Verbal and nonverbal intelligence scores within the context of poverty

Abstract

This study explored the relationship between verbal and nonverbal intelligence scores in students from low-income households. Researchers investigated how the verbal and nonverbal scores of these students differed both by ability level and within each grade. The Reynolds Intellectual Assessment Scale was administered to 1935 kindergarten through fifth-grade students enrolled in free and reduced-price lunch programs. Large gaps were found between students’ verbal and nonverbal intelligence scores; such gaps were not specific to any particular grade or ability level but rather were seen in all of the examined grades. Gaps were even larger for students with higher nonverbal IQ scores. Results suggest that assessments based solely on nonverbal scores should be interpreted with caution.

Keywords

Poverty verbal scores nonverbal scores verbal–nonverbal gap gifted

Cattell and Horn’s fluid–crystallized (Gf-Gc) theory of intelligence is perhaps the best known and most widely used comprehensive theory of intelligence. Horn and Cattell (1966) claimed that primary mental abilities could be organized into two general types of intelligence: fluid and crystallized. Many current intelligence tests include both verbal and nonverbal subtests to assess these abilities. For example, the Reynolds Intellectual Assessment Scale (RIAS; Reynolds and Kamphaus, 2003) is an intelligence test with verbal and nonverbal subtests that assess both crystallized and fluid intelligences.

Fluid intelligence is defined as the use of purposeful and controlled mental operations to solve novel problems that cannot be solved automatically or by simple memorization. Examples of these types of mental operations include problem-solving, identifying relations, comprehending implications, concept formation, classification, drawing inferences, transforming information, and inductive and deductive reasoning (McGrew, 2009; McGrew and Evans, 2004). Fluid intelligence is related to the g factor (Blair, 2006) and conceptualized as a causal factor for learning, especially in novel situations (Kvist and Gustafsson, 2008; Watkins et al., 2007). Fluid intelligence also supports the acquisition of skills and knowledge that leads to the formation of crystallized intelligence (Cattell, 1971). Factor analytic studies on the structure of human cognitive abilities routinely find that nonverbal intelligence tests can accurately measure fluid intelligence (Gustafsson and Undheim, 1996).

Crystallized intelligence, on the other hand, includes acquired knowledge and skills as well as language and vocabulary usage (Horn and Cattell, 1966). As people age and compile learning experiences, they accumulate knowledge and their crystallized intelligence becomes more robust. For example, vocabulary acquisition is an example of crystallized intelligence. Willis et al. (2011: 44) state “Most verbal subtests of intelligence scales are classified primarily as measuring crystalized intelligence.”

A comprehensive measure of intelligence should include both verbal and nonverbal components, as a measure of one of these components represents only part of an individual’s total intellectual ability. Such partial testing can result in construct underrepresentation (Lohman, 2005; Warne, 2009) and thus represent a primary threat to test score validity (Braden, 2000).

Many school districts have difficulties increasing the diversity of students enrolled in their gifted education programs (Baer, 1980; Borland and Wright, 1994; Callahan, 2001; Renzulli, 1982). To address this problem, nonverbal ability tests have increasingly been used to place underrepresented minority students into gifted programs (see Lakin and Lohman, 2011; Lewis et al., 2007; Saccuzzo et al., 1992). Although this tactic may increase the diversity of students in these programs, such assessment practices also have drawbacks. Verbal intelligence scores tend to have greater predictive validities of achievement than do nonverbal scores (Benbow and Stankey, 1996; Lakin and Lohman, 2011), while nonverbal tests often yield low validities for predicting reading and math achievement (Lakin, 2012). Specifically, verbal tests have predictive validities for achievement that range from r = 0.6 to 0.8, while the predictive validities of nonverbal abilities for achievement range from r = 0.3 to 0.6 (McCallum et al., 2001; Powers and Barkan, 1986).

Poverty is associated with lower verbal skills and, relatedly, school achievement. Gifted students from low-socioeconomic status (SES) families represent the largest proportion of underrepresented students in gifted programs. Many of the low-SES gifted students enrolled in public schools are not identified (Aguirre, 2003; Barkan and Bernal, 1991). The primary reason for such underrepresentation appears to be low verbal scores. Lower verbal skills are associated with other sociocultural factors such as changing schools, poor physical health, and absent parents who might otherwise influence achievement (Bornstein and Bradley, 2003). Nonverbal tests are often used to ameliorate score differences and increase the diversity of students admitted to gifted programs. However, as children from low-SES households tend to have decreased levels of verbal achievement (Cummins, 1979; Duncan et al., 1998), which may be difficult for students completing rigorous academic coursework required by gifted programs. Few researchers, however, have studied the verbal–nonverbal intelligence score discrepancy in low-income students.

