The Aberrant Behavior Checklist and the Behavior Problems Inventory: convergent and divergent validity

Research in Developmental Disabilities 24 (2003) 391–404

The Aberrant Behavior Checklist and the Behavior Problems Inventory: convergent and divergent validity Johannes Rojahna,*, Michael G. Amanb, Johnny L. Matsonc, Erik Mayvilled a

George Mason University, Center for Cognitive Development, 4400 University Drive, MSN 2C6, Fairfax, VA 22030, USA b The Ohio State University Nisonger Center, Columbus, OH, USA c Louisiana State University, Baton Rouge, LA, USA d Pathways Strategic Teaching Center, Warwick, RI, USA

Received 6 January 2003; received in revised form 7 February 2003; accepted 28 February 2003

Abstract This study was designed to compare and cross-validate two rating instruments [the Aberrant Behavior Checklist (ABC) and the Behavior Problems Inventory (BPI)] for assessing maladaptive behavior. The BPI assesses three types of behavior problems: SelfInjurious Behavior (SIB), Stereotyped Behavior and Aggressive/Destructive Behavior. The ABC assesses five domains including these three. We collected data on 226 adults, mostly with severe or profound mental retardation, from a medium-sized developmental center. Individuals with elevated BPI scores generally had higher ABC scores; however, the extent of covariation differed across subscales. Similarly, multiple regression analyses showed that BPI subscales significantly but selectively predicted ABC subscale scores. Measures of differential diagnostic value (positive and negative predictive power, sensitivity, specificity and overall correct diagnostic efficiency) confirmed the anticipated partial overlap between instruments. Both instruments were used to rate participants with and without a Diagnosis of Stereotyped Movement Disorder. BPI, SIB and Stereotypy subscale composite had stronger positive predictive power than the ABC Stereotypy scale, while the ABC had higher negative predictive power and greater overall diagnostic efficiency. Thus, the ABC and the

*

Corresponding author. Tel.: þ1-703-993-4241. E-mail address: [email protected] (J. Rojahn).

0891-4222/$ – see front matter # 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0891-4222(03)00055-6

392

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

BPI cross-validated one another where expected, and they diverged for subscales thought to have little relationship. # 2003 Elsevier Ltd. All rights reserved. Keywords: Challenging behavior; Behavior problems; Psychopathology; Mental retardation; Assessment; Rating scales; Validity

Despite advances in our knowledge during the past decades, behavior problems continue to be one of the primary concerns for individuals with mental retardation and those responsible for their welfare. Dangerous behaviors such as aggressive outbursts or severe self-injury can be common and present barriers to successful integration into the community and to unrestricted access to available educational, vocational and leisure opportunities. Assessment instruments for behavior problems in mental retardation are important tools for clinical research and outcome studies in this field. For instance, they are used in selecting individuals for clinical trails, classifying them for research protocols, and for assessing intervention effects. One can distinguish between two kinds of behavior rating scales, depending on their scope: (a) generalized (or broad-band) and (b) specialized (or narrow-band) behavior scales. Generalized scales cover a relatively wide array of different conditions such as psychopathology, which consists of a broad spectrum of heterogeneous conditions such as mood disorders, schizophrenia or personality disorders. Some of these generalized scales were developed deductively, where items were developed to mirror diagnostic categories from nosological systems such as the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition Text Revision (DSM-IV) (American Psychiatric Association, 2000). Examples of such instruments are the Reiss Screen for Maladaptive Behavior (Reiss, 1988), and the Diagnostic Assessment for the Severely Handicapped-II (Matson, Gardner, Coe, & Sovner, 1991). Other rating scales were developed inductively, starting with clinically observed behaviors and developing diagnostic categories (subscales) by factor analysis. One example is the Aberrant Behavior Checklist (ABC) (Aman & Singh, 1986). More specialized behavior rating scales, on the other hand focus on narrowly defined conditions such as the Compulsive Behavior Checklist and the Stereotypy Checklist (Bodfish et al., 1995), the Tic Checklist (Rosenquist, Bodfish, & Thompson, 1997), the Dyskinesia Identification System: Condensed User Scale (Sprague, Kalachnik, & Slaw, 1989), and the Behavior Problems Inventory (BPI) (Rojahn, Matson, Lott, Esbensen, & Smalls, 2001). The ABC and the BPI are both empirically developed scales for undesirable behavior. Whereas the ABC assesses a wider range of behavior disorders, including self-injury, stereotypy, aggressive and destructive behaviors, the BPI addresses these three behavior categories exclusively and in greater detail. We had three aims. First, we wanted to assess the extent to which BPI and ABC scores correlated, with the expectation that the three BPI subscales would predict analogous ABC subscales scores. Second, we anticipated that the two instruments would yield both overlapping and independent information. Third, we tested the

