Abstract
Previous research has shown that reading ability is a stronger predictor of cognitive functioning than years of education, particularly for African Americans. The current study was designed to determine whether the relative influence of literacy and education on cognitive abilities varies as a function of race or socioeconomic status (SES). We examined the unique influence of education and reading scores on a range of cognitive tests in low and high SES African Americans and Whites. Literacy significantly predicted scores on all but one cognitive measure in both African American groups and in low SES Whites, while education was not significantly associated with any cognitive measure. In contrast, both education and reading scores predicted performance on many cognitive measures in high SES Whites. These findings provide further evidence that reading ability better predicts cognitive functioning than years of education and suggest that disadvantages associated with racial minority status and low SES affect the relative influence of literacy and years of education on cognition.
Keywords: reading, income, demographics, ethnicity, neuropsychological testing, cognition
Associations between literacy and cognitive ability have been well documented. Although there are some exceptions (Deloche et al., 1999; Manly et al., 2004; Manly et al., 1999; Reis et al., 2003), the preponderance of studies that compare the test performance of literate and illiterate individuals or that use continuous measures of literacy have shown effects of reading ability on a range of cognitive tasks, including measures of orientation, visual and verbal memory, visuospatial ability, attention, language, calculation, and praxis (Ardila et al., 1989; Deloche et al., 1999; Manly et al., 1999; Matute et al., 2000; Reis & Castro-Caldas, 1997; Reis et al., 1994; Reis et al., 2003; Rosselli et al., 1990). Longitudinal associations between literacy and cognitive decline have also been reported. Manly and colleagues (2003) found that although ethnically diverse elders with both high and low reading levels declined in immediate and delayed memory over time, the decline was more rapid among elders with a low reading level.
A number of recent studies have shown that reading ability is a better predictor of cognitive performance than education, despite the traditional use of years of education for neuropsychological test norm development and as a demographic correction in neuropsychological research. Reading level predicts cognitive performance even when controlling for education (Albert & Teresi, 1999; Byrd et al., 2005; Johnson et al., 2006; Manly et al., 2004; Manly et al., 2002; Mayeaux et al., 1995; B. D. Weiss et al., 1995). For example, in a sample of African Americans who were primarily of low socioeconomic status, we (Dotson et al., 2008) found that literacy, but not education years, significantly predicted performance on a battery of neuropsychological tests, including measures of visual and verbal memory, attention and executive functions, semantic fluency, and visuospatial abilities. Reading ability had a highly significant incremental contribution to test scores after the effect of education was partialled out. In contrast, education did not contribute to test scores after accounting for the effect of literacy.
It is hypothesized that reading is a better predictor of cognitive performance than years of education because it is a better measure of quality of education (Manly et al., 2002; Manly et al., 1999). Factors such as teaching methods, teacher quality, pupil-teacher ratios, presence of special facilities, length of school year, peer characteristics, and per pupil expenditures (Gurland et al., 1992; Manly et al., 2002) affect quality of education but are not reflected in years of education. Reading level, on the other hand, correlates with these direct measures of quality of education (Hedges et al., 1994) and with overall academic achievement (Wilkinson, 1993).
The impact of unequal educational quality may be particularly salient for African Americans, whose educational opportunities have been limited due to historical factors such as segregation (Anderson, 1988), which resulted in lower education expenditures, shorter school years, and higher student-teacher ratios for African American students (Loewenstein et al., 1994; Manly et al., 2002; Ryan et al., 2005; Whitfield, 2003). Indeed, the impact of unequal educational quality on achievement, test performance, and outcomes such as wage earnings in African Americans is well documented (Baker et al., 1996; Hanushek, 1989; Margo, 1986). Moreover, numerous studies have shown that African Americans read at a grade level that is significantly lower than their reported years of education (Albert & Teresi, 1999; Baker et al., 1996; Johnson et al., 2006; Manly et al., 2002; O’Bryant et al., 2005; F. L. Wilson, 1995; F. L. Wilson & McLemore, 1997; F. L. Wilson et al., 2003) and that the discrepancy between years of education and reading level is greater in African Americans and other minority groups than in Whites (Ryan et al., 2005).
Because of these findings, investigations of the relative influence of education and literacy on cognitive performance have primarily focused on African Americans. However, demographic factors other than race may contribute to education-reading ability discrepancies. For example, socioeconomic status (SES) is associated with cognitive functioning, perhaps because higher SES individuals have greater access to high quality education and to resources that increase the chances for participation in cognitively stimulating activities (Farah et al., 2006; Noble et al., 2007; L. G. Weiss et al., 2006; R. S. Wilson et al., 1999). Consequently, low SES, regardless of race, may be associated with poor educational quality, and thus a greater influence of literacy than education on cognitive performance. Furthermore, the discrepancy between reading ability and years of education may vary within the African-American community as a function of SES, with less of a discrepancy in higher, compared to lower, SES African Americans. The current study was aimed at investigating the unique influences of education and literacy on cognitive performance in a sample stratified by race (African American, White) and SES (low income, higher income). We hypothesized that literacy would be a better predictor of cognitive performance than education across domains of cognition, particularly for low-SES participants and for African Americans. This study extends our previous work (Dotson et al., 2008) in African Americans by examining the relative influence of literacy and education as a function of both SES and race.
Method
Participants
Data for the present study were obtained from the Healthy Aging in Neighborhoods of Diversity across the Life Span (HANDLS) study at the National Institute on Aging. HANDLS is a multidisciplinary, prospective epidemiologic longitudinal study that is collecting data from a representative sample of African Americans and Whites between 30-64 years old. A fixed cohort of participants is being recruited by household screenings from an area probability sample of twelve census segments in Baltimore, Maryland. After the baseline recruitment is complete in 2008, participants will be re-examined every three years. Data for the present study are from baseline examinations, which began in November 2004. The National Institute on Aging Intramural Research Program Institutional Review Board approved this study and all subjects gave written informed consent, in compliance with the Declaration of Helsinki.
