Abstract

The Spanish version of the third edition of the Wechsler Adult Intelligence Scale (WAIS-III) by TEA Ediciones is an excellent addition to available instruments for Spanish speakers. The Spanish norms function similarly to US norms for individuals aged 16–35. The norms become increasingly different for individuals 35 and older, seemingly because of the lower levels of formal education of the older Spanish cohorts. Using data from a random half of the Spanish sample, the authors developed regression equations to adjust the scaled scores for individuals with a low level of education. The adjustment is made to the level that would have been expected if the individual had 12 years of education, the median level of education of the US norms. The article includes the methodology and values necessary to make the adjustments. The scaled scores were then adjusted for individuals on the second random half of the Spanish sample and compared to the United States norms. The results showed the adjustments succeed in bringing the Spanish norms closer to the US norms.

Introduction

Until recently, there was no standardized Spanish-language version comparable with the third edition of the Wechsler adult intelligence scale (WAIS-III; The Psychological Corporation, 1997), the most widely used intelligence test for adults in the USA and many parts of the world. Although the Puerto Rican version of the Escala de Inteligencia Wechsler para Adultos (EIWA; The Psychological Corporation, 1968), a Spanish-language version of the first edition of the WAIS (Wechsler, 1955) had been available, it inflated the full-scale intelligence quotient (IQ) by 20 points in comparison with the U.S. version of the test (Melendez, 1994). Moreover, as the IQs of people in industrialized countries rise significantly over time (for review seeFlynn, 2007, 2008), it is estimated that intelligence tests become outdated in a decade.

TEA Ediciones, a publisher of psychological tests in Spain, has developed a Spanish-language version of the WAIS-III. This version has been shown to have adequate psychometrics with a U.S. sample of Hispanics (Renteria, Li, & Pliskin, 2008). The Spanish norms for this version (TEA Ediciones, 2001) are composed of average scores that are similar to the U.S. norms of the WAIS-III up to age 35. As can be seen in Fig. 1, the two norms differ for the older cohorts, with the Spanish WAIS-III norms showing increasingly lower average scores. The lower means lead to progressively higher standard scores, when compared with the U.S. norms, producing the same inflationary effect from which the original EIWA suffered in the USA. The reason for the increased discrepancy of these norms appeared to be the decreasing level of education of the Spanish older cohorts (Krueger, de la Torre, Fuentes, & Choca, 2005), a phenomenon that may have been affected by historical events. Specifically, those older cohorts were growing up during the Spanish Civil War and/or the Franco dictatorship when the access to education was limited.

Fig. 1.

Mean raw scores for the USA and Spanish versions of the third edition of the Wechsler adult intelligence scale.

Fig. 1.

Mean raw scores for the USA and Spanish versions of the third edition of the Wechsler adult intelligence scale.

The main factor contributing to cultural differences in cognitive ability testing has been shown to be the degree of formal education (Ardila, Roselli, & Rosas, 1989; Boone, Victor, Wen, Razani, & Pontón 2007; Crawford-Nutt, 1977; da Silva, Petersson, Faísca, Ingvar, & Reiss, 2004; Greenfield, 1997; Hsieh & Tori, 2007; Kendall, Verster, & von Mollendorf, 1988; Mathuranath et al., 2003; Nell, Myers, Colvin, & Rees, 1993; Ostrosky-Solís, López, & Ardila, 2000; Ostrowsky-Solis, Lozano Gutierrez, Ramirez Flores, & Ardila, 2007; Ratcliff et al., 1998). Although there is typically no difference in test scores for children of different cultures examined before the age of five (Verster & Prinsloo, 1988), the negative impact of illiteracy and poor reading ability on most intellectual abilities has been repeatedly demonstrated (Ardila, 2000; Ardila, Ostrowsky, & Mendoza, 2000; da Silva et al., 2004; Folia & Kosmidis, 2003; Gonzalez da Silva, Petersson, Faisca, Ingvar, & Reiss, 2004; Manly et al., 1999; Matute, Leal, Zarabozo, Robles, & Cedillo, 2000; Ostrosky, Ardila, Rosselli, Lopez-Arango, & Uriel-Mendoza, 1998; Ratcliff et al., 1998; Reis & Castro-Caldas, 1997; Reis, Guerreiro, & Petersson, 2003). The negative impact of low education, in turn, has been shown to lead to overestimates of the degree of impairment with Hispanics, even when non-verbal instruments are used (Cherner et al., 2007).

Similarly, the impact of education on cognitive abilities scores of older persons has been well-documented, although not uniformly across abilities. The Mayo's Older American Normative Studies, for instance, showed that education had a greater association with verbal measures of remote memory than with measures of novel visuo-spatial problem-solving; this study suggested that the higher IQs were due to higher exposure to information in school (Malec, Ivnik, & Smith, 1992). The Framingham Heart Study examined 1,893 older persons and showed increasing age to be associated with lower levels of education, and fewer years of education to be associated with lower scores within the same age groups (Elias, Elias, & D'Agostino, 1997).

The effect of education on WAIS-III scores can be the source of error with the English-speaking population. Heaton and his associates note that the percentile of individuals who score 1 SD below the mean on the verbal comprehension index goes from approximately 3% for the college educated group to 45% for individuals with 8 years or less of schooling. The diagnostic problem can affect both ends of the continuum: Not only are people with a low level of education more likely to be misclassified as showing cognitive decline, but highly educated individuals may be erroneously seen as normal when they are in fact showing cognitive decay (Heaton, Taylor, & Manly, 2003). Similar concerns have been expressed by researchers working with English-speaking individuals in Africa (Shuttleworth-Edwards et al., 2004).

