Professional Documents
Culture Documents
Cattell 1961
Cattell 1961
Cattell 1961
A comprehensive, empirical system of per- than first-order factors, just as there are
sonality factor dimensions based on factor fewer first-order factors than there are tests
analyses of a battery of objective tests has (Cattell, 19S2). Second-order factors may
been described in previous publications (Cat- therefore be looked upon as relatively broad
tell, 19S5a; Cattell, 1957a). The initial em- descriptive categories, interpretable as repre-
phasis in the development of this system has senting general organizing influences in per-
been to establish these as first-order factors, sonality. Due to the limited number of ex-
that is factors which are based on correlations planatory concepts which the human mind
between tests, and which therefore adhere seems able to juggle at one time, as well as
closely to the detailed information available to the analytic defects of behavioral analysis
in tests. Recently, as first-order factor de- at the level of premetric general and clinical
scription has attained increasing replication, observation, the concepts of most psychol-
permitting more exact conceptualization and ogists fit factors at the second-order rather
measurement, it has become possible to shift than the first-order level. For example, the
attention to second-order factors, that is fac- personality concept measured by the ques-
tors based on the correlations between first- tionnaire second-order extraversion factor in
order factors (Cattell, 1956, 19S7a). These the 16 PF (Cattell, 1956; Cattell, Saunders,
correlations may be determined either (a) by & Stice, 1957) is probably referred to more
measuring the factors by factor batteries, in often than are concepts measured by the first-
which case they are attenuated by test un- order components: Cyclothymia, Surgency,
reliability and disturbed by invalidity, or (b) Parmia (Cattell, 1957a; Cattell, Saunders, &
from the plots of rotations to simple structure Stice, 1957). Both levels have their use and,
in the first-order domain. In the latter case, although secondaries (second-orders) lose
there is no attenuation effect and relatively some of the predictive power of primaries
slight disturbance by invalidity, if and as the (first-orders), knowledge of the second-order
rotation is good and enough variables are em- structure is extremely important for under-
ployed to define the hyperplanes. Research standing personality, developmentally and in
needs to take both approaches, since the action.
degree of consensus throws light on the mag- The position achieved by research to date,
nitude of error. Both approaches are con- in respect to orders and media of observation,
sidered in this article. is as follows:
Since factor analysis is a parsimonious
operation, there will be fewer second-order 1. The establishment of first-order person-
1
The opinions expressed are those of the writers
ality factors in the questionnaire (verbal,
and are not necessarily shared by the Department of self-evaluative) and life record (behavior
the Navy. rated in situ) media of measurement (e.g.
345
346 R. B. Cattell, R. R. Knapp, and I. H. Scheier
oblique rotations, as was the case in the other four one study and others are indicated by enclosing the
studies. atypical value in brackets. A positive sign on a load-
ing means that the positive pole of that factor, as it
Second-order factorization in all of the five studies is standardly written in various textbooks (Cattell,
used tests of completeness of factor extraction de- 1957a; Cattell & Scheier, 1961) goes positively with
scribed elsewhere (Cattell, 1952) and pursued simple the second-order factor pole as written, and oppo-
structure blindly to a plateau at which hyperplane sitely for a minus loading. Since, in reading titles
count proved unimprovable. At that point the factor on the left, readers like to see at once which "go
patterns were inspected and a process of matching together," we have followed the practice of writing
(correlation of loading patterns) was begun, factor in the pole which is consistent with the positive direc-
by factor, with each of the other four studies. This tion of the second-order factor. That is to say, when
matching revealed tolerably good convergence of re- a loading in the numerical columns is negative (for
sults in the five studies and the matched factors and the positive pole of the first-order factor), we have
their loadings are therefore set out in Tables 1 already reversed the title to put the negative pole of
through 7 below, each table representing a second- the factor in this title column. Therefore, the verbal
order objective test factor which we now believe to title of the first-order already gives the direction in
be reasonably well confirmed. All five studies replicate which (pole at which) it is related to the second-
the same seven second-order factors, and although a order factor. The signs on the loadings merely tell
given factor is often less clear in one series than in whether this is the positive (high score) or negative
another, no alternative matching would be nearly as (low score) pole of the first-order, as it is usualh
satisfactory as that presented. scored and thought of.