For the purposes of this study, only the 1,610 participants with available cognitive test data and no missing demographic data were extracted from the total sample. Based on self-report data, 103 participants were excluded due to significant cardiovascular disease (e.g., coronary artery disease, myocardial infarction), and an additional 148 participants were excluded due to head injury with loss of consciousness or neurological conditions (e.g., epilepsy, stroke). Fourteen participants who reported a diagnosis of schizophrenia were also excluded, resulting in a final sample of 1,345 participants (747 women, 598 men). A summary of participant demographic information for the final sample is presented in Table 1. Participants ranged in age from 30 to 64 years (mean = 47.35, SD = 9.12), and ranged from 1 to 21 years of formal education (mean = 12.41, SD = 3.10).
Table 1.
Demographics and reading levels of the four groups
| Low SES African Americans | High SES African Americans | Low SES Whites | High SES Whites | |
|---|---|---|---|---|
| N | 487 | 270 | 209 | 379 |
| Sex, no. women/men | 280/207 | 135/135 | 135/74 | 197/182 |
| Age, M (S.D.) | 46.84 (8.62) | 48.47 (9.10) | 46.43 (9.35) | 47.71 (9.52) |
| Education, M (S.D.) | 11.58 (2.27) | 12.68 (2.81) | 11.57 (3.00) | 13.75 (3.73) |
| WRAT-3 Reading raw score, M (S.D.) | 38.84 (8.29) | 41.07 (7.00) | 42.00 (7.91) | 46.19 (7.54) |
| WRAT-3 Reading grade equivalent | ||||
| 5th grade or lower, % | 28.95 | 16.67 | 18.18 | 7.65 |
| 6th to 8th grade, % | 15.61 | 12.96 | 12.92 | 5.28 |
| High school or post-high school, % | 55.44 | 70.37 | 68.90 | 87.07 |
| Reading level = reported grade, % | 34.70 | 41.85 | 38.28 | 59.63 |
| Reading level > reported grade, % | 12.94 | 10.74 | 25.84 | 19.79 |
| Reading level < reported grade, % | 52.36 | 47.41 | 35.89 | 20.58 |
Note. SES = socioeconomic status.
Participants were self-defined as African American (N = 757) or White (N = 588). Individuals reporting multi-racial backgrounds were asked which race they identified with primarily and were categorized as such. SES was defined by self-reported income. Participants who reported income below 125% of poverty level as defined by the Department of Health and Human Services (DHHS) (Department of Health and Human Services, 2003) were considered low SES, while participants with reported income above 125% of poverty level were considered higher SES. For example, participants with families of four with income of $23,000 or lower were considered low SES because the poverty guideline for a family of four is $18,400.
Four groups were formed based on race and SES: low SES African Americans, higher SES African Americans, low SES Whites, and higher SES Whites. Analysis of variance revealed an effect of group on age [F(3, 1341) = 2.81, p = .04], however, posthoc Tukey’s HSD test did not reveal any significant pairwise comparisons. As expected, groups differed in years of education [F(3, 1341) = 44.95, p <.0001], with post hoc tests revealing significant differences for all group comparisons except for the comparison between low SES African Americans and low SES Whites. Groups also differed in the proportion of men and women [X2(3) = 12.99, p = .005].
Measures and Procedure
The neuropsychological measures were administered as part of a larger evaluation that involved cognitive evaluation, physical examination, and an in-home interview that included questionnaires about the participant’s health status, psychosocial factors, neighborhood characteristics, and demographics. Neuropsychological measures were administered by psychometrists who were trained and supervised by a research psychologist (MKT).
The reading subtest of the Wide Range Achievement Test – 3rd Edition (WRAT-3) (Wilkinson, 1993) was administered to assess participants’ ability to recognize and name letters and words. The total score was used as a continuous measure of literacy. The Benton Visual Retention Test – 5th Edition (BVRT) (Sivan, 1991) and a modified version of the California Verbal Learning Test (CVLT) (Delis et al., 1987) served as measures of short-term visual and verbal memory, respectively. For the CVLT, three, rather than five, learning trials were administered, and cued recall trials were not administered. Animal fluency assessed language and generative abilities. The Card Rotation Test (Ekstrom et al., 1976) served as a measure of visuospatial ability. The Brief Test of Attention (BTA) (Schretlen et al., 1996) and the Wechsler Adult Intelligence Scale, Revised (Weschler, 1981) Digit Span subtest measured attention and immediate verbal memory. The Trail Making Test (TMT) was administered to assess attention, cognitive control, processing speed, and visuomotor scanning (Reitan, 1992). The Identical Pictures Test (Ekstrom et al., 1976) also measured processing speed. Raw neuropsychological test scores were used in all analyses.
Data Analyses
Multiple regression analyses were used to examine the effects of literacy and years of education on cognitive performance, controlling for age and sex (women = 0, men = 1). Separate models were run for each cognitive test within each of the four groups. Using a Bonferroni correction to control for multiple comparisons, p values of less than .001 were considered significant.
In secondary analyses aimed at examining whether the contribution of education and literacy to cognitive test performance differed significantly across groups, pairwise comparisons of the parameters estimates from the multiple regression analyses were performed using Wald tests.
Results
Results of the regression analyses are summarized in Tables 2-5. For both low and high SES African Americans and for low SES Whites, reading scores were significant predictors of each cognitive measure except for TMT part A (p <.001), while education did not have a significant unique effect on any of the cognitive measures after Bonferroni correction. In contrast, both education (p <.0001) and literacy (p <.0001) were significant predictors of CVLT trials 1-3, CVLT long delay free recall, BVRT errors, Animal fluency, and Identical Pictures in high SES Whites. Neither literacy nor education was associated with the BTA or TMT part A in this group. Literacy (p <.001), but not education, significantly predicted scores on the remaining measures (Card Rotations, Digits Forward and Backwards, and TMT part B) in high SES Whites.
Table 2.
