Child behavior checklist behavioral profiles of children with autism spectrum disorders
TABLE OF CONTENTS
List of Tables vii List of Figures viii Abstract ix 1. INTRODUCTION 1 2. REVIEW OF THE LITERATURE 6 Broad Behavior Screening Instruments 6 Components of Broad Behavior Screening Instruments 6 Frequency of Use of Broad Behavior Screening Instruments 6 Commonly Used Broad Behavior Screening Instruments 7 Child Behavior Checklist (CBCL) 9 History 9 Evaluation of the CBCL 10 Reliability-Item Scores 11 Reliability-Scale Scores 12 Internal Consistency 12 Test-Retest Reliability 12 Cross Informant Agreement 13 Stability 14 Validity 14 Content Validity 14 Criterion-Related Validity 15 Screening for Unique Populations Using the CBCL 15 Autism Spectrum Disorders 17 Prevalence 17 Importance of Early Identification 18 Screening for ASD in the Schools 19 Screening for ASD Using the Child Behavior Checklist 19
Research Questions 24 ASD Versus Normal Controls 24 ASD Versus Clinic-Referred 24 3. METHODOLOGY 25 Participants 25 Instruments 29 Child Behavior Checklist/6-18 29 Data Analysis 29 4. RESULTS 31 Analysis I: ASD Versus Controls 32 Analysis II: ASD Versus Clinic-Referred 35 5. DISCUSSION 38 Summary and Interpretation of Results 38 Analysis I: ASD Versus Controls 38 Analysis II: ASD Versus Clinic-Referred 41 Limitations of the Study 43 Future Research 44 APPENDIX A 46 APPENDIX B 48 APPENDIX C 50 REFERENCES 56 BIOGRAPHICAL SKETCH 68
LIST OF TABLES
1. CBCL - Test-Retest Reliabilities and Alpha Coefficients 12 2. CBCL - Cross-Informant Agreement on Scale Scores 13 3. CBCL - Stabilities on Scale Scores 14 4. Demographic Characteristics of the Samples 26 5. Differential Diagnosis of the ASD Sample 28 6. CBCL Raw Score Subscale Means and Standard Deviations 32
LIST OF FIGURES
1. CBCL Profiles for ASD versus nonASD Controls 33 2. CBCL Profiles for ASD versus Clinic-Referred for Behavior Disorders 36
ABSTRACT Behavior checklists have been utilized by psychologists since the early 1900’s and continue to play integral roles in the screening and monitoring of behavior based disorders (Achenbach & Rescorla, 2001). The Achenbach System of Empirically Based Assessment - Child Behavior Checklist (CBCL) is one of the few widely used broad-based behavior rating scales that have excellent psychometric properties (Achenbach & Rescorla, 2001). Recent research has focused on the CBCL’s ability to screen for behaviors associated with autism spectrum disorders (ASD; Bolte, Dickhunt, & Poustka., 1999; Duarte, Bordin, Oliveira, & Bird, 2003; Rescorla, 1988, Sikora, et al, 2008). A combination of increasing rates of ASD and the need for earlier identification has created an urgency to find broadband screening tools that identify accurately this group of individuals since ASD-specific screening tools are not readily available. Therefore, the purpose of this study was to determine if patterns of normative strengths and weaknesses on the subscales of the CBCL are similar for individuals with ASD, nonASD controls, and those clinic-referred for behavior disorders. Profile analysis was used to compare the behaviors of (a) individuals with ASD to their normal controls, and (b) individuals with ASD to individuals clinic-referred for behavior difficulties. Results revealed that individuals with ASD displayed significantly higher scores on seven out of the eight CBCL subscales when compared to their nonASD controls. When compared to individuals clinic-referred for behavior difficulties, the ASD group displayed significantly higher scores on two of the CBCL areas. These findings support the use of the CBCL as a screening tool. Implications and conclusions are discussed and areas for future research are provided.
