TRAITS REVISITED

GORDON W. ALLPORT
Harvard University

Y EARS ago I ventured to present a paper cow on the doormat: John has a trait of aggres-
before the Ninth International Congress at sion. The result is the fallacy of misplaced con-
New Haven (G. W. Allport, 1931). It was creteness.
entitled "What Is a Trait of Personality?" For me The general positivist cleanup starting in the
to return to the same topic on this honorific oc- 1930s went even further. It swept out (or tried
casion is partly a sentimental indulgence, but partly to sweep out) all entities, regarding them as ques-
too it is a self-imposed task to discover whether tion-begging redundancies. Thus Skinner (1953)
during the past 36 years I have learned anything writes:
new about this central problem in personality When we say that a man eats because he is hungry, smokes
theory. a great deal because he has the tobacco habit, fighls be-
In my earlier paper I made eight bold assertions. cause, of the instinct of pugnacity, behaves brilliantly be-
A trait, I said, cause of his intelligence, or plays the piano well because of
his musical ability, we seem to be referring to causes. But
1. Has more than nominal existence. on analysis these phrases prove to be merely redundant
2. Is more generalized than a habit. descriptions [p. 31],
3. Is dynamic, or at least determinative, in be- It. is dear that this line of attack is an assault
havior. not only upon the concept of trait, but upon all
4. May be established empirically. intervening variables whether they be conceived in
5. Is only relatively independent of other traits. terms of expectancies, attitudes, motives, capacities,
6. Is not synonymous with moral or social judg- sentiments, or traits. The resulting postulate of
ment. the "empty organism" is by now familiar to us all,
7. May be viewed either in the light of the person- and is the scientific credo of some. Carried to its
ality which contains it, or in the light of its logical extreme this reasoning would scrap the con-
distribution in the population at large. cept of personality itself—an eventuality that seems
merely absurd to me.
To these criteria I added one more:
More serious, to my mind, is the argument
8. Acts, and even habits, that are inconsistent with against what Block and Bennett (19SS) called
a trait are not proof of the nonexistence of the "traitology" arising from many studies of the vari-
trait. ability of a person's behavior as it changes from
situation to situation. Every parent knows that an
While these propositions still seem to me de-
offspring may be a hellion at home and an angel
fensible they were originally framed in an age of
when he goes visiting. A businessman may be
psychological innocence. They now need reexami-
hardheaded in the office and a mere marshmallow
nation in the light of subsequent criticism and re-
in the hands of his pretty daughter.
search.
Years ago the famous experiment by La Piere
(1934) demonstrated than an innkeeper's prejudice
CRITICISM OF THE CONCEPT OP TRAIT
seems to come and go according to the situation
Some critics have challenged the whole concept confronting him.
of trait. Carr and Kingsbury (1938) point out the In recent months Hunt (1965) has listed various
danger of reification. Our initial observation of theories of personality that to his mind require
behavior is only in terms of adverbs of action: revision in the light of recent evidence. Among
John behaves aggressively. Then an adjective them he questions the belief that personality traits
creeps in: John has an aggressive disposition. Soon are the major sources of behavior variance. He,
a heavy substantive arrives, like William James' like Miller (1963), advocates that we shift atten-
1
2 AMERICAN PSYCHOLOGIST

