People and Organizational Culture

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 22

PEOPLE AND ORGANIZATIONAL CULTURE: PEOPLE AND

ORGANIZATIONAL CULTURE: A PROFILE


COMPARISON APPROACH TO A PROFILE
COMPARISON APPROACH TO ASSESSING
PERSON-ORGANIZATION FIT CHARLES A. O'REILLY III
University of California, Berkeley JENNIFER CHATMAN Northwestern University
DAVID F. CALDWELL Santa Clara University This article brings together three current
themes in organizational be- havior: (1) a renewed interest in assessing person-situation
interac- tional constructs, (2) the quantitative assessment of organizational cul- ture, and (3)
the application of "Q-sort," or template-matching, ap- proaches to assessing person-situation
interactions. Using longitudinal data from accountants and M.B.A. students and cross-
sectional data from employees of government agencies and public accounting firms, we
developed and validated an instrument for assessing person- organization fit, the
Organizational Culture Profile (OCP). Results sug- gest that the dimensionality of individual
preferences for organiza- tional cultures and the existence of these cultures are interpretable.
Further, person-organization fit predicts job satisfaction and organiza- tional commitment a
year after fit was measured and actual turnover after two years. This evidence attests to the
importance of understand- ing the fit between individuals' preferences and organizational
cul- tures. The notion of organizational culture has been important in the study of
organizational behavior for the past decade (e.g., Barley, Meyer, & Gash,
1988; O'Reilly, 1989; Smircich, 1983). In spite of disagreements over some
elements of definition and measurement, researchers seem to agree
that cul- ture may be an important factor in determining how well an
individual fits an organizational context (e.g., Kilmann, Saxton, & Serpa, 1986;
Schein, 1985). Implicit in writing on this theme is a logic of person-culture fit fun-
damentally drawn from an interactional psychology perspective in which
aspects of both individual and situation combine to influence a focal indi-
vidual's response to a given situation (e.g., Chatman, 1989; Schneider, 1987;
Terborg, 1981). In this regard, aspects of individuals, such as values and The authors
gratefully acknowledge the research support of the Institute of Industrial Re- lations and the
Institute for Personality Assessment and Research at the University of California, Berkeley.
We also thank John Sheridan and Blair Sheppard for providing data access. 487488
Academy of Management Journal September expectations, interact with facets of
situations, such as incentive systems and norms, to affect the individuals' attitudinal
and behavioral responses. As with similar fit theories of careers (Holland, 1985), job
choice (Hackman & Oldham, 1980), work adjustment (Lofquist & Dawis, 1969), and
organiza- & Oldham, 1980), work adjustment (Lofquist & Dawis, 1969), and
organiza- tional climate (Joyce & Slocum, 1984), the validation of the construct of
tional climate (Joyce & Slocum, 1984), the validation of the construct of person-
culture fit rests on the ability to assess relevant aspects of both person and culture.
This measurement problem is a significant and sometimes con- troversial issue
(Keon, Latack, & Wanous, 1982; Rousseau, 1990)-one that is at the center of the
person-situation debate, that is, the controversy over the degree to which personality
or context variables explain attitudes and behavior (Bem & Allen, 1974; Davis-Blake
& Pfeffer, 1989; Kenrick & Funder, 1988). The purpose of this research was to
examine person-culture fit and its implications for work attitudes and behavior. We
draw on recent develop- ments in the applications of Q-sort, or template-matching,1
approaches to resolve some of the measurement issues that have hindered previous
re- search on fit (Bem & Funder, 1978; Block, Block, & Morrison, 1981; Caldwell &
O'Reilly, 1990). THEORETICAL BACKGROUND Person-Situation Fit The general
notion of fit, or congruence, has long been important in psychology and
organizational behavior (Nadler & Tushman, 1980). In studying person-situation fit,
organizational behavior researchers have typ- ically taken one of two broad paths.
One has led to exploration of the inter- action of individual characteristics and broad
occupational attributes, the other to exploration of the fit between specific
characteristics of an organi- zation and the people in it. Examples of the second
approach range from studying the match of individual skills to job requirements to
studying the relationship between individual characteristics and organizational
climate (e.g., Downey, Hellriegel, & Slocum, 1975). For example, the two major theo-
ries of vocational choice (Holland, 1985; Super, 1957) both postulate that an
individual will select a career or occupation that is similar to or that fits with that
person's self-concept. Empirical results have typically supported the hypothesis that
congruence between individuals' personalities and the de- mands of their
occupations are associated with positive affect (Mount & Muchinsky, 1978; Spokane,
1985) and a high likelihood of their staying in their jobs (Meir & Hasson, 1982). A
similar logic characterizes a series of studies of work adjustment conducted
by Lofquist and Dawis (1969). For ' Q-methodology (Stephenson, 1953) is a well-
established assessment technique. In a typ- ical Q-sort procedure, the individual is
presented with a large number of statements or char- acteristics and asked to sort the items
into categories (normally nine) according to some crite- rion (usually the extent to which the
item is characteristic of the individual). Constraints are typically imposed to force
respondents to place fewer items in the extreme categories and more items in the middle
categories.1991 O'Reilly, Chatman, and Caldwell 489 instance, they proposed that
satisfaction results from "a harmonious rela- tionship between the individual and his
environment, suitability of the in- dividual to the environment and vice versa" (1969:
45). Tom (1971) recast this notion of person-situation complementarity to focus on
person- organization fit. He studied the role of personality and organizational images
in the recruiting process and found that the greater the similarity between an in the
recruiting process and found that the greater the similarity between an individual's
self-concept and his or her image of an organization, the more individual's self-
concept and his or her image of an organization, the more that individual preferred
that organization. In a similar vein, Keon and col- leagues (1982) found that
individuals with a positive self-image sought to enter graduate schools with positive
organizational images. Other studies have reported generally consistent findings
(e.g., Graham, 1976). More recent studies of early career adjustment and person-job
fit have also invoked notions of congruence. For example, in a programmatic effort,
Wanous and his colleagues showed how an accurate understanding of
job requirements can enhance a person's adjustment to a job (Wanous, 1977).
