Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Developmental Psychology

1969, Vol. 1, No. 2, 169-177

Cohort Effects in Cognitive Development of


Children as Revealed by Cross-Sectional
Sequences1
PAUL B. BALTES2 GUENTHER REINERT
AND
West Virginia University Universitaet Mannheim

A recently proposed bifactorial developmental model consisting of the fac-


tors of age and cohort was tested by using cross-sectional sequences as data-
collection strategies. In each of two studies, age was varied in two levels
(8:4 and 8:8, 9:4 and 9:8) and cohort (birthdate) in three levels (July
1955, November 1955, and March 1956; July 1956, November 1956, and
March 1957). Four subtests of intelligence from the Begabungstestsystem
were administered to two random samples of 420 subjects, with 35 boys and
35 girls in each treatment condition. Using an analysis of variance design,
significant main effects of age and cohort as well as significant interactions
were found for some of the four measures of cognitive functioning. It is
argued that the obtained cohort effects represent differential amounts of
schooling rather than season-of-birth effects. The findings clearly support the
necessity of the bifactorial developmental model which calls for sequential
strategies to disentangle age from cohort effects in developmental research.

In an attempt to solve the ongoing con- sumptions, these sequential strategies are
troversy about the validity of various age expected to yield the unconfounded effects
curves, alternative strategies to the conven- of age, cohort, and time of measurement
tional cross-sectional and longitudinal de- and thus disentangle some of the sources
signs have been proposed (Baltes, 1968; of error confounding the age differences ob-
Schaie, 1965). Using Bell's (1953) notion tained by application of single conven-
of an "accelerated longitudinal approach" tional cross-sectional or longitudinal meth-
as a starting point, Schaie (1965) formu- ods. Schaie (1965) appears to explain the
lated a well-explicated general model for effects of the three components as the con-
the study of developmental problems which sequences of maturation (age), environ-
is based on three components: age, cohort ment (time of measurement), and of
(birthdate), and time of measurement. In heredity (cohort), an interpretation long
order to analyze his trifactorial devel- sought in developmental psychology.
opmental model, Schaie derived three While accepting the basic setup of
types of sequential strategies: the cross- the general developmental model, Baltes
sequential, the cohort-sequential, and the (1967, 1968) criticized Schaie's analysis
time-sequential method. Under certain as- and interpretation of the developmental
1
model by pointing out some shortcomings
This study was conducted in conjunction with in the formal definitions and the measurabil-
research supported by a grant from the Wis- ity of the components age, cohort, and time
senschaftliche Gesellschaft der Universitaet des
Saarlandes, Germany. The authors are grateful to of measurement. He proposed to give up
Heide Diesel, Elisabeth Haas, and Gisela Labouvie the explanatory status of the components
for their assistance in data analysis and to J. R. (maturation, heredity, environment) and to
Nesselroade and O. H. Cross for their critical transform Schaie's trifactorial model into a
readings of the manuscript. bifactorial developmental model using the
2
Requests for reprints should be sent to Paul B.
Baltes, Department of Psychology, West Virginia descriptive components of age and cohort
University, Morgantown, West Virginia, 26506. only. As was true for Schaie's sequential
169
170 PAUL B. BALTES AND GUENTHER REINERT

