Journal of Applied Psychology
1997, Vol. 82, No. 4, 546-561
Copyright 1997 by the American Psychological Association, Inc.
Interviewers' Perceptions of Person-Organization Fit
and Organizational Selection Decisions
Daniel M. Cable
Georgia Institute of Technology
Timothy A. Judge
University of Iowa
A model of person-organization fit and organizational hiring decisions is developed and
tested, using data from 38 interviewers making hiring decisions about 93 applicants.
Results suggest that interviewers can assess applicant-organization values congruence
with significant levels of accuracy and that interviewers compare their perceptions of
applicants' values with their organizations' values to assess person-organization fit.
Results also suggested that interviewers' subjective person-organization fit assessments
have large effects on their hiring recommendations relative to competing applicant charac-
teristics, and that interviewers' hiring recommendations directly affect organizations'
hiring decisions (e.g., job offers).
Much of the recent interest in the concept of person-
organization (P-O) fit (see Kristof, 1996) can be traced
to Schneider's attraction-selection±attrition (ASA)
framework (Schneider, 1987), which suggests that people
and organizations are attracted to one another based on
their similarity. Although an emerging body of research
is supportive of the ASA framework, much of it remains
untested (Schneider, Goldstein, & Smith, 1995). For ex-
ample, although P-O fit is affected by posthire socializa-
tion practices (e.g., Chatman, 1991 ), little is known about
how organizations can establish P-O fit during the selec-
tion process (Bretz, Rynes, & Gerhart, 1993).
One selection device that may be critical in establishing
P-O fit is the employment interview (Chatman, 1991;
Judge & Ferris, 1992). The interview enables organiza-
tions and job applicants to interact through organizational
representatives, presumably allowing each party to deter-
mine if the other demonstrates congruent values and inter-
ests (Bowen, Ledford, & Nathan, 1991 ). Organizational
Daniel M. Cable, DuPree School of Management, Georgia
Institute of Technology; Timothy A. Judge, College of Business
Administration, University of Iowa.
A preliminary version of this article was presented in the
Human Resources Division of the 1995 Academy of Manage-
ment annual meetings in Vancouver, British Columbia, Canada.
We thank Chuck Parsons and Scott Shane for helpful comments
and Karin Ash, Ted Bartlett, and Kim Bloom for their help with
data collection and entry.
Correspondence concerning this article should be addressed
to Daniel M. Cable, who is now at the Kenan-Flagler Business
School, University of North Carolina, Campus Box 3490, Car-
roll Hall, Chapel Hill, North Carolina 27599-3490. Electronic
mail may be sent via Internet to firstname.lastname@example.org.
interviewers also readily declare the goal of locating and
hiring applicants who fit (e.g., Rynes & Gerhart, 1990).
Although the interview appears to be a means for orga-
nizations to select for P-O fit, much past research on
the topic has been practitioner oriented and exploratory
(Bowen et al., 1991; Judge & Ferris, 1992). However,
some research has examined the relationship between P-O
fit and organizational hiring decisions. Rynes and Gerhart
(1990) found that P-O fit is a separate construct from
general employability and that interviewers evaluate P-O
fit according to their organizations' attributes, not just
their personal preferences. These researchers also found
that applicants' interpersonal attributes (e.g., leadership
and warmth) were related to interviewers' fit assessments.
Whereas Rynes and Gerhart (1990) suggested that the
employment interview may be a means to assess and hire
for P-O fit, two more recent studies have explored differ-
ent aspects of fit in the selection process. Bretz et al.
(1993) reported that interviewers most often mentioned
job-related courses, experience, and general applicant
characteristics (e.g., attractiveness) when responding to
open-ended questions about their subjective fit evalua-
tions. Adldns, Russell, and Werbel (1994) examined work
values congruence between applicants and organizations
and found that values congruence did not affect interview-
ers' subjective P-O fit perceptions and that interviewers'
P-O fit perceptions had little effect on organizations' se-
The summary of past research just presented demon-
strates that it is difficult to compare results from the initial
studies in an incipient area of research. Although these
past investigations appear to conflict in some ways, note
that fundamentally different questions were inherent in the
methodologies. For example, Rynes and Gerhart (1990)
P-O FIT AND SELECTION DECISIONS 547
examined perceived congruence (similarity between an
interviewer's perceptions of an applicant's and their orga-
nization' s attributes ), whereas Adkins et al. (1994) exam-
ined actual congruence (similarity between an applicant' s
attributes and an organization's attributes as indepen-
dently reported by each party). The present study devel-
ops past research on interviewers' P-O fit perceptions by
examining both actual and perceived congruence and by
examining congruence relative to variables suggested
by Bretz et al. (1993) and other researchers (e.g., inter-
viewer liking, applicant physical attractiveness, and work
Past research on P-O fit and the interview also may
appear to conflict because different attributes were used
to represent the fit construct (Judge & Ferris, 1992). Al-
though interviewers' perceptions of P-O fit may be based
on many applicant attributes (e.g., demographics or per-
sonality), one important aspect of both individuals and
organizations that can be compared directly and meaning-
fully is values (Chatman, 1989, 1991; Schein, 1990).
Values are enduring beliefs that a specific mode of conduct
or end state is preferable to its opposite, therefore guiding
individuals' attitudes, judgments, and behaviors (Chat-
man, 1989, 1991; Rokeach, 1973). Thus, the present study
attempts to unify fit research by defining P-O fit as the
congruence between individuals' and organizations' val-
ues (Adkins et al., 1994; Chatman, 1989, 1991; Kristoff,
1996; O'Reilly, Chatman, & Caldwell, 1991).
Figure 1 depicts the conceptual model of the role of P-
O fit in organizations' hiring decisions. Each link in the
model is discussed in turn.
Applicant-Organization Values Congruence
Implicit in the belief that employment interviews enable
organizations to establish P-O fit is the assumption that
interviewers can assess applicants' values with some de-
gree of validity. However, decades of research have indi-
cated that interviewers are not adept at assessing appli-
cants' personal characteristics (se6 Arvey & Campion,
1982). In fact, interviewers' prevalent beliefs that they
can accurately assess personality characteristics has been
called the interview illusion and has been interpreted as
evidence of the fundamental attribution error (Gilovich,
1991 ). Thus, past research indicates that interviewers may
base their P-O fit judgments on inaccurate perceptions of
applicants, suggesting that actual values congruence and
perceived values congruence may be unrelated.
Although interviewers' abilities to assess applicants' val-
ues has not been the focus of most interview research
(Paunonen, Jackson, & Oberman, 1987), some research
supports interviewers' claims that they can assess appli-
cants' values accurately. Integrating results from a number
of laboratory studies, Jackson and his colleagues suggested
that judges can "interpret verbal, self-referent statements
made in an interview context in terms of personality charac-
teristics and relate these to expectations of suitability and
performance" (Jackson, Peacock, & Holden, 1982, p. 1).
However; this research was conducted under experimental
conditions in which personality information about "paper
applicants" was manipulated in interview transcripts. Sup-
port for interviewers' abilities to assess applicants' values
also is provided by the consensus-at-zero-acquaintance re-
search paradigm (e.g., Albright, Kenny, & MaUoy, 1988).
This research literature has demonstrated above-chance
agreement between self-reported personality traits and rat-
ings by observers who have had minimal exposure with
the targets (Albright et al., 1988; Gangestad, Simpson,
DiGeronimo, & Biek, 1992). Furthermore, Borkenau and
Liebler (1993) found that strangers' personality assess-
ments not only were related to targets' self-reports but
also were correlated with ratings given by targets' close
On the basis of interpretations of the interview as an
opportunity to assess applicants' values (Chatman,
1991), the rigorous consensus-at-zero-acquaintance re-
search (Gangestad et al., 1992), and interviewers' own
claims of their abilities (Rynes & Gerhart, 1990), we
expected actual values congruence to positively predict
perceived values congruence:
Hypothesis 1: Actual values congruence positively affects
perceived values congruence.
Interviewers' Subjective Person-Organization Fit
The previous section concerns values congruence, re-
ferring to a comparison between an organization's culture
and an applicant's values. In addition to comparisons of
specific values, it generally is believed that interviewers
make subjective P-Ofit evaluations, or holistic judgments
about applicants' fit with their organizations (Rynes &
Gerhart, 1990). Conceptually, values congruence and sub-
jective P-O fit perceptions are related but distinct con-
structs (Adkins et al., 1994). In fact, Schneider's (1987)
ASA framework indicates that P-O fit evaluations should
be based on the congruence between an organization's
values and an applicant's values.