Similarly, the congruence between a person and a job have embodied no- tions of fit
(e.g., O'Reilly, 1977): the degree to which individuals are suited to a job depends on
their motives and needs and the job's requirements (Hackman & Oldham, 1980). The
cited studies and more general studies of person-situation interac- tions (e.g., Diener,
Larsen, & Emmons, 1984; Pervin, 1968) rest on the prem- ise that positive
responses will occur when individuals fit or match the requirements of a situation.
Although broadly used and intuitively compel- ling, the person-situation framework
has spawned a number of disagree- ments. For example, a recent issue of the
Academy of Management Review (1989, vol. 14, no. 3) contained articles dealing
with those controversies. One important question raised is how fit should be defined
(Chatman, 1989). The definition of fit remains a critical and largely unanswered
question (Rynes & Gerhart, 1990). For instance, most studies of person-situation fit
in organizations have used normative measures of personality to assess indi- vidual
characteristics and relatively broad classifications of tasks, occupa- tions, or jobs to
characterize situations. Thus, people are described with one language, or set of
characteristics, and situations with a totally different language. This failure to
describe people and situations along commensurate dimensions limits scholars'
ability to develop a coherent theory of person- situation interactions (Graham,
1976; Pervin, 1968; Springfield, 1988) and makes it difficult to determine the real
impact of person-situation effects (Terborg, 1981). Further, the use of a very
limited set of descriptions of person and situation may make the detection of
any true interaction effect difficult. Davis-Blake and Pfeffer, for example,
observed that similar jobs in different firms may vary greatly and
concluded that the "measurement of job characteristics requires moving
beyond crude occupational surrogates to measures which actually
reflect the characteristics of a particular job as it is structured in a particular
organizational setting" (1989: 394). A similar crit- icism can be made of the extant
treatments of personality, in which most researchers have relied on a few normative
measures that may not only fail490 Academy of Management Journal September to
describe individuals adequately but may also assess personality charac- teristics not
relevant to the people or situations under study (O'Reilly, Cald- well, & Mirabile, 1990;
Weiss & Adler, 1984). Thus, although suggestive, previous research has generally
failed to describe people and situations in a comprehensive manner along
commensurate and relevant dimensions. This failure has hindered the development
and empirical assessment of coherent theories of person-situation interaction.
Recent work in interactional psychology has begun to identify the char- Recent work
in interactional psychology has begun to identify the char- acteristics of effective
techniques for addressing person-situation effects. acteristics of effective techniques
for addressing person-situation effects. Bem and Funder (1978) argued that, in
addition to providing comprehensive measurements, effective techniques for
assessing persons and situations should allow for holistic comparisons across
multiple dimensions. Such an approach can be thought of as "semi-idiographic" in
that it is idiographic (i.e., compares the relative strength of attributes within a single
individual) with respect to individual attributes but permits comparisons of person-
situations (Luthans & Davis, 1982; Springfield, 1988). Since any given trait
dimension will not be applicable to all individuals, we want to be able to select only
those personological variables that are pertinent to a focal indi- vidual. Doing so
requires an idiographic approach rather than a nomothetic one in which all
individuals are rated in terms of a given attribute (Lamiell, 1981). The difficulty,
however, with an idiographic approach is that it isn't clear what to do once a rating
has been made. What is then needed is to be able to compare individuals even
though descriptors may be differentially relevant to them. Using "Q-methodology"
(Stephenson, 1953), Bem and Allen (1974) de- veloped a template-matching
technique to accommodate this dual concern with relevance and comparability. This
approach focuses on the salience and configuration of variables within a person
rather than on the relative standing of persons across each variable. Since not all
characteristics apply to all people and since what differentiates people from each
other is the set of traits salient to each individual, an assessment of
person-situation fit must permit such idiographic measurement of each person
while also allowing comparisons across situations. Such an approach requires a
large number of items or descriptors that comprehensively describe individuals and
are rel- evant to particular situations. For instance, Bem and Funder
(1978) created a 100-item profile of the ideal person for successful performance in
an array of specific situations. How well individuals might do in a situation
was predicted by how well they matched the ideal person-in-situation profile. Thus,
rather than comparing a person and situation on a few dimensions, an appropriate
person-situation investigation would attempt to determine the overall fit of the
person to the set of relevant situational attributes. Drawing on the Q-sort
technique used for template matching, Caldwell and O'Reilly (1990) and O'Reilly and
colleagues (1990) developed a profile- matching process to assess person-job fit.