strategies, the bifactorial developmental sign with repeated observations over age (longitu-
model can be conceptualized as a design dinal sequences). Such reasoning follows from the
fact that the increase in sensitivity of a repeated-
with independent as well as repeated meas- measurement design concerns only those factors
urements over the age factor. It was pro- where repeated observations are possible, that is,
posed that the data-collection strategies with regard to the bifactorial model, only the age
needed to apply the bifactorial develop- factor. In addition, the interpretation of cross-sec-
mental model as a design with independent tional sequences is not impaired by the existence
of either testing, selective-sampling, or selective
measurements be called cross-sectional se- dropout effects as is the interpretation of longitu-
quences; and in the case of repeated meas- dinal sequences. The analysis of cohort effects be-
urements, longitudinal sequences. If certain ing the primary purpose of the present study, it
assumptions are met, the use of independent therefore seemed logical to choose cross-sectional,
as well as repeated measurements, that is, rather than longitudinal, sequences as data-collec-
tion strategies.
the application of cross-sectional as well as As part of a larger study on the differentiation
longitudinal sequences, permits the separa- hypotheses of intelligence (Reinert, 1969; Reinert,
tion of the effects of age and cohort in de- Baltes, & Schmidt, 1965), it was possible to col-
velopmental phenomena. lect data suited for the bifactorial developmental
model. Four independent cross-sectional studies
As holds true for all models, their useful- separated by 4-month intervals (November 1964,
ness is dependent upon their heuristic value, March 1965, July 1965, November 1965) were
that is, upon the empirical relevance of age conducted. The population of subjects consisted
and cohort effects in the case of the bifac- of all 8- to 10-year-old pupils (Grades 2-4) attend-
torial developmental model. While the ex- ing 64 elementary schools in the city of Saar-
bruecken (Germany) and its surrounding com-
istence of age effects on psychological be- munities. Each of the 64 elementary schools was
havior is widely acknowledged, the presence randomly assigned to one of the four occasions of
of cohort effects, that is, differences between measurement, so that the subjects in each cross-
people born at different points of time, is section came from 16 elementary schools. The
city of Saarbruecken has a population of about
not generally recognized. Thus far, there are 150,000 and is the capital of the Saarland, one of
few studies which have attempted to analyze the states of the Federal Republic of Germany. In
cohort effects directly (e.g., Riegel, 1965; general, the population can be described as being
Riegel, Riegel, & Meyer, 1967; Schaie, & fairly representative (IQ, socioeconomic status) of
Strother, 1968, 1969), most of the evidence German children from urban and suburban areas.
A more detailed description of the population is
being of an indirect nature (Damon, 1965; given elsewhere (Reinert, 1969).
Kuhlen, 1963). In general, one is inclined From the pool of subjects (N = 9686), two
to concede that there might be differences different designs (Study I, Study II) each consist-
between cohorts born several decades apart, ing of two age and three cohort levels were set
up to test the bifactorial model (Table 1).
for example, differences between genera- The subpopulations defined by their levels of
tions; however, one is hesitant to predict age and cohort were each measured only at one
differences between subpopulations of or- time of measurement, that is, either in November
ganisms born only a few years, or even 1964, March 1965, July 1965, or November 1965.
months, apart. The present study tests such Then, from each of the various subpopulations a
subsample of 70 subjects (35 male, 35 female)
lower bounds of the cohort dimension with was drawn at random, resulting in a total sample
regard to cognitive development in children. of 420 subjects for each of the two studies. Thus,
a 2 (Age) X 3 (Cohort) x 2 (Sex) analysis of
variance setup was obtained for both Study I and
Method Study II.
Research Design and Subjects Measurement Variables and Analysis
One has to observe at least two cohorts at the As measures of cognitive functioning, a selected
same two age levels to realize the minimum se- battery of subtests from the Leistungspruefsystem
quential design derived from the bifactorial de- (LPS) and the Begabungstestsystem (BTS) were
velopmental model. This can be accomplished, used. Both the LPS (Horn, 1962) and the BTS
however, either by cross-sectional or longitudinal (Horn, 1956) are group tests of intelligence con-
sequences. It has been argued (Baltes, 1968) that structed to cover the domain of intelligence as
a design with independent observations over both conceptualized by Thurstone's primary mental
age and cohort (cross-sectional sequences) is as abilities for the German language area. Thus far,
sensitive to the analysis of cohort effects as a de- however, factor-analytic studies have not been
COHORT EFFECTS IN COGNITIVE DEVELOPMENT 171

TABLE 1
BlFACTORIAL DEVELOPMENTAL MODEL: Two RESEARCH DESIGNS USING
EACH THREE LEVELS or COHORT AND Two LEVELS OF AGE

Age"
Cohort"
8:4 8:8 9:4 9:8

Study I

July 1956 November 1964 March 1965 —


November 1956 March 1965 July 1965
March 1957 July 1965 November 1965 —

Study II

July 1955 November 1964 March 1965


November 1955 March 1965 July 1965
March 1956 — July 1965 November 1965

Note.—Body entries representing four times of measurement.


a
Bimonthly range in each level of cohort and age.

