Differences in Self‐Concept among Children with Mathematics Disabilities and their Average and High Achieving Peers

Abstract

Self‐concept ratings of children with mathematics disabilities (MD), average mathematics achievement (AA), and high mathematics achievement (HA) who attended regular classes in grades 4 through 6 were compared. Twenty‐four children in each group, who were selected from an original pool of 811 children, and who were matched one‐to‐one by grade, gender, and nonverbal ability participated in the study. The Culture‐Free Self‐Esteem Inventories (Battle, 1992), a nonverbal ability test, and mathematics achievement tests were administered to all children. Results indicated a marginally significant difference in mathematics self‐concept between the MD and AA groups but no difference between the AA and HA groups. However, children with MD rated themselves more negatively than their HA peers on mathematics, academic, and global subscales. Interestingly, the group differences on academic and general self‐concept measures did not emerge when variations in mathematics self‐concept were controlled, limiting the difference between the MD and HA groups to mathematics self‐concept only. The results suggest that by focusing on children with a specific learning disability and by using domain‐specific academic self‐concept measures, the group differences could be shown to be limited to a subject‐specific self‐concept that is pertinent to the children's area of disability.

Notes

Department of Special Needs Education, University of Oslo, Post Box 1140 Blindern, 0318 Oslo, Norway. Email: [email protected]

This study used a curriculum‐based test for each grade. Specifically, the tests were developed in such a way that the items correspond to the curriculum both in content and difficulty. Such tests compare favourably with standardised achievement tests for they could fit closely national or local curriculum objectives.

A table of specifications is a two‐way grid. One dimension contains the contents to be covered by the test whereas the other contains learning objectives. Each cell contains the number of items to be developed from a content area and a corresponding learning objective. After listing the contents and formulating objectives (mostly at the levels of knowledge and comprehension but at the level of application as well), the amount of time a teacher needs to teach each content area was estimated. Each test was then developed in such a way that the number of items constructed from each content area was roughly proportional to the approximate time needed to teach it. Each test contained a sample of items that represented the contents from which they were developed.

The tests were tried out on a pilot sample of 277 children who were randomly selected from all 4th‐, 5th‐, and 6th‐grade classes of one primary school. However, children from this school were not included in the main study. The purpose of the pilot study, among other things, was to obtain data for item analysis. Accordingly, although each test initially included 80 multiple‐choice items (30 mathematical computations, 30 mathematical concepts, and 20 mathematical problem‐solving items), several (16 items) were discarded, several others (15 items) were revised so as to improve their quality, and most (49 items) were retained without revision because of their good qualities in terms of item difficulty and discrimination indices. As a result, each test comprised 64 items (24 computation, 24 mathematical concepts, and 16 word problem items).