Perceprual and Motor Skills, 1979, 48, 699-702.

@ Perceptual and Motor Skills 1979

ITPA INTERCHANNEL DIFFERENCES THOMAS J. WENTLAND1

University of Mississippi Summary.-The extent to which auditory and visual channel part-score means differ typically was studied by reference to the performance of 962 average children in the normative sample of the Illinois Test of Psycholinguistic Abilities. A standard error of measurement difference of 1.87 scaled score points was obtained, such indicating comparatively little difference in the mean performance of average children on the two communication channels of this test. Differences between means of auditorg and visual channels greater than 3.09 or 4.36 scaled score points are presented as statistically significant at the 5 or 1 percent level of confidence, respectively. Significant part-score differences suggest atypical performance in the channel with the lesser mean scaled score and may indicate the presence of a disability among behaviors assessed by subtests within that channel.

Various aspects of a subject's performance on standardized tests can be studied by part-score procedures. It is a common practice for a diagnostician to group obtained scores, compute a mean for the group of scores, then compare this mean to some other group mean. By way of example, a subject's mean performance on a group of verbal tasks when compared to his mean performance on a group of non-verbal tasks, can be studied for particular strengths or weaknesses in a group of tasks. In the case of che Illinois Test of Psycholinguistic Abilities (Kirk, McCarthy, & Kirk, 1968), part-score procedures have been recommended for analysis of a subject's performance by channel, level, or process (Kirk, 1974, p. 95). It has been held that part-score analyses provide important information for the determination of psycholinguistic abilities or disabilities (Lombardi & Lombardi, 1977, pp. 22-82), subsequent diagnosis (Bateman, 1968, pp. 22-45), and remediation (Kirk & Kirk, 1971, pp. 77-185; 1978). Further, part-score analyses seem to hold promise for the measurement of the effects of remediation of selected psycholinguistic disabilities. Two communication channels are presented in the construct of this test, an auditory-verbal (auditory) channel and a visual-motor (visual) channel. Five subtests comprise each channel. By computing the mean of the five auditory subtests and the mean of the five visual subtests and subtracting the smaller from the larger mean, an interchannel difference can be determined. From this simple part-score procedure, a diagnostician is able to analyze deficits, which, in the aggregate, constitute a problem in one of these two communication channels. When the difference between these means is large, the diagnos'Department of Communicative Disorders, University of Mississippi, University, Mississippi 38677.

700

T. J. WENTLAND

tician would conclude that a disability is present in the channel having the smaller mean. Of course, the diagnostician would conclude that a disability is not present when the difference between these means is small. Although the clinical literature refers often to channel disabilities and what such disabilities indicate (Kirk & Kirk, 1971, pp. 79-85; Lombardi & Lombardi, 1977, pp. 37-42), this same literature or other primary references relating to the test (Paraskevopoulos & Kirk, 1969; Kirk & Kirk, 1978) do not provide the necessary statistical references for the evaluation of regular or irregular patterns of interchannel differences. In the absence of such reference points for determining the significance of such differences, the entire notion of a channel disability depends on clinical intuition of questionable reliability, a position clearly presented by Sattler (1974, pp. 198-204). The necessary statistical reference points include the standard error of measurement (SE,,,) for each of the channels of communication and the standard error of measurement difference ( S E m d ) . It is the purpose of this report to provide these statistical reference points. The data were those from the normative sample of the Revised Illinois Test of Psycholinguistic Abilities. Extensive description of the subjects in this normative sample and test administration and related procedures are available in Paraskevopoulos and Kirk (1969). By way of summary, the normative sample included 962 children ranging in chronological age from 2-7 to 10-1. Subjects have been described as average children drawn from English-speaking families within four communities in the upper-Midwest. Socioeconomic status indicators approximated a national distribution. The sample was equally divided by sex at each age level. Only 4% of subjects were black children.

