The Information and Picture Completion Dyad of the WISC-III as a Screening Test for Gifted Referrals.

The study investigated the relationship between Wechsler Intelligence Scale for Children-Third Edition Full Scale IQ scores and subtest dyads from the scale. The Information and Picture Completion dyad correlated with the Full Scale score and acted as an accurate screening instrument for gifted assessment. Although the Information and Picture Completion dyad was not found to be the best predictor of the Full Scale score, it was found to correlate significantly with Full Scale scores. The screening formula was aisle to accurately detect 70 of the 86 referrals (81%). Discriminant analysis revealed that the Picture Completion/Information dyad identified subjects significantly more accurately than the proportional chance criterion. The chosen cutoff sum of scaled scores for combined Picture Completion and Information (26) was found to be the most accurate gifted screen for that dyad.

Psychologists have searched for an accurate way to screen gifted referrals because, ultimately, only half of those nominated qualify for gifted programs (Fineman & Carran, 1986; Gear, 1976). Several studies have found screening instruments to be useful in gifted assessment. Benefits include saved time, money, and effort (Lustberg, Motta, & Naccari, 1990). Kaufman, Ishikuma, and Kaufman-Packer (1991) and Kaufman, Kaufman, Balzopal and McLean (1996) suggest that several factors be included in selecting short forms. These include administration and scoring time as well as psychometric and clinical utility. Fineman & Carran (1986) researched a short form consisting of the Vocabulary, Block Design, and Similarities WISC-R subtests (Wechsler, 1972). When that screening instrument was applied to the records of gifted referrals, 173 of 200 students were accurately identified, a success rate of 87%.

The WISC-III is a recent revision with little research published to date dealing with screening of gifted children. The WISC-III Full Scale Score correlates highly with that of the WISC-R (Sevier, Bain, & Hildman, 1994; Wechsler, 1991). Because of these strong correlations, research conducted on the WISC-R also might be applicable to the WISC-III.

Kaufman et al. (1996) and Sattler (1992) analyzed WISC-III standardization data to select short forms. Although a number of combinations exhibited high reliability, Kaufman et al. (1996) recommended a short form which included the Similarities, Arithmetic, Picture Completion, and Block Design subtests. Kaufman et al. (1996) concluded that this short form represented a compromise of psychometric and clinical soundness and practical utility. Sattler (1992) noted that if speed of administration and ease of scoring were priorities, then Information and Picture Completion represented a dyad worth cautious consideration. Caution is urged because the dyad does not provide as much clinical information as does the Vocabulary-Block Design dyad. An important advantage of the Information-Picture Completion dyad is that standard order of test administration is not disrupted. The Information-Picture Completion dyad also has a high correlation with the Full Scale as shown in Silverstein's (1970) study (r = .889) with the WISC. Therefore, if it is determined that a comprehensive evaluation is needed, the full test can be administered without switching test instruments and without discarding the information already obtained. Individually, Picture Completion was computed to have the second highest correlation to the Full Scale score from the Performance scale, while Information had the second highest correlation from the Verbal Scale (Wechsler, 1991). The present research was conducted to see whether the Information and Picture Completion dyad is an accurate screening device in gifted assessment, and if the linear combination of the Information and Picture Completion subtests is a significant predictor of the Full Scale score.

Method

The data included all children referred for gifted certification since the introduction of the WISC-III in a rural Tennessee school system. The WISC-III was administered by licensed school psychologists or interns in school psychology registered with the State Department of Education.

Participants

Information was obtained for each student describing age, gender, grade, ethnicity, and eligibility for gifted services. The Tennessee Department of Education (1993) requires that a child meet two of the following three criteria to be eligible for gifted services: a Full Scale intelligence test score of at least 127, achievement in at least one academic area which measures at or above the 96th percentile, and superior ideas and products as evidenced by projects or work samples.