Using a structured interview job analysis with job incumbents and experts, they first
developed a compre- hensive set of competencies required for successful job
performance. This1991 O'Reilly, Chatman, and Caldwell 491 set (typically 60-90 items)
was then used to construct a consensus profile of the job. Individual profiles were
then obtained by using peers and superiors as assessors. Person-job fit was
measured by correlating the two profiles. Results of a series of studies have shown
that person-job fit predicts perfor- mance, satisfaction, and turnover across a variety
of jobs. Like template matching (Bem & Funder, 1978), the profile comparison
process comprehen- sively assesses individuals and situations using a common
language, allows for the ipsative measurement of individual characteristics by
arraying at- tributes in terms of their salience to the individual, and provides a direct
measure of person-situation fit. The profile comparison process goes
beyond measure of person-situation fit. The profile comparison process
goes beyond template matching by using items that are highly specific to a target
situation template matching by using items that are highly specific to a target
situation and equally relevant to a person and a situation. Thus, the application of a
Q-sort technique appears to be a useful way to obtain semi-idiographic as-
sessments of fit and offers a way to resolve a number of the measurement problems
that have characterized earlier studies of person-situation interac- tion. Person-
Culture Fit Although a number of earlier studies have explored the general notion of
person-organization fit (e.g., Graham, 1976; Joyce & Slocum, 1984; Tom, 1971),
more recent interest has centered on the idea that organizations have cultures that
are more or less attractive to certain types of individuals (e.g., Wilkins & Ouchi,
1983). As Barley and colleagues (1988) noted, the concept of
organizational culture has a long history, dating to early sociological studies (e.g.,
Gouldner, 1954; Selznick, 1949). Only recently, however, has the term
"organizational culture" become prominent. Drawing on theories from
anthropology, sociology, and social psy- chology, researchers have made a
number of efforts to understand the be- havior of individuals and groups in
organizations using cultural concepts such as semiotics, rituals, ceremonies,
stories, and language (e.g., Ouchi & Wilkins, 1985; Smircich, 1983;
Swidler, 1986; Trice & Beyer, 1984). This process has generated a series of
debates over issues such as the definition of "culture," the appropriate
methodology for investigating it, and the proper level of analysis for its study.
Barley (1983) pointed out that all studies of culture, whatever their theoretical origin,
use reasonably similar terms and constructs. Differences exist among
researchers in how objective or subjec- tive, conscious or unconscious their
use of these terms and constructs is and in what they see as appropriate elements to
study. Typically, researchers have agreed that culture can be thought of as a set of
cognitions shared by members of a social unit (e.g., Geertz, 1973; Smircich, 1983).
Rousseau (1990) provided an excellent description of the common
elements in such sets and suggested a framework including
fundamental assumptions, values, behavioral norms and expectations, and
larger patterns of behavior. Re- search on culture usually begins with a set of
values and assumptions (Enz, 1988; Martin & Siehl, 1983; Schein, 1985;
Weiner, 1988). These values, whether conscious or unconscious, typically act as the
defining elements492 Academy of Management Journal September around which norms,
symbols, rituals, and other cultural activities revolve. Thus, Parsons argued that a
cultural tradition emerges around values, de- fined as elements "of a shared
symbolic system which serves as a criterion or standard for selection among the
alternatives of orientation which are intrinsically open in a situation" (1951: 11-12).
Rokeach offered a very similar definition, proposing that "a value is an enduring
belief that a spe- cific mode of conduct or end-state of existence is personally or
socially preferable to an opposite or converse mode of conduct or end-state of exis-
tence" (1973: 5). In this vein, basic values may be thought of as internalized
normative beliefs that can guide behavior. When a social unit's members share
values, they may form the basis for social expectations or norms. Should these be
even more widely shared throughout a larger social group- Should these be even
more widely shared throughout a larger social group- ing, an organizational culture
or value system may exist. Thus, researchers ing, an organizational culture or value
system may exist. Thus, researchers who investigate culture by focusing on norms
(e.g., Cooke & Rousseau, 1988; Harrison & Carroll, 1991; O'Reilly, 1989) are
studying social expectations that are based on underlying values. Others who study
culture through rit- uals, stories, or myths (e.g., Louis, 1983; Martin & Siehl, 1983;
Trice & Beyer, 1984) are examining phenotypic outcroppings that reflect underlying
beliefs and values. The pervasiveness and importance of values in organizational
culture are fundamentally linked to the psychological process of identity formation in
which individuals appear to seek a social identity that provides meaning and
connectedness (Ashforth & Mael, 1989). A substantial body of research has shown
that individuals tend to classify themselves into social categories, such as gender,
race, ethnicity, and organizational affiliation, and to use those categories to
define themselves. For instance, people appear particu- larly able to discriminate
between in-groups and out-groups and to be at- tracted to those seen as similar to
themselves (Brewer, 1979; Moreland, 1985). Drawing on underlying values,
individuals may manage their lives in ways that help them choose congruent roles,
occupations, and even organi- zations (Albert & Whetten, 1985; Sampson, 1978;
Stryker & Serpe, 1982). Schneider (1987) proposed that individuals may be attracted
to organiza- tions they perceive as having values similar to their own. In addition,
organ- izations attempt to select recruits who are likely to share their values. New
entrants are then further socialized and assimilated, and those who don't
fit leave. Thus, basic individual values or preferences for certain modes of conduct
are expressed in organizational choices and then reinforced within organizational
contexts. Just as research has shown that similar back- grounds, attitudes, and
experience can increase liking between individuals (Tsui & O'Reilly, 1989), it may be
