A qualitative bias review of the adaptation of the WAIS-III for English-speaking South Africans
- Authors: Aston, Susan
- Date: 2006
- Subjects: Wechsler Adult Intelligence Test -- South Africa , Intelligence tests -- South Africa
- Language: English
- Type: Thesis , Masters , MA
- Identifier: vital:9939 , http://hdl.handle.net/10948/453 , http://hdl.handle.net/10948/d1011931 , Wechsler Adult Intelligence Test -- South Africa , Intelligence tests -- South Africa
- Description: In response to the growing demand for a test of cognitive ability for South African adults, the Human Sciences Research Council (HSRC) adapted the Wechsler Adult Intelligence Scales, third edition (WAIS-III) for English-speaking South Africans. However, since the publication of the South African adaptation in 2001, there has been some concern whether some bias may not still be present in the measure. Consequently, this study aimed to qualitatively explore the item content of the South African adaptation of the WAIS-III in order to identify items that might still be potentially biased with specific reference to cultural and linguistic issues. The researcher employed purposive sampling, a non-probability technique, to select psychology professionals to act as expert reviewers. The sample consisted of 20 registered psychologists, psychology lecturers and psychology interns from the predominant culture and language groups of the Eastern and Western Cape. Qualitative and quantitative data were collected by means of a Bias Review Questionnaire which consisted of fixed-choice questions, as well as sections in which specific comments could be indicated. The questionnaire was distributed to 20 selected participants who were qualified to administer psychometric measures and were registered with the Health Professionals Council of South Africa (HPCSA). Qualitative data was analysed using Tesch’s model of thematic content analysis, and five main themes relating to potential bias were identified, namely: Culture, Language, Education, Socio-economic status, and Format. The qualitative data was re-submitted to the expert review group for approval and further comment. Quantitative data obtained from the Bias Review Questionnaire was analysed by obtaining frequency counts, which were converted to percentages, per subtest and aspect reviewed. The major findings of the study were as follows: 1. Cultural bias is evident in the South African adaptation of the WAIS-III. The differential meanings ascribed to concepts, terms, phrases and words by diverse cultures impact on test performance, as does familiarity with stimulus materials used in the measure. 2. Language bias was detected against groups with English as their second or third language owing to the unfamiliar, and sometimes archaic, terms and vocabulary used in the measure. The intended meanings of the translated items are unclear and confusing for test-takers. 3. South African test-takers have experienced vastly divergent educational opportunities which impact on intelligence test performance. The measure contains potential bias against individuals from disadvantaged educational backgrounds. 4. The content of the items is not relevant to test-takers from deprived socioeconomic backgrounds that have been unable to access good-quality education. 5. Format bias was found to exist with respect to the timed tests. Cultural groups who do not value speededness are disadvantaged by the structure and format of the measure. Based on the findings of this study, certain recommendations were made. These include: 1. Additional qualitative investigation of the WAIS-III , with a view to further adaptation in an effort to eradicate bias on cultural or linguistic grounds, and 2. Conducting further Differential Item Functioning (DIF) analyses and establishing construct equivalence across cultural and linguistic groups to promote this objective. Among the contributions believed to have been made by this study are the identification of potentially biased subtests and items on the grounds of culture and language, which will facilitate possible further revision and adaptation of the measure. Additionally, the variables that were considered important influencing factors of cognitive test performance have been documented and are accessible for future research that will supplement the present study.
