Info

Age Group

Figure 5.13

Mean K-BIT Vocabulary (Gc) and Matrices (Gf) standard scores (based on norms for ages 25-34), adjusted for education, for eight groups of adults between the ages of 20 and 90 years.

Source: Reprinted from Wang and Kaufman (1993, Figure 1) with permission.

with educational attainment (self-education) and age as predictors of each of the seven subtests. The results of these regression analyses (based on N = 860 adults ages 25-94) permitted computation of age-education gradients, similar to the gradients reported by Heaton et al. (1986), shown in Table 5.7 for Wechsler's adult scales.

KEY FINDINGS. The measures of Gc and Gq displayed maintained age-by-age patterns (see Figure 5.14), consistent with Horn's (1989; Horn & Hofer, 1992) predictions and with results on Wechsler's V-IQ, VCI, and Arithmetic subtest. In contrast, but in keeping with Horn's predictions, the tests that measure Gf and Gv demonstrated quite vulnerable aging patterns. Though the Gv subtest (K-SNAP Gestalt Closure) is a Wechsler-

like measure of Perceptual Organization, most resembling Picture Completion, the Gf subtest (K-SNAP Four-Letter Words) is totally different from Wechsler's Performance subtests. This novel task requires no visual-spatial ability, but, instead, measures fluid reasoning ability with semantic stimuli. Note in Figure 5.14 that the graph for the Gf subtest has a plateau during the middle of the adult life span, similar to KAIT Fluid IQ and K-BIT Matrices, but different from the steady decline (after age 34) of the Kaufman measure of Gv and the steady decline of Wech-sler's Performance subtests.

On the maintained abilities, Gc and Gq, Tukey's HSD test revealed that mean scores increased significantly from the adolescent years (ages 15-19) to the young adult groups. Then,

Age Group

Figure 5.14

Mean Gc (K-FAST Reading), Gf (K-SNAP Four-Letter Words), Gq (K-FAST Arithmetic), and Gv (K-SNAP Gestalt Closure) standard scores (based on norms for ages 15-94 years), adjusted for education, for 14 groups of adults between the ages of 15 and 94 years.

Note Gc = (K-FAST Reading); Gq = (K-FAST Arithmetic); Gf = (K-SNAP Four-Letter Words); Gv = (K-SNAP Gestalt Closure).

Source: Graphs developed from data presented by Kaufman et al. (1996, Tables 4, 5, and 7) with permission.

abilities were maintained through the late 60s before scores declined in the 70s, notably at age 75 and older. Peak performance was at ages 40-44 on both Gc and Gq. On the vulnerable abilities of Gf and Gv, adults ages 50-54 scored nearly 0.5 SD (6-7 IQ points) higher than adults ages 60-64 and about 1.0 SD (14-17 IQ points) higher than the oldest sample (ages 75-94). Peak performance was at ages 20-24 on Gf and at ages 17-19 on Gq.

Horn and Hofer (1992) classified Glr (also called TSR, long-term storage and retrieval) as a maintained ability, as opposed to short-term memory, which they consider vulnerable: "several studies showing that in the same samples in which Gsm declines with age in adulthood, Glr does not decline and, in some samples, increases"

(p. 69). However, Horn and his colleagues have typically studied Glr with tasks that involve words and verbal learning. The KAIT measures of Glr include one task (Auditory Delayed Recall) that resembles Horn's tasks and one that does not (Rebus Delayed Recall). Figure 5.15 displays age-by-age education-adjusted means for these two measures of long-term retrieval and the graphs are quite different from each other. The Auditory task displayed a maintained pattern across the life span, but the Rebus task revealed a vulnerable pattern. Both tasks measure recall of information learned about 30-45 minutes earlier in the KAIT administration. The Delayed Auditory subtest measures retention of information presented as part of a mock news

Age Group

Figure 5.15

Mean standard scores on two measures of Glr (KAIT Rebus Delayed Recall and KAIT Auditory Delayed Recall), based on norms for ages 15-94 and adjusted for education, for 14 groups of adults between the ages of 15 and 94 years.

