You are on page 1of 24

Journal ofhttp://jca.sagepub.

com/
Career Assessment

The Integration of Interests, Aptitudes, and Personality Traits: A Test of Lowman's Matrix
Andrew D. Carson
Journal of Career Assessment 1998 6: 83
DOI: 10.1177/106907279800600106

The online version of this article can be found at:


http://jca.sagepub.com/content/6/1/83

Published by:

http://www.sagepublications.com

Additional services and information for Journal of Career Assessment can be found at:

Email Alerts: http://jca.sagepub.com/cgi/alerts

Subscriptions: http://jca.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://jca.sagepub.com/content/6/1/83.refs.html

>> Version of Record - Jan 1, 1998

What is This?

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


83

The Integration of Interests, Aptitudes, and


Personality Traits: A Test of Lowman’s Matrix
Andrew D. Carson
The Ball Foundation, Glen Ellyn, IL
Characteristics of Holland’s (1959, 1985/1992) six personal
orientations were determined through construction of a table

reporting multiple-domain data (interests, aptitudes, personality


traits) collected from a sample of 547 career clients. The resulting
table was used to test Lowman’s (1991) Matrix of expected
characteristics across Holland’s types.
In his theory of vocational personalities and work environments, Holland
(1985/1992, 1996) described basic types of people (and environments) in a
way that could be applied to many problems, both within and outside the
vocational area. He identified six types-Realistic (R), Investigative (I),
Artistic (A), Social (S), Enterprising (E), and Conventional (C)-as having
distinct interests, abilities, personality traits, and other characteristics. A
number of studies have suggested that the types fit a two-dimensional
hexagonal structure reasonably well, following the familiar RIASEC order
(see Spokane, 1996).
The primary concern motivating this investigation lay in how these types
related to abilities and personality traits, following Lowman’s (1991, pp. 186-
187 ; see Table 1) proposal of a table for integrating the interest based
personal orientations (one for each of Holland’s six types), on the one hand,
with abilities and personality traits, on the other. That table is hereafter
referred to as Lowman’s Matrix. Also adopted in the following discussion is
Holland’s term personal orientations from his original 1959 theoretical
statement, used to describe types identified solely on the basis of interests,
with his later (1985/1992) term vocational personality type reserved to
refer to the broader construct that incorporates interests, abilities,
personality traits, and other characteristics (also see Spokane, 1996, p. 40,
on this point). In all, Lowman described 14 different abilities and nine

This article was based on a paper presented as part of the symposium Contemporary
Issues in Scientific Psychology: Inter-Domain Relationships and Applications, R. L.
Lowman (Chair), at the 1996 meeting of the American Psychological Association,
Toronto, Ontario. The author gratefully acknowledges the suggestions of E. B. Bizot to
that paper.
The dataset (ann2l.sav) on which this study is based is available from the author.
Correspondence concerning this article and requests for offprints should be addressed
to Andrew D. Carson, The Ball Foundation, 800 Roosevelt Rd., C-120, Glen Ellyn, IL
60137-5850.

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


84

Table 1
Hypothesized Characteristics of Personal Orientations
Described Through Lowman’s Matrix

Note. + high scores, - low scores, and o medium scores. R Realistic, I


= = = = =

Investigative, A Artistic, S Social, E Enterprising, and C Conventional.


= = = =

aLowman’s descriptions of the features of individuals with Artistic orientation include


a variety of options depending on the type of artistic endeavor (e.g., writers high, actors

average or below; see Lowman, 1991, p. 187). Parentheses denote that the trait is one
of a set of traits required by at least one of the subtypes of Artistic orientation.
Individuals with Artistic orientation would have high need for self-control in work and
low need for self-control in personal life. Lowman describes individuals with Realistic
orientation as being either low or average in adjustment and self-control. bIndicates that
this trait was not investigated as part of either Carson (1996) or the present study.
Adapted from The Clinical Practice of Career Assessment: Interests, Abilities, and
Personality (pp. 186-187), by R. L. Lowman, 1991, Washington, DC: American
Psychological Association. Copyright @ 1991 by the American Psychological Association.
Adapted with permission.

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


85

different personality traits (assuming agreeableness and likeability to be one


and the same), yielding a list of 23 traits across six types, producing a total
of 138 cells. Each cell represents the intersection of a type and a trait and,
in the summary presented in Table 1, the contents of each cell indicate
whether people of that type are generally high, medium, or low on the trait.
Lowman (1991) filled in the cells of his matrix primarily (but not
exclusively) on the basis of published empirical research. Although his
literature review was extensive, he found it necessary to supplement his
research based review with clinical experience when constructing the
matrix. He reasonably suggested that the elements of the matrix should,
therefore, be empirically validated before it was used as a routine guide to
career assessment practice. Thus, the goal in this investigation was to test
as many of his 138 matrix based predictions as possible, using a rich (i.e.,

many variables) data set.


