Eric,
There is a nice overview of IQ testing and culture in the book "development of
children" by Cole and Cole. Just in case there is not retroactive permission
of the author (hopefully there will -- this was shared at one time, n'est
pas?), There is also the reference to the book at the end of what I'm reposting
here. It brought me up to speed, and it might help this discussion too.
From the book:
Population Differences and the Nature-Nurture Controversy
Along with their disagreements about what intelligence means and whether it is
specific or general, theorists also disagree about why people’s
intelligence-test performances vary. The current debate dates back to the
beginning of World War I, when Robert Yerkes proposed that all military
recruits be given an intelligence test to determine their fitness to serve in
military capacities. The testing also generated data about the intelligence of
the U.S. population as a whole (Yerkes, 1921). Approximately 1.75 million men
were given IQ tests in groups-- written tests for those who could read and
write English, a picture-completion test for those who could not (see Figure
13.11). Never before had IQ tests been administered to such large groups of
people at one time, or to people for whom the language of the tests was not
their native language.
Yerkes’s research began a controversy that has continued to the present time.
Two results appeared to be particularly problematic. First, the average mental
age of native-born Anglo-Americans was assessed at 13 years. Since, by the
standards of the time, a mental age of 8 to 12 years was considered subnormal
for an adult, it appeared that a substantial part of the Anglo population
consisted of "morons." Second, there was a substantial difference between the
scores obtained by recruits of European-American and African-American origin.
Overall, the average for recruits of European origin was a mental age of 13.7
years, whereas African-Americans averaged slightly more than 10 years.
Several of the pioneer testers of intelligence interpreted such differences as
the results of innate, immutable differences in natural intelligence
("nature"). According to this innatist hypothesis of intelligence, some people
are born generally smarter than others, and no amount of training or variation
in the environment can alter this fact. The generally lower test scores of
members of ethnic minority groups and the poor (who often, but not always, are
the same people) were widely interpreted to mean that such groups are innately
and irrevocably inferior (Herrnstein & Murray, 1994).
During the 1930s and 1940s the general-intelligence, innatist position was
balanced by an environmentalist hypothesis of intelligence, which asserted that
intelligence is both specific and heavily dependent on experience (Klineberg,
1980). It was demonstrated, for example, that after people had moved from rural
areas to the city, their intelligence test scores rose (Klineberg, 1935), and
that when orphans were removed from very restricted early environments, their
intelligence test scores improved markedly (see Chapter 7).
One of the most striking lines of evidence for the environmental hypothesis of
intelligence is the fact that worldwide there has been a steady increase in IQ
test performance since testing began (Flynn, 1999). Although the amount of
improvement differs somewhat according to the kind of test that is used and the
particular country in which it is administered, the general result for the 20
countries where such testing has been widely carried out for many decades
indicates that IQ scores have been going up an average of 10 to 20 points for
every generation. This means, for example, that the average African-American
adult in 1990 had a higher IQ than the average European-American adult in 1940,
and that the average English person in 1900 would score at the level currently
considered to indicate mental retardation.
There is no clear consensus about what environmental factors are causing IQ
scores to go up, but it is certain that the change is environmental in origin,
since it cannot be from a rapid change in the genetic constitution of people
all over the world. As Flynn (1999) points out, it is almost impossible to
determine how the environment contributes to the development of intelligence
because all the possible causal factors are closely connected with each other,
and all lead to changes that are in the same direction. The list of the
possible causal factors ranges from improved nutrition and increasing years of
education to an increase in the complexity of life, and even to the spread of
interactive video games. (For extensive discussions of what environmental
factors might be at work in raising IQ scores, see Neisser, 1998).
IQ performance and the Logic of Testing
At the present time, no responsible scholar believes that the variation in
intelligence-test scores from person to person can be attributed entirely to
either environmental or genetic factors As we noted in Chapter 2, a number of
large-scale studies report significant heritability of IQ test performance. At
the same time, even those who believe that genetic variation plays a major role
in variations in IQ readily acknowledge a significant role for the environment
(Ceci & Hembrooke, 1995; Mackintosh, 1998). Those who claim that genetic
heritage makes a large contribution to academic success agree that all
behavior, including performance on IQ tests and in school, is an aspect of a
person’s phenotype (that is, one’s observable characteristics) and that the
phenotype arises from the joint action of the genotype (the set of genes one
inherits) and the environment.