Although theory suggests that values congruence
should affect P-O fit evaluations, some past interview re-
search has not supported this prediction. For example, in
the most recent study of values congruence and interview-
ers' P-O fit evaluations, Adkins et al. (1994) found that
applicant-organization values congruence had little effect
on interviewers' P-O fit perceptions. However, these re-
CABLE AND JUDGE
J Interviewer Person- L
Attractiveness of Applicant
Experience Applicant GPA
~Org~it~on ~ '
I Applicant Hiring Applicant Race
Figure 1. Hypothesized model of person-organization fit in organizational selection decisions
(GPA = grade point average).
searchers examined only actual value s congruence. Other
researchers have emphasized the importance of social and
personal constructions of organizations, where individu-
als' attitudes and behaviors are based on their perceptions
rather than a more objective standard (e.g., Ferris & Judge,
1991; Salancik & Pfeffer, 1978; Schneider & Reichers,
1983). For example, Pulakos and Wexley (1983) found
that actual similarity between managers and subordinates
was less predictive of performance evaluations than per-
ceived similarity. Extending this research to the context
of the interview, interviewers' perceptions of applicants'
values congruence should have a more proximal influence
on their subjective fit evaluations than actual values con-
gruence. Presumably, interviewers evaluate applicants on
the basis of information that is available to them, even if
this information is inaccurate. Thus, to extend past re-
search and provide a new test of Schneider' s (1987) ASA
framework, we hypothesized the following:
Hypothesis 2: Perceived values congruence positively af-
fects interviewers' subjective assessments of P-O fit.
Interviewer Person-Organization Fit Assessments
and Hiring Decisions
One of the most established findings in social psychol-
ogy is that individuals are more attracted to others per-
ceived as similar to themselves than those viewed as dis-
similar (e.g., Byrne, 1969). Theoretically, individuals are
more attracted to others who fit because of reduced cogni-
tive dissonance, improved communication, and increased
predictability in social interactions (e.g., Festinger, 1954;
Swann, 1984; Tsui & O'Reilly, 1989). In the context of
organizational selection, interviewers should prefer job
seekers who appear to fit their organization (Bowen et
al., 1991; Schneider, 1987). Although theory suggests that
interviewers' subjective P-O fit assessments should affect
their selection decisions, little empirical research has ex-
amined the role of fit perceptions in organizational hiring
decisions. Accordingly, we next consider the relation-
ships among interviewers' subjective P-O fit perceptions,
their hiring recommendations, and organizations' hiring
Interviewer recommendation to hire. Almost all orga-
nizations use the interview when selecting new employ-
ees, and it is widely believed that interviewers' evaluations
of applicants are crucial determinants of organizations'
hiring decisions (e.g., Dipboye, Smith, & Howell, 1994;
McDaniel, Whetzel, Schmidt, & Maurer, 1994). Accord-
ingly, it is important to examine the impact of P-O fit
perceptions on interviewers' hiring recommendations. Al-
though past research has demonstrated that P-O fit is a
P-O FIT AND SELECTION DECISIONS
separate construct from general employability (e.g., Ad-
kins et al., 1994; Rynes & Gerhart, 1990), it has not
been established that P-O fit affects interviewers' hiring
recommendations, especially relative to other applicant
attributes. Thus, on the basis of past theory and research,
we hypothesized the following:
Hypothesis3: Interviewers' subjective evaluations of
applicants' P-O fit positively affect their hiring
terview is a means to establish P-O fit in organizations,
interviewers' evaluations of applicants must affect organi-
zations' hiring decisions. Both anecdotal reports and sur-
vey research indicate that the interview is a critical com-
ponent of organizations' selection strategies (Dipboye,
1992). In fact, Dipboye et al. (1994) reported that the
majority of those responsible for final hiring decisions
considered the interviewer's impression of the applicant
to be the most important factor in their decision. Although
much past interview research has overlooked the final
outcomes of applicant-organization fit, it is important to
examine the impact of interviewers' recommendations on
organizational hiring decisions, to understand the pro-
cesses through which P-O fit operates (e.g., Adkins et al.,
1994). On the basis of this discussion, we hypothesized
If the selection in-
Hypothesis 4: Interviewers' hiring recommendations posi-
tively affect organizations' hiring decisions.
Other Relevant Influences
Past interview research has revealed many applicant
characteristics that affect interviewers' hiring recommen-
dations (Arvey & Campion, 1982). Thus, as indicated in
Figure 1, relevant variables were added to the model in
addition to the hypothesized variables, to better estimate
the role of P-O fit in the interview. First, it generally
is acknowledged that applicants' physical attractiveness
affects interviewers' hiring recommendations (e.g., Ar-
vey & Campion, 1982; Dipboye, 1992; Gilmore, Beehr, &
Love, 1986; Raza & Carpenter, 1987); therefore, this link
was included. Research also has indicated that interview-
ers' personal liking of applicants affects their hiring rec-
ommendations (Campion, 1978; Keenan, 1977; Raza &
Carpenter, 1987), so this relationship was examined in
the present study. Not surprisingly, research indicates that
interviewers' evaluations are affected by applicants' hu-
man capital (e.g., grade point average (GPA) and work
experience; Campion, 1978; Raza & Carpenter, 1987;
Singer & Bruhns, 1991); therefore, these relationships
were included in the model as well. Finally, we estimated
the relationships between applicants' demographics (sex
and race) and interviewers' hiring recommendations, in
keeping with the findings of past research (Gilmore,
Beehr & Love, 1986; Hitt & Barr, 1989; McDonald &
Hakel, 1985; Raza & Carpenter, 1987).
With respect to the predictors of interviewers' P-O fit
evaluations, our a priori expectations were motivated by
past research that examined P-O fit in the interview con-
text. Accordingly, we examined the effects of applicant
attractiveness on P-O fit evaluations, as suggested by
Rynes and Gerhart (1990) and Bretz et al. (1993). We
also estimated the effect of liking on P-O fit evaluations
because Rynes and Gerhart (1990) suggested that inter-
viewers' personal preferences or liking affect their fit per-
ceptions. Finally, because both Bretz et al. (1993)and
Adkins et al. (1994) suggested that applicants' human
capital were related to P-O fit evaluations, these relation-
ships also were included in the hypothesized model.
Overview of Data Collection
To circumvent potential confounds of nonexpert judges (e.g.,
Barr & Hitt, 1986), this study examined interviewers making
authentic applicant assessments and hiring recommendations.
Also, to mitigate the effects of common-method variance, data
were collected in three stages from multiple sources, which are
Time 1. In the spring and fall recruiting sessions of 1994,
64 organizations recruited for positions through the career office
in the industrial relations school of a large northeastern univer-
sity. With the support of the career office, recruiters were con-
tacted and asked to participate in the study before they arrived
on campus. Confidentiality of recruiters' responses was assured.
Forty-two recruiters from 35 organizations (55% response rate)
completed surveys about their organizations (e.g., culture).
Of responding recruiters, 5% held staff positions, 57% were
in management positions, 18% were vice presidents, and 20%
held positions of director or partner. Of the recruiters, 55% were
men and 45% were women, and 91% were Caucasian. The
average interviewer had been employed with his or her company
for 8.1 years and had been interviewing for 7 years. On average,
recruiters interviewed 11 applicants at this university and were
recruiting for seven positions across multiple universities. Of
the recruiters, 38% were alumni of the university investigated
in this study. Recruiters represented organizations averaging
46,845 employees and gross revenues of $41,795,880. Available
information on nonrespondents was gathered, both in terms of
recruiters (gender, university alumni status, position level in
the organization, and number of applicants interviewed) and
organizations (net income, number of employees, and earnings
per share). No significant differences existed between respon-
dents and nourespondents, at least in terms of these variables.
Interviewers were asked to complete applicant-evaluation sur-
veys as soon as possible after their interviews. In keeping with
the methodology used by Bretz et al. (1993) and Adkins et al.