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

393

diagnostic value of the two instruments in relation to a clinical DSM-IV Diagnosis of Stereotyped Movement Disorder.

1. Method 1.1. Participants Participants were 226 adults with mental retardation who lived at a mediumsize developmental center in Louisiana. Demographic characteristics (gender, age and level of mental retardation) are shown in Table 1. Over 45% of the participants had at least one clinical psychiatric diagnosis. One individual had three psychiatric diagnoses, 13 had two, and 103 had one. Stereotyped Movement Disorder (n ¼ 39; 17.3%) was the most prevalent diagnosis, followed by Bipolar Disorder (n ¼ 11; 4.9%), Pervasive Developmental Disorder (n ¼ 10; 4.4%), and Autistic Disorder (n ¼ 10; 4.4%). Diagnoses of schizophrenia and psychosis were rare and no diagnosis of major depressive disorder was found. 1.2. Instruments Aberrant Behavior Checklist. The ABC (Aman & Singh, 1986) is an informantbased problem behavior rating scale that was developed by factor analysis on a large number of residents in developmental centers and group homes (Aman, Table 1 Demographic features (N ¼ 226) n

%

Gender Male Female

126 100

55.8 44.2

Level of mental retardation Mild Moderate Severe Profound Unspecified

7 16 41 141 22

3.1 7.1 18.1 61.9 9.7

Age in years 20–38 39–46 47–57 58–91

48 51 66 61

21.2 22.6 29.2 27.0

Ethnicity White Black Unknown

157 49 20

69.5 21.7 8.8

394

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

Singh, Stewart, & Field, 1985a). The ABC was primarily intended for treatment (e.g., psychotropic medication) evaluation (Aman, 1991), although it has been employed as a general assessment tool as well (e.g., for behavior phenotype studies). It consists of 58 items, each scored on a 4-point scale (0: not a problem, through 3: problem is severe in degree). The items fall into five subscales: (1) Irritability, Agitation, Crying (15 items), (2) Lethargy, Social Withdrawal (16 items), (3) Stereotypic Behavior (7 items), (4) Hyperactivity, Non-Compliance (16 items), and (5) Inappropriate Speech (4 items). A number of psychometric studies have shown that the ABC is a reliable and valid behavior rating instrument (e.g., Aman, Singh, Stewart, & Field, 1985b; Aman, Singh, & Turbott, 1987; Bihm & Pointdexter, 1991; Paclawskyi, Matson, Bamburg, & Baglio, 1997; see Aman, 2002 for summary). Behavior Problems Inventory. The BPI is a 49-item informant-based behavior rating scale for behavior problems in individuals with mental retardation. The BPI-01 edition, described by Rojahn et al. (2001), was used in this study. This BPI was originally designed as a survey instrument for epidemiological research (Rojahn, 1984, 1986), and it has been refined further since. It has three subscales that were validated by confirmatory factor analysis (Rojahn et al., 2001; Sturmey, Sevin, & Williams; 1995): (a) Self-Injurious Behavior (14 items), (b) Stereotyped Behavior (24 items), and (c) Aggressive/Destructive Behavior (11 items). To be rated, the behavior must have occurred at least once during the previous 2 months. Each item is scored on two scales, a 5-point frequency scale (never: 0, through hourly: 4) and a 4-point severity scale (no problem: 0, through severe problem: 3). Each of the three subscales is prefaced by a generic definition that applies to all items within the subscale. The BPI was found to be reasonably reliable and valid (e.g., Rojahn et al., 2001; Sturmey, Fink, & Sevin, 1993). 1.3. Respondents and interviewers The ABC was administered either by a doctoral student in clinical psychology, by a licensed Ph.D. psychologist, a senior graduate student, or a master’s level psychologist; doctoral students administered the BPI. A licensed Ph.D. psychologist trained all raters. The informants were direct care staff members who worked with the client for at least 6 months. Interviewers tried to enlist staff members who were most familiar with a given participant.