As we pointed out in Chapter 2, the attempt to tease apart the specific
gene-environment interactions that shape human beings is especially difficult
in relation to traits, like intelligence, that are polygenic—that is, that they
are shaped by several or many genes acting in combination in a given set of
environmental conditions. Thus, even when it has been possible to estimate the
genetic contribution to a trait, little can be said about precisely which genes
are interacting with the environment in what way. Efforts to separate the
various influences of nature and nurture on the phenotype are further
complicated by the fact that parents contribute both to their children’s
genetic constitution and to the environment in which their children grow up.
And then there is the final knot in the parsing of gene-environment
interaction: in response to both genetic and environmental influences, children
actively shape their own environments.
Attempts to understand how genetic and environmental factors combine to create
the phenotypic behavior called "intelligence" face another, even greater
difficulty. As we noted earlier, psychologists disagree profoundly about what,
precisely, they are measuring when they administer an intelligence test. All
they can say with any confidence is that these tests predict later school
performance to a moderate degree. (The typical correlation between test
performance and school performance is .50 [Neisser et al., 1996]). We can
understand this problem better if we compare the gene-environment interactions
that might determine intelligence with those that determine height.
To determine how environmental variation influences height, we might study sets
of monozygotic (identical) and heterozygotic (fraternal) twins. Suppose that
the twins to be studied were all born in Minnesota. Suppose further that some
of the twins were separated, with one member of each pair being sent to live
among the !Kung Bushmen of the Kalahari Desert. The environments of Minnesota
and the Kalahari Desert do not represent the most extreme variations compatible
with human life, but they are sufficiently different in climate, diet, daily
activities, and other relevant factors to represent a plausible test of the
relative importance of genetic and environmental contributions to height.
If, within this environmental range, genetic factors dominate the expression of
the phenotype (measured height), then we would expect two facts to emerge:
o The heights of identical twins should be roughly as much alike when the twins
are raised far apart as when they are raised in the same family.
o The similarity between the heights of identical twins should be greater than
the similarity between the heights of fraternal twins. In fact, the similarity
of the heights of identical twins raised in very different environments might
be greater than that of fraternal twins raised in the same environment.
Note that whether the children are in Minnesota or in the Kalahari Desert, we
can be pretty confident about our measure of height. Whether we use a yardstick
or a metric scale, we have a valid standard for measuring the twins’ heights,
regardless of the context in which we use it.
At first glance, IQ tests may appear to be standard measures logically similar
to a yardstick. But this appearance is an illusion. Precisely because
intelligence tests derive their validity from their correlation with academic
achievement, they are rooted in the schooled society in which they are
developed and bound to the graphic systems of representation that are central
to all schooling. But these modes of representation are generally absent in
nonliterate societies. To be administered to a !Kung child, every existing
intelligence test would thus require some modification, if only translation
from English to !Kung. If, for example, one of the test questions asks how many
fingers are on two hands, the tester’s might assume that the test could be
adapted to !Kung with only minimal modification—but that assumption would be
wrong. The number system used by the !Kung is not the same as that used by
Minnesotans, and it plays a different role in their lives. In !Kung society,
the relative importance of knowing the number of fingers on a hand is less
important than knowing how to tie knots with those fingers.
When it comes to the tests that require interpretation of pictures or some form
of written answer, even more serious difficulties arise. The !Kung have no
tradition of either drawing or writing, and research with nonliterate peoples
in several parts of the world (Segall et al., 1997) and with young children in
the United States ( Pick, 1997) shows that people without such experience do
not automatically interpret two-dimensional pictures of objects as they would
the objects themselves. For them, interpreting the pictures requires additional
mental work. As a result, tests that use pictures or require copying figures
graphically are inappropriate, as are any tests that depend on the ability to
read. We thus cannot assume that an IQ test is like a yardstick, yielding
equivalent measures in all cultural environments.
Various attempts have been made to create "culture-free" tests, but no
generally satisfactory solution has yet been found: all tests of intelligence
draw on a background of learning that is culture-specific (Cattell, 1949;
Davis, 1948). (More recent attempts to deal with the difficulties of comparing
intelligence across racial and cultural lines are described in Irvine & Berry,
1987, and Neisser et al., 1996.)
The fact that intelligence cannot be tested independently of the culture that
gives rise to the test greatly limits the conclusions that can be drawn from IQ
testing in different social and cultural groups. A number of studies have used
comparisons of identical and fraternal twins to distinguish genetic from
environmental contributions to intelligence, but those studies suffer an
important limitation. According to the logic of twin studies, the twins’
environments must differ enough for it to be possible to detect their
differential contributions with the test. But if the environmental variation is
very great, as in the case of a child transported from Minnesota to the
Kalahari Desert, both twins’ intelligence cannot be validly measured by the
same test.
