| The Sixteen
Personality Factor Questionnaire (16PF; Cattell,
Eber, & Tatsuoka, 1970) and its companion
questionnaires for younger persons, the High School
Personality Questionnaire (HSPQ; Cattell, Cattell,
& Johns, 1984) and the Children's Personality
Questionnaire (CPQ; Porter, Cattell, & Staff,
1975), published by The Institute for Personality
and Ability Testing (IPAT), are useful to counselors
in that they provide a sample of the client's
verbal self-presentation that is broad based and
structured. The parent instrument of the three,
the 16PF, is consistently ranked among the most
used and recommended personality questionnaires
(Piotrowski & Keller, 1989), and it ranks
high among personality questionnaires in the number
of research articles that make reference to it
(see, for example, Graham & Lilly, 1984, p.
234).
Several books have been published on the 16PF
that are recommended to the practitioner. Karson
and O'Dell's (1976) Guide to the Clinical Use
of the 16PF deserves mention as a book for practitioners,
as does Heather Cattell's (1989) The 16PF: Personality
in Depth. Krug's (1981) Interpreting 16PF Profiles
provides a unique approach, as a compendium of
data about various typical patterns of 16PF scores.
Krug's (1977) Psychological Assessment in Medicine,
though a bit dated, gives the practitioner a solid
grasp of the interaction of 16PF variables and
physical well-being. For the counselor in primarily
clinical settings, Meyer's (1989) Clinician's
Handbook puts scores from the 16PF and other popular
tests in the context of Diagnostic and Statistical
Manual of Mental Disorders, revised edition, (DSM-III-R;
American Psychiatric Association, 1987) diagnosis
and treatment.
Background
To understand the place of the 16PF in the field
of human assessment generally, it is helpful to
think of three domains:
Ability -- This is the domain of what a person
can do--his or her talents, skills, proficiencies.
The archetypal statement in this realm is "I
can do X."
Motive -- This is the domain of movement, drive,
or tendency toward something or someone. The archetypal
statement in this realm is "In this circumstance,
I want X."
Personality -- This is the domain of personal
style. The archetypal statement in this realm
is "In this circumstance, I am X."
Again, there are three ways to know about a person,
three data sources (Cattell & Johnson [1986])
or levels (Leery, 1957) of assessment:
Life data -- From this data source, personality
is revealed in everyday behavior, usually reported
by someone other than the person being evaluated.
Examples are behavioral observations, ratings,
school grades, and interview observations.
Question data -- This data source consists of
the person's own conscious verbal self-presentation,
in a given circumstance. The presentation may
be oral, as in an interview, written in essay
form, or in responses to multiple-choice questions,
as on a personality questionnaire.
Test data -- From this data source, personality
is revealed by a person's response to an artificially
contrived situation, such as an ability test or
a projective test. It is not conscious self-presentation,
because the overt task is not self-description.
Personality characteristics are inferred from
what the person does rather than from direct statements
about what kind of person one is. More than from
question data, this data source can reveal a person's
fantasy projection, as on projective tests.
the 16PF samples a broad base of questions about
normal personality style. This broad sampling
is, in fact, a distinguishing characteristic of
the 16PF. The authors of the 16PF (R. B. Cattell
and his collaborators) set out consciously to
sample the realm of personality in verbal description.
They began with Allport and Odbert's ( 1936) collection
of several thousand personal descriptors and reduced
it by content analysis to a shorter list. They
made rating scales from this smaller number and
had fraternity brothers, who might be expected
to know each other well, rate one another on all
the descriptors. Ratings obtained this way represent
an optimum in L-data ratings, averaged over many
raters. For example, in any given fraternity,
if 30 young men rated each other, any given person's
score on a descriptor would be the average among
the other 29 fraternity brothers.
The ratings were then submitted to a statistical
procedure called factor analysis, the purpose
of which is to discover, in a large number of
variables, some smaller set that will serve to
explain the entire domain. At this point in the
research, the factor analysis indicated that 12
variables were enough to account for the domains
sampled. From these 12 rating factors (L-data)
of personality, the researchers wrote multiple-choice
items to represent the domain in a similar way.
These questions were tried out on other large
groups and the resultant "question data"
(Q-data) were factor analyzed, yielding 16 factors
of personality. These became the primary factors
of the 16PF.