The layout of these tables follows the usual con- 4. The second column gives the average loading,
ventions, as follows: which, by reason of the attenuation in Ni, would be
1. Each table is headed by a contingent verbal title expected to err systematically, slightly below the true
for the second-order factor and a roman numeral. value. The given rank order in listing the primaries
The verbal title may be modified as more information is not literally the declining order of their mean load-
becomes available on the factor, but its roman ing on the second-order, but an estimate of the order
numeral remains constant for easy identification in of importance of the primaries, made by taking into
present and future studies. Each verbal title is bi- account also the consistency and frequency of replica-
polar, the positive (high score) pole being on top and tion of the result.
and the negative in parentheses. 5. The tables report data only on the first-orders
2. First-order factors—the "variables" at this order which are most highly and consistently associated
of analysis—are identified in the left-hand column of with second-orders. The full tables, showing the load-
each table, (a) by their Universal Index or UI num- ing of all first-orders present in each study, including
bers (Cattell, 1957b), which are constant across all those with lesser to essentially zero relationship, are
published researches, and (6) by their verbal titles. preserved in tables available from the American
Documentation Institute.*
3. The figures in Columns 3 through 7 are the
4
loadings of first-orders on the second-order, each Complete loading tables for all first-order factors
column being designated by the symbol for the re- have been deposited with the American Documenta-
search in which the results were found. These values tion Institute. Order Document No. 6758 from the
diverge from the true values, in the case of C5, Co, ADI Auxiliary Publications Project, Photoduplication
Rj, and Ra, by reason of imperfect simple structure Service, Library of Congress; Washington 25, D. C.,
positions in first- and second-order relations, and in remitting in advance $1.25 for microfilm or $1.25
the case of Ni by attenuation due to unreliability of for photocopies. Make checks payable to: Chief,
the factor batteries. Inconsistencies of sign between Photoduplication Service, Library of Congress.
TABLE 1
F(T)I: TIED SOCIALIZATION OR SUPEREGO (vs. ABSENCE OF CULTURAL INTROJECTION)
Loadings in Studies
TABLE 2
F(T)II: EXPANSIVE EGO (vs. HISTORY OF DIFFICULTY IN PROBLEM SOLVING)
Loadings in Studies
DATA AND INTERPRETATION FOR critical?), but the central factors are UI
SECOND-ORDER FACTORS 20, Comention (conventional acceptance of
Factor F(T)I, Tied Socialization, has values) and UI 28, Rigidity of Superego,
claims to being the largest second-order fac- which we believe the contingent title aptly
tor. There is high internal psychological con- designates.
sistency (with the possible exception of low Except for UI 23 ( — ) , there is again consid-
intelligence) in terms of the characteristics of erable psychological consistency in F(T)II,
the first-order factors involved. A clinician here in the sense of willfulness ("will power")
might well claim it to be proof of a broad and a high development of the self-sentiment.
superego organization, and some solid support Once again, too, the clinician might claim
for this lies in the finding (Cattell & Scheier, that this particular clustering of first-order
1961) that neurotics are significantly higher factors gives support to an observational con-
than normals on this second-order factor. cept at the second-order level, in this case,
However, in the label "Tied Socialization" the concept of a broad dynamic organization
we have favored an hypothesis which may that is called "ego development" or "ego
specialize to a superego definition, but which strength." The essential unifying concept,
is at present broader. It implies that this covering higher intelligence and the assertive
pattern represents the extent to which the characteristics, appears to be, if an environ-
individual has accepted the culture patterns mental explanation is adopted, a history of
and standards of the group. It has overtones success in managing ergic (drive) satisfac-
of extraversion and realistic contact, subdued- tions. This interpretation would also fit psy-
ness, receptivity, and lower intelligence (un- choanalytic concepts of ego strength. The
TABLE 3
F(T)III: TEMPERAMENTAL ARDOR (vs. TEMPERAMENTAL APATHY)
Loadings in Studies
UI 21 Exuberance, Energetic
Spontaneity +31 + 70 +40 +23 + 11 + 13
UI 20 Comention, Conformity to
Cultural Standards +27 +48 + 12 +08 +38
UI 1 Low General Intelligence -28 -15 -41
UI 19 Promethean Will +21 +21 + 12 + 18 +54 +01
UI 27 Alert Control -19 -29 -09
UI 32 Extraversion -16 (10) -34 -55 (03) -06
Second-Order Personality Factor Structure 349
TABLE 4
F(T)IV: HIGH EDUCATED SELF-CONSCIOUSNESS (vs. Low EDUCATED SELF-CONSCIOUSNESS)
Loadings in Studies
UI 22 Corticalertia, Cortical
Alertness +64 +40 +08 (-02) +47
UI 18 Shrewdness +28^ +66 +24 +40 +07 +01
UI 25 Imaginative Tension +17-, +45 + 12 +02 +08
UI 36 Self-Sentiment Development /+51.) +51
UI 30 High General Reactivity ^TO" -03 -08 (+05) -57
UI 29 Low Adaptation Energy -15 -01 -08 -53 (+04)
UI 33 Dourness +20 +20
chief criterion evidence on this factor shows genetic influence covering all primaries con-
it distinguishing even more powerfully than tributing to this second-order factor. How-
F(T)I between neurotics and normals, the ever, it is difficult to imagine in polygenic
normals being significantly higher than neu- determination how a series of genes would
rotics in Expansive Ego (Cattell & Scheier, happen to coincide in repeatedly affecting
1961). several things in the same way. This is a pro-
A noteworthy feature of F(T)III is that it vocative finding for psychological genetics.