- Full Text:
- Date Issued: 2006
Cross-cultural differences in IQ test performance: extension of an existing normative database on WAIS-III test performance
- Authors: Gaylard, Emma K
- Date: 2006
- Subjects: Wechsler Adult Intelligence Scale , Wechsler Adult Intelligence Scale -- South Africa , Black people -- Education , Black people -- Social conditions , Intelligence tests -- South Africa
- Language: English
- Type: Thesis , Masters , MA
- Identifier: vital:2979 , http://hdl.handle.net/10962/d1002488 , Wechsler Adult Intelligence Scale , Wechsler Adult Intelligence Scale -- South Africa , Black people -- Education , Black people -- Social conditions , Intelligence tests -- South Africa
- Description: Prior research (Shuttleworth-Edwards et al., 2004) presented preliminary normative data for the Wechsler Adult Intelligence Scale – III (WAIS-III) for Southern Africa, stratified according to gender (female vs. male), language (black African vs. white English), level of education (matric/12+ years of education vs. Graduate/15+ years of education) and quality of education (disadvantaged – Department of Education and Training vs. advantaged - Private/Model C). IQ scores for black African language and white English Southern Africans were comparable with the United States of America (USA) standardization when level and quality of education were equitable. (‘White English’ is the term used to denote those of European descent whose first language is English). A limitation of the research was the lack of control for language for most of the black groups and particularly in the Private/Model C Graduate group, where sixty percent of the participants originated from Zimbabwe. These represented a particularly elite group whose education was equitable to that of the white participants throughout their education (i.e. at primary, secondary and tertiary level). In order to rectify the lack of homogeneity of language, all non- Xhosa first language participants were excluded from the black sample and sixteen additional Xhosa first language participants were tested on the WAIS-III. Data analyses found no significant differences between the original and new groups, except in the comparison between Mixed African language Private/Model C Graduates and the Xhosa first language Private/Model C Graduate/15+ years of education, where there was a lowering of WAIS-III subtest, index and IQ scores in the newly constituted group. This lowering in test performance is explained in that the new Xhosa first language 15+ years of education group was a less advantaged group than the original Mixed African Language Private/Model C Graduate group, as the new group generally had less advantaged primary school education and had generally studied less at a tertiary level. Overall, these results demonstrate an incremental increase in WAIS-III test performance for sample groups on a continuum of quality of education from least to most advantaged education. This was true for both verbal and non-verbal subtests.
- Full Text:
- Date Issued: 2006
The effects of level and quality of education on a South African sample of English and African first language speakers, for WAIS-III digit symbol-incidental learning
- Authors: Donnelly, Martin Joseph Rhodes
- Date: 2001
- Subjects: Wechsler Adult Intelligence Scale , Intelligence tests -- South Africa , Wechsler Adult Intelligence Scale -- South Africa
- Language: English
- Type: Thesis , Masters , MA
- Identifier: vital:2965 , http://hdl.handle.net/10962/d1002474 , Wechsler Adult Intelligence Scale , Intelligence tests -- South Africa , Wechsler Adult Intelligence Scale -- South Africa
- Description: This study examined the effects of level and quality of education on WAIS-III Digit Symbol-Incidental Learning performance. The Pairing and Free Recall measures were administered to a South African sample (N = 68, age range 19-30), which was stratified for English and African language, level of education attained (Grade 12s and Graduates) and quality of education (advantaged and disadvantaged schooling). Results yielded no significant main or interaction effects between acculturation factors of level and quality of education. Normative guidelines of 13 or more pairs and 8 or more free recall symbols, appropriate to a non-clinical sample in a multicultural setting, are provided. Digit Symbol-Incidental Learning proved to be a culture-fair test, which contributes to its clinical utility as a sensitive memory screening tool.
- Full Text:
- Date Issued: 2001
An investigation into the effects of socio-economic and education factors on WAIS-III performance in a stratified South African sample
- Authors: Kemp, Ryan
- Date: 2000
- Subjects: Wechsler Adult Intelligence Scale -- South Africa , Intelligence tests -- South Africa
- Language: English
- Type: Thesis , Masters , MA
- Identifier: vital:3002 , http://hdl.handle.net/10962/d1002511 , Wechsler Adult Intelligence Scale -- South Africa , Intelligence tests -- South Africa
- Description: The present study examined the effects of socio-economic status, quality and level of education on performance on the Wechsler Adult Intelligence Scale 3rd Edition r.:w AIS-III). The study was conducted within the context of the present Human Sciences Research Council (HSRC) W AIS-III standardisation process, cross-cultural psychometric research and the notion of test-wiseness. A South African sample (N = 68) comprising African first language (n = 40) and English first language (n = 28) participants, stratified for age (19 - 30 years), gender and educational attainment (Matric and Graduate) were utilised. Effects due to quality of education were determined by dividing the African language participants into those with private/model C schooling (n = 20) and those with DET schooling (n = 20). Effects due to level of education were determined by dividing the entire sample into those with Matric level education (n = 34) and those with Graduate level education (n = 34). Detailed demographic and socio-economic information were recorded for all participants, who completed the entire American version of the W AIS-III. Results revealed a highly significant positive correlation between high socio-economic status and W AIS-III Full Scale IQ. In addition the respective impact of the relatively poor quality of education and lower level of educational attainment on W AIS-III performance was substantial. These results were discussed in terms of their implications for cross-cultural research, the HSRC standardisation of the W AIS-III and the practical use of the W AIS-III in neuropsychological assessment.