Source: Reprinted from Kaufman et al. (1996, Figure 1) with permission.

broadcast in the Crystallized Auditory Comprehension subtest; the Delayed Rebus task assesses how well individuals remember the words that are paired with unfamiliar symbols (rebuses) in the Fluid learning task known as Rebus Learning. Even though the initial subtests are administered a few minutes apart, as are the two delayed-recall tasks, how much a person remembers from each initial subtest is quite dependent on his or her age.

As Figure 5.15 indicates, adolescents ages 15-16 earned mean scores that were about 9 points higher on Rebus Delayed Recall than on Auditory Delayed Recall; in contrast, elderly adults (ages 65 and above) scored about 3 points lower on the Rebus subtest than on the Auditory subtest. Horn (personal communication, September 1994) con sidered the different patterns for the two Glr tasks suggestive of "intriguing hypotheses" about the storage, consolidation, and retrieval of information. In the Kaufman, Kaufman, Chen et al. (1996) study, the aging pattern for long-term retrieval tasks depended on the material to be recalled: When the stimuli were learned during a Gc task, the amount of retention displayed a Gc-like maintained pattern; when the stimuli were learned during a Gf task, the pattern of retention was vulnerable to the effects of aging. Data for the WJ III Glr scale, which measures long-term retrieval with tasks that resemble KAIT's Fluid subtest (Rebus Learning), are consistent with Kaufman et al.'s finding: This WJ III scale demonstrates a vulnerable pattern (see Chapter 14), although the data are tentative because education was not controlled. The results with the KAIT and WJ III require cross-validation, especially with longer delays between initial learning and later recall. Though the KAIT interval of 30 to 45 minutes clearly qualifies as measuring Glr and not Gsm from a Horn standpoint, it is nonetheless true that Horn and colleagues (e.g., Horn & Hofer, 1992; Horn & Risberg, 1989) have been primarily interested in storage for hours, weeks, or years when investigating and theorizing about Glr. Furthermore, the use of Horn's terminology to define memory tasks is one of many theoretical approaches that can be taken. For example, in the next section, the aging patterns observed for WMS-III scales are interpreted within an episodic versus semantic memory context (Tulving, 1983), and there are numerous other ways of categorizing distinctions among memory tasks, and of interpreting changes due to aging from cognitive and neuropsychological perspectives (Craik & Salthouse, 2000).

Age-education gradients were computed for each of the seven subtests investigated by Kaufman, Kaufman, Chen et al. (1996) based on a multiple regression analysis conducted for ages 25-94 years. Table 5.10 presents these gradients, listed from the most age-related to the most education-related subtest. The measure of Gv was easily the most related to age, followed by the measure of Gf, whereas Gq and Gc were most related to education. These results are quite similar to the findings for the WAIS, WAIS-R, and WAIS-III subtests (see Table 5.7). The age-education gradients show again how different the two Glr tasks were, with the gradient for the Rebus task fairly similar to the gradient for the Gf measure and the gradient for the Auditory task close to the value for the Gc test.

TABLE 5.10 Age-education gradients on Kaufman and Kaufman subtests that measure different Horn abilities

Horn Ability (Kaufman Subtest)

% Age Variance

% Education Variance

Age-

Education Gradient

Gv—Broad visualization (K-SNAP Gestalt Closure)

Gf-—Fluid reasoning (K-SNAP Four-Letter Words) Glr—Long-term storage & retrieval (KAIT Rebus Delayed Recall)

Gsm—Short-term acquisition & retrieval (K-SNAP Number Recall)

Glr—Long-term storage & retrieval (KAIT Auditory Delayed Recall) Gc—Crystallized knowledge (K-FAST Reading)

Gq—Quantitative thinking (K-FAST Arithmetic)

26.4 21.9 +4.5 18.0 22.2 -4.2 13.0 20.0 -7.0 13.0 31.8 -18.8

Note: Age-education gradient equals age variance minus education variance. Positive values denote the more age-related subtests, while negative values indicate the tasks more heavily dependent on education. Subtests are listed in order of gradients for the seven Kaufman and Kaufman subtests.