lion from traits to interactions among people, and raphy deals with distal forces—personality study
look for consistency in behavior chiefly in situation- with proximal forces. The fact that the inn-
ally defined roles. Helson (1964) regards trait as keeper's behavior varies according to whether he
the residual effect of previous stimulation, and thus is, or is not, physically confronted with Chinese
subordinates it to the organism's present adapta- applicants for hospitality tells nothing about his
tion level. attitude structure, except that it is complex, and
Scepticism is likewise reflected in many investiga- that several attitudes may converge into a given
tions of "person perception." To try to discover act of behavior.
the traits residing within a personality is regarded Nor does it solve the problem to explain the
as either naive or impossible. Studies, therefore, variance in terms of statistical interaction effects.
concentrate only on the process of perceiving or Whatever tendencies exist reside in a person, for a
judging, and reject the problem of validating the person is the sole possessor of the energy that
perception and judgment. (Cf. Tagiuri & Petrullo, leads to action. Admittedly different situations
1958.) elicit differing tendencies from my repertoire. I
Studies too numerous to list have ascribed chief do not perspire except in the heat, nor shiver ex-
variance in behavior to situational factors, leaving cept in the cold; but the outside temperature is
only a mild residue to be accounted for in terms not the mechanism of perspiring or shivering. My
of idiosyncratic attitudes and traits. A prime ex- capacities and my tendencies lie within.
ample is Stouffer's study of The American Soldier To the situationist I concede that our theory
(Stouffer et al., 1949). Differing opinions and of traits cannot be so simpleminded as it once
preferences are ascribed so far as possible to the was. We are now challenged to untangle the
GJ's age, martial status, educational level, location complex web of tendencies that constitute a per-
of residence, length of service, and the like. What son, however contradictory they may seem to be
remains is ascribed to "attitude." By this pro- when activated differentially in various situations.
cedure personality becomes an appendage to demog-
raphy (see G. W. Allport, 19SO). It is not the ON THE OTHER HAND
integrated structure within the skin that determines In spite of gunfire from positivism and situation-
behavior, but membership in a group, the person's ism, traits are still very much alive. Gibson
assigned roles—in short, the prevailing situation. (1941) has pointed out that the "concept of set or
It is especially the sociologists and anthropologists attitude is nearly universal in psychological think-
who have this preference for explanations in terms ing." And in an important but neglected paper-—
of the "outside structure" rather than the "inside perhaps the last he ever wrote—McDougall (1937)
structure" (cf. F. H. Allport, 19SS, Ch. 21). argued that tendencies are the "indispensable postu-
I have mentioned only a few of the many lates of all psychology." The concept of trait falls
varieties of situationism that flourish today. While into this genre. As Walker (1964) says trait,
not denying any of the evidence adduced I would however else defined, always connotes an enduring
point to their common error of interpretation. If tendency of some sort. It is the structural counter-
a child is a hellion at home, an angel outside, he part of such functional concepts as "expectancy,"
obviously has two contradictory tendencies in his and "goal-directedness."
nature, or perhaps a deeper genotype that would After facing all the difficulties of situational and
explain the opposing phenotypes. If in studies of mood variations, also many of the methodological
person perception the process turns out to be com- hazards such as response set, halo, and social de-
plex and subtle, still there would be no perception sirability, Vernon (1964) concludes, "We could
at all unless there were something out there to go a long way towards predicting behavior if we
perceive and to judge. If, as in Stouffer's studies, could assess these stable features in which people
soldiers' opinions vary with their marital status or differ from one another [p. 181]." The powerful
length of service, these opinions are still their own. contributions of Thurstone, Guilford, Cattell, and
The fact that my age, sex, social status help form Eysenck, based on factor analysis, agree that the
my outlook on life does not change the fact that search for traits should provide eventually a satis-
the outlook is a functioning part of me. Demog- factory taxonomy of personality and of its hier-
 TRAITS REVISITED

archical structure. The witness of these and other regardless of our shortcomings in comprehending
thoughtful writers helps us withstand the pessi- it. Since traits, like all intervening variables, are
mistic attacks of positivism and situationism. never directly observed but only inferred, we must
It is clear that I am using "trait" as a generic expect difficulties and errors in the process of dis-
term, to cover all the "permanent possibilities for covering their nature.
action" of a generalized order. Traits are cortical, The incredible complexity of the structure we
subcortical, or postural dispositions having the ca- seek to understand is enough to discourage the
pacity to gate or guide specific phasic reactions. realist, and to tempt him to play some form of
It is only the phasic aspect that is visible; the positivistic gamesmanship. He is tempted to settle
tonic is carried somehow in the still mysterious for such elusive formulations as: "If we knew
realm of neurodynamic structure. Traits, as I enough about the situation we wouldn't need the
am here using the term, include long-range sets and concept of personality"; or "One's personality is
attitudes, as well as such variables as "perceptual merely the way other people see one"; or "There
response dispositions," "personal constructs," and is no structure in personality but only varying de-
"cognitive styles." grees of consistency in the environment."
Unlike McClelland (1951) I myself would re- Yet the truly persistent realist prefers not to
gard traits (i.e., some traits) as motivational abandon his commitment to find out what the other
(others being merely stylistic). I would also insist fellow is really like. He knows that his attempt
that traits may be studied at two levels: (a) di- will not wholly succeed, owing partly to the com-
mensionally, that is as an aspect of the psychology plexity of the object studied, and partly to the
of individual differences, and (b) individually, in inadequacy of present methods. But unlike Kant
terms of personal dispositions, (Cf. G. W. Allport, who held that the Ding an Sick is doomed to re-
1961, Ch. 15.) It is the latter approach that brings main unknowable, he prefers to believe that it is at
us closest to the person we are studying. least partly or approximately knowable.
As for factors, I regard them as a mixed bless- I have chosen to speak of heuristic realism, be-
ing. In the investigations I shall soon report, fac- cause to me special emphasis should be placed on
torial analysis, I find, has proved both helpful and empirical methods of discovery. In this respect
unhelpful. My principal question is whether the heuristic realism goes beyond naive realism.
factorial unit is idiomatic enough to reflect the Taking this epistemological point of view, the
structure of personality as the clinician, the coun- psychologist first focuses his attention on some
selor, or the man in the street apprehends it. limited slice of personality that he wishes to study.
Or are factorial dimensions screened so extensively He then selects or creates methods appropriate to
and so widely attenuated—through item selection, the empirical testing of his hypothesis that the
correlation, axis manipulation, homogenization, cleavage he has in mind is a trait (either a di-
and alphabetical labeling—-that they impose an mensional trait or a personal disposition). He
artifact of method upon the personal neural net- knows that his present purposes and the methods
work as it exists in nature? chosen will set limitations upon his discovery. If,
however, the investigation achieves acceptable
A HEURISTIC REALISM
standards of validation he will have progressed far
This question leads me to propose an eptistemo- toward his identification of traits. Please note, as
logical position for research in personality. Most with any heuristic procedure the process of dis-
of us, I suspect, hold this position although we covery may lead to important corrections of the
seldom formulate it even to ourselves. It can be hypothesis as originally stated.
called a heuristic realism. Empirical testing is thus an important aspect
Heuristic realism, as applied to our problem, of heuristic realism, but it is an empiricism re-
holds that the person who confronts us possesses strained throughout by rational considerations.
inside his skin generalized action tendencies (or Galloping empiricism, which is our present occu-
traits) and that it is our job scientifically to dis- pational disease, dashes forth like a headless horse-
cover what they are. Any form of realism assumes man. It has no rational objective; uses no rational
the existence of an external structure ("out there") method other than mathematical; reaches no ra-
 AMERICAN PSYCHOLOGIST