2. Results Data analysis was performed on subscale and total scores. Only the frequency scales of the BPI were used here since correlations between the frequency and severity scales were high (.93 for Self-Injurious Behavior, .92 for Stereotyped Behavior, and .87 for Aggression/Destruction; N ¼ 226). Table 2 contains the means, standard deviations, and the range of the ABC and BPI subscale scores. Neither the ABC nor the BPI had normally distributed data.1

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

395

Table 2 Means, standard deviations and range of the BPI and the ABC scores (N ¼ 226) M BPI Subscales

Self-Injurious Behavior Stereotypy Aggression/Destruction

Total score

Minimum

Maximum

1.8 4.1 2.1

3.3 6.4 4.3

0 0 0

25 39 25

8.1

10.2

0

64

3.4 3.0 2.0 3.5 0.4

5.5 5.1 3.7 6.4 1.4

0 0 0 0 0

35 33 18 40 12

12.2

15.6

0

92

Total score ABC Subscales

SD

Irritability Lethargy Stereotypic Behavior Hyperactivity Inappropriate Speech

2.1. Categorical or group comparisons The first part of the data analysis was designed to test the hypothesis that individuals with elevated BPI subscale scores should have higher scores on some ABC subscale scores, as compared to individuals with low BPI scores. Three sets of analyses were performed. The first one was based on a split of the total group into two groups: One group with a BPI total score of 0, and a second group with BPI total scores higher than 0 (see Fig. 1). We conducted a MANOVA with the ABC subscale scores as the dependent variables. The top panel in Table 3 shows that the multivariate test (Wilks’ l ¼ :92, p ¼ :003) and the between-subjects tests for the ABC subscales Irritability, Lethargy, Stereotypy, and Hyperactivity were significant. The two groups did not differ significantly with respect to their Inappropriate Speech scores. Since the group split of the first analysis could be seen as arbitrary, we broke down the sample again, this time into three groups: (a) one with a BPI total score of 0 (i.e., no BPI behavior problems); (b) one with BPI total scores larger than 0, but with no items with a score of 4 on the SIB or the Aggression/Destruction subscales (i.e., individuals with moderate behavior problems); (c) one with at least one item 4 on either the SIB or the Aggression/Destruction subscales (i.e., individuals with highly frequent behavior problems; see Fig. 2). Again, a MANOVAwas computed with the ABC subscale scores as the dependent variables. The multivariate effect was again significant (Wilks’ l ¼ :88, p ¼ :000). Once again, all subscales except Inappropriate Speech showed significant group differences (see Table 3, second panel). In the third series of group comparisons the unique effect of each of the three BPI subscale scores on the ABC subscale scores was determined, while controlling the variance of the two other BPI subscales. Therefore, three MANCOVAs were computed, one for each subscale. (a) In the first MANCOVA the BPI–SIB dichotomous subscale score (group 1: SIB ¼ 0; group 2: SIB > 0) was the

396

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404 5

Mean ABC Subscale Score

4

3

2

1 BPI = 0 BPI > 0

0 Irritability

Stereotypy Lethargy

Speech Hyperactivity

Fig. 1. Mean ABC subscale scores for individuals with a BPI total score of 0, and higher than 0.