Contributions of demographic variables and literacy to test performance in low SES African Americans
| Cognitive Test | Predictor | Squared Semi-Partial Correlation | t | p |
|---|---|---|---|---|
| BVRT Errors (R2 = .106) | Age | .02 | 3.22 | .001* |
| Sex | .04 | -4.66 | <.0001* | |
| WRAT Reading | .03 | -3.55 | <.0001* | |
| Education | .01 | -2.10 | .036 | |
| CVLT-A (R2 = .147) | Age | .01 | -2.52 | .012 |
| Sex | .03 | -4.49 | <.0001* | |
| WRAT Reading | .09 | 7.48 | <.0001* | |
| Education | .00 | .00 | .996 | |
| LDFR (R2 = .145) | Age | .02 | -3.43 | .001* |
| Sex | .02 | -3.32 | .001* | |
| WRAT Reading | .09 | 7.48 | <.0001* | |
| Education | .00 | .66 | .512 | |
| Card Rotations (R2 = . 188) | Age | .05 | -4.91 | <.0001* |
| Sex | .09 | 6.74 | <.0001* | |
| WRAT Reading | .04 | 4.48 | <.0001* | |
| Education | .00 | .32 | .749 | |
| Digits Forward (R2 = . 139) | Age | .00 | .68 | .494 |
| Sex | .00 | -.61 | .543 | |
| WRAT Reading | .14 | 9.07 | <.0001* | |
| Education | .00 | .07 | .944 | |
| Digits Backward (R2 = . 182) | Age | .00 | -.70 | .487 |
| Sex | .00 | .15 | .878 | |
| WRAT Reading | .18 | 1.54 | <.0001* | |
| Education | .00 | .39 | .697 | |
| Animals (R2 = . 102) | Age | .02 | -3.66 | <.0001* |
| Sex | .03 | 4.13 | <.0001* | |
| WRAT Reading | .04 | 4.76 | <.0001* | |
| Education | .00 | 1.56 | .119 | |
| Identical Pictures (R2 = . 245) | Age | .16 | -1.16 | <.0001* |
| Sex | .00 | -1.75 | .082 | |
| WRAT Reading | .04 | 4.94 | <.0001* | |
| Education | .01 | 3.01 | .003 | |
| TMT A (R2 = . 091) | Age | .09 | 6.46 | <.0001* |
| Sex | .00 | .00 | .998 | |
| WRAT Reading | .00 | .17 | .868 | |
| Education | .00 | -.62 | .533 | |
| TMT B (R2 = .070) | Age | .03 | 3.65 | <.0001* |
| Sex | .00 | .68 | .499 | |
| WRAT Reading | .03 | -3.75 | <.0001* | |
| Education | .00 | -1.46 | .144 | |
| BTA (R2 = . 129) | Age | .01 | -2.23 | .026 |
| Sex | .00 | .41 | .683 | |
| WRAT Reading | .10 | 6.94 | <.0001* | |
| Education | .00 | 1.42 | .157 |
Note. Women were coded as 0; men were coded as 1. WRAT = Wide Range Achievement Test; BVRT = Benton Visual Retention Test; CVLT = California Verbal Learning Test; LDFR = Long-delay Free Recall; BTA = Brief Test of Attention; TMT = Trail Making Test.
p <.001 (Bonferroni corrected).
Table 5.
Contributions of demographic variables and literacy to test performance in higher SES Whites
| Cognitive Test | Predictor | Squared Semi-Partial Correlation | t | p |
|---|---|---|---|---|
| BVRT Errors (R2 = .278) | Age | .04 | 4.38 | <.0001* |
| Sex | .01 | -2.24 | .025 | |
| WRAT Reading | .08 | -5.31 | <.0001* | |
| Education | .05 | -4.30 | <.0001* | |
| CVLT-A (R2 = .279) | Age | .02 | -3.60 | <.0001* |
| Sex | .01 | -1.99 | .047 | |
| WRAT Reading | .07 | 5.02 | <.0001* | |
| Education | .08 | 5.25 | <.0001* | |
| LDFR (R2 = .270) | Age | .03 | -4.30 | <.0001* |
| Sex | .01 | -2.06 | .040 | |
| WRAT Reading | .05 | 4.18 | <.0001* | |
| Education | .09 | 5.50 | <.0001* | |
| Card Rotations (R2 = . 234) | Age | .06 | -4.93 | <.0001* |
| Sex | .02 | 3.18 | .002 | |
| WRAT Reading | .11 | 5.56 | <.0001* | |
| Education | .01 | 1.47 | .144 | |
| Digits Forward (R2 = . 204) | Age | .00 | -1.49 | .138 |
| Sex | .00 | 1.16 | .247 | |
| WRAT Reading | .17 | 7.40 | <.0001* | |
| Education | .00 | .88 | .381 | |
| Digits Backward (R2 = . 269) | Age | .01 | -2.28 | .023 |
| Sex | .00 | .95 | .341 | |
| WRAT Reading | .23 | 9.12 | <.0001* | |
| Education | .00 | .60 | .546 | |
| Animals (R2 = . 239) | Age | .02 | -3.47 | .001* |
| Sex | .00 | .66 | .507 | |
| WRAT Reading | .07 | 4.76 | <.0001* | |
| Education | .06 | 4.61 | <.0001* | |
| Identical Pictures (R2 = . 376) | Age | .17 | -9.57 | <.0001* |
| Sex | .00 | -.94 | .347 | |
| WRAT Reading | .05 | 4.29 | <.0001* | |
| Education | .06 | 4.97 | <.0001* | |
| TMT A (R2 = . 113) | Age | .03 | 3.31 | .001* |
| Sex | .00 | 1.11 | .268 | |
| WRAT Reading | .02 | -2.11 | .035 | |
| Education | .03 | -2.92 | .004 | |
| TMT B (R2 = .260) | Age | .00 | .85 | .399 |
| Sex | .00 | 1.34 | .181 | |
| WRAT Reading | .17 | -7.44 | <.0001* | |
| Education | .02 | -2.40 | .017 | |
| BTA (R2 = . 032ns) | Age | .00 | -.22 | .828 |
| Sex | .00 | -.02 | .981 | |
| WRAT Reading | .02 | 2.22 | .027 | |
| Education | .00 | .78 | .438 |
Note. Women were coded as 0; men were coded as 1. WRAT = Wide Range Achievement Test; BVRT = Benton Visual Retention Test; CVLT = California Verbal Learning Test; LDFR = Long-delay Free Recall; BTA = Brief Test of Attention; TMT = Trail Making Test.
p <.001 (Bonferroni corrected).