Tyler-Wood and Carri (1993) conducted a study of 80 students from fourth to eighth grades. They compared the scores of low-, average, and above average SES students on the Cognitive Abilities Test, the Stanford-Binet, and Slosson intelligence tests. They found that on all three tests, low-SES students’ verbal scores were significantly lower than those of other groups. Lohman et al. (2008) similarly found that students qualifying for free or reduced-price lunch status scored 6 IQ-like points lower than other students, even when ethnicity and bilingualism were held constant. In a similar study, Carman and Taylor (2010) held ethnicity constant and found scored 8.1 IQ-like points lower in students from poverty.

Even when students from low-SES backgrounds are appropriately identified and placed in gifted education programs, they may not succeed at the same level as their peers. Juntune et al. (2011) gathered 2 years of data from 572 low-SES, mixed-ability students using the RIAS. They found large statistically significant gaps between students’ verbal and nonverbal intelligence scores, with the nonverbal scores being significantly higher. A 23.3 point difference existed between the verbal and nonverbal intelligence scores of the first graders, a 22.1 point difference for second graders, and a 20.2 point difference for third graders. The Cohen’s d effect size ranged from 1.27 to 1.44, representing large effect sizes. These results suggested verbal skills low enough to subsequently affect achievement in advanced academic courses.

Purpose of the study

The purpose of this study was to better understand the differences in verbal and nonverbal intelligence scores of students from poverty. It was hypothesized that students from poverty would demonstrate a gap between their verbal and nonverbal IQ scores. Data analyzed in this study demonstrated how students’ verbal and nonverbal scores differed for each grade. Scores across several elementary-grade levels were also examined for any differences existing in the ability level across the grades. It was hypothesized that the verbal–nonverbal gap would become smaller as the students’ grade levels and associated nonverbal scores increased. Research questions guiding the study were as follows: (1) What is the difference between the verbal and nonverbal intelligence scores of students from poverty? (2) How does this difference differ across the various grade levels?

Method

Participants

The sample consisted of 1935 kindergarten through fifth-grade students from several schools within a single Texas school district. All students in the sample were enrolled in either the free or reduced-price lunch program. Participation in the US lunch program was used as a proxy for living in poverty as it is a well-established metric frequently used by other researchers (i.e. Goddard et al., 2009; Jones and Gansle, 2010; Skiba et al., 2005; Warne et al., 2013). While it is acknowledged that “poverty” entails more than simply participating in a free or reduced-price lunch program, it is believed that this categorization was accurate, given the environmental context of the sample. The school district from which the sample came experiences one of the highest poverty levels in the nation, based on multiple socioeconomic factors including income level, education level, unemployment rate, received health, education, and social services. However, in this study, the phrase “students receiving free or reduced-price lunch” will be used in place of poverty to clarify the measure.

Table 1 displays the demographics of the sample. According to the Texas Education Agency (2013), there were 82.6% Hispanic, 10.6 White, and 4.1% African American students enrolled in this school district during the 2012–2013 academic year. In addition, 69.7% were labeled as economically disadvantaged. The percentage of students enrolled in bilingual or English as a second language program was 21.8%; however, students attending these programs were not included in the current sample, and only students whose primary language was English were included.

Table 1.

Gender, ethnicity, and school status of the participants.

	Kindergarten	Grade 1	Grade 2	Grade 3	Grade 4	Grade 5	Total
Total	138 (7.1%)	472 (24.4%)	327 (16.9%)	418 (21.6%)	293 (15.1%)	287 (14.8%)	1935 (100%)
Gender
Male	79 (57.2%)	233 (49.4%)	152 (46.5%)	214 (51.2%)	139 (47.4%)	139 (48.4%)	956 (49.4%)
Female	59 (42.8%)	239 (50.6%)	175 (53.5%)	204 (48.8%)	154 (52.6%)	147 (51.2%)	978 (50.6%)
Ethnicity
African American	6 (4.3%)	24 (5.1%)	14 (4.3%)	11 (2.6%)	13 (4.4%)	9 (3.1%)	77 (4%)
Hispanic	121 (87.7%)	407 (86.2%)	292 (89.3%)	361 (86.4%)	254 (86.7%)	268 (93.4%)	1703 (88%)
White	10 (7.2%)	35 (7.4%)	19 (5.8%)	41 (9.8%)	25 (8.5%)	8 (2.8%)	138 (7.1%)
Other	0 (0%)	5 (1.1%)	2 (0.6%)	3 (0.7%)	1 (0.3%)	1 (0.3%)	12 (0.6%)

