Journal o f Educational Measurement

Spring 1989, Vol. 26, No. 1, pp. 1-16

The Raven Progressive Matrices: A Review of National

Norming Studies and Ethnic and Socioeconomic
Variation Within the United States
John Raven
Edinburgh, S c o t l a n d

In this paper, some recent results relating to the stability o f scores on the Raven
Progressive Matrices Test f o r different subgroups within and between the
United Kingdom, the United States, and other Western societies are summar-
ised. Subsequent sections deal with variation over time. A possible explanation
f o r the variation in norms over time and between ethnic groups within the
United States is offered.

Raven's Progressive Matrices (RPM) and Mill Hill Vocabulary (MHV)

Scales were developed to assess, as straightforwardly as possible, the two
components of general intelligence identified by Spearman in 1923 (see Spear-
man, 1927). These are, respectively, (a) eductive ability, that is, the ability to
educe correlates, the ability to generate high level schemata, which make it easy
to handle complex events, and (b) reproductive ability, that is, the ability to
recall acquired information.
The available evidence suggests that Raven was successful in developing
measures of these abilities. Many researchers have demonstrated that the
Progressive Matrices tests are among the purest available measures of g, and the
Mill Hill Vocabulary Test, which can be administered in a few minutes,
correlates well over .9 with full-length intelligence tests (see Raven, Court, &
Raven, 1987, for a review of the literature). Nevertheless, within age groups, the
Progressive Matrices Test and the Mill Hill Vocabulary Scale correlate only .5.
Thus, as Spearman observed, reproductive and eductive abilities would appear to
be distinct.
Spearman also noted that, despite their distinct psychological natures, educ-
tive and reproductive abilities interpenetrate and cooperate, contributing cumu-
latively to effective life performance. Furthermore, the types of activity that
people with relatively high eductive ability undertake well are very different from
the types of activity that people with high reproductive ability excell in. This
finding has not received sufficient attention in educational institutions, which, on
the whole, favour students with high reproductive ability (see Hope, 1985;
Raven, 1977; Raven, Johnstone, & Varley, 1985).
The Raven Progressive Matrices Tests are available in three forms: the
Coloured version (for children), the Standard version (for the entire age range),
and the Advanced version (designed to spread the scores of the top 10% on the
Standard version). Two versions of the Vocabulary scales are available: the
Crichton Vocabulary Scale (for children) and the Mill Hill Vocabulary Scale
John Raven

(for a cross-section of ability). The MHV scale itself is available in senior and
junior forms. Both sets of tests have been revised from time to time over the past
40 years.

Standardisations of the Progressive Matrices

The Progressive Matrices Tests have been used in over 1,600 published
psychological studies (see Court, 1988; Court & Raven, 1977 and 1982) and are
widely used by applied psychologists.
The Standard Progressive Matrices (SPM) was first fully standardised by
Raven on 1,407 children in Ipswich, England, in 1938 (Raven, 1941). Over the
years several more normative studies were carried out. The first of these was by
Raven himself when he standardised the MHV scale in Colchester, England, in
1943-44 (Raven & Walshaw, 1944). The SPM norms obtained in that study
were consistently two raw score points lower than the Ipswich norms. In 1952,
Adams reported norms from 11,621 12-year-old children in Surrey, England.
The Surrey data were, within the limits of sampling error, very similar to
Raven's. Tuddenham et al. (1958), in one of the few studies that attempted to
establish the appropriateness or otherwise of the British norms in the United
States, tested several school classes of Californian children. They concluded that
the British norms were acceptable. In 1963-65, Skanes (see Raven, 1981) tested
4,017 children aged 91/2 to 14 in St. Johns, Newfoundland. The similarity
between Skanes's results and the 1938 Ipswich norms is striking (Raven, 1981).
In 1967, he also tested the entire population of 2,097 Corner Brook, Newfound-
land, children aged 101//2 to 141/2. The results consistently lagged behind the
Ipswich norms. In 1972, Byrt and Gill (1973), working with the author, collected
data from a nationally representative sample of 3,464 primary school children
aged 51/2 to I ll/z in the Republic of Ireland. The urban norms seemed to
correspond to the 1938 lpswich norms, although the figures for the rural areas
lagged behind (Raven, 1981). Up to 1979, therefore, there was apparent
uniformity in normative results.
From 1979 on the story began to change. In 1979, Kratzmeier and Horn
reported norms from a large German study that were well above those obtained
in England in 1938. Melhorn's (1980) East German data were similar. The 1979
British norms appeared to be broadly similar to those obtained in the two
German studies (Raven, 1981). Holmes (1980) reported results for British
Columbia that were similar, if slightly lower. The New Zealand Council for
Educational Research reported closely corresponding results for New Zealand in
1984. Ferjencik (1985) reported data for the Coloured Progressive Matrices for
Czechoslovakia that corresponded to a recently reported British study. Work
carried out in the United States by Raven, Summers et al. (1986) shows that,
although the overall U.S. norms lag behind these international figures, the norms
for whites do not. Finally, Zhang (unpublished) has collected data for urban
mainland China and shown that, despite the norms collected by Chan (1983) for
Hong Kong, they correspond closely with those obtained elsewhere.
 Raven Progressive Matrices