that organizations that manifest and re- ward characteristic outcomes and behaviors
will be more or less attractive to different types of people. Values provide the starting
point, with the joint processes of selection and socialization acting as complementary
means to insure person-organization fit (Chatman, 1988). Thus, congruency
between an individual's values and those of an organization may be at the crux of
person-culture fit.1991 O'Reilly, Chatman, and Caldwell 493 The Assessment of Person-
Culture Fit There are some fundamental areas of agreement in the definition of
culture, but less agreement exists about its measurement. As Rousseau noted at the
very outset of her review, "Quantitative assessment of culture is controversial" (1990:
1). She discussed how advocates of qualitative meth- ods for studying culture have
argued that much of what constitutes a culture may be a unique social construction
of reality, perhaps unconscious on the part of the culture's members (e.g., Schein,
1985; Smircich, 1983). Acknowl- edging that some aspects of organizational culture
may not be easily acces- sible, Rousseau also asserted that certain dimensions of
culture may be ap- propriately studied using quantitative methods, indeed suggesting
that quantitative assessments offer an opportunity to understand the systematic
effects of culture on individual behavior. One way to assess culture quantitatively is
to focus on the central values One way to assess culture quantitatively is to focus on
the central values that may be important to an individual's self-concept or identity as
well as that may be important to an individual's self-concept or identity as well as
relevant to an organization's central value system. Weiner suggested this
perspective, noting that "when a number of key or pivotal values concerning
organization-related behaviors and state-of-affairs are shared across units and
levels-by members of an organization, a central value system is said to exist" (1988:
535). To characterize an organization's culture in terms of its central values requires
first that the range of relevant values be identified and then that an assessment be
made of how much intensity and consensus there is among organizational members
about those values (Enz, 1988; Saf- fold, 1988). O'Reilly (1989), drawing on earlier
research on measuring norms, noted two important characteristics of strong cultures.
One is inten- sity on the part of organization members, that is, displaying approval or
disapproval to those who act in certain ways; the second is the presence of
crystallization, or widespread agreement on values, among members. If there is no
substantial agreement that a limited set of values is important in a social unit, a
strong culture cannot be said to exist. If there is strong and widespread agreement
about the salience and importance of specific values, a central value system or unit
culture may exist. Much previous research has suggested that person-
culture fit increases commitment, satisfaction, and performance, but very little
empirical re- search on these relationships has been done. The general
research question examined here was the following: To what extent is person-
culture fit asso- ciated with individual commitment, satisfaction, and longevity with an
or- ganization (Holland, 1985; Meir & Hasson, 1982; Mount & Muchinsky, 1978)? We
expected to find that high levels of person-culture fit would be positively associated
with those outcomes. Addressing that question required two types of analyses. First,
we needed to demonstrate that preferences individuals have for
organizational cultures are comparable to cultures that exist. Second,
the relationship be- tween individual preferences and organizational culture
needed to be as- sessed across a broad range of values. Therefore, we tested our
general re-494 Academy of Management Journal September search question by
developing a method of assessing culture based on the extant values of
organizations and measuring person-culture fit through a semi-idiographic technique
based on the profile comparison process (Cald- well & O'Reilly, 1990; O'Reilly et al.,
1990). METHODS Overview To investigate person-culture fit, we developed an
instrument we called the Organizational Culture Profile (OCP). This instrument
contains a set of value statements that can be used to idiographically assess both
the extent to which certain values characterize a target organization and an
individual's preference for that particular configuration of values. Person-culture fit
can be calculated by correlating the profile of organizational values with
the profile of the individual's preferences. In a set of related investigations using
multiple sets of respondents, we explored the characteristics of the OCP and
demonstrated its ability to assess both organizations and people. In addition, we
explored the relationship both organizations and people. In addition, we explored the
relationship between preferences for organizational values and individual personality
between preferences for organizational values and individual personality variables.
Finally, we used the OCP to assess person-culture fit and test for the relationship
between fit and work-related outcomes. Development and Use of the Organizational
Culture Profile The OCP was developed and used to measure person-organization fit.
The approach to its development followed Caldwell and O'Reilly (1990).
The OCP contains 54 value statements that can generically capture individ- ual and
organizational values. Following the general procedure for generat- ing
Q-sort profiles (Block, 1978), we had respondents sort the 54 items into nine
categories, ranging, for instance, from most to least desirable or from most to least
characteristic, and to put a specified number of statements in each category; the
required item-category pattern is 2-4-6-9-12-9-6-4-2. Fewer items are required at the
extremes than in the central, more neutral categories. The question respondents
were asked to keep in mind while sorting the deck varied according to
whether they were describing their own preferences or the value system
or culture of a focal organization. To develop a profile of an
organization's culture, we instructed respondents familiar with the
organization to sort the 54 value statements according to the extent to
which the items were characteristic of the organization. For individual
preferences, individuals were asked to sort according to their personal pref- erences
for each value in their ideal organization. With this procedure, sep- arate groups of
individuals can be used to assess a firm's culture and provide ratings of preferences.