Heaton and Colleagues' Cross-Sectional Investigation of Wechsler Memory Scale—III

In addition to analyzing education-controlled aging data for the WAIS-III, Heaton and colleagues (Heaton et al., 2001; Manly et al., 2000) conducted similar analyses for the major memory indexes yielded by the WMS-III. In each WMS-III study, data were analyzed for ages 20 to 89 years, although sample sizes, composed of standardization cases and additional "education oversampling" cases differed in the two reports: N = 885 for Heaton et al. (2001) and N = 1,089 for Manly et al. (2000). To permit age-by-age comparisons, all standard scores were based on norms for ages 20-34 years and statistically corrected for education.

KEY Findings. Heaton et al. (2001) compared z scores for the age group earning the lowest education-corrected mean index (ages 85-89 in each case) and the highest mean index (usually ages 2024), as well as the difference between these indexes in SD units. The fact that the peak age is 20-24 years for six of the indexes (Auditory Delayed Recognition and Working Memory were exceptions) indicates the vulnerability of these abilities to the normal aging process. The differences between high and low means in SD units, which exceeded 1.5 SD for five indexes and ranged from 1.20 to 1.76 SD, further illustrates the vulnerability of both the immediate and delayed recall scales on the WMS-III, although the three measures of delayed recall were clearly the most vulnerable: The largest effect sizes (about 1.6-1.7 SD) were obtained for General Memory, Auditory Delayed, and Visual Delayed. Compare the WMS-III effect sizes with the values of the WAIS-III indexes, also reported by Heaton et al. (2001) using analogous techniques: PSI (1.9), POI (1.5), WMI (1.1), and VCI (0.6). The values for most WMS-III indexes are commensurate with the value for POI.

The delayed recall indexes, both auditory and visual, are both quite vulnerable to the effects of age. In contrast, the auditory delayed recognition index is a more maintained ability, as depicted in graphs presented by Manly et al. (2000). The auditory delayed recall and delayed recognition tasks are a combination of performance on delayed memory of a story (which has a Gc component) and delayed recall of paired associates (which is a learning task with a Gf component). The visual delayed recall test is a combination of memory of faces and memory of family pictures, both with a substantial Gv component, resembling Gv subtests on the Woodcock-Johnson Tests of Cognitive Ability, both the WJ-R and WJ III (see Chapter 14). The immediate and delayed conditions are separated by about 25 to 35 minutes, similar to the interval for the KAIT Delayed Recall subtests.

The findings for the WMS-III reinforce the findings for the KAIT delayed memory subtests and the WJ III Glr scale (see Chapter 14). Contrary to Horn's prediction that long-term retrieval is unilaterally a maintained ability, the nature of the material to be recalled is instrumental in determining the shape of the aging curve. When the initial learning involves a vulnerable ability like Gf or Gv, then the long-term retrieval likewise displays a vulnerable pattern.

A comparison of the aging patterns for the highly vulnerable auditory delayed recall index versus the moderately maintained auditory delayed recognition index also suggests an additional amendment to Horn's predictions: Even when the content of the tasks is held constant, different aging patterns may emerge whether adults are asked to recall the material or recognize it.