CHAPTER 1 INRODUCTION Behavior rating scales and checklists are commonly used tools in the assessment of internal and external behaviors, social skills, and emotional functioning (Heckamena, Conroy, East, & Chait, 2000). These screening tools are capable of screening for a range of behavior disorders and are utilized in multiple settings. Contributing factors to their growing popularity include (a) provision of quantifiable information, which can be held to standards of reliability and validity; (b) efficient completion and scoring; (c) provision of systematic and organized information; (d) inclusion of normative data, allowing for comparisons of individual behaviors to larger groups; and (e) ability to compare ratings of multiple respondents across settings (Hosp, Howell, & Hosp, 2003). In addition, there are numerous behavior rating scales commercially available (i.e., Behavioral Assessment System for Children [Reynolds & Kamphaus, 1992], Behavioral and Emotional Rating Scale [Epstein & Sharma, 1998], Behavior Rating Profile [Brown, 1990], Burks’ Behavior Rating Scales [Burks, 1996], Child Behavior Checklist [Achenbach & Rescorla, 2001], Conner’s Rating Scales [Conners, 1997], Revised Behavior Problem Checklist [Quay & Peterson, 1987], Social-Emotional Dimension Scale [Hutton & Roberts, 1986], and The Walker-McConnell Scale of Social Competence and School Adjustment [Walker & McConnell, 1995]. The Achenbach System of Empirically Based Assessment (ASEBA) Child Behavior Checklist (CBCL) is the most well-known dimensional approach to behavior assessment (Achenbach, 1991). It is widely used, reliable, valid, and typically referred to in research and relied upon in clinical practice (Hosp, Howell, & Hosp, 2003). This empirically based system uses three broad band syndromes: (a) Total Problems, (b) Internalizing, which include items that are problematic for the child rather than for the child’s environment; and (c) Externalizing, which include items that are disruptive for the child’s environment. Underlying the two broad- band dimensions are eight narrow-band syndromes: Anxious/Depressed, Withdrawn/Depressed, Somatic Complaints, Social Problems, Thought Problems, Attention Problems, Rule-Breaking Behavior, and Aggressive Behavior. The CBCL was developed in 1966 while scoring clinical records with a symptom checklist (Achenbach, 1966). There have been multiple revisions and current versions encompass the lifespan (ages 1.5 to 90+ years of age). For children ages six to
18 there are three versions: (a) CBCL, (b) Teacher’s Report Form (TRF), and (c) Youth Self- Report Form (YSR). The CBCL has excellent psychometric properties and a large body of research that demonstrates its reliability and validity in both clinical and nonclinical practices (Achenbach, 1991). Many studies have examined the validity of the CBCL in screening for unique populations such as ADHD subtypes, bipolar depression, mania, maladjustment, and anxiety (Aschenbrand, Angelosante, & Kendall, 2005; Biederman, Wozniak, Kiely, Ablon, Faraone, Mick, Mundy, & Kraus, 1995; Bird et al., 1988; Krol et al., 2006; Rescorla et al., 2007). However, one population that has received little attention regarding the use of the CBCL are those with autism spectrum disorders (ASD). Autism spectrum disorders are a group of developmental disorders defined by impairments in the areas of communication and socialization, as well as patterns of restricted or repetitive behaviors (American Psychiatric Association [APA], 2000). Recent epidemiological studies indicate one out of 150 children in the general population experiences symptoms of ASD (National Center on Birth Defects and Developmental Disabilities [NCBDDD], 2007). Increased prevalence rates, along with heightened public awareness, have generated a sense of urgency to better identify individuals with ASD for specific medical, educational, and behavioral interventions. With mounting evidence demonstrating the effectiveness of early intervention, there is constant pressure to improve early identification of ASD so critical interventions can be implemented (Wetherby et al., 2004). Given that there is no biological or genetic marker for ASD, screening and diagnosis must be based on the behavioral features of ASD (Fillipek et al., 1999). Therefore, research on behavior rating scales, such as the CBCL, is imperative to the ability of professionals working with this unique population to accurately identify and diagnose as early as possible. Only four studies have investigated whether the CBCL can identify a general behavior pattern in children with ASD (Bolte, Dickhunt, & Poustka., 1999; Duarte, Bordin, Oliveira, & Bird, 2003; Rescorla, 1988, Sikora, et al, 2008). The first study to examine the CBCL was a cluster analytic study conducted by Rescorla (1988) on the symptom profiles of a clinical sample comprised of 204 three to five year old boys. The clinical sample included 79 autistic and autistic-like children, 82 children with reactive and more typical behavior disorders, and an