The degree to which the organization's values are consistently shared can be
investigated by the intercorrelation among raters using a variation of the
Spearman-Brown general prophecy formula (Nun- nally, 1978). A more
complete description of the development and general use of the OCP follows. 1991
O'Reilly, Chatman, and Caldwell 495 Step 1 -Describing organizational values. The set
of value statements was developed on the basis of an extensive review of academic
and practi- tioner-oriented writings on organizational values and culture (cf. Davis,
1984; Deal & Kennedy, 1982; Kilmann, 1984; Ouchi, 1981; Peters & Water- man,
1982; Schein, 1985). The purpose of this review was to identify a comprehensive set
of values that could be used to characterize both indi- viduals and organizations. An
attempt was made to identify items that (1) could be used to describe any
person or organization, (2) would not be equally characteristic of all people or
organizations, and (3) would be easy to understand. The initial pool of items
consisted of over 110 items. For the final set, we used four criteria: (1) generality-an
item should be relevant to any type of organization, regardless of industry, size, and
composition; (2) discrimina- bility-no item should reside in the same category for all
organizations; (3) readability-the items should be easily understandable to facilitate
their hav- ing commonly shared meanings; and (4) nonredundancy-the items should
have distinct enough meanings that they could not substitute for one another
consistently. Therefore, in addition to the literature search, we made a sys-
consistently. Therefore, in addition to the literature search, we made a sys- tematic
empirical check to insure that the items met those criteria. tematic empirical check to
insure that the items met those criteria. We asked 38 undergraduates
participating in a vocational interest feed- back program, all seniors majoring
in business administration, and four fac- ulty members to screen the 110 items,
identifying items that were redundant, irrelevant, or difficult to understand.
They were also asked to identify any unincluded items that would be important
descriptors of an organization's culture. We made a similar check with respondents
from a set of accounting firms. After several iterations, we obtained a final
set of 54 value-based characteristics, which are listed in the Appendix. Step 2-
Assessing characteristics of firms. To obtain profiles of the cultures of firms, we
identified sets of key informants with broad experience and asked them to sort the 54
items in terms of how characteristic each was of their organization's culture.
Respondents received the following defini- tion and instructions: "Important
values may be expressed in the form of norms or shared expectations about what's
important, how to behave or what attitudes are appropriate. Please sort the
54 values into a row of nine categories, placing at one end of the row those
cards that you consider to be the most characteristic aspects of the culture of your
organization, and at the other end those cards that you believe to be the
least characteristic...." To study eight accounting firms, Chatman (1988)
used an average of 16 accountants per firm with an average tenure of
eight years. She constructed separate firm profiles by averaging the
responses of the raters within each firm. The extent to which individuals in a
firm described it in a consistent way was assessed through a variation of the
Spearman-Brown general proph- ecy formula. The eight profiles showed
substantial reliability, with an aver- age alpha of .88, representing a
range of .84 to .90, indicating relatively high levels of agreement among the
raters in each firm. The similarity of the cultures of the eight firms was assessed by
correlating the overall firm pro-496 Academy of Management Journal September files
with one another. These correlations ranged from .29 to .85, suggesting substantial
variability in the extent to which any two firms had similar cultures. Reliabilities for
the government agency and other accounting firms used in the studies reported here
were also high. Step 3-Assessing individual preferences. To assess individual prefer-
ences for organizational cultures, respondents were asked to sort the 54-item deck
into the nine categories by responding to the question, "How important is it for this
characteristic to be a part of the organization you work for?" The answers ranged
from "most desirable" to "most undesirable." To assess the test-retest reliability, or
stability, of such preferences, we had 16 M.B.A.- degree candidates complete Q-
sorts 12 months apart. The instructions for these respondents were identical to those
for the other groups providing data for the overall effort reported here. Correlations
over the year averaged .73, suggesting a high stability of preferences. To investigate
possible social de- sirability bias in the sorts, we gave eight doctoral students at the
same uni- versity a description and definition of what constitutes social desirability
bias and asked them to sort the 54 items in the most socially desirable way. Such
biases, if undetected, could limit the variability of the profiles. In order to minimize
such limitation, we cast all items in the OCP in socially neutral or slightly positive
terms. This social desirability profile was then compared or slightly positive terms.
This social desirability profile was then compared to firm profiles for the eight
accounting firms. No evidence of social desir- to firm profiles for the eight accounting
firms. No evidence of social desir- ability bias emerged-organizations' members did
not appear to be sorting the OCP in a way calculated to make their firms look like
good places to work. Step 4-Calculating the person-organization fit score. We
calculated a person-organization fit score for each individual by correlating the
individ- ual preference profile with the profile of the firm for which the person worked.