The 1979 British Standardisation

We may now examine the 1979 British Standardisation in a little more detail.
It was based on a nationally representative sample of over 3,250 children aged 6
to 16.
SPM scores correlated .16 with region of the country, but when the effect of
socioeconomic status (SES) was partialled out, the correlation dropped to .07.
The correlation between the SPM and SES was .22. However, because age
accounted for 46% of the variance, SES accounted for 8.9% of the variance that
is not attributable to age. This is equivalent to a within-age correlation between
SES and the SPM of .30.
SPM score correlated .68 with age. Thus, more than half the variance was not
explained by age. One result of this is that the top 10% of 71/2-year-oids do better
than the bottom 10% of 151/2-year-olds. Although these corollaries of the initial
observation will be familiar to anyone involved with the measurement of abilities,
the first runs counter to the widely held belief that such tests simply measure
"intellectual maturity," and the second seems to be largely ignored by the
educational system.
As in the 1938 standardisation, item characteristic curve-based (item response
theory or Rasch-type) item analyses were carried out separately within each
socioeconomic and age group. The items scaled in virtually the same way in all
cases. It is therefore difficult to maintain that, in any general sense, the test is
foreign to the way of thought of children from certain backgrounds, although
that does not mean that it is not unfair to particular children.

U.S. Data
Between 1983 and 1987 some 30 norming studies, involving more than 30,000
students aged 5 to 18 years, were carried out in school districts across the United
States of America (Raven, Summers ct al, 1986).~ Each sample was chosen to be
representative of the school district from which it was drawn.
The norms that were obtained varied markedly from one school district to
another and, within districts, between socioeconomic and ethnic groups. Both
ethnicity and SES made independent contributions to the variance. Differences
between the norms for school systems catering to white students of differing SES
were as great as the ethnic differences within school districts. The ethnic
differences seem to correspond to differences in birth weight, infant mortality,
and the incidence of serious childhood illness (U.S. Bureau of the Census,
1984).
The data hint that the Hispanic-white difference may be declining: There was
no major Hispanic/white difference in the data collected by Stallard (see Raven,
Summers et ai, 1986) in the Ontario-Montclair school district of California. Yet
there had been a marked difference between these two groups in the adjacent
county of Riverside when A. R. Jensen (1980) collected his data 15 years earlier.
Although Jensen's results have been confirmed in studies conducted in other
areas by Burciaga (1973) and Hoffman (1983), Stallard's results are confirmed
John Raven

in as yet unpublished data recently collected by Felmlee in San Luis valley,

Colorado.
Item analyses were run separately for a number of ethnic groups. The results
can be indicated briefly by saying that the intercorrelations between the item
difficulties established separately for these groups ranged from .98 to 1.00. In
addition, Hoffman (1983, 1986) demonstrated that the regression lines of RPM
on achievement for the different ethnic groups were parallel but with different
intercepts. Thus, although ethnic groups score at different levels on both
achievement and matrix tests, the RPM has equal predictive validity within each
group.

The Mill Hill Vocabulary Scale

In the 1979 British Standardisation there was, as with the SPM, no variance in
MHV scores within region once the effect of SES was partialled out. SES
explained 16.2% of the variance not explained by age. MHV scores are,
therefore, related more to background SES than SPM scores.
Age accounted for 58% of the MHV variance. MHV scores did not plateau in
the same way as SPM scores, and growth continued at approximately one and a
half words per 6-month interval through age 15 I/2.The top 10% of 9-year-olds did
better than the bottom 10% of ! 51/2-year-olds.
Separate item analyses were carried out again within eight SES groups. Once
more, the reproducibility of the scale properties across groups averaged .97. It
would appear to be untrue that children from different backgrounds learn
different subsets of dictionary words.
The overall U.S. norms again lagged behind the international data, but the
U.S. white norms once more corresponded fairly closely to other cross-cultural
data. The test scaled in the same way for (English-speaking) students from
different socioeconomic and ethnic backgrounds. This conclusion is similar to
that of Deltour (1984) who found that the MHV translated into French with few
changes, scaled in the same way, and yielded similar norms.
In sum, this quick and simple measure of reproductive ability has proved to be
remarkably robust, and the information it has yielded challenges assumptions
commonly made by both psychologists and laymen.

Stability and Change Over Time

The data that have been summarised show that the RPM scales in the same
way for children from different socioeconomic backgrounds in both the 1938 and
1979 British standardisations, from different ethnic groups in the 1986 U.S.
standardisation, and from different cultures. In addition, the computer-drawn
item characteristic curves from the 1979 British standardisation (Raven, 1981)
are very similar to the hand-drawn curves from the 1938 standardisation (Raven
1941).
As can be seen from Figure 1, however, the growth curves for British children
in 1979 are considerably ahead of those for 1938. (The circled points are those
where the unsmoothed 1938 and 1979 data coincide.)
If we can trust the earlier data (and, as we have seen, there is considerable
4
 Raven Progressive Matrices

el)

1979 9Slho*,le

1979 9Otn**,le

SO ~ ] 9 7 9 ?Slh ".,le

,~1~ ~ 1938 th~o, e

45 •

40,

35

T . / / ,- -~r; / ,.Sx x , S ~ - - ".4" / + .+ - +~-*,-.-

- - *~ -I- 1979 5,~, o,lv
SPM
SCORE
/ // /'Iv .t / , ~ . . . . ,93B ,o,,,~....
zS I iV ¢1 / I _-+--"+'- /

15

Io

q" - - - - - - + 1 9 7 9

s •- - . 1938

6'~ 7 7'.~ 8 Bh 9 9' ~ I0 1Oh 11 I1'; ,2 1~': 13 13'. 14 14'~ 15 15'.