“other” category which included 43 children with relatively mild impairments. Given that the data was archival from the 1960’s and 1970’s, parent report on the CBCL was not available; therefore, behaviors were rated by coders based on existing chart material (Rescorla, 1988). The CBCL’s syndrome scales were not used due to the differences in age and proportion of autistic children included in the clinical sample from those in the CBCL’s original standardization sample. Patterns of results across 2-, 3-, 4-, 5-, and 6-cluster solutions presented, with the first cluster to emerge as the Autistic/Bizarre cluster, which included five of the items from the CBCL (Rescorla, 1988). Overall, this study showed that when the eight-factor CBCL symptom profiles of preschool children manifesting a wide range of psychiatric and developmental disturbances were submitted to cluster analysis, autistic males differentiated from preschool boys with other behavior disturbances (Rescorla, 1988). A second study conducted in Germany also suggested that the CBCL can identify a general behavior pattern in children with Autistic Disorder (AD) (Bolte, Dickhunt, & Poustka, 1999). When compared to normative (N=2,856) and clinical samples (N= 1,655), a sample of 77 individuals with AD, ages 4 to 18, scored more than two standard deviations higher on the Total Problems scale of the CBCL. A closer look at the individual syndrome scales revealed that the individuals with AD scored higher than the normal and clinical group on all but one of the eight subscales. The exception was Somatic Complaints, on which individuals with AD scored below the clinical and normal population. This finding was not surprising given the lack of or inappropriate communication style that is typical of individuals with AD. The greatest score elevation for the group with AD was on the Social, Thought, and Attention subscales. In summary, the authors conclude that the CBCL shows sufficient ability to record clinically relevant behavioral patterns of autistic behavior (Bolte, Dickhunt, & Poustka, 1999). The ability of the CBCL to identify children with autism and related conditions in ages four through 11 was further investigated by Duarte, Bordin, de Oliveira, & Bird (2003). They found that the broad band scores on the CBCL (Internalizing Problems, Externalizing Problems, and Total Problems) did not distinguish children with autism and related conditions from children with other psychiatric disorders. The Externalizing Problems broad band score of autistic children was similar to those of normal schoolchildren and the Internalizing Problems broad band score did not distinguish children with autism and related conditions from children
with other psychiatric disorders. The authors also investigated the narrow band scales on the CBCL and found that both the Thought Problems and Autistic/ Bizarre subscales (the latter was developed by Rescorla, 1988) provided differentiation between the two groups (autism and related conditions and school children) with the Thought Problems scale alone supplying the best differentiation (Duarte, Bordin, de Oliveira, & Bird, 2003). When comparing autistic children with children with other psychiatric disorders, the Thought Problems, Autistic/Bizarre, and Aggressive Behavior Scales were able to distinguish the two groups. The most recent study examined the utility of the CBCL (1.5/5), which was developed for children ages 18 months to five years, 11 months of age, and the Gilliam Autism Rating Scale (Gilliam, 1995) in discriminating children with ASD (ages 36 to 71 months) from individuals referred for ASD evaluations but not found to be ASD (Sikora et al., 2008). Results revealed that the Withdrawn and Pervasive Developmental Problem subscales of the CBCL were higher among children with ASD than nonASD children (Sikora et al., 2008). The sensitivity and specificity of the CBCL were also better than that of the GARS suggesting that the CBCL is a useful behavior