(1994), interviewers were asked to evaluate at least one success-
ful applicant and one unsuccessful applicant, but to complete
surveys about as many applicants as time permitted. This
550 CABLE AND JUDGE
method was used because interviewers seldom reported about
negative applicants unless they were asked directly, limiting the
variability of their responses. Interviewers rated a total of 112
applicants; the mean number of applicants rated was 3.15, rang-
ing from 1 to 6. Post hoc analyses revealed that the number
of applicants rated by a recruiter shared no relationship with
judgments made about those applicants (e.g., liking, willingness
to hire, or P-O fit evaluation).
After completing the recruiting cycle, the job appli-
cants who had interviewed through the career office were asked
to complete a survey that assessed personal characteristics (e.g.,
values and demographics) and their success in obtaining second
interviews and job offers from each organization with which
they had interviewed during that cycle. Although approximately
3 months separated Time 1 (campus interviews) and Time 2
(applicants reported job offers), some job seekers indicated that
they were still waiting to hear from organizations regarding
their job prospects. We subsequently contacted those job seekers
before graduation to complete our data collection. Confidential-
ity of individuals' responses was assured, and participation was
voluntary. Of the applicants who were rated by interviewers,
91% completed surveys.
Applicants' ages ranged from 19 to 45 years, with an average
age of 23 years. Of the respondents, 55% were women and
45% were men, and 72% were Caucasian. Degree-related work
experience ranged from 0 to 19 years, with an average of 1.6
years. Of the applicants, 58% were undergraduates, and appli-
cants' GPAs ranged from 2.3 to 4.0, with a mean of 3.42. Avail-
able data on nonrespondents (degree, gender, GPA, and work
experience) were collected from applicants' r6surn6s, which
were on file for recruiters in the career office. No significant
differences were found on these variables between respondents
and nonrespondents. Respondents' r6sumgs also were used to
confirm the GPAs and years of work experience that respondents
had reported on the survey. The average difference between
respondents' r6sum6s and their direct reports were 0.01 and 3.8
months for GPAs and work experience, respectively (r = .95
and .96, respectively; both ps < .01), indicating that applicants
provided us with the same information that was provided to
Six months after completing surveys about the ap-
plicants they had interviewed, interviewers again were contacted
and asked to complete a short survey that reassessed their orga-
nizations' values. Thirty-eight (90%) of the interviewers who
responded to the initial surveys also completed this final survey,
and usable data were available for 93 applicant-interviewer
ments for interactional research (Bern & Allen, 1974; Chatman,
1989), the assessment of individual and organizational values
should be idiographic, so that the relevance of particular values
and the uniqueness of patterns of values across people and orga-
nizations are represented and commensurate, so that people and
organizations can be compared. Also, because many competing
work values are socially desirable (e.g., few individuals would
care to characterize themselves as lacking fairness or tolerance
According to the conceptual require-
until they are forced to make a choice between them), forced
ranking (ipsative measurement) is the most appropriate method
of values assessment (Chatman, 1989; Meglino, Ravlin, & Ad-
Of the available measures of work values, the Organizational
Culture Profile (OCP; O'Reilly et al., 1991 ) appeared to be best
suited for the present study. The OCP was expressly developed
and validated to assess P-O fit (O'Reilly et al., 1991) and was
recommended specifically by Rynes and Gerhart (1990) to un-
derstand how organizations use the interview as a means to
establish P-O fit. As recommended by O'Reilly et al. (1991),
interviewers reported their perceptions of their organizations'
values by sorting values into nine categories ranging from most
characteristic of my organization to least characteristic of my
organization. Interviewers reported their organizational cultures
both at Time 1 and at Time 3.
In the present study, the number of items in the original OCP
was reduced from 54 to 40. A pilot study with organizational
recruiters suggested that several of the items were too similar
for the task of describing applicants (e.g., "flexibility" and
"adaptability") and that 54 items simply took too long to com-
pare and sort. To reduce the number of items, 10 organizational
researchers were given the OCP and were asked to make the 54
values more manageable by grouping similar values together
but retaining each value that was truly unique. Each respondent
removed at least 15 items, and only the values that all respon-
dents agreed were highly similar were removed. The 40 values
used in the reduced OCP appear in the Appendix.
Because interviewers reported their organizations' values at
two different times (approximately 6 months apart), it was
possible to assess the consistency of their perceptions. Ac-
cording to the approach recommended and used by Chatman
( 1991 ) and O'Reilly et al. ( 1991 ), the stability of interviewers'
perceptions of their organizations was assessed by calculating
test-retest reliabilities. For the 38 interviewers who responded
to both culture assessments, the mean reliability was .61 (all
ps < .01 ). To assess the reliability of the most defining values
of organizations, a second test-retest reliability was computed,
using only those values rated as 9, 8, or 7 (very characteristic)
and 1, 2, or 3 (very uncharacteristic). The mean test-retest
reliability was .87. Not surprisingly, values seen as defining an
organization were more stable than those "neither characteristic
nor uncharacteristic" of an organization. Overall, these results
indicate significant stability in recruiters' perceptions of their
The reduced OCP also was used to as-
sess interviewers' perceptions of applicants' values and appli-
cants' perceptions of their own values. After their interviews
(at Time 1), recruiters sorted the values into nine categories
ranging from most characteristic to least characteristic, ac-
cording to the question, "To what degree is this a characteristic
of the applicant I interviewed?" At Time 2, applicants per-
formed the same Q sort according to the question "How charac-
teristic is this attribute of me?"
At Time 1, interviewers reported their
subjective assessment of applicants' P-O fit, according to the
question, "To what degree did this applicant match or fit your
organization and the current employees in your organization,"
evaluated on a 5-point graphic scale ranging from not at all
P-O FIT AND SELECTION DECISIONS 551
(1) to completely (5). Although one-item measures are not
inherently deficient, their reliability is questionable. Accordingly,
we assessed the reliability of the one-item subjective P-O fit
scale during the 1994 fall recruiting session by adding a second
item ("Do you think this applicant's values reflect your own
organization's values and 'personality?' "). Thus, the fall 1994
data collection (approximately 50% of the total database) en-
abled us to investigate the reliability of the original P-O fit item.
The coefficient alpha reliability estimate resulting from the two-
item P-O fit scale was .83, indicating that the one-item P-O fit
measure provided an adequate measure of interviewers' P-O fit
judgments, which therefore is used in all subsequent analyses.
Recommendation to hire.
The variable representing inter-
viewers' hiring recommendation comprised three items. Be-
cause interviewers decide whether applicants receive further
consideration for a job or are eliminated from the selection
process (Adkins et al., 1994), second interview invitations are
a direct behavioral measure of interviewers' hiring intentions.
Second interview invitations were measured during data collec-
tion Time 2, when applicants reported which organizations of-
fered them second interviews (coded 1 = yes, 0 = no). Re-
sponses to this item were combined with interviewers' reported
likelihood that they would recommend that applicants be hired
(assessed at Time 1 and ranging from very unlikely ( 1 ) to very
likely (5)) and with interviewers' responses to the statement
"Please give your overall evaluation of this candidate" (as-
sessed at Time 1 and ranging from very negative  to very
positive ). When subjected to a factor analysis, these three
items resulted in a single factor that explained 87.1% of the
variance, and the internal consistency of this three-item scale
was .93. Due to differences in scale formats between the items,
responses were standardized before the scores were computed.
Organizations' hiring decisions were
measured during Time 2, when applicants reported which orga-
nizations had offered them jobs. A dummy variable was coded
as 1 if the organization had extended an offer to the applicant
and as 0 otherwise. As noted above, if applicants indicated that
they still were waiting to hear from an organization, we con-
meted them again before graduation and recorded whether they
had received a job offer.
At Time 1, interviewers rated the
physical attractiveness of applicants, according to the question,
"Please rate the overall level of attractiveness of this applicant
(appearance, dress, etc.)," on a 5-point rating scale ranging
from very unattractive (1) to very attractive (5).
At Time 1, interviewers rated their per-
sonal liking of each applicant, according to the statement,
"Please estimate how well you personally liked this applicant."
Responses were to a 5-point scale ranging from very little (1)
to very well (5).
Profile similarity scores.