tends to pick out those factors which individ- Meanwhile, we shall label the second-order
ually appear to have relatively high genetic "Temperamental Ardor," since, except for
determination (Cattell, 1957a). Previous UI 20, the central character is a willful exu-
multivariate experiments (Cattell, Stice, & berance and ardor of temper. Consistent with
Kristy, 1957) using groups of identical and this interpretation, F(T)III has been found
fraternal twins and siblings, some of which to be significantly higher for hospitalized
were reared together and some apart, have neurotics of sociopathic type than it is for
attempted to assess the contribution which normals (Cattell & Scheier, 1961).
heredity and environment may make to cer- F(T)IV tends to be loaded by what appear
tain of the objective test factors. In particular to be largely environmentally determined fac-
it appeared that heredity was the main deter- tors. In previous investigations (Cattell, Stice,
miner in F(T)III Factors UI 1, Intelligence, & Kristy, 1957) environment appeared to be
and UI 20, Comention. Hereditary and envi- the main determiner in F(T)IV Factors UI
ronmental influences appeared to be about 22, Corticalertia, and UI 29, Low Adaptation
equal in determining Factors UI 19, Prome- Energy-vs.-Overresponsiveness. It probably
thean Will, and UI 27, Alert Control, which has something to do with education in the
are found on F(T)III. Conceivably, these sense of developing alert, shrewd, and imag-
genetic determinations all arise from a single inative qualities. It also involves much ex-
TABLE 5
F(T)V: HISTORY OF INHIBITING, RESTRAINING ENVIRONMENT (vs. LAXNESS)
Loadings in Studies
TABLE 6
F(T)VI: NARCISTIC DEVELOPMENT (vs. ENVIRONMENTAL CONTACT AND INVESTMENT)
Loadings in Studies
plicit self-awareness and "canniness." Possi- In F(T)VII, we may have a form of ten-
bly, an upbringing in a critical, competitive, sion wider than Free Anxiety, UI 24, and
high-standard home atmosphere could gen- therefore perhaps representing total drive ten-
erate such a pattern. The above interpretation sion as it is oriented and controlled toward
is, at least, not inconsistent with the fact that achievement. On the other hand, an argument
neurotics with marked sociopathic trends could be made from present inclusions and
have significantly lower scores than normals exclusions that the central influence in the
on this factor (Cattell & Scheier, 1961). factor is insecurities connected with the self-
Our hypothesis is that F(T)V represents concept. We shall keep interpretive eventual-
a dimension of personality resulting from an ities open by a descriptive title indicating
environment in which considerable inhibition anxiety and related drive tensions centered on
prevails. The somewhat unexpected role of insecurity, and under control of an achieve-
UI 23, here as in F(T)II, suggests that there ment goal.
may be some need to modify our hypotheses Now that the patterns of second-order ob-
about this primary. However, it is reasonable jective test factors have been presented and
to interpret UI 23 as unused reserves of en- tentatively interpreted, it is logical to ask if
ergy and it might then fit here as the cum- very broad third-order factors can be found.
ulative result of inhibited, undischarged re- Accordingly, we computed the correlations
activity. among second-order T factors, in the three
F(T)VI is a rather narrow factor which most recent studies (Rj, R2, NJ and aver-
nevertheless has a consistent character of aged these values in Table 8. These correla-
narcistic and autistic development. It is in tions among second-orders are generally very
some sense a false ego development, a moving low. Over half of the r's are .10 or less, and
out of contact with reality. This hypothesis only one, between F(T)III and F(T)VI,
is supported by its being significantly high could confidently be called statistically signif-
in neurotics with sociopathic trends, but not icant. It is possible that a third-order might
in typical neurotics (Cattell & Scheier, 1961). be found, including the second-orders of Tem-
TABLE 7
F(T)VII: HIGH TENSION TO ACHIEVE, CONTROLLED DRIVE TENSION LEVEL
(vs. Low TENSION TO ACHIEVE)
Loadings in Studies
CATTELL, R. B., & SCHEIER, I. H. Extension of mean- SCHEIER, I. H. What is an "objective" test? Psychol.
ing of objective test personality factors: Especially Rep., 1958, 4, 147-157.
into anxiety, neuroticism, questionnaire, and phys- SCHEIER, I. H., & CATTELL, R. B. Confirmation of
ical factors. /. gen. Psychol., 1959, 61, 287-315. objective test factors and assessment of their rela-
CATTEIX, R. B., & SCHEIER, I. H. The meaning and
tion to questionnaire factors: A factor analysis of
measurement oj neuroticism and anxiety. New
York: Ronald Press, 1961. 113 rating, questionnaire and objective test meas-
CATTELL, R. B., STICE, G. F., & KRISTY, N. F. A first urements of personality. /. ment. Set., 1958, 104,
approximation to nature-nurture ratios for eleven 608-624.
primary personality factors in objective tests.
J. abnorm. soc. Psychol., 1957, 54, 143-159. (Received July 13, 1960)