Instruments and Scales
By reason of the items and population involved
in its origins, the 16PF is appropriate for persons
no younger than 16 years of age. For adolescents
and children, two other questionnaires were developed
by modifying items from the adult research, trying
them out with youngsters of varying ages, and
submitting the results to factor analyses. By
a series of overlapping studies, two new questionnaires
were developed, the HSPQ for youngsters ages 11
to 17 or 18 years, and the CPQ for children ages
8 to 12 years. Thus, the researches of Cattell
and his co-workers yielded three linked instruments,
with similar scales, over an age span from childhood
to full adulthood. At the high end, the 16PF has
been used successfully with persons older than
70 years of age. This ability to provide scores
on similar variables over a wide age range is,
to our knowledge, unique to the 16PF family.
Administration and Scoring
The assessment of personality by questionnaire--the
16PF, HSPQ, or CPQ--involves recording a person's
conscious self-presentation in some specific circumstance.
One's self-presentation may be expected to change
depending on context, on the degree of confidence
one has in the person administering the inventory,
on the perceived purpose of the results, and so
on. For this reason, it is important to establish
a comfortable, unambiguous context for the person
completing the inventory. Otherwise, the administration
instructions are straightforward, and because
the questionnaires are untimed, they do not require
particularly active supervision. About 80% of
persons finish Form A of the 16PF in 40 minutes.
Form C, being shorter, is usually completed in
half an hour.
Scoring any of the inventories by hand for primary
factors only is quick and simple. Raw scores are
obtained by placing two grids over the answer
sheet and adding points for marked choices, factor
by factor. Raw scores are transformed to standard
scores (STENS) by a look-up table. For an experienced
scorer, the whole process takes 6 or 7 minutes.
Getting supplementary scores, secondary factors
and others, takes another 5 or 6 minutes for an
experienced person.
Computerized scoring is available from the test
publisher and from several other licensed sources.
Practitioners may mail the answer sheets, access
the computer using their own computer by modem,
or, in some circumstances, lease scoring software.
Computerized scoring services usually provide
not only the 16 primary scores but also the secondary
factors, as well as derived scores specific to
application areas--career counseling, marriage
counseling, or clinical diagnosis and treatment.
Reliability
In this section I address three issues that are
traditionally called "reliability":
stability of scores--how they hold up over time;
the internal consistency of the scales--how highly
the items in a scale correlate with each other;
and interform correlations--the degree to which
the various forms of the 16PF, HSPQ, and CPQ seem
to measure the same thing.
Stability, Internal Consistency And Interform
Correlations
There has been some criticism of the 16PF family
of inventories in respect to stability and internal
consistency, but in-depth study of the issue has
shown that the concerns are based on misunderstandings.
Table 1 presents data (smoothed) from a meta-analysis
by Schuerger, Zarrella, and Hotz (1989), which
summarized 106 sources and more than eight instruments.
Of the studies, 19 involved the 16PF, 5 of them
the HSPQ. CPQ data are from the handbook. Values
for the second-order factors are from Zarrella
(1989). These values are comparable to those for
the other instruments, as far as temporal stability
is concerned. Internal consistency reliability,
taken from a single occasion, unlike temporal
stability (test-retest) reliability, is calculated
solely from item intercorrelations and the number
of items per scale. In this respect, the 16PF
has item intercorrelations slightly above average
for personality questionnaires, but has relatively
few items per scale. It has far fewer items (about
16 per scale) than does the MMPI, for example,
which averages about 50 per scale. Because of
this small number of items, the internal consistency
values for the 16PF and its junior tests are lower
than is common among personality questionnaires,
except for the second-order values, which compare
favorably with those for other instruments.
Table 2 presents averaged correlations among
forms and between instruments. They may be compared
to two kinds of standards: (a) typical stabilities,
as in Table 1, and (b) interform correlations
from other instruments. The figures in the table
are calculated by correlating the same-named scale
across forms or across instruments, as indicated.
For example, Factor A-Form A is correlated with
Factor A-Form B, Factor B--Form A is correlated
with Factor B--Form B, and so on.
The interform and interinstrument correlations
presented in the table are all variants of internal
consistency reliability, in that they are functions
of the correlations among items and the number
of items per scale. The relative magnitude of
the values in the table illustrates this function.
For example, Forms A and B, which have on average
about 12 items per scale, show the highest average
correlation between similarly named scales.
These figures are close to those found for the
straight internal consistency reliability of the
instruments presented in the last column of Table
1. They may be compared with those found in other
instruments for which alternate forms are available.