In the following we describe the results of research that has tested
specific elements of Lowman’s (1991) Matrix, and it is important to note an
important feature of Holland’s (1985/1992), Lowman’s (1991), and others’
(e.g., Randahl, 1991; Spokane, 1996) discussions of the characteristics of the
six types. All included general cognitive ability (g) or intelligence as a
variable, and they all discussed specific abilities (such as spatial ability),
but they did not discuss specific abilities with the effects of g partialed out
(i.e., g-free abilities). An exception to this observation may be a recent
review of the literature by Ackerman and Heggestad (1997), who do discuss
general intelligence, other aptitudes, and multiple other traits as they
relate to personal orientations. This may sound esoteric, but it has practical
implications for the counselor, because it bears directly on the question of
whether or not it is worth administering objective tests of specific abilities
in addition to a measure of g. It is generally established that g has
ubiquitous predictiveness of success in training and to a somewhat lesser
degree on the job across all personal orientations (Ree & Earles, 1994,
1996; see Schmidt, 1988).
However, it is conceivable that leaving g admixed with measures of
specific abilities might obscure the presence of individual differences across

personal orientations in such abilities. This problem seems never to have


been directly addressed in the context of Holland’s theory, although Randahl
(1991) came close to doing so through her study of aptitude profiles across
Holland’s types (by considering the profiles of aptitudes within types, and
not aptitude scores per se, one effectively partials out the effects of g on
aptitudes). However, this general issue has been a long-standing one in
psychology. Early in this century Spearman (1927; Spearman & Jones,
1950) proposed the usefulness of considering specific abilities (e.g., spatial
ability) with the effects of g removed (i.e., g-free). More recently, Desmarais
and Sackett (1993) and others (Rischall, Bizot, & Lobsenz, 1995)
demonstrated that statistically removing the effects of g from specific
abilities resulted in a more fine-grained differentiation of jobs within the
Occupational Aptitude Patterns (OAP) Map created by L. Gottfredson
(1986). Desmarais and Sackett’s findings, coupled with Carroll’s (1993,
p. 637, 1996) recent discussion of the theoretical justification of g-free

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


86

specific abilities, led me to examine the relative magnitude of g-free


specific abilities across personal orientations as part of the evaluation of
Lowman’s Matrix.
The present study was an extension of an earlier one completed in
Montreal (Carson, 1996), which examined g and g-free aptitudes across
Holland’s personal orientations. A sample of career clients (N 117, average =

age = 26 years) completed computerized versions of the Strong Interest


Inventory (Strong; Hansen & Campbell, 1985) and the Differential Aptitude
Tests-Adaptive (DAT-A; McBride, 1986; Moe, 1988). Carson assigned clients
to personal orientations on the basis of high-point scores on the General
Occupational Themes (GOTs) of the Strong, which resulted in quite small
samples for some types, especially Realistic and Investigative. With a
measure of g derived from the first unrotated factor in a principal

components analysis of the DAT-A (and therefore sample specific), Carson


used regression analysis to compute g-residual (g-free) scores for each of the
specific abilities (as recommended by Ree & Earles, 1994). Correlations
between g and interest scales conformed to the simplex pattern reflected in
Holland’s hexagonal organization of types, with a moderately high positive
correlation found between g and Investigative (and to some degree Realistic
and Artistic) interests and a moderately negative one observed between g
and Enterprising (and to some degree Conventional) interests (rs of .26, .40,
.18, .03, -.23, -.14, for R, I, A, S, E, and C, respectively).
Investigative type clients had the highest average scores on most of the
aptitude tests, but then they also had the highest levels of g. Statistically
removing the effects of g affected some things and not others. In particular,
Investigative types no longer dominated the entire spectrum of g-free
aptitudes. Realistic types had by far the highest g-free Mechanical Reasoning
and Space Relations scores, and Conventional types had by far the highest
g-free Numerical Ability score as well as the highest Clerical Speed score.
In general, these results were consistent with Holland’s descriptions of the
personal orientations. However, with g removed, Investigative types obtained
the highest average ability scores only in the areas of language usage and
spelling, suggesting that beyond &dquo;raw&dquo; g, the Investigative type was
relatively the strongest type only in some specific linguistic areas.
Method
Sample and Procedure
The present archival study drew from a sample of 270 men (mean age =

30.24 years, S’D 10.08 years) and 277 women (mean age
= 31.27 years, =

SD 9.98 years), all career clients who had completed the Ball Aptitude
=

Battery (BAB; Ball Foundation, 1995), the Strong, and the Sixteen
Personality Factor Questionnaire (16PF; Cattell, Eber, & Tatsuoka, 1970;
Russell & Karol, 1994), all as part of career assessment services offered
through The Ball Foundation/CAREER VISION at its offices in a suburb of
Chicago. The sample represented all records from the Ball Foundation
archives of clients with the following characteristics: (a) completed each of
these tests during the period 1985-1994, (b) had a single high (scaled) score
on the GOTs of the Strong (i.e., no ties for highest scores; that required

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


87

dropping 29 cases), and (c) had counselor recommendations available (such


recommendations were not, however, investigated as part of this study). The

sample was primarily White (96%) and drawn mostly from suburban
Chicago. Most, but not all, of the clients had paid a fee for career assessment
services (ranging from $200-$400 over the period covered). Referral sources
included self-referral (e.g., from advertisements), word-of-mouth from
former clients, and counselors or other professionals in the community.
Client activities included approximately 7 hours of testing (which included
a brief intake interview) followed at least 1 day later by 2 to 3 hours of
individual counseling, during which results of testing were discussed with
the client.