lional conclusion. It lets the discordant data sing types of subjective evaluation or Lebensjormen.
for themselves. By contrast heuristic realism says, Adopting this rational starting point we ourselves
"While we are willing to rest our case for traits took the second step, to put the hypothesis to
on empirical evidence, the area we carve out for empirical test. We asked: Are the six dimensions
study should be rationally conceived, tested by proposed—the theoretic, the economic, the esthetic,
rational methods; and the findings should be social, political, and religious—measurable on a
rationally interpreted." multidimensional scale? Are they reliable and
valid? Spranger defined the six ways of looking
THREE ILLUSTRATIVE STUDIES at life in terms of separate and distinct ideal types,
although he did not imply that a given person
It is now time for me to illustrate my argument
belongs exclusively to one and only one type.
with sample studies. I have chosen three in which
It did not take long to discover that when con-
I myself have been involved. They differ in the
fronted with a forced-choice technique people do
areas of personality carved out for study, in the
in fact subscribe to all six values, but in widely
methods employed, and in the type of traits estab-
varying degrees. Within any pair of values, or any
lished. They are alike, however, in proceeding
quartet of values, their forced choices indicate
from the standpoint of heuristic realism. The
a reliable pattern. Viewed then as empirical con-
presentation of each study must of necessity be
tinua, rather than as types, the six value directions
woefully brief. The first illustrates what might
prove to be measurable, reproducible, and con-
be called meaningful dimensionalism; the second
sistent. But are they valid? Can we obtain ex-
meaningful covariation; the third meaningful
ternal validation for this particular a priori concep-
morphogenesis.
tion of traits? The test's Manual (Allport &
Vernon, 1931) contains much such evidence. Here
Dimensions of Values
I would add a bit more, drawn from occupational
The first illustration is drawn from a familiar studies with women subjects. (The evidence for
instrument, dating almost from the stone age, The men is equally good.) The data in Table 1 are
Study of Values (Allport & Vernon, 1931). While derived partly from the Manual, partly from
some of you have approved it over the years, and Guthrie and McKendry (1963) and partly from
some disapproved, I use it to illustrate two im- an unpublished study by Elizabeth Moses.
portant points of my argument. For present purposes it is sufficient to glance at
First, the instrument rests on an a priori analy- the last three columns. For the theoretic value
sis of one large region of human personality, we note that the two groups of teachers or teachers
namely, the region of generic evaluative tendencies. in preparation select this value significantly more
It seemed to me 40 years ago, and seems to me often than do graduate students of business ad-
now, that Eduard Spranger (1922) made a per- ministration. Conversely the young ladies of
suasive case for the existence of six fundamental business are relatively more economic in their
choices. The results for the esthetic value prob-
TABLE 1 ably reflect the higher level of liberal arts back-
MEAN SCORES FOR OccurATTONAT. GROUPS OT WOMEN: ground for the last two groups. The social
STUDY OT VALUES (philanthropic) value is relatively low for the busi-
ness group, whereas the political (power) value is
Female Graduate Graduate relatively high. Just why nurses should more often
collegiate nurses students of Peace Corps
norms training for business ad-
ministration
teachers endorse the religious value is not immediately clear.
teaching
N = 2,475 N = 328 N = 77 2V = 1.11 Another study of external validation, showing the
Theoretical 36.5 40.2 37.3 40.6
long-range predictive power of the test is an un-
Economic 36,8 32.9 40.4 29.9 published investigation by Betty Mawardi. It is
Esthetic 43.7 43.1 46.8 49.3 based on a follow-up of Wellesley graduates IS
Social 41.6 40.9 35.0 41.2 years after taking the Study of Values.
Political 38.0 37.2 41.8 39.7
Religious 43.1 45.7 38.7 39.2
Table 2 reports the significant deviations (at the
5% level or better) of various occupational groups
 TRAITS REVISITED