Finally, Heaton et al. (2001) data indicate that the WMS-III immediate memory indexes generally have substantially larger high-low differences in SD units (1.3-1.5) than the WAIS-III WMI (1.1). The mildly vulnerable pattern found for the WAIS-III WMI, akin to Horn's Gsm, is a blend of three distinctly different patterns for each of the three component subtests (as mentioned earlier and depicted in Figure 5.9). The subtest that most captures the essence of Horn's Gsm is WAIS-III Digit Span. None of the WMS-III subtests is very similar to the basically simple Digit Span. The auditory paired associates task, for example, requires learning ability and Gf, and is not really an immediate recall task in the true sense of the term. Other WMS-III subtests, as indicated, have components of Gc and Gv and are far more complex than Digit Span. Interestingly, both the WAIS-III and WMS-III yield Working Memory Indexes, yet they are quite different from each other even though both include Letter-Number Sequencing. On the WAIS-III, the WMI reflects a blend of skills, whereas the two-subtest Index of the same name on the WMS-III includes Spatial Span— which, like Letter-Number Sequencing, measures Gv as well as Gsm—leading to the extreme vulnerability of the WMS-III Working Memory Index.

The WMS-III results are consistent with the literature on memory and aging, which is often interpreted within the context of Tulving's (1983) distinction between episodic and semantic memory. Episodic memory refers to personally experienced events or episodes and is assessed experimentally with immediate or delayed recall of word lists, geometric designs, text, faces, and so forth (as well as with more personally oriented tasks requiring individuals to recall things that occurred to them at a particular time or within a specific context). In contrast, semantic memory reflects general world knowledge and is often assessed with tests of information, naming ability, or vocabulary. All of the WMS-III tasks and scales fit into the category of episodic memory. The vulnerability of the WMS-III scales is consistent with the burgeoning literature on age changes in episodic memory that usually reports notable declines with aging on immediate recall, delayed recall, and delayed recognition of a variety of verbal and nonverbal stimuli (e.g., Korten et al., 1997; Souchay, Isingrini, & Espagnet, 2000). Even an elite sample of elderly professors at Berkeley (ages 60-71) performed much more poorly than middle-aged professors (ages 45-59) and young professors (ages 30-44) on a verbal paired-associate learning task (Shimamura, Berry, Mangels, Rusting, & Jurica, 1995). In contrast, semantic memory displays a maintained pattern with little variability across most of the adult life span before declining systematically and gradually in very old age (Backman, Small, Wahlin, & Larsson, 2000).

In general, the experimental psychology literature on memory and aging mimics the findings of the studies of the Wechsler and Kaufman tests: Episodic memory—like fluid intelligence, processing speed, visualization, and the memory abilities assessed by the WMS-III and KAIT—is quite vulnerable to the effects of aging, in contrast to semantic memory and the related construct of crystallized intelligence, both of which are maintained abilities that do not decline appreciably until old age. There also seems to be a maintained aging pattern for tasks that might technically fall within the episodic-memory domain, but have a clear-cut Gc component such as recalling the main facts in a mock news broadcast (KAIT Auditory Comprehension and Delayed Auditory Recall) or repeating prose passages that contain factual content. Both of the KAIT memory tasks displayed maintained aging patterns (Kaufman & Horn, 1996; Kaufman et al., 1996). Also, the elderly Berkeley professors who performed so poorly on a paired-associate memory task performed as well as young and middle-aged professors when repeating prose passages about a woman who was robbed (WMS-R Logical Memory), the elements that make up the earth's atmosphere, and the tribal cultures in the Missis-sippian period (Shimamura et al., 1995).

Overview of Cross-Sectional Investigations

The Wechsler adult scales have been in use for more than 60 years, and data on several versions of these scales have been analyzed cross-sectionally, with controls for cohort differences in education, for several generations. Though occasionally the results of an investigation have led to conclusions that suggested little or no decline in P-IQ with advancing age (e.g., Green's, 1969, study of Puerto Rican adults with very limited formal education), the preponderance of evidence accumulated over time has supported a steady and sometimes dramatic decline in P-IQ as individuals age from adolescence to old age. This decline has been accompanied by maintenance of V-IQ through middle age and occasionally the 60s, before a notable decrease in verbal ability as adults reach their seventh and eighth decades of life.