The educational background of Hispanics in the USA varies widely, with the majority attaining a lower level of education compared with African Americans and non-Hispanic Whites. Current census data show that 24% of Hispanics in the USA have fewer than 9 years of education and 41% do not have a high-school diploma (U.S. Census Bureau, 2006). Given the discrepancy in formal levels of education, it would be useful to have a way of adjusting the Spanish WAIS-III scaled and standard scores to take into account the level of education. The scores from the test manual would provide comparisons with same aged peers, whereas the adjusted score would provide gender and education adjusted comparisons. If the global indices were then computed with the adjusted scaled scores, those standard scores would also be adjusted. The goal of this study is to provide a way of adjusting the norms of the Spanish WAIS-III to take into account demographic variables that would otherwise adversely affect the scores obtained. The test, and the adjustments proposed in this paper, could be useful in evaluation of Spanish-speaking individuals in the USA as well as in Spanish-speaking countries.

Materials and Methods

The standardization data for the Spanish WAIS-III were obtained from TEA Ediciones. The sample is representative of adults in Spain aged 16–94 years. The sampling plan was based on data gathered in 1991 and 1996 by Spanish Bureau of the Census and stratified along age, gender, educational level, type of residential area, and geographic region. Individuals with known neurological or psychiatric problems were excluded from the sample. Six age groups were established (16–19, 20–24, 25–34, 35–54, 55–69 and ≥70) with a minimum of 136 participants on older groups, and a maximum of 435 participants in the 35–54 age group. The residential setting was coded with three levels according to population density: Urban, intermediate, and rural. For the geographic region, Spain was divided into four major regions (North, South, East, and Central), and the number of participants recruited from each of those areas was proportional to the population of that area. A total of 1,369 Spanish adults were tested.

The Spanish sample was coded with the following educational-level categories: (1) did not complete primary school or <8 years of education, (2) primary school was the highest degree obtained, (3) secondary school was the highest degree obtained, and (4) a university degree was obtained. Unfortunately, only this ordinal data were available. In order to create a continuous variable for analyses, we used the following values. For the first category, we used 4 as an estimate of the mean of the group having studied between 0 and 8 years of school, but never having finished primary school. For categories 2–4, we used the minimal number of years a person would have obtained to be considered in that category. Category 2, primary school, was coded as 8 years. Category 3, secondary school, was coded as 11 years. Category 4, university degree, was coded as 16 years. Table 1 provides detailed information for the standardization sample by age and educational levels.

Table 1.

Number of participants with the different levels of education at the different age-brackets

Age band Educational level categories
 
 Total 
16–19 39 123 163 
20–24 29 105 14 153 
25–34 11 72 145 44 272 
35–54 85 164 118 41 408 
55–69 111 88 27 11 237 
≥70 88 40 136 
Total 301 432 525 111 1,369 
Age band Educational level categories
 
 Total 
16–19 39 123 163 
20–24 29 105 14 153 
25–34 11 72 145 44 272 
35–54 85 164 118 41 408 
55–69 111 88 27 11 237 
≥70 88 40 136 
Total 301 432 525 111 1,369 

The total sample was randomly divided into two halves. Regression equations, taking into account age, gender, and years of education, were developed using the raw scores from the first half of the sample. The target educational level used to develop these equations was 12 years. The mean level of education for the U.S. norms was approximately 12 years and, if the difference between the two norms were due to the lower level of education of the older Spanish cohorts, the target of 12 years should adjust the Spanish norms to the same level as the U.S. norms. A separate regression equation was developed for each of the subtests. The second half was used to validate the regression equations derived from the first half. Tables 2 and 3 offer the mean and standard deviation for both the derivation and the validation samples. A comparison of these two halves on all of the variables used in this study did not show any significant differences.

Table 2.