The hypotheses to be tested in
the present study required the calculation and use of values
congruence scores (e.g., Hypothesis 1 and Hypothesis 2). Con-
sistent with theoretical conceptualizations of P-O fit (e.g., Chat-
man, 1989; Kristof, 1996) and as recommended and used in
past research (Bem& Allen, 1974; O' Reilly et al,, 1991 ), values
congruence scores were calculated by correlating two values
profiles assessed with the OCP. To reduce potential common-
method variance concerns, these congruence scores were com-
puted with data collected from different sources and times. Spe-
cificaUy, the scores representing perceived values congruence
were based on interviewers' reports of applicants (assessed dur-
ing Time 1 data collection) and interviewers' reports of their
organizations (assessed during Time 3). The scores representing
actual values congruence were based on interviewers' reports
of their organizations (assessed during Time 3) and applicants'
reports of their own values (assessed during Time 2).
The means, standard deviations, and correlations
among all variables appear in Table 1. Because the same
group of applicants were interviewing for multiple posi-
tions, some applicants were rated by multiple interview-
ers. Although most of the variables under investigation
were reported independently about each applicant (after
each individual interview), some variables were dupli-
cated across cases (e.g., applicant demographics). Dupli-
cation of variables across cases can lead to correlated
errors that violate statistical assumptions. To investigate
the hypothesis that duplicated applicant characteristics led
to a positive correlation between error terms, we com-
puted the Durban-Watson statistic, a test designed for
exactly this purpose (Greene, 1993). In the present study,
the Durbin-Watson statistic (d) confirmed the null hy-
pothesis of no autocorrelation (d = 1.86, ns), indicating
that the disturbances were not significantly correlated (r
= .07, ns).
Because interviewers provided data for more than one
case, it also was possible that interpretation of the results
was confounded by systematic differences across inter-
viewers in their decision making (e.g., general interviewer
tendencies to evaluate all applicants positively or nega-
tively). To test the prevalence of interviewer effects, we
conducted an analysis of variance on the endogenous
study variables (perceived values congruence, P-O fit
evaluations, recommendation to hire, applicant physical
attractiveness, and interviewer liking), using interviewer
as the grouping variable. Each of the 38 interviewers in
the study was assigned a unique number to distinguish
them. None of the F ratios were significant for any of
the variables, indicating that there were no significant
differences across interviewers in terms of their mean
ratings. Although this analysis does not exclude the possi-
bility of idiosyncratic interviewer effects, it does suggest
that variation in interviewer ratings is due to the applicant
rather than interviewer effects alone.
To test the model presented in Figure 1, we estimated a
covariance structure model using LISREL 8 (JSreskog &
S6rbom, 1993), which allows researchers to model entire
systems of relationships (e.g., sequential dependent vari-
ables). Also, LISREL permits direct comparisons of alter-
native models, providing information about the relative
adequacy of the hypothesized model. One concern when
Correlations Between Study Variables
CABLE AND JUDGE
1 2 3 4 5 6 7 8 9 10 11
1. Actual values congruence
2. Perceived values congruence
3. Interviewer P-O fit evaluation
4. Applicant physical attractiveness
5. Interviewer liking of applicant
6. Interviewer hiring recommendation
7. Applicant sex
8. Applicant race
9. Applicant work experience
10. Applicant GPA
11. Organizational hiring decision
Note. N = 93. Correlations greater than .19 are significant at the .05 level. P-O fit = person-organization fit; GPA = grade point average.
using covariance structure analyses is sample size. Bentler
(1985) suggested that a sample-size-to-parameter ratio of
5:1 is adequate to achieve reliable estimates. Because that
ratio was 7:1 in the present study, we considered the sam-
ple size adequate for the analyses. A second potential
concern with the analyses was the dichotomous nature of
job offer decisions. Because LISREL assumes that endog-
enous variables are continuous, modeling organizations'
selection decisions with covariance structure analysis
would violate this assumption. To examine the effects of
this violated assumption, we predicted hiring decisions
using logistic regression, an approach designed specifi-
cally to model dichotomous dependent variables (Greene,
1993). Results from this reduced-form equation were
nearly identical to the LISREL output, indicating that the
dichotomous nature of the dependent variable had little
effect on the LISREL analysis and results.
Before discussing support for the specific hypotheses,
it is important to evaluate the overall fit of the theoretical
model to the data. Fit statistics are the central means
through which alternative factor structures are compared,
and there are numerous statistics that can be used to de-
scribe a model's fit to the data. Table 2 contains the fit
statistics for the hypothesized model, as well as several
alternative models, which will be discussed later. The most
widely used measures of fit are chi-square and the good-
ness-of-fit index (GFI). Chi-square statistics that are not
statistically significant suggest that a model adequately
fits the data, and GFIs at or above .90 are believed to
indicate acceptable fit (Medsker, Williams, & Holahan,
1994). Thus, according to these indices, the hypothesized
model provides a good fit to the data. 1 However, because
chi-square and GFI depend on sample size (Hu & Bentler,
1995), researchers have suggested alternative fit statistics
that depend less on the sample size and are thought to
provide better information on model fit (Marsh, Balla, &
McDonald, 1988). Three of these statistics provided by
LISREL 8 are the normed fit index (NFI), the compara-
tive fit index (CFI), and the incremental fit index (IFI).
As with GFI, levels at or above .90 for these statistics
imply adequate fit, and each of these fit indices indicate
that the hypothesized model provides a good fit to the
data. Next, we discuss the substantive results from the
i According to the LISREL program (see JOreskog & S~r-
bom, 1993, p. 28), degrees of freedom are provided by the
df = (no. variables) (no. variables + 1)/2
- no. freed parameters.
The hypothesized model contains 11 variables and 14 freed
parameters (as indicated in the parameter specifications matrix
provided in the LISREL output). According to this formula, the
proper degrees of freedom for the hypothesized model should
be 52. However, the LISREL output indicated that the degrees
of freedom were 24. Apparently, this discrepancy was due to
a miscalculation in the LISREL program using the FIXED-X
command. Because it often is unnecessary to specify a measure-
ment model for exogenous variables, LISREL allows a FIXED-
X command to be used, which fixes the covariances among the
exogenous variables (contained in the Phi matrix) to be equal
to their values in the sample covariance matrix (see J6reskog &
SSrbom, 1993, p. 9). The apparent error was that when calculat-
ing degrees of freedom; LISREL treated these fixed parameters
as freed. In the present study, because there are 7 exogenous
variables in the hypothesized model, the Phi matrix contains 7
variances and 21 covariances, thus, 28 parameters. If one sub-
tracts these 28 parameters from the proper 52 degrees of free-
dom, one arrives at the 24 degrees of freedom provided in the
LISREL output. However, because the elements in Phi are not
freed parameters, we believe this to be erroneous and therefore
will use the proper 52 degrees of freedom for the hypothesized
model. (We thank an anonymous reviewer for bringing this
discrepancy to our attention.) One way to resolve this apparent
error is to directly fix elements in Phi to be equal to the elements
in the covariance matrix. In fact, when we did this, 52 degrees
of freedom were obtained, and the model parameters were the
P-O FIT AND SELECTION DECISIONS
Fit Statistics for Alternative Models
Model X 2
df GFI NFI CFI IF]
Alternative Model 1
Hiring recommendation predicts P-O fit perception.
Alternative Model 2
P-O fit perception predicts perceived values
Alternative Model 3
Human capital predicts values congruence.
Alternative Model 4
Demographics predict values congruence.
Alternative Model 5
Interviewer liking and applicant attractiveness
predict perceived values congruence.
52 .94 .90 .96 .97
52 .77 .67 .69 .71
52 .93 .88 .94 .94
50 .94 .91 .96 .97
50 .95 .92 .98 .98
50 .96 .94 1.00 1.00
Note. N = 93. GFI = goodness-of-fit index; NFI = normed fit index; CFI = comparative fit index; IFI
= incremental fit index; P-O fit = person-organization fit.
*p < .05.
Figure 2 contains the maximum-likelihood parameter
estimates, significance levels, and coefficients of determi-
nation for the hypothesized model. Actual values congru-
ence significantly predicted perceived values congruence.
Thus, Hypothesis 1 was supported. Perceived values con-
gruence significantly influenced interviewers' subjective
P-O fit evaluations, providing support for Hypothesis 2.