These aging patterns have been interpreted from a Horn-Cattell (Horn & Cattell, 1966, 1967) standpoint as denoting the vulnerability of fluid intelligence or Gf in the face of maintenance of crystallized intelligence or Gc. Some researchers, notably Woodcock (1990) and Flanagan and McGrew (1997), insist that Wechsler's P-IQ denotes only Gv, with virtually no Gf at all, but that position is arguable (Kaufman, 1994a, 2000b). Horn believes that Wechsler's Performance subtests (except for highly speeded tasks like Digit Symbol-Coding) measure a blend of Gf and Gv (Horn & Hofer, 1992), and that interpretation is consistent both with the accumulated research and with examiners' clinical observations of the clear-cut problem-solving components of tasks like Block Design and Picture Arrangement. Yet, the tasks, including WAIS-III Matrix Reasoning, definitely involve visual-spatial ability also. P-IQ and POI seem to be dependent on both Gf and Gv, and untangling them seems futile.

Data from the Kaufman and Kaufman tests clarify the issue to some extent. The KAIT includes measures of Gf, notably Mystery Codes and Logical Steps, that emphasize reasoning ability without stressing visualization. K-SNAP includes one measure of Gf (Four-Letter Words) that has no discernible Gv component at all and includes one measure of Gv (Gestalt Closure) that apparently requires no Gf. All of these Kaufman subtests demonstrated extremely vulnerable patterns for measures that are primarily Gf and for measures that are primarily Gv; these findings are also reinforced by the growth curves for WJ III cognitive scales, although those curves were not adjusted for educational attainment (see Chapter 14, especially Figures 14.1-14.3). The WMS-III immediate and delayed scales, which include subtests that are dependent on Gv, also evidenced extreme vulnerability to aging. Therefore, the best conclusion is that the aging declines observed for WAIS, WAIS-R, and WAIS-III P-IQ—and for WAIS-III POI— reflect the vulnerability of both Gf and Gv from Horn's theoretical perspective. Even more dramatically, a decline occurs for Gs or broad speed-iness. From Baltes's theory, the "mechanics" component is vulnerable to the effects of normal aging and subsumes all of these Horn abilities: reasoning, spatial orientation, and perceptual speed (Baltes, 1997; Baltes, Staudinger, & Lin-denberger, 1999).

Taken together, the data from the Kaufman tests (KAIT, K-BIT, K-SNAP, K-FAST) and the Wechsler adult scales offer broad-based support for the increase and then maintenance of Gc ("pragmatics" to Baltes) across much of the life span before notable declines occur during the 70s and 80s. At the same time, this accumulation of data from the Wechsler and Kaufman tests offers equally pervasive support for the peaking of Gf and Gv abilities in early adulthood (usually 20-24 years) followed by declines that continue throughout adulthood and old age. The findings of maintained Gc abilities and vulnerable Gf abilities—including the approximate magnitude of the declines in Gf ability with increasing age, and the ultimate decline in Gc in very old age— are also consistent with age-related changes in intelligence observed in well-designed, large-scale, cross-sectional studies conducted in Europe (i.e., Rabbitt's, 1993, investigation of more than 6,000 adults ages 50-96 in the United Kingdom, and Baltes & Lindenberger's, 1997, study of 687 adults ages 25-103 in Germany). Data collected prior to 1980 suggest that Gc, once adjusted for educational differences among age groups, peaks in the 60s. However, the more recent data integrated in this chapter from the Kaufman tests (late 1980s and early 1990s) and WAIS-III (mid-1990s) indicate an earlier peak in the 40s. Why this generational shift has occurred is not clear.

