//img.uscri.be/pth/54e32b526050d2965de20f78f56ba750fdacec7f
La lecture en ligne est gratuite
Le téléchargement nécessite un accès à la bibliothèque YouScribe
Tout savoir sur nos offres
Télécharger Lire

Heritability of some personality source traits : évidence from MAVA design, Maximum Likelihood Analysis, and the OA Battery - article ; n°2 ; vol.81, pg 429-451

De
25 pages
L'année psychologique - Année 1981 - Volume 81 - Numéro 2 - Pages 429-451
Summary
1 221 12-18 years old boys, in pairs, from four family constellations (identical and fraternal twins, brothers and unrelated boys raised logether) plus a fifty covering the general population constellation (1 543 cases) were measured on four personality source traits (UI 21, 25, 32 and 33) by objective (performance) batteries from the 0-A Kit.
Nine équations linked the seven unknown abstract variances (within family genetic, between family threptic, etc.) with the nine observed variances in the standard MA VA model.
The maximum likelihood method was used to solve for the unknown contributing variances, and to test the goodness of fit of the data to three simplifications of MA VA more parsimonious as to parameters. In the case of UI 21, 32 and 33 a more parsimonious model fitted.
The results are compared with the one earlier study and with two different modes of analysis of the present data. There are noticeable dis-crepancies in one or two instances, but it its concluded that an overall view of this and other evidence indicates that heritabilities of all of these traits are very low to moderate. The order of increasing heritability—0 to .37 for Hp—is UI 33, UI 32, UI 21 and UI 25. However, UI 25, Reality-contact-vs-Tensidia, has a between family heritability possibly reaching .62, consistent with its significant association with psychosis. The low heritability of exvia-invia suggests a social inhibition rather than a temperamental origin of the second order factor as such. On the law of coercion to the biosocial mean the evidence of between family negative correlations rbgbtls supportive, though in these traits the within family r's are unusually positive and substantial.
Résumé
A l'aide de batteries extraites du 0-A Kit, on a mesuré quatre traits (UI 21, 25, 32, 33) de source de personnalité chez 1 221 paires de garçons âgés de 12 à 18 ans appartenant à 4 constellations familiales (jumeaux homozygotes et hétérozygotes, frères, garçons élevés ensemble) et chez 1 543 cas représentatifs de la constellation de la population générale.
Neuf équations relient les sept variances théoriques inconnues aux neuf variances observées dans l'analyse de variance standard multivariée (MAVA model).
La méthode du maximum de vraisemblance a été utilisée pour estimer les variances et tester l'ajustement des données à trois modèles d'analyse de variance plus économiques en paramètres. En ce qui concerne les traits UI 21, 32 et 33, on a trouvé un bon ajustement pour un modèle plus économique.
Les données ont été analysées à l'aide de deux méthodes et les résultats ont été comparés à ceux obtenus dans une étude antérieure. On observe des différences notables dans un ou deux cas, mais la conclusion générale est que le degré d'héritabiliié des traits étudiés est faible ou modéré.
23 pages
Source : Persée ; Ministère de la jeunesse, de l’éducation nationale et de la recherche, Direction de l’enseignement supérieur, Sous-direction des bibliothèques et de la documentation.
Voir plus Voir moins

R. B. Cattell
D. C. Rao
L. R. Schmidt
D.S Vaughan
Heritability of some personality source traits : évidence from
MAVA design, Maximum Likelihood Analysis, and the OA
Battery
In: L'année psychologique. 1981 vol. 81, n°2. pp. 429-451.
Citer ce document / Cite this document :
B. Cattell R., Rao D. C., R. Schmidt L., Vaughan D.S. Heritability of some personality source traits : évidence from MAVA
design, Maximum Likelihood Analysis, and the OA Battery. In: L'année psychologique. 1981 vol. 81, n°2. pp. 429-451.
doi : 10.3406/psy.1981.28384
http://www.persee.fr/web/revues/home/prescript/article/psy_0003-5033_1981_num_81_2_28384Abstract
Summary
1 221 12-18 years old boys, in pairs, from four family constellations (identical and fraternal twins,
brothers and unrelated boys raised logether) plus a fifty covering the general population constellation (1
543 cases) were measured on four personality source traits (UI 21, 25, 32 and 33) by objective
(performance) batteries from the 0-A Kit.