Instruments

The RIAS (Reynolds and Kamphaus, 2003) was used. The RIAS was developed to measure the standardized IQ scores of people aged 3–94. Reynolds and Kamphaus (2003: 13) argue that “the RIAS provides a reliable and valid measurement of g, the general intelligence factor, and its two components, verbal and nonverbal intelligence. It also has correspondence with crystallized and fluid intelligence.” The RIAS produces six subtest scores, three index scores, and a global intelligence score. The Guess What (GWH) and Verbal Reasoning (VRZ) subtests were used to yield the Verbal Intelligence Index (VIX). The GWH measures verbal reasoning in combination with vocabulary and language development. In the GWH, examinees are given a set of two to four clues and asked to deduce the object or concept being described. The VRZ measures verbal–analytical reasoning ability by requiring the examinee to supply one or two words that complete a complex analogy. The Odd-Item Out (OIO) and What’s Missing (WHM) subtests were used to yield the Nonverbal Intelligence Index (NIX). The OIO measures nonverbal reasoning skills and requires both spatial ability and visualization. The WHM requires nonverbal reasoning, where the individual must conceptualize a picture, analyze its gestalt, and deduce what essential element is missing (Reynold and Kamphaus, 2003). The Verbal Memory and Nonverbal Memory subtests were used to yield the Composite Memory Index. The Composite Intelligence Index (CIX) is a global intelligence score derived from the VIX and NIX. Memory subtests were not included in the present study, because research on the factor structure of the RIAS has already indicated that memory subtests do not contribute to the CIX and thus should be separated from the main IQ battery when interpreting intelligence scores (Dombrowski et al., 2009; Nelson et al., 2007; Reynold and Kamphaus, 2003).

Multiple measures were used to establish the reliability of the scores on the RIAS in the original normative sample (i.e. internal consistency, test–retest, and interscorer). Mean reliability coefficients ranged from 0.90 to 0.95 for the scores on the six RIAS subtests and from 0.94 to 0.96 for the scores on the four RIAS indexes.

The test–retest reliability of the four index scores ranged from 0.83 to 0.91. The interscorer reliability of the scores on the six subtests was very high and ranged from 0.95 to 1.00. In the current study, the internal consistency coefficients ranged from 0.83 to 0.90 for the scores on the six RIAS subtests and from 0.85 to 0.91 for the scores on the four RIAS indexes. The validity of scores on the RIAS (utilizing theory-based, logic-based, and empirically based construct validities, factor analysis, and correlations with other measures of intelligence and academic achievement) supported this interpretation of the test scores. The content of the RIAS appeared to be an excellent sampling of items used for the measurement of intelligence and memory; it had high levels of internal consistency for both the individual subtests and the RIAS indexes. A factor analysis of the RIAS scores yielded a two-factor solution involving a verbal and nonverbal domain across the various age, gender, and ethnic groups. The scores on the RIAS indexes correlated between 0.60 and 0.78 with the WISC-III Full Scale IQ and 0.70 and above with the WAIS-III Full Scale IQ. The RIAS VIX scores correlated 0.73 with the Wechsler Individual Achievement Test Composite scores (Reynold and Kamphaus, 2003).

Procedures

The school district administered the RIAS each spring semester in order to identify gifted students. Once a year, for 3 consecutive years, the school district collected RIAS scores on each student. Intelligence scores, along with students’ basic demographic information, were obtained from the school district’s academic service department for the purpose of this analysis.

Data analysis

The data were analyzed using IBM SPSS software. Analyses began with the management of any missing data. Forty students did not complete any or both of the verbal subtests and four students did not complete any or both of the nonverbal subtests. Thus, these scores were treated as missing data. Missing values were replaced according to the series mean method provided by the SPSS software. In response to the first research question, descriptive analyses (including means, standard deviations (SDs), skewness, and kurtosis) were conducted. A paired sample t test and a Pearson product–moment correlation coefficient were calculated to explore the relationship between the students’ verbal and nonverbal intelligence scores.

For the second question, students were divided into three ability groups according to their nonverbal intelligence scores: above average ability, average ability, and below average ability. The above average ability group consisted of students whose scores were equal to or higher than 115, one SD above the mean of 100. A cutoff point of 115 was also used by the school district for gifted identification. The average ability group consisted of students whose scores were between 1 SD above (114) and below (86) the mean of 100. Students whose scores were equal to or lower than 85 constituted the below average ability group. No kindergarten students were represented in the below average ability group as scores from these students had higher nonverbal means.

The verbal and nonverbal means, along with other descriptive statistics, were calculated across the three ability groups and for each grade. In addition, a Pearson product–moment correlation coefficient was computed for each grade to see the relationship between the verbal and nonverbal scores across the different grade levels. Finally, a multivariate analysis of variance (MANOVA) was conducted to compare the verbal and nonverbal scores across all six grades (kindergarten through fifth grade).

Results

The skewness and kurtosis statistics were within the range of ±1, indicating that the score distributions were approximately normal. The verbal and nonverbal mean scores were 89.1 and 109.4, respectively (see Table 2). Results showed that the verbal mean for this study was 10.9 points lower than the mean of the normal distribution of IQ scores. The nonverbal mean for this study was 9.4 points higher than the mean of the normal distribution of IQ scores. Thus, the difference between the verbal and nonverbal means obtained was more than 1 SD.

Table 2.

Descriptive statistics for the verbal and nonverbal intelligence scores.