checklist for screening ASD. The CBCL was not originally intended to identify individuals with ASD, yet it has recently been used to screen for particular behaviors that are characteristic of ASD prior to pursuing more formal evaluations. While there have been four studies that have looked at the CBCL with various populations, there has yet to be a study to look at the distinct profiles produced by the CBCL with a group of six to 18 year olds in the United States. This line of research is important for several reasons. Primarily it would assist in identifying characteristics that differentiate groups of individuals (e.g., ASD, nonASD Controls, and Clinic-Referred for behavior difficulties). If differences are found, it would be helpful to know which areas certain groups of individuals are likely to be identified as requiring further screening and evaluation. This would be of extreme interest to psychologists and other professionals who specialize in working with unique populations of children, such as ASD. Secondly, this additional knowledge of all the CBCL subscales that contribute to these unique populations profiles is important for identification (e.g., individuals in need of further screening and evaluation) and intervention (e.g., individuals in need of behavioral remediation). Therefore, the purpose of this study is to
determine if patterns of normative strengths and weaknesses on the subscales of the CBCL are similar for individuals with ASD, nonASD Controls, and Clinic-Referred for behavior.
CHAPTER 2 REVIEW OF LITERATURE Broad Behavior Screening Instruments Behavior rating scales and checklists are commonly used tools in the assessment of internalizing and externalizing behaviors, social skills, and emotional functioning (Heckamena, Conroy, East, & Chait, 2000). These screening tools are capable of screening for a range of behavior disorders and are utilized in multiple settings. Contributing factors to their growing popularity include (a) provision of quantifiable information, which can be held to standards of reliability and validity; (b) efficient completion and scoring; (c) provision of systematic and organized information; (d) inclusion of normative data, allowing for comparisons of individual behaviors to larger groups; and (e) ability to compare ratings of multiple respondents across settings. (Hosp, Howell, & Hosp, 2003). In addition, there are numerous behavior rating scales commercially available (i.e., Behavioral Assessment System for Children [Reynolds & Kamphaus, 1992], Behavioral and Emotional Rating Scale [Epstein & Sharma, 1998], Behavior Rating Profile [Brown, 1990], Burks’ Behavior Rating Scales [Burks, 1996], Child Behavior Checklist [Achenbach & Rescorla, 2001], Conner’s Rating Scales [Conners, 1997], Revised Behavior Problem Checklist [Quay & Peterson, 1987], Social-Emotional Dimension Scale [Hutton & Roberts, 1986], and The Walker-McConnell Scale of Social Competence and School Adjustment [Walker & McConnell, 1995]). Components of Broad Behavior Screening Instruments In searching for a comprehensive broad behavior screener to identify individuals with an array of presenting problems, it is important to focus on some key factors: a) the ability of the instrument to be read and understood, b) allowance of information from multiple sources, c) ability to distinguish the individual from others, d) efficiency and economic availability, e) interpretability, f) high internal consistency, g) availability in multiple languages, h) norms based on large representative samples, i) consistency among scales, j) the ability to document for further decisions, and k) the ability to detect outcome evaluations. Frequency of Use of Broad Behavior Screening Instruments Assessment methods commonly associated with the process of behavioral assessment and screening, such as structured interviews, behavior checklists, rating scales, and systematic
