Children's Depression Scale (CDS)

Devised By:

Lang & Tisher (1978) developed the Research Edition of the Childrens Depression Scale (CDS).  Moshe Lang is a Clinical Psychologist, originally located at Williams Road Family Therapy Centre in Melbourne, Australia.  Miriam Tisher is also a Clinical Psychologist, originally located at Community Services of the Chisholm Institute of Technology in Melbourne, Australia.  The CDS was copyrighted and developed into the Second Research Edition (Lang & Tisher, 1983), following much research and psychometric evaluation, particularly in Australia.  Patton & Burnett (1993) produced a further revision of the CDS, namely the CDS-R, based on significant factor analyses of the CDS.  However, it is unclear whether the CDS-R has been developed and analysed further, with no evidence in the literature for its use in research studies.

Type of Instrument/Description:

The CDS is a box and card-sort questionnaire, with 66 items displayed on colour-coded cards, and 5 response boxes ranging from ‘Very Wrong’ to ‘Very Right’.  The 66 items are divided into 48 depressive items and 18 positive items, yielding 2 main scales and 6 sub-scales.  The CDS was devised for use with children and adolescents ranging in age from 9-16 years old, and purports to measure depression in children, as well as providing sub-scales of depressive symptomatology.  The different sub-scales of the CDS were developed according to definitions of childhood depression in the mid-1970’s (Lang &Tisher, 1978).  The Second Research Edition contains a revised format for parents, siblings, teachers, or significant others, namely a paper and pencil questionnaire, and rephrasing of questions from ‘I’ to ‘He/She’ (Lang & Tisher, 1983).  The CDS can also be used with younger or learning disabled children, with the clinician reading out items as necessary (Lang & Tisher, 1983).

The CDS yields two main scales, namely the Depressive scale and the Positive scale.  The Depressive scale is composed of 5 sub-scales and miscellaneous depressive items, whereas the Positive scale is composed of 1 sub-scale and miscellaneous positive items.  The sub-scales are posited as follows:  

Depressive sub-scales (D):

1)      Affective Response –AR (feeling, state, mood of respondent – 8 items)

2)      Social Problems – SP (social interaction, loneliness, isolation – 8 items)

3)      Self-Esteem – SE (attitudes and concepts in relation to own worth and value – 8 items)

4)      Preoccupation with Sickness and Death – SD (7 items)

5)      Guilt – GL (self-blame – 8 items)

Miscellaneous D Items – MD (9 items)

Positive sub-scales (P):

1)      Pleasure and Enjoyment – PE (fun, enjoyment, happiness – 8 items)

Miscellaneous P Items – MP (10 items)

Reliability:

Split Half/Cronbach’s Alpha:

Internal consistency for the CDS is reportedly high, ranging from 0.82 to 0.97, depending on the age of the sample, and the person completing the questionnaire.  Lang & Tisher (1983) analysed 226 children record forms, reporting an alpha coefficient of 0.96.  The total raw scores have yielded alpha scores of 0.92 (Tonkin & Hudson, 1981; cited in Tisher, Lang-Takac & Lang, 1992) to 0.94, with a Guttman split-half coefficient of 0.90 (Bath & Middleton, 1985).  Rotundo & Hensley (1985) reported an alpha of 0.97 for the child D score, and 0.96 for the parent D score.  Knight, Hensley, & Waters (1988) reported coefficients for the child D score = 0.94, child P score = 0.82, parent D score = 0.95 and parent P score = 0.82 in a prepubertal sample.  Kazdin (1987) reported similar results with psychiatric inpatients, with a child D score = 0.94, child P score = 0.85, parent D score = 0.92 and parent P score = 0.84.

Reynolds (1994) suggested that high internal consistency of the CDS could be expected due to the large numbers of items on the scale.  Patton & Burnett (1993) stated that the high levels of alpha for the CDS might be indicative of significant item redundancy.

Test/Retest Reliability:

Tonkin & Hudson (1981; cited in Tisher et al., 1992) administered the CDS twice to their subjects (9-13 year old Australian children), within a period of 7-10 days, reporting a correlation of 0.74.  However, no other test-retest analyses have been completed with the CDS (Costello & Angold, 1988).

Alternate Form and Inter-rater Reliability:

The CDS does not provide an alternate form measure, aside from the possibility that the Children and Parent forms may be analysed as alternate forms (Lang & Tisher, 1983).  However, the significant differences between child self-reports and parental reports of depression in children suggest that these reports are not particularly compatible (Tisher et al, 1992). Coefficients for alternate or inter-rater reliability are not reported in the numerous research studies for the CDS (Costello & Angold, 1988).