grouping variable, the ABC subscale scores were the multiple dependent variables, and BPI Stereotyped Behavior and Aggression/Destruction subscale scores were the covariates. The overall multivariate effect was significant (Wilks’ l ¼ :91, p ¼ :001), and so were the effects for Irritability, Lethargy, and Hyperactivity (Table 3, third panel). Participants in these SIB groups did not differ with regard to their ABC Stereotypy and Inappropriate Speech scores. (b) The next MANCOVA showed that the multivariate test for BPI Stereotyped Behavior was statistically significant (Wilks’ l ¼ :93, p < :01), and that the between-subjects tests for ABC Irritability, Lethargy, Stereotypy, and Hyperactivity were also significant (see Table 3, fourth panel). (c) In the third MANCOVA for BPI Aggression/ Destruction, the multivariate test was significant (Wilks’ l ¼ :94, p < :05), as were the effects for ABC Irritability, Hyperactivity, and Inappropriate Speech. 2.2. Linear regression In addition to the categorical group comparisons linear regression analyses were computed to examine which of the three BPI subscales would significantly predict the five ABC subscales (see Table 4). The ABC Irritability subscale was significantly, but weakly, predicted by the BPI subscales SIB and Aggression/ Destruction (R2 ¼ :12). The ABC Lethargy and ABC Stereotypy subscales were significantly predicted by BPI SIB and Stereotypy. The highest coefficient of

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

397

Table 3 Summary table of the MANOVA and MANCOVA results Tests

F

df

Significance

MANOVA: two-group comparisona Multivariate: Wilks’ l ¼ .92, between-subjects Irritability Lethargy Stereotypy Hyperactivity Speech

3.77 9.83 10.93 5.79 8.67 0.29

5/220 1/224 1/224 1/224 1/224 1/224

.003 .002 .001 .017 .004 .594

MANOVA: three-group comparisonb Multivariate: Wilks’ l ¼ .88, between-subjects Irritability Lethargy Stereotypy Hyperactivity Speech

2.77 8.25 6.16 3.17 8.30 0.27

10/438 2/222 2/222 2/222 2/222 2/222

.003 .000 .002 .044 .000 .767

MANCOVA: SIB ¼ 0 versus SIB > 0 Multivariate: Wilks’ l ¼ .91, between-subjects Irritability Lethargy Stereotypy Hyperactivity Speech

4.39 9.88 12.61 2.61 8.30 0.37

5/218 1/222 1/222 1/222 1/222 1/222

.001 .002 .000 .108 .004 .545

3.39 7.74 6.48 8.90 5.50 0.54

5/218 1/222 1/222 1/222 1/222 1/222

.006 .006 .012 .003 .020 .465

MANCOVA: Aggression/Destruction ¼ 0 versus Aggression/Destruction > 0 Multivariate: Wilks’ l ¼ .94, between-subjects 2.59 5/218 Irritability 7.63 1/222 Lethargy 0.24 1/222 Stereotypy 0.00 1/222 Hyperactivity 6.01 1/222 Speech 4.58 1/222

.027 .006 .622 .984 .015 .033

MANCOVA: Stereotypy ¼ 0 versus Stereotypy > 0 Multivariate: Wilks’ l ¼ .93, between-subjects Irritability Lethargy Stereotypy Hyperactivity Speech

a

Group 1: BPI total score: 0; group 2: BPI total score: >0. Group 1: BPI total score: 0; group 2: BPI total score: >0, but with SIB or Aggression/ Destruction items: <4; group 3: BPI total scores: >0, and at least one SIB or the Aggression/ Destruction item of: 4. b

multiple determination was found for the ABC Stereotypy subscale, suggesting that 40% in the variation in the ABC Stereotypy subscale score could be explained by variations in the BPI subscale scores (Schroeder, Sjoquist, & Stephan, 1986). The BPI subscale SIB predicted Hyperactivity, and the ABC Inappropriate Speech was not predicted by any of the three BPI subscales. In summary, the BPI SIB

398

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404 7

Mean ABC Subscale Score

6

5

4

3

2 BPI = 0 1

BPI > 0 SIB or agg frequent

0 Irritability

Stereotypy Lethargy

Speech

Hyperactivity

Fig. 2. Mean ABC subscale scores for (1) individuals without behavior problems (BPI total score of 0), (2) individuals with moderate behavior problems according to the BPI (BPI total score larger than 0, but without a SIB or Aggression/Destruction item with a score of 4), and (3) individuals with at least one severe behavior problem (BPI total score larger than 0 and with at least one item on the SIB or the Aggression/Destruction scale of 4).