very successful in reproducing the assumed dimen- of variance was performed for each of the four in-
sions (Reinert et al., 1965) in a satisfactory and telligence measures separately for Study I and
consistent manner, as they did not suggest any Study II, yielding a total of eight analyses of
other clear-cut interpretation. Therefore, in the variance.
present study the authors have kept to the content
validity of the subtests, instead of referring to their Results
assumed factorial validity. For the present analysis
•only the subtests of the BTS included in the bat- The results are summarized in Table 3
tery are used. Table 2 gives an overview on some and Table 4. In describing the two tables,
of their test characteristics including their assumed it should be noted that the treatment factors
factorial validity. age and cohort have to be conceptualized
After the distribution of the raw scores and the
homogeneity of variances were found to be satis- as fixed factors; that is, in a strict sense it is
factory for the four variables, a 2 x 3 x 2 analysis feasible only to state differences between

TABLE 2
FOUR MEASURES OF INTELLIGENCE FROM THE BEGABUNGSTESTSYSTEM (BTS)

Factor identification
Subtest Number of item
Thurstone system Guilford system

Regelerkennen (RE)a Induction Cognition of symbolic systems 28


Buchstabenraten (BR)b Verbal comprehension Cognition of semantic units 30
Grundrechnen (GR)° Numerical facility Convergent production of 38
symbolic implications
Buchstabenzaehlen (BZ)d Perceptual speed Evaluation of symbolic units 96

a
Letter Series.
b
c
Word Completion.
d
Basic Arithmetic.
Letter Counting.
172 PAUL B. BALTES AND GUENTHER REINERT

TABLE 3 Age Effects


MEANS IN FOTJB SUBTESTS OF INTELLIGENCE" FOR As was expected from numerous other
DIFFERENT LEVELS OF COHORT AND AGE
studies on the development of intelligence
and from the standardization data of the
Age BTS, the subtests Buchstabenraten (BR),
Subtestb Cohort Grundrechnen (GR), and Buchstabenzaeh-
8:4 8:8 9:4 9:8
len (BZ) show a main effect of age which,
of course, signifies that with increasing age
there is an increase in intellectual perform-
Study I ance as shown in Table 3. This does not
apply to the same extent to both studies,
7, 1956 11.3 12.0 _ since in Study I (ages 8:4 and 8:8) the
RE 11, 1956 11.8 12.5 — — three subtests, BR, GR, and BZ, show sig-
3, 1957 12.2 11.8 — — nificant age differences, while in Study II
7, 1956 5.2 6.9 — — (Ages 9:4 and 9:8) only Subtest BZ re-
BR 11, 1956 5.7 7.7 — — veals a significant main effect of age.
3, 1957 6.2 7.3 — — On the other hand, some of the main ef-
fects of age cannot be interpreted unambig-
7, 1956 13.2 18.0 — — uously because of the significant interac-
GR 11, 1956 16.4 18.3 — —
3, 1957 17.1 18.0 — — tions. This will be dealt with in a subsequent
section in more detail.
7, 1956 32.7 35.5 — —
BZ 11, 1956 33.9 32.6 — —
— Cohort Effects
3, 1957 31.3 36.5 ~ ~
Although overall main effects of cohort
are not so numerous as age effects, they are
Study II present in both studies, equalling the num-
ber of age effects in Study II, Subtest GR re-
7, 1955 13.4 14.1 veals in both studies a significant main ef-
RE 11, 1955 — 13.8 14.5 fect of cohort. The a posteriori analysis
— 14.2
3, 1956 — — 14.3 shows (see Table 3) that in both studies it
7, 1955 — — 7.9 9.6 is the first cohort level (7, 1956 and 7,
BR 11, 1955 — 9.8 10.4 1955) which is significantly different from

3, 1956 — — 9.7 10.4 the other two cohort levels. The significant
interactions between age and cohort, again,
7, 1955 18.1 21.5
GR 11, 1955