tOi~

FIGURE I. Comparison o f 1938 and 1979 S P M growth curves

evidence suggesting that we can), it would appear that children now master the
abilities tested by the Matrices at an earlier age, and that the scores of the less
able plateau above their previous level. However, for the majority of children
aged 11 t/2 and over, there has been little increase in score.
Data assembled by Flynn (1987a) both supports and challenges this conclu-
sion. The Progressive Matrices scores of young adults (military conscripts) from
a wide variety of westernised societies appear (see, e.g., Bouvier's data in Figure
2) to have been going up steadily--Flynn says "dramatically"--over time.
(Flynn cites a round figure of about one standard deviation per generation. 2)
Flynn suggests that what seems to be a more moderate increase in the United
Kingdom--that is, the increase documented in Figure l - - m i g h t be explained if
the 1938 British norms were too high. This explanation is certainly consistent
with the available data. However, an alternative explanation is that British social
and welfare provisions were, at that time, well above the international average
but have now fallen behind. What is perhaps most striking about Flynn's
compendium is that differences between the various norms summarised previ-
ously which, taken by themselves, could be dismissed as sampling errors, turn
out, in a broader context, to be meaningful.
It would appear from the results summarised earlier that there has been, and
still is, considerable--if far from perfect--stability in SPM scores from one
Western society to another at any given point in time. However, in common with
other test norms (see, e.g., Bouvier, 1969; Garfinkel & Thorndike, 1976;
John Raven

38

French Speaking A
/.

g 35
N

/ ~F" S9 ] 6200 25,400

33 /" 60 17,0130 25.000
~r" 6! 16,800 25.900
~r" 62 18.600 20,500
~F" 63 19,700 31.700
32 J" c~ 17,.509 28,400
~" 65 22,900 32.900
J 66 22.100 31.900
~V 67 23,600 33,100
31
I I I I I I I I I I
58 59 60 61 62 63 64 65 66 67
Year
F I G U R E 2. Mean SPM scores of Belgian military conscripts (1958-1967)
(Redrawnfrom Bouvier. 1969)

Thorndike, 1975, 1977; and the large number of published and unpublished
studies brought together by Flynn, 1984b, 1987b), there has been an increase in
mean scores over time, particularly for the less able, at least since 1970 and,
according to Flynn, continuously throughout the period (although accelerating
more recently).
Thorndike (1977) and Garfinkel and Thorndike (1976) suggest several
hypotheses that might explain the increase. However, the data available on the
Progressive Matrices do not really support any of them. Thorndike suggests, for
example, that the observed acceleration in development may be due to earlier
maturity. However, if maturity is a factor, the growth curves for boys and girls
should differ more than the evidence summarised in Raven (1981), Raven,
Summers et al (1986), and Court (1983) suggests that they do. Thorndike
suggests that the differences may be due to the nature of early school education,
but the fact that there was little difference between the RPM norms obtained in
Scotland and England in the course of the 1979 standardisation suggests that this
is unlikely because Scottish infant education remains very formal (HMI, 1980).
Thorndike suggests that television may have had an effect. However, television
was widely available in Ireland when what can now be seen to be low Irish norms
were collected.
Others have suggested that the increases in Progressive Matrices scores over
time may be attributed to schools using matrix-type problems to teach problem
 Raven Progressive Matrices

solving. However, Thorndike showed that performance on all the subscales of the
Binet had improved.
Flynn (1987a) (having, in 1984b, queried Thorndike's hypotheses concerning
the Binet results) concluded that most of the explanations for the RPM increase
that readily come to mind do not hold up. He shows, through a detailed analysis
of Leeuw and Meester's (1984) data, that changes in level of education could
account for only one of the 20 IQ-point gains in RPM scores documented among
servicemen. Changes in the intellectual quality of the home environment, as
indexed by SES, could account for little more.
The variation in mean scores between ethnic groups within the United States
does seem to correspond to variation between the same groups in height, birth
weight, and infant mortality. Height and birth weight have, like intelligence test
scores, increased over the past 80 years. 3 The most probable explanation of the
increase in scores over time therefore seems to be that it is due to the same
variables as increases in height and birth weight and decline in infant mortali-
t y - t h a t is, to improved nutrition, welfare, and hygiene. What it is about these
variables that is important is as obscure for height and birth weight as it is for
intellectual ability. However, the fact that such variables do have important
effects on RPM score as well as birth weight and height is shown in a remarkable
study carried out in Aberdeen, Scotland, most of the results from which have
never been published (Baird & Scott, 1953; Scott, Illsley, & Thomson, 1956). In
this study, calcium intake was used as an index of quality of diet, and it was
shown that this has a marked impact on all three of the outcomes mentioned, and
that the relationship held both within and between socioeconomic groups. More
recently, Benton and Roberts (1988) have shown that vitamin and mineral
supplements appear to have a rapid and marked effect on eductive, but not
reproductive, ability.

The Origins of Variance

It follows from the results just summarised that, despite their persistence over
time (see also the persistence of the French-Flemish speaking difference in
Belgium shown in Figure 2), the differences between ethnic groups within the
United States cannot be regarded as immutable: Most of the current ethnic
norms in the United States lie between the 1938 and 1979 British norms, and, as
we have seen, there is some evidence that the Hispanic-white difference may be
disappearing. (The significance of the increases over time for the arguments
pointing to a genetic explanation of the ethnic differences have been spelled out
by Flynn, 1987b).
There have been a number of empirical studies of the factors that increase or
decrease RPM scores. The results surprise many psychologists. Eductive ability
has turned out to be more easily influenced by appropriate educational and
developmental experience than reproductive ability. However, the variables that
influence the development of eductive ability are not the obvious cultural and
socioeconomic variables that divide society and on which sociologists have
focused so much attention. Acquired information is more influenced by these
variables than is the ability to perceive and think clearly, but these background
John Raven