Measures

Aptitudes
The BAB is a multiple aptitude battery that has been used in both career
assessment and industrial selection applications. The BAB tests used in this
study were Vocabulary (VO; 80 items; task is to choose the word option
that best matches the prompt word), Writing Speed (WS; involves repeatedly
writing a sentence for a minute), Paper Folding (PF; 24 items; task is to
identify a drawing option that shows how a hole-punched paper would
appear if unfolded), Shape Assembly (SA; 6 items; involves manual
reassembly of complex shapes that have been separated into parts), Inductive
Reasoning (IR; 30 items; task is to identify which three of a set of six
pictures have something in common), Analytical Reasoning (AR; 18 items;
task is to identify the most logical hierarchical semantic network for
organizing sets of five, six, or seven words), Numerical Reasoning (NR; 20
items; task involves predicting the next member of a sequence of numbers),
Numerical Computation (NC; 20 items; involves solving arithmetic
problems), Idea Generation (IG; involves writing as much as possible on a
subject for 10 minutes), Idea Fluency (IF; 4 items; involves generating
alternative uses for an object), Clerical (CL; 240 items; involves determining
whether series of numbers are the same or different), Finger Dexterity
(FD; involves rapidly inserting sets of pins in holes), and Grip Strength (GS;
repeated trials with a dynamometer to assess hand squeezing strength).
The BAB manual (Ball Foundation, 1995; see also Carson, Stalikas, &
Bizot, 1997) summarizes evidence for the reliability and validity of the
battery’s tests. Reliability ranged from moderately high to high across the
tests: 6-week test-retest reliability scores are available for all tests except
NR and NC (N 261), and they ranged from .70 to .98; alpha reliability was
=

.91 for NR and .87 for NC. The tests correlate with scales of other aptitude
measures, such as the General Aptitude Test Battery (GATB; U.S.
Department of Labor, 1970) and the Differential Aptitude Tests (DAT;
Bennett, Seashore, & Wesman, 1982) in predictable ways. Finally, BAB
scores correlate positively with academic achievement and job performance
(with patterns that support the criterion-related validity of the specific
tests), and a variety of occupational groups may be distinguished on the basis
of BAB scores. Scores for FD and GS were based on summations of raw scores
from lateralized indicators of the same traits found in the BAB (i.e., Finger
Dexterity Left, Finger Dexterity Right).

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


88

An index for g (hereafter, BAB-g) was derived from the first unrotated
factor from a principal components analysis of the BAB scales that had
first been transformed into Z-score form. The first factor had an eigenvalue
of 3.77 and accounted for 29% of the total variance, and the factor loadings
(correlations between observed test score and factor score) were consistent
with previous observations of BAB tests (Ball Foundation, 1995): NR, .72;
AR, .71; PF, .66; NC, .64; SA, .58; IF, .54; IR, .54; IG, .48; FD, .48; CL, .44;
VO, .42; WS, .37; GS, .06. BAB-g correlated at .86 (N 547, p < .01) with =

the BAB’s index for estimating the G scale of the GATB (see Bizot, 1993;
Goldman, 1987), lending support that BAB-g provides a good index of the
construct. Because GS scores proved almost entirely distinct from g, and
because GS has shown such strong sex-related differences (Ball Foundation),
GS was dropped from further analyses. As described in relation to Carson’s
(1996) Montreal study, g-free ability scores were also obtained.
Interests and Personal Orientations
Client data were sorted into Holland’s (1985/1992) six personal
orientations using high-point codes on the GOTs of the Strong, as derived
from standard scores (within-gender norms were not used). The psychometric
characteristics of the GOTs have been so thoroughly enumerated elsewhere
that they are not reviewed here. The sample included a respectable
proportion of clients from each of the six types 14% Realistic (67 men, 7
women), 9% Investigative (32 men, 16 women), 24% Artistic (43 men, 90
women), 22% Social (36 men, 82 women), 18% Enterprising (54 men, 46
women), and 14% Conventional (38 men, 36 women). Only data from those
clients with a high score on a single GOT were retained for further analyses.

Personality Traits
Weighted composites of the 16 primary (first-stratum) scales of the fourth
edition of the 16PF (as indicated on the combined sex norms table on p. 129
of Cattell et al., 1970) provided indicators for each of the Big Five personality
traits, namely, extraversion, anxiety, tough-mindedness, independence,
and self-control. Russell and Karol (1994, p. 76) reported that these scales
correlate .81, .79, .38, .70, and .65, respectively, with their counterparts in
the fifth edition of the 16PF. Lowman (1991, pp. 186-187) tends to use the
16PF-style terminology for the Big Five traits, so use of such trait terms
provides some advantage in testing his model. It seems reasonable to
suggest-based on material in the 16PF manual-that these 5 scales might
serve as proxies for the more standard Big Five traits of extraversion,
neuroticism (anxiety), openness (tough-mindedness; reverse scored),
agreeableness (independence; reverse scored), and conscientiousness (self-
control). Two primary traits, E and F, were used in addition to the global
composites, because they might serve as proxies for two traits Lowman
included in his matrix (E for ascendance, F for reserved).
Results
large number of variables included in the set of data, it was
Because of the
necessary to multivariate method to investigate group differences. This
use a
method ensured that experiment-wise error would not be inflated and that

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


89

all relationships between variables could be considered simultaneously in


the analysis (Fish, 1988; see Vacha-Haase et al., 1994). Because the primary
research question related to differences in g, g-free aptitudes, and personality
traits across personal orientation groups, only those scales were included
in a descriptive discriminant analysis (described by Vacha-Haase et al.). The
discriminant analysis of the standard scores produced a canonical correlation
of .57, Wilks’s lambda =
.42, x2(100) 459.18, approximate F 4.93
= =

