The Reduction of Socioeconomic Bias in Intelligence Testing

Abstract

SUMMARY. This study was designed to investigate the extent of socioeconomic bias in selected intelligence tests, with a view to identifying and constructing instruments which might better estimate the intellectual potential of pupils from deprived cultural and educational backgrounds. For such pupils the conventional intelligence test is unsuitable as a predictor of later educational progress if adaptive treatment reduces the effects of their educational handicaps.Nine promising culture-reduced tests and subtests were administered to a representative Edmonton sample of 27 1 twelve and thirteen-year-old children, and the results analysed in relation to measures of socio-economic status, verbal intelligence, achievement and school marks. City-wide testing surveys conducted during the previous four years provided data on the same children, making longitudinal comparisons possible. The results supported the hypothesis that it is possible to measure a broad component of intellectual ability with significantly less cultural bias than is found in the conventional intelligence test. Raven's Progressive Matrices proved the most useful test in the battery since it showed high g loading, consistent and minimal relation with socioeconomic status, no evidence of cultural bias by items, and moderate correlations with school marks. The Cattell test of g and the Lorge-Thorndike Figure Analogies subtest showed similar promise, since they met most of the criteria of the study in shorter testing time.

Sub‐cultural Differences on Selected Cognitive Tasks

“…To return to the issue foreshadowed in the introduction, the present results lend support to the notion that certain standardised ability measures-PPVT in this case-are loaded against particular sub-groups in an English-speaking community. Indeed, the present data concur with the results of an earlier investigation of SES bias with children from the same cultural milieu (MacArthur and Elley, 1963). In that study, lower SES children were relatively less handicapped on ' culture-reduced ' tests (e.g., Raven's Matrices) than on more conventional verbal tests.…”

Ego‐Attitudes in Second Language Learning: an analysis of sex‐related differences

“…(7) Raven's Progressive Matrices: a test of non-verbal intelligence (Raven, 1965) with a reported reliability of 0.83-0.93. This is generally regarded as culture free (McArthur & Elley, 1963;Venables, 1963 andVernon, 1965) and educational content free (Biesheuvels, 1962).…”

Social Class Differentials in Vocabulary Expansion

“…Mean ages of M boys were precisely 10 and 14, those of W boys one month higher. MacArthur and Elley (1963) demonstrated the relative insensitivity of Progressive Matrices, the non-verbal test used, to social status differences : the younger boys were given the Coloured and the older ones the Standard Progressive Matrices. For the vocabulary assessment, the Self-Judging Vocabulary Scale (S. J .-Scale) devised by Heim and Watts (1961) was used, form aA a t age 10 and form A E a t age 14.…”

“…Mean ages of M boys were precisely 10 and 14, those of W boys one month higher. MacArthur and Elley (1963) demonstrated the relative insensitivity of Progressive Matrices, the non-verbal test used, to social status differences : the younger boys were given the Coloured and the older ones the Standard Progressive Matrices.…”