TABLE 2
SIGNIFICANT DEVIATIONS OP SCORES ON THE STUDY OE VALUES KOR OCCUPATIONAL GROUPS
ov WELLESLEY ALUMNI EROM WELLESLEY MEAN SCORES

Occupational groups N Theoretical Economic Esthetic Social Political Religious

Business workers 64 Lower Higher

Medical workers 42 Higher Lower Lower
Literary workers 40 Higher Lower Higher
Artistic workers 37 Higher Lower
Scientific workers 28 Higher Lower
Government workers 24' Higher Lower Lower
Social workers 26 Higher
Religious workers 11 Lower' Higher

from the mean scores of Wellesley students. In factorsL_are_rj.eeded==som6~,LhaLiS!r.e..n.eed'-m'ore. And

virtually every case we find the deviation mean- in all cases the clusters that emerge seem strange
ingful (even necessary) for the occupation in ques- an unnamable. Here is a case, I believe, where
tion. Thus women in business are significantly high our empiricism should submit to rational restraint.
in economic interests; medical, government, and The traits as defined are meaningful, reliably meas-
scientific workers in theoretical; literary and artistic ured, and validated. Why sacrifice them to gal-
workers in esthetic; social workers in social; and loping gamesmanship?
religious workers in religious values.
One must remember that to achieve a relatively Covariation: Religion and Prejudice
high score on one value, one must deliberately Speaking of covariation I do not mean to imply
slight others. For this reason it is interesting to that in restraining our empirical excesses we should
note in the table the values that are systematically fail to explore the patterns that underlie covaria-
slighted in order to achieve a higher score on the tion when it seems reasonable to do so.
occupationally relevant value. (In the case of Take, for example, the following problem. Many
social workers it appears that they "take away" investigations show conclusively that on the broad
more or less uniformly from other values in order average church attenders harbor more ethnic preju-
to achieve a high social value.) dice than nonattenders. (Some of the relevant
Thus, even at the college age it is possible to studies are listed by Argyle, 1959, and by Wilson,
forecast in a general way modal vocational activity 1960.) At the same time many ardent workers for
IS years hence. As Newcomb, Turner, and Con- civil rights are religiously motivated. From Christ
verse (1965) say, this test clearly deals with "in- to Gandhi and to Martin Luther King we note
'-——jiusfeie values" or with "basic value postures" that equimindedness has been associated with re-
whose generality is strikingly broad. An evaluative ligious devoutness. Here then is a paradox: Re-
posture toward life saturates, or guides, or gates ligion makes prejudice; it also unmakes prejudice.
(choose your own metaphor) specific daily choices First we tackle the problem rationally and form
over a long expanse of years. a hypothesis to account for what seems to be a
One reason I have used this illustration of trait curvilinear relation. A hint for the needed hy-
research is to raise an important methodological pothesis comes from The Authoritarian Personality
issue. The six values are not wholly independent. (Adorno, Frenkel-Brunswik, Levinson, & Sanford,
There is a slight tendency for theoretic and esthetic 1950) which suggests that acceptance of institu-
values to covary; likewise for economic and politi- tional religion is not as important as the way in
cal values; and so too with social and religious. which it is accepted. Argyle (1959) sharpens the
Immediately the thought arises, "Let's factor the hypothesis. He says, "It is not the genuinely
whole matrix and see what orthogonal dimensions devout who are prejudiced but the conventionally
emerge." This step has been taken several times religious [p. 84]."
(see Manual); but always with confusing results. In our own studies we have tentatively assumed
Some investigators discover that fewer than six that two contrasting but measurable forms of re-
 AMERICAN PSYCHOLOGIST