significantly predicted the ABC Irritability, Lethargy, Stereotypy and Hyperactivity, while the BPI Stereotypy predicted ABC Lethargy and Stereotypy. BPI subscale Aggression/Destruction only predicted ABC Irritability. 2.3. Differential diagnostic value Behavior rating scales are often used to classify individuals into ‘‘cases’’ (individuals with a certain characteristic) and ‘‘non-cases’’ (individuals without that characteristic). Their precision in doing so is typically expressed by five interrelated measures of differential diagnostic value which range from 1.0 to 1.0: positive and negative predictive power, sensitivity, specificity (Brenner, 1996) and overall correct diagnostic efficiency (Blanchard, Jones-Alexander, Buckley, & Forneris, 1996; cf. Andrykowski, Cordova, Studts, & Miller, 1998). Positive predictive power is the proportion of individuals correctly classified as cases. Negative predictive power is the proportion of non-cases correctly classified as non-cases. Sensitivity is the probability that a case is correctly identified as a case. Specificity is the probability that a non-case is correctly identified as a

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

399

Table 4 Predicting ABC subscales from BPI subscales with the predictors Aggression, Stereotypy, and SIB by multiple regression analysis R2

B

SE

b

Irritability

.12

SIB Stereotypy Aggression

.41 .01 .18

.12 .06 .09

.25** .02 .14*

Lethargy

.31

SIB Stereotypy Aggression

.34 .15 .08

.12 .05 .08

.22** .19** .07

Stereotypy

.40

SIB Stereotypy Aggression

.18 .19 .06

.08 .04 .06

.16* .34*** .07

Hyperactivity

.34

SIB Stereotypy Aggression

.54 .05 .11

.15 .07 .10

.28*** .05 .08

Inappropriate Speech

.13

SIB Stereotypy Aggression

.02 .00 .03

.03 .02 .02

.05 .02 .10

R2  100 measures the percentage of the variation in the dependent variable that is explained by variations in the independent variable. * p < :05. ** p < :01. *** p < :001.

non-case. Overall correct diagnostic efficiency is the probability that individuals are correctly classified as cases or non-cases. As the ABC was designed to assess SIB, stereotypic behaviors, and aggressive and destructive behaviors, differential diagnostic values can be determined for the ABC in comparison to the BPI. Diagnostic efficiency shifts with changing cut-off criteria. Therefore, measures could be calculated for countless cut-off score scenarios. For the purpose of this paper, diagnostic efficiency was restricted to a single scenario, namely the dichotomous ABC total score (0 vs. >0) predicting dichotomous BPI total and subscale scores (0 vs. >0). Table 5 indicates that the positive predictive value of the ABC total was very high with regard to BPI total (namely .87; see first row last column in Table 5) and also with specific behavior problems (ranging from .89 to .91; see last column in the top rows of each block). This means that if a subject has an ABC total score larger than 1, 87% of the cases will have a BPI total score larger than 1 as well; 89% will have a BPI SIB score larger than 1, 91% will have a BPI Stereotypy score larger than 1, and 89% will have a BPI Aggressive/Destructive Behavior score larger than one. Negative predictive power, which is inversely related to positive predictive power, was lower ranging from .26 to .36. Sensitivity of ABC total ranged from .40 to .74. The overall diagnostic efficiency of the ABC total score with regard to the BPI total score was .71 (and ranged from .48 to .61 for BPI

400

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

Table 5 Sensitivity, specificity, positive and negative predictive power, and overall diagnostic efficiency of the ABC vis-a´ -vis behavior problems as indicated by the BPI scores BPI

ABC Irritability

Lethargy

Stereotypy

Hyperactivity

Speech

Total

Total

Pos. Pred. Neg. Pred. Sensitivity Specificity O.D.E.

.58 .54 .73 .38 .57

.58 .68 .79 .43 .61

.42 .83 .84 .40 .55

.64 .65 .80 .46 .64

.18 .90 .80 .34 .41

.87 .36 .74 .57 .71

SIB

Pos. Pred. Neg. Pred. Sensitivity Specificity O.D.E.