- Full Text:
- Date Issued: 2000
An investigation of male and female cognitive ability on the WAIS-III
- Authors: Muirhead, Joanne
- Date: 2000
- Subjects: Wechsler Adult Intelligence Scale , Sex differences (Psychology) , Intelligence tests -- South Africa
- Language: English
- Type: Thesis , Masters , MA
- Identifier: vital:3028 , http://hdl.handle.net/10962/d1002537 , Wechsler Adult Intelligence Scale , Sex differences (Psychology) , Intelligence tests -- South Africa
- Description: This study, which formed part of a larger research project, investigated the effect of gender on test performance on the Wechsler Adult Intelligence Scale - Third Edition (WAIS-III). The WAIS-III was administered to a sample of 68 participants in the Eastern Cape following the initiative of the Human Sciences Research Council to standardise the WAIS-III for a South African population. The participants, aged 19 to 30, were stratified according to language of origin (African or English First Language), educational attainment (matriculant or graduate), quality of education (Department of Education and Training or private/"Model C" school) and gender. Analyses of variance and two sample t tests were used to compare male and female test performance. For the total sample, no significant difference between males and females on Verbal, Performance and Full Scale IQ were found. On the factor indices, females scored marginally higher than males on Processing Speed at a level which was approaching significance (p = 0.105), but no significant differences were found. On subtest performance, females significantly outperformed males on Digit Symbol (p = 0.020). Differences which were approaching significance were found on Information (p = 0.133) in favour of males, and on Matrix Reasoning (p = 0.092) in favour of females. For subgroups of the total sample, the most significant differences in test performance were found for the African First Language private/"Model C" school cohort in favour of females. Thus the overriding implication that emerged from this research was that on this relatively highly educated sample, no significant gender differences in cognitive ability were apparent.
- Full Text:
- Date Issued: 2000
The development of a WAIS-III short form for use in South Africa
- Authors: Rust, Annegret L
- Date: 2000
- Subjects: Wechsler Adult Intelligence Scale , Intelligence tests -- South Africa , Wechsler Adult Intelligence Scale -- South Africa
- Language: English
- Type: Thesis , Masters , MSocSc
- Identifier: vital:3047 , http://hdl.handle.net/10962/d1002556 , Wechsler Adult Intelligence Scale , Intelligence tests -- South Africa , Wechsler Adult Intelligence Scale -- South Africa
- Description: The Wechsler Adult Intelligence Scale - 3rd Edition (WAIS-III) is the newest of the internationally recognised Wechsler family of intelligence tests. It has been improved in terms of its psychometric properties, neuropsychological assessment abilities and its content. This test is in the process of being standardised by the Human Sciences Research Council (HSRC) in South Africa. As the adapted South African version will be available shortly for use in the multicultural circumstances of South Africa, the application of the various aspects of this test needs to be investigated. This test is very comprehensive and thorough, however its one disadvantage is that it takes on average three hours to administer in its entirety. Thus there is a need to find ways in which to abbreviate the test for particular purposes when time is limited, for example in research, brief clinical assessments or neuropsychological screenings. The concept of abbreviating tests, including the earlier Wechsler intelligence test can be traced back to 1917, when it was asked if all the items on the Binet-Simon scale were required to give an accurate assessment of IQ (Levy, 1968). Since then there have been many short form suggestions made, with many different considerations in mind. These can be divided into two main approaches or methods. Firstly, the number of subtests of the scale can be reduced. Thus with the WAIS-III which consists of 14 subtests in total, an option is to use, for example only four of the subtests to get an estimate of a person's IQ. Secondly, the number of items in each subtest can be reduced. Thus only half the items or even only a third of the items on a subtest can be administered to get an estimate of the persons' performance on each subtest and in this way estimate their overall IQ. Both methods have been used on the WAIS and