Descriptive statistics for the different subtests of the derivation sample

Subtest Level of education N Mean Standard deviation 
Arithmetic 4.00 166 8.95 2.619 
 8.00 204 11.35 3.270 
 11.00 259 14.29 3.413 
 16.00 55 15.58 3.287 
 Total 684 12.22 3.931 
Symbol search 4.00 166 17.32 9.403 
 8.00 203 26.11 11.094 
 11.00 258 34.96 7.656 
 16.00 55 35.65 8.845 
 Total 682 28.09 11.756 
Block design 4.00 166 13.15 5.876 
 8.00 204 17.00 5.250 
 11.00 259 20.99 4.896 
 16.00 55 22.82 4.695 
 Total 684 18.05 6.213 
Comprehension 4.00 166 23.22 10.984 
 8.00 204 34.77 13.736 
 11.00 258 45.59 10.791 
 16.00 54 46.80 11.409 
 Total 682 37.01 14.931 
Coding 4.00 165 38.74 24.211 
 8.00 203 57.79 24.016 
 11.00 259 76.77 17.549 
 16.00 55 79.53 18.758 
 Total 682 62.14 26.476 
Digit span 4.00 166 11.36 3.696 
 8.00 204 13.94 4.176 
 11.00 259 17.03 3.920 
 16.00 55 17.09 4.001 
 Total 684 14.74 4.578 
Picture completion 4.00 166 12.83 5.868 
 8.00 203 17.35 4.682 
 11.00 259 20.76 2.551 
 16.00 55 20.96 3.600 
 Total 683 17.84 5.342 
Picture arrangement 4.00 165 7.16 5.187 
 8.00 204 11.38 5.484 
 11.00 259 15.47 4.178 
 16.00 55 16.29 4.054 
 Total 683 12.31 5.923 
Information 4.00 166 10.77 4.906 
 8.00 204 14.50 4.759 
 11.00 259 19.80 4.445 
 16.00 55 21.60 3.828 
 Total 684 16.17 6.038 
Letter number 4.00 166 6.21 2.898 
 8.00 204 8.79 3.215 
 11.00 259 11.36 2.666 
 16.00 55 11.62 2.896 
 Total 684 9.36 3.588 
Matrix reasoning 4.00 166 9.05 4.805 
 8.00 204 14.52 6.065 
 11.00 259 19.37 4.014 
 16.00 55 19.93 4.294 
 Total 684 15.46 6.458 
Object assembly 4.00 166 19.95 9.250 
 8.00 204 28.95 9.783 
 11.00 259 34.61 7.814 
 16.00 55 34.02 8.801 
 Total 684 29.31 10.587 
Similarities 4.00 166 11.24 5.233 
 8.00 204 15.93 5.676 
 11.00 259 20.97 5.403 
 16.00 55 22.44 5.284 
 Total 684 17.22 6.793 
Vocabulary 4.00 166 23.46 12.987 
 8.00 204 35.37 10.953 
 11.00 259 45.28 8.012 
 16.00 55 48.71 8.157 
 Total 684 37.31 13.720 
Subtest Level of education N Mean Standard deviation 
Arithmetic 4.00 166 8.95 2.619 
 8.00 204 11.35 3.270 
 11.00 259 14.29 3.413 
 16.00 55 15.58 3.287 
 Total 684 12.22 3.931 
Symbol search 4.00 166 17.32 9.403 
 8.00 203 26.11 11.094 
 11.00 258 34.96 7.656 
 16.00 55 35.65 8.845 
 Total 682 28.09 11.756 
Block design 4.00 166 13.15 5.876 
 8.00 204 17.00 5.250 
 11.00 259 20.99 4.896 
 16.00 55 22.82 4.695 
 Total 684 18.05 6.213 
Comprehension 4.00 166 23.22 10.984 
 8.00 204 34.77 13.736 
 11.00 258 45.59 10.791 
 16.00 54 46.80 11.409 
 Total 682 37.01 14.931 
Coding 4.00 165 38.74 24.211 
 8.00 203 57.79 24.016 
 11.00 259 76.77 17.549 
 16.00 55 79.53 18.758 
 Total 682 62.14 26.476 
Digit span 4.00 166 11.36 3.696 
 8.00 204 13.94 4.176 
 11.00 259 17.03 3.920 
 16.00 55 17.09 4.001 
 Total 684 14.74 4.578 
Picture completion 4.00 166 12.83 5.868 
 8.00 203 17.35 4.682 
 11.00 259 20.76 2.551 
 16.00 55 20.96 3.600 
 Total 683 17.84 5.342 
Picture arrangement 4.00 165 7.16 5.187 
 8.00 204 11.38 5.484 
 11.00 259 15.47 4.178 
 16.00 55 16.29 4.054 
 Total 683 12.31 5.923 
Information 4.00 166 10.77 4.906 
 8.00 204 14.50 4.759 
 11.00 259 19.80 4.445 
 16.00 55 21.60 3.828 
 Total 684 16.17 6.038 
Letter number 4.00 166 6.21 2.898 
 8.00 204 8.79 3.215 
 11.00 259 11.36 2.666 
 16.00 55 11.62 2.896 
 Total 684 9.36 3.588 
Matrix reasoning 4.00 166 9.05 4.805 
 8.00 204 14.52 6.065 
 11.00 259 19.37 4.014 
 16.00 55 19.93 4.294 
 Total 684 15.46 6.458 
Object assembly 4.00 166 19.95 9.250 
 8.00 204 28.95 9.783 
 11.00 259 34.61 7.814 
 16.00 55 34.02 8.801 
 Total 684 29.31 10.587 
Similarities 4.00 166 11.24 5.233 
 8.00 204 15.93 5.676 
 11.00 259 20.97 5.403 
 16.00 55 22.44 5.284 
 Total 684 17.22 6.793 
Vocabulary 4.00 166 23.46 12.987 
 8.00 204 35.37 10.953 
 11.00 259 45.28 8.012 
 16.00 55 48.71 8.157 
 Total 684 37.31 13.720 
Table 3.