	N	M	SD	Skewness	Kurtosis
Verbal	1935	89.1	16.6	−0.13	0.61
Nonverbal	1935	109.4	16.9	0.39	−0.07

A paired sample t test was used to compare the verbal and nonverbal scores. As indicated in Table 3, there was a statistically significant difference between the students’ verbal and nonverbal scores (p < 0.001). The standardized effect size index of the verbal and nonverbal means resulted in Cohen’s d = 1.2, indicating a large effect size (Cohen, 1992). A Pearson product–moment correlation coefficient was also computed to examine the relationship between the verbal and nonverbal intelligence scores. There was a positive correlation between the two variables, r(1935) = 0.51, p < 0.01. The coefficient of determination was calculated as r ² = 0.26.

Table 3.

Paired t-test results for the verbal and nonverbal intelligence scores.

		Paired differences				t	df	Significance
		M	SD	95% CI
		M	SD	Lower	Upper
Pair 1	Verbal–Nonverbal	−20.3	16.6	−21.1	−19.6	−54.01	1934	0.000^a

CI: confidence interval.

^a p < 0.05 (two tailed).

Table 4 indicates the verbal and nonverbal mean scores along with the verbal–nonverbal gap for students in each grade and for each ability group. It should be noted that the verbal–nonverbal gaps were relatively small for the below average ability groups in each grade, while the average ability groups had larger gaps than the below average ability groups. However, the verbal–nonverbal gaps for the above average ability groups were the largest across all grade levels. Students with higher nonverbal intelligence scores also had larger discrepancies between their verbal and nonverbal scores, regardless of their grade. In addition, this pattern not only persisted across all six grades but increased as the academic grade increased.

Table 4.

Verbal–nonverbal gaps across ability groups for grades kindergarten through fifth.

Grade	Ability group	Verbal N	Verbal M	SD	Nonverbal N	Nonverbal M	SD	Gap
Kindergarten	Average	38	87.6	15.7	38	100.4	7.9	12.8
	Above	99	101.5	21.2	99	138.6	12.6	37.1
	Total	138	98.0	20.8	137	128.0	20.6	30.0
First	Below	33	69.4	19.7	33	80.9	4.5	11.5
	Average	246	80.9	17.3	246	99.8	8.3	18.9
	Above	191	99.9	16.6	191	129.9	9.8	30.0
	Total	470	88.1	20.0	470	110.7	18.8	22.6
Second	Below	25	76.9	17.7	25	81.5	3.5	4.6
	Average	190	82.0	14.3	190	100.9	7.6	18.9
	Above	111	96.5	13.7	111	126.1	8.4	29.6
	Total	326	86.6	16.0	326	108.0	15.9	21.4
Third	Below	22	73.9	15.9	22	77.1	6.6	3.2
	Average	261	86.9	13.4	261	103.0	7.7	16.1
	Above	135	99.5	13.3	135	123.8	8.5	24.3
	Total	418	90.1	14.9	418	108.4	14.5	18.3
Fourth	Below	15	77.3	14.9	15	81.0	4.6	3.7
	Average	212	87.7	12.8	212	103.0	7.1	15.3
	Above	66	99.5	16.6	66	121.5	5.4	22.0
	Total	293	89.6	14.8	293	106.0	11.7	16.4
Fifth	Below	21	75.5	12.8	21	78.8	5.5	3.3
	Average	202	85.4	11.2	202	101.6	7.3	16.2
	Above	64	96.1	11.2	64	125.3	8.8	29.2
	Total	287	87.0	12.5	287	105.2	14.4	18.2

Students’ predicted verbal scores, standard error of estimates, and upper and lower values of 95% confidence intervals were calculated based on the nonverbal scores using the regression equation of the normative sample (Ŷ = 0.5 × X + 50). Most of the students (79.41%) had verbal scores between the lower and upper confidence interval values. Percentages of verbal scores were lower than the lower confidence interval value and higher than the upper confidence interval values were calculated as 20.54% and 0.05%, respectively. This means that 388 (20.54%) students were observed to have verbal scores significantly lower than the expected verbal scores (based on their nonverbal scores) when compared to the normative sample.

Pearson product–moment correlation coefficients were calculated to investigate the relationship between the verbal and nonverbal intelligence scores for each grade. The correlation coefficients, coefficients of determination, and Cohen’s d values are presented in Table 5. The Cohen’s d values were calculated based on r scores using Friedman’s (1968: 246) formula. All of the correlations were moderate, positive, and statistically significant (p < 0.01). The smallest correlation was found for the fourth graders (r = 0.43), while the highest correlation was found for the first graders (r = 0.56). Consistently, any increase or decrease in the verbal intelligence scores was significantly related to the increase or decrease in the nonverbal intelligence scores and vice versa. An analysis was also conducted to determine if the correlations statistically differed by grade level. The only statistically significant difference was found between the first (r = 0.56) and fourth (r = 0.43) grades (p = 0.02).

Table 5.

Correlation results between the verbal and the nonverbal scores for each grade.

