Sei sulla pagina 1di 14

PERSONNEL PSYCHOLOGY

1991.50

IDENTIFYING POTENTIALLY USEFUL PERSONALITY


CONSTRUCTS FOR EMPLOYEE SELECTION

PATRICK H. RAYMARK
The Ohio State University-Newark
MARK J. SCHMIT
Department of Management
University of Florida
ROBERT rd. GUION
Bowling Green State University

The Personality-RelatedPositionRequirements Form (PPRF), a job anal-


ysis form to be used in making hypotheses about personality predic-
tors of job performance, is described. The Big Five personality factors
provided an organizing framework for the PPRF. Subsequent develop-
ment resulted in identifying 12 specific sets of items for facets of each
of the Big Five. A study was conducted by gathering job descriptions
on 260 different jobs to determine if the PPRF could reliably differ-
entiate jobs; such evidence was found. The PPRF is offered to both
researchers and practitioners for use, refinement, and further testing
of its technical merits and intended purposes.

Description of the Practice

Douglas Jackson, in his presidential address to the Division of Eval-


uation and Measurement of the American Psychological Association
(Jackson, 1990), said that the dormancy in the use of personality vari-
ables for personnel selection can be laid to job analytic procedures
that do not encourage their consideration. Although some job analy-
sis forms appear to identify some aspects of personality-related position
requirements (e.g., Position Analysis Questionnaire, PAQ; McCormick,

The authors, in addition to Pilar Delaney, Michelle Brodke, Bob Hayes, Sandra
Martens, Karen Mattimore, Laura Mattimore, and Murray Weaver were the members
of the team that developed the instrument we describe in this article; though their tenures
in the groupvaried widely, all of them made significant contributions. Gracious thanks are
extended to all the individuals who contributed their time, effort, opinions, or comments
on the project.
Mark J. Schmit is now with Personnel Decisions International, Minneapolis,MN.
Correspondence and requests for reprints should be addressed to Patrick H. Raymark,
Department of Psychology, The Ohio State University-Newark Campus, 1179 University
Drive, Newark, OH 43055.
Ken Perlman with Lucent technologies was the guest Editor on this article.

COPYRIGHT 0 1997 PERSONNEL PSYCHOLOGY, INC.

723
724 PERSONNEL PSYCHOLOGY

Mecham, & Jeanneret, 1977), these attempts have been rather unsys-
tematic and incomplete in their coverage of potentially relevant person-
ality variables.
Ideas for selection procedures may come from a hunch, other psy-
chologists, or test catalogs, but modern literature urges the development
of testable hypotheses about predictor and criterion construct variables
and the choice of valid instruments for measuring them (Binning & Bar-
rett, 1989; Landy, 1986). Selection hypotheses generally emerge from
an understanding of jobs based on job analysis. Most job analysis in-
ventories are quite clear in providing help for hypothesizing ability or
aptitude variables tliat might make good predictors but are less clear
for those traits more closely associated with personality variables. If a
job analysis method emphasizes only cognitive or psychomotor aspects
of jobs, it is likely that only cognitive or psychomotor predictors will be
hypothesized. Therefore, an approach to job analysis explicitly directed
to generating hypotheses about relevant personality variables would be
helpful. This paper discusses the development of such an instrument,
the Personality-RelatedPosition Requirements Form (PPRF).
This instrument was designed to identify aspects of work potentially
related to individual differences in personality. If personality traits are
relevant to some aspects of job performance, and if they are not identi-
fied and measured, they will be overlooked for selection. Although the
PPRF was designed to aid in the identification and development of se-
lection hypotheses that certainly does not guarantee that the identified
personality variables will successfully predict performance. The value
of the PPRF is as a supplement to other job analysis techniques which
will allow a more thorough examination of the job performance domain,
specifically that part of the domain that is related to personality vari-
ables.