(hypothesized df = 100), p .01. Thus, about 32% of the variability in the


<
discriminant scores was attributable to between-group differences (.572 =

.32). Wilks’s lambda describes the proportion of total variability not


explained by group differences; a moderately large 16% of group differences
(effect size) may be explained by the variance in the aptitude and personality
scores (using the formula l-W1/S, where W Wilks’s lambda, and S L-1,
= =

where L is the number of levels in the independent variable; 1-[.42]1~5 =

1-[.42]~2 1-.84 .16). Overall, 47% of cases were correctly classified into
= =

the GOT-based personal orientation groups. A separate discriminant analysis


(results not shown) using only g and g-free aptitudes as predictors resulted
in a 38% correct classification rate, suggesting that addition of the
personality variables added at least moderately (a 9% improvement) to the
ability to correctly classify cases into personal orientations.
Discriminant analysis provides the opportunity to examine both
(standardized) function and structure coefficients to evaluate variable
importance. The function coefficients are the multipliers of dependent
variables that yield the group differences; however, the function coefficients
must be interpreted with care, because highly correlated variables share
their discriminant weights. Structure coefficients are ordinary Pearson
product-moment correlations between the various dependent variables and
scores on the discriminant function, and they are therefore not influenced
by correlations between the various dependent variables. Three canonical
discriminant functions (with eigenvalues of .47, .27, .15) resulted in
significant F values (at p < .01); therefore, only they were retained for
analysis (see Table 2).
Table 3 provides descriptive statistics across personal orientations and
shows the results of both univariate F tests and individual multivariate tests
for those variables included in the discriminant analysis. Among the
univariate tests, all variables except analytical reasoning, finger dexterity,
and vocabulary (on the aptitude side) and anxiety (on the personality side)
showed significant differences across personal orientations. However,
because many of these variables are highly correlated (Bartlett’s test of
sphericity 6022.67, df = 190, p < .01), the results of the separate univariate
=

tests are not necessarily informative regarding the importance of particular


variables in describing group differences. (Individual multivariate tests
were also conducted using planned comparisons in which the predicted
and average score for the variable were contrasted. Because not all of the
personal orientation groups could thus be tested via planned comparisons,
the Enterprising group was dropped, because it has traditionally been
described as being the most difficult to predict from aptitudes. Due to space
considerations, the complete results are not shown in Table 3, but are