— — 22.1 21.8 point out that the main effects of cohort
3, 1956 — —
— 21.2 20.2 also cannot be generalized in a straightfor-
ward manner over all other treatment con-
7, 1955 33.9 38.7 ditions, but have to be qualified with regard
BZ 11, 1955 — 38.5 38.6
3, 1956 —
•— — 38.1 40.7
to the age level.
Sex Effects
Note.—Total sample for each study: n = 420. The means
combined over sexes are based on n = 70. Neither Study I nor Study II revealed a
» Begabungstestsystem (Horn, 1956).
b
RE = Regelerkennen (Letter Series); BR, = Buchstaben- significant main effect of sex, so that, in
raten (Word Completion); GR = Grundrechnen (Basic the two studies, sex does not pose a serious
Arithmetic); BZ = Buchstabenzaehlen (Letter Counting).
problem. There is, however, one significant
interaction between age and sex for the
mean performances at the various levels Regelerkennen (RE) variable in Study II.
without generalizing to the total range of Since sex differences are not involved in
age or cohort under consideration. The those variables which are affected by co-
level of significance for interpreting differ- hort effects (the prime concern of the pres-
ences was set at the .02 level of confidence ent paper), no further discussion of the
for all treatment conditions. Age X Sex interaction will be pursued.
COHORT EFFECTS IN COGNITIVE DEVELOPMENT 173

TABLE 4
SUMMARY OP ANALYSES OF VARIANCE FOR STUDY I AND STUDY II

Subtest"

Source df RE BR GR BZ

Study I Study II Study I Study II Study I Study II Study I Study II

Age (A) 1,408 ** #* * **

Cohort (C) 2,408 ** **

Sex (S) 1,408


AXC 2,408 ** **
*
AXS 1,408 **

c x s 2,408
A X C X S 2,408

a
Note.—For levels of ay;e and levels of cohort in both studies see Table 1.
RE = Regelerkennen (Letter Series); RR = Buchstabenraten (Word Completion); GR = Grundrechnen (Basic Arithmetic);
BZ = Buchstabenzaehlen (Letter Counting).
* V < .02.
** p < .01.

Interaction Effects Interpretation of Cohort Differences


A total of three significant interactions Contrary to Schaie's proposal (Schaie,
between age and cohort was obtained. Since 1965), it has been argued previously (Bal-
there is no evidence of heterogeneity of tes, 1968) that it is not justified to interpret
variances (indicating a violation of the age or cohort differences as representing the
analysis of variance model), the striking effects of different functional developmen-
stability of the significant Age X Cohort in- tal sources (maturation, heredity, etc.)
teraction for GR in both Study I and Study since such sources cannot be teased out by
II may tentatively be interpreted. As can be the arrangement of the general developmen-
seen from Figure 1, the Age X Cohort in- tal model itself. However, it seems reason-
teraction signifies that the main effect of able to hypothesize about potential sources
cohort concerns only the lower age groups when one is able to obtain additional infor-
(8:4 and 9:4), while the older age groups mation on the treatment conditions, that is,
in both studies obtained about the same on the levels of cohort or age. Since the
mean performance across all levels of co- main purpose of the present paper is to test
hort. The other side of the same interaction the existence of cohort differences, it is con-
is, of course, that the main effect of age for cerned only with hypotheses about their
Subtest GR obtained in Study I holds only interpretation. Two rival interpretations of
with one of the cohort levels (7, 1956). the nature of the obtained cohort differ-
The interaction term Age X Cohort, ob- ences seem most reasonable, although the
tained for Variable BZ in Study I, shows present authors favor the second one.
that the significant main effect of age (Ta- One interpretation is based on the no-
ble 3) does not apply to Cohort 11, 1956; tion of season of birth effects on intelligence
that is, the mean scores obtained by the as discussed, for example, by Pintner and
younger and older age groups of this cohort Forlano (1943), Orme (1962, 1963), or
level do not differ significantly. However, Craddick (1966). This season of birth ef-
the mean scores of the other two cohort fect refers to the notion that subjects born
levels do exhibit significant age differences. at various seasons of the year (spring,
174 PAUL B. BALTES AND GUENTHER REINERT

19

18

17

„«

fi
$14
u
213

12 STUDY I

II

7,1966 11,1965 3,1956 7,1966 11.1986


COHORTS COHORTS
FIG. 1. Age x cohort interaction for GR in Study I and Study II.