Secondary analyses comparing regression parameters across groups (Table 6) revealed that the association of WRAT-3 reading scores after adjusting for demographic measures were significantly smaller in low SES African Americans compared to low SES Whites for Card Rotations (Wald z = -2.06, p <.05) and Identical Pictures (Wald z = -2.13, p <.05), and compared to high SES Whites for Card Rotations (Wald z = -2.48, p <.01), Digits Forward (Wald z = -2.14, p <.05) and Backward (Wald z = -2.93, p <.01), and TMT part B (Wald z = 3.08, p <.01). Education estimates were significantly larger in high SES Whites compared to low and high SES African Americans for CVLT trials 1-3 (low SES Wald z = -4.61, p <.001; high SES Wald z = -2.31, p <.05), CVLT long delay free recall (low SES Wald z = -4.42, p <.001; high SES Wald z = -2.69, p <.001), and Animal fluency (low SES Wald z = -3.24, p <.001; high SES Wald z = -2.61, p <.001). Education regression parameters were also larger in high SES Whites compared to low SES African Americans for Identical Pictures (Wald z = -3.24, p <.001), and compared to low SES Whites for CVLT trials 1-3 (Wald z = -2.14, p <.05).
Table 6.
Wald z scores from the comparison of group regression parameters for the WRAT-3 Reading score and years of education
| Low SES African Americans vs. Higher SES African Americans | Low SES African Americans vs. Low SES Whites | Low SES African Americans vs. Higher SES Whites | Higher SES African Americans vs. Low SES Whites | Higher SES African Americans vs. Higher SES Whites | Low SES Whites vs. Higher SES Whites | |
|---|---|---|---|---|---|---|
| WRAT-3 | ||||||
| Reading | ||||||
| BVRT Errors | 1.47 | 1.65 | 1.12 | -.15 | -.60 | -.56 |
| CVLT-A | -.19 | -.43 | -.84 | -.22 | -.54 | -.30 |
| LDFR | .14 | -.06 | .35 | -.16 | .18 | .32 |
| Card Rotations | -.65 | -2.06* | -2.48** | -1.32 | -1.59 | -.15 |
| Digits Forward | -.49 | -1.28 | -2.14* | -.64 | -1.37 | -.74 |
| Digits Backward | -1.77 | -1.45 | -2.93** | .21 | -.96 | -1.14 |
| Animals | -.83 | -.77 | -1.54 | .03 | -.59 | -.61 |
| Identical Pictures | -.59 | -2.13* | -.96 | -1.36 | -.32 | 1.04 |
| TMT A | 1.20 | 1.88 | 1.52 | .34 | -.12 | -.58 |
| TMT B | 1.07 | 1.87 | 3.08** | .97 | 1.55 | .16 |
| BTA | .65 | -.34 | .32 | -.83 | -.07 | .50 |
| Education | ||||||
| BVRT Errors | -.52 | -.45 | .25 | .10 | .91 | .84 |
| CVLT-A | -1.42 | -.93 | -4.61*** | .16 | -2.31* | -2.14* |
| LDFR | -.68 | -1.39 | -4.42*** | -.67 | -2.69** | -1.67 |
| Card Rotations | -.11 | .37 | -.20 | .46 | -.07 | -.55 |
| Digits Forward | -.53 | -1.04 | -.60 | -.45 | -.01 | .46 |
| Digits Backward | .57 | -1.10 | -.51 | -1.29 | -.84 | .61 |
| Animals | .28 | -1.34 | -3.24*** | -1.28 | -2.61** | -1.07 |
| Identical Pictures | -.68 | .35 | -2.21* | .78 | -1.02 | -1.74 |
| TMT A | .56 | .01 | 1.60 | -.46 | .38 | 1.06 |
| TMT B | .77 | .15 | .92 | -.34 | -.08 | .32 |
| BTA | .28 | .21 | .15 | -.05 | -.07 | -.02 |
Note. SES = socioeconomic status; WRAT = Wide Range Achievement Test; BVRT = Benton Visual Retention Test; CVLT = California Verbal Learning Test; LDFR = Long-delay Free Recall; TMT = Trail Making Test; BTA = Brief Test of Attention.
p <.05
p <.01
p <.001
Discussion
The purpose of this study was to examine the unique influence of literacy and education on cognitive performance in a sample stratified by race and SES. Given the associations of both race and SES with quality of education, we expected literacy to be a better predictor of cognitive functioning than education in African Americans and in low SES participants.
Results confirmed our hypotheses. Consistent with our previous work (Dotson et al., 2008) as well as the work of others (Albert & Teresi, 1999; Byrd et al., 2005; Johnson et al., 2006; Manly et al., 2004; Manly et al., 2002; Mayeaux et al., 1995; B. D. Weiss et al., 1995), literacy was a stronger predictor of cognitive performance than years of education in African Americans. While significant WRAT-3 reading effects were observed, education did not have a significant effect on any measure once reading ability was taken into account. This relationship held for both verbal and nonverbal measures, and was found for all but one test. Both low and high SES African Americans showed this pattern of results, thus, literacy appears to be a stronger predictor of cognitive functioning than education regardless of SES in African Americans. In contrast, findings varied by SES in White participants. Low SES Whites were similar to African Americans; that is, literacy was a significant predictor of all but one measure, while education did not significantly predict any measure. However, findings were more variable for high SES Whites, with both literacy and education showing significant relationships with some measures, while for other measures, literacy, but not education, was a significant predictor. The differential effects of SES on our results in White and African-American participants may be related to social mobility. Participants with currently low SES may have come from a low SES background, suggesting that quality of education during the school years would have been poor. The high SES groups, on the other hand, may comprise a mix of individuals, some of whom came from a high SES background and others who may have changed social status in adulthood. It is possible that our high SES African-American group is more likely to consist of individuals who were raised in a lower SES environment but in adulthood were able to benefit from increasing opportunities for African Americans. In that case, their quality of education as a child may not have substantially differed from that of individuals in the low SES African-American group. In contrast, the high SES White group may be more heterogeneous in regards to childhood SES, and thus have more variable educational quality. Because information about childhood SES was not available for the current study, were unable to test this possibility.