Development of the PPRF

On the basis of several job analysis forms, personality measures, and


job analysis reviews (Gael, 1988, Harvey, 1991), it was decided to use
the following stem for all items: Effective performance in this position
requires the person to.. . A second decision was to use the following
3-point response scale format: 0 = not require4 1 = helpful and 2 = es-
sentiaL The initial stage of item generation (conducted without the aid of
a theoretical framework) resulted in a list of 185 statements that might

A technical report, containing the entire PPRF, scoring instructions, and additional
details concerning the development and scaling of the PPRF is available upon request
from the first author.
PATRICK H.RAYMARK ET AL. 725

be used in the analysis of jobs and lead to the formation of hypothe-


ses about personality variables as predictors of job performance. An
attempt to identify a work-related structure by sorting the position re-
quirement items was unsuccessful; therefore, existing taxonomies were
considered (Browne & Howarth, 1977; Goldberg, 1981; Norman, 1963;
Wiggins, 1980). The taxonomy that seemed to fit the position require-
ment items best was that tagged as the Big Five (see Barrick & Mount,
1991; Digman, 1990; Goldberg, 1993; McCrae & John, 1992).
Research group members sorted the 185 items into five separate
stacks to represent the five factors. Items not readily placed in any
of these categories were placed in a sixth labeled Other. Only those
items placed in the same major dimension in 80% or more of the sorts
were retained, reducing the number of items to 92. Although these re-
maining items were consistently allocated to their respective Big Five
factor, items within the five groups seemed to suggest a variety of per-
sonality characteristics. At this point it became clear that the Big Five
factors were too broad to describe work-related employee characteris-
tics, a problem noted by others (e.g., Costa & McCrae, 1992; Goldberg,
1981, 1990; Hogan, 1987; Hough, 1992; McCrae & Costa, 1985; Nor-
man, 1963). Accordingly, the items were subdivided within the major
factor groupings into smaller but more homogenous clusters. An initial
attempt resulted in 54 clusters of items; subsequent analyses winnowed
this down to 12 clusters. Each cluster of items was then named on the
basis of the items contributing to it and on the Big Five factor to which
it had been assigned. Some subdimensions (i.e., item clusters) had very
few items, so 44 new items were written to ensure that the content do-
main of each subdimension had been adequately captured; therefore,
the item count was increased to 136. The taxonomy was hierarchical;
each subdimension was conceptuaily linked to one of the Big Five fac-
tors. The 12 subdimensions, their definitions, and the Big Five dimen-
sion to which they were assigned are presented in Bble 1.
Forty-four psychologists with extensive knowledge of psychological
aspects of work, personality theory, or both were asked to judge whether
each item was relevant to any of three subdimensions (the task was lim-
ited to three subdimensions because a request to make judgments for
all 136 items on 12 subdimensions would have imposed unreasonable
time demands on judges). Responses were received from 41 of the 44
judges. Xventy-four items were not reliably allocated to the appropri-
ate subdimension and were therefore dropped, leaving 112 items. A
new questionnaire was developed to scale (i.e., weight) the importance
of the items within the subdimensions. Each set of position require-
ments was formed into a paired-comparisonquestionnaire (i.e., one set
of paired items was developed for each of the 12 dimensions, using each
726 PERSONNEL PSYCHOLOGY

TABLE 1
Brief Definitions and Sample Items for the Subdimenswns of the PPRF
~

I. Surgency
1. General Leadership: a tendency to take charge of situations or groups, to
influence or motivate behavior or thinking of other persons. Sample items: Lead
group activities through exercise of power or authority; Xike control in group situ-
ations.
2. Interest in Negotiation: an interest in bringing together contesting parties
through mediation or arbitration or as a contesting party, an ability and willingness
to see and understand differing points of view. Sample items: Negotiate on behalf
of the work unit for a fair share of organizational resources; Mediate and resolve
disputes at individual, group, or organizational levels.
3. Achievement StrivinK an ambition and desire to achieve, to win, or to do
better than others, a desire to exert effort to advance, to do better than ones
own prior achievement. Sample items: Work beyond established or ordinarywork
period to perfect servicesor products; Work to excel rather than work to perform
assigned tasks.
11. Agreeableness
4. Friendlv Diswsition: a tendency to be outgoing in association with other
people, to seek and enjoy the company of others, to be gregarious, to interact easily
and well with others. Sample items: Represent and promote the organization in
social contacts away from work; Attract new clients or customers through friendly
interactions.
5. Sensitivitv to Interest of Others: a tendency to be a caring person in relation
to other people, to be considerate,understanding, and to have genuine concern for
others. Sample items: Listen attentively to the work-related problems of others;
Give constructive criticism tactfully.
6. Cooperative or Collaborative Work bndenw: a desire or willingness towork
with others to achieve a common purpose and to be part of a group, a willingness
and interest in assisting clients, customers, or coworkers. Sample items: Work as
part of an interacting work group; Work with one or more co-workersto complete
assigned tasks.
Ill. Conscientiousness
7. General Ifustworthiness:a pattern of behavior that leadsone to be trusted by
other people with property, money, or confidential information, a demonstration
of honesty, truthfulness, and fairness. Sample items: Refuse to share or release
confidential information; Make commitments and follow through on them.
8. Adherence to a Work Ethic: a tendency to work hard and to be loyal, to give
a full days work each day and to do ones best to perform well, a tendency to
follow instructions and accept company goals, policies, and rules. Sample items:
See things that need to be done and do them without waiting for instructions;Work
until task is done rather than stopping at quitting time.
9. Thorounhness and Attentivenessto Details: a tendency to carry out tasks
with attention to every aspect, a meticulous approach to ones own task perfor-
mance. Sample items: Examine all aspectsof written reports to be sure that nothing
has been omitted; Remain attentive to details over extended periods of time.
PATRICK H.RAYMARK ET AL. 727