.69 .55 .48 .75 .61

.64 .59 .48 .73 .61

.48 .75 .54 .71 .65

.73 .57 .51 .78 .63

.19 .87 .46 .64 .61

.89 .26 .42 .80 .50

Stereotypy

Pos. Pred. Neg. Pred. Sensitivity Specificity O.D.E.

.62 .54 .57 .60 .58

.67 .67 .66 .68 .67

.51 .83 .75 .64 .68

.68 .58 .61 .65 .63

.21 .90 .66 .54 .56

.91 .31 .56 .78 .61

Aggression

Pos. Pred. Neg. Pred. Sensitivity Specificity O.D.E.

.68 .54 .45 .75 .59

.52 .52 .38 .66 .52

.38 .69 .41 .67 .58

.68 .53 .45 .75 .58

.28 .92 .66 .70 .69

.89 .26 .40 .80 .48

Pos. Pred.: positive predictive power; Neg. Pred.: negative predictive power; O.D.E.: overall diagnostic efficiency.

subscales). The remaining cells of Table 5 show the relationships for all subscales of the ABC and BPI. Finally, ABC Stereotypy and the combined score from the BPI SIB and Stereotypy subscales were used to assess differential diagnostic values in relation to a clinical DSM-IV Diagnosis of Stereotyped Movement Disorder. Table 6 shows that the combined SIB and Stereotypy total score of the BPI had stronger positive predictive power than the ABC Stereotypy scale (.71 vs. .61). The ABC on the other hand, was Table 6 Sensitivity, specificity, positive and negative predictive power, and overall diagnostic efficiency of the ABC Stereotypy subscale, the BPI SIB and/or Stereotypy subscale, and the total score classifying a clinical DSM-IV Diagnosis of Stereotyped Movement Disorder

Positive predictive power Negative predictive power Sensitivity Specificity Overall diagnostic efficiency

ABC Stereotypy

BPI SIB/Stereotypy

.61 .72 .33 .89 .70

.71 .42 .21 .87 .47

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

401

superior in terms of negative predictive power and overall diagnostic efficiency. Thus, whereas the ABC more correctly classified individuals into cases and noncases (particularly non-cases), the BPI tended to identify cases more correctly.

3. Discussion To recap, this study was designed to assess cross-validity of the ABC and BPI. This was accomplished by exploring three questions. We predicted that individuals with elevated BPI scores would also have higher ABC subscale scores than individuals with low BPI scores. The data unequivocally confirmed that, and they also indicated that the covariance between the two instruments differed across subscales. One would expect this if the scales were to cross-validate each other in a convergent and divergent fashion. For instance, the three MANCOVAs revealed that the ABC Irritability subscale scores were significantly higher among BPI SIB, stereotypy, and aggression/destruction cases in comparison to their respective non-case peers (bottom panels of Table 3). This positive relationship is consistent with similarities in item contents. The ABC Irritability subscale includes items such as ‘‘Injures self,’’ ‘‘Does physical harm to self,’’ which correspond to the BPI SIB and Aggression/Destruction subscale content. This can be considered as a form of convergent validity. On the other hand, the regression analyses determined that only 12% of the variance of ABC Irritability was explained by BPI SIB and Aggression/Destruction (see Table 4). Clearly, these subscales identify different sorts of problem behavior. ABC Lethargy subscale scores were significantly higher among BPI SIB cases and stereotypy cases (Table 3), and the same BPI subscales predicted the ABC Lethargy, while the BPI subscale Aggression/Destruction did not (see Table 4). This is consistent with the fact that the ABC Lethargy content describes autisticlike behavior, a condition associated with repetitive and Self-Injurious Behavior. The fact that the ABC Stereotypy subscale scores were significantly associated with BPI Stereotypy and that 40% of its variance was explained by the BPI SIB and Stereotypy subscales can also be seen as strong evidence of convergent validity. Conversely, the lack of correspondence between BPI Aggression/ Destruction and ABC Stereotypy and between BPI Aggression/Destruction and ABC Stereotypy are evidence for divergent validity. As for the ABC Hyperactivity/Non-compliance subscale, MANCOVAs showed significantly higher scores for SIB cases, stereotypy cases, and aggression/ destruction cases (Table 3). However, multiple regression analysis indicated that only the BPI SIB scores significantly predicted Hyperactivity subscale scores (Table 4). The item content of the ABC Hyperactivity and the BPI SIB subscales are not similar, and one can only speculate about the reasons for their relationship. ABC Inappropriate Speech did not correlate with most BPI scores. However, BPI Aggression/Destruction scores did correspond with ABC Inappropriate Speech scores. While it seems intuitively reasonable that the BPI SIB and BPI Stereotypy subscales would be unrelated to ABC’s Inappropriate Speech, it is not