ligious orientation exist. The first form we call the This second item is of considerable interest, for
eytrtnsic orientation, meaning that for the church- many studies have found that it is the irregular
goer religious devotion is not a value in its own atlenders who are by far the most prejudiced (e.g.,
right, but is an instrumental value serving the mo- Holtzmann, 1956; Williams, 1964). They take
tives of personal comfort, security, or social status. their religion in convenient doses and do not let it
(One man said he went to church because it was regulate their lives.
the best place to sell insurance.) Elsewhere I have Now for a few illustrative results in Table 3. If
defined this utilitarian orientation toward religion we correlate the extrinsicness of orientation with
more fully (G. W. Allport, 1960, 1963). Here I various prejudice scales we find the hypothesis con-
shall simply mention two items from our scale, firmed. Likewise, as predicted, intrinsicness of
agreement with which we assume indicates the ex- orientation is negatively correlated with prejudice.
trinsic attitude: In view of the difficulty of tapping the two com-
What religion offers me most is comfort when sorrows
plex traits in question, it is clear from these studies
and misfortune strike. that our rationally derived hypothesis gains strong
support. We note that the trend is the same when
One reason for my being a church member is that such different denominations are studied in relation to
membership helps to establish a person in the community.
differing targets for prejudice.
By contrast the intrinsic_onentation regards faith Previously I have said that empirical testing has
as a supreme value in its own right. Such faith the ability to correct or extend our rational analysis
strives to transcend self-centered needs, takes seri- of patterns. In this particular research the follow-
ously the commandment of brotherhood that is ing unexpected fact emerges. While those who
found in all religions, and seeks a unification of approach the intrinsic pole of our continuum are on
being. Agreement with the following items indicates the average less prejudiced than those who approach
an inlrinsic orientation: the extrinsic pole, a number of subjects show them-
selves to be disconcertingly illogical. They accept
My religious beliefs arc what really lie behind my whole both intrinsically worded items and extrinsically
approach to life. worded items, even when these are contradictory,
If not prevented by unavoidable circumstances, I attend such as:
church, on the average (more than once a week) (once a
My religious beliefs are what really lie behind my whole
week) (two or three times a month) (less than once a
approach to life.
month).
Though I believe in my religion, I feel there are many
more important things in my life.
TABLK 3
CORKKLATIONS I1ETWEKN MEASURES OT? RELIGIOUS It is necessary, therefore, to inspect this sizable
ORIENTATION AMONG CHURCHGOERS AND VARIOUS group of muddleheads who refuse to conform to our
PREJUDICE SCALES neat religious logic. We call them "inconsistently
proreligious." They simply like religion; for them
Denominational sample N r
it has "social desirability" (cf. Edwards, 1957).
Unitarian 50 The importance of recognizing this third mode of
Intrinsic—anti-Catholicism .56 religious orientation is seen by comparing the
Fnt rinsic— anl i-Catholicism -.36 prejudice scores for the groups presented in Table
Intrinsic- anti-Mexican .54 4. In the instruments employed the lowest possible
Intrinsic—anti-Mexican - .42
prejudice score is 12, the highest possible, 48. We
Catholic 66
Kxliinsic -anti-Negro .36 note that the mean prejudice score rises steadily
Intrinsic—anti -Negro -.49 and significantly from the intrinsically consistent to
Nazarene 39 the inconsistently proreligious. Thus subjects with
Kxtrinsic- anti-Negro .41 an undiscrlounated,prcu^gious-i£spofl^
Fill rinsic— ant i-Ncgro -.44
Mixed"
the ayccagejnjost-ptejwliced of all.
207
Kxl rinsic—anti-Semitic .65 Having discovered the covariation of prejudice
with both the extrinsic orientation and the "pro"
' From response set, we are faced with the task of rational
 TRAITS REVISITED