lhble 1 (continued)
I\! Emotional Stability
10. Emotional Stabilitv: a calm, relaxed approach to situations, events, or peo-
ple, emotionally controlled responses to changes in the work environment situa-
tions. Sample items: Adapt easily to changes in work procedures; Keep cool when
confronted with conflicts.
V Intellectance
11. Desire to Generate Ideas: a preference for situations in which one can de-
velop new things, ideas, or solutions to problems through creativity or insight, or
try new or innovative approaches to tasks or situations. Sample items: Help find
solutions for the work problems of other employeesor clients; Develop innovative
approaches to old or everyday problems.
12. Rndencv to Think Thinas Throuah: a habit of mentally going through pro-
cedures or a sequence of probable events before taking action, a tendency to seek
and evaluate information, and toconsider consequences. Sample items: Solve corn-
plex problems one step at a time; Analyze past mistakes when faced with similar
problems

pair once). For each pair of position requirements, the respondents were
instructed to Check the statement in each pair that, in your judgment,
more strongly suggests a need for the trait. Questionnaires were sent to
145 psychologists, including the original 44. Completed questionnaires
were returned by 100of the 145 psychologists. Responses were analyzed
using Thurstones Case V (Ghiselli, Campbell, & Zedeck, 1981), the
results of which were used to eliminate 5 items and to scale the remaining
107 items for the final form of the job analysis questionnaire.
This paper describes an initial test of the usefulness of the form for
differentiating various jobs on personality-relateddimensions. Reliable
differentiation of jobs is a requirement for an effective approach to job
analysis. Accordingly, it was important to determine the extent to which
(a) the 12 subdimensions could be discriminated, and (b) jobs could be
described reliably.

Sample Charactetistics

A sample of positions was obtained through various recruitment ef-


forts. Each member of the research team sent forms to friends, relatives,
and acquaintances; a call for participants was made at an annual confer-
ence of the Society for Industrial and Organizational Psychology; and
several conscientious senior-level undergraduate and graduate students
were recruited to distribute forms to friends and relatives. Data collec-
tors were instructed to recruit participants who had held their current
position in an organization for more than 6 months and worked at least
728 PERSONNELPSYCHOLOGY