402

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

immediately clear why ABC Inappropriate Speech and BPI Aggression/Destruction were related. The expectation that the two instruments would yield partly overlapping information but that they would generate independent data too was confirmed by the results of the diagnostic efficiency parameters. If any two of the instruments or their subscales had been completely redundant, the respective sensitivity or positive predictive values in Table 5 should have been 1.00. Obviously, this was not the case. In other words, it seems as if each of the instruments contributed unique information and that information would be lost if only one of the two instruments were used. This was particularly true when the main focus of assessment was SIB, Stereotyped Behavior, or Aggressive/Destructive Behavior. Last, we compared the predictive value of the two instruments against an external criterion, namely the presence or not of a DSM-IV Diagnosis of Stereotyped Movement Disorder. The combined BPI SIB and Stereotypy scores correctly classified a higher proportion of Stereotyped Movement Disorder cases than the ABC. The ABC on the other hand, was superior to the BPI subscales in terms of correctly identifying a larger proportion non-Stereotyped Movement Disorder cases and in terms of the overall diagnostic efficiency. Strengths of this study included a prospective design, use of parallel behavior rating scales with established psychometric properties, and specially trained data collectors. Perhaps the main weakness was the select group of subjects who were rated. It remains to be seen to what extent the present data may generalize to higher-functioning individuals based in the community. Also, the subscale means appeared somewhat lower than expected based on our previous experience. This may be due in part to the advanced ages of many of our subjects, 56% of whom were 47 years of age or older. Although the data were not normally distributed, skewed distributions reflecting maladaptive behavior data are common with most symptom scales and are very difficult to avoid. Although MANOVA and MANCOVA are relatively robust, even with non-normal data, the results should be considered with appropriate caution. To conclude, SIB, Stereotyped Behavior, and Aggressive/Destructive Behavior as assessed by the BPI were significantly and positively related to a number of ABC subscales in a selective and clinically reasonable fashion. The BPI and the ABC generated related but not entirely overlapping information. Despite the similarities in some of the subscale labels between these instruments, they also yielded information that was unavailable from the other instrument. This indicates that clinicians need to familiarize themselves with both scales’ item content if they wish to use one instrument only. Alternatively, it is reasonable to employ both instruments for assessing the same individuals.

Note 1. Lack of a normal distribution is characteristic for maladaptive behavior data. Typically, they have a unimodal distribution with positive skewness

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

403

and kurtosis, which was the case with the present data set as well. The BPI total score had a skewness of 2.0 and a kurtosis of 6.2; the ABC total score had a skewness of 2.1 and a kurtosis of 5.1.

Acknowledgments This research was supported in part by Grant #R215K010121 from the US Department of Education. The authors appreciate the critical comments and helpful suggestions by John W. Jacobson on an earlier draft of this manuscript.