The overall findings for IQ versus memory variables, regarding patterns of maintenance and vulnerability, can be interpreted jointly. As noted, Gc and semantic memory tasks display maintained aging patterns across most of the life span (with both declining in old age), while Gf, Gv, and episodic memory tasks are extremely vulnerable to the normal aging process. "There are interesting parallels with regard to the cognitive processes involved during task performance, with semantic memory and crystallized intelligence drawing largely on prior knowledge and episodic memory and fluid intelligence requiring new learning and flexible adjustments to new situational demands" (Backman, Small, Wahlin, & Larsson, 2000, p. 503). In addition, tasks akin to episodic memory are included—along with tests of fluid reasoning, processing speed, and visualization—as measures of mechanics from Baltes's pragmatics-mechanics dichotomy (Lindenberger & Baltes,

1997). The evidence is conflicting on whether the parallel declines in old age on Gf and episodic memory are due to the same cause or set of causes. There is some evidence from a study by Isingrini and Taconnat (1998) with 318 adults (aged 20-40 and 60-85) that the simultaneous declines on Gf and episodic memory tasks are two fairly separate phenomena. In addition, data from a longitudinal study of 387 healthy old people (ages 70-88 years) who were retested four years later revealed differences in prediction from time 1 to time 2 for measures of fluid ability (matrices) versus episodic memory (Wechsler's Logical Memory) (Deary, Starr, & MacLennan,

1998). Demographic variables, blood pressure, and measures of premorbid IQ accounted for nearly 40% of the reliable variance in fluid intelligence compared to 12% of the memory variance; also, blood pressure at the initial testing was related to subsequent fluid ability but not to episodic memory differences (Deary et al., 1998). In contrast to the studies that suggest separate explanations for the vulnerability of Gf and episodic memory is a series of studies with adults in their 70s and 80s by Backman, Hill, and their colleagues (e.g., Backman et al., 1998) that report substantial correlations between measures of Gf and both verbal and nonverbal episodic re call (though the heavy speed component of the Gf tasks in the Backman-Hill studies clouds the nature of the relationship to some extent).

Cautions Associated with Cross-Sectional Investigations

Despite the careful experimental designs of the cross-sectional investigations conducted on Wech-sler's and Kaufman's adult scales, this type of study has a few built-in problems that must be considered, namely issues concerned with (1) equating on educational attainment, (2) cohort and time-of-measurement effects, and (3) internal and external validity.

Equating on Educational Attainment

All of the results discussed and integrated in this section are based on the interpretation of cross-sectional data that were matched or otherwise equated on adults' educational attainment. Mat-arazzo (1972) wondered whether years of formal education is "a variable with identical meaning across generations" (p. 115), and his concern has merit (Kaufman, 2001). Equating groups that differ substantially in age on educational attainment is an inexact science and must be considered as approximate correction for a changing society's inequalities. For example, schooling beyond high school, commonplace now, was enjoyed primarily by the elite in the 1950s. Indeed, about 50% of each age group between 20-24 and 45-54 had at least one year of college (see Table 5.2), making post-high school education "average" for young and middle-age adults in the WAIS-III sample. In contrast, only about 15-20% of adults ages 20-54 in the early 1950s (when the WAIS was standardized) had some college, as shown in Table 5.2. The meaning of "attended college" or "graduated college," therefore, is not a constant across generations; analogously "high school dropout" has a far greater stigma for younger than older adults in the 2000s than a half-century earlier.

"Years of formal schooling" is clearly not a perfect yardstick. Nonetheless, despite logical arguments to the contrary, there is some empirical evidence that this term may have a fairly constant meaning across the adult age range. Consider the provocative data in Table 5.4, which show that, regardless of age, and with only mild aberrations, individuals with a comparable amount of education earned similar scores on the WAIS-R Verbal Scale. Certainly Verbal skills are closely related to formal education; other things being equal, the greater the years of schooling, the greater the success on tests of general information, word meaning, and arithmetic ability. As shown in Table 5.4, mean Verbal sums of scaled scores for those with 0-8 years of schooling averaged about 40 regardless of age, and similar consistency across age was obtained for other educational categories as well. Thus, Verbal IQ, long known to be a maintained ability across much of the adult age range (Horn & Noll, 1997), was maintained within each of five different levels of educational attainment. Although this finding does not trivialize the concern of the inequality of the educational attainment yardstick, it does provide some empirical support for statistically controlling for education in the various cross-sectional investigations.