20 hours per week. In a few cases, consultants, whose contracts with


clients had included extensive job analysis, completed forms from the
results of their analyses, using information they had obtained in consen-
sus meetings with job incumbents.
This unusual approach to job sampling was intentional. Alternative
approaches might have included administeringthe PPRF for jobs within
one or a few organizations, or perhaps a mailing asking for a sample of
jobs within a national survey of organizations. The first of these options
seemed poor because it could exacerbate organizational idiosyncracies
in job definitions; organizational characteristics would be seriously con-
founded with job-characteristics. Although some such confounding is
realistic, too much of it could spuriously reduce the extent to which the
dimensions discriminate among jobs. A sample by mail for a prelimi-
nary pilot study seemed questionable partly because of expense and even
more because of the absence of any real control over the care exerted in
completing the descriptions. The procedure chosen was expected to pro-
vide a substantialvariety of jobs described by people who, either because
of personal involvement or professional standards, could be expected to
complete the forms carefully.
Job descriptions were obtained for 260 differentjobs from a total of
283 completed forms. For eight of these jobs, more than one incum-
bent in a single organization completed the inventory for the same job;
their responses were consolidated into a single consensus form. That
is, an item by item comparison was made across forms and a final rat-
ing was based on majority agreement on the item; where there was no
clear majority, the higher rating was favored (i.e., the rating leaning to-
ward essential). The 260 jobs were demonstrably different, but not
as diverse as desired. They were assigned six-digit occupational codes
from the Dictionav of Occupational Titles (US.Department of Labor,
1991). Two members of the research group assigned a code number,
based, if possible, on the job title given by the incumbent and/or on per-
sonal knowledge of the incumbentsjob. Where the job title was vague or
could not be matched in the DOT, a code deemed appropriate by both
researchers was assigned, and consensus was sought from a second pair
of coders for a final decision. Consensus was not reached for six of the
job titles.
The first digit of the code identifies 1 of 10 primary occupational
categories; tallying these categories shows the danger of the sampling
method used. About 4 jobs in 10 were managerial, more than a quar-
ter of them were in the clerical or sales classifications, and nearly that
many were professional or technical jobs. Very few of what are typically
called blue-collar jobswere in this sample. Specifically,the tally was: 0-
Professional and technical occupations (21.5%); 1-Managerial occupa-
PATRICK H. RAYMARK ET AL. 729

TABLE 2
Correlations and Internal Consistency Reliabilities
of the 12 Subsets of the PPRF ( N = 260)

1 (35)
2 .65 (.84)
3 .29 .37 (.83)
4 .24 .42 .25 (.72)
5 .48 .52 .18 .38 (.76)
6 .2E .33 .41 .13 .36 (.78)
7 .22 .32 .25 .46 -35 .19 (.72)
8 .24 .29 .44 .I9 .21 .38 .33 (.a)
9 .23 .37 .58 .14 .20 .36 .24 .49 (.92)
10 .34 .38 .19 .38 .56 .26 .41 .32 .17 (.78)
11 .45 54 .56 .29 .33 .39 .21 .40 S O .24 (.%I)
12 .39 .53 .59 .26 .32 .39 .22 .43 .70 .29 .68 (.88)
Note: Numberson the diagonal are alpha coefficients;all correlationsare significantat
p < .05.

tions (38.8%); 2-Clerical and sales occupations (27.7%); 3Service oc-


cupations (7.3%); &Agriculture, fishery, forestry, and related (0%); 5-
Processing occupations (0.8%); &Machine trade occupations (0%); 7-
Benchwork occupations (0.8%); &Structural work occupations (0.8%);
%Miscellaneous occupations (2.3%).
It is not surprising that the data-people-thingspart of the classifica-
tion code indicated that many of these jobs were fairly highly concerned
with data or with people, but most were quite low in responsibility for
things. Insofar as personality variables are concerned with interpersonal
relationships, and many of them are, this skewing of the distribution may
not pose a serious problem. The paucity of jobs emphasizingthings, how-
ever, means that some of the 12 subdimensions may be under repre-
sented in the current data.

Evidence Supporting the fiactice

Diflerentiation of Occupational Categories

The 12 set scores were determined for each of the 260 job descrip-
tions using the scaled items. The correlations among the set scores were
then calculated to determine if there was a useful level of scale inde-
pendence. Tdble 2 presents the correlations among the 12 sets and the
730 PERSONNEL PSYCHOLOGY