variables still account for only a small proportion of the total variance. The
development of the ability to perceive and think clearly appears to be promoted
by "democratic" childrearing practices in the home and by "open" educational
activities in the school (Chan, 1981; McGillicuddy-DeLisi, DeLisi, Flaugher, &
Sigel, 1987; Miller, Kohn, & Schooler, 1985; Raven, 1980; Raven et al., 1985;
Sigel, 1986; Sigel & McGillicuddy-DeLisi, 1984; Stallings & Kaskowitz, 1974).
It is also promoted by work that demands high levels of problem solving and
responsibility (Jaques, 1976; Kohn & Schooler, 1978; Lempert, 1986). However,
it appears that the child-rearing and educational practices that increase
Progressive Matrices scores depress reading, writing, and arithmetical ability if
these abilities are measured by conventional educational tests that load heavily
on reproductive ability (Sigel, 1986; Stallings & Kaskowitz, 1974). 4
Experimental attempts to teach the strategies required to solve matrix
problems (e.g., Budoff & Corman, 1976; Budoff, Corman, & Gimon, 1976;
Feuerstein, 1979, 1980; Haywood et al. 1982; Jacobs, 1977; Wortman, 1968)
yield dramatic short-term results, but it is not clear that these affect eductive
ability in any basic sense (Savell, Twohig, & Rachford, 1986). What is notable
about at least some "democratic" parents and "open" educators is that they
extend children's utterances and encourage them to ask questions and to invent
ways of perceiving and conceptualising things for themselves. They also encour-
age them to share in their own problematizing and thinking about the long-term
consequences of their actions in the context of their own personal views of how
society works. 5 In this context it is, however, important to note that attempts to
introduce such activities into schools in general have not succeeded even in
getting teachers to change their behaviour (Fraley, 1981; HMI 1980; Raven et
al., 1985). The same has been true of attempts to get parents to change their
behaviour (Raven, 1980). This is almost certainly why overall evaluations of
intervention programmes such as Head Start and Follow-Through (e.g., Bock,
Stebbins, & Proper, 1977; Spitz, 1986) show no effect.
The evidence that child-rearing practices can influence scores suggests that
there is possibly a more psychologically based explanation of the increase in
RPM scores over time than the explanation in terms of nutrition and hygiene
favoured earlier. This is that the move toward smaller families, combined with
welfare provisions that satisfy the more fundamental needs in Maslow's (1954)
hierarchy, has permitted more parents to adopt child-rearing practices that
facilitate their children's cognitive development. These parents may encourage
their children to think things out for themselves; to share in their own thought
processes and, in particular, to think about the operation of complex social
processes (society) and consider the probable long-term, and as yet intangible,
social consequences of their actions. If they do, these experiences might well be
reflected in increases in both cognitive ability and the development of internal-
ised codes to guide behaviour. When these more socialised children reach school,
their teachers can change their discipline strategies and treat them in a more
developmental way. And when they get to work these young people might find
that it, too, has become more cognitive, requiring them to take more responsibil-
ity for the personal, organisational, and societal consequences of their actions. If
 Raven Progressive Matrices

work has changed in this way, Jaques (1976), Kohn and Schooler (1978), and
Lempert (1986) already have demonstrated that it further promotes the develop-
ment of problem-solving ability. What is important about this possible explana-
tion of the observed increase in scores is that it is not disconfirmed by the
observations that earlier led us to reject Thorndike's hypotheses.
It is important to conclude this section on a less "environmentalist" note by
juxtaposing the fact that Progressive Matrices scores can be influenced by and
have changed over time with the fact that the bulk of the variance is still between
children from similar ethnic and socioeconomic backgrounds. Indeed, as Max-
well (1969) and Jencks et al. (1973) have shown, two-thirds of the variance in
intelligence test scores is between children from the same families.

Concluding Comments
The first general conclusion that may be drawn out of this material is that it
does not really support Flynn's (1987a) contention that "IQ tests do not measure
intelligence but rather a correlate with a weak causal link to intelligence." It is
easiest to demonstrate this by citing parallel data on height. According to Flynn
(personal communication), height has, since well before the turn of the century,
been increasing at about two thirds of a standard deviation per generation. This is
of the same order of magnitude as the one standard deviation per generation
figure he cites for RPM scores. For height, too, sex and ethnic differences
abound, and the differences persist despite intergenerational increases. Neither
the intergenerational increases nor the sex and ethnic differences imply that
rulers cannot validly be used to measure height. Flynn buttresses his conclusion
that IQ scores have little meaning by suggesting that, if people's intelligence
really has improved, more patents should have been filed. This has not happened.
But no one would suggest that a prerequisite to rulers being regarded as valid
measures of height would be that the number of Olympic high jump gold
medalists should have gone up. (Incidentally, would not the number of books and
articles published be a better criterion against which to judge the construct
validity of IQ tests than the number of patents filed? The number has certainly
increased over time.)
My own view is that it would have been virtually impossible to have obtained
meaningful evidence on the issues in question if the tests had not been available.
The impressive stability in the norms across cultures at any given point in time,
the stability in the test properties and item statistics across time and socioeco-
nomic and ethnic group, the consistency of the increases from year to year, and
the persistence of the ethnic differences imply that we are dealing with
psychological processes of fundamental importance.
Despite my disagreement with Flynn's general conclusion, I feel that his
compilation of data is of the greatest theoretical and practical importance. These
data demonstrate that some, as yet unidentified, features of the environment
have dramatic effects on eductive ability.
Another set of insights that stem from the material summarised in this article
are those that relate to the controversy about test bias which has troubled
American education for two decades. Because RPM and MHV scores are
John Raven