trait names, and if these may be used in combina-

oi' R K . U C I O I I S ORIKNTATION A N D tions, there is no real point in arguing that the use
MKAN P R K J U D I C K SCORES of the available lexicon of a language necessarily
makes all trait studies purely nomothetic (dimen-
Mean prejudice scores sional).
Moderately Extremely
A series of 172 published Letters jrom Jenny (G.
Consistently Consistently inconsistent inconsistent W. Allport, 196S) contains enough material for a
intrinsic extrinsic (prorcligion) (proretigion)
rather close clinical characterization of Jenny's
Anti-Negro 28,7 33.0 35.4 37.9 personality, as well as for careful quantitative and
Anti-Semitic 22.6 24.6 28.0 30.1 computational analysis. While there is no possi-
Noto.—^V = 309, mixed denominations. All differences significant
bility in this case of obtaining external validation
at .01 level. for the diagnosis reached by either method, still by
employing both procedures an internal agreement
explanation. One may, I think, properly argue that is found which constitutes a type of empirical vali-
these particular religious attitudes aie_JnsjTum£0£al dation for the traits that emerge.
in nature; they provide safety, security, and status The clinical method in this case is close to com-
•—all within a self-serving frame. Prejudice, we mon sense. Thirty-nine judges listed the essential
know, performs much the same function within characteristics of Jenny as they saw them. The
some personalities. The needs for status, security, result was a series of descriptive adjectives, 198 in
comfort, and a feeling of self-rightness are served number. Many of the selected trait names were
by both ethnic hostility and by tailoring one's re- obviously synonymous; and nearly all fell readily
ligious orientation to one's convenience. The econ- into eight clusters.
omy of other lives is precisely the reverse: It is The quantitative analysis consisted of coding the
their religion that centers their existence, and the letters in terms of 99 tag words provided by the
only ethnic attitude compatible with this intrinsic lexicon of the General Inquirer (Stone, Bales, Na-
orientation is one of brotherhood, not of bigotry. menwirth, & Ogilvie, 1962). The frequency with
This work, along with the related investigations which these basic tag words are associated with one
of Lenski (1963), Williams (1964), and others, another in each letter forms the basis for a factor
signifies that we gain important insights when we analysis (see G. W. Allport, 1965, p. 200).
refine our conception of the nature of the religious
Table S lists in parallel fashion the clusters ob-
sentiment and its functions. Its patterning prop-
tained by clinical judgment based on a careful read-
erties in the economy of a life are diverse. It can
ing of the series, along with the factors obtained by
fuse with bigotry or with brotherhood according to
Jeffrey Paige in his unpublished factorial study.
its nature.
In spite of the differences in terminology the
As unfinished business I must leave the problem
of nonattenders. From data available it seems that
the unchurched are less prejudiced on the average TAHLR5
than either the extrinsic or the inconsistent church- CENTRAL TKAITS IN JENNY'S PEKSONALITY AS
goers, although apparently more prejudiced on the DETKHMINED BY Two METHODS

average than those whose religious orientation is

Common-sense ti'aits B'actorial traits
intrinsic. Why this should be so must form the
topic of future research. Quarrelsome-suspicious^
Aggressive j Aggression
Personal Dispositions: An Idiomorphic Approach Self-centered (possessive) Possessivencss
Sentimental /Need for affiliation
The final illustration of heuristic realism has to \Need for family acceptance
do with the search for the natural cleavages that Independent-autonomous Need for autonomy
mark an individual life. In this procedure there is Esthetic-artistic Sentience
no reference to common dimensions, no comparison Self-centered (self-pif ying) Martyrdom
(No parallel) Sexuality
with other people, except as is implied by the use of Cynical-morbid (No parallel)
the English language. If, as Allport and Odbert Dramatic-intense ("Overstate")
(1936) have found, there are over 17,000 available
 AMERICAN PSYCHOLOGIST

general paralleling of the two lists establishes sAme ality study is irrelevance, by which I mean that
degree of empirical check on both of them./We we frequently impose dimensions upon persons
can say that the direct common-sense perception when the dimensions fail to apply. (I am re-
of Jenny's nature is validated by quantification, minded of the student who was told to interview
coding, and factoring./ (Please note that in this women patients concerning their mothers. One
case factor analysis does not stand alone, but is patient said that her mother had no part in her
tied to a parallel rational analysis.) problem and no influence on her life; but that
While this meaningful validation is clearly pres- her aunt was very important. The student n-
ent, we gain (as almost always) additional insights swered, "I'm sorry, but our method requires that
from our attempts at empirical validation of the you tell about your mother." The method required
traits we initially hypothesize. I shall point to it, but the life did not.)
one instance of such serendipity. The tag words In ascribing a list of traits to Jenny we may
(i.e., the particular coding system employed) are seem to have used a dimensional method, but such
chiefly substantives. For this reason, I suspect, is not the case. Jenny's traits emerge from her
sexuality can be identified by coding as a minor own personal structure. They are not imposed by
factor; but it is not perceived as an independent predetermined but largely irrelevant schedules.
quality by the clinical judges. On the other hand,
the judges, it seems, gain much from the running CONCLUSION
style of the letters. Since the style is constant
What then have I learned about traits in the last
it would not appear Jn a factoriaLjjnalysis
4 decades? Well, I have learned that the problem
which deals__QnJy..Jffiith variance within_the_jyjiol€."
cannot be avoided—neither by escape through
Thus the common-sense traits cynical-morbid and
positivism or situationism, nor through statistical
dramatic-intense are judgments of a pervading
interaction effects. Tendencies, as McDougall
expressive style in Jenny's personality and seem
(1937) insisted, remain the "indispensable postu-
to be missed by factoring procedure.
lates of all psychology."
Here, however, the computer partially redeems
itself. Its program assigns the tag "overstate" to Further, I have learned that much of our re-
strong words such as always, never, impossible, search on traits is overweighted with methodo-
etc., while words tagged by "understate" indicate logical preoccupation; and that we have too few
reserve, caution, qualification. Jenny's letters score restraints holding us to the structure of a l i f e a s
exceedingly high on overstate and exceedingly low jt~is-"rtv^d. We find ourselves confused by our
r on understate, and so in a skeletonized way the intemperate empiricism which often yields un-
method does in part detect the trait of dramatic namable factors, arbitrary codes, umntdjrgible
intensity. -interaction effects, and sheer flatulence from our
One final observation concerning this essentially computors.
idiomorphic trait study. Elsewhere I have reported As a safeguard I propose the restraints of
a small investigation (G. W. Allport, 1958) show- "heuristic realism" which accepts the common-sense
ing that when asked to list the "essential character- assumption that persons are real beings, that each
istics" of some friend, 90% of the judges employ has a real neuropsychic organization, and that our
between 3 and 10 trait names, the average number job is to comprehend this organization as well as
being^ 7.2^ An "essential characteristic" is denned we can. At the same time our profession uniquely
as "any trait, quality, tendency, interest, that demands that we go beyond common-sense data
you regard as of major importance to a description and either establish their validity or else—more
of the person you select." There is, I submit, food frequently—correct their errors. To do so requires
for thought in the fact that in these two separate that we be guided by theory in selecting our trait
studies of Jenny, the common-sense and the fac- slices for study, that we employ rationally relevant
torial, only 8 or 9 central traits appear. May it methods, and be strictly bound by empirical verifi-
not be that the essential traits of a person are cation. In the end we return to fit our findings to
few in number if only we can identify them? an improved view of the person. Along the way
The case of Jenny has another important bearing we regard him as an objectively real being whose
on theory. In general our besetting sin in person- tendencies we can succeed in knowing—at least
 TRAITS REVISITED