More thorough descriptions of this approach to assessing fit appear in Chatman
(1989) and in Caldwell and O'Reilly (1990). Respondents Data to develop and test
the OCP were obtained from five separate groups of respondents. The first consisted
of 131 first-year M.B.A. students at a west coast university voluntarily participating in
a managerial and per- sonality assessment project (group 1). They
completed the OCP to assess their preferences for organization values and
provided substantial person- ality data, completing the Adjective Check List (Gough
& Heilbrun, 1980), a well-developed and widely used measure of personality. Men
comprised 57 percent of group 1, women 43 percent; the average age was 27.7 and
the average years of work experience was 3.2. We used this group in two ways, first
in combination with other groups to assess the structure of individual
preferences for organizational values, and second, to investigate the rela- tionship
between personality and preferences for organizational cultures. A second group of
93 M.B.A. students at a midwestern university pro- vided OCP data on individual
culture preferences (group 2). Men comprised 54 percent of this group; the average
age was 26 and the average number of 1991 O'Reilly, Chatman, and Caldwell 497 years
of work experience was two. We combined this group with others to assess the
structure of individual preferences. The third group of respondents was part of a
longitudinal study that tracked new accountants as they entered and proceeded
through their first two years in west coast offices of eight of the largest U.S. public
accounting firms (Chatman, 1988). In each of the eight firms, most or all of the
accoun- tants hired in 1986 into the audit function participated in the study; for all the
firms, there were 171 respondents, representing 84 percent of the eligible employees
(the mean per firm was 22, s.d. = 5). The demographic charac- teristics of group 3
were as follows: 47 percent were men; the average age was 24 in the fall of 1986; all
had bachelor's degrees; and 25 percent had master's degrees. At the time of data
collection, salaries were nearly identi- cal across the eight firms (3x = $21,500,
s.d. = $1,000), and all respondents entered their firms with the title of staff
accountant. These individuals provided data on their preferences by completing the
OCP. In addition, we surveyed them approximately 12 months later regard- ing their
job satisfaction, organizational commitment, and intent to leave. A measure of actual
turnover was obtained approximately 24 months after the OCP data were
collected. Data from this group were used in two ways. First, we combined
their OCP responses with those obtained from the first two groups to assess the
structure of individual preferences for organizational values. Second, we correlated
individual OCP data with firm-level measures values. Second, we correlated
individual OCP data with firm-level measures of corporate culture (obtained
from a fourth group of respondents) to provide of corporate culture
(obtained from a fourth group of respondents) to provide a measure of
person-culture fit. This fit score was related to the individual outcome variables of
job satisfaction, commitment, intent to leave, and ac- tual turnover. Data
were also obtained from 128 senior accountants employed by the eight firms
taking part in the longitudinal study of new accountants. Ap- proximately
18 individuals completed the OCP for each firm, describing the pattern of values that
characterized the organization. All these informants had at least two years
experience in the firm. We developed an overall profile of the culture of each firm by
averaging the individual responses. There was a high level of agreement
among the members of each firm, as shown by Spearman-Brown
coefficients ranging from .84 to .90. We used data from this sample to
generate the aggregate profiles describing the cul- ture of each of the eight
accounting firms. These were then used to calculate the person-culture fit scores for
the newly hired accountants. The fourth group of respondents consisted of 96
certified public accoun- tants from six offices of major accounting firms in the west
central United States. Of group 4, 63 percent were men; the group's
average tenure was 85 months and 55 percent were over 30 years of
age. These respondents pro- vided assessments of firm culture. We combined
these data with data from a fifth group of respondents to analyze the structure of
OCP descriptions of firm cultures. The fifth group consisted of 730 middle-level
managers employed by a government agency who were attending a continuing
management develop-498 Academy of Management Journal September ment program
at an east coast university. In group 5, 88 percent of the respondents were men; the
average age was 45.9 and the average tenure with the employing agency was 21.4
years. All described the values of their or- ganization by completing the OCP. We
combined these data with those from group 4 to analyze the structure of the values
defining corporate culture. Measures To test the general hypothesis that person-
organization fit is related to work outcomes, we measured a number of other
variables: Person-organization fit. The new accountants in group 3 sorted the items
in the OCP in terms of their own preferences for organizational culture. The senior
accountants in group 3 sorted the items in terms of how descrip- tive they were of
their firms. Thus, we developed a profile of the culture of each firm and calculated
person-organization fit by correlating the rankings of the set of 54 individual
preferences obtained from the new accountants with the rankings of the 54
values obtained from the senior accountants in the firm that employed them.