Nine équations linked the seven unknown abstract variances (within family genetic, between family
threptic, etc.) with the nine observed variances in the standard MA VA model.
The maximum likelihood method was used to solve for the unknown contributing variances, and to test
the goodness of fit of the data to three simplifications of MA VA more parsimonious as to parameters. In
the case of UI 21, 32 and 33 a more parsimonious model fitted.
The results are compared with the one earlier study and with two different modes of analysis of the
present data. There are noticeable dis-crepancies in one or two instances, but it its concluded that an
overall view of this and other evidence indicates that heritabilities of all of these traits are very low to
moderate. The order of increasing heritability—0 to .37 for Hp—is UI 33, UI 32, UI 21 and UI 25.
However, UI 25, Reality-contact-vs-Tensidia, has a between family heritability possibly reaching .62,
consistent with its significant association with psychosis. The low of exvia-invia suggests a
social inhibition rather than a temperamental origin of the second order factor as such. On the law of
coercion to the biosocial mean the evidence of between family negative correlations rbgbtls supportive,
though in these traits the within family r's are unusually positive and substantial.
Résumé
A l'aide de batteries extraites du 0-A Kit, on a mesuré quatre traits (UI 21, 25, 32, 33) de source de
personnalité chez 1 221 paires de garçons âgés de 12 à 18 ans appartenant à 4 constellations
familiales (jumeaux homozygotes et hétérozygotes, frères, garçons élevés ensemble) et chez 1 543 cas
représentatifs de la constellation de la population générale.
Neuf équations relient les sept variances théoriques inconnues aux neuf variances observées dans
l'analyse de variance standard multivariée (MAVA model).
La méthode du maximum de vraisemblance a été utilisée pour estimer les variances et tester
l'ajustement des données à trois modèles d'analyse de variance plus économiques en paramètres. En
ce qui concerne les traits UI 21, 32 et 33, on a trouvé un bon ajustement pour un modèle plus
économique.
Les données ont été analysées à l'aide de deux méthodes et les résultats ont été comparés à ceux
obtenus dans une étude antérieure. On observe des différences notables dans un ou deux cas, mais la
conclusion générale est que le degré d'héritabiliié des traits étudiés est faible ou modéré.L'Année Psychologique, 1981, 81, 429-452
HE RIT ABILITY
OF SOME PERSONALITY SOURCE TRAITS :
EVIDENCE FROM MAVA DESIGN,
MAXIMUM LIKELIHOOD ANALYSIS,
AND THE OA BATTERY
by R. B. Cattell1, D. C. Rao2,
L. R. Schmidt3 and D. S. Vaughan*
;*■/ INSTITUT
BIRUCTKEQï
iERora
RÉSUMÉ
A Vaide de batteries extraites du 0-A Kit, on a mesuré quatre traits
(UI 21, 25, 32, 33) de source de personnalité chez 1 221 paires de garçons
âgés de 12 à 18 ans appartenant à 4 constellations familiales (jumeaux
homozygotes et hétérozygotes, frères, garçons élevés ensemble) et chez
1 543 cas représentatifs de la constellation de la population générale.
Neuf équations relient les sept variances théoriques inconnues aux
neuf variances observées dans l'analyse de variance standard multivariée
(MAVA model).
La méthode du maximum de vraisemblance a été utilisée pour estimer
les variances et tester l'ajustement des données à trois modèles d'anal
yse de variance plus économiques en paramètres. En ce qui concerne les
trails UI 21, 32 et 33, on a trouvé un bon ajustement pour un modèle plus
économique.
Les données ont été analysées à l'aide de deux méthodes et les résultats
ont été comparés à ceux obtenus dans une étude antérieure. On observe des
différences notables dans un ou deux cas, mais la conclusion générale est
que le degré d'héritabiliié des traits étudiés est faible ou modéré.
1. Psychology Department, University of Illinois.
2. Population Genetics Department, University of Hawaii.
3. Psychiatry Universität of Trier.
4. Psychology University of Texas. R. B. Cattell et al. 430
INTRODUCTION
1. Objectives of the research
Strategic behavior genetics research begins with two ques
tions: (1) On what traits is it most important, for theory and for
psychological practice, first to establish degrees of inheritance?
and, (2) Which of the two main analytical methods, applicable
to humans, namely, the twin method and the MAVA method,
is more effective?