summer, etc.) show differences in certain taneously different amounts of schooling.


psychological behaviors. Such effects are In other words, children of the same age
generally interpreted as a function of either have had different lengths of schooling, de-
exogenous or endogenous factors related to pending upon the time they were measured
the differential time of birth. Since in both during the year. The population that the
present studies the cohorts which exhibit present study was conducted with started
significantly lower intellectual performances school in midspring. Since at any time of the
on GR consist of subjects born at the same year it is possible to get children of all
time of the year (June and July), one might ages, and grades generally include a 1-year
argue that the obtained cohort effects rep-
resent season-of-birth effects. As a matter TABLE 5
of fact, such an interpretation parallels a LENGTH OF SCHOOLING FOR DIFFERENT COHORTS
number of studies showing that summer-
born children tend to have lower IQs. How-
Age
ever, one wonders why such a season-of-
birth effect should be present in the younger
age groups only (8:4 and 9:4) and why it 8:4 8:8 9:4 9:8
should affect only such a measurement var-
iable as GR which generally is conceived to Study I
be largely dependent on training opportuni-
ties. 7, 1956 | 7 11
The alternative interpretation, on the U, 1956 11 15 — —
other hand, seems much more plausible to 3, 1957 15 19 ~ "
the present authors and might even be taken
as a basis for solving the ongoing discus- Study II
sion on the existence and/or on the meaning
of season-of-birth effects with regard to 7, 1955 19 23
11, 1955 — 23 27
psychological variables. 3, 1956 ~ . — 27 31
As can be seen from Table 5, the differ-
ent levels of cohort besides being born at dif-
Note,—Body entries representing months, second grade
ferent times of the year, experienced simul- base line.
COHORT EFFECTS IN COGNITIVE DEVELOPMENT 175

range, subjects at the four times of measure- for season-of-birth differences without be-
ment (November 1964, March 1965, July ing able to put it to a rigorous test, since in
1965, November 1965) showed differential the present study different length of school-
amounts of schooling. For example, the 8:4 ing and season of birth (cohort) are co-
year-olds tested in November 1964 had varying simultaneously. A study to sepa-
about 7 months of schooling beyond second- rate schooling from season-of-birth effects
grade entrance, 8:4 year-olds tested in would have to be conducted, for example,
March 1965 had 11 months, and 8:4 year- cross-culturally, so that one could obtain an
olds tested in July 1965 even had 15 months independent variation of age, length of
of schooling beyond second-grade entrance. schooling, and season of birth.
Since grades include generally a 1-year The fact that the present data do not
range, the greatest difference in length of show a continuous additive effect of cohort
schooling one would be able to find in chil- levels, that is, not a linear increase in effect
dren of about the same age is, of course, 1 as cohort levels are separated by greater
full year. The differences in schooling for the time intervals, complicates the situation.
cohort levels in the present study amount to One could argue, for instance, that such an
4 and 8 months. outcome suggests the existence of a type of
Considering the differential amounts of periodicity in cohort effects as a function
schooling the various cohorts had (Table of two different kinds of sources. One refers
5), the obtained significant differences be- to the within-year levels of cohort, and the
tween cohorts as well as the interactions be- second refers to between-year levels of co-
tween age and cohort become more mean- hort. In other words, our biological and
ingful. The Subtest GR (arithmetic) is a cultural system may be of such a kind that
subtest which is heavily loaded with ma- individuals born at different times in the
terial that is covered in the grade levels year are provided with distinctive develop-
used; therefore, it makes sense that it is mental conditions which may be different
mainly GR which is affected by cohort dif- from those developmental conditions that
ferences. With regard to the interaction distinguish cohorts born several years apart,
term Age x Cohort, it seems reasonable to that is, the usual generational differences.
assume that the cohort effect is most dis- Such a distinction between within-year co-
tinct with that cohort level which shows the hort differences and between-year cohort
lowest amount of schooling, that is, either differences, however, needs further investi-
7 months in second grade (Study I) or 7 gation.
months in third grade (Study II). The pres-
ent authors, however, cannot offer any ob- Discussion
vious interpretation of the significant inter- In general, the results are not as clear-
action between age and cohort obtained in cut and parsimonious as one might desire,
Study I concerning the Variable BZ. since they give a rather complicated pic-
In general, then, it seems plausible to in- ture of the effects of age and cohort and
terpret the cohort effects found mainly as their interaction in the cognitive develop-
a differential schooling effect. This interpre- ment of children. The outcome, however,
tation is, of course, a rival interpretation for demonstrates the utility of the bifactorial
season-of-birth effects, in general, also, a developmental model and illustrates some
notion which was first touched by Lodge of the pitfalls one has to deal with when
(1938) and subsequently by Williams using single cross-sectional or longitudinal
(1964). As a matter of fact, the notion of studies. For neither of the two subtests af-
differential schooling implies that the con- fected by age and cohort, that is, GR and
flicting evidence on season-of-birth effects BZ, would the application of single cross-
may be only a result of different times of sectional or longitudinal studies have re-
measurements and different terms of sulted in valid and generalizable age effects.
school entries in various countries and As has been shown elsewhere (Baltes,
states. However, the present study can only 1968), single cross sections would have
suggest such an alternative interpretation yielded age effects which are a confounded
176 PAUL B. BALTES AND GUENTHER REINERT