Interviewers' liking of applicants and applicants' physical
I, ~'~-.53) ,~
/ lnterviewerPerson- /
I ~ ~** 22**/
ApplicantAttractivenessPhysical IntervleWerof ApplicantLiking
VJ 'i mm tion"
K** Organizational Hiring [
Xn..',,iewe, L "V °'"
Experience r~7,, I (g~.78) J .04 I
• ~.01 I
~ ApplicantSex Applicant Race
organization fit in organizational selection decisions (GPA = grade point average). Statistics
are standardized path coefficients; *p < .05, two-tailed; * *p < .01, two-tailed. R 2 = coefficient
of determination for each endogenous link.
Maximum-likelihood parameter estimates for the hypothesized model of person-
554 CABLE AND JUDGE
Total Effects of Variables and Hiring Recommendations and Hiring Decisions
hiring decisions Variable hiring recommendations
Actual values congruence
Perceived values congruence
Applicant physical at~activeness
Interviewer liking of applicant
Subjective P-O fit evaluation
Applicant work experience
Note. Total effects are standardized. P-O fit = person-organization fit; GPA = grade point average.
*p <.05. **p <.01.
attractiveness also positively and significantly influenced
interviewers' perceptions of applicants' P-O fit. Inter-
viewers' assessments of applicants' P-O fit significantly
influenced their hiring recommendations, providing sup-
port for Hypothesis 3. Hypothesis 4, that interviewers'
hiring recommendations would positively affect organiza-
tions' final selection decisions, also was supported. In
terms of the control variables, applicants with higher
GPAs were judged to fit significantly better than appli-
cants with lower GPAs. Applicants who were personally
liked by interviewers were more likely to be recommended
for hire than less-liked applicants, as were applicants with
more work experience. Finally, female applicants were
significantly more likely to be recommended for hire than
Because the ultimate outcomes in this study are inter-
viewers' hiring recommendations and organizations' hir-
ing decisions, it is informative to examine the relative total
effects of each exogenous variable on these outcomes.
Accordingly, Table 3 contains the standardized total ef-
fects of each antecedent variable provided by the LISREL
software. Furthermore, to illustrate the practical results of
interviewers' P-O fit perceptions, we have presented in
Table 4 the percentage effect sizes of P-O fit on actual
job offers. The results in Table 3 also reveal that perceived
values congruence, interviewer liking of the applicant, and
subjective fit perceptions have at least moderately strong
effects on interviewing outcomes. Although actual con-
gruence has a statistically significant total effect on hiring
recommendations and hiring decisions, its effect is rela-
Both Medsker et al. (1994) and Hayduk (1987) recom-
mended evaluating a hypothesized model relative to plau-
sible alternative models. In the present study, there appear
to be five alternative models that should be tested. First,
it is possible that interviewers' P-O fit perceptions are not
antecedents to, but consequences of, their hiring recom-
mendations, so that interviewers rate applicants as having
a good fit with their organization if they already have
decided to recommended that they be hired. We tested
this alternative model by reversing the direction of the
link between interviewers' recommendation to hire and
interviewers' P-O fit evaluation in Figure 2, leaving
all other aspects of the original hypothesized model
A second possibility is that recruiters' reports of appli-
cants' values are a result rather than an antecedent of their
P-O fit perceptions. Thus, it is possible that interviewers
Relationship Between Interviewer Judgments of Applicant Person-Organization Fit
and Organizational Selection Decisions
"To what degree did this applicant match
or fit your organization and the current
employees in your organization?"
No. applicants falling
% applicants in category
receiving job offers
Not at all
To almost no degree
To some degree
To a large degree
Note. Interviewer judgments of applicant fit were made directly after their interview with applicants (Time
1); information on whether a job offer was received was reported by applicants after organizational selection
decision (Time 2).
P-O FIT AND SELECTION DECISIONS 555
manufactured applicants' values profiles to look like their
organization if they believed they fit, broadly speaking.
We tested this alternative model by reversing the direction
of the link between interviewers' P-O fit evaluations and
perceived values congruence in Figure 2, again leaving
all other aspects of the hypothesized model unchanged.
Third, it is possible that applicants' human capital attri-
butes affect interviewers' perceptions of their values, re-
suiting in higher perceived values congruence for appli-
cants' with certain qualifications. This argument is consis-
tent with Spence's (1973)signaling theory, which
suggests that employers make inferences about applicants'
tacit qualities based on their overt characteristics. We
tested this third alternative model by adding links between
applicants' human capital (GPA and work experience)
and perceived values congruence, again leaving all other
aspects of the hypothesized model unchanged.
Similarly, it also appears plausible that applicants' de-
mographic characteristics might affect interviewers' per-
ceptions of their values, so that perceived values congru-
ence might be higher for some groups than others. Al-
though no past research
relationships, Heilman's (1983) lack-of-fit model would
suggest that interviewers' perceptions of applicants' val-
ues congruence may be affected by the applicants' demo-
graphics due to stereotypes (also see Taylor, Fiske, Et-
coff, & Ruderman, 1978). We tested this fourth alternative
model by adding links between applicants' demographics
(sex and race) and perceived values congruence.
Finally, it appears plausible that interviewers' liking
of applicants and applicants' physical attractiveness may
affect perceived values congruence. First, in keeping with
past research indicating that people's cognitions often are
affected by their affective reactions to stimuli (e.g., Za-
jonc, 1980), interviewers' initial liking of applicants may
cause them to alter their judgments of applicants' values.
Furthermore, research has indicated that interviewers' ini-
tial impressions of applicants influence subsequent judg-
ments about them (Macan & Dipboye, 1988 ). Thus, appli-
cants' physical attractiveness also may affect interviewers'
perceptions of their values, because both attractiveness
and likability are applicant attributes that can be identified
early in the interview. In the present study, we tested this
fifth alternative model by freeing the links between liking
and perceived values congruence and between attrac-
tiveness and perceived values congruence.
Table 2 contains fit statistics from each of the alternative
models. The first alternative model could be rejected im-
mediately, because it provided a poor fit to the data. Spe-
cifically, the chi-square was significant, and each of the
fit indices fell below the .90 criterion suggested by Med-
sker et al. (1994). The hypothesized model also provided
a better fit than the second alternative model, across all
of the fit indices. Specifically, the second alternative model
has investigated these
was an average of .02 lower across the four standardized
fit statistics in Table 2, and researchers have suggested
that alternative LISREL models may be judged by whether
they differ by at least .01 across standardized fit statistics
(Widaman, 1985). The third alternative model did not
improve on the fit of the hypothesized model. Because
this model was less parsimonious than the hypothesized
model, it would need to fit significantly better to be pre-
ferred. Furthermore, neither of the new links (GPA and
work experience) of this alternative model was significant
(both p > .10).
The fourth and fifth alternative models both appeared to
demonstrate a better fit with the data than the hypothesized
model. Results from the fourth model indicated that sex
(but not race) significantly predicted perceived values
congruence, so that female applicants were significantly
more likely to be perceived as having the same values
as organizations. Results from the fifth alternative model
indicated that interviewers' liking (but not applicants'
physical attractiveness) significantly predicted'perceived
values congruence, so that applicants who were liked by
interviewers were significantly more likely to be perceived
as having congruent values. Although these relationships
in the alternative models were significant, they were not
added to the hypothesized model in the present study
because it generally is not appropriate to modify a hypoth-
esized model in the midst of testing it (MacCallum, Roz-
nowski, & Necowitz, 1992).
Researchers have proposed that a critical function of
the employment interview is the assessment of applicants'
values congruence with recruiting organizations (Bowen
et al., 1991; Chatman, 1991; Rynes & Gerhart, 1990).
Experimental research has indicated that judges can assess
individuals' personal characteristics and organizational
suitability (Jackson et al., 1982), and conversations with
actual interviewers reveal that the interview is believed to
be a means of assessing applicants' values congruence
with organizations. However, little field research has ex-
amined the determinants of interviewers' fit judgments,
and no research has demonstrated that interviewers' P-O
fit perceptions affect either theft hiring recommendations
or organizations' hiring decisions. Given the perva-
siveness of the interview in selection systems (Arvey &
Campion, 1982; McDaniel et al., 1994) and the positive
outcomes of P-O fit (Chatman, 1991; Govindarajan,
1989), the lack of answers to these basic questions is a
substantial gap in the literature.