observations have gained more prominence and acceptance over time. Shapiro and Heick (2004) surveyed 1000 practicing psychologist at a national convention about their use of assessment instruments with students who were referred for social, behavioral, and/or emotional problems. Results of the study indicated that although the use of intelligence, achievement, and visual- motor assessments remain a popular choice for school based psychological assessments, these remain only a portion of the assessments. The use of interviews, rating scales, and observations were reported in 60% to 90% of cases (Shapiro & Heick, 2004). These data suggest that the use of rating scales has substantially increased over the past 10 years. It is important to note that broad behavior screening instruments are typically included with referral information; thus psychologists and educational diagnosticians may look at the child’s behavior before anything else. Therefore, these screening instruments also play an integral role in planning the evaluation. Commonly Used Broad Behavior Screening Instruments Commonly used behavior rating scales and screeners are the Conner’s Rating Scales (Conners, Parker, Sitarenios, & Epstein, 1998), the Behavior Assessment System for Children (Reynolds & Kamphaus, 1992), and the Achenbach Child Behavior Checklist (Achenbach, 1991). The Connors Rating Scale – Revised (CRS-R) is a research and clinical tool for assessing classroom behaviors in children ages three to 17. Although it is mainly used for the assessment of ADHD, the subscales provide useful information for the assessment of conduct problems, cognitive problems, family problems, emotional problems, anger control, and anxiety problems (Conners, et al., 1998). The CRS-R provides three scales (parent, teacher, and adolescent) with long and short versions of each scale. While the long versions require more time to complete, they correspond more closely to the ADHD criteria of the DSM-IV. The short version is useful when time is limited or when repeated administrations are needed. CRS-R scales include 1) Oppositional, 2) Cognitive Problems/Inattention, 3) Hyperactivity, 4) Anxious-Shy, 5) Perfectionism, 6) Social Problems, 7) Psychosomatic, 8) Conners’ Global Index, 9) DSM-IV Symptom Subscales, and 10) ADHD Index. Computer and hand scoring are available which calculates standardized T-scores from raw scores, and provides a graphic display and a report of the results.
The coefficient alphas for internal reliability were highly satisfactory for the normative groups. For the long form there was a range from .73 to .94 and .86 to .94 for the short form of the CRS-R, indicating that the CRS-R items measure the same construct. The results indicate that the CRS-R is able to identify childhood and adolescent ADHD behavioral problems and psychopathology. Validity studies are continuing. The normative sample consisted of over 8000 cases. Data for parent, teacher, and self-report are from over 200 schools, 45 states, and 10 provinces throughout the U.S. and Canada. The Behavior Assessment System for Children (BASC; Reynolds & Kamphaus, 1992) is a coordinated system of instruments that evaluates the behaviors, thoughts, and emotions of children and adolescents ages four to 18. It focuses on assessing both adaptive and maladaptive behaviors. The BASC, parent and teacher forms, are measures described as providing a multidimensional approach to evaluating dimensions of behavior and personality in children, both positive and negative (Reynolds & Kamphaus, 2002). These questionnaires require a parent–guardian or teacher to rate a number of observable behaviors according to frequency evidenced during the past 6 months. Regarding school-age children, the BASC–PRS (Parent Rating Scale) has nine clinical scales including Hyperactivity, Aggression, Conduct Problems, Anxiety, Depression, Somatization, Atypicality, Withdrawal, and Attention Problems, and three adaptive scales including Adaptability, Social Skills, and Leadership. This emotional/behavioral measure also yields four composite scores including Externalizing Problems, Internalizing Problems, Behavioral Symptoms Index, and Adaptive Skills. The BASC–TRS (Teacher Rating Scale) includes the aforementioned nine scales and also adds the Learning Problems clinical scale, as well as the Study Skills adaptive scale and a School Problems composite score (Reynolds & Kamphaus, 2002). Each scale yields a mean T score of 50 and a standard deviation of 10. Confidence intervals and percentile ranks also are reported for each scale. Adequate reliability and validity has been shown for this measure (Reynolds & Kamphaus, 2002). The parent report version, test-retest reliability is reported at .88, interrater reliability is .57, internal consistency ranges from .80 to .90, and criterion validity was assessed and found to be acceptable. The teacher report version, test-retest reliability is purported at .90, interrater reliability ranges from .63 to .83, internal consistency ranges from .62 to .95, and criterion validity was assessed and found to be acceptable.