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


90

V
0 ~
flg
O U
#

~, gP V £CCS
d m, ~ <
~ _~
S
(3) -41

Cd Cd
cd
2im
cdV
f
< M

m~ cS
<2~ ~U
2
~~~
V
CO
-i-a
~:s
a
4) qj ~£ < rU-i
’S t6 £d
&t:
dquo;14 ,<j
U O U
a)
o U41s
CdU m
U
4)
h
§Q~J#§
< h u
:3
-~
CQ £ fl
# X£6m
:3

c~cc
~
-<j « £j#% p>~g£ m >140.
0) &dquo;0 cd§Q t6 <
4 0
~s
0
O+~
1% ;z cd
E-4 0 41 ~ s
0 ~~ ~
-W
u y s
a -41
Cd
:3
r-14
m-s~
C)
S M
-
0 0~O
Cd
ce

j
0
s s-s
.p4 CJ CO OJ
$4Q
rA cd cdm
-4--~4
5

Cd0M fj f~, j
~~.s-~
Cd n
CIS &dquo;0 .~
C>Z~~~Can)~c~3
41 cC .~

~s >M
Sc/S
Al

LObJ)
ll n g
g ’°g~’’
ll oU
/ o’
~ a~ ~+, ~&dquo;
~ h ~ o
CO

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


91

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


92

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


93

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


94

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


95

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


96

0 o
’40o -
Cd
cd
~!F bio

Cd
~=!o
~£3
gh m
6C
C U
~0
*~ N
ca ~ +~
a
.r., g<fl
O
~3
,£Cd
, flg
a
0
tW CJ ~ ,

6C
;4 ~A~~
.Alq~
0)
(~ II ll £
6C £d,°
0
h #£§k
~<0) ? _«m
t6g£Q
4-i
o
~ 5~m
.5S ~~~
Cd -0
o cd
§~
O
v
CCS
££>
Cd

M Cd
V£Ui
~~ 4.
0
J3 60
~
H O 4&dquo;i ~
~
~~.s
C
ca
O
0 cd
a-) cn
ce
.# o
.r.,
£m<M
,bDTJ~
N ~ ~fJ
.r., ~ !-< ~ 0
66 0 Cd - _,<-)-.
cc ~£~fl<5 flj ,
a)
.« ~Q£ m g _ £d
.r., <D -~ 0 ~
.&’s ~~~5
0o
> u
u ~0~~
0)
0
A ~ +, . ~
Cd
ai ~ ~ a~ ~’
#t6£~
~ ~ T.~-m,
.r
c~3

II
11%~ ’8
g,f1°£
C1 ~ U
Q§Q.2§f
ou

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


97

available from the author; however, the table does indicate which individual
multivariate tests were significant.)
Taken together, the results shown in Tables 2 and 3 suggest that some
variables are more important for describing group differences than are
others. The g-free aptitudes all had large standardized function coefficients,
but, consistent with the univariate F tests, neither finger dexterity, analytical
reasoning, nor vocabulary had structure coefficients above .30 for the three
functions. Consistent with the univariate F tests, anxiety likewise failed to
show a structure coefficient above .30. Thus, it appears reasonable to
conclude that these four variables contributed little to the discrimination
between groups. However, three variables that showed significant univariate
F tests failed to generate structure coefficients above .30: writing speed,
tough-mindedness, and ascendance. It would appear that, given the rest of
the variables, these three variables also failed to contribute much to
discrimination across groups when the effects of other variables were taken
into consideration.
Some comments on the interpretation of the functions appear warranted,
although the sample size and the eigenvalues suggest that a more reliable
interpretation of coefficients is more likely for the first two functions than
for the third. The first discriminant function (Function I) describes a
dimension in which all of the g-free aptitudes have their largest (and all
positive) function coefficients (absolute value); the variables for which
structure coefficients are highest (absolute terms) for this function suggest
that an underlying orientation to people and things characterizes this
function. The pattern of structure coefficients suggests that the dimension
is more Social at its positive pole (due to high extraversion, low reserved [i.e.,
high impulsivity], and high verbal productivity [idea generation, writing
speed]) and more Realistic at its negative one (spatial abilities). For the
second discriminant function (Function II), the set of highest structure
coefficients suggest an underlying orientation to both the Conventional
and Investigative types (with emphases on numerical abilities) and Artistic
type (with emphases on independence, analytical reasoning, and vocabulary).
Lastly, the third discriminant function (Function III) suggests an underlying,
but loose, orientation to the Enterprising type (e.g., for high idea fluency and
relatively lower g) and Investigative type (as associated with high g and low
inductive reasoning). Correlations between the discriminant functions and
the GOTs were as follows (all in RIASEC order, all with
p < .05 unless otherwise noted; highest and lowest correlations for each set
italicized): Function I, -.56, -.24, .20, .44, .19, and -.07 (ns); Function II,
-.06 (ns), .12, .42, -.13, -.17, -.22; Function III, -.12, .07 (ns), -.11, -.15,
-.23, .20. Thus, the simplex model reflected in Holland’s hexagon held for
the first two sets of correlations (Functions I and II) but not for the third
(Function III).
Table 4 reorganizes the data from Table 3 to show the rank order of
personal orientation groups by raw score on each of the variables. Removing
g tended to lower the rankings in aptitude scores of the Investigative, and
to a lesser degree the Artistic, personal orientations, and in some cases it
raised those of the Enterprising and Social orientations. Contrary to

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


98

expectations, Investigative type emerged as having the highest (g-


clients
retained) aptitudes only Paper Folding, Idea Fluency, and
on three tests
Numerical Computation. Once g was removed, Investigative type individuals
were dominant in none of the aptitudes. Particularly on the Inductive
Reasoning test, Investigative types also performed poorly relative to the
other types (except Enterprising) even in the g-retained condition. Realistic
types dominated the spatial aptitude tests, especially in the g-free condition.
Even in the g-retained condition, Realistic types dominated the Shape
Assembly test, a hands-on measure of spatial aptitude, and they placed
second to (the higher g) Investigative types in the more mental Paper
Folding test. The comparative strengths of the Artistic person would appear
to lie in verbal areas, where the Artistic group performed well in both the
g-retained and g-free conditions. However, although in the g-retained
condition Artistic individuals outperformed others in creativity tests (Idea
Generation and Idea Fluency, in the latter of which Artistic types scored just
behind Investigative types), in the g-free condition Artistic individuals fell
behind their more extraverted Social and Enterprising peers. Conventional
types tended to dominate a set of aptitudes in both the g and g-free
conditions: Finger Dexterity, Clerical, and Numerical Aptitudes, although
at least in the g-retained condition Investigative types also showed strengths
in these areas.
The pattern of correlations between g and interest scales was fairly
consistent with Carson’s (1996) findings in the Montreal study, with
correlations of .15, .29, .21, .01, -.07, and .12 between g and R, I, A, S, E,
and C, respectively (all correlations except those for S and E were significant
at the .05 level, N = 547). Investigative interests had the strongest positive
correlations with g, and Enterprising interests had the weakest. As was the
case for Carson, the pattern of correlations mirrored Holland’s hexagon.

Table 5 compares Lowman’s hypotheses (as shown in Table 1) to the