The person-organization fit correlations ranged from -.36 to + .62. To establish the
predictive validity of person-organization fit, follow-up data were collected from the
entry-level accountants, through surveying re- spondents from seven of the
eight firms about one year after the initial data collection. Because of a
delay in collecting initial data at the eighth firm, it was impossible to
survey the individuals at this firm, although the firm provided information
about turnover. This limitation reduced the potential number of
respondents in the second survey to 144. Of that number, 6 had number
of respondents in the second survey to 144. Of that number, 6 had left
their firms and 2 were on extended leaves of absence at the time of the
left their firms and 2 were on extended leaves of absence at the time of
the follow-up survey. Eligible respondents returned a total of 127
surveys, yield- ing a response rate of about 92 percent. Questions in the survey
used related to commitment, job satisfaction, and intent to leave. Organizational
commitment. Commitment was measured using O'Reilly and Chatman's (1986)
12-item scale. A principal components anal- ysis with varimax rotation
yielded two factors. One factor, normative com- mitment, was defined by
eight items representing commitment based on an acceptance of an
organization's values. The second factor, instrumental com- mitment,
was defined by four items describing commitment based on ex- change,
or in response to specific rewards. These factors are consistent with
recent findings (Caldwell, Chatman, & O'Reilly, 1990). We calculated sepa- rate
factor scores for normative and instrumental commitment and used them in
subsequent analyses. Job satisfaction. Overall satisfaction with a job was
measured using the single-item Faces Scale (Kunin, 1955). Recent
research has shown this scale to be the most balanced job satisfaction
measure in terms of capturing pos- itive and negative affect and
cognitions (Brief & Roberson, 1989). Intent to leave. Intentions of leaving an
organization were measured with four 7-point Likert-type questions: (1) "To
what extent would you prefer another more ideal job than the one you
now work in?" (2) "To what1991 O'Reilly, Chatman, and Caldwell 499 extent have
you thought seriously about changing organizations since be- ginning to work here?"
(3) "How long do you intend to remain with this organization?" and (4) "If you have
your own way, will you be working for this organization three years from now?" Since
a principal components anal- ysis of the questions yielded a single factor, we
calculated one factor score to measure intent to leave. Turnover. Approximately one
year after the administration of the sec- ond survey and two years after person-
organization fit was measured, each of the eight firms provided a list of individuals
who had left and the dates of their departures. Although responses to the
commitment, satisfaction, and intent-to-leave questions were available from only 127
individuals, actual turnover data were available for all the original respondents. Of
the 171 individuals for whom person-organization fit scores were available, about 28
percent (N = 47) had left their firms in the two-year period. Control variables. Since
the initial person-organization fit data were obtained very soon after individuals
joined their firms, during what is po- tentially a key time in shaping future attitudes
(e.g., Louis, 1980), we con- trolled tenure with a firm at the time initial data
were collected (x = 19 days, s.d. = 27). Age and gender were also used as
control variables. Although respondents differed in the degrees they had obtained,
we did not use degree as a control because all these individuals had identical jobs
and because the quality of the programs from which they had graduated varied
substantially. It therefore seemed that issues degree might normally index, such as
expec- tations and career prospects, would not be related to that variable for the
individuals studied. RESULTS RESULTS Although the initial development of the
OCP showed good internal and Although the initial development of the OCP showed
good internal and test-retest reliability, validity remained a major concern; did the
OCP dis- criminate among individuals and organizations in terms of their central
value systems and did the measure of individual-culture fit have predictive validity?
To test those questions, we used two general types of analyses. First, we conducted
separate factor analyses of the individual (Table 1) and organizational profiles
(Table 3) to examine the dimensionality underlying the OCP. To be
useful, the dimensions of individual preferences and organ- izational
cultures should be comparable. Evidence of such comparability would indicate
that the types of cultures individuals indicate they want are generally
equivalent to the cultures organizations offer, and lack of compa- rability would
reduce the meaningfulness of person-organization fit. In ad- dition, evidence that the
individual dimensions of culture are associated with characteristically different
personality types would suggest that the underlying factors are
psychologically meaningful (Table 2). In addition to seeking evidence of
discriminant validity, we used a second set of analyses based on person-
organization fit scores to predict satisfaction, commitment, and tenure (Tables 4-
6). Taken together, the results of these analyses dem-500 Academy of
Management Journal September Results of Factor Analysis Individual Preferencesa
Organizational Atention Outcome Emphasis on Team
Culture Inovation: to Detail: Orientation: Agresivenes: Suportivenes: Rewards: Decisivenes: Profile
Item Factor 1 2 3 Stability -
4 5 6 7 8

.6 .04 .25 .05 .03 .01 .06 Inovation .51 -.05


07 .07 02 09 Experimenting .59 -.12 03 05 08 04
Risk taking .65 -.04 06 .2 20 10 08 05 Careful
-.42 .3 25 1 15 07 .06 .16 Rule oriented -
.43 .38 .06 - .04 .16 .02 .07 .09 Security -.53
24 30 06 .10 .15 03 .13 Highly organized -
.47 .24 21 01 17 05 03 24 Analytical .01 .56 .13 -
.09 .03 .06 Atention to detail -.08 .75 05 .02 06
03 Precise -.09 .75 .12 .01 05 1 Calm -
.16 .10 46 .08 - .04 .0 .12 Achievement
oriented - .14 .04 .62 .08 .0 .25 .03 .05
Demanding .19 -.01 .57 .21 15 07 .02 High
expectations .12 -.03 .65 .08 14 06 .01 Results
oriented .07 -.01 .49 15 18 .20 .05 14
Oportunities .17 .01 - .08 .5 .05 .25 .04 .10
This content downloaded from 36.84.1.157 on Sun, 28 Oct 2018 00:39:36 UTC
All use subject to https://about.jstor.org/terms 1991 O'Reilly, Chatman, and Caldwell
501 Organizational Atention Outcome Emphasis on Team
Culture Inovation: to Detail: Orientation: Agresivenes: Suportivenes: Rewards: Decisivenes: Profile
Item Factor 1 2 3 4 5 6 7 8 Agresive .09 -.08 .13 .75 1 09
12 14 Socialy responsible .1 - .19 .06 -.51 .09 .28 .05
Competitive .0 - .09 .18 .5 .40 .13 .04 Shares
information .30 - .08 .0 .12 .4 .01 .21 .24
Suportive -.08 .03 21 .63 03 .14 .09 Praises
performance - .19 .1 .06 .0 .54 .10 .13 .07 Long
hours -.01 .21 .14 .20 53 .02 .12 .16
Profesional growth - .08 .05 .12 .15 .68 .03
performance .07 - .08 .14 .03 .16 .6 .20 .06
Fiting in -.23 02 .0 09 .17 .41 .03 .21
Autonomy .19 -.02 04 27 21 0 45 07 Team
oriented -.10 .02 06 .01 .07 .03 .75
Colaboration .04 - .09 .01 .19 .12 .70 .03
Predictability - .3 .2 .03 .04 .0 .02 .4
Decisivenes - .03 .10 .02 .09 .04 .06 .15 .65
Low conflict - .26 .20 .25 .15 .09 .56
Eigenvalues 5.28 4.16 3.1 2.3 1.93 1.73 1.61.49
acounted for .10 .08 .06 .04 .03 a N = 395. Boldface
statistics represent loadings greater than .40 on that factor.