Participants included 86 students who had been referred for comprehensive psychoeducational evaluations which utilized the WISC-III standard battery to determine eligibility for gifted certification. The mean age of the sample was 10 years, 3 months, with a standard deviation of 2.23 years. Ages ranged from 6 years, 0 months to 14 years, 3 months. The sample included 44 males and 42 females resulting in a ratio for the entire sample of 51.16% male to 48.84% female. There were 81 white students, 2 African-American students, and 3 Asian students. Of the 86 participants, 58 were determined to be eligible for the gifted program resulting in 67.40% being certified and 32.60% not being certified. All 86 students met the achievement criterion. Of the 58 who were certified gifted, 43 (48.84% of the original sample) met eligibility requirements by having intelligence test scores of at least 127. The other 15 students were considered eligible after committee review of their products and ideas.

Instruments

The Full Scale IQ score, Verbal Scale IQ score, Performance Scale IQ score, and subtest scaled scores were obtained from the Wechsler Intelligence Scale for Children--Third Edition (WISC-III; Wechsler, 1991). The WISC-III is an individually administered intellectual measure designed for use with children ages 6 years to 16 years, 11 months. The WISC-III yields three composite scores which have a mean of 100 and a standard deviation of 15: the Full Scale, the Verbal Scale, and the Performance Scale. It contains 13 subtests, 3 of which are considered supplementary to the main battery. The subtests have a mean scale score of 10 and a standard deviation of 3. The WISC-III has excellent reliability across the entire age range with reliability coefficients [is greater than or equal to] .89 for all three scales. Concurrent and construct validity have been found to be adequate. Average correlations between the standard battery subtests and Full Scale scores ranged from .33 to .74.

Procedures

After the data were collected, screening results were compared to actual results. Additionally, a screening method modeled after the design used by Fineman and Carran (1986) was applied to the 86 records. The sum of the Information and Picture Completion scaled scores was compared to a preset cutoff score. For the purposes of this study a score of greater than or equal to 26 (83rd percentile) was selected to screen the student into the eligible pool for further testing. A score of 25 or lower indicated that the subject was screened out and the testing should cease. The results of this screen were then compared with the actual placement results.

Results

The standard scores from the WISC-III were analyzed to determine the range, central tendency, and standard deviation of the students referred for gifted evaluation. Table 1 provides means and standard deviations for the WISC-III IQ's for the gifted, non-gifted, and total samples. The sample's descriptive statistics were: Full Scale IQ range = 92 to 148, M = 123.66, SD = 9.26; Verbal Scale IQ range = 89 to 145, M = 121.72, SD = 9.43; Performance Scale IQ range = 96 to 151, M = 121.56, SD = 11.77. Of the 42 female participants, 29 were found to be gifted. The same number of male participants qualified as gifted. The selection ratios were 69% and 66% respectively. The chi-square Which compared these selection rates was insignificant, p [is greater than] .05. Because of school district privacy policies, socio-economic status was not available for the sample. As there only were two African-American and three Asian-American participants, race was not interpreted.

Table 1
Means and Standard Deviations of WISC-III IQ's and Subtest Scores

                                  Total              Gifted
                                  M = 86             M = 58
                              M            SD           M
Full Scale IQ              123.66         9.26       127.93
Verbal IQ                  121.72         9.42       124.69
Performance IQ             121.56        11.77       125.86
Picture Completion          12.97         2.63        13.90
Information                 13.51         2.16        14.09
Coding                      12.66         2.93        12.83
Similarities                13.65         2.34        14.26
Picture Arrangement         14.06         3.03        14.72
Arithmetic                  13.69         2.54        14.34
Block Design                13.57         2.82        14.19
Vocabulary                  13.56         2.55        14.12
Object Assembly             12.43         2.70        13.19
Comprehension               14.26         2.98        14.98

                           Gifted           Not Gifted
                           M = 58             M = 28
                             SD            M            SD
Full Scale IQ               6.34        114.82         8.07
Verbal IQ                   7.38        115.57        10.31
Performance IQ              9.90        112.64        10.35
Picture Completion          2.09         11.04         2.62
Information                 1.88         12.32         2.26
Coding                      2.87         12.32         3.08
Similarities                2.14         12.39         2.27
Picture Arrangement         2.95         12.68         2.75
Arithmetic                  2.45         12.32         2.18
Block Design                2.52         12.29         3.00
Vocabulary                  2.10         12.39         3.01
Object Assembly             2.50         10.86         2.43
Comprehension               2.82         12.75         2.78