This study suggests that interviewers base their P-O fit
evaluations on the congruence between their perceptions
of applicants' values and their organizations' values. Fur-
thermore, as indicated in Table 3, perceived values congru-
556 CABLE AND JUDGE
ence had significant effects on interviewers' hiring recom-
mendations and organizations' hiring decisions. Thus,
work values appear to be an important element of the
interviewing process. Results also indicated that inter-
viewers' subjective P-O fit perceptions were important
predictors of their hiring recommendations, even after
controlling for the relevant variables suggested by past
interview research (e.g., attractiveness and demograph-
ics). Furthermore, other organizational members placed
considerable weight on interviewers' assessments of ap-
plicants, because job offers were based in large part on
interviewers' evaluations (see Tables 3 and 4). In fact,
when we used methods of calculating percentage effect
sizes in logistic regression (Hanusheck & Jackson, 1977 ),
the LOGIT results indicated that a one-unit increase in
fit from the mean would result in a 44% increase in the
probability of receiving a job offer. Thus, an applicant
who was perceived to fit the organization fairly well (be-
ing rated as a 4 on the 1-5 scale) would be 44% more
likely to receive a job offer than the average fitting appli-
cant, even after controlling for interviewer liking, appli-
cant attractiveness, work experience, GPA, sex, and race.
This cumulative set of findings is consistent with past
practitioner-oriented research on the topic (e.g., Bowen
et al., 1991 ) and also confirms the selection component
of Schneider's (1987) ASA model.
Results from this study perhaps are made more interest-
ing by the finding that the relationship between actual
values congruence and perceived values congruence, al-
though significant, was relatively small (fl = .25, p <
.01). This finding implies that interviewers' inferences
about applicants' values and P-O fit often are inaccurate,
at least from applicants' perspectives. These results are
consistent with past performance appraisal research indi-
caring that perceived similarity is a more effective pre-
dictor of subordinate evaluations than actual similarity
(Pulakos & Wexley, 1983 ). Results from the present study
also replicate Adkins et al.'s (1994) recent finding that
actual values congruence had little direct effect on inter-
viewers' P-O fit perceptions. However, the present findings
also help clarify and integrate conflicting results from past
research in the following way: Values congruence affects
interviewers' P-O fit perceptions, but apparently it is per-
ceived congruence rather than actual congruence that best
predicts fit judgments. The relatively weak effect of actual
congruence on selection decisions is likely due to the fact
that it is a relatively distal influence: In the hypothesized
model, actual congruence is mediated by three other vari-
ables (perceived congruence, subjective fit perceptions,
and recommendation to hire) before it influences actual
Although the present study integrated several discrep-
ancies between past research investigations, note the im-
portant methodological differences that remain. Specifi-
cally, although we examined the effect of applicants' GPA
and work experience, Bretz et al. (1993) found that re-
cruiters also were interested in applicants' articulateness
and job-related course work. Thus, it is possible that a
more fully specified model would have affected our re-
suits, perhaps restraining the relative effects of the P-O
fit variables. Furthermore, because Bretz et al. interviewed
recruiters from four different colleges, whereas the present
study examined data from only one college, it is possible
that differences in results can be attributed to different
samples, so that the relevance of values congruence and
P-O fit depends on the occupational field. Another meth-
odological difference is that this study used the OCP,
whereas Adkins et al. (1994) used the Comparative Em-
phasis Scale. Although both of these scales measure work
values ipsatively, the CES examines 4 values, but the OCP
used in this study examined 40. Finally, it is important to
recognize the effects that prescreening decisions may have
on the importance of P-O fit relative to other variables
(e.g., human capital). Because the percentage of pre-
screening permitted in different interviewing situations
may vary and because prescreening decisions may affect
the variance of applicants' work experience, course work,
GPA, and demographics (Macan & Dipboye, 1988), pre-
screening may affect the impact of P-O fit variables rela-
tive to other applicant characteristics. In summary, al-
though it is difficult for a single study to examine all
relevant variables, it is important for P-O fit research to
continue to integrate past research and determine the ef-
fects of methodological differences.
Finally, results from the comparison of the hypothe-
sized model with the alternative models merit some dis-
cussion. The hypothesized model was motivated by past
interview and P-O fit research, and the LISREL fit statis-
tics indicated that this system of relationships provided a
good fit to the data. Of the five alternative models tested
(including eight different links), only two new links ap-
peared to significantly improve the fit of the hypothesized
model. Although this set of findings increases confidence
in the validity and robustness of the hypothesized model,
the results also suggest that future research would benefit
from examining the relationships between perceived val-
ues congruence, applicant sex, and interviewer liking.
Specifically, results indicated that applicants' sex may af-
fect interviewers' perceptions of their values due to stereo-
typing, consistent with Heilman's (1983) model. Like-
wise, the significant relationship between interviewer lik-
ing and perceived values congruence may highlight the
construed nature of perceived values congruence, helping
to explain the relatively small effect of actual values con-
gruence. Future research is needed to cross-validate the
findings of these alternative models (MacCallum et al.,
P-O FIT AND SELECTION DECISIONS 557
Limitations and Strengths
This study has several limitations. First, interviewers
reported applicants' values at the same time that they
reported how well applicants fit their organization. This
data collection methodology allows for a reverse interpre-
tation of Hypothesis 2 (that interviewers first form a fit
perception, then use this perception to complete the OCP
so that applicants' values reflect their fit perception).
However, several factors weigh in favor of the causal di-
rection contained in the hypothesized model. First, organi-
zational theory suggests that values affect P-O fit (Chat-
man, 1989, 1991; Schneider, 1987), not the reverse causa-
tion. Second, a direct test of the alternative model
described above indicated a considerably worse fit with
the data than the model derived from past theory and
research (see Table 2). Finally, if perceived values con-
gruence was simply a function of interviewers' P-O fit
judgments, then perceived values congruence would not
have been related to actual values congruence because
actual values congruence was reported independently of
perceived values congruence. If perceived values congru-
ence is simply a manifestation of P-O fit judgments, actual
values congruence and perceived values congruence will
be unrelated. However, actual values congruence was sig-
nificantly and positively related to perceived values con-
gruence, consistent with P-O fit theory.
A second potential weakness of this study concerns the
assessment of P-O fit with profile similarity indices
(PSIs). Although PSIs are consistent with theoretical con-
ceptualizations of P-O fit and therefore have been recom-
mended and used by P-O fit researchers (e.g., Chatman,
1989, 1991; Rynes & Gerhart, 1990), they also have been
criticized. As noted in Kristof's (1996) review, PSIs ob-
fuscate the contribution of individual elements to the over-
all score, discard information regarding the direction and
magnitude of the differences between two entities, and
may make unjustified assumptions about the congruence
relationship. Although polynomial regression may cir-
cumvent these problems, several aspects of the present
study made PSIs practicable and even necessary. First,
as discussed above, it is necessary to use an ipsative,
idiographic measurement approach to assess values
(Chatman, 1989; Meglino et al., 1989), which precludes
polynomial regression. Also, the polynomial regression
approach requires that each variable, interaction variable,
and higher order term be entered into the regression equa-
tion simultaneously ( 159 additional variables in the pres-
ent study), which is obviously not feasible given the sam-
ple size (Schneider et al., 1995). Finally, as noted by
Kristof (1996), the statistics emanating from polynomial
regression may not accurately represent the P-O fit con-
struct. Fortunately, the problems with profile similarity
assessments imply that these methods provide conserva-
tive estimates of true relationships, not that the methods
provide upwardly biased (inflated) estimates. Thus, al-
though PSIs have their limitations, the consequences of
not using them were judged to be even more restrictive.