Validity:

Criterion (Predictive) Validity:

Lang & Tisher (1983) highlighted the difficulties in selecting appropriate criteria for childhood depression, as measures did not exist prior to the CDS, and there was incomplete agreement regarding the meaning of ‘childhood depression’ (Kovacs, 1977).  Initially, Lang & Tisher (1983) utilised the relationship of scores of the same persons, and relationship to other measures.  Hence, the IPAT (Personality Assessment) was used as a predictor of childhood depression, based on empirical and theoretical literature regarding correlations between depression and personality factor scores.  The CDS demonstrated significant correlations with many of the personality factors associated with depression (Lang & Tisher, 1983).  Significant correlations (p<.001) were also reported with regard to psychiatric diagnoses, obtained by experienced clinicians who knew the subject via detailed case histories and discussion with colleagues (Lang & Tisher, 1983).

Rotundo & Hensley (1985), in a teenage sample of one control and three experimental groups, reported that the CDS discriminated significantly between normal vs clinical groups, depressed vs other clinical diagnoses, and sad vs non-sad children.  Kazdin (1987) also reported significant discrimination between depressed and non-depressed clinical children on the CDS, with Knight et al. (1988) demonstrating that depressed children endorsed more items on the CDS than non-depressed children.

Costello & Angold (1988) suggest that criterion validity is strongly related to the ability of a measure to screen for ‘cases’ of depressive disorder in children.  In this regard, there are no studies that give a measure of ‘screening efficiency’ for the CDS, possibly due to the lack of clarity regarding depressive symptoms in children (Costello & Angold, 1988).  Tisher et al. (1992) concede that the parent form of the CDS is not as predictive as the child version, hypothesising that parents may be less willing to accept depression in their children.  However, Tisher et al. (1992) concluded that the CDS has good criterion validity across a range of studies, which seems an accurate assessment of the literature.

Content Validity:

Lang & Tisher (1983) cited three main sources of content validity for the CDS.  Firstly, that the CDS items were developed from the ‘universe of knowledge’ regarding childhood depression at the time.  Secondly, a pilot study was performed with a group of depressed clinic patients, who stated that the items of the CDS validated their depressive feelings, and suggested modifications to specific items.  Finally, independent judges (7 child psychiatrists) rated 53% of CDS items to be consistent with depression, in comparison to separate items measuring anxiety.  Tisher & Lang (1983) also cited many references in the literature regarding the ‘universe of knowledge’ that led to the development of specific sub-scales and items. 

Bath & Middleton (1985) suggested that 10 of the 66 items should be dropped, as they displayed correlations <0.30 with the total score, whereas Rotundo & Hensley (1985) found negative loadings for 4 specific items, recommending that they also be dispensed with.  Tisher et al. (1992) highlighted that the CDS was based on DSM-III definitions of childhood depression, and that further development of item construction needed to occur.

Interestingly, Costello & Angold (1988) warn that content and criterion-related validity may be opposing factors in measures of childhood depression, suggesting that the CDS is more of a ‘symptom checklist’ (content validity), rather than a screening device for ‘cases’ of depression (criterion validity).

Construct Validity:

Lang & Tisher (1983) emphasised that the Depressive scale and Positive scale correlated well negatively (r=0.53), with both discriminating between experimental and control groups.  On all the scales and sub-scales, depressed children obtained the highest scores, normal children obtained the lowest scores, and clinical children scored between these groups as expected, providing a ‘clear statement of support for the construct validity of the CDS’ (Lang & Tisher, 1983).  Factor analyses demonstrated strong support for a general ‘D’ factor, with most items showing acceptable levels of loading on the general factor (56 items >.04).  Lang & Tisher (1983) also reported that 35/48 ‘D’ items discriminated between depressed and normal children, and that 11/18 ‘P’ items also discriminated between depressed and normal children, suggesting that the CDS measures a relatively homogeneous property.

However, the 6 sub-scales of the CDS are not well supported in the literature, with many and varied factors reported within the overall scores.  Bath & Middleton (1985) found 9 different factors accounting for the variance, with little correspondence to the sub-scales identified by Lang & Tisher (1978).  Rotundo & Hensley (1985) found a 13-factor solution to the CDS, again displaying little support for the sub-scales presented in the CDS.  The Dutch version of the CDS has constructed four sub-scales, namely Listlessness, Lability, Guilt, and Positive (Tisher et al., 1992).  Patton & Burnett (1993) presented a revised CDS based on four distinct factors, namely Loneliness/Death, School related depression, Positive Self-Esteem, and Self-Downing.