relatively unambiguous, the data that have been reviewed direct attention toward
a constructive search for explanations of differences and for educational policies
that reflect the socioeconomic realities, values, and educational needs of the
groups concerned. (See Raven (1987) and the article by Tharp, Gallimore, and
others (Tharp et al., 1984) for a fuller discussion the way in which this might be
done.) Nevertheless, to fully capitalize on these observations it is essential for
psychologists to undertake the fundamental research that is needed to develop
measures of a wider range of human talents. It seems that we are still trying to
administer our educational system with the aid of concepts and tests developed
by Binet at the turn of the century. (A new attempt to provide a conceptual
framework and measurement model appropriate to thinking about the assess-
ment, development and release of a wider range of talents is available in Raven,
1984, 1988, in press-a.)
It may be useful to expand on what was said in the last paragraph by taking an
example. The material that has been presented shows that, if Federal US funds
for Gifted Education were administered on the basis of National US norms, those
funds would be channelled into what are already very wealthy school districts. If
this is regarded as inequitable, and if it is therefore felt that local instead of
national norms should be used to determine the level of provision of, and
eligibility for entry to, both "gifted" and "remedial" forms of special education,
then the logic of the argument points to the use of local ethnic norms. In the end,
therefore, decisions about which norms to use depends on judgments about
whether Special Education is a good thing or a bad thing: does it help or does it
label? By introducing these facts and considerations into the debate it is possible
to direct attention to the goals of policy and the steps that are required to reach
them.
Psychologists should be arguing much more forcefully for a major role in
identifying policy goals and the processes to be used to reach them and in
developing the tools needed to administer the programmes and assess their
effects. If they did this, they would find themselves developing a whole new range
of understandings of the qualities which are to be fostered in the course of
educational programmes, the processes to be used to nurture them, and the
psychometric models to be used to assess educational processes and outcomes
(Raven, 1977, 1984, 1988).

Summary
At its most basic level, the data briefly summarised in this paper suggest that
there is considerable stability in SPM and M H V performance both within and
between societies with a literary tradition at any given point in time. This is true
at the level of item statistics as well as both mean scores and variance. On the
other hand, there has been an impressively continuous increase in scores over
time. Despite these increases, the differences between socioeconomic and ethnic
groups, both within the United States and across cultures, remain.
From a more fundamental point of view the material suggests that the use of
theoretically based tests which have good psychometric properties has:
• demonstrated that we are dealing with psychological processes that are of
10
 Raven Progressive Matrices

fundamental importance. The reproducibility of the scale across time and

socioeconomic and ethnic groups, the stability of the norms across cultures, the
persistence of ethnic differences, and the consistency of the rate of increase all
support this conclusion.
• shown that it is not true that all human abilities have basically the same
patterns of intercorrelation with independent and dependent variables. This is
important because it opens the way to solutions to social and educational
problems which are not available to those who have uncritically accepted the
belief that the bulk of the variance in human performance is attributable to a
single factor.
• made it possible to redirect efforts to find socially acceptable solutions to the
socioeducational problems that have been associated with previous attempts to
use psychological concepts and tools for the enormous range of abilities and
talents that are present in any school class.
• contributed significantly, if in unexpected ways, to answering some of the
highly contentious and politicised research questions that led to these tests being
developed in the first place. The use of the same test in a wide variety of studies,
in field surveys, laboratory experiments, and field trials, in many cultures, and
over an extended period of time has, on the one hand, demonstrated that the
environment can have a previously unsuspected impact on eductive ability and,
on the other, shown that most of the hypotheses commonly advanced to explain
the increase in scores over time are invalid. Firm evidence of an increase over
time has eluded those who have employed tests of General Intelligence that
sample a wide range of content and process because these tests include items that
measure knowledge which is either culturally specific or which becomes increas-
ingly well-known or obsolete. The massive effect of the environment has also
eluded researchers who have worked with twins. But the results are not only
disconcerting to those who believe that g is largely inherited. They also pose
problems for those who are inclined to believe that socioeconomic and educa-
tional environments markedly influence intellectual performance, and particu-
larly those who seek to explain low scores by reference to a cultural-deficit
hypothesis.
The available evidence suggests that the factors responsible for both the
increase over time and the ethnic differences will be found to be among those that
account both for increases over time and ethnic differences in height, birth
weight, and incidence of infant mortality. However, the evidence is far from
conclusive and more psychologically oriented hypotheses are still tenable.

Notes
~Compilation of thcsc norms is continuing. The author would welcome correspondence
from anyone able to contribute to the process and, in particular, from anyone able to
establish norms for an urban, black sample.
2There is no doubt that there has been a major increase in scores over time and that
Flynn has done an outstanding service by unearthing and bringing together studies that
demonstrate this. Likewise, there is no doubt it was necessary for Flynn, like most other
researchers, to employ statistics that do not fully satisfy the relevant statistical assump-
11
John Raven

tions in order to make progress. Nevertheless, it is important to recognise that expression