internal consistency reliability (i.e., alpha) of each set. All but one subdi-
mension (i.e., Set 8, Adherence to a Work Ethic) had alpha coefficients
higher than .70, indicating a satisfactory degree of internal consistency
within subdimensions. Further, the low variability of the means for Set 8
across jobs (as will be seen in Bble 3 below) suggests that this low co-
efficient alpha may be artifactual. Overall, the pattern of correlations
suggested that an acceptable degree of scale independence had been
achieved; the mix of correlations was substantial, all positive but many
were low.
'Ib demonstrate that the 12subdimensionswere useful in differentiat-
ing amongjobs, the 260job descriptionswere clustered into occupational
groups to form several samplejob descriptionprofiles for different types
of jobs. The first three digits of the DOT code assigned to each of the
260 jobs represented the occupational group to which the jobs belonged,
the first digit indicating the broadest category. The 3-digit occupational
groups with more than one job in them were selected for each of the nine
1-digit (i.e., broadest) DOT categories, providing 12 profiles; these are
presented in %ble 3. All of the mean set scores are in a range of 0 to 2,
consistent with the 0 = not required 1 = helpful, and 2 = essential scale
format.
An examination of Xble 3 suggests that the 12 sets differentiate
among the occupational groups in logical ways. For example, general
leadership scores were highest in management (DOT #s 187 and #189),
education (DOT #092), and firefighter (DOT #373) occupation cate-
gories, and lowest for janitorial (DOT #382), customer service (DOT
#239), and cashier categories (DOT #211). Friendly disposition scores
were highest in the sales clerk (DOT #290) and cashier occupatjon
groups, and lowest in the janitorial categories. The personnel admin-
istration (DOT #166), management, and accounting (DOT #160) oc-
cupation groups had high scores for the thoroughness and attentive-
ness to details dimension and the tendency to think things through di-
mension. General trustworthiness scores were highest for cashiers and
tellers. These are just a few examples of logical relationships between
the PPRF set scores and the job duties of the represented occupational
groups; more can be gleaned by close examination of lhble 3.
Also noteworthy is the lack of differentiation for some dimensions
across occupation groups. Consistent with the finding that conscien-
tiousness is important across manyjobs (Bamck & Mount, 1991), at least
two of the three conscientiousness dimensions (i.e., subdimensions 7,8
and 9) had mean scores greater than 1for each of the occupation groups.
Each of the 12 profiles in Bble 3 represented a group of jobs, rather
than one particular job. Thus, it is important to estimate the degree
of consistency of mean ratings across groups within each occupational
TABLE 3
Means and Standard Deviarwpls of PPRF Set Scores for Several Job Caregot+s
Job caterzorv
A* B* C* D* I?* F*
DOT#032 DOT#092 DOT#lM) DOT#I66 DOT#187 DOT#189
H=X N=6 N = 8 N = 10 N = 8 N = 22
PPRF Dimensions M SD M SD M SD M SD M SD M SD n
Set 1: General leadership .94 .MI 1.42 .28 1.10 .49 1.12 .41 1.65 .24 1.25 SO
Set 2: Interest in negotiation .58 .25 .69 .34 .81 .39 1.09 .42 1.30 .44 1.08 .44
Set 3: Ambition 1.01 .SO 1.18 .58 1.36 .41 1.63 .35 1.10 .41 1.51 .46
Set 4: Friendly disposition 1.19 .45 .80 .52 .69 .30 1.17 24 1.24 .37 1.19 .48
Set 5: Sensitivity to interests of others .7S .23 1.17 .38 .73 .64 1.25 38 1.09 .35 .87 .42
Set 6: Cooperative or collaborative work tendency 1.1 1 .39 1.04 .28 1.41 .43 1.65 .30 1.46 .48 1.35 .46 s
Set 7: General trustworthiness 1.50 .27 .92 .34 1.18 .I9 1.26 .12 1.56 .32 1.23 .43
Set 8: Adherence to a work ethic
Set 9: Thoroughnessand attentiveness todetails
1.26
.93
.34
.38
1.31
1.03
.33
.68
1.4.5
1.48
.32
.41
1.58
1.66
.25
.36
1.48
1.00
.44
.65
1.52
1.54
.20
.44
F
F
Set 10 Emotional stability 1.53 .36 1.53 .38 1.11 51 1.43 35 1.27 .37 1.21 .35
Set 11: Desire to generate ideas .84 .42 -90 .43 1.12 .49 1.24 .29 .98 .44 1.18 .47
Set 1 2 Xndency to think things through 1.01 .3X .X3 .58 1.41 .43 1.60 37 1.06 .65 1.40 .48
Rater agreement .a3 .66 .77 .S1 .73 .86
-4
w
'lhble 3 (continued) N
~~ ~~~

Job category
G' H* I* J* K* L*
DOT #2 I 1 DOT #239 DOT #290 DOT #373 DOT #382 DOT #922
N=9 N=4 N=7 N=4 N=3 N=2
PPRF Dimensions M SD M SD M SD M SD M SD M SD

