Date of Award

8-1-1995

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Teaching & Learning

Abstract

The purpose of the study was to determine the validity and reliability of a nonverbal learning disability evaluation (NLDE) scale using an instrument intended to screen for nonverbal learning disabilities in classroom settings. Scholars believe that there are at least four distinct subtypes of learning disabilities, each with its own characteristics and interventions. Validity was examined via an ANOVA, discriminant functions analysis, and factor analysis. Reliability was examined via use of Cronbach's alpha ( a = .93).

The sampled populations were special education and regular education teachers in North Dakota and 61 of the students they served. The 43 LD students had identified disabilities in learning and represented grade levels from 3-8. The 18 non-LD children were all enrolled at the elementary school level.

The results of the ANOVA and Tukey's pairwise comparisons on NLDE composite scores indicated that the three group means (NLD, VLD, and Non-LD children) were significantly different from one another. The results of the canonical discriminant functions analysis indicated that there were two functions (factors) required. Can 1, mostly made up of motor behavior, was needed to separate the NLD group from each of the other two. Can2., mostly behaviors learned in a classroom setting in the areas of language and mathematics, was needed to separate both LD groups from the non-LD group. A factor analysis revealed that the instrument was univariate, deriving only one factor. This factor was correlated with over 80% of the scale items. The DISCRIM analysis correctly classified children into their original a priori groups.

Results supported the existence of an NLD subtype, a group of students discriminate from other LD individuals. Further development of the scale is suggested by dropping some items that didn't load on some of the functions or factors and re-evaluating the scale and using a much larger research population. Changes in scale scoring were also suggested.

Download

Share

COinS