of the amount of the increase in IQ units gives an unjustified impression of precision. The
problems inherent in the use of IQs are fully discussed in Raven, Summers et al. (1986)
and cannot be discussed here. The easiest way to hint at them is by taking an example. The
within-age distributions of raw scores on many tests deviate markedly from a Gaussian
("normal") curve. They are frequently both biomodal and have tails that do not
correspond to the shape of the Gaussian curve. The result is that the IQ of the same child,
on the same test, judged against the same standardisation sample, can be 40 if the
statistician who calculated the norms made one set of assumptions and 65 if he or she
made other assumptions. It is important to recognise that these technicalities mean that it
would be inappropriate for researchers to embark on a debate about the precise magnitude
of the rate of increase.
~Although the Progressive Matrices data are limited to 50 years, Tuddenham's (! 948)
army data go back to 1914 and the Binet data earlier still (Thorndike, 1975).
q t is not known what the results would be if these academic abilities were measured by
tests that take into account the use of such abilities to find information one wants,
communicate what is important to one, and lead one's life effectively. Raven et al. (1985),
and especially Raven (1988b, in press), discuss the practical implications of these
limitations of conventional educational tests.
5 For fuller discussions see Raven (1987, in press-b, in press-c, in press-d), but the main
references are Tough (1973), McGillicuddy-DeLisi (1985), Sigel and Kelley (1988),
Gallimore, Boggs, and Jordan (1974), Tharp and Gallimore (1984), Raven (1980), and
Raven et al. (1985).

References
Adams, A. E. (1952). Analysis o f Raven's matrics scores: Preliminary report. Surrey,
England: Surrey Educational Research Association.
Baird, D., & Scott, E. M. (1953). Intelligence and childbearing. Eugenics Review, 45,
139-154.
Benton, D., & Roberts, G. (1988). Effect of vitamin and mineral supplementation on
intelligence of a sample of schoolchildren. Lancet, 140-143.
Bock, G., Stebbins, L. B., & Proper, E.C. (1977). Education as experimentation: A
planned variation model (Vol. IV-B). Effects of follow through models. Cambridge,
MA: Abt Associates.
Bouvier U. (1969). Evolution des cotes a quelques test. Belgium: Centre de Recherches,
Forces Armees Beiges.
Budoff, M., & Corman, L. (1976). Effectiveness of a learning potential procedure in
improving problem-solving skills of retarded and non-retarded children. American
Journal o f Mental Deficiency, 81,260-264.
Budoff, M., Corman, L., & Gimon, A. (1976). An educational test of learning potential
assessment with Spanish speaking youth, lnteramerican Journal of Psychology, 10,
13-24.
Burciaga, L. E. (1973). A research study on the Raven Coloured Progressive Matrices
among school children of the E1 Paso public schools. Doctoral dissertation, University
of Texas, El Paso.
Byrt, E., & Gill, P. E. (1973). Standardisation of Raven's Standard Progressive Matrices
and Mill Hill Vocabularly for the Irish population ages 6-12. Master's thesis, National
University of Ireland, University College Cork.
Chart, J. (1981). Correlates of parent-child interaction and certain psychological variables

12
 Raven Progressive Matrices

among adolescents in Hong Kong. In J. L. M. Binnie-Dawson (Ed.), Perspectives in

Asian cross-cultural psychology. Lisle, Netherlands: Zwets and Zeitlinger.
Chan, J. (1983). Norms for Hong Kong. In J. C. Raven, J. H. Court, & J. Raven. A
manual for Raven's progressive matrices and vocabulary tests (1983 ed. "Standard
Progressive Matrices" Section, Table SPM XII). London: Lewis.
Court, J. H. (1983). Sex differences on Raven's Progressive Matrices: A review. Alberta
Journal of Educational Research, 29, 54-74.
Court, J. H. (1988). A researcher's bibliography for Raven's Progressive Matrices and
Mill Hill Vocabularly scales (7th ed.). Published by J. H. Court, 500 Goodwood Rd.,
Cumberland Park, South Australia 5041.
Court, J.H., & Raven, J. (1977 & 1982). Summaries of reliability, validity, and
normative studies for Raven's Progressive Matrices and Vocabulary Scales. In A
manual for Raven's progressive matrices and Mill Hill vocabularly scales (Research
Supp. No. 2). London: H. K. Lewis; San Antonio, Texas: The Psychological Corpora-
tion.
Deltour, J.J. (1984). Une nouvelle methode d'evaluation rapide de la deterioration
mentale. Unpublished manuscript, University of Liege, Laboratoir de Psychologie
Appliquee.
Ferjencik, J. (1985). Manual: Coloured Progressive Matrices. Bratislava, Czechoslo-
vakia: Psychodiagnosticke, a Didakticke Testy.
Feuerstein, R. (1979). The dynamic assessment of retarded performers. Baltimore, MD:
University Park Press.
Feuerstein, R. (1980). Instrumental enrichment. Baltimore, MD: University Park Press.
Flynn, J.R. (1984a). IQ gains and the Binet decrements. Journal of Educational
Measurement, 21,283-290.
Flynn, J.R. (1984b). The mean IQ of Americans: Massive gains 1932 to 1978.
Psychological Bulletin. 95, 29-5 I.
Flynn, J. R. (1987a). Massive IQ gains in 14 nations: What IQ tests really measure.
Psychological Bulletin, 101 (2).
Flynn, J. R. (1987b). Race and IQ: Jensen's case refuted. In S. Modgil & C. Modgil
(Eds.), Arthur Jensen: Consensus and controversy (pp. 221-232). Lewes, England:
Falmer Press.
Fraley, A. (1981). Schooling and innovation: The rhetoric and the reality. New York:
Tyler Gibson.
Gallimore, R., Boggs, J. W., & Jordan, C. (1974). Culture, behavior and education.
Beverly Hills, CA: Sage.
Garfinkel, R., & Thorndike, R. L. (1976). Binet item difficulty: Then and now. Child
Development, 47, 959-965.
Haywood, H. C., & Arbitman-Smith, R. (1981). Modification of cognitive functions in
slow-learning adolescents. In P. Mittler (Ed.), Frontiers of knowledge in mental
retardation: 1Iol 1. Social, educational, and behavioural aspects, (pp. 129-140).
Baltimore, MD: University Park Press.
Haywood, H. C., Arbitman-Smith, R., Bransford, J. D., Towery, V. R., Hannel, I. L., &
Hannel, M. V. (1982). Cognitive education with adolescents." Evalution of instrumen-
tal enrichment. Paper presented at the International Association for the Study of
Mental Deficiency.
Her Majesty's Inspectors (Scotland) (1980). Learning and teaching in primary 4 and
primary 7. Edinburgh: Her Majesty's Stationery Office.
Hoffman, H.V. (1983). Regression analysis of test bias in the Raven's Progressive