effect of age and cohort differences, and vestigation to include additional levels of
any single cohort-specific longitudinal study age and cohort and other measurement var-
would not have been able to indicate the re- iables in order to check further into the
strictions one has to consider when general- empirical relevance of the cohort factor in
izing the cohort-specific age effects to the study of age-related phenomena.
other levels of cohort. Furthermore, it is
evident that even in investigating such a REFERENCES
small range of age and cohort, cross- BALTES, P. B. Sequenzmodelle zum Studium von
sectional and longitudinal methods do not Altersprozessen: Querschnitt- und Laengsschnitt-
sequenzen. In F. Merz (Ed,), Bericht ueber den
necessarily result in equivalent age trends, 25. Kongress der Deutschen Gesellschaft fuer
since both conventional methods are af- Psychologic in Muenster 1966. Goettingen:
fected differently by existing cohort differ- Hogrefe, 1967.
ences. BALTES, P. B. Longitudinal and cross-sectional se-
In addition, the present results appear to quences in the study of age and generation ef-
fects. Human Development, 1968, 11, 145-171.
have some significance for more specific BELL, R. Q. Convergence: An accelerated longi-
problems in developmental psychology. On tudinal approach. Child Development, 1953, 24,
the one hand, for instance, the results sug- 145-152.
gest that the emphasis in the interpretation CRADDICK, R. A. WISC and WAIS as a function of
season of birth. Psychological Reports, 1966, 18,
of season-of-birth effects on intelligence 259-264.
should be shifted from birthdate character- DAMON, A. Discrepancies between findings of
istics to variables which are associated with longitudinal and cross-sectional studies in adult
different times of measurement during the life: Physique and physiology. Human Develop-
year. On the other hand, the findings sug- ment, 1965, 8, 16-22.
HORN, W. Begabungstestsystem. B-T-S. Handan-
gest that it might not be sufficient to stand- weisung, Durchfuehrung, Auswertung, Interpre-
ardize an intelligence test for children at tation. Goettingen: Hogrefe, 1956.
one specific time during the year, but that it HORN, W. Leistungspruefsystem. L-P-S. Handan-
might be advisable to perform several weisung fuer die Durchfuehrung, Auswertung
standardizations within 1 year to take into und Interpretation. Goettingen: Hogrefe, 1962.
KUHLEN, R. G. Age and intelligence: The signifi-
account cohort effects. This demand was cance of cultural change in longitudinal vs.
first formulated some decades ago by Lodge cross-sectional findings. Vita humana, 1963, 6,
(1938), but was never really implemented 113-124.
in subsequent standardization procedures LODGE, T. Variation in Stanford-Binet IQs of pre-
school children according to the months in
of intelligence tests for children. which examinations were given. Journal of Psy-
While the implications of the present chology, 1938, 6, 385-395.
findings should not be overgeneralized, the ORME, J. E. Intelligence and season of birth. Brit-
obtained results clearly support the empiri- ish Journal of Medical Psychology, 1962, 35,
cal significance of the bifactorial develop- 233-234.
ORME, J. E. Intelligence, season of birth and cli-
mental model. In other words, in studying matic temperature. British Journal of Psychol-
age-related phenomena one should not be ogy, 1963, 54, 273-276.
satisfied with the application of single con- PINTNER, R., & FORLANO, G. Season of birth and
ventional methods, but one should attempt intelligence. Psychological Bulletin, 1943, 40,
to disentangle age from cohort effects by 25-35.
REINERT, G. Die Differenzierungshypothese der
use of the bifactorial developmental model Intelligenz: Untersuchungen zum Problem der
and the sequential strategies derived there- Invariant der faktoriellen Intelligenzstruktur*
from. While Schaie and Strother (1968, Mannheim: Psychologisches Institut der Uni-
1969) have demonstrated the presence of versitaet, 1969, in press.
REINERT, G., BALTES, P. B., & SCHMIDT, L. R.
cohort effects in age changes of adults using Faktorenanalytische Untersuchungen zur Dif-
7-year intervals, the present results argue ferenzierungshypothese der Intelligenz: Die Lei-
for the consideration of cohort effects in stungsdifferenzierungshypothese. Psychologische
cognitive development of children in deal- Forschung, 1965, 28, 246-300.
RIEGEL, K. Age and cultural differences as deter-
ing with cohort populations separated by minants of word associations: Suggestions for
intervals as small as a few months. It seems their analysis. Psychological Reports, 1965, 16,
a next logical step to extend the present in- 75-78.
COHORT EFFECTS IN COGNITIVE DEVELOPMENT 177
RlEGEL, K. F., RlEGEL, R. M. ( & MEYER, G. SOCIO- of age changes in cognitive behavior. Multivari-
psychological factors of aging: A cohort-sequen- ate Behavioral Research, 1968, 3, 259-293.
tial analysis. Human Development, 1967, 10, SCHAIE, K. W., & STROTHER, C. R. A cross-sequen-
27-56. tial study of age changes in cognitive behavior.
SCHAIE, K. W. A general model for the study of Psychological Bulletin, 1968, 70, 671-680.
developmental problems. Psychological Bulletin, WILLIAMS, P. Date of birth, backwardness and
1965, 64, 92-107. educational organization. British Journal of
Educational Psychology, 1964, 34, 247-255.
SCHAIE, K. W., & STROTHER, C. R. The effect of
time and cohort differences on the interpretation (Received May 2, 1968)