in part—beyond the level of unaided common decades of labor on my part, and in spite of my
sense. In some respects this recommended pro- efforts in the present paper—I suppose I should in
cedure resembles what Cronbach and Meehl (19SS) all decency cry "uncle" and retire to my corner.
call "construct validation," with perhaps a dash
more stress on external validation. REFERENCES
I have also learned that while the major foci ADORNO, T. W., FRENKEL-BRUNSWIK, ELSE, LEVJNSON, D.
of organization in a life may be few in number, J., & SANFORD, R. N. The authoritarian personality.
the network of organization, which includes both New York: Harpers, 1950.
minor and contradictory tendencies, is still elusively ALLPORT, F. H. Theories of perception and the concept of
structure. New York: Wiley, 19SS.
complex. ' ' ""''
ALLPORT, G. W. What is a trait of personality ? Journal of
^t5ne~reason for the complexity, of course, is the Abnormal and Social Psychology, 1931, 25, 368-372.
need for the "inside" system to mesh with the ALLPORT, G. W. Review of S. A. Stouffer et al., The Ameri-
"outside" system—in other words, with the situa- can soldier. Journal of Abnormal and Social Psychology,
tion. While I do_not_beliey_e_ that traits__can__be_ 19SO, 45, 168-172.
ALLPORT, G. W. What units shall we employ? In G.
defined_in__terms of iotoaetion^effects (since all Lindzcy (Ed.), Assessment of human motives. New
tendencies draw their energy from within the per- York: Rinehart, 1958.
son), still the vast vanability ofjoehavior cannot ALLPORT, G. W. Religion and prejudice. In, Personality
be overlooked. In this respect I have learned that and social encounter. Boston: Beacon Press, 1960. Ch.
my earlier views seemed to neglect the variability 16.
ALLFORT, G. W. Pattern and growth in personality. New
induced by ecological, social, and situational fac- York: Holt, Rinehart & Winston, 1961.
tors. This oversight needs to be repaired through ALLPORT, G. W. The general and the unique in psycho-
an adequate theory that will relate the inside and logical science. Journal of Personality, 1962, 30, 405-422.
outside systems more accurately. ALLPORT, G. W. Behavioral science, religion and mental
health. Journal of Religion and Health, 1963, 2, 187-197.
The fact that my three illustrative studies are
ALLPORT, G. W. (Ed.) Letters from Jenny. New York:
so diverse in type leads me to a second concession: Harcourt, Brace & World, 1965.
that trait studies depend in part upon the investi- ALLPORT, G. W., & ODBERT, H. S. Trait-names: A, psycho-
gator's own purposes. He himself constitutes a lexical study. Psychological Monographs, 1936, 47(1,
situation for his respondents, and what he obtains Whole No. 211).
from them will be limited by his purpose and his ALLPORT, G. W., & VERNON, P. E. A study of values.
method. But this fact need not destroy our belief Boston: Houghton-Mifflin, 1931. (Reprinted: With
G. LINDZEY, 3rd ed., 1960.)
that, so far as our method and purpose allow, we
ARGYLE, M. Religious behaviour. Glencoe, 111.: Free
can elicit real tendencies. Press, 1959.
Finally, there are several problems connected BLOCK, J., & BENNETT, LILLIAN. The assessment of com-
with traits that I have not here attempted to re- munication. Human Relations, 1955, 8, 317-325.
visit. There are, for example, refinements of dif- CARR, H. A., & KINGSBURY, F. A. The concept of trait.
ference between trait, attitude, habit, sentiment, Psychological Review, 1938, 45, 497-524.
need, etc. Since these are all inside tendencies of CRONBACH, L. J., & MEEHL, P. E. Construct validity in
psychological tests. Psychological Bulletin, 1955, 52,
some sort, they are for the present occasion all 281-302.
"traits" to me. Nor am I here exploring the EDWARDS, A. L. The social desirability variable in person-
question to what extent traits are motivational, ality assessment and research. New York: Drydiui Press,
cognitive, affective, or expressive. Last of all, and 1957.
with special restraint, I avoid hammering on the GIBSON, J. J. A critical review of the concept of set in con-
temporary experimental psychology. Psychological Bul-
distinction between common (dimensional, nomo- letin, 1941, 38, 781-817.
thetic) traits such as we find in any standard GUTI-IRIE, G. M., & MCKENDRY, MARGARET S. Interest
profile, and individual traits (personal dispositions) patterns of Peace Corps volunteers in a teaching project.
such as we find in single lives, e.g., Jenny's. (Cf. Jo^^rnal of Educational Psychology, 1963, 54, 261-267.
G. W. Allport, 1961, Ch. IS, also 1962.) Nevitt KELSON, H. Adaptation-level theory. New York: Harper
& Row, 1964.
Sanford (1963) has written that by and large
HOLTZMAN, W. H. Attitudes of college men toward non-
psychologists are "unimpressed" by my insisting segregation in Texas schools. Public Opinion Quarterly,
on this distinction. Well, if this is so in spite of 4 1956, 20, 559-569.
10 AMERICAN PSYCHOLOGIST