13
John Raven

Matrices for Anglos and Mexican-Americans. Unpublished doctoral dissertation,

University of Arizona, Department of Educational Psychology.
Hoffman, H.V. (1986). In J. Raven, J. H. Court, & J. Raven, Manual for Raven's
progressive matrices and vocabularly tests. (A compendium o f North American
normative and validity studies, Research Supp. No. 3). London: H. K. Lewis.
Holmes, B.J. (1980). British Columbia norms for Wechsler Intelligence Scale for
Children--revised; Peabody Picture Vocabulary Test; Slosson Intelligence Test;
Standard Progressive Matrices; and Mill Hill Vocabulary Scale. University of British
Columbia: Faculty of Education.
Hope, K. (1985). As others see us: Schooling and social mobility in Scotland and the
United States. New York: Cambridge University Press.
Jacobs, P. I. (1977). Up the IQ t. New York: Wyden Books.
Jaques, E. (1976). A general theory of bureaucracy. London: Heinemann.
Jencks, C., Smith, M., Acland, H., Bane, M. J., Cohen, D., Gintis, H., Heyns, B., &
Michelson, S. (1973). Inequality: A reassessment o f the effect o f family and schooling
in America. New York: Basic Books; London: Penguin.
Jensen, A. R. (1980). Bias in mental testing. New York: Free Press.
Jensen, M. R. (1985). The LPAD and low functioning children. In, To sing our own songs
(Report of a Workshop for American Indian Educators) New York: Association on
American Indian Affairs.
Kohn, M. L., & Schooler, C. (1978). The reciprocal effects of the substantive complexity
of work and intellectual flexibility. A longitudinal assessment. American Journal of
Sociology, 84, 24-52.
Kratzmeier, H., & Horn, R. (1979). Manual: Raven-Matrizen-Test. Standard pro-
gressive matrices. Weinheim, West Germany: Beltz Test.
Leeuw, J. de, & Meester, A. C. (1984). Over het intelligente-onderzoek bij de militaire
keuringen vanaf 1925 tot henden [Intelligence--as tested at selections for the military
service from 1925 to the present]. Mens en Maatschappij, 59, 5-26.
Lempert, W. (1986). Sozialisation und Personlicckeitsbildung in beruflichen Schulen,
dargestellt am Beispiel der Entwicklung moralischer Orientierung. Die berufsbildende
Schule, 38, 148-160.
Maslow, A. H. (1954). Motivation and personality. New York: Harper.
Maxwell, J. N. (1969). Sixteen years on. Edinburgh: Scottish Council for Research in
Education.
McGillicuddy-DeLisi, A.V. (1985). The relationship between parental beliefs and
children's cognitive level. In I. E. Sigel (Ed.), Parental belief systems." The psychologi-
cal consequences for children. Hillsdale, N J: Erlbaum.
McGillicuddy-DeLisi, A. V., DeLisi, R., Flaugher, J., & Sigel, I.E. (1987). Family
influences on planning. In S. L. Friedman, E. K. Scholnick, & R. R. Cocking (Eds.),
Blueprints for thinking." The role of planning in cognitive development. New York:
Cambridge University Press.
Melhorn, H.G. (1980). Aspekte der Geistigen Entwicklung Jugendlicher. In W.
Freidrich & H. Muller (Eds.), Zur Psychologie der 12 bis 22 Jahrigen. East Berlin:
VEB Deutscher Verlag der Wissenschaften.
Miller, K. A., Kohn, M. L., & Schooler, C. (1985). Educational self-direction and the
cognitive functioning of students. Social Forces, 63, 923 944.
New Zealand Council for Educational Research. (1984). Standard Progressive Matrices:
New Zealand norms supplement. Wellington: Author.
Raven, J. (1977). Education, values and society." The objectives of education and the
nature and development of competence. London: H. K. Lewis.