Grade	Verbal N	Nonverbal N	r	r ²	Cohen’s d	r _{norm (6–11 years)}
Kindergarten	138	138	0.47^a	0.22	1.05	0.50
First	472	472	0.56^a	0.31	1.34
Second	327	327	0.49^a	0.24	1.12
Third	418	418	0.52^a	0.27	1.23
Fourth	293	293	0.43^a	0.18	0.95
Fifth	287	287	0.48^a	0.23	1.08

^aCorrelation is significant at the 0.01 level (two tailed).

A separate coefficient of determination was also calculated for each grade. The r ² provided an explanation of how much variance in the verbal scores of each grade could be explained by the nonverbal scores and vice versa. The Cohen’s d was then computed, based on the corresponding r for each grade. Both the r ² and Cohen’s d statistics were found in the same order, from highest to lowest. As with the correlation coefficients, the relationship between the verbal and nonverbal scores of the first-grade students produced the highest effect size, while those of the fourth-grade students produced the lowest effect size. Furthermore, it should be noted that all of the observed effect sizes were high. An effect size of 0.95 was found in Grade 4, while the effect sizes in all other grades were above 1.

In looking at the verbal and nonverbal means across the grade levels, kindergarteners had the highest verbal mean (M = 98, SD = 20.8), while second graders had the lowest verbal mean (M = 86.6, SD = 16). The verbal mean scores fluctuated throughout the six grades. However, there was a continuous decrease in nonverbal mean scores from kindergarten (M = 128, SD = 20.6) through fifth grade (M = 105.2, SD = 14.4).

Finally, MANOVA was conducted to compare the verbal and nonverbal scores as well as to explore the effect of the grade-level verbal and nonverbal mean score changes across the various grades. As shown in Table 6, a statistically significant multivariate main effect was found for grade level, F(10, 3764.000) = 25.785, p < 0.05; Wilks’ λ = 0.876; η ² = 0.064. The observed power to detect the effect was 1.00. A Tukey post hoc test revealed that only kindergarteners’ verbal scores were statistically significantly different from those of students in other grades. As for the nonverbal scores, there were statistically significant differences between kindergarteners and students in all other grades, as well as differences between students in first through fourth grade and those in first through fifth grade.

Table 6.

MANOVA results for the effect of grade on verbal and nonverbal mean scores.

Independent variable	Dependent variable	Wilks’ λ	df	F	Significance	η ²	Noncentrality parameter	Observed power
Grade level	VIX and NIX	0.876	10	25.785	0.000^a	0.064	257.845	1.000

VIX: verbal intelligence index; NIX: nonverbal intelligence index.

^a p < 0.05.

Discussion

The difference between the verbal and nonverbal intelligence scores in this sample was examined by mean comparison and correlation analysis. In general, the nonverbal mean (M = 109.4, SD = 16.9) was 20.3 points higher than the verbal mean (M = 89.1, SD = 16.6). This difference was statistically significant at the p < 0.05 level and produced a large effect size (d = 1.2). These effect sizes are similar to those previously reported by Juntune et al. (2011). The observed verbal–nonverbal gap can be better understood when the statistics for the current sample are compared to the statistics for the RIAS (Reynolds and Kamphaus, 2003) normative sample. The verbal IQ means for the norm samples ranged from 98.8 (SD = 16.4) to 102.1 (SD = 16.9), while the nonverbal IQ means ranged from 100.1 (SD = 17.7) to 101 (SD = 16.2).

In this study, the obtained verbal mean was 89.1, which was 10.9 points lower than the mean of a normal distribution. In addition, the nonverbal mean for this study was 109.4, which was 9.4 points higher than the mean of a normal distribution. The obtained SDs, however, did not differ significantly from the SD values for a normal distribution, and a normality of scores was found in the sample.

There was a statistically significant and positive correlation between the verbal and nonverbal scores, r(1935) = 0.51, p < 0.01. The coefficient of determination was calculated as r ² = 0.26. Reynolds and Kamphaus (2003) reported a correlation of r = 0.50 between the verbal and nonverbal scores of students of age 6 to 11 years, which gave the r ² = 0.25. After considering the discrepancy between the verbal and nonverbal intelligence scores in the current sample, a lower verbal and nonverbal relationship was expected; however, the obtained correlation and coefficient of determination was close to that of the norm sample (r = 0.51 and r ² = 0.26). Although the verbal–nonverbal correlation in this study was moderate and explained the 26% of variance, it should be remembered that the remaining variance had other predictors including factors of biology, parental background (education and SES), and environmental stimulation (Rindermann et al., 2010).

Discrepancies between the verbal and nonverbal intelligence scores were examined for each grade level, from kindergarten through fifth grade. Within each grade, the nonverbal mean scores were significantly higher than the verbal mean scores. The verbal–nonverbal gap was even larger for students with higher nonverbal IQ scores.

When the verbal and the nonverbal IQ means were examined across the grade levels, a continuous decline in the nonverbal mean scores from kindergarten through fifth grade was found. Concurrently, there was a progressive decrease in the verbal mean scores from kindergarten through fifth grade. In sum, all students, regardless of their grade level, demonstrated large gaps between their verbal and nonverbal intelligence scores. In addition, no significant increase or decrease in these score gaps was observed at any grade level represented in the sample.