Cohort and

Time-of-Measurement Effects

Regardless of consistencies across studies, instruments, and generations, inferences from cross-sectional studies about developmental (on-togenetic) changes in intelligence are speculative at best. As long as different individuals compose the separate age samples, one can only guess at the nature of the age-related changes in intelligence in the same individuals over time. When education level is controlled, one aspect of cohort differences is eliminated to some extent. However, numerous other nonage and nonedu-cation variables associated with growing up at a given period of time are either unknown, unmeasured, or unquantifiable. Yet such variables as motivation level, historical events, social customs and mores, the availability of television and personal computers, child-rearing techniques, nutrition, the quality and extent of prenatal care and knowledge, and the impact of mass media will affect apparent age-related changes in scores on mental tests.

In addition, time-of-measurement effects interact with performance on intelligence tests. Real changes either in mental ability or in test-taking ability could affect how every group of adults (regardless of cohort) performs on a given test. These sweeping cultural changes could affect individuals aging from 25 to 35 in much the same way that they affect others who age from 40 to 50 during the same time frame. For example, in the 1920s, tests were uncommon for everyone, and scores would likely be relatively low for a person of 20 or 40 or 60 tested on unfamiliar items like verbal or figure analogies; people of the same ages tested in the 1960s or 1970s would likely score relatively higher on these same tests because such tests had become a familiar part of U.S. culture. This type of control for cultural change was used by Owens (1966) in his landmark longitudinal study (discussed later in this chapter).

Not all cultural changes relate to test-taking ability, however, as Flynn (1984, 1987) has made abundantly clear (see Chapter 2). Indeed, Flynn has probably come as close as anyone to quantifying these cultural or time-of-measurement effects by using cross-sectional data to show systematic IQ gains across generations. That these gains differ dramatically from country to country stresses their cultural-environmental origin. Because differences in IQs earned in different eras by individuals of the same age reflect both time-of-measurement and cohort effects, Kausler (1982, 1991) prefers to use the term time lag to denote these changes in intelligence scores.

Internal and External Validity

By controlling for education level in various cross-sectional studies, the investigators have conducted studies high in internal validity, permitting both the identification of causative factors and the generalization of these causative factors to other samples (Kausler, 1982, 1991).

Thus, apparent age-related declines in verbal intelligence may be attributed to educational attainment; declines with age in mean Gs, Gf, Gv, and Gsm scores are due partly to education, but mostly to age differences plus an unknown proportion of cohort variation. The downside of the high internal validity of the Wechsler and Kaufman age-education studies is low external validity, meaning poor generalization of the "adjusted" age differences to the population at large; in fact, in the real world, older individuals are less well educated than younger adults. Consequently, the actual, unadjusted values come closer to describing true differences in the mean scores of different age groups. With the WAIS-III, though, even the unadjusted values may not validly describe true differences in the population at large in view of the unusual number of exclusionary criteria applied to the selection of the standardization sample. However, unadjusted values cannot be used to infer causality of the differences, and they have limited value for implying developmental change. But education-balanced groups, according to Kausler (1982), "give a truer picture of on-togenetic change than our previous contrasts between educationally imbalanced age groups that were, nevertheless, representative of their respective populations" (p. 67). Because of the very nature of the limitations of cross-sectional research, it is essential that any conclusions about aging and IQ be buttressed by the results of longitudinal research.

Brain Research And Your Child

Brain Research And Your Child

Enchanted Learning Experiences -Why They Should Be The Norm For Our Children. The latter part of the twentieth century has seen more discoveries about the human brain than in all previous history of mankind. It is as though we have been paddling in the shallows of a vast ocean hitherto unaware of its existence.

Get My Free Ebook


Post a comment