Statistical Analysis A special Fortran computer program was designed to treat subtest raw scores from the normative sample in a manner equivalent to that reported by Paraskevopoulos and Kirk (1969, pp. 80-88) for transformation of raw scores to scaled scores. From these scaled scores, the following- summary data were computed: ( a ) a grand mean scaled score for each channel, ( b ) a standard deviation of channel mean scaled scores about the grand mean scaled score for each channel, and ( c ) internal consistency coefficients for subtests grouped according to channel. The procedure outlined by Paraskevopoulos and Kirk (1969, p. 101) was followed for this latter data using averaged reliability coefficients corrected for restricted intelligence range across age levels. Subsequent to these computations, hand computed yielded a SE,,, for each channel and the SEmd for channel means. RESULTS Table 1 displays the results of the above statistical analysis. The grand

ITPA INTERCHANNEL DIFFERENCES

TABLE 1 CHANNEL PERFOWNCEOF THE ITPA NORMATIVE SAMPLE Channel Grand M SD r* SE, SEmi Auditory 36.01 3.66 .88 1.24 Visual 36.08 "Internal consisrency coefficient.

3.5 1

.84

1.40

1.87

mean performance of subjects in the normative sample was 36.01 and 36.08 for the auditory and visual channels, respectively. The variability of means about the grand mean of each channel is reflected by standard deviations of 3.66 and 3.51 for the auditory and visual channels, respectively. Average reliability coefficients corrected for restricted intelligence range ( r ) of .88 and .84 for channels were obtained for all ages of subjects. Using the averaged reliability coefficients and the obtained standard deviations, the standard errors of measurement (SE,) of 1.24 and 1.40 were calculated for the auditory and visual channels, respectively. From these values, the standard error of measurement for differences between channel means was computed to be 1.87 scaledscore units.

DISCUSSION The standard error of measurement for channel means, the value of 1.87, provides the test user with an interpretive value not available previously. Using this value as the denominator in a formula where the numerator is the mean difference between channels, the resulting Z statistic can be interpreted with respect to the probability of a true difference between a subject's auditory and visual channel performance. An interchannel difference of 3.09 and 4.36 scaled-score points is required for statistical significance at the 5 and 1% levels of confidence, respectively. As a rule of thumb, a mean difference of 4 scaled-score points would present a difference sufficiently great to conclude that a subject's performance in the channel with the lower mean scaled score is due to real, rather than chance, difference in abilities tested. Although a mean difference of 4 scaled-score points can be considered a consemative cut-off point for the interpretation of atypical-channel difference, some other considerations must be tendered. For example, clinical experience has indicated that some subjects may possess clinically meaningful disabilities in one or the other communication channel in the presence of a mean difference smaller than 4 scaled-score points. T o the contrary, clinically meaningful disabilities may not be present among some subjects who show mean differences greater than 4 scaled-score points. The psychological and educational impact of interchannel differences must, of course, be weighed with respect to a host of subject-related variables, i.e., sociocultural background, compensatory abilities, etc., and performance on other measures that can assist in the delineation of patterns of disabilities.

T. J. WENTLAND REFERENCES BATEMAN, B. Interpretation o f the 1961 Illinois Test o f Psycholinguistic Abilities. Seade: Special Child Publ.. 1968. KIRK, S., & KIRK, W. Psycholinguistic learning disabilities: diagnosis and remediation. Urbana, Ill.: Univer. of Illinois Press, 1971. KIRK,S.. & KIRK, W. Uses and abuses of the ITPA. Journal of Speech and Hearing Disorders, 1978, 43, 58-75. KIRK, S., MCCARTHY, J., & KIRK, W. Examiner's manual: lllinois Test o f Psycholinguistic Abilities. Urbana, Ill.: Univer. of Illinois Press, 1968. KIRK,W. aid^ and precautions i n administering the Illinois Test of Psycholinguistic Abilities. Urbana, Ill.: Univer. of Illinois Press, 1974. LOMBAmI, T., 81 LOMBARDI,E. ITPA: clinical interpretarion and remediation. Seattle: Special Child Publ., 1977. PARASICEVOPOULOS, J., & S. T h e development and psychometric characteristics o f the Revise Zlltnots Test of Psycholinguistic Abilities. Urbana, Ill.: Univer. of Illinois Press, 1969. SATIZBR,J. Assessment o f children's intelligence. Philadelphia: Saunders, 1974.

Accepted March 12, 1979.

ITPA interchannel differences.

Perceprual and Motor Skills, 1979, 48, 699-702. @ Perceptual and Motor Skills 1979 ITPA INTERCHANNEL DIFFERENCES THOMAS J. WENTLAND1 University of...
137KB Sizes 0 Downloads 0 Views