HUNT, J. McV. Traditional personality theory in the light SKINNER, B. F. Science and human behavior. New York:
of recent evidence. American Scientist, 1965, 53, 80-96. Macmillan, 1953.
LA PIERE, R. Attitudes vs. actions. Social Forces, 1934, SPRANGER, E. Lebensformen. (3d ed.) Halle: Nieraeyer,
230-237. 1922. (Translated: P. Pigors. Types oj men. Halle:
LKNSKI, G. The religious factor. Garden City, N. Y.: Niemeyer, 1928.)
Doublcday, 1961. STONE, P. J., BALES, R. F., NAMENWIRTII, J. Z., & OGILVIE,
MCCLELLAND, D. C. Personality. New York: Dryden D. M. The general inquirer: A computer system for
Press, 1951. content analysis and retrieval based on the sentence as a
McDorGALL, W. Tendencies as indispensable postulates of unit of information. Behavioral Science, 1962, 7(4),
all psychology. In, Proceedings oj the XI International 484-498.
Congress on Psychology: 1037. Paris: Alcan, 1938. Pp. STOUITER, S. A., KT At. The American soldier. Princeton:
157-170. Princeton Univer. Press, 1949. 2 vols.
MILLER, D. R. The study of social relationships: Situa- TAGITJRI, R., & PETRULLO, L. Person perception and inter-
tion, identity, and social interaction. In S. Koch (Ed.), personal behavior. Stanford: Stanford Univer. Press,
Psychology: A study oj a science. Vol. 5. The process 19S8.
areas, the person, and some applied fields: Their place in VERNON, P. E. Personality assessment: A critical survey.
psychology and the social sciences. New York: Mc- London: Methuen, 1964.
Graw-Hill, 1963. Pp. 639-737. WALKER, E. L. Psychological complexity as a basis for a
NISWCOMB, T. M., TURNER, H. H., & CONVERSE, P. E. theory of motivation and choice. In D. Levine (Ed.),
Social psychology: The study of human interaction. Nebraska symposium on motivation: 1964. Lincoln:
New York: Holt, Rinehart & Winston, 1965. Univer. Nebraska Press, 1964.
SANVORD, N. Personality: Its place in psychology. In S. WILLIAMS, R. M., JR. Strangers next door. Englewood
Koch (Ed.), Psychology: A study of a science. Vol. 5. Cliffs, N. J.: Prentice-Hall, 1964.
The process areas, the person, and some applied fields: WILSON, W. C. Extrinsic religious values and prejudice.
Their place in psychology and in science. New York: Journal oj Abnormal and Social Psychology, 1960, 60,
McGraw-Hill, 1963. Pp. 488-592. 286-288.