According to Horn (1980) and Cattell (1971), fluid abilities, such as nonverbal intelligence, develop at an early age, peak after puberty, and gradually decline through adulthood. In contrast, crystallized abilities, such as verbal intelligence, develop out of the Gf factor, along with acquisitions made through acculturation. These abilities continue to develop throughout the individual’s life span and may never reach a maximum peak. Cattell (1971) stated that Gc arises through schooling, experience, and acculturation as students progress through the grade levels. In this study, however, there was no continuous increase in verbal scores; in fact, a progressive decrease was observed.

These results suggest that the verbal abilities of these low-income students did not develop at a rate similar to that of their peers. Jensen (1998) commented that people with higher Gf tend to acquire more Gc from their experiences and opportunities than do those with lower Gf as Gf affects people’s learning experiences throughout their life span. The current study found that students’ verbal and nonverbal scores decreased together, even though their verbal scores were expected to increase.

Implications

Several scholars (e.g., Cattell, 1971; Cattell and Horn, 1978; Reynolds and Kamphaus, 2003) have stressed that nonverbal scores indicate the potential to learn, while verbal scores indicate readiness to learn. Repeatedly, research has shown that verbal abilities, as measured through cognitive tests, are one of the best predictors of academic achievement and school-related performance (Flanagan et al., 1997). Therefore, relying only on nonverbal scores may not be useful for predicting students’ academic achievement, given these demonstrated score discrepancies. In addition, Benbow and Stankey (1996) and Lohman (2005) maintained that measured academic abilities should be matched by the demands that academic programs make on students. If a gifted education program aims to identify those students who are most likely to achieve in the program, they should use verbal or composite scores. If the aim of the gifted education program is to locate students with the greatest potential, examining both verbal and nonverbal scores as well as the gap between the two will provide a valuable diagnostic tool for educators responsible for program design and service implementation.

The context of gifted identification is crucial to understanding the implications of this research. Given the lower verbal abilities of students from poverty, school districts sponsoring gifted programs may seek to identify those students with the highest potential. If so, the gifted education program should focus on curriculum methods and content that nurture students’ verbal abilities; fluid intelligence supports the formation of crystallized intelligence (Cattell, 1971). Accordingly, developing verbal abilities would produce academic achievement. In contrast, if school districts use verbal tests to identify students with the highest verbal scores, the identified students will likely be successful in the gifted programs; however, students from poverty are likely to be underrepresented, given their lower verbal scores. The decision of what test or subtest to use and what program content to prepare should be made in response to the context.

In addition, the abilities of gifted students and the content of gifted programs should be matched to one another. While nonverbal testing might allow a higher number of students from poverty to be placed in gifted programs, such placements may create unintended academic consequences as nonverbal tests do not appear to provide predictive validity for verbal or quantitative achievement (Lakin, 2012). As a result, students identified as gifted through nonverbal intelligence tests alone may be placed at an academic disadvantage when the academic content in the gifted program requires high verbal ability as is often the case. Lohman (2005) and Warne (2009) argued that measuring the construct of intelligence requires measuring both verbal and nonverbal abilities. Findings from this study support such an argument. Assessing both verbal and nonverbal abilities within the context of poverty provides valuable information about the identification and formation of program content.

Appropriately designed individualized programs should be offered to gifted students so they can be successful in gifted programs. In particular, gifted education programs should pay special attention to the development of verbal abilities, especially in the early grades. Intervention programs that build the verbal abilities of children from poverty must be implemented early so that verbal abilities can be developed before placement decisions for advanced programs commence, typically in the fifth grade.

According to Olszewski-Kubilius and Thomson (2010), identifying gifted children at a young age, building programs around the needs of the specific students to be served, and designing programs with multiple support structures for students are all features of successful programs for minority students. In addition, preparatory programs, such as after-school, summer, and weekend courses, can help students entering advanced and accelerated programs (Olszewski-Kubilius and Thomson, 2010). Furthermore, when students with lower verbal scores are placed in gifted programs through nonverbal intelligence scores alone, close tracking of academic performance is recommended. Such tracking will provide information about the students’ academic achievement as well as assist in preventing dropouts from gifted programs.

Limitations and further research

Additional longitudinal studies are needed to track students’ development of verbal and nonverbal abilities from their early years of elementary school onward. In addition, while ethnicity and bilingualism were not directly accounted for in the current study, this sample came from a school district that was predominantly Hispanic. Future studies should examine the variables of ethnicity and bilingualism as these variables often aggregate with poverty (Lohman and Gambrell, 2012). Also, the roles of poverty, ethnicity, and bilingualism in students’ verbal and nonverbal scores should be further investigated. One example would be to examine how verbal scores change from a pretest to a posttest when an intervention program designed to build verbal intelligence is implemented.