Descriptive statistics for the different subtests of the validation sample

Subtest Level of education N Mean Standard deviation 
Arithmetic 4.00 130 9.08 2.653 
 8.00 223 11.45 3.328 
 11.00 266 14.08 3.463 
 16.00 56 15.09 4.179 
 Total 675 12.33 3.906 
Symbol search 4.00 129 17.94 10.718 
 8.00 223 27.66 12.328 
 11.00 265 34.43 9.007 
 16.00 56 35.88 7.872 
 Total 673 29.15 12.192 
Block design 4.00 130 13.62 5.723 
 8.00 224 16.63 5.399 
 11.00 266 20.61 4.995 
 16.00 56 23.05 4.338 
 Total 676 18.15 6.026 
Comprehension 4.00 129 25.28 13.368 
 8.00 223 36.14 13.691 
 11.00 266 45.36 10.556 
 16.00 55 44.95 11.562 
 Total 673 38.42 14.424 
Coding 4.00 127 38.80 23.749 
 8.00 223 59.17 24.839 
 11.00 266 74.25 18.978 
 16.00 56 77.84 18.726 
 Total 672 62.84 25.792 
Digit span 4.00 130 11.28 3.564 
 8.00 224 13.83 3.915 
 11.00 266 16.83 4.346 
 16.00 56 18.43 4.339 
 Total 676 14.90 4.674 
Picture completion 4.00 129 12.67 5.510 
 8.00 224 17.21 4.975 
 11.00 266 20.17 2.897 
 16.00 56 21.14 2.882 
 Total 675 17.84 5.123 
Picture arrangement 4.00 130 7.12 5.326 
 8.00 223 11.67 5.529 
 11.00 266 14.97 4.325 
 16.00 56 16.00 4.306 
 Total 675 12.45 5.798 
Information 4.00 130 10.02 4.403 
 8.00 224 14.15 5.490 
 11.00 266 19.25 4.168 
 16.00 56 22.00 3.521 
 Total 676 16.01 6.054 
Letter number 4.00 130 6.05 2.861 
 8.00 222 8.95 3.074 
 11.00 266 11.26 2.737 
 16.00 56 12.09 2.459 
 Total 674 9.56 3.506 
Matrix reasoning 4.00 130 8.97 5.140 
 8.00 224 14.33 5.788 
 11.00 266 18.87 4.292 
 16.00 56 19.43 4.480 
 Total 676 15.51 6.277 
Object assembly 4.00 130 22.49 10.456 
 8.00 222 28.05 10.295 
 11.00 265 34.69 8.300 
 16.00 56 33.79 7.670 
 Total 673 30.07 10.476 
Similarities 4.00 130 10.67 5.392 
 8.00 224 14.98 6.026 
 11.00 266 20.45 4.926 
 16.00 56 22.09 5.334 
 Total 676 16.89 6.730 
Vocabulary 4.00 129 23.40 12.397 
 8.00 224 33.75 12.365 
 11.00 265 44.27 8.310 
 16.00 56 50.02 9.156 
 Total 674 37.26 13.697 
Subtest Level of education N Mean Standard deviation 
Arithmetic 4.00 130 9.08 2.653 
 8.00 223 11.45 3.328 
 11.00 266 14.08 3.463 
 16.00 56 15.09 4.179 
 Total 675 12.33 3.906 
Symbol search 4.00 129 17.94 10.718 
 8.00 223 27.66 12.328 
 11.00 265 34.43 9.007 
 16.00 56 35.88 7.872 
 Total 673 29.15 12.192 
Block design 4.00 130 13.62 5.723 
 8.00 224 16.63 5.399 
 11.00 266 20.61 4.995 
 16.00 56 23.05 4.338 
 Total 676 18.15 6.026 
Comprehension 4.00 129 25.28 13.368 
 8.00 223 36.14 13.691 
 11.00 266 45.36 10.556 
 16.00 55 44.95 11.562 
 Total 673 38.42 14.424 
Coding 4.00 127 38.80 23.749 
 8.00 223 59.17 24.839 
 11.00 266 74.25 18.978 
 16.00 56 77.84 18.726 
 Total 672 62.84 25.792 
Digit span 4.00 130 11.28 3.564 
 8.00 224 13.83 3.915 
 11.00 266 16.83 4.346 
 16.00 56 18.43 4.339 
 Total 676 14.90 4.674 
Picture completion 4.00 129 12.67 5.510 
 8.00 224 17.21 4.975 
 11.00 266 20.17 2.897 
 16.00 56 21.14 2.882 
 Total 675 17.84 5.123 
Picture arrangement 4.00 130 7.12 5.326 
 8.00 223 11.67 5.529 
 11.00 266 14.97 4.325 
 16.00 56 16.00 4.306 
 Total 675 12.45 5.798 
Information 4.00 130 10.02 4.403 
 8.00 224 14.15 5.490 
 11.00 266 19.25 4.168 
 16.00 56 22.00 3.521 
 Total 676 16.01 6.054 
Letter number 4.00 130 6.05 2.861 
 8.00 222 8.95 3.074 
 11.00 266 11.26 2.737 
 16.00 56 12.09 2.459 
 Total 674 9.56 3.506 
Matrix reasoning 4.00 130 8.97 5.140 
 8.00 224 14.33 5.788 
 11.00 266 18.87 4.292 
 16.00 56 19.43 4.480 
 Total 676 15.51 6.277 
Object assembly 4.00 130 22.49 10.456 
 8.00 222 28.05 10.295 
 11.00 265 34.69 8.300 
 16.00 56 33.79 7.670 
 Total 673 30.07 10.476 
Similarities 4.00 130 10.67 5.392 
 8.00 224 14.98 6.026 
 11.00 266 20.45 4.926 
 16.00 56 22.09 5.334 
 Total 676 16.89 6.730 
Vocabulary 4.00 129 23.40 12.397 
 8.00 224 33.75 12.365 
 11.00 265 44.27 8.310 
 16.00 56 50.02 9.156 
 Total 674 37.26 13.697 

Fractional polynomials were used in order to explore whether the iterative algorithms comparing all sets of predictors would improve the final optimal fit. In the case of most subtests, the fit was identical as to what had been obtained with the linear regression. In the cases where there were some differences in the predicted scores (e.g., vocabulary and symbol search), the differences were trivial.

Results

Table 4 shows the values that were derived from the first half of the sample in order to calculate the adjusted scaled score for each subtest. The process of generating the adjusted scaled score involved three steps: (a) calculating the predicted raw score, (b) calculating the standardized residual, and (c) calculating the adjusted scaled score. The equation to calculate the predicted raw score is as follows:  

formula
where the constant and the coefficients β-Ed, β-Age, and β-Sex are obtained from the row for the particular subtest in Table 4; education and age are entered in years, and sex is entered as 1 for men and 0 for women. For example, for a 25-year-old man with 6 years of education, the predicted raw score for vocabulary would be 34.364 or  
formula
The standardized residual is then calculated as follows:  
formula
where the actual raw score is the score the person obtained on the subtest, the predicted raw score is taken from the first equation above, and the SD(e) is the square root of the MS residual from Table 4. If the 25-year-old man with 6 years of education had an actual vocabulary raw score of 20, then the standardized residual would be  
formula
To convert the standardized residual to the usual scaled score metric with a mean of 10 and a standard deviation of 3, we compute what will become our adjusted scaled score in the following manner:  
formula
Thus, the adjusted scaled score for our hypothetical gentleman becomes  
formula
That figure would be rounded off to the closest integer, 6.

Table 4.