Set I: General leadership .63 .24 .52 .31 .95 .34 1.68 .05 .33 .23 .68 .58
Set 2 Interest in negotiation .61 .S 1 50 .25 .83 -38 .88 .42 .26 .25 .39 .3s 3
Set 3; Amhition 1.30 .44 1.25 .40 1.48 .43 1.09 .ss 1.39 .81 1.09 .39
Set 4 Friendly disposition 1.35 .39 .HI .61 1.48 .1s .a9 55 .s2 .06 .67 .h3 g
Set S: Sensitivity to interests of others .87 .43 .93 .67 .81 .33 1.20 .64 .50 .35 35 .21
Set 6: Wpcrativc or collaborative work tendency 1.00 .40 1.21 .28 1.32 31 1.4 .27 1.2s 1.08 .79 .77 F
Set 7: General trustworthiness 1.71 39 .90 .3s 1.43 .3S 1.40 .49 1.23 .47 .ho .42 z
Set 8: Adherence to a work ethic 1.21 .20 1.37 .I7 1.44 39 1.58 .32 1.69 .27 1.19 .27 2
Set 9 Thoroughnessand attentiveness to details 1.24 .49 1.20 .7S .94 .32 1.08 .38 1.13 .70 1.10 .8S 5
Set I0 Emotional stability 1.38 .26 1.30 .45 1.48 . I 1 1.87 .27 .% 54 .70 .42 P
Set 11: Desire to generate ideas .47 .43 .SO .42 .93 .43 .77 .39 .% .72 .47 57
Set 12: Tendency to think things through .83 .17 .88 .67 .84 .31 1.20 .65 .90 .8S .SS .64
Rater agreement .92 .74 .8S .82 .81 .76
~ ~ ~~

A* = Occupations in computer systems user support; 3* = Occupations in preschool. primary school, and kindergarten education;
C* = Accountants, audilors, and related occupations; D* = Personnel administration occupations; E* = Service industry managers and
officials; F* = Miscellaneous managers and officials; G* = Cashiers and tellers; H' = Information and message distribution occupations;
I* = Sales clerks; J' = Firefighters, fire department; K* = Janitors; L* = Owupations in moving and storing materials and products
PATRICK H.RAYh4ARK ET AL. 733

category. Thus, an intraclass correlation of an average (Guilford &


Fruchter, 1973; Mean Square between set rating minus Mean Square
for residuals divided by Mean Square for residuals) was calculated using
each set score as a rating. For example, in the DOT #032 group, a 12 x 8
matrix of subdimensionsby raters was analyzed. The average ratings had
adequate levels of rater agreement ranging from .66 to .92 (see 'Ihble 3).
These coefficients represent the expected correlations of the averaged
set scores with averaged set scores from a similar sized group of raters
drawn from the same population.

Reiiubility of the PPRF in Describing Jobs

The responses across raters of the same job in the same organization
can be used to determine how reliable the PPRF is in describing jobs.
Data were available for four positions for which PPRFs were completed
independently by incumbents in the same organization: Brewery Super-
visor (7 incumbents); Production-line Manager (5); Sales Associate (5);
and Computer Programmer (4). An intraclass correlation of the average
ratings was computed for each of the 12 sets for each job. An overall
intraclass correlation was then derived for each job by calculating the
mean of these 12 coefficients. For example, for Set 1 (i.e., 9 items) rat-
ings by the Brewery Supervisors,a 9 x 7 matrix of items by raters was an-
alyzed. The same was done for Sets 2-12; these 12 coefficients were then
summed and divided by 12. The average intraclass correlations were .97
for the Brewery Supervisor job, .89 for the Production-line Manager,
.90 for the Sales Associate job, and .85 for the Computer Programmer
job. These findings suggest that incumbents can consistently agree on
the importance of the PPRF's behavior clusters for a job, and that a sta-
tistical combination of rating forms might yield reliable job information
relevant to the personality domain.

Implicatiomfor Practice

The PPRF has now gone through several rounds of revision and field
testing. A large number of people have been able to use it to describe a
variety of jobs. The PPRF has dimensions of important work behaviors
that were found to be internally consistent and not redundant. The 12
sets appeared to be effective in the differentiation of occupational cate-
gories. Finally, the intrarater reliability of the 12 sets was demonstrated.
Indeed, the use of four or five raters to describe a job with the PPRF was
found to result in very good reliability estimates. In summary, the evi-
dence suggests that this instrument can prove useful in the development
734 PERSONNEL PSYCHOLOGY