results of the present study and those of Carson (1996). High, medium,
and low designations for traits across personal orientation groups were
assigned through reference to the results reported in Table 3 and in Carson
(p. 326). For all groups except Enterprising, differences were deemed
significant if the individual multivariate tests resulted in differences at the
.05 level. For example, on inductive reasoning the Realistic group was high
and the Investigative group low. This left the problem of deciding for which
variables Enterprising individuals differed from the average. Inspection of
the table suggested that a cutoff for the absolute value of Z > .25 would
provide a conservative indicator of a real effect, given that in 97% of the
instances in which a variable had an absolute Z score > .25, the
corresponding individual multivariate test was significant at the .05 level
(30/31 .97). This cutoff probably provided a conservative indicator of a
=

significant effect for Enterprising types, given that the overall level of
agreement between a cutoff score of absolute Z > .25 agreed with the results
of individual multivariate tests in only 91% of cases (91/100 .91), with the =

Z > .25 cutoff being the more conservative. Because 8 of Lowman’s 23


variables had no corresponding variable in either Carson’s or the present
data set, only 90 of the specific cells in Lowman’s Matrix could be directly

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


99

Table 4
Rank Order of Personal Orientation Groups by Mean Scores
on Aptitude (g and g-free) and Personality Scales

Note. Abbreviations for personal orientation, determined by high-point score on General


Occupational Themes of the Strong Interest Inventory: R Realistic, I Investigative,
= =

A Artistic, S Social, E Enterprising, and C Conventional. BAB Ball Aptitude


= = = = =

Battery, and 16PF Sixteen Personality Factor Questionnaire.


=

tested with the data sets at hand. Overall, 57% of Lowman’s hypotheses were
confirmed, with 3% (n 3) of the mismatches being extreme (i.e., when
=

Lowman predicted a low [or high] score, but in the observed data an extreme
score in the other direction was observed). Within traits and across groups,
agreement with Lowman’s Matrix ranged from a low of 33% (chance level)
to a high of 83% (five of six groups). Within groups and across traits,
agreement ranged from a low of 47% (Investigative, Artistic, and Social) to
a high of 73% (Realistic).

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


100

Table 5
Differences From Lowman’s Matrix Identified
Through This Study and Carson (1996)

Note. R Realistic, I = Investigative, A= Artistic, S


=
Social, E Enterprising, and
= =

C Conventional. Degree of match of the results to Lowman’s hypotheses was determined


=

through two steps: (a) assign score-level categories based on cut-off scores (low with Z
score < .25, high with Z score > .25, otherwise medium) and (b) assigned degree of
match categories based on combinations of score-level categories between Lowman’s
hypotheses (see Table 1) and the present study (match or &dquo;a&dquo; if the same, moderate
mismatch or &dquo;x&dquo; if the combinations are low + medium or high + medium, major mismatch
or &dquo;X&dquo; if the combinations are low + high).

fiUnderlined tests are from the Differential Aptitude Tests-Adaptive (g general =

cognitive ability, MR Mechanical Reasoning, CSA Clerical Speed and Accuracy,


= =

VR Verbal Reasoning); the remaining tests are from the Ball Aptitude Battery (NC
= =

Numerical Computation, IG Idea Generation, g =


general cognitive ability, CL = =

Clerical, AR Analytical Reasoning, SA Shape Assembly).


= =

Discussion
This study suffers from the following limitations: use of a primarily
clinical career client sample that was also higher than the population
average on g, with both factors affecting generalizability to the general
population in unknown ways; reliance on a nearly all-White sample; use of
a relatively untested method of determining the Big Five personality traits
(with the fourth edition 16PF); restrictions of analyses to combined sex
groups, despite the discovery of some sex-related differences in mean scores

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


101

on some variables and in distributions across personal orientations; selection


of clients for the sample on the basis of GOT high-point code, leaving issues
of level (high versus low scores) and profile differentiation unexamined; and
an incomplete sampling of the universe of abilities, given that not all
current BAB tests were available for this sample (e.g., tests for memory,
degree of verbal socialization, and musical aptitude) and given that tests of
other abilities were not used. All of these limitations are unavoidable,
given the archival nature of the study (in the case of sample ethnicity,
level of g, and reliance on earlier editions of the interest and personality
inventories used), or would require an excessively lengthy report or a much
larger sample (in the case of sex-related differences and alternative methods
of examining impact of interest score elevation and differentiation). Efforts
to extend or replicate this study might attend to any or all of these
limitations. The addition of other aptitude tests that extend the breadth of
aptitude coverage would be particularly welcome; perhaps Lowman’s (1991)
review of ability measures across Holland’s types might guide such
investigations.
The primary goal of this study was to test Lowman’s Matrix. Over half of
the specific predictions he made were confirmed through this study. However,
several of the findings of this study and those of Carson (1996) were at
variance with aspects of his matrix. Of course, such differences may be
due to a variety of factors irrelevant to the matrix. For example, these
samples may have been unusual in some way or an indicator used might not
have measured a construct equivalent to one Lowman described. For
example, the sample used in this study had a higher than average level of
g; Gottfredson (1996) suggested, based on extrapolations from the structure
of her (1986) OAP Map, that this may have affected several relations
between variables, perhaps particularly between g and interest scores.
This possibility certainly merits further investigation, particularly with a
sample less highly selected on g.
The use ofg-free aptitude scores in this and in Carson’s (1996) study may
account for some of the differences from Lowman’s Matrix. Such a strategy,
based on the data reported in Tables 3 and 4, clearly lowers the absolute
scores on all abilities for Investigative types and raises them a comparable
amount for all Enterprising ones. Many of the high specific aptitudes often
attributed to Investigative types would appear to be due to the high level
of g that characterizes that type, rather than to specific aptitudes per se. An
exception to this was the high Spelling (from the DAT-A) scores of
Investigative types (reported in Carson, 1996), but spelling performance
presumably depends on academic learning, rather than an underlying
aptitude. Thus, Investigative types might be characterized as having high