Coefficients used for the regression equation adjustment of the scaled scores

Scale Constant β-education β-age β-sex MS residual SD(e) 
Vocabulary 24.485 2.078 −0.144 1.011 107.299 10.35852 
Similarities 14.716 0.795 −0.111 0.240 27.672 5.260418 
Arithmetic 8.928 0.484 −0.046 1.967 8.946 2.990986 
Digit span 15.202 0.334 −0.092 0.846 13.941 3.733765 
Information 7.862 0.959 −0.032 2.441 20.757 4.555985 
Comprehension 12.289 0.813 −0.041 0.621 27.581 5.251762 
Letter-number sequencing 9.876 0.305 −0.085 0.672 6.980 2.641969 
Picture completion 19.565 0.447 −0.137 0.015 14.785 3.845127 
Coding 75.165 1.924 −0.761 3.200 330.319 18.17468 
Block design 39.501 1.282 −0.377 3.988 108.091 10.39668 
Matrix reasoning 15.839 0.645 −0.159 1.149 19.167 4.378013 
Picture arrangement 13.706 0.487 −0.146 0.818 18.730 4.327817 
Symbol search 31.974 0.975 −0.316 1.359 66.118 8.131298 
Object assembly 33.520 0.716 −0.264 0.987 65.615 8.100309 
Scale Constant β-education β-age β-sex MS residual SD(e) 
Vocabulary 24.485 2.078 −0.144 1.011 107.299 10.35852 
Similarities 14.716 0.795 −0.111 0.240 27.672 5.260418 
Arithmetic 8.928 0.484 −0.046 1.967 8.946 2.990986 
Digit span 15.202 0.334 −0.092 0.846 13.941 3.733765 
Information 7.862 0.959 −0.032 2.441 20.757 4.555985 
Comprehension 12.289 0.813 −0.041 0.621 27.581 5.251762 
Letter-number sequencing 9.876 0.305 −0.085 0.672 6.980 2.641969 
Picture completion 19.565 0.447 −0.137 0.015 14.785 3.845127 
Coding 75.165 1.924 −0.761 3.200 330.319 18.17468 
Block design 39.501 1.282 −0.377 3.988 108.091 10.39668 
Matrix reasoning 15.839 0.645 −0.159 1.149 19.167 4.378013 
Picture arrangement 13.706 0.487 −0.146 0.818 18.730 4.327817 
Symbol search 31.974 0.975 −0.316 1.359 66.118 8.131298 
Object assembly 33.520 0.716 −0.264 0.987 65.615 8.100309 

After the regression equations were derived with the first random half of the sample, the adjustments were applied, using the method described earlier, to the second half of the sample. The results were then plotted against the U.S. norms to ensure that the procedure had the desired effect of adjusting for the lower level of education of the older cohorts of the Spanish sample. Fig. 2 shows the results obtained. As can be seen, the adjusted Spanish norms are basically identical to the U.S. norms after the age of 30. There were no significant differences between the values shown.

Fig. 2.

Mean raw scores for the USA and Spanish versions of the third edition of the Wechsler adult intelligence scale, with the Spanish norms adjusted to 12 years of education.

Fig. 2.

Mean raw scores for the USA and Spanish versions of the third edition of the Wechsler adult intelligence scale, with the Spanish norms adjusted to 12 years of education.

The regression work was done again with the derivation sample separated by gender. The values derived following this system were not appreciably different from the values offered in Table 4.

Discussion

The Spanish WAIS-III is a popular test of intelligence that has contemporary norms for Spanish speakers. The instrument promises to be useful in the examination of Spanish-speaking individuals in the USA and Latin America. Although this instrument appeared to perform well with younger examinees, it inflated—from the point of view of U.S. standards—the scores of older individuals. The inflation was thought to be due to the lower level of education of the older cohorts in the Spanish sample. It should be noted, for instance, that whereas approximately 20% of the older participants in the U.S. norms had an elementary education or less, that level of education was characteristic of 90% of the Spanish sample.

The inflation of scores that would occur with the unadjusted norms of the test would lead to inaccurate assessment of older persons when the test is used in the USA. Diagnosticians examining persons with a low level of education are faced with the challenge of disentangling the effects of the low education from the effect of an acquired deficit on the test scores. Using the standardization sample of the Spanish WAIS-III, we were able to create regression equations that adjust for the level of education. These adjustments may help clinicians more accurately capture a person's present abilities, given a variety of educational levels.

Perhaps, a few examples can serve to highlight the usefulness of the adjustments. Table 5 shows the scores of a 59-year-old Puerto Rican construction worker with no formal education, sent for an evaluation of a work-related injury. The adjustments served to show that he was functioning within the broadly normal range for his age and education. The scores obtained by a 45-year-old Mexican woman with a history of mental retardation and a second-grade education are shown in Table 6. In this case, the system adjusted the scores upward as it had with the previous case, but in a way that was still consistent with the history of mental deficiency. Finally, Table 7 has the scores of a college-educated 78-year-old Cuban woman referred for a dementia work-up. In this case, the unadjusted scores inflated her performance and would have camouflaged the intellectual decay that she was experiencing. Because of her high level of education, the system adjusted the scores downward and was instrumental in showing her deficiencies.

Table 5.

Scores obtained by 59-year-old Puerto Rican construction worker

Subtest Raw score Scaled score Predicted raw score Standardized residual Adjusted scale score IQ Adjusted IQ 
Vocabulary 22 17.0 0.483 11   
Similarities 11 8.4 0.493 11   
Arithmetic 8.2 −0.395   
Digit span 10.6 −0.702   
Information 8.4 −0.969   
Comprehension 13 10.5 0.478 11   
Letter number 5.5 −0.202   
Picture completion 12 11.5 0.131 10   
Coding 24 33.5 −0.521   
Block design 12 21.2 −0.889   
Matrix reasoning 10 7.6 0.090 10   
Picture arrangement 5.9 −1.135   
Verbal scale      81 96 
Performance scale      83 88 
Full scale      80 91 
Subtest Raw score Scaled score Predicted raw score Standardized residual Adjusted scale score IQ Adjusted IQ 
Vocabulary 22 17.0 0.483 11   
Similarities 11 8.4 0.493 11   
Arithmetic 8.2 −0.395   
Digit span 10.6 −0.702   
Information 8.4 −0.969   
Comprehension 13 10.5 0.478 11   
Letter number 5.5 −0.202   
Picture completion 12 11.5 0.131 10   
Coding 24 33.5 −0.521   
Block design 12 21.2 −0.889   
Matrix reasoning 10 7.6 0.090 10   
Picture arrangement 5.9 −1.135   
Verbal scale      81 96 
Performance scale      83 88 
Full scale      80 91 
Table 6.