Conclusions

Low-income students in this study evidenced large gaps, up to 2 SDs, between their verbal and nonverbal intelligence scores. This finding supports previous results regarding the verbal abilities of children living in economically deprived situations (e.g. Cummins, 1979; Duncan et al., 1998). In addition, these verbal–nonverbal gaps were not specific to any grade or ability level. Rather, all grades (kindergarten through fifth) demonstrated large gaps between their verbal and nonverbal scores. These gaps were even wider for those with higher nonverbal IQ scores. Procedures used to identify gifted students and design gifted programs should consider the verbal–nonverbal score discrepancy often found in high-ability students from low-income households.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Aguirre

(2003) ESL students in gifted education. In: Castellano

(ed) Special Populations in Gifted Education. Boston: Allyn and Bacon, pp. 17–27.

Baer

(1980) Programs for the gifted: a present or a paradox? Phi Delta Kappan 61: 621–623.

Barkan

Bernal

(1991) Gifted education for bilingual and limited English proficient students. Gifted Child Quarterly 35(3): 144–147. DOI: 10.1177/001698629103500306.

Benbow

Stankey

(1996) Inequity in equity: how “equity” can lead to inequity for high-potential students. Psychology, Public Policy, and Law 2(2): 249–292. DOI: 10.1037//1076-8971.2.2.249.

Blair

(2006) How similar are fluid cognition and general intelligence? A developmental neuroscience perspective on fluid cognition as an aspect of human cognitive ability. Behavioral and Brain Sciences 29: 109–160. DOI: 10.1017/S0140525X06009034.

Borland

Wright

(1994) Identifying young, potentially gifted, economically disadvantaged students. Gifted Child Quarterly 38: 164–171.

Bornstein

Bradley

(2003) Socioeconomic Status, Parenting, and Child Development. Mahwah: Lawrence Erlbaum.

Braden

(2000) Perspectives on the nonverbal assessment of intelligence. Journal of Psychoeducational Assessment 18: 204–210. DOI: 10.1177/073428290001800301.

Callahan

(2001) Fourth down and inches. Journal of Secondary Gifted Education 12: 148–156.

10.

Carman

Taylor

(2010) Socioeconomic status effects on using the Naglieri Nonverbal Ability Test (NNAT) to identify the gifted/talented. Gifted Child Quarterly 54: 75–84. DOI: 10.1177/0016986209355976.

11.

Cattell

(1971) Abilities: Their Structure, Growth, and Action. Boston: Houghton-Mifflin.

12.

Cattell

Horn

(1978) A check on the theory of fluid and crystallized intelligence with description of new subtest designs. Journal of Educational Measurement 15: 139–164. DOI: 10.1111/j.1745-3984.1978.tb00065.x.

13.

Cohen

(1992) A power primer. Psychological Bulletin 112: 155–159. DOI: 10.1037//0033-2909.112.1.155.

14.

Cummins

(1979) Linguistic interdependence and the educational development of bilingual children. Review of Educational Research 49(2): 222–251. DOI: 10.2307/1169960.

15.

Dombrowski

Watkins

Brogan

(2009) An exploratory investigation of the factor structure of the Reynolds Intellectual Assessment Scales (RIAS). Journal of Psychoeducational Assessment 27: 494–507.

16.

Duncan

Yeung

Brooks-Gunn

. (1998) How much does childhood poverty affect the life chances of children? American Sociological Review 63: 406–423. DOI: 10.2307/2657556.

17.

Flanagan

Andrews

Genshaft

(1997) The functional utility of intelligence tests with special education populations. In: Flanagan

Genshaft

Harrison

(eds) Contemporary Intellectual Assessment: Theories, Tests, and Issues. New York: Guilford, pp. 457–483.

18.

Friedman

(1968) Magnitude of experimental effect and a table for its rapid estimation. Psychological Bulletin 70: 245–251. DOI: 10.1037/h0026258.

19.

Goddard

Salloum

Berebitsky

(2009) Trust as a mediator of the relationships between poverty, racial composition, and academic achievement: evidence from Michigan’s public elementary schools. Educational Administration Quarterly 45(2): 292–311.

20.

Gustafsson

Undheim

(1996) Individual differences in cognitive functions. In: Berliner

Calfee

(eds) Handbook of Educational Psychology. New York: Simon and Schuster Macmillan, pp. 186–242.

21.

Horn

(1980) Concepts of intellect in relation to learning and adult development. Intelligence 4: 285–317. DOI: 10.1016/0160-2896(80)90025-2.

22.

Horn

Cattell

(1966) Refinement and test of the theory of fluid and crystallized general intelligence. Journal of Educational Psychology 57: 253–270. DOI: 10.1037/h0023816.

23.

Jensen

(1998) The g Factor: The Science of Mental Ability. Westport: Praeger.

24.

Jones

Gansle

(2010) The effects of a mini-conference, socioeconomic status, and parent education on perceived and actual parent participation in individual education program meetings. Research in the Schools 17: 23–38.