(Continued from page 133)

Determinants of Boys' Perceptions of Verbal Reinforcers: Daniel Solomon* and Judy Yaeger: Re-
search Program in Child Development, 232 East Ohio Street, Chicago, Illinois 60611.
Simplicial Structures of Middle-Class and Lower-Class Pupils' Attitudes toward Teachers: Albert H.
Yee* and Philip J. Runkel: School of Education, University of Wisconsin, 606 State Street,
Madison, Wisconsin 53706.
Role of Measurement Operations in the Acquisition of Conservation: David J. Bearison*: Department
of Psychology, Clark University, Worcester, Massachusetts 01610.
Sibling-Age-Spacing Effects upon Cognition: B. G. Rosenberg* and B. Sutton-Smith: Department of
Psychology, University of California, Berkeley, California 94720.
Solution Process in Children's Matching-to-Sample: Gerald R. Levin* and Daphne J. McDonnell: De-
partment of Psychology and Education, Swarthmore College, Swarthmore, Pennsylvania 19081.
Fruit-Tree Study as a Measure of Associative Thinking and Imagery in Children of Different Ages:
Leonore L. Adler*: American Museum of Natural History, Central Park West at 79th Street,
New York, New York 10024.
Modification of the Life Concept in Children: William R. Looft and Don C. Charles*: Department
of Psychology, Iowa State University of Science and Technology, Ames, Iowa 50010.
Growth Hormone and Environmental Complexity Effects on Behavior in the Rat: Oakley S. Ray*
and Sandra Hochauser: Psychology Research Laboratories, Veterans Administration Hospital, Leech
Farm Road, Pittsburgh, Pennsylvania 15206.
* Asterisk indicates author for whom address is supplied.

You might also like