Scores obtained by a 45-year-old mentally defective Mexican woman

Subtest Raw score Scaled score Predicted raw score Standardized residual Adjusted scale score IQ Adjusted IQ 
Vocabulary 17 24.2 −0.693   
Similarities 11.8 −0.721   
Arithmetic 11.8 −2.260   
Digit span 13.4 −1.184   
Information 13.2 −1.805   
Comprehension 13.3 −1.392   
Letter number 8.0 −1.894   
Picture completion 14.3 −2.165   
Coding 19 51.2 −1.770   
Block design 11 33.1 −2.123   
Matrix reasoning 12.3 −1.661   
Picture arrangement 9.7 −2.021   
Verbal scale      58 69 
Performance scale      52 58 
Full scale      52 58 
Subtest Raw score Scaled score Predicted raw score Standardized residual Adjusted scale score IQ Adjusted IQ 
Vocabulary 17 24.2 −0.693   
Similarities 11.8 −0.721   
Arithmetic 11.8 −2.260   
Digit span 13.4 −1.184   
Information 13.2 −1.805   
Comprehension 13.3 −1.392   
Letter number 8.0 −1.894   
Picture completion 14.3 −2.165   
Coding 19 51.2 −1.770   
Block design 11 33.1 −2.123   
Matrix reasoning 12.3 −1.661   
Picture arrangement 9.7 −2.021   
Verbal scale      58 69 
Performance scale      52 58 
Full scale      52 58 
Table 7.

Scores obtained by a 78-year-old neurologically impaired college educated Cuban woman

Subtest Raw score Scaled score Predicted raw score Standardized residual Adjusted scale score IQ Adjusted IQ 
Vocabulary 37 13 46.5 −0.917   
Similarities 14 14 18.8 −0.908   
Arithmetic 11 15 13.1 −0.697   
Digit span 14 15 13.4 0.169 11   
Information 20.7 −2.570   
Comprehension 22.1 −3.256   
Picture completion 8.1 −0.805   
Coding 31 112 16.0 −2.609   
Block design 39 16 46.6 −0.858 12   
Matrix reasoning 12 30.6 0.807   
Picture arrangement 10 13.8 −1.315   
Verbal scale      118 71 
Performance scale      110 75 
Full scale      115 71 
Subtest Raw score Scaled score Predicted raw score Standardized residual Adjusted scale score IQ Adjusted IQ 
Vocabulary 37 13 46.5 −0.917   
Similarities 14 14 18.8 −0.908   
Arithmetic 11 15 13.1 −0.697   
Digit span 14 15 13.4 0.169 11   
Information 20.7 −2.570   
Comprehension 22.1 −3.256   
Picture completion 8.1 −0.805   
Coding 31 112 16.0 −2.609   
Block design 39 16 46.6 −0.858 12   
Matrix reasoning 12 30.6 0.807   
Picture arrangement 10 13.8 −1.315   
Verbal scale      118 71 
Performance scale      110 75 
Full scale      115 71 

A distinction is often made between “population” or “national” norms and “with-in group” or “local” norms (Strauss, Sherman, & Spreen, 2006). Population norms are typically well-defined because the representation of the various groups is guided by census data. They are most appropriate when the interest is comparing the individual being assessed to people in general. Population norms may not be able to show how the individual compares to other people with a particular characteristic. A local norm, on the other hand, is a norm collected with a subgroup that shares a particular attribute. In this case, representation is not dictated by the national census but by the inclusionary and exclusionary criteria used in collecting the data. At times, a local norm may be much more meaningful than the population norm. A student applying to college is more interested on how he/she ranks among the applicants to a school of interest than how he/she ranks among all the college applicants in the nation. There are advantages and disadvantages to both types of norms (Choca, 2004) because each type gives a different vantage point. In the context of this discussion, the kind of demographic adjustments proposed in this paper provide a way to use population norm figures to create a local norm. By adjusting the scores to simulate a 12th grade level of education, the regression equations allow us to compare a particular individual to other individuals with that level of education.

The ease of travel and the amount of communication available at this time has made national norms appear like local norms of a particular nation. Our work was done with the national norms for Spain and assumed that these norms would be very similar to what would be the population norms of all Spanish-speaking people. Although that may not be the case, this test and these norms are the most appropriate method for assessing Spanish-speaking individuals in the USA at this time.

The secondary education in Spain, roughly equivalent to the U.S. high school, is only 11 years. This fact does not pose much of a problem for the work done in this study since our goal was to adjust scores to the U.S. high-school level, and the U.S. 12 year standard could be used as the target level. The study, however, assumes that 1 year of study in Spain is roughly equivalent to 1 year of study in the USA, a presumption that cannot be easily validated.

Another complication faced by the diagnostician will occur when the adjustments are used for the WAIS-III, whereas other scores are collected through instruments that may not offer demographic adjustments (Strauss et al., 2006). Depending on the educational level that characterize the norms of the other instrument, the comparison with the adjusted WAIS-III scores may or may not be appropriate.

Another caveat from this work concerns the amount of variation seen in the raw scores of the different subtests. Fig. 3 offers a representative example with the vocabulary subtest in the 35–44 age-bracket. As can be seen, with only 4 years of education, some people demonstrated better vocabulary skills than individuals with 16 years of education. This variation means that although it is possible to adjust the scores of the entire group with accuracy, the system may not function as well with individuals who are at the peripheries of the range. There is also the issue of a possible discrepancy between the self-reported years of education and the “true educational attainment,” as reflected by the person's reading level (O'Bryant et al., 2007). More research is needed to understand how the adjustment equations can best be employed. Ultimately, the adjustments will only be useful if they increase the diagnostic or predictive validity of the test scores, an issue that would also have to be examined with further research.

Fig. 3.

Example of the variance of raw scores at specified levels of education. The variance of raw scores of the vocabulary subtest for individuals of 35–44 years of age at different levels of education.

Fig. 3.

Example of the variance of raw scores at specified levels of education. The variance of raw scores of the vocabulary subtest for individuals of 35–44 years of age at different levels of education.

The use of the adjustments will require clinician judgment. We recommend the use of those adjustments with people who are 35 years of age or older. With younger individuals, the adjustments are likely to alter the scaled scores to a level that would be different from the U.S. norms. With older people with low levels of education, the adjustments would have the opposite effect, and increase the scores to a level that would make them similar to the U.S. norms. Older individuals with a high level of education are likely to obtain inflated scores from the unadjusted norms since they would look much more capable than the poorly educated Spanish sample. In that case, as was the case with the example of Table 7, the adjustments will lower the person's score.