The variability of the sub-scale factors prompted Tisher et al. (1992) to conclude that the sub-scales of the CDS in their current form need to be reviewed, due to the lack of empirical support.  However, they maintain that the scales are still reportedly useful in a clinical, qualitative, or descriptive context (Tisher et al., 1992).  By far the strongest feature of the CDS appears to be the ‘D’ score (Reynolds, 1994), suggesting that this should be the primary clinical score for the CDS.

Convergent Validity:

The CDS has been compared to several different measures in the literature, with the Childrens Depression Inventory (CDI) prominent.  Comparisons of the CDS and CDI have yielded correlations of 0.48 (Kazdin, 1987), 0.76 (Knight et al, 1988), and 0.84 (Rotundo & Hensley, 1985).  The latter study used an adolescent sample, as opposed to prepubertal children, which may account for the higher correlation reported.  Lang & Tisher (1978) demonstrated high correlations with the IPAT and EPQ personality questionnaires related to depression.  Kazdin (1987) reported significant correlations with the Bellevue Index of Depression – Revised (0.51), Hopelessness Scale (0.38), Child Behaviour CheckList (0.57), and Self-Esteem Inventory (-0.69).  Rotundo & Hensley (1985) also correlated sub-scale scores with therapist ratings of child functioning in six areas, reporting significant correlations in all areas.  Tisher et al. (1992) concluded that the CDS correlates highly with other measures of depression, implying good convergent validity.

Discriminant Validity:

Whilst the CDS is able to discriminate between depressed and non-depressed children, there may be a high false positive rate of up to 25% (Knight et al., 1988).  Tisher et al. (1992) suggested that further research was warranted due to high correlations with self-esteem and self-concept scales, with the concern that they may be measuring similar constructs.

Scoring: (adapted from Lang & Tisher, 1983)

The examiner is required to remove the cards from the boxes, and score them accordingly:

Very Wrong = 1; 
Wrong = 2; 
Don’t Know/Not Sure = 3; 
Right = 4; 
Very Right = 5.

These scores are then transferred to the record form, where items are reproduced in numerical order.  Cooperative scoring with the child may also be useful as a clinical tool.

Scores for each sub-scale are added together to provide the sub-scale scores, with raw scores entered on the front page of the record form.  The Total D score is obtained by adding the five D scales and miscellaneous D, and the Total P score is obtained by adding the PE sub-scale to the miscellaneous P items.  On the front of the record form, two tables of deciles are provided, one for children and one for parents.  Deciles are presented as ‘Normal’ scores, and lines can be drawn to produce profiles of depression if necessary.  Corresponding decile scores are provided on the form, with deciles >= 8 being regarded as significant and indicative of depressive symptomatology.

Norms:

The original validation sample consisted of 96 children, with 76 mothers and 54 fathers (Lang & Tisher, 1978).  Children were aged between 9-16 years, and attending Bouverie Clinic, a child psychiatric clinic in inner city Melbourne.  The Experimental group consisted of cases of severe school refusal, cited as experiencing comorbid depression (Lang & Tisher, 1983).  The Clinic group consisted of a variety of clinical disorders other than school refusal, and the Control group consisted of regular school attenders who had not missed more than ten days of school during the year (matched with experimental group on age, sex, school, and year).   Means and standard deviations are provided for experimental and control group, with significant differences reported between the two groups as follows:

Experimental                 Depression Score                     Control Depression Score

             _                                                                                 _

Parents X                            159.77                              Parents X                 102.78

            SD                            24.82                                          SD                 18.59

                 _                                                                                  _

Childrens X                         156.95                              Childrens X              116.89

               SD                         28.65                                             SD              35.78

                                         (p < NS)                                                            (p <.03)

           

Lang & Tisher (1983) provide more detailed means and standard deviations for each of the sub-scales as follows:

Experimental                 AR         SP         SE        S/D      GL       Pscale       PE

             _

Parents X                  27.74      28.22   26.06    21.82   23.93    49.15      22.23

            SD                   5.82       5.51      5.25      5.53     4.37    10.46        5.97

                 _

Childrens X                25.70     26.55   26.55    21.83   25.15    53.27      23.83

               SD                6.44       6.47      5.16      5.55     5.35    11.45        5.94

              (p<)               NS         NS        NS        NS       NS       .00           .04

 

Control             AR         SP         SE         S/D      GL       Pscale      PE

             _

Parents X                   15.32     16.14   17.26    13.79   17.12     38.59     15.32

            SD                   3.81      3.68      3.79     2.74     3.44       4.79       2.18

               _

Children X                   17.43    17.38   21.19    16.08   20.43     41.54     16.00

              SD                 6.09      7.00      7.29      5.04     7.12       9.07       4.90

             (p<)                  NS        NS        .00        .01       .01         .02         NS

Sex, Age and Social Class Differences:

Differences were not found to be statistically significant for all of these variables, although interpretations must proceed with caution due to small sample size (Lang & Tisher, 1983).  Whilst studies have explored prepubertal and teenage samples, it is yet to be demonstrated that the CDS is valid for the entire age range covered by the scale (Knight et al, 1988).