The first surely receives the answer that we should deal
with factorially-established, functionally-unitary, source traits
rather than the infinity of specific, narrow, situationally-tied
behaviors, concerning which chance of agreement of operation
and pursuit of replication by different psychologists are remote.
The bases of source trait patterns here have been fully discussed
by Nesselroade and Delhees (1966). In this case the patterns have
been checked (Cattell, Schmidt and Pawlik, 1973) as meaningful
and sufficiently constant across different countries. They have
numerous "non-test" criterion associations (Cattell and Schuerger,
1978). The second — the need for a more powerful method of
analysis — calls for use of the MAVA (Multiple Abstract Variance
"lame" twin method, wherever Analysis) rather than the relatively
data resources permit.
The present article is one of a series of investigations by MAVA
on the heritability of the principal primary personality source
traits. Half of them (Cattell, Rao, Schuerger and Vaughan, 1981 a
and b; Cattell, Schuerger and Klein, 1981; Cattell, Schmidt, Klein
and Schuerger, 1980; Cattell, Rao, Vaughan and Ahern, 1981;
Cattell, Klein, Graham and Kameoka, 1981) are
based on measurement by the objective tests of the O-A Battery.
The other half (Cattell, Schuerger, Klein and Ahern, 1981;
Cattell, Rao and Schuerger, 1981; Cattell, Kameoka, Klein and
Schuerger, 1981; and Kameoka, 1980) are based on questionnaire
scale measurement as in the 16PF (Cattell, 1973 {b)) and HSPQ
(Schumacher and Gattell, 1974). The heritability values across
these seven or eight studies are comparable in virtue of identity
of method and identity or similarity also of population samples.
They are published separately because together they would
overwhelm journal space, but particularly because different sets of personality traits 431 Hcritability
of personality traits interest psychologists in different areas and
it is important to have space for discussion of the implications
of the bare genetic results for personality theory.
This article concentrates on the results for four major per
sonality primaries already widely studied (Gattell and Schuerger,
1978; Schmidt, Hacker and Cattell, 1975) as to other properties
and criterion relations in clinical and educational psychology
(Gattell and Bjersted, 1967; Gattell, Schimdt and Bjersted, 1972;
Gattell and Schuerger, 1978; Eysenck and Eysenck, 1968;
Hundleby, Pawlik and Gattell, 1965; Ishikawa, 1977; Killian
and Cattell, 1967; Schmidt, Hacker and Gattell, 1975; Tatro,
1968). The titles used, adjusted to the present theories concerning
their natures, are however still to be regarded as tentative. For
this reason universal index (UI) numbers have meanwhile been
used to anchor them, simply as stable factor patterns in objective
test behaviors.
UI 21, Exuberance, is marked by quick and confident decisions,
high fluency, capacity to complete uncompleted pictures, and
other variables shown in Table I. Neurotics are significantly
below average on UI 21 (Gattell and Scheier, 1961) and schizo
phrenics still more so (Gattell, Schmidt and Bjersted, 1972).
UI 25 has been called, in its negative direction, psychoticism, by
Eysenck and Eysenck (1968), but Cattell and his associates,
notably in the several studies by Killian and Cattell (1967);
Tatro (1968); Cattell, Schmidt and Bjersted (1972) and others,
have shown that it is as powerful in distinguishing neurotics,
drug addicts, etc., as psychotics, and have found more special
pathological associations to it. From this and other evidence
Cattell has given it the more specific theoretical meaning of the
tital Realism (Reality contact) versus Tensidia. The latter,
negative pole defines a tense inner subjective preoccupation present
in various pathologies but perhaps outstandingly in schizo
phrenics (who, nevertheless, also differ significantly on at least
four other source traits). It correlates (in the positive direction,
Realism) with school performance and job performance, intell
igence being partialled out. UI 32 has been confidently identified
with extraversion (according to the content of the second order
questionnaire factor, QI which it matches in cross media factor
ings). In order to set Protean popular descriptions of extraversion
aside, the core factor pattern in U I 32 has been re-named exvia-vs-
invia. UI 33 is less explored than the others, being lower in 432 E. B. Caiiell et al.