A discriminant analysis was used to test the accuracy of the discriminant function using the Information and Picture Completion dyad. The following discriminant function was obtained: Gifted group: [Z.sub.G] = -35.68 + 2.96 x Information + 2.09 x Picture Completion, Non-Gifted group: [Z.sub.NG] = -26.20 + 2.64 x Information + 1.60 x Picture Completion. A multivariate analysis of variance (MANOVA) showed that the linear combination of Information and Picture Completion had significant discriminating power for the Gifted and Non-Gifted groups, F (2, 83) = 18.19, p = .0001 (Wilks Lambda = .70). Subsequent analyses of variance (ANOVAs) revealed that both Picture Completion, [F (1, 84) = 29.89, p = .0002], and Information, [F (1, 84) = 14.59, p = .0003] were significant contributors for the discriminant function. It was determined that 70 of 86 students (81%) would have been accurately placed by the screen. This hit-ratio was highly significant compared to the proportional chance criterion [PCC = [(58/86).sup.2] + [(28/86).sup.2] = .5608], t (85) = 4.30, p [is less than] .01.

Table 2 displays the results of the Information and Picture Completion dyad screening with actual assessment results. True positives are those subjects correctly classified as gifted (N = 50). False positives are those participants identified by the screening device as potentially gifted but not meeting the entry requirements of the program (N = 8). True negatives are participants who were correctly screened out (N = 20). False negatives are participants who were incorrectly screened out (N = 8).

Table 2

Results of the WISC-III Screen for Giftedness Compared to True Subject Status

                                       True Status

                                Gifted Program Placement
                                Gifted         Not Gifted

Screen                    True Positives     False Positives
 Results
                Gifted             50                  8
                          False Negatives    True Negatives
             Non-gifted             8                 20

Correlations were determined between the individual subtests and Full Scale scores. All subtests were significantly correlated with the Full Scale score, p [is less than] .05. Correlations also were determined for all possible subtest dyads and the Full Scale scores. All subtest combinations were found to correlate significantly with the Full Scale score, p [is less than] .01. The correlations of the Full Scale with all Verbal-Performance dyads are presented in Table 3. These are uncorrected for the contamination of the overlap of the subtests with the Full Scale. The Picture Completion and Similarities dyad was most highly correlated with the Full Scale score, r(84) = .79, p [is less than] .01. The Information and Picture Completion dyad was also significant, r(84) = .72, p [is less than] .01. The difference between these correlations was tested with Fishers Z. It was not significant, Z = 1.211, p [is greater than].05.

Table 3

Correlations(*) of Each Possible Verbal-Performance Subtest Dyad with the Full Scale IQ

                       Picture        Coding        Picture
                     Completion                    Arrangement

Information              .72            .53            .70
Similarities             .79            .57            .70
Arithmetic               .77            .52            .65
Vocabulary               .78            .59            .64
Comprehension            .72            .50            .61

                        Block         Object
                       Design        Assembly
Information              .65            .67
Similarities             .79            .76
Arithmetic               .64            .69
Vocabulary               .74            .73
Comprehension            .75            .74

Note: These correlation are not corrected for contamination from overlap of the subtests with the Full Scale.

Discussion

The study found that the Information/Picture Completion dyad is an appropriate screen for gifted referrals. The screen would have accurately placed 50 of 58 students who eventually qualified for the gifted program, and it would have eliminated 28 complete exams. However, 8 of those students eliminated would have been inaccurately screened out of the assessment process. The ideal screening dyad would be as convenient as the Information/Picture Completion dyad and also accurately place all qualified subjects in the gifted program. It would also eliminate the number of superfluous assessments by screening out those that were unnecessary.