References Aman, M. G. (1991). Assessing psychopathology and behavior problems in persons with mental retardation: A review of available instruments. Rockville, MD: U.S. Department of Health and Human Services. Aman, M. G. (2002). Annotated bibliography on the Aberrant Behavior Checklist (ABC). Washington, DC: Author. Aman, M. G., & Singh, N. N. (1986). Aberrant Behavior Checklist: Manual. East Aurora, NY: Slosson Educational Publications. Aman, M. G., Singh, N. N., Stewart, A. W., & Field, C. J. (1985a). The Aberrant Behavior Checklist: A behavior rating scale for the assessment of treatment effects. American Journal on Mental Deficiency, 89, 491–495. Aman, M. G., Singh, N. N., Stewart, A. W., & Field, C. J. (1985b). Psychometric characteristics of the Aberrant Behavior Checklist. American Journal on Mental Deficiency, 89, 492–502. Aman, M. G., Singh, N. N., & Turbott, S. H. (1987). Reliability of the Aberrant Behavior Checklist and the effects of variations in instructions. American Journal on Mental Deficiency, 92, 237–240. American Psychiatric Association. (2000). Diagnostic and Statistical Manual of Mental Disorders, 4th Edition Text Revision. Washington, DC: Author. Andrykowski, M. A., Cordova, M. J., Studts, J. L., & Miller, T. W. (1998). Posttraumatic stress disorder after treatment for breast cancer: Prevalence of diagnosis and use of the PTSD ChecklistCivilian Version (PCL-C) as a screening instrument. Journal of Consulting and Clinical Psychology, 66, 586–590. Bihm, E., & Pointdexter, A. R. (1991). Cross-validation of the factor structure of the Aberrant Behavior Checklist. American Journal on Mental Retardation, 96, 209–211. Blanchard, E. B., Jones-Alexander, J., Buckley, T. C., & Forneris, C. A. (1996). Psychometric properties of the PTSD Checklist (PCL). Behavior Research and Therapy, 34, 669–673. Bodfish, J. W., Crawford, T. W., Powell, S. B., Parker, D. E., Golden, R. N., & Lewis, M. H. (1995). Compulsions in adults with mental retardation: Prevalence, phenomenology, and co-morbidity with stereotypy and self-injury. American Journal on Mental Retardation, 100, 183–192. Brenner, H. (1996). Measures of differential diagnostic value of diagnostic procedures. Journal of Clinical Epidemiology, 49, 1435–1439. Matson, J. L., Gardner, W. I., Coe, D. A., & Sovner, R. (1991). A scale for evaluating emotional disorders in severely and profoundly mentally retarded person. British Journal of Psychiatry, 159, 404–409. Paclawskyi, T. R., Matson, J. L., Bamburg, J. W., & Baglio, C. S. (1997). A comparison of the Diagnostic Assessment for the Severely Handicapped-II (DASH-II) and the Aberrant Behavior Checklist (ABC). Research in Developmental Disabilities, 18., 289–298. Reiss, S. (1988). The Reiss Screen for Maladaptive Behavior test manual. Worthington, OH: IDS.

404

J. Rojahn et al. / Research in Developmental Disabilities 24 (2003) 391–404

Rojahn, J. (1984). Self-injurious behavior in institutionalized, severely/profoundly retarded adults: Prevalence data and staff agreement. Journal of Behavioral Assessment, 6, 13–27. Rojahn, J. (1986). Self-injurious behavior and stereotypic behavior in non-institutionalized mentally retarded people. American Journal on Mental Deficiency, 91, 268–276. Rojahn, J., Matson, J. L., Lott, D., Esbensen, A. J., & Smalls, Y. (2001). The Behavior Problems Inventory: An instrument for the assessment of self-injury, stereotyped behavior and aggression/ destruction in individuals with developmental disabilities. Journal of Autism and Developmental Disorders, 31, 577–588. Rosenquist, P. B., Bodfish, J. W., & Thompson, R. (1997). Tourette syndrome associated with mental retardation: A single-subject treatment study with haloperidol. American Journal on Mental Retardation, 101, 497–504. Schroeder, L. D., Sjoquist, D. L., & Stephan, P. E. (1986). Understanding regression analysis. Newbury Park, CA: Sage. Sprague, R. L., Kalachnik, J. E., & Slaw, K. M. (1989). Psychometric properties of the Dyskinesia Identification System: Condensed User Scale. Mental Retardation, 27. Sturmey, P., Fink, C., & Sevin, J. (1993). The Behavior Problems Inventory: A replication and extension of its psychometric properties. Journal of Developmental and Physical Disabilities, 5, 327–336. Sturmey, P., Sevin, J., & Williams, D. E. (1995). The Behavior Problems Inventory: A further replication of its factor structure. Journal of Intellectual Disability Research, 39, 353–356.