Table I. — Source Traits Involved
and the Chief Subtesis Used in Obtaining
Scores on Them
Master Index
Nürnberg) Direction and nature of performance
UI 21 Exuberance
MI 3356 Faster speed following directions
MI 271 Higher ideational fluency
MI 853 More concrete completion of drawings
MI 699garbled words guessed in tautophone
MI 7 Faster closure in perceiving incomplete drawings
MI 8 High frequency of alternating perspective
UI 25 Realism-us-Tensidia
MI 1006 Less pessimistic insecurity in life attitudes
MI 2411 Better immediate memory
MI 2408 Great accuracy in digit span reproduction
MI 144 More agreement with homely wisdom perspectives
UI 32 Exvia-vs-Invia
MI 733 Greater willingness to decide on vague data
MI 1169 Less influenced on assigned punishments by extenuating
circumstances
cards" in the CMS (Cursive MI 15 More excessive use of "trump
miniature situation) test
MI 2 a Lower motor-perceptual rigidity
UI 33 Discouragement-vs-Sanguine Temperament
MI 108 Less confidence in unfamiliar situations
MI 473 Fewer people who appreciate one as a friend written down
MI 1245 Higher score on depression scale
MI 112 Greater expectation of unfavorable consequences of action
The scope of this article is too wide to permit detailed description of
these tests and indeed these are not all, for 6 to 8 were used for each factor.
The full tests and descriptions are available in Cattell and Schuerger,
1978, and Cattell and Warburton, 1967.
(x) The Master Index numbers are those used in some 30 researches and
organized with test numbers (locating the test as distinct from the behavior
measured) in Cattell and Warburton's compendium (1967).
Concept (construct) validities of batteries against
the factor itself
UI 21 Exuberance .80
UI 25 Realism-vs-Tensidia .74
UI 32 Exvia-vs-Invia .71
UI 33 Discouragement-vs-Sanguiness .85 of personality traits 433 Herilability
variance, but by content (see Table I) and its association (nega
tively) with clinical depression, it has well been called sanguine
temperament. It is indeed not distinguishable from Galen's
description, except that we have today actual performance tests
with unique loadings to define it, as shown in Table I.
OUTLINE OF THE MULTIPLE ABSTRACT VARIANCE (mAVAJ METHOD
Let us next turn from the measures to the experimental and
analytical method. First one must recognize that the MAVA
method differs fundamentally from the more limited and far
more universally used twin method. The latter, in most actual
practice, commonly compares simply the variance of identical
and fraternal twins (or sibs), since identical twins raised apart
have never been gathered in sufficient samples to extend the
method. See the excellent survey of the twin method and its
results by Mittler (1971).
MAVA differs by using the variances (between children) of a
whole series of family "constellations"; identical twins, fraternal
twins, sibs raised together, sibs raised apart unrelated, children
reared together, half-sibs, unrelated children raised apart (general
population) and so on. These are listed in Table II.
Table II. — Definition of 9 concrete observed
variance components
No. Symbol Definition
1 s^ Between identical (MZ) twins raised together
2 sFTTfraternal (DZ)
3 s|T Between sibs raised together
4 sl„runrelated boys raised together
5 «Iittf Between families of MZ twins
6 sbfttfof DZ
7 s|NF Between families of sibs
8 s|SFof unrelated boys
9 s|p General population variance
An equation is then set up in the MAVA model for each such
empirically obtained concrete variance accounting for it in terms 434 B. B. Callell et al.
of a combination of so-called abstract variances such as within
family genetic variance, a^, within family threptic variance,
o^t , and so on. These are listed in Table III.
Table III. — Definition
of 7 abstract variance components
No. Symbol Definition
1 a^j t Within family threptic variance, for twins
2 a*genetic variance
3 rwgwt Correlation of and threptic deviations
within the family
4 o^j t Within family threptic variance, for sibs
5 o|t Between variance
6 a\family genetic
7 rbgbt Correlation of and threptic deviations
between families
MA VA is now well accepted among geneticists as a deve
lopment of biométrie genetics particularly suited to human
families and psychological analysis. One would hope that the
model and its assumptions and advances have been sufficiently
discussed elsewhere, Cattell (1953, 1960, 1963, 1965, 1973 (a),
1981 (a), Loehlin (1965), Jinks and Fulker (1970), Vandenberg
(1965) to permit us to pass on here, but as some reviewers have
raised Loehlin's (1965) criticisms a moment's digression is neces
sary. MA VA is not so much a single model as a family of models,
and different sets of equations have been set up according to the
number of family constellations available. Loehlin's criticisms
(below) apply to more limited sets of equations, and the present
writers' position is that in those circumstances admitted approxi
mations have to be made. The limiting assumptions and the
tangible advantages over twin, adoptive and other "fragmentary
MA VA" method can be briefly listed.