Secondary analyses, which compared the associations of reading scores and education across groups, revealed smaller reading parameter estimates in low SES African Americans compared to low and high SES Whites for some measures. This is not surprising considering that the regression models tended to account for less variance in low SES African Americans (9%-25%) than in low and high SES Whites (11%-38%). For low SES African Americans, although reading level is a better predictor of cognitive performance than education, the association between reading and cognitive functioning is smaller compared to Whites for some cognitive functions. Thus, other factors that were not included in our regression models may be important in predicting the cognitive performance of low SES African Americans. For some measures, secondary analyses also revealed larger education estimates in high SES Whites compared to the other groups. Combined with the finding that both literacy and education predicted performance on some measures in high SES Whites, the secondary analyses suggest that education is a better predictor of cognitive abilities in high SES Whites than in other groups.
Our findings highlight the importance of considering an individual’s reading level when interpreting performance on cognitive tasks. Previous studies have shown that literacy is a better predictor of cognitive performance than years of education, presumably because it is a better measure of quality of education (Albert & Teresi, 1999; Byrd et al., 2005; Johnson et al., 2006; Manly et al., 2004; Manly et al., 2002; Mayeaux et al., 1995; B. D. Weiss et al., 1995). Although research in this area has focused on African Americans, our results suggest that reading ability may be a more important consideration than education years for some cognitive abilities in Whites as well, particularly in those with low SES. The finding that education predicted performance on some measures in high SES Whites but was not associated with any cognitive measure in African Americans and low SES Whites is consistent with the idea that individuals from disadvantaged groups are more likely to obtain poor quality education. As a result, education is less likely to accurately reflect educational achievement or predict cognitive performance in these groups. The extension of previous research in African Americans to another disadvantaged group (i.e., those with low SES) underscores the need for research that examines predictors of cognitive performance in myriad groups with limited educational opportunities.
The potential impact of intellectual functioning on our findings is unclear. Both educational attainment and reading ability are associated with intelligence, and word reading tests are frequently used as estimates of premorbid intelligence (Bright et al., 2002; Crawford et al., 2001). Because of these relationships, including intelligence scores in our statistical models would have resulted in problems with multicollinearity. It is possible that group differences in intelligence affected the relative contributions of reading ability and education level to cognitive test scores. Another possibility is that reading ability is a stronger predictor of cognitive functioning in low SES and African-American participants because it has a stronger correlation with intelligence than does education years in those groups. Because the HANDLS study does not include intelligence estimates other than word reading ability, we were unable to explore these possibilities in the current study.
We chose to perform separate analyses for each group and for each cognitive test in order to avoid obscuring differences between tests caused by forming composite scores and to provide the most straightforward demonstration of group differences in the relative contribution of education years and literacy on cognitive functioning. Although the number of analyses in this study was inflated, we do not consider this to be a limitation of the study because the results withstood Bonferroni correction, which is a very conservative correction for multiple comparisons.
Participants were not given a learning disability evaluation. As a result, it is possible that undiagnosed cases of reading disabilities were present in our sample. Although this may have contributed to the observed discrepancy between reading ability and reported grade level, particularly in low SES groups, it is unlikely to have affected our analysis of the relative contribution of education years and reading level to cognitive scores. Indeed, because individuals with reading disabilities would be expected to have reading skills that are much lower than other cognitive abilities, it is likely that the presence of undiagnosed learning disability would have reduced the impact of reading scores on our cognitive tests. The magnitude and consistency of our finding that literacy is a better predictor of cognitive functioning than education years despite the possible inclusion of individuals with a learning disability attests to the strength of our findings.
The proportion of women was greater in the high SES White group (65%) compared to the other groups (50%-58%). The inclusion of more women may have contributed to the differential findings in this group. However, this possibility was minimized by the inclusion of sex as a covariate in the statistical analyses. Although years of education for our sample ranged from 1-21 years, the majority of participants had 9-13 years of education. Thus, the limited variability in education years may have obscured education effects, since education is known to have a nonlinear effect on cognition (Ardila et al., 2000). Moreover, the limited variability in education years suggests that the present results may not generalize to individuals with extremely low or extremely high levels of education, as they were not adequately sampled in this study. The categorical definition of low and high SES groups based solely on current income and the relatively smaller sample sizes in the low SES White and high SES African-American groups are additional limitations of our study.
Nonetheless, the present results are useful in that they provide further evidence that reading ability better predicts cognitive functioning than years of education, and they suggest that disadvantages associated with racial minority status and low SES affect the relative influence of literacy and years of education on cognition. Our findings contribute to the existing literature by providing evidence that 1) despite the previous focus on African Americans, literacy is a better predictor of cognitive functioning than education in both African Americans and Whites, and 2) SES affects the relative contribution of reading ability and education to cognitive performance in Whites, but not in African Americans. These results also suggest that minority status and socioeconomic status have independent effects on cognitive performance. Additional research is needed to examine the effects of education and literacy on cognitive performance in different ethnic groups. Moreover, our understanding of group differences in neuropsychological test performance will be enhanced by further exploration of intra-group differences and the impact of diverse cultural experiences on cognitive performance.
Table 3.