One example is the correlations found between
the Wechsler Adult Intelligence Scale (WAIS) subtests
and those on the new WAIS-R, which average .57
(Jackson, 1984). The correlations between corresponding
scales on the COP System Interest Inventory (COPS)
and the revised COPSR average .72 over 14 subtests
(Knapp & Knapp, 1984). In both of these instances
the scales purport to measure the same constructs.
Correlations from the 16PF family are somewhat
lower than are those from the WAIS-WAIS-R comparison,
and quite a bit lower than those from the COPS.
The differences may be attributed to number of
items for the comparison with the WAIS, but not
for that with the COPS, which actually has about
the same number of items per scale as the 16PF.
The practical implication of these correlations
is that one usually advises caution in making
inferences from a form or instrument based on
experience with an alternate form or instrument,
except with the second-order factors.
Validity: The 16PF In Career, School, And Clinic
Data On Career And School Application
For both career and school applications, it is
convenient to distinguish job or school proficiency
from job or school satisfaction. Anticipating
a person's job satisfaction from personality information
is done by noting the ways a person's characteristics
are like those of others persons in various jobs--Holland's
"congruence" principle (Holland, 1973).
It is the old principle of "like to like."
The more similar a person is to others on the
job, the more likely the person will feel comfortable.
There are, of course, many ways to be dissatisfied
with a job, so the principle works only approximately,
but it is a useful way to conceptualize fit of
a person to a job.
Predicting a person's performance on a job is
not the same as predicting a person's satisfaction
or comfort with a job. One need only think of
persons one knows who are very proficient at a
job but uncomfortable with it.
Finding A Job That Fits: Job Satisfaction
To advise a person about fit to an occupation
(Holland's congruence principle) one must know
what personality profiles characterize various
occupations. Much of this kind of information
exists for the 16PF in the handbook and elsewhere.
For example, DiFiore (1981), Franklin (1983),
Nowak (1980), and Anonsen (1985) have all contributed
to the understanding of specific occupations in
terms of 16PF occupational patterns. Guyer (1984),
Johns (1985), and Nasvytis (1988) have treated
broader issues of fit to occupation in terms of
personality. Some of their efforts are reflected
in Table 3 (after Schuerger et al., 1990), which
presents average scores for a number of occupations,
arranged according to Holland's typology. To apply
Holland's principle concretely here, one simply
compares the client's score patterns with those
of the various known occupational patterns.
The question is best put "In what ways do
this client's scores compare with those of the
various occupations?" For example, the client
in Figure I resembles Realistic and Artistic type
occupations in her introversion. She differs from
Realistic type occupations in her low self-control,
and in this respect is more like Artistic type
occupations. Her low Practicality score, in the
sensitive direction, puts her more in the artistic
group as well. She is unlike those in Enterprising
type occupations in most respects but shares the
sensitivity with Social types. With Conventional
types, she shares introversion and low independence.
This process of comparison might be done by a
practitioner and can also be done mathematically
by a computer. Results of calculating similarity
to the various occupations for large numbers of
persons have been encouraging. Among a sample
of 1,339 persons representing 21 occupations,
approximately 50% of the sample showed their highest
similarity to an occupation with a Holland type
appropriate to their own current occupation. Counselors
will recognize that using only the highest similarity
is a very stringent test and that the result compares
very favorably with what one usually sees reported
for interest measures, for example. Schuerger
et al. ( 1990) also reported that similarities,
when measured on a small group of graduate students,
were very stable over time.
After 8 months, 85% of the students had the identical
highest occupational similarity. A total of 93%
had as highest either the identical occupation
or an occupation within the same Holland type
as the Holland type of their prior highest occupational
similarity. As many as 7 years later, almost half
had the identical highest similarity, and 87%
had either identical highest similarity or of
same Holland type. In another small sample, of
undergraduate psychology majors, the durability
of these similarities was tested when students
were instructed to try to "look like a personnel
officer." Under these conditions, 60% had
the identical highest similarity on the second
occasion, and 90% were within the same Holland
type. Parenthetically, very few managed to increase
their similarities to the target occupation, personnel
officer.
New Directions
The major need for the 16PF family of instruments
is to make validity information available in forms
more familiar to clinicians and vocational practitioners.
These instruments have seen a great many researches--the
16PF is in the top 10 instruments in this respect,
and the HSPQ on its own seems to be in the top
50--but most of the information is not made available
in convenient form. That is, for a practitioner,
it is useful to know that if a certain scale is
over 8, say, and another scale is below 3, then
the likelihood of a certain outcome or even of
a certain broad description is high. Better yet,
that it is 3 out of 5, or some such. To develop
such availability, new research is needed, but
much can be done with current information.
|