CBCL History The Achenbach System of Empirically Based Assessment (ASEBA) approach originated in the 1960s with Dr. Thomas M. Achenbach’s efforts as a graduate student in personality research to develop a differentiated picture of child and adolescent psychopathology (1966). The first scientific report of ASEBA findings was presented at the Society for Research in Child Development (Achenbach, 1965), and the first scientific publication was a monograph in the American Psychological Association’s Psychological Monographs series (Achenbach, 1966). The 1966 publication carefully describes the advancement in achieving a classification system based upon theoretical principles. The purpose of that paper was to “attempt to elucidate, in the child symptom domain, the relationship between the general symptom clusters found by Hewitt and Jenkins (1946), Phillips and Rabinovitch (1958), and Guertin (1952) and the specific functional syndromes employed in adult psychiatry and found in the Wittenborn (1953) study” (Achenbach, 1966). Additional purposes of the study were a) to obtain more differentiated empirical classification of child psychiatric cases, b) to classify individual cases according to factors, and c) explicate relationships between empirical groupings of symptoms. To answer the above questions, case histories from 300 males and 300 females, ages 4 to 16, were obtained from the University of Minnesota Hospital Child Psychiatry Unit. Each case history was rated on a 91 item symptom checklist that was constructed from items regularly appearing in prior studies, which involved minimal inference, mutually exclusive, and “not excessively molecular” (Achenbach, 1966). The symptoms were intercorrelated and factor analyzed, separately for each sex, by the principal factor method, and the factors were rotated to the varimax, quartimax, and oblimin criteria for simple structure. The first principal to emerge for both sexes was bipolar, with antisocial behavior “Externalizing” at one end and internal problems “Internalizing” at the other. For both sexes, factors labeled Somatic Complaints, Obsessions, Compulsions and Phobias, Rule Breaking Behavior, Aggressive Behavior, Hyperreactive Behavior, and Schizoid Thinking and Behavior were found. For the boys, a factor labeled Sexual Problems was revealed, and for the girls alone, factors labeled Depressive Symptoms, Anxiety Symptoms, Neurotic and Rule Breaking Behavior, Enuresis and Other Immaturities, and Obesity were found. Overall
results showed that the factors obtained can be used directly for the classification of child psychiatric cases for research purposes (Achenbach, 1966). In 1971, Achenbach and Lewis applied the empirically based approach in new research and laid the groundwork for the Child Behavior Checklist (CBCL). The conceptual framework for the ASEBA was outlined in relation to the developmental study of psychopathology in the first and second editions of the book Developmental Psychopathology (Achenbach, 1974, 1982). Based on the framework presented in Developmental Psychopathology, the first CBCL Manual was published in 1983 in collaboration with Craig Edelbrock, Ph.D. The CBCL Manual was followed by Manuals for the Teacher’s Report Form (TRF; Achenbach & Edelbrock, 1986) and the Youth Self-Report (YSR; Achenbach & Edelbrock, 1987). In 2001, the most recent edition of the CBCL/6-18 was released with the statistical foundation and normative data (Achenbach & Rescorla, 2001). Evaluation of the Child Behavior Checklist (CBCL) The CBCL was designed to address the problem of defining child behavior problems through direct observation and/or experience. It also has a teacher (Teacher Report Format-TRF), youth (Youth Self Report-YSR), and adult (Adult Self-Report) extension of the checklist. The CBCL can be self-administered or administered by an interviewer and consists of 118 items related to behavior problems which are scored on a 3-point scale ranging from not true to often true of the child. The CBCL is intended to serve as only one component of a multiaxial empirically based assessment, which includes parental report, teacher report, cognitive assessment, physical assessment, and direct assessment of the child. The CBCL exceeds other broad behavior screeners in that it includes 20 social competency items used to obtain parents’ reports of the amount and quality of their child’s participation in sports, hobbies, games, activities, organizations, jobs and chores, friendships, how well the child gets along with others and plays and works by himself/herself, and school functioning, which provides the clinician with a more intimate look at the child’s strengths and weaknesses. Although behavior rating scales such as the Conner’s and the BASC gather useful information, they do not allow the responder to personalize the child being assessed. However, the Achenbach Child Behavior Checklist provides open-ended questions, fill-in-the-blank answers, and has a special area devoted to independence, adaptive behavior, and personal