14
 Raven Progressive Matrices

Raven, J. (1980). Parents, teachers and children. Edinburgh: Scottish Council for
Research in Education.
Raven, J. (1981 ). The 1979 British standardisation of the standard progressive matrices
and Mill Hill vocabulary scales, together with comparative data from earlier studies in
the UK, US, Canada, Germany, and Ireland. In J. C. Raven, J. H. Court, & J. Raven
(1987), A manual for Raven's progressive matrices and vocabulary tests (Research
Supp. No. 1). London: H. K. Lewis.
Raven, J. (1984). Competence in modern society: Its identification, development and
release. London: H. K. Lewis.
Raven, J. (1987). Values, diversity and cognitive development. Teachers College Record,
89, 21-38.
Raven, J. (1988). The assessment of competencies. In H. D. Black & W. B. Dockrell
(Eds.), New developments in educational assessment: British Journal of Educational
Psychology, Monograph Series No. 3. 98-126.
Raven, J. (in press-a). A model of competence, motivation and its assessment. In H.
Berlak (Ed.), Assessing academic achievement: Issues and problems. Madison, WI:
National Center for Effective Secondary Schools.
Raven, J. (in press-b). Equity in diversity: The problems posed by values--and their
resolution. In F. Macleod (Ed.), Families and schools: Issues in accountability and
parent power. Brighton, England: Falmer Press.
Raven, J. (in press-c). Parents and schools in early childhood. British Journal of
Educational Psychology (Special issue).
Raven, J. (in press-d). Progressive matrices." Solution strategies and their implications
for education. Madrid: Ministry of Education.
Raven J., & Court, J. H. (in press). Additional normative, reliability, and validity
data--to 1988. Research supplement No. 4 to Raven, J. C., Court, J. H., & Raven, J.
Manuel for Raven's Progressive Matrices and Vocabulary Scales. London: H.K.
Lewis. San Antonio, TX: Psychological Corporation.
Raven, J., Johnstone, J., & Varley, T. (1985). Opening the primary classroom.
Edinburgh: Scottish Council for Research in Education.
Raven, J., Summers, W. A., et al. (1986). A compendium of North American normative
and validity studies. In J. C. Raven, J. H. Court, & J. Raven, A manual for Raven's
progressive matrices and vocabulary tests (Research Supp. No. 3). London: H. K.
Lewis; San Antonio, Texas: The Psychological Corporation.
Raven, J. C. (1941). Standardisation of progressive matrices. British Journal of Medical
Psychology, 19, 137-150.
Raven, J.C., Court, J.H., & Raven, J. (1987). A manual for Raven's Progressive
Matrices and Vocabulary Tests. London: H.K. Lewis; San Antonio, Texas: The
Psychological Corporation.
Raven, J. C., & Walshaw, J. B. (1944). Vocabulary Tests. British Journal of Medical
Psychology, 20, 185-194.
Savell, J. M., Twohig, P. T., & Rachford, D. L. (1986). Empirical status of Feuerstein's
"Instrumental Enrichment" (FIE) Technique as a method of teaching thinking skills.
Review of Educational Research, 56, 381-409.
Scott, E. M., lllsley, R., & Thomson, A. M. (1956). A psychological investigation of
primigravidae: II. Maternal social class, age, physique and intelligence. Journal of
Obstetrics and Gynaecology of the British Empire, 63, 338-343.
Sigel, I.E. (1986). Early social experience and the development of representational
competence. In W. Fowler (Ed.), Early experience and the development of competence.
New Directions for child development (No. 32). San Francisco: Jossey-Bass.

15
John Raven

Sigel, I. E., & Kelley, T. D. (1988). A cognitive developmental approach to questioning.

In J. Dillon (Ed.), Classroom questioning and discussion: A multidisciplinary study.
Norwood, N J: Ablex.
Sigel, I. E., & McGillicuddy-DeLisi, A. V. (1984). Parents as teachers of their children:
A distancing behaviour model. In A.D. Pellegrini & T.D. Yawkey (Eds.), The
development of oral and written language in social contexts.Norwood, N J: Ablex.
Skanes, G. R., Sullivan, A. M., Rowe, E.J., & Shannon, E. (1974). Intelligence and
transfer. Journal Educational Psychology, 66, 563-568.
Spearman, C. (1927). The nature o f "intelligence" and the principles of cognition (2nd
ed.). London: Macmillan.
Spitz, H. H. (1986). The raising of intelligence." A selected history of attempts to raise
retarded intelligence. Hillsdale, N J: Erlbaum.
Stallings, J., & Kaskowitz, D. (I 974). Follow through classroom observation evaluation
1972-1973. (Report No. URU-7370). Menlo Park, CA: Stanford Research Institute.
Tharp, R. G., Jordan, C., Speidel, G. E., Au, K. H. P., Klein, T. W., Calkins, R. P., Sloat,
K. C. M., & Gallimore, R. (1984). Product and process in applied developmental
research: Education and the children of a minority. In M. E. Lamb, A. L. Brown, & B.
Rogoff (Eds.), Advances in developmental psychology (Vol. 3, pp. 91-144). Hillsdale,
N J: Lawrence Erlbaum.
Thorndike, R. L. (1975). Mr. Binet's test 70 years later. Presidential address presented at
the annual meeting of the American Educational Research Association,
Thorndike, R. L. (1977). Causation of Binet IQ decrements. Journal of Educational
Measurement, 14, 197-202.
Tough, J. (1973). Focus on meaning: Talking with some purpose to young children.
London: Allen and Unwin.
Tuddenham, R.D. (1948). Soldier intelligence in World Wars I and 11. American
Psychologist, 3, 54-56.
Tuddenham, R. D., Davils, L., Davison, L., & Schnindler, R. (1958). An experimental
group version for school children of the progressive matrices. University of California
(Mimeo). (See Abstract: J Consult. Psychol., 22, 30.)
United States Bureau of the Census. (1984). Statistical abstract of the United States,
1983. Washington, DC: U.S. Government Printing Office.
Wortman, R. A. (1968). Coaching and teaching in retardates." The Raven Matrices as a
learning situation (Final Report, Project 6-8441). Washington, DC: U.S. Department
of Health, Education, and Welfare,

Author
JOHN RAVEN, Consultant, 30 Great King St., Edinburgh EH3 6QH, Scotland.
Degrees: BSc, University of Aberdeen; Dip. Soc. Psych., London School of Economics;
PhD, University of Dublin. Specializations: broadly based educational evaluation, links
between education and society.

16