .91 for total and all subscales for all age and sex groups. Also, since the CDI 2 purports to estimate temporary attributes (i.e., state) rather than stable attributes (i.e., trait; Hong & O’Neil, 2001), transient test-retest stability is significant. Test-retest reliability of the self-report forms was tested for 79 children (53.2% male, mean age = 12.06 years, SD = 2.92 years) within a 2- to 4-week interval (M = 16.1 days). The results showed excellent short-term stability with nearly no change during the time interval. Test-retest reliability of the parent and teacher report of CDI 2 was not conducted.
Validity
The construct validity of CDI 2: SR was analyzed through a hierarchical third order (four subscales–two scales -total) 1 CFA model (Hong & O’Neil, 2001). The model fit was very good with the CFA indexes: NFI = .95, NNFI = .94, CFI = .96 and RMSEA = .07. The high correlation between two scales (Emotional Problems and Functional Problems) at r = .77 and the moderately high correlations among the subscales from .58 to .69 indicate a hierarchical association among total, scales, and subscales (Hong & O’Neil, 2001).
CDI 2: SR[S] was derived from the full version of CDI 2: SR by examining an effect size (Cohen’s d) and multiple regression analyses for screening in school or community settings. The Cohen’s d presenting the mean difference between 108 children with major depression and 108 normal children ranged from 0.39 to 2.09 for all items of CDI 2: SR[S]. This suggests that CDI 2: SR[S] works well in screening. The Pearson correlation between the regular and short form was .95 (p < .001), which shows high construct similarity. CFAs for construct validities of CDI 2: P and CDI 2: T presented good model fits with NFI = .94, NNFI = .91, CFI = .94, and RMSEA = .09 for the parent report and NFI = .99, NNFI = .99, CFI = 1.00, and RMSEA = .06 for the teacher report. The scales on Emotional Problems and Functional Problems have moderate associations on both the parent and teacher report with r = .61(p < .001) and r = .65 (p < .001), respectively.
Univariate and multivariate analyses of covariance (ANCOVA and MANCOVA) for discriminative validity were employed to test how well the CDI 2 can distinguish youth with major depressive symptoms (MDD) from youth without them. The results from MANCOVAs and ANCOVAs with the rigorous criterion (p < .01) showed that the MDD group had statistically significantly higher scores than did all other groups, which included control group matched on sex, age, and race/ethnicity; generalized anxiety disorder (GAD); conduct/oppositional defiant disorder (CD/ODD); and attention deficit hyperactivity disorder (ADHD). Discriminant function analysis (DFA) on total scores and the scale scores on Emotional Problems and Functional Problems of both self-reports generally showed appropriate accuracy in differentiating the MDD group from other groups.
As for convergent validity, 266 children (214 from the standardization sample and 52 from the clinical sample) completed either the Beck Depression Inventory–Youth version (BDI-Y; Beck, Beck, Jolly, & Steer, 2001) or Conners Comprehensive Behavior Rating Scales (Conners CBRS; Conners, 2008) along with the CDI 2, and the CDI 2 scores were correlated with similar scales on the other two instruments. Furthermore, all correlation coefficients for total and scales between the CDI 2 self-report, parent report, and teacher report were statistically significant (p < .01) with moderate effect size. This result suggests that the different forms measure the same construct but offer distinctive information as a multiperspective assessment of depression.
Commentary and Recommendations
The sophisticated features of the CDI 2 such as specific scales for Emotional and Functional Problems, additional subscales, and age and sex norms firmly support the theoretical model of depressive symptoms in youth. Also, the scales on Functional Problems regarding peers, school, and family show relevance to the targeted population.
The CDI 2 sampling approaches such as matching and stratifying based on age, sex, race/ethnicity, and geographic regions in the United States strengthened the representativeness of the 7- to 17-year-old age group. In addition, the specific categorization of the clinical sample by the primary DSM-IV diagnoses enhanced the discriminative validity.
The reliability estimates of high or adequate level indicate the appropriateness of the CDI 2 for practical and academic applications. Particularly, the correlation coefficients between the CDI 2 self-report, parent report, and teacher report presented statistically significant relationships to each other but with modest values. This suggests that all forms of the CDI 2 measure the same construct but represent exclusively distinctive perspectives on the child’s functioning. Additional validity evidence is provided by the CFAs which support the well-organized constructs of the CDI 2, results from the two approaches to test discriminative validity which show the groups could be discriminated based on diagnostic attributes, and the statistically significant correlations of the CDI 2 with the other two inventories which suggest correspondence with other instruments that assess depression.
In terms of relative weaknesses, the CDI 2 has mainly written-text formats, so measurement alternatives for special youth with cognitive or physical disabilities are necessary. As the standardization samples were selected only within the United States, the norms may not represent the populations in other countries with different social and cultural backgrounds which impact on affective issues. Therefore, with more translations and norms, the CDI 2 could be used much more effectively with youth from diverse backgrounds.
Even though CDI 2 measurements reflect the degree or intensity of ongoing depressive manifestations, the results should be integrated with careful diagnostic judgment as part of a larger information-gathering process. Considering the properties of the CDI 2, repeated measurements to assess changes in depressive symptoms over time are recommended, but use with children under involuntary conditions is not recommended.