interests of the individual. These open-ended areas allow the clinician to gain a much needed personal look at the individual’s personal strengths and weaknesses according to the caregiver. This crucial information is then added to the quantitative data to determine the extent the child’s behavior is affecting his or her daily functioning. The 118 items of the CBCL break down each child’s behavior into the “broad band” scores of Internal Behavior Problems, External Behavior Problems, and Total Behavior Problems to give the clinician an idea of how the child is processing their emotions and outputting their behaviors. The 118 items yield scores for eight “narrow band” or problem scales: Anxious/Depressed, Withdrawn/Depressed, Somatic Complaints, Social Problems, Thought Problems, Attention Problems, Rule-Breaking Behavior, and Aggressive Behavior. The standard scores for these eight subscales have a mean of 50 and a standard deviation of 10. These scales were derived and defined by the items that make up each scale. For example, the Thought Problems scale is defined by the ratings items: can’t get his/her mind off certain thoughts/ obsessions; deliberately harms self or attempts suicide; hears sounds or voices that are not there; nervous movements or twitching; picks nose, skin, or other parts of body; plays with own sex parts in public; plays with own sex parts too much; repeats certain acts over and over; compulsions; sees things that are not there; sleeps less than most kids; stores up too many things he/she does not need; strange behavior; strange ideas; talks or walks in sleep; and trouble sleeping. Reliability-Item Scores Reliability of the item scores were computed using intraclass correleation coefficients from one way ANOVA’s. Inter-Interviewer ICC’s using three interviewers and 241 children matched for age, gender, ethnicity, and SES for a total sample of 723 were computed for the 20 competence items (.93) and for the 118 specific problem items (.96), both of which were significant (p<.001), indicating high inter-interviewer reliability (Achenbach & Rescorla, 2001). Test-retest reliability was computed over one week intervals. Overall ICC was 1.00 for the 20 competence items and .95 for the 118 specific problem items, which indicates high test-retest reliability (Achenbach & Rescorla, 2001).
Reliability-Scale Scores Internal Consistency. Cronbach alphas for each scale were moderately high, ranging from .63 to .79. These alphas are about as high as expected for scales that have as few as four items. For the syndrome/problem scales, the alphas ranged from .78 to .97. For the DSM-oriented scales, scales of items judged to be “very consistent” with the DSM-IV diagnostic categories, the alphas ranged from .72 to .91. Table 1 displays Cronbach alphas for each scale.
Table 1. Test-Retest Reliabilities and Alpha Coefficients for CBCL Scale Scores Empirically Based Test-Retest Alpha Coefficients Scales r
Test-Retest Reliability. To assess reliability in both the rank ordering and magnitude of scale scores, Pearson r correlations and t tests of differences were calculated between CBCL ratings of both referred and nonreferred children at mean intervals of 8 to 16 days. Total scale scores correlations ranged from .91 to .94. Table 1 displays high reliability coefficients for most scales with the majority of test-retest rs ranging from .80 to .90. It should be noted that there were four significant declines or test-retest attenuation effects (p<.05) in scores from the 8-day interval to the 16-day interval, which are marked with subscript a. All were expected by chance Anxious/Depressed .82 .84 Withdrawn/Depressed .89a,b .80 Somatic Complaints .92 .78 Social Problems .90 .82 Thought Problems .86 .78 Attention Problems .92 .86 Rule-Breaking Behavior .91 .85 Aggressive Behavior .90 .94 Internalizing .91a .90 Externalizing .92 .94 Total Problems .94a,b .97 Mean r .90