Contributions of demographic variables and literacy to test performance in higher SES African Americans
| Cognitive Test | Predictor | Squared Semi-Partial Correlation | t | p |
|---|---|---|---|---|
| BVRT Errors (R2 = .190) | Age | .01 | 1.78 | .076 |
| Sex | .02 | -1.95 | .052 | |
| WRAT Reading | .13 | -5.17 | <.0001* | |
| Education | .01 | -1.41 | .161 | |
| CVLT-A (R2 = .116) | Age | .00 | .67 | .504 |
| Sex | .02 | -2.73 | .007 | |
| WRAT Reading | .07 | 4.26 | <.0001* | |
| Education | .01 | 1.39 | .164 | |
| LDFR (R2 = .117) | Age | .00 | -.74 | .463 |
| Sex | .03 | -3.02 | .003 | |
| WRAT Reading | .07 | 4.35 | <.0001* | |
| Education | .00 | .80 | .424 | |
| Card Rotations (R2 = . 106) | Age | .01 | -1.59 | .114 |
| Sex | .02 | 2.03 | .044 | |
| WRAT Reading | .07 | 4.04 | <.0001* | |
| Education | .00 | .80 | .422 | |
| Digits Forward (R2 = . 130) | Age | .01 | 1.50 | .133 |
| Sex | .01 | 2.17 | .031 | |
| WRAT Reading | .11 | 5.62 | <.0001* | |
| Education | .00 | .49 | .621 | |
| Digits Backward (R2 = . 190) | Age | .00 | .01 | .994 |
| Sex | .00 | .90 | .369 | |
| WRAT Reading | .19 | 7.69 | <.0001* | |
| Education | .00 | -.19 | .848 | |
| Animals (R2 = . 089) | Age | .01 | -1.62 | .106 |
| Sex | .01 | 1.89 | .059 | |
| WRAT Reading | .07 | 4.21 | <.0001* | |
| Education | .00 | .07 | .948 | |
| Identical Pictures (R2 = . 336) | Age | .24 | -9.42 | <.0001* |
| Sex | .01 | -1.54 | .124 | |
| WRAT Reading | .05 | 3.96 | <.0001* | |
| Education | .01 | 1.81 | .071 | |
| TMT A (R2 = . 153) | Age | .13 | 6.11 | <.0001* |
| Sex | .00 | .77 | .440 | |
| WRAT Reading | .01 | -1.96 | .051 | |
| Education | .00 | -.28 | .779 | |
| TMT B (R2 = .145) | Age | .03 | 3.12 | .002 |
| Sex | .00 | -.96 | .337 | |
| WRAT Reading | .08 | -4.74 | <.0001* | |
| Education | .01 | -1.27 | .204 | |
| BTA (R2 = . 130) | Age | .05 | -3.65 | <.0001* |
| Sex | .01 | -1.74 | .083 | |
| WRAT Reading | .04 | 3.23 | .001* | |
| Education | .01 | 1.28 | .203 |
Note. Women were coded as 0; men were coded as 1. WRAT = Wide Range Achievement Test; BVRT = Benton Visual Retention Test; CVLT = California Verbal Learning Test; LDFR = Long-delay Free Recall; BTA = Brief Test of Attention; TMT = Trail Making Test.
p <.001 (Bonferroni corrected).
Table 4.
Contributions of demographic variables and literacy to test performance in low SES Whites
| Cognitive Test | Predictor | Squared Semi-Partial Correlation | t | p |
|---|---|---|---|---|
| BVRT Errors (R2 = .233) | Age | .05 | 3.95 | <.0001* |
| Sex | .02 | -2.15 | .033 | |
| WRAT Reading | .13 | -5.44 | <.0001* | |
| Education | .01 | -1.36 | .176 | |
| CVLT-A (R2 = .137) | Age | .00 | -.56 | .578 |
| Sex | .04 | -3.27 | .001* | |
| WRAT Reading | .08 | 4.10 | <.0001* | |
| Education | .00 | .71 | .481 | |
| LDFR (R2 = .146) | Age | .00 | -.92 | .357 |
| Sex | .03 | -2.94 | .004 | |
| WRAT Reading | .08 | 4.06 | <.0001* | |
| Education | .01 | 1.41 | .160 | |
| Card Rotations (R2 = . 176) | Age | .07 | -3.94 | <.0001* |
| Sex | .01 | 1.38 | .170 | |
| WRAT Reading | .11 | 4.29 | <.0001* | |
| Education | .00 | -.28 | .783 | |
| Digits Forward (R2 = . 236) | Age | .00 | -1.02 | .309 |
| Sex | .02 | 2.65 | .008 | |
| WRAT Reading | .20 | 6.94 | <.0001* | |
| Education | .00 | .40 | .691 | |
| Digits Backward (R2 = . 253) | Age | .00 | -.63 | .530 |
| Sex | .00 | -.30 | .768 | |
| WRAT Reading | .22 | 7.32 | <.0001* | |
| Education | .00 | .99 | .325 | |
| Animals (R2 = . 147) | Age | .02 | -2.11 | .036 |
| Sex | .00 | .64 | .526 | |
| WRAT Reading | .08 | 3.92 | <.0001* | |
| Education | .02 | 1.90 | .059 | |
| Identical Pictures (R2 = . 315) | Age | .20 | -7.54 | <.0001* |
| Sex | .01 | -1.83 | .069 | |
| WRAT Reading | .09 | 4.56 | <.0001* | |
| Education | .00 | 1.04 | .298 | |
| TMT A (R2 = . 122) | Age | .06 | 3.64 | <.0001* |
| Sex | .02 | 2.32 | .021 | |
| WRAT Reading | .03 | -2.49 | .014 | |
| Education | .00 | -.56 | .574 | |
| TMT B (R2 = .116) | Age | .00 | .87 | .386 |
| Sex | .02 | 2.32 | .022 | |
| WRAT Reading | .09 | -3.96 | <.0001* | |
| Education | .00 | -.19 | .848 | |
| BTA (R2 = . 152) | Age | .00 | .79 | .432 |
| Sex | .00 | -.50 | .618 | |
| WRAT Reading | .12 | 4.69 | <.0001* | |
| Education | .01 | 1.12 | .265 |
Note. Women were coded as 0; men were coded as 1. WRAT = Wide Range Achievement Test; BVRT = Benton Visual Retention Test; CVLT = California Verbal Learning Test; LDFR = Long-delay Free Recall; BTA = Brief Test of Attention; TMT = Trail Making Test.
p <.001 (Bonferroni corrected).