Despite these difficulties, a large literature has grown up around studies of
twins’ IQ test performance, along with studies of children of interracial
marriages and of children adopted across racial and ethnic lines (reviewed in
Mackintosh, 1998). Controversy continues to attend this work, but the following
conclusions appear to be the most defensible.
1. Some part of individual differences in performance on IQ tests is
attributable to inheritance. The degree of heritability is in dispute: some
investigators claim that it is very high (Herrnstein & Murray, 1994); some
claim that it is very low or indeterminate (Bronfenbrenner & Ceci, 1994). One
influential summary estimates that perhaps 50 percent of the variation in test
performance within population groups is controlled by genetic factors (Plomin
et al., 1997).
2. There are significant differences between ethnic groups in their average IQ
scores. Americans of European origin score about 15 points higher than
African-Americans, while Asian-Americans score a few points higher than
European Americans. Other ethnic groups in the United States, such as Native
Americans and Hispanics, score at some intermediate level (Herrnstein & Murray,
1994; Mackintosh, 1998).
3. There is no evidence that the average difference in scores between ethnic
groups in the United States is the result of inherited differences in
intelligence, however defined.
At first glance, the first two facts may appear to conflict with the third: if
inheritance is responsible for a part of the differences between individuals in
tested intelligence, and if there are differences between groups in tested
intelligence, why wouldn’t it be reasonable to conclude that the source of the
differences between groups is the same as the source of the differences between
individuals?
There are two answers to this question, one logical and the other empirical.
The logical answer was provided by Richard Lewontin (1976). It can be
illustrated by an example from plant genetics (see Figure 13.12). Suppose that
a farmer has two fields, one fertile and the other depleted of nutrients. He
randomly takes corn seed from a bag containing several genetic varieties and
plants them in the two fields. He cares for them equally. When the plants have
reached maturity, he will discover that in each field some plants have grown
taller than others. Since all the plants within a given field experienced
roughly the same environment, their variation can be attributed to genetic
factors. But the farmer will also discover variation between the fields: the
plants grown in the fertile field will, on average, be taller than the plants
grown in the nutrient-poor field. The explanation for this average difference
in the heights of the plants lies in their environments, even though the
degrees of heritability in the two fields may be equal.
This same argument applies to variations in test performance between ethnic and
racial groups. Even though the heritability of intelligence within ethnic or
racial groups may be the same, the average difference in performance between
groups may still be caused not by their genetic endowment but by differences in
the environments in which the children have been raised.
Lewontin’s example also illustrates another important point about heritability
that applies equally to IQ. As pointed out in Chapter 2, heritability is a
population statistic: it applies to groups, not to individuals. If the
heritability statistic for height in a field of corn or a set of IQ scores is
.50, it does not mean that 50 percent of the height of each corn plant or each
IQ score is determined by genetic factors. It means, rather, that 50 percent of
the variation in height in the entire field of corn or of the variation in
scores in the entire set of IQ scores can be traced to genetic differences. The
other 50 percent of the variation must somehow be explained in terms of
environmental factors.
Suggestive evidence concerning environmental factors that account for ethnic,
racial, and class differences in tested IQ comes from a study of ethnicity and
IQ among a large sample of 5-year-old African-American and European-American
low-birth-weight premature children, most of whom were from relatively poor
families (Brooks-Gunn et al, 1996). The researchers studied these children
from birth, and in addition to giving the children IQ tests when they were 5
years old, they collected data on neighborhood and family poverty, the social
structure of the families, maternal characteristics such as education and IQ,
and the degree of cognitive stimulation in the home environment. In line with
prior research, the African-American children’s IQ scores were significantly
lower than the scores for the white children (85 vs. 103). When adjustments
were made for ethnic-group differences in poverty, however, the difference in
IQ score was reduced by a little over half. When the differences in the
cognitive stimulation provided in the home environment were also controlled for
statistically, the ethnic differential in IQ was reduced by another 28 percent.
As the authors note, these results do not imply that heredity has no role in
IQ, because they are not based on twin studies that allow estimates of
heritability. They do show clearly, however, that when socioeconomic
differences in the lives of African-American and European-American children are
taken into account, IQ differences between the groups are all but eliminated.
The Development of Children (4th ed.), M. Cole & S. Cole, 1996
__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/
This archive was generated by hypermail 2b29 : Mon Oct 01 2001 - 01:02:19 PDT