g and being dominant only in those ability areas that arise from assiduous
attending to the fine points of routine academic work. Investigative types
also appear to be low on Inductive Reasoning, which requires essentially a
holistic, parallel-processing approach to detection of visual and conceptual
similarities across objects.
Realistic types are very different from Investigative ones when it comes
to Inductive Reasoning, and to some degree Mechanical Reasoning, where

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


102

Realistic individuals achieve the highest scores. Another difference between


Investigative and Realistic types is that the latter type performs well at both
hands-on and paper-and-pencil spatial tasks (the latter involving a purely
mental paper-folding task), but Investigative types have a relative strength
(at least compared to Realistic individuals) only in the paper-and-pencil
activity. Similarly, Realistic types have some deficits where Investigative
ones are strong, in spelling and language usage. An exception to this trend

may be Verbal Reasoning (from the DAT-A, an analogies test), on which


Realistic types scored high. Why Realistic individuals should have an island
of relative verbal strength on a test that essentially taps analogical reasoning
is not clear, but such results are not necessarily inconsistent with a
nonverbal strength in inductive reasoning. (This result, in particular, needs
replication, given the small sample size for the Realistic type in the Montreal
study.) Given space constraints, the reader is referred to Tables 3 and 4 for
further independent exploration of differences across types. In summary, the
data reported in this study should contribute to moving Lowman’s Matrix
closer to the point where it may be used to guide routine clinical practice.
The results of this study raise additional questions about career
assessment as related to Lowman’s (1991, 1993), Holland’s (1985/1992),
and others’ theories. First, how should information about g gained through
career assessment affect interpretation and assignment of personal
orientation or vocational personality type? Investigative and Artistic types
appear to be high on g, while Enterprising ones are, on average, low.
Lowman assigned Enterprising types a high g, quite at odds with the
proposed revision. However, to be a successful Enterprising type person, one
would probably benefit from high g, especially for leadership of organizations.
It almost goes without saying that what one needs for success and what one
is interested in doing may diverge; this may be especially true for
Enterprising type individuals. Nevertheless, both this study and the
Montreal one found negative correlations between Enterprising interests
and g. Perhaps the oft-cited assumption that g and Enterprising interests
are positively correlated has arisen from a need to view our leaders as
intelligent and competent, able to properly lead and take care of us. On the
other hand, Gottfredson (1996) has suggested that this finding might be the
result of a restriction of range at the high end of g, which, given the
relations between g and Enterprising-related categories in her (1986) OAP
Map, might also account for the present finding. This alternative
explanation therefore suggests the need for replication using a sample
less highly selected on g, for example, an entire class of 10th graders at a
public high school.
Second, how do the rank orderings of personal orientation groups by
mean aptitude scores (see Table 5) provide the basis for theory extensions
related to the nature of interdomain constructs such as creativity? The
examination of such rank order data is similar to that reported by Randahl
(1991), and such an approach also bears similarity to the identification of
core profiles of different groups on tests such as the Wechsler (McDermott,
Fantuzzo, & Glutting, 1990, pp. 296-298) as well as to the other traditional
ipsative approaches to profile interpretation such as those advocated in
leading assessment texts (e.g., Kaufman, 1990; Sattler, 1988).

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


103

In particular, the present results have bearing on Holland’s (1985/1992)


hypothesis relating vocational personality to creative performance, which
states that &dquo;The likelihood of creative performance increases the more a
personality pattern resembles the following order from high to low: Artistic,
Investigative, Social, Enterprising, Realistic, and Conventional&dquo; (p. 31).
Holland, Johnston, Hughey, and Asama (1991) provided a brief review of the
literature regarding this hypothesis as well as new additional data
supporting it. The findings of the present study are also generally supportive
of the hypothesis, although with qualifications. The two BAB tests that
most directly assess creativity are Idea Generation and Idea Fluency, which
(in the g-retained condition) had associated personal orientation patterns
of ASIECR and IASERC, respectively. These are fairly close to the AISERC
predicted by Holland and for which he has found repeated support across
studies (Holland et al.). However, in the g-free condition the results are quite
different, the pattern for Idea Generation showing the Social type dominating
(SAEICR) and the pattern for Idea Fluency showing the Enterprising type
dominating (EASIRC). These results suggest that Social and Enterprising
types have previously unsuspected creative strengths that tend to come to
the fore only in situations in which g-associated performance is not required.
One is even tempted to draw linkages between these findings and
Gardner’s (1983) theory of multiple intelligences; it seems possible that g-
free creativity in the form of Idea Generation scores may provide an indicator
of the more empathic antrapersonal intelligence (on which Social types
would be expected to be higher than Enterprising ones), whereas g-free
creativity in the form of Idea Fluency scores may provide an indicator of the
more persuasive interpersonal intelligence (on which Enterprising types
would be expected to be higher than Social ones). Also, high scores on g-free
Idea Fluency may indicate strengths in brainstorming, a valued skill in many
organizations (cf., de Bono, 1971), particularly those that are more
Enterprising in nature.
Third, and given that the results of this study provide further empirical
support for the characterizations of Holland’s (1985/1992) and Lowman’s
(1991) descriptions of person-based types through separate consideration of
g and g-free abilities, it seems to follow that this same approach could be
applied profitably to the study of work environments. Namely, work-related
tasks might be conceptualized as having features that to varying degrees
benefit for their successful completion from either g or various specific
abilities. Academic tasks and those tasks involved in the early period of
learning on the job presumably relate quite closely to g, but subsequent job
performance may depend relatively more on the quality of performance on
a host of tasks that, perhaps with the exception of Investigative