A weakness of the selected dyad was that the screener left out 8 students who were, in actuality, eligible for the gifted program. Fineman and Carran (1986) added a third step in their study. If the student scored within 4 points of the cutoff score, then a third subtest (Similarities) was administered. If the scale score on this subtest was [is greater than or equal to] 15, then the comprehensive WISC-III battery was administered.

To retain the aforementioned benefits of following standard subtest administration in the present study, the third subtest on the standard WISC-III protocol (Coding) would need to be administered. If Coding had been administered in this manner, and if a cutoff of [is greater than or equal to] 15 had been used, it would have included five of the eight students that had been eliminated. However, overall accuracy would have been reduced to 79% because an additional seven students would have been unnecessarily tested.

An easier way of dealing with this lack of sensitivity would be to lower the cutoff score. In this study, lowering the cutoff score to 23 would have accurately screened all 58 of the gifted referrals. However, lowering the cutoff score in this way would have caused the accuracy in overall placement to drop to 80% because an additional nine students would have been unnecessarily tested. Even lowering the cutoff score one point to 25 would have caused an overall loss in accuracy to 79%. If the cutoff score had been raised one point to 27, 14 more students would have been inaccurately eliminated which would have lowered accuracy to 69%. Therefore the chosen cutoff score with an accuracy of 81% proved to be the most accurate when compared to actual placement. However, deciding which screening formula is most useful in assessment is a judgment best left up to district-level program planners.

A limitation of the study was that the sample size (N = 86) was relatively small and geographically limited. Replications of the present study with a larger sample of gifted data would allow for a more accurate identification of an appropriate cutoff level.

Caution is essential when short forms are used (Kaufman et al., 1996). The best time to discuss the utilization of a short form for giftedness is during a screening team meeting. During this meeting the practical application of a short form screen could be suggested to the team to determine if it is appropriate for that particular referral. Assessment specialists can present probabilities and arguments for results, but ultimately, the team including the child's parents must accept the results for them to be effective.

References

Fineman, C., & Carran, D. (1986). An epidemologic approach to screening gifted students utilizing WISC-R subtests. Psychology in the Schools, 21, 142-147.

Gear, G. H. (1976) Accuracy of teacher judgment in identifying intellectually gifted children: A review of the literature. The Gifted Child Quarterly, 20, 478-489.

Kaufman, A.S., Ishikuma, T., & Kaufman-Packer, J.L.. (1991) Amazingly short forms of the WAIS-R. Journal of Psychoeducational Assessment, 9, 4-15.

Kaufman, A.S., Kaufman, J.C., Balzopal, R., & McLean, J.E., (1996) Comparison of three WISC-III short forms: Weighing Psychometric, clinical and practical factors. Journal of Clinical Child Psychology, 25, 97-105.

Lustberg, R., Motta, R., & Naccari, N. (1990). A model using the WISC-R to predict success in programs for gifted students. Psychology in the Schools, 27, 126-130.

Sattler, J. M. (1992). Assessment of children (3rd ed.). San Diego, CA: Author.

Sevier, R. C., Bain, S. K., & Hildman, L. K. (1994). Comparison of the WISC-R and WISC-III for Gifted Students. Roeper Review, 17(1), 39-42.

Silverstein, A.B. (1970). Reappraisal of validity of WAIS, WISC, and WPPSI short forms. Journal of Consulting and Clinical Psychology, 34, 1, 12-14.

Tennessee Department of Education. (1993). Special education manual (Publication Authorization No. 331033). Nashville: Division of Special Education.

Wechsler, D. (1972). Wechsler intelligence scale for children-revised. San Antonio: The Psychological Corporation.

Wechsler, D. (1991). Wechsler intelligence scale for children-third edition. San Antonio: The Psychological Corporation.

Jason N. Linville, James O. Rust and Jwa K. Kim, Department of Psychology, Middle Tennessee State University.

Correspondence concerning this article should be addressed to James O. Rust, P.O. Box 533, MTSU, Murfreesboro, TN 37132. E-mail: [e-mail].