of hypotheses about the relationshipsbetween personality predictors and


performance criteria.
The major question left to be answered is whether the PPRF actually
does assist in identifyingvalid personalitypredictors of job performance.
This is likely to be a difficult question to answer, given the current tech-
nology for the measurement of personality traits. For now, users must
infer, on the basis of limited information about personality measures,
which instrument might fit the scales and then conduct the necessary
empirical validation work to test these hypotheses.
It is worth noting that the search for potentially useful personal-
ity variables on the basis of the PPRF need not be limited to those
subdimensions that are considered helpful or essential to effective
performance.2 It is possible that a subdimension evaluated as not re-
quired for effective performance may be indicative of a predictor that
is negatively correlated with performance (e.g., although Surgency may
be rated as not required for structured team settings, low Surgency
scores may identify higher performing workers). Therefore, although
the PPRF is intended to help in hypothesis development, it should not
be viewed as a substitute for a usersprofessionaljudgment. If one or two
of the 12 item sets dominate the description of a job, the user may infer
both personality traits and criterion constructs appropriate to those sets
(or to the most descriptive items in them), but neither trait nor criterion
constructs are discrete categories. As Hofstee, de Raad, and Goldberg
(1992) pointed out, different uses of a personality trait may emphasize
different aspects of the trait (or item set). Different instruments may
differ in how well they match different aspects of a trait, that is, have dif-
ferent construct validities for the different specifics of trait definition.
Nevertheless, evidence should be accumulated to show whether hy-
potheses developed with the help of the PPRF tend to be supported
in practical use. So far, one study has been conducted that used the
PPRF to form hypotheses about the relationship between personality
test scores and both training and job performance. Valid predictions
were made by matching scale content definitions of scales from the Mil-
lon Index of Personality Styles (Millon, 1994) with definitions of PPRF
dimensions judged by subject matter experts to be essential for success-
ful performance (pp. 103-106). Clearly, much more evidence is needed
before the scores on the PPRF scales can be assumed to provide solid
evidence that related predictors will be valid. For example, further in-
vestigations may examine the extent to which the predictors identified

We thank an anonymous reviewer for suggestingthis point.


PATRICK H.RAYMARK ET AL. 735

using the PPRF result in higher validities than other predictors not hy-
pothesized to be related to job performance for a particular job. This
process will have to be conducted for many different types of jobs.
The criterion space also must be considered invalidationefforts. The
PPRF should assist in the development of criterion measures related
to the assumed underlying personality predictor traits. The behavioral
statements on the PPRF are very general in nature, so that they can be
used with any job. These general statements can be linked to specific
tasks that are thought to lead to successful performance. Performance
ratings for these specific tasks, or outcomes associated with the tasks,
may be useful for building either specific or global job performance
criterion measures.
Certainly, use of the PPRF is likely to be an improvement over the
uninformed use of personality predictors in personnel selection. Some
deficiencies in the current study, however, make the results tentative.
First, there is limited information about its use in describing blue collar
jobs. Second, the Big Five was relied upon as an organizing taxonomy.
However, this taxonomy may be incomplete in its coverage of work-
related traits (cf. Hough, 1992). For example, there are several low inter-
correlationswithin the clusters of PPRF dimensions falling under each
of the Big Five factors. Because the PPRF dimensions have been linked
to their most appropriate Big Five factor and have been shown to be in-
ternally consistent, the low within-categoryscale intercorrelationswould
appear to be due to a limitation in the organizing taxonomy. Therefore,
future refinements of the PPRF should consider additional higher or-
der constructs and subdimensions. Referring to work styles rather than
to personality, Borman, McKee, and Schneider (1995) provided a sim-
ilar taxonomy of 7 (not 5) higher-order constructs divided further into
a total of 17 lower-order constructs; their taxonomy may offer additions
to the 12 dimensions now in the PPRF. Despite the above limitations,
the current paper presents a start toward the development of an instru-
ment specificallyaimed at personality-relatedposition requirements. At
this point, the PPRF is offered to researchers and practitioners so that
improvements, refinements, and additional tests of the efficacy of the
instrument in generating hypotheses can be conducted on a broad front.
REFERENCES
Barrick MR, Mount MK (1991). The Big Five personality dimensions and job perfor-
mance: A meta-analysis. PERSONNEL PSYCHO^, 44,1-26.
Binning JF, Barrett GV (1989). Validity of personnel decisions: A conceptual analysis of
the inferential and evidential bases. Journal ofApplied Psychology, 74, 478494.
Borman WC, McKee AS, Schneider RJ.(1995). Work styles. In kterson NG, Mumford
MD, Borman WC, Jeanneret PR, Fleishman EA (Eds.),Devefopment ofapmto&x
OCCUpatWMl Information Netwok (O*NET)content model (Vols. 1-2, pp. 1-26).
Salt Lake City, UT: Utah Department of Employment Security.
736 PERSONNEL PSYCHOLOGY