Acknowledgments
This research was supported (in part) by the Intramural Research Program of the NIH, National Institute on Aging.
Footnotes
Portions of this data were presented in Dotson et al. (2008).
The authors have no conflicts of interest to disclose.
References
- Albert SM, Teresi JA. Reading ability, education, and cognitive status assessment among older adults in Harlem, New York City. American Journal of Public Health. 1999;89(1):95–97. doi: 10.2105/ajph.89.1.95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Anderson JD. The education of Blacks in the south, 1860-1935. Chapel Hill: University of North Carolina Press; 1988. [Google Scholar]
- Ardila A, Ostrosky-Solis F, Rosselli M, Gomez C. Age-related cognitive decline during normal aging: the complex effect of education. Archives of Clinical Neuropsychology. 2000;15(6):495–513. [PubMed] [Google Scholar]
- Ardila A, Rosselli M, Rosas P. Neuropsychological assessment in illiterates: Visuospatial and memory abilities. Brain and Cognition. 1989;11(2):147–166. doi: 10.1016/0278-2626(89)90015-8. [DOI] [PubMed] [Google Scholar]
- Baker FM, Johnson JT, Velli SA, Wiley C. Congruence between education and reading levels of older persons. Psychiatric Services. 1996;47(2):194–196. doi: 10.1176/ps.47.2.194. [DOI] [PubMed] [Google Scholar]
- Bright P, Jaldow ELI, Kopelman MD. The National Adult Reading Test as a measure of premorbid intelligence: A comparison with estimates derived from demographic variables. Journal of the International Neuropsychological Society. 2002;8(6):847–854. doi: 10.1017/s1355617702860131. [DOI] [PubMed] [Google Scholar]
- Byrd DA, Jacobs DM, Hilton HJ, Stern Y, Manly JJ. Sources of errors on visuoperceptual tasks: role of education, literacy, and search strategy. Brain and Cognition. 2005;58(3):251–257. doi: 10.1016/j.bandc.2004.12.003. [DOI] [PubMed] [Google Scholar]
- Crawford JR, Deary IJ, Starr J, Whalley LJ. The NART as an index of prior intellectual functioning: a retrospective validity study covering a 66-year interval. Psychological Medicine. 2001;31(3):451–458. doi: 10.1017/s0033291701003634. [DOI] [PubMed] [Google Scholar]
- Delis DC, Kramer JH, Kaplan E, Ober BA. California Verbal Learning Test: Adult Version. San Antonio, TX: The Psychological Corporation; 1987. [Google Scholar]
- Deloche G, Souza L, Braga LW, Dellatolas G. A calculation and number processing battery for clinical application in illiterates and semi-literates. Cortex. 1999;35(4):503–521. doi: 10.1016/s0010-9452(08)70815-3. [DOI] [PubMed] [Google Scholar]
- Department of Health and Human Services. Annual Update of the HHS Poverty Guidelines. Federal Register. 2003;68(26):6456–6458. [Google Scholar]
- Dotson VM, Kitner-Triolo M, Evans MK, Zonderman AB. Literacy-based normative data for low socioeconomic status African Americans. The Clinical Neuropsychologist. 2008;22(6):989–1017. doi: 10.1080/13854040701679017. [DOI] [PubMed] [Google Scholar]
- Ekstrom RB, French JW, Harman HH. Manual for kit of factor referenced cognitive tests. Princeton, NJ: Educational Testing Service; 1976. [Google Scholar]
- Farah MJ, Shera DM, Savage JH, Betancourt L, Giannetta JM, Brodsky NL, Malmud EK, Hurt H. Childhood poverty: Specific associations with neurocognitive development. Brain Research. 2006;1110(1):166–174. doi: 10.1016/j.brainres.2006.06.072. [DOI] [PubMed] [Google Scholar]
- Gurland BJ, Wilder DE, Cross P, Teresi J, Barrett VW. Screening scales for dementia: Toward reconciliation of conflicting cross-cultural findings. International Journal of Geriatric Psychiatry. 1992;7(2):105–113. [Google Scholar]
- Hanushek EA. The Impact of Differential Expenditures on School Performance. Educational Researcher. 1989;18(4):45–51. [Google Scholar]
- Hedges LV, Laine RD, Greenwald R. Does Money Matter? A Meta-Analysis of Studies of the Effects of Differential School Inputs on Student Outcomes. Educational Researcher. 1994;23(3):5–14. [Google Scholar]
- Johnson AS, Flicker LJ, Lichtenberg PA. Reading ability mediates the relationship between education and executive function tasks. Journal of the International Neuropsychological Society. 2006;12(1):64–71. doi: 10.1017/S1355617706060073. [DOI] [PubMed] [Google Scholar]
- Loewenstein DA, Arguelles T, Arguelles S, Linn-Fuentes P. Potential cultural bias in the neuropsychological assessment of the older adult. Journal of Clinical and Experimental Neuropsychology. 1994;16(4):623–629. doi: 10.1080/01688639408402673. [DOI] [PubMed] [Google Scholar]
- Manly JJ, Byrd D, Touradji P, Sanchez D, Stern Y. Literacy and cognitive change among ethnically diverse elders. International Journal of Psychology. 2004;39(1):47–60. [Google Scholar]
- Manly JJ, Jacobs DM, Ferraro FR. Minority and cross-cultural aspects of neuropsychological assessment. Lisse, Netherlands: Swets & Zeitlinger; 2002. Future directions in neuropsychological assessment with African Americans; pp. 79–96. [Google Scholar]
- Manly JJ, Jacobs DM, Sano M, Bell K, Merchant CA, Small SA, Stern Y. Effect of literacy on neuropsychological test performance in nondemented, education-matched elders. Journal of the International Neuropsychological Society. 1999;5(3):191–202. doi: 10.1017/s135561779953302x. [DOI] [PubMed] [Google Scholar]