There may be situations where the adjustments could be seen as unfair to a particular individual, and perhaps unjustifiable. Take, for instance, the case of a person of low abilities who has been convicted of a crime. It is possible that the adjustments of the scores would place that person above the mentally defective range and would preclude consideration of the low abilities as a mitigating factor in the sentencing. In that case, it could be argued that the score that should be considered would be the unadjusted score, since that score reveals more accurately how the individual would compare with the entire Spanish population. Additionally, making the adjustments could serve to prevent an individual from obtaining needed services (Manly & Echemedia, 2007). The Psychological Corporation warns against using demographically adjusted scores in any case where the goal is to determine the functional level of the individual in comparison to the general population. The company proposes that demographically adjusted scores are best used in neuropsychological assessments where the aim is to minimize the impact of confounding variables on the diagnosis of cognitive impairment (The Psychological Corporation, 2002). In other words, the diagnosticians “need to balance the risks and benefits of using (the adjustments), and use them with a full understanding of their implications and the situations in which they are most appropriate” (Strauss et al., 2006, p. 51).

Conflict of interest

S. C. works for TEA Ediciones, the company that sells the test.

Acknowledgements

Readers are welcome to obtain an Excel spreadsheet that calculates the adjustments described in this article at http://faculty.roosevelt.edu/choca/EIWA_Adjustments.XLS.