25.

Juntune

Kaya

Ramos

(2011) Another look at high ability students from low SES populations. Tempo 32(4): 14–19.

26.

Kvist

Gustafsson

(2008) The relation between fluid intelligence and the general factor as a function of cultural background: a test of Cattell’s investment theory. Intelligence 36: 422–436.

27.

Lakin

(2012) Assessing the cognitive abilities of culturally and linguistically diverse students: predictive validity of verbal, quantitative, and nonverbal tests. Psychology in the Schools 49(8): 756–768.

28.

Lakin

Lohman

(2011) The predictive accuracy of verbal, quantitative, and nonverbal reasoning tests: consequences for talent identification and program diversity. Journal for the Education of the Gifted 34(4): 595–623. DOI: 10.1177/016235321103400404.

29.

Lewis

DeCamp-Fritson

Ramage

. (2007) Selecting for ethnically diverse children who may be gifted using Raven’s standard progressive matrices and Naglieri nonverbal abilities test. Multicultural Education 15(1): 38–42.

30.

Lohman

(2005) The role of nonverbal ability tests in identifying academically gifted students: an aptitude perspective. Gifted Child Quarterly 49: 111–138. DOI: 10.1177/001698620504900203.

31.

Lohman

Gambrell

(2012) Using nonverbal tests to help identify academically talented children. Journal of Psychoeducational Assessment 30(1): 25–44. DOI: 10.1177/0734282911428194.

32.

Lohman

Korb

Lakin

(2008) Identifying academically gifted English language learners using nonverbal tests: a comparison of the Raven, NNAT, and CogAT. Gifted Child Quarterly 52: 275–296.

33.

McCallum

Bracken

Wasserman

(2001) Essentials of Nonverbal Assessment. Hoboken: Wiley.

34.

McGrew

(2009) CHC theory and the human cognitive abilities project: standing on the shoulders of the giants of psychometric intelligence research. Intelligence 37: 1–10. DOI: 10.1016/j.intell.2008.08.004.

35.

McGrew

Evans

(2004) Internal and External Factorial Extensions to the Cattell–Horn–Carroll (CHC) Theory of Cognitive Abilities: A Review of Factor Analytic Research Since Carroll’s Seminal 1993 Treatises. St. Cloud: Institute for Applied Psychometrics.

36.

Nelson

Canivez

Lindstrom

. (2007) Higher-order exploratory factor analysis of the Reynolds Intellectual Assessment Scales with a referred sample. Journal of School Psychology 45: 439–456.

37.

Olszewski-Kubilius

Thomson

(2010) Gifted programming for poor or minority urban students: issues and lessons learned. Gifted Child Today 33(4): 58–64.

38.

Powers

Barkan

(1986) Concurrent validity of the Standard Progressive Matrices for Hispanic and non-Hispanic seventh-grade students. Psychology in the Schools 23: 333–336.

39.

Renzulli

(1982) Myth: the gifted constitute 3-5% of the population. Gifted Child Quarterly 26: 11–14.

40.

Reynolds

Kamphaus

(2003) Reynolds Intellectual Assessment Scales and the Reynolds Intellectual Screening Test Professional Manual. Lutz: Psychological Assessment Resources.

41.

Rindermann

Flores-Mendoza

Mansur-Alves

(2010) Reciprocal effects between fluid and crystallized intelligence and their dependence on parents’ socioeconomic status and education. Learning and Individual Differences 20: 544–548.

42.

Saccuzzo

Johnson

Russell

(1992). Verbal versus performance IQs for gifted African-American, Caucasian, Filipino, and Hispanic children. Psychological Assessment 4: 239–244. DOI: 10.1037//1040-3590.4.2.239.

43.

Skiba

Poloni-Staudinger

Simmons

. (2005) Unproven links: can poverty explain ethnic disproportionality in special education? The Journal of Special Education 39: 130–144.

44.

Tyler-Wood

Carri

(1993) Verbal measures of cognitive ability: the gifted low SES student’s albatross. Roeper Review 16(2): 102–105. DOI: 10.1080/02783199309553550.

45.

Warne

(2009) Comparing tests used to identify ethnically diverse gifted children: a critical response to Lewis, DeCamp-Fritson, Ramage, McFarland, and Archwamety. Multicultural Education 17(1): 48–53.

46.

Warne

Anderson

Johnson

(2013) The impact of race and ethnicity on the identification process for giftedness in Utah. Journal for the Education of the Gifted 36(4): 487–508. DOI: 10.1177/062353213506065.

47.

Watkins

Lei

Canivez

(2007) Psychometric intelligence and achievement: a cross–lagged panel analysis. Intelligence 35: 59–68. DOI: 10.1016/j.intell.2006.04.005.

48.

Willis

Dumont

Kaufman

(2011) Factor–analytic models of intelligence. In: Sternberg

Kaufman

(eds) The Cambridge Handbook of Intelligence. New York: Cambridge University Press, pp. 39–57.