Cross-Cultural Norms:

The CDS has been translated for use in several languages, namely Japanese, Italian, Dutch, French, Spanish, and Hindi (Tisher et al, 1992), with a recent study providing validation specifically for use with Iranian children (Khosravi, 1996).  The CDS is also published in North America (Lang & Tisher, 1987), Brazil, Holland, Italy and Spain, with some local norms provided (Tisher et al, 1992).

Some criticisms of the norms include a lack of sufficient normative data (Reynolds, 1994), with norms largely based on Australian children.  Also, no effort was made to cross-validate the original CDS and the Revised version with a more appropriate criterion group (Knight et al, 1988).  Many studies have also pointed to the lack of evidence justifying the use of school refusal children in the normative sample (see Tisher et al, 1992).

References:

Bath, H.J. & Middleton, M.R. (1985). The Childrens Depression Scale: Psychometric properties and factor structure.  Australian Journal of Psychology, Vol 37 (1), 81-88.

Costello, E.J. & Angold, A. (1988). Scales to assess child and adolescent depression: Checklists, screens, and nets.  Journal of the Academy of Child and Adolescent Psychiatry, Vol 27 (6), 726-737.

Kazdin, A.E. (1987). Childrens Depression Scale: Validation with Psychiatric Inpatients. Journal of Child Psychology and Psychiatry, Vol 28 (1), 29-41.

Khosravi, Z. (1996).  The role of cognitive bias in encoding and recalling information by depressed and happy children. Dissertation Abstracts International Section A: Humanities and Social Sciences. Vol 57 (1-A): 0102.

Knight, D., Hensley, V.R., & Waters, B. (1988). Validation of the Childrens Depression Scale and the Childrens Depression Inventory in a prepubertal sample. Journal of Child Psychology and Psychiatry, Vol 29 (6), 853-863.

Kovacs, M. (1977). Conclusions and recommendations of the sub-committee on Assessment.  In J.G. Schulterbrandt & A. Raskin (Eds.), Depression in childhood: Treatment and Conceptual models. New York: Raven Press.

Lang, M. & Tisher, M. (1978). Childrens Depression Scale: Research Edition. Melbourne: The Australian Council for Educational Research Limited.

Lang, M. & Tisher, M. (1983). Childrens Depression Scale: Second Research Edition. Melbourne: The Australian Council for Educational Research Limited.

Lang, M. & Tisher, M. (1987). Childrens Depression Scale Manual, North American Edition. Palo Alto, CA: Consulting Psychologists Press.

Patton, W. & Burnett, P.C. (1993). The Childrens Depression Scale: Assessment of factor structure with data from a normal adolescent population. Adolescence, Vol 28 (110), 315-324.

Reynolds, W.M. (1994). Assessment of depression in children and adolescents by self-report questionnaires.  In W.M. Reynolds & H.F. Johnston (Eds.), Handbook of Depression in Children and Adolescents. New York: Plenum Press.

Rotundo, N. & Hensley, V.R. (1985). The Childrens Depression Scale: A study of its validity. Journal of Child Psychology and Psychiatry, Vol 26 (6), 917-927.

Tisher, M. & Lang, M. (1983). The Childrens Depression Scale: Review and further developments. In D. Cantwell & G. Carlson (Eds.), Affective Disorders in Childhood and Adolescence. New York: SP Scientific Books.

Tisher, M., Lang-Takac, E. & Lang, M. (1992). The Childrens Depression Scale: Review of Australian and overseas experience. Australian Journal of Psychology, Vol 44 (1), 27-35.

Tonkin, G. & Hudson, A. (1981). The Childrens Depression Scale: Some further psychometric data. Australian Council for Educational Research Bulletin for Psychologists, Vol 30, 11-18.

(Actual Questionnaire or Record forms not available due to Copyright)

Above written by: Mr John Miksad

Reviewed, edited and approved by: Dr. Grant J. Devilly