environments, call primarily for g-free abilities. Therefore, job analyses
might profit from the separation of g-intensive and g-free components of
work-related tasks. Empirical research on the g and g-free features of work
environments would facilitate the creation of person-job matching tools
useful in both counseling and industrial applications (as anticipated by
Levine, 1949). Thus, the present findings appear to have implications not
only for counseling-related career assessment, but also for assessment

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


104

related to industrial applications such as selection, classification, and


organizational development.
References
Ackerman, P. L., & Heggestad, E. D. (1997). Intelligence, personality, and interests:
Evidence for overlapping traits. Psychological Bulletin, 121, 219-245.
Ball Foundation. (1995). Ball Aptitude Battery technical manual (Rev. ed.). Glen Ellyn, IL:
Ball Foundation.
Bennett, G. K., Seashore, H. G., & Wesman, A. G. (1982). Differential Aptitude Tests:
Administrator’s handbook. San Antonio, TX: The Psychological Corporation.
Bizot, E. B. (1993). BAB-GATB reanalysis. Unpublished manuscript. (Available from
author, Ball Foundation, Glen Ellyn, IL)
Carroll, J. B. (1993). Human cognitive abilities: A survey of factor-analytic studies.
Cambridge, England: Cambridge University Press.
Carroll, J. B. (1996). A three-stratum theory of intelligence: Spearman’s contribution. In
I. Dennis & P. Tapsfield (Eds.), Human abilities: Their nature and measurement (pp. 1-18).
Mahwah, NJ: Erlbaum.
Carson, A. D. (1996). Aptitudes across Holland’s types Implications for school-based
counselling. McGill Journal of Education, 31, 319-332.
Carson, A. D., Stalikas, A., & Bizot, E. B. (1997). Correlations between the Myers-Briggs
Type Indicator (MBTI) and measures of aptitudes. Journal of Career Assessment, 5, 81-104.
Cattell, R. B., Eber, H. W., & Tatsuoka, M. M. (1970). Handbook for the Sixteen Personality
Factor Questionnaire ( ), (4th ed.). Champaign, IL: Institute for Personality and Ability
16 PF
Testing.
de Bono, E. (1971). Lateral thinking for management: A handbook of creativity. (No city)
England: The Presidents Association.
Desmarais, L. B., & Sackett, P. R. (1993). Investigating a cognitive complexity hierarchy
of jobs. Journal of Vocational Behavior, 43, 279-297.
Fish, L. (1988). Why multivariate methods are usually vital. Measurement and Evaluation
in Counseling and Development, 21, 130-137.
Gardner, H. (1983). Frames of mind: The theory of multiple intelligences. New York: Basic
Books.
Goldman, S. H. (1987). BAB tests and their relation to GATB, and prediction of success on
66 occupational aptitude patterns (OAP) (Technical Rep. No. 25). Glen Ellyn, IL: Ball
Foundation.
Gottfredson, L. S. (1986). Occupational Aptitude Patterns Map: Development and
implications for a theory of job aptitude requirements [Monograph]. Journal of Vocational
Behavior, 29, 254-291.
Gottfredson, L. S. (1996, August). Inter-domain relationships and applications: A discussion.
In R. L. Lowman (Chair), Contemporary issues in scientific psychology Inter-domain
relationships and applications. Symposium conducted at the meeting of the American
Psychological Association, Toronto, Ontario.
Hansen, J.-I. C., & Campbell, D. P. (1985). Manual for the Strong Interest Inventory (4th
ed.). Stanford, CA: Stanford University Press.
Holland, J. L. (1959). A theory of vocational choice. Journal of Counseling Psychology, 6,
35-45.
Holland, J. L. (1992). Making vocational choices A theory of vocational personalities and
work environments (2nd ed.). Odessa, FL: Psychological Assessment Resources. (Original
work published 1985)

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014


105

Holland, J. L. (1996). Exploring careers with a typology: What we have learned and some
new directions. American Psychologist, 51, 397-406.
Holland, J. L., Johnston, J. A., Hughey, K. F., & Asama, N. F. (1991). Some explorations
of a theory of careers: VII. A replication and some possible extensions. Journal of Career
Development, 18, 91-100.
Kaufman, A. S. (1990). Assessing adolescent and adult intelligence. Boston: Allyn & Bacon.
Levine, A. S. (1949). Correcting special ability test scores for general ability. Journal of
Applied Psychology, 33, 566-568.
Lowman, R. L. (1991). The clinical practice of career assessment: Interests, abilities, and
personality. Washington, DC: American Psychological Association.
Lowman, R. L. (1993). The inter-domain model of career assessment and counseling.
Journal of Counseling and Development, 71, 549-554.
McBride, J. R. (1986, August). A computerized adaptive edition of the Differential Aptitude
Tests.Paper presented at the meeting of the American Psychological Association, Washington,
DC.
McDermott, P. A., Fantuzzo, J. W., & Glutting, J. J. (1990). Just say no to subtest analysis:
A critiqueon Wechsler theory and practice. Journal of Psychoeducational Assessment, 8,
290-302.
Moe, K. C. (1988). User manual for the Differential Aptitude Tests: Computerized adaptive
edition, IBM PC&reg; version. San Antonio, TX: The Psychological Corporation.
Randahl, G. J. (1991). A typological analysis of the relations between measured vocational
interests and abilities. Journal of Vocational Behavior, 38, 333-350.

Ree, M. J., & Earles, J. A. (1994). The ubiquitous predictiveness of g. In M. J. Rumsey,


C. B. Walker, & J. H. Harris (Eds.), Personnel selection and classification (pp. 127-136).
Hillsdale, NJ: Erlbaum.
Ree, M. J., & Earles, J. A. (1996). Predicting occupational criteria: Not much more than
g. In I. Dennis & P. Tapsfield (Eds.), Human abilities: Their nature and measurement (pp. 151-
166). Mahwah, NJ: Erlbaum.

Rischall, R. Y., Bizot, E. B., & Lobsenz, R. (1995, May). Differential abilities, personality,
and interests across job content categories. Paper presented at the meeting of the Society for
Industrial and Organizational Psychology, Orlando, FL.
Russell, M., & Karol, D. (1994). The 16PF fifth edition administrator’s manual. Champaign,
IL: Institute for Personality and Ability Testing.
Sattler, J. M. (1988). Assessment of children (3rd ed.). San Diego, CA: Author.
Schmidt, F. L. (1988). The problem of group differences in ability test scores in employment
selection. Journal of Vocational Behavior, 33, 272-292.
Spearman, C. (1927). The abilities of man: Their nature and measurement. London:
Macmillan.
Spearman, C., & Jones, LL. W. (1950). Human ability. London: Macmillan.
Spokane, A. R. (1996). Holland’s theory. In D. Brown & L. Brooks (Eds.), Career choice and
development (3rd ed., pp. 33-64). San Francisco, CA: Jossey-Bass.
U.S. Department of Labor. (1970). Manual for the USES General Aptitude Test Battery
Section III Development. Washington, DC: U.S. Government Printing Office.
Vacha-Haase, T., Walsh, B. D., Kapes, J. T., Dresden, J. H., Thomson, W. A., Ochoa-
Shargey, B., & Camacho, Z. (1994). Gender differences on the Values Scale for ethnic minority
students. Journal of Career Assessment, 2, 408-421.

Downloaded from jca.sagepub.com at UNIVERSITY OF WINDSOR on October 23, 2014

You might also like