- Manly JJ, Touradji P, Tang MX, Stern Y. Literacy and memory decline among ethnically diverse elders. Journal of Clinical and Experimental Neuropsychology. 2003;25(5):680–690. doi: 10.1076/jcen.25.5.680.14579. [DOI] [PubMed] [Google Scholar]
- Margo RA. Education Achievement in Segregated School Systems: The Effects of “Separate-But-Equal”. American Economic Review. 1986;76(4):794–801. [Google Scholar]
- 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;6(6):668–672. doi: 10.1017/s1355617700666043. [DOI] [PubMed] [Google Scholar]
- Mayeaux EJ, Jr, Davis TC, Jackson RH, Henry D, Patton P, Slay L, Sentell T. Literacy and self-reported educational levels in relation to Mini-mental State Examination scores. Family Medicine. 1995;27(10):658–662. [PubMed] [Google Scholar]
- Noble KG, McCandliss BD, Farah MJ. Socioeconomic gradients predict individual differences in neurocognitive abilities. Developmental Science. 2007;10(4):464–480. doi: 10.1111/j.1467-7687.2007.00600.x. [DOI] [PubMed] [Google Scholar]
- O’Bryant SE, Schrimsher GW, O’Jile JR. Discrepancies between self-reported years of education and estimated reading level: potential implications for neuropsychologists. Applied Neuropsychology. 2005;12(1):5–11. doi: 10.1207/s15324826an1201_2. [DOI] [PubMed] [Google Scholar]
- Reis A, Castro-Caldas A. Illiteracy: A cause for biased cognitive development. Journal of the International Neuropsychological Society. 1997;3(5):444–450. [PubMed] [Google Scholar]
- Reis A, Guerreiro M, Castro-Caldas A. Influence of educational level of non brain-damaged subjects on visual naming capacities. Journal of Clinical and Experimental Neuropsychology. 1994;16(6):939–942. doi: 10.1080/01688639408402705. [DOI] [PubMed] [Google Scholar]
- Reis A, Guerreiro M, Petersson KM. A sociodemographic and neuropsychological characterization of an illiterate population. Applied Neuropsychology. 2003;10(4):191–204. doi: 10.1207/s15324826an1004_1. [DOI] [PubMed] [Google Scholar]
- Reitan R. Trail Making Test: Manual for Administration and Scoring. Tucson, AZ: Reitan Neuropsychological Laboratory; 1992. [Google Scholar]
- Rosselli M, Ardila A, Rosas P. Neuropsychological assessment in illiterates. II. Language and praxic abilities. Brain and Cognition. 1990;12(2):281–296. doi: 10.1016/0278-2626(90)90020-o. [DOI] [PubMed] [Google Scholar]
- Ryan EL, Baird R, Mindt MR, Byrd D, Monzones J, Bank SM. Neuropsychological impairment in racial/ethnic minorities with HIV infection and low literacy levels: effects of education and reading level in participant characterization. Journal of the International Neuropsychological Society. 2005;11(7):889–898. doi: 10.1017/S1355617705051040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schretlen D, Bobholz JH, Brandt J. Development and psychometric properties of the Brief Test of Attention. The Clinical Neuropsychologist. 1996;10(1):80–89. [Google Scholar]
- Sivan AB. The Benton Visual Retention Test: Fifth Edition Manual. San Antonio, TX: The Psychological Corporation; 1991. [Google Scholar]
- Weiss BD, Reed R, Kligman EW, Abyad A. Literacy and performance on the Mini-Mental State Examination. Journal of the American Geriatrics Society. 1995;43(7):807–810. doi: 10.1111/j.1532-5415.1995.tb07057.x. [DOI] [PubMed] [Google Scholar]
- Weiss LG, Harris JG, Prifitera A, Courville T, Rolfhus E, Saklofske DH, Holdnack JA. WISC-IV interpretation in societal context. In: Weiss LG, Saklofske DH, Prifitera A, Holdnack JA, editors. WISC-IV: Advanced clinical interpretation. Burlington, MA: Academic Press/Elsevier; 2006. pp. 1–58. [Google Scholar]
- Weschler D. Weschler Adult Intelligence Scale - Revised. New York: Psychological Corporation; 1981. [Google Scholar]
- Whitfield KE, Wiggins S. The impact of desegregation on cognition among older African Americans. Journal of African American Psychology. 2003;29:275–291. [Google Scholar]
- Wilkinson GS. Wide Range Achievement Test–Revision 3. Wilmington, DE: Jastak Association; 1993. [Google Scholar]
- Wilson FL. Measuring patients’ ability to read and comprehend: a first step in patient education. NursingConnections. 1995;8(4):17–25. [PubMed] [Google Scholar]
- Wilson FL, McLemore R. Patient literacy levels: a consideration when designing patient education programs. Rehabilitation Nursing. 1997;22(6):311–317. doi: 10.1002/j.2048-7940.1997.tb02124.x. [DOI] [PubMed] [Google Scholar]
- Wilson FL, Racine E, Tekieli V, Williams B. Literacy, readability and cultural barriers: critical factors to consider when educating older African Americans about anticoagulation therapy. Journal of Clinical Nursing. 2003;12(2):275–282. doi: 10.1046/j.1365-2702.2003.00711.x. [DOI] [PubMed] [Google Scholar]
- Wilson RS, Bennett DA, Beckett LA, Morris MC, Gilley DW, Bienias JL, Scherr PA, Evans DA. Cognitive activity in older persons from a geographically defined population. Journals of Gerontology Series B: Psychological Sciences and Social Sciences. 1999;54(3):P155–160. doi: 10.1093/geronb/54b.3.p155. [DOI] [PubMed] [Google Scholar]