WAIS-R, although the reduction of the subtests is favoured. Both should now be validated and considered for use with the WATS-III in South Africa. Wechsler tests and their constituent subtests have been found to be differentially effected by race, education, language and socio-economic status (Kaufman, McLean & Reynolds, 1988; Nell 1999). These differences have also been found to impact on the short forms which are suggested, as certain subtests are considered to be more biased towards particular groups than others. Vocabulary and Block Design in particular bias testees who are not as westernised or acculturated towards a largely American and European culture (Kaufman, McLean & Reynolds, 1988). These differences, although often ascribed to race, language or socio-economic status can best be understood more broadly in terms of degree of acculturation (i.e. westernisation) (Shuttleworth-Jordan, 1996) and test-wiseness (Nell, 1999). In South Africa in particular, with its extreme cultural diversity these factors need to be carefully considered when developing short forms. In the present study the development of a short form appropriate to South Africa's diverse cultural circumstances will be approached, through a sample which has been stratified according to gender, first language (English vs. African), quality of schooling received (Private/Model C vs. DET) and level of education achieved (Matric vs. Graduate). Both a subtest reduction method and an item reduction method will be considered to arrive at a short form. The subtest reduction method will be considered further in an attempt to clarify which subtests would be more or less appropriate to include in a short form considering group differences. Finally the thesis will develop suggestions as to which short forms would be best for use in South Africa.
- Full Text:
- Date Issued: 2000
Standardization of Raven's standard progressive matrices for secondary school African pupils in the Grahamstown region
- Authors: Vass, Vasili Arthur
- Date: 1992
- Subjects: Black people -- Education -- South Africa , Intelligence tests -- South Africa , Educational tests and measurements -- South Africa , Cognition and culture , Raven's Progressive Matrices
- Language: English
- Type: Thesis , Masters , MA
- Identifier: vital:3078 , http://hdl.handle.net/10962/d1002587 , Black people -- Education -- South Africa , Intelligence tests -- South Africa , Educational tests and measurements -- South Africa , Cognition and culture , Raven's Progressive Matrices
- Description: Arising out of a need, expressed by Clinical Psychologists in the Grahamstown region, for the fair assessment of secondary School African pupils, norms for the Raven's Standard Progressive Matrices (SPM) were established. Two methods of presentation were used, the first using the original instructions of John Raven translated into the students' first language of Xhosa, hence-forth referred to as the Alternate sample. The second method of presenting the instructions was adapted, to include the use of visual aids and active participation in the instruction phase of the test, hence-forth referred to as the Normative sample. In addition to the establishment of norms, the two methods of test presentation were investigated to see if the method of presentation had an effect on the results of the Raven's SPM. The population was drawn from the three African Secondary Schools in the Grahamstown municipality. The total population consisted of 3 232 students. Classes were randomly sampled across the three schools, with the average age of the Normative sample being 19.3 years. The sample consisted of 812 pupils, 711 in the Normative sample and 101 in the Alternate Sample. The following results and conclusions arose from the study: 1) Norms generated were considerably lower than previously established norms in similar studies. 2) t-Test results indicated that the method of test presentation on the Raven's SPM is important when assessing students that may be regarded as 'disadvantaged'. 3) Respondents scored significantly higher when the method of presenting the instructions ensured a greater understanding of the task demanded of the respondents. 4) The analyses of covariance indicate that male subjects score significantly higher than female subjects, and that there is a significant difference between the ages and educational standard, on the scores of the Raven's SPM. 5) The differences found contradict previous findings using the test, and question the cultural fairness of the test.
- Full Text:
- Date Issued: 1992