Other limitations of this study resulted from the brevity
of our surveys, which reflected interviewers' busy sched-
ules. For example, it would have been advantageous to
have multiple-item scales for each measure (e.g., inter-
viewer liking and applicant attractiveness), because mea-
surement error may have constrained several relationships
in the model. Similarly, this study examined only values
congruence when predicting interviewers' subjective P-O
fit evaluations. Although the relationship between values
congruence and P-O fit has been well established (Chat-
man, 1989), past research also has suggested that other
aspects of applicants and organizations may contribute to
P-O fit perceptions. For example, it would be interesting
for future research to examine applicant-organization
goal congruence or the congruence between applicants'
personality and organizations' human resource systems,
when predicting interviewers'
The potential weaknesses of this investigation are ac-
companied by a number of strengths. The determinants
of P-O fit perceptions and the effects of those perceptions
on selection decisions were modeled based on a careful
review of past theory and research concerning both P-O
fit and the interview. The conceptual model included both
actual and perceived values congruence, distinct con-
structs which are directly relevant to P-O fit theoretically
but have not been examined simultaneously to date. Fi-
nally, the inclusion of competing variables reduced the
possibility that the results were affected by omitted vari-
A second strength of this study was that interviewers
rated applicants immediately after they interviewed for ac-
tual jobs, thus avoiding potential confounds due to nonex-
pert judges. In addition to extending the generalizability of
the results, the examination of actual interviewers reduces
concerns about survey priming because it appears unlikely
that interviewers would alter their actual evaluations based
on an optional, confidential survey. This argument is rein-
forced by significant direct and indirect effects of inter-
viewers' evaluations on organizations' job offer decisions.
Finally, this study examined data collected from differ-
ent sources and times, and interviewer-supplied data were
complemented with applicants' reports of their second
interview and job offers, demographics, values, and hu-
man capital. Furthermore, several variables reported by
applicants were confirmed against their r6sum6s. Because
expected, significant relationships were found between
variables reported by both applicants and interviewers,
concerns about common-method variance were reduced.
Threats of priming and common-method variance also
subjective P-O fit
558 CABLE AND JUDGE
were mitigated in the context of assessing interviewers'
fit perceptions because the calculation of values congru-
ence was based on data collected 6 months apart.
Implications and Future Research
Surprisingly, P-O fit has not been considered in the
context of the employment interview in any of nine com-
prehensive literature reviews of interview research (Har-
ris, 1989). Results from our study appear particularly
interesting in this respect because interviewers' P-O fit
perceptions were the best predictors of hiring recommen-
dations relative to the determinants discussed in those
same literature reviews. These findings suggest that future
research on the employment interview should continue
to examine P-O fit perceptions and should investigate
interviewers' perceptions of applicants' characteristics be-
cause presumably this is what they and their organizations
rely on for hiring decisions. It also is important for future
research to determine the sources of interviewers' percep-
tions of applicants' values. A limitation of this study is
that we did not examine applicants' performance in the
interview (e.g., dress, oral communication skills, and non-
verbal behavior), and therefore, we were unable to exam-
ine the effects of interview performance on interviewers'
P-O fit judgments. It would be interesting to reveal which
applicant attributes or actions are related to interviewers'
perceptions of their values.
Although the relationship between actual and perceived
values congruence was significant, the effect size was
relatively small. Furthermore, tests of plausible alternative
models suggested that perceived values congruence is re-
lated to applicants' sex and interviewers' liking, relation-
ships that have not been examined in past interview
research and that were not expected in the hypothesized
model. These results suggest that actual and perceived
applicant-organization congruence are not always
aligned, and this has serious implications for theories of
person-environment fit as they currently are interpreted.
From the perspectives of person-environment fit (Per-
vin, 1968) and similarity-attraction research (Byrne,
1969), actual congruence should affect the relationships
between organizations and individuals regardless of
whether it is perceived explicitly (e.g., through improved
communication). If it is not congruence per se but per-
ceived congruence affecting the composition of organiza-
tions, a new interpretation of P-O fit must allow for condi-
tions under which congruence and perceived congruence
disconnect (e.g., Ferris & Judge, 1991). Much past re-
search has investigated the fit of relatively concrete attri-
butes, such as demographics (Tsui & O'Reilly, 1989).
It is possible that when examining relatively subjective
characteristics such as values, perceived and actual fit may
become unaligned and that perceived congruence is more
predictive of decisions and outcomes than actual congru-
ence (Pulakos & Wexley, 1983). This perspective is en-
tirely consistent with theories of social memory (e.g.,
Wyer & Carlston, 1994), which suggest that judges' cog-
nitive representations of people and events, and not the
original stimuli, govern subsequent judgments and behav-
iors. As noted by Wyer and Carlston (1994), "Conse-
quently, it is important to understand the nature of these
mediating cognitive representations to predict the influ-
ence of information on perceivers' judgments or behav-
ioral decisions" (p. 42).
The construed-reality approach of P-O fit has important
implications for recruiting organizations. If interviewers
rely on their fit perceptions when making hiring recom-
mendations and these perceptions are based primarily on
misinterpreted values, the function of the interview as a
means to assess and establish values congruence is called
into question. Data from the present study corroborate
experimental findings that interviewers can evaluate appli-
cants' personal characteristics and make P-O fit judgments
with significant degrees of accuracy. However, it remains
for future research to resolve whether interviewers' level
of accuracy in values assessment has positive utility after
factoring in the financial costs and social benefits of the
Finally, results from this study have implications for
how recruiters structure their interviews. Research has
indicated that the validity of an interview increases when
it is structured around a job analysis and is consistent
across applicants (Campion, Pursell, & Brown, 1988; Mc-
Daniel et al., 1994). However, job-based structured inter-
views appear to be somewhat incompatible with assessing
applicants' values and P-O fit, because these criteria ex-
tend well beyond immediate job-related factors (Rynes &
Gerhart, 1990). In fact, subjective fit impressions typi-
cally are what structured interviews remove from inter-
viewer decision making (Campion et al., 1988). Accord-
ingly, it is important to determine whether the interview
is most valuable as a verbal, job-based ability test or as
an opportunity to assess P-O fit. Because theoretical and
empirical research suggest that organizations may max-
imize performance when P-O fit is established (e.g., Bar-
ney, 1991; Govindarajan, 1989), the interview may have
its greatest utility as a P-O fit assessment device (Bowen
et al., 1991; Chatman, 1991). Furthermore, as noted by
Judge and Ferris (1992), it may be possible to improve
interviewers' P-O fit judgments by structuring interviews
around organizational cultures (rather than specific jobs)
and by assessing applicants' personal characteristics that
are relevant to the fit criterion.
Adkins, C. L., Russell, C. J., & Werbel, J. D. (1994). Judgments
of fit in the selection process: The role of work value congru-
ence. Personnel Psychology, 47, 605-623.
P-O FIT AND SELECTION DECISIONS
Albright, L., Kenny, D. A., & Malloy, T. E. (1988). Consensus
in personality judgments at zero acquaintance. Journal of
Personality and Social Psychology, 55, 387-395.
Arvey, R. D., & Campion, J. E. (1982). The employment inter-
view: A summary and review of recent research. Personnel
Psychology, 35, 281-322.
Barney, J. (1991). Firm resources and sustained competitive
advantage. Journal of Management, 17, 99-120.
Barr, S. H., & Hitt, M. A. (1986). A comparison of selection
decision models in manager versus student samples. Person-
nel Psychology, 39, 599-617.
Bern, D.J., & Allen, A. (1974). On predicting some of the
people some of the time. Psychological Review, 81, 506-520.
Bentler, P. M. (1985). Theory and implementation of EQS: A
structural equations program. Los Angeles: BMDP Statistical
Borkenau, P., & Liebler, A. (1993). Convergence of stranger
ratings of personality and intelligence with self-ratings, part-
ner ratings, and measured intelligence. Journal of Personality
and Social Psychology, 65, 546-553.
Bowen, D. E., Ledford, G. E., & Nathan, B. R. (1991). Hiring
for the organization, not the job. Academy of Management
Executive, 5, 35-51.
Bretz, R.D., Rynes, S.L., & Gerhart, B. (1993). Recruiter
perceptions of applicant fit: Implications for individual career
preparation and job search behavior. Journal of Vocational
Behavior, 43, 310-327.
Byrue, D. (1969). Attitudes and attraction. In L. Berkowitz
(Ed.), Advances in experimental social psychology (Vol. 4,
pp. 35-89). New York: Academic Press.
Campion, M. A. (1978). Identification of variables most influ-
ential in determining interviewers' evaluations of applicants
in a college placement center. Psychological Reports, 42,
Campion, M. A., Pursell, E. D., & Brown, B. K. (1988). Struc-
tured interviewing: Raising the psychometric properties of
the employment interview. Personnel Psychology, 41, 25-42.
Chatman, J. (1989). Improving interactional organizational re-
search: A model of person-organization fit. Academy of Man-
agement Review, 14, 333-349.