This content downloaded from 36.84.1.157 on Sun, 28 Oct 2018 00:39:36 UTC
All use subject to https://about.jstor.org/terms 502 Academy of Management Journal
September onstrate that person-organization fit possesses predictive validity and is
or- ganizationally useful. Discriminant Validity and the Factor Structure of the
Organizational Culture Profile Recall that each respondent was asked to sort the 54
items in the OCP into nine categories ranging from most to least desirable according
to how important it was for the person that the characteristic represented be a part of
an organization's culture. Following Block (1978), we analyzed data from the two
groups of M.B.A. students and the new accountants (groups 1, 2, and 3, N = 395)
using principal components analysis with varimax rotation. Results of an analysis of
all 54 items revealed 33 items with loadings of greater than .40 on a single factor.
Table 1 shows results; items with signif- icant cross-loadings are not reported. From
a scree test, eight interpretable factors with eigenvalues greater than 1.0 and defined
by at least three items emerged. This pattern shows that an organization's culture
can be characterized by innovation and risk taking (factor 1), attention to detail
(factor 2), orientation toward outcomes or re- sults (factor 3), aggressiveness and
competitiveness (factor 4), supportive- ness (factor 5), emphasis on growth and
rewards (factor 6), a collaborative and team orientation (factor 7), and decisiveness
(factor 8). These eight or- thogonal factors are unambiguously defined. In general,
they approximate many of the dimensions to which the qualitative literature on
culture has often referred (e.g., Deal & Kennedy, 1982; Peters & Waterman, 1982). It
would support the reasonableness of those dimensions if different types of
individuals reported preferences for cultures differing along the dimensions. To
investigate, we computed factor scores for the west coast M.B.A. students (group 1,
N = 131). We then correlated those scores with a very well-developed measure of
normal personality, the Adjective Check List (Gough & Heilbrun, 1980). The
instrument, which provides multiple measures of the strength of motives such as
affiliation, aggression, and achievement, has demonstrated substantial reliability and
validity, with a median scale alpha of .76. One indication of the validity of the OCP
dimen- median scale alpha of .76. One indication of the validity of the OCP dimen-
sions would be the emergence of distinctive preferences for different organ- sions
would be the emergence of distinctive preferences for different organ- izational
cultures among respondents with characteristically different per- sonality attributes.
Table 2 presents the correlations among a set of Adjec- tive Check List measures
and the eight OCP factor scores. Although we do not report specific hypotheses here,
good support for seven of the eight factors can be seen in the form of easily
interpretable patterns of personality and cultural preferences. For instance,
individuals with high needs for achievement show a significant preference for
aggres- sive, outcome-oriented cultures. Respondents with high needs for
autonomy show a preference for innovative cultures and negativity toward those
char- acterized by an emphasis on supportiveness and teamwork. Only the detail-
orientation factor, defined by a preference for precision, analysis, and
atten- tion to detail, shows no correlation with any personality dimension. This1991
O'Reilly, Chatman, and Caldwell 503 Corelations Betwen Adjective Check List Measures and
Organizational Culture Profile Factorsa

Adjective
Emphasis Check List
Atention Outcome on Team Scales Inovation to Detail

Orientation Agresivenes Abasement -


Suportivenes Rewards Decisivenes

.30* .09 .06 .28* .19* .05 .07 .03 Achievement -


.12 .24* .25* .13 .04 .03 Afiliation -
.01 .10 .05 .03 .02 .19* .20* .29* Agresion .26* -
.01 .06 .16 .27* .09 .13 Autonomy .3* -.07
03 .10 2* .05 - .21* 16 Change .30* - .13 -
.01 .05 18* .04 07 05 Creativity -.1 .02 .10
19* .30* 03 15 Deference -.40* .05 - .02 .15 .21
* .07 .17 .13 Dominance .26* -
.1 .18* .28* .19* .05 .01 .04
Endurance .02 .04 .19* .13 .01 - .08
Exhibition .23* -.05 .05 .1 21* .09 07
Nurturance - .