(1) Like all other actually used models it supposes additive
action of genetic and environmental effects.
(2) Nevertheless it considers the full role of possible covari-
ances between genetic and environmental effects lacking from
virtually all twin method studies. However, as Loehlin correctly
points out it assumes that one of these correlations, namely that of personality trails 435 Heritability
of the within family genetic deviation of one sib with the within
family environmental deviation of another is zero. This is not
soluble as non-zero with the present equation set, but we point
out (a) that it is likely to be very small and (6) it wil be still
smaller when there are several sibs.
(3) Where the same covariance-causing effects exist in family
constellations with different actual variance magnitudes one can
assume either that (a) the correlation or (b) the regression
coefficient will stay the same. Loehlin favors the latter and
points out that the solution for the separate terms in a^ + a^t
+ ^rwawt °w awt *s then not possible. The choice is debatable
and we have followed (a).
(4) Surprise has been expressed at the 2 y2 coefficient in
Equation (5) Table IV. It will be seen that where single variances
are involved in the covariance, as in Equation (2) the expression
2r is used, and when double size variances are involved in both
the terms in covariance, as in Equation (6), the coefficient
(2 V 2 X 2) is 4. In Equation (5) with one double and one single
it follows that it is 2 V 2. (This was accepted by both R. G. Rao
and R. A. Fisher.)
(5) Among its advantages are (a) that it does not accept the
assumption of equality of within family environmental variances
for sibs and twins, which our results elsewhere (Cattell, 1981)
show to be quite untenable (b) that it supplies evidence on the
covariances (c) that it yields three heritabilities: within family
Hw, between family Hp, and total population, Hp . A general
expression of this comparison with other behavior genetic
methods is given elsewhere (Cattell, 1981 (a)).
In the present case we were able to get data for what has been
called the Limited MAVA, with constellations (1) ITT, identical
twins raised together (2) FTT, fraternals together (3) ST, sibs
together (4) UT, unrelated boys raised together, and (5) GP the
general population.
With these (sample sizes given in Table V) we can set up nine
equations (Table IV) from which we hope to solve for seven
unknowns III).
Parenthetically, from here on we shall substitute the Greek-
derived term threptic for the clumsy "environmentally pro
duced" — as our co-workers have also done in a dozen other
publications. The habit of speaking of "environmental variance" R. B. Caliell et al. 436
obfuscates discussion by confusing the variance in environment
itself, e.g. in years of education, temperature, social status, with
the threptic variance which the environment generates in the
given irait. Later we may wish, in the comparative MAVA method
(Cattell, 1973 (a)) to relate the threptic variance in a given trait
to particular environmental variances that have produced it, and
if these are verbally and conceptually confused from the begin
ning we make a bad start. For example, we cannot then use the
concept of "nurturability" or "culturability" of a trait, which is
the ratio of threptic to environmental variance. Incidentally, this
possible investigation of threptic-environmental correlations
(and partial correlations) under the concept of "teachability",
"culturability", or "nurturability" should remind psychologists
that behavior genetic research findings are quite as important
an element in cultural learning theory as in behavior genetics
(Cattell, 1981 (a), Vol. 2).
DATA GATHERING AND THE NATURE OF SAMPLES
Our research aimed to get personality measures, on the four
traits listed in Table I, on pairs of children in 7 constellations
(including the general population, i.e. the randon variance). Due
to some unfortunate social restrictions on this research, two
categories — brothers raised apart and the half-brothers — finished
up too small for use and so only the 5 constellations in Table II
were finally used. Incidentally, since identical twins are same
sex, we confined the whole study to males, e.g. brothers raised
together, etc. Hopefully another investigator will repeat with
girls. The subjects were obtained by going through the
schools — one high school in the down state Illinois small towns;
random choice in Chicago and Cleveland — asking in classes for
those with brothers in the school or supposed twins. Brothers
were checked by reciprocity and school records; twins were color-
photographed on a grid, and information was obtained (Klein and
Cattell, 1977) by Klein, experienced in twin diagnosis, on birth
delivery, etc. (At this time blood typing was not permitted by
regulations and the decision rested on a summation of identities.)
The administration of tests, in two sessions of 1£ hours each,
was done in a classroom situation, but mainly in small groups of
four to six. Scoring is as given in the Handbook (Cattell and
Schuerger, 1978).