Chatman, J. ( 1991 ). Matching people and organizations: Selec-
tion and socialization in public accounting firms. Administra-
tive Science Quarterly, 36, 459-484.
Dipboye, R. L. (1992). Selection interviews: Process perspec-
tives. Cincinnati, OH: College South-Western.
Dipboye, R. L., Smith, C. S., & Howell, W. C. (1994). Under-
standing industrial and organizational psychology: An inte-
grated approach. Fort Worth, TX: Harcourt Brace.
Ferris, G. R., & Judge, T. A. ( 1991 ). Personnel/human resource
management: A political influence perspective. Journal of
Management, 17, 447-488.
Festinger, L. (1954). A theory of social comparison processes.
Human Relations, 7, 117-140.
Funder, D. C., & Colvin, C. R. (1988). Friends and strangers:
Acquantainceship, agreement, and the accuracy of personality
judgment. Journal of Personality and Social Psychology, 55,
Gangestad, S.W., Simpson, J.A., DiGeronimo, K., & Biek,
B. (1992). Differential accuracy in person perception across
traits: Examination of a functional hypothesis. Journal of
Personality and Social Psychology, 62, 688-698.
Gilmore, D. C., Beehr, T. A, & Love, K. G. (1986). Effects of
applicant sex, applicant physical attractiveness, type of rater,
and type of job on interview decisions. Journal of Occupa-
tional Psychology, 59, 103-109.
Gilovich, T. ( 1991 ). How we know what isn't so: The fallibility
of human reason in everyday life. New York: Free Press.
Govindarajan, V. (1989). Implementing competitive strategies
at the business unit level: Implications of matching managers
to strategies. Strategic Management Journal, 27, 25-41.
Greene, W.H. (1993). Econometric analysis. New York:
Hanusheck, E. A., & Jackson, J. E. (1977). Statistical methods
for social scientists. Orlando, FL: Academic Press.
Harris, M.M. (1989). Reconsidering the employment inter-
view: A review of recent literature and suggestions for future
research. Personnel Psychology, 42, 691-726.
Hayduk, L. A. (1987). Structural equation modeling with LIS-
REL: Essentials and advances. Baltimore: Johns Hopkins
Heilman, M. E. (1983). Sex bias in work settings: The lack of
fit model. In L. L. Cummings & B. M. Staw (Eds.), Research
in organizational behavior (Vol. 5, pp. 269-298). Green-
wich, C'12. JAI Press.
Hitt, M, A., & Barr, S. H. (1989). Managerial selection decision
models: Examination of configural cue processing. Journal
of Applied Psychology, 74, 53-61.
Hu, L., & Bentler, P. M. (1995). Evaluating model fit. In R. H.
Hoyle (Ed.), Structural equation modeling: Concepts, issues,
and applications (pp. 76-99). Thousand Oaks, CA: Sage.
Jackson, D. N., Peacock, A. C., & Holden, R. R. (1982). Profes-
sional interviewers' trait inferential structures for diverse oc-
cupational groups. Organizational Behavior and Human Per-
formance, 29, 1-20.
J6reskog, K. G., & S6rbom, D. (1993). LISREL 8 user's refer-
ence guide. Chicago: Scientific Software.
Judge, T. A., & Ferris, G. R. (1992). The elusive criterion of
fit in human resources staffing decisions. Human Resource
Planning, 15, 47-67.
Keenan, A. (1977). Some relationships between interviewers'
personal feelings about candidates and their general evaluation
of them. Journal of Occupational Psychology, 50, 275-283.
Kristof, A. L. (1996). Person-organization fit: An integrative
review of its conceptualizations, measurement, and implica-
tions. Personnel Psychology, 49, 1-49.
Macan, T. H., & Dipboye, R. L. (1988). The effects of inter-
viewers' initial impressions on information gathering. Orga-
nizational Behavior and Human Decision Processes, 42,
MacCallum, R. C., Roznowski, M., & Necowitz, L. B. (1992).
Model modification in covariance structure analysis: The
problem of capitalization on chance. Psychological Bulletin,
Marsh, H. W., Balla, J. R., & McDonald, R. P. (1988). Good-
ness-of-fit indexes in confirmatory factor analysis: The effect
of sample size. Psychological Bulletin, 103, 391-410.
McDaniel, M.A., Whetzel, D.L., Schmidt, E L., & Maurer,
S. D. (1994). The validity of employment interviews: A tom-
560 CABLE AND JUDGE
prehensive review and meta-analysis. Journal of Applied Psy-
chology, 79, 599-616.
McDonald, T., & Hakel, M.D. (1985). Effects of applicant
race, sex, suitability, and answers on interviewer's questioning
strategy and ratings. Personnel Psychology, 38, 321-334.
Medsker, G. J., Williams, L. J., & Holahan, P. J. (1994). A re-
view of current practices for evaluating causal models in orga-
nizational behavior and human resource management re-
search. Journal of Management, 20, 439-464.
Meglino, B. M., Ravlin, E. C., & Adkins, C. L. (1989). A work
values approach to corporate culture: A field test of the value
congruence process and its relationship to individual out-
comes. Journal of Applied Psychology, 74, 424-432.
O'Reilly, C. A., Chatman, J., & Caldwell, D. E ( 1991 ). People
and organizational culture: A profile comparison approach to
assessing person-organization fit. Academy of Management
Journal, 34, 487-516.
Pannonen, S.V., Jackson, D.N., & Oberman, S.M. (1987).
Personnel selection decisions: Effects of applicant personality
and the letter of reference. Organizational Behavior and Hu-
man Decision Processes, 40, 96-114.
Pervin, L. W. (1968). Performance and satisfaction as a function
of individual-environment fit. Psychological Bulletin, 69,
Pulakos, E.D., & Wexley, K.N. (1983). The relationship
among perceptual similarity, sex, and performance ratings in
manager-subordinate dyads. Academy of Management Jour-
nal 26, 129-139.
Raza, S.M., & Carpenter, B.N. (1987). A model of hiring
decisions in real employment interviews. Journal of Applied
Psychology, 72, 596-603.
Rokeach, M. (1973). The nature of human values. New York:
Rynes, S.L., & Gerhart, B. (1990). Interviewer assessments
of applicant "fit": An exploratory investigation. Personnel
Psychology, 43, 13-35.
Salancik, G. R., & Pfeffer, J. (1978). A social information pro-
cessing approach to job attitudes and task design. Administra-
tive Science Quarterly, 23, 224-252.
Schein, E.H. (1990). Organizational culture. American Psy-
chologist, 45, 109-119.
Schneider, B. (1987). The people make the place. Personnel
Psychology, 40, 437-453.
Schneider, B., Goldstein, H.W., & Smith, D. B. (1995). The
ASA framework: An update. Personnel Psychology, 48, 747-
Schneider, B., & Reichers, A. E. (1983). On the etiology of
climates. Personnel Psychology, 36, 19-39.
Singer, M. S., & Bruhns, C. ( 1991 ). Relative effect of applicant
work experience and academic qualification on selection in-
terview decisions: A study of between-sample generalizabil-
ity. Journal of Applied Psychology, 76, 550-559.
Spence, M. A. (1973). Job market signalling. Quarterly Journal
of Economics, 87, 355-374.
Swarm, W. B., Jr. (1984). Quest for accuracy in person percep-
tion: A matter of pragmatics. Psychological Review, 91, 457-
Taylor, S.E., Fiske, S.T., Etcoff, N.L., & Ruderman, A.J.
(1978). Categorical and contextual bases of person memory
and stereotyping. Journal of Personality and Social Psychol-
ogy, 36, 778-793.
Tsui, A. S., & O'Reilly, C. A. (1989). Beyond simple demo-
grap~c effects: The importance of relational demography in
superior subordinate dyads. Academy of Management Jour-
nal, 32, 402-423.
Widaman, K. E (1985). Hierarchically nested covariance struc-
ture models for multitralt-multimethod data. Applied Psycho-
logical Measurement, 9, 1-26.
Wyer, R. S., & Carlston, D. E. (1994). The cognitive representa-
tion of persons and events. In R. S. Wyer & T. Srull (Eds.),
Handbook of social cognition (pp. 41-98). Hillsdale, NJ:
Zajonc, R.B. (1980). Feeling and thinking: Preferences need
no inferences. American Psychologist, 35, 151-175.