References

Ardila
A.
Evaluación cognoscitiva en analfabetos [neuropsychological assessment in illiterates]
Revista de Neurología
 , 
2000
, vol. 
30
 (pg. 
465
-
468
)
Ardila
A.
Ostrowsky
F.
Mendoza
V.
Learning to read is much more than learning to read: A neuropsychologically based learning to read method
Journal of the International Neuropsychological Society
 , 
2000
, vol. 
6
 (pg. 
789
-
801
)
Ardila
A.
Rosselli
M.
Rosas
P.
Neuropsychological assessment in illiterates: Visuospatial and memory abilities
Brain and Cognition
 , 
1989
, vol. 
11
 (pg. 
147
-
166
)
Boone
K. B.
Victor
T. L.
Wen
J.
Razani
J.
Pontón
M.
The association between neuropsychological scores and ethnicity, language, and acculturation variables in a large patient population
Archives of Clinical Neuropsychology
 , 
2007
, vol. 
22
 (pg. 
355
-
365
)
Cherner
M.
Suarez
P.
Lazzaretto
D.
Artiola i Fortuny
L.
Rivera Mindt
M.
Dawes
S.
, et al.  . 
Demographically corrected norms for the Brief Visuospatial Memory Test–Revised and Hopkins Verbal Learning Test–Revised in monolingual Spanish speakers from the U.S.–Mexico border region
Archives of Clinical Neuropsychology
 , 
2007
, vol. 
22
 (pg. 
343
-
353
)
Choca
J. P.
Georgas
J.
Choca
J.
National versus local test norms
Culture and intelligence testing—Cross-cultural issues involving the WISC-III and WAIS-III
 , 
2004
Symposium conducted at the annual convention of the American Psychological Association, Honolulu, Hawaii
Crawford-Nutt
D. H.
The effect of educational level on the test scores of the people in South Africa
Psychologia Africana
 , 
1977
, vol. 
17
 (pg. 
49
-
59
)
da Silva
C. G.
Petersson
K. M.
Faísca
L.
Ingvar
M.
Reiss
A.
The effect of literacy and education on the quantitative and qualitative aspects of semantic verbal fluency
Journal of Clinical and Experimental Neuropsychology
 , 
2004
, vol. 
26
 (pg. 
266
-
277
)
Elias
M. F.
Elias
P. K.
D'Agostino
R.
Role of age, education, and gender on cognitive performance in the Framingham Heart Study: Community-based norms
Experimental Aging Research
 , 
1997
, vol. 
23
 (pg. 
201
-
235
)
Flynn
J. R.
What is intelligence?
 , 
2007
Cambridge
Cambridge University Press
Flynn
J. R.
¿Qué es la inteligencia?
 , 
2008
Madrid
TEA Ediciones
Folia
V.
Kosmidis
M. H.
Assessment of memory skills in illiterates: Strategy differences or test artifact?
The Clinical Neuropsychologist
 , 
2003
, vol. 
17
 (pg. 
143
-
152
)
Gonzalez da Silva
C.
Petersson
K. M.
Faisca
L.
Ingvar
M.
Reiss
A.
The effect of literacy and education on the quantitative and qualitative aspects of semantic verbal fluency
Journal of Clinical and Experimental Neuropsychology
 , 
2004
, vol. 
26
 (pg. 
266
-
277
)
Greenfield
P. M.
You can't take it with you: Why ability assessments don't cross cultures
American Psychologist
 , 
1997
, vol. 
52
 (pg. 
1115
-
1124
)
Heaton
R. K.
Taylor
M. J.
Manly
J.
Tulsky
D.
Saklofske
D. H.
Chelune
G. J.
Heaton
R. K.
Ivnik
R. J.
Bornstein
R.
, et al.  . 
Demographic effects and use of demographically corrected norms with the WAIS-III and WMS-III
Clinical interpretation of the WAIS-III and WMS-III
 , 
2003
San Diego, CA
Academic Press
(pg. 
181
-
210
)
Hsieh
S. J.
Tori
C. D.
Normative data on cross-cultural neuropsychological tests obtained from Mandarin-speaking adults across the life span
Archives of Clinical Neuropsychology
 , 
2007
, vol. 
22
 (pg. 
283
-
296
)
Kendall
I. M.
Verster
M. A.
von Mollendorf
J. W.
Irvine
S. H.
Berry
J. W.
Test performance of black in South Africa
Human abilities in cultural context
 , 
1988
Cambridge, England
Cambridge University Press
(pg. 
239
-
299
)
Krueger
K. R.
de la Torre
G. G.
Fuentes
H.
Choca
J. P.
Choca
J.
Comparing Spanish and English norms for the WAIS and WMS
Spanish language adaptations of the Wechsler scales—WISC-IV, WAIS-III/WMS-III
 , 
2005
Symposium at the annual convention of the American Psychological Association, Washington, DC
Malec
J. F.
Ivnik
R. J.
Smith
G. E.
Mayo's Older Americans Normative Studies: Utility of corrections for age and education for the WAIS-R
The Clinical Neuropsychologist
 , 
1992
, vol. 
6
 (pg. 
31
-
47
)
Manly
J. J.
Echemedia
R. J.
Race-specific norms: Using the model of hypertension to understand issues of race, culture, and education in neuropsychology
Archives of Clinical Neuropsychology
 , 
2007
, vol. 
22
 (pg. 
319
-
325
)
Manly
J. J.
Jacobs
D. M.
Sano
M.
Bell
K.
Merchant
C. A.
Small
S. A.
, et al.  . 
Effect of literacy on neuropsychological test performance in non-demented, education-matched elders
Journal of the International Neuropsychological Society
 , 
1999
, vol. 
5
 (pg. 
191
-
202
)
Mathuranath
P. S.
George
A.
Cherian
P. J.
Alexander
A.
Sarma
S. G.
Sarma
P. S.
Effects of age, education and gender on verbal fluency
Journal of Clinical and Experimental Neuropsychology
 , 
2003
, vol. 
25
 (pg. 
1057
-
1064
)
Matute
E.
Leal
F.
Zarabozo
D.
Robles
A.
Cedillo
C.
Does literacy have an effect on stick construction tasks?
Journal of the International Neuropsychological Society
 , 
2000
, vol. 
6
 (pg. 
668
-
672
)
Melendez
F.
The Spanish version of the WAIS: Some ethical considerations
The Clinical Neuropsychologist
 , 
1994
, vol. 
8
 (pg. 
388
-
393
)
Nell
V.
Myers
J.
Colvin
M.
Rees
D.
Neuropsychological assessment of organic solvent effects in South Africa: Test selection, adaptation, scoring and validation issues
Environmental Research
 , 
1993
, vol. 
63
 (pg. 
301
-
318
)
O'Bryant
S. E.
Lucas
J. A.
Willis
F. B.
Smith
G. E.
Graff-Radford
N. R.
Ivnik
R. J.
Discrepancies between self-reported years of education and estimated reading level among elderly community-dwelling African-Americans: Analysis of the MOANS data
Archives of Clinical Neuropsychology
 , 
2007
, vol. 
22
 (pg. 
327
-
332
)
Ostrosky
F.
Ardila
A.
Rosselli
M.
Lopez-Arango
G.
Uriel-Mendoza
V.
Neuropsychological test performance in illiterates
Archives of Clinical Neuropsychology
 , 
1998
, vol. 
13
 (pg. 
645
-
660
)
Ostrosky-Solís
F.
López
G.
Ardila
A.
Sensitivity and specificity of the mini-mental state examination in a Spanish-speaking population
Applied Neuropsychology
 , 
2000
, vol. 
7
 (pg. 
25
-
31
)
Ostrowsky-Solis
F.
Lozano Gutierrez
A.
Ramirez Flores
M.
Ardila
A.
Same or different? Semantic verbal fluency across Spanish-speakers from different countries
Archives of Clinical Neuropsychology
 , 
2007
, vol. 
22
 (pg. 
367
-
377
)
Ratcliff
G.
Ganguli
M.
Chandra
V.
Sharma
S.
Belle
S.
Seaberg
E.
, et al.  . 
Effects of literacy and education on measures of verbal fluency
Brain and Language
 , 
1998
, vol. 
61
 (pg. 
115
-
122
)
Reis
A.
Castro-Caldas
A.
Illiteracy: A cause for biased cognitive development
Journal of the International Neuropsychological Society
 , 
1997
, vol. 
5
 (pg. 
444
-
450
)
Reis
A.
Guerreiro
M.
Petersson
K. M.
A sociodemographic and neuropsychological characterization of an illiterate population
Applied Neuropsychology
 , 
2003
, vol. 
10
 (pg. 
191
-
204
)
Renteria
L.
Li
S. T.
Pliskin
N. H.
Reliability and validity of the Spanish language Wechsler adult intelligence scale (3rd edition) in a sample of American, urban, Spanish-speaking Hispanics
Clinical Neuropsychologist
 , 
2008
, vol. 
22
 (pg. 
455
-
470
)
Shuttleworth-Edwards
A. B.
Kemp
R. D.
Rust
A. L.
Muirhead
G. G. L.
Hartman
N. P.
Radloff
S. E.
Crosscultural effects on IQ performance: A review and preliminary normative indications on WAIS-III test performance
Journal of Clinical and Experimental Neuropsychology
 , 
2004
, vol. 
26
 (pg. 
903
-
920
)
Strauss
E.
Sherman
E.
Spreen
O.
A compendium of neuropsychological tests
 , 
2006
3rd ed.
New York
Oxford University Press
TEA Ediciones.
WAIS-III, Escala de Inteligencia Wechsler para Adultos – III.
 , 
2001
Madrid, Spain
Author
The Psychological Corporation.
Escala de Inteligencia Wechsler para Adultos.
 , 
1968
New York
Author
The Psychological Corporation.
Wechsler adult intelligence scale—Third Edition.
 , 
1997
San Antonio, TX
Author
The Psychological Corporation.
WAIS-III/WMS-III updated technical manual
 , 
2002
San Antonio, TX
Author
U.S. Census Bureau.
Current population survey, annual social and economic supplement
 , 
2006
Washington, DC
Author
Verster
J. M.
Prinsloo
R. J.
Irvine
S. H.
Berry
J. W.
The diminishing test performance gap between English speakers and Africans speakers in South Africa
Human abilities in cultural context
 , 
1988
Cambridge, England
Cambridge University Press
(pg. 
534
-
559
)
Wechsler
D.
Wechsler adult intelligence scale.
 , 
1955
New York
The Psychological Corporations