Browne JA, Howarth E. (1977). A comprehensivefactor analysis of personality question-


naire items: A test of 20 positive factor hypotheses. Multivariate Behavioml Re-
search, 12 399-427.
Costa PT, McCrae RR. (1992). NEO PI-Rpmfesswnal manuaL Odessa, F L Psychological
Assessment Resources.
Digman JM. (19%). Personality structure: Emergence of the five-factor model. Annual
Review of fychology, 41,417-440.
Gael S (Ed.). (1988). The job anabsis handbook for business, industry, and govenunent
(Vols. 1-2). New York Wiley.
Ghiselli EE, Campbell JP,Zedeck S. (1981). Measurementtheoryforthebehaviomlsciences.
San Francisco: Freeman.
Goldberg LR. (1981). Lang3age and individual differences: The search for universals in
personalitylexicons. In Wheeler L (Ed.), Review ofpersonality andsocialpsycholqp
(Vol. 2, pp. 141-165). Beverly Hills, CA: Sage.
Goldberg LR. (1990). An alternative description of personality: The big five factor
structure. Journal of &rsonali@ and Social Pychology, 59, 1216-1229.
Goldberg LR. (1993). The structure of phenotypic personality traits. American Psycholo-
gkt, 4t$ 26-34.
Guilford JP, Fruchter B. (1973). Fwrdamental statistics in psychology and &tion (5th
ed.). New York McCiraw-Hill.
Harvey RJ. (1991). Job analysis. In Dunnette MD, Hough LM (Ed&), Handbook of
industrial and organizationalpsychology (2nd ed., Vol. 2, pp. 71-163). Pa10 Alto,
CA: ConsultingPsychologists Press.
Hofstee WKB, de Raad B, Goldberg LR. (1992). Integration of the Big Five and circum-
plex taxonomiesof traits. Journal of PerSoMlify and Social Pycholqp, 63, 146-163.
Hogan R. (1987). Personality psychology: Back to basics. In Aronoff J, Rabin AI, Zucker
RA (Eds.), The emergence ofpersonab (pp. 141-188). New York Springer.
Hough LM. (1992). The Big Five personality variables-construct confusion: Descrip
tion versus prediction. Human performnc~5, 139-155.
Jackson DN. (1990, August). @antifarive perspectives on the personality-jobpedormance
relationship. Presidential address to the Division of Evaluation, Measurement, and
Statisticsat the American Psychological Association Annual Convention, Boston.
Landy FJ. (1986). Stampcollecting versus science: Validation as hypothesistesting. Amer-
ican Psychologist, 41, 1183-1192.
McCormick El, Mecham RC, Jeanneret PR. (1977). Technical manual for the Posirion
Analysis Questionnaire (PAQ) (System 2). West Layfayette, I N University Book
Store.
McCrae RR, Costa PT Jr. (1985). Updating Normans adequate taxonomy: Intelligence
and personality dimensions in natural language and in questionnaires. Journal of
PerSoMlify and Social Pycholqp, 49,710-721.
McCrae RR. John OR (1992). An introduction to the five-factor model and its applica-
tions. Joumal of hona&v, 60,175-215.
Millon 7: (1994). Millon Index of firsonulily Styh: Manual San Antonio, Tx.The
Psychological Corporation.
Norman W. (1%3). lbward an adequate taxonomy of personality attributes. J o w ~ofl
Abnormal and Social Psychology, 66,574583.
US.Department of Labor. (1991). Dictionary of occupationaltitles (Vols.1-2). Mhing-
ton, D C Employment and Ttaining Administration, U.S. Department of Labor.
Wiggins JS. (1980). Circumplex models of interpersonal behavior. In Wheeler L (Ed.),
Review ofpersonal@ and socialpsychology (Vol. 1, pp. 265-294). Beverly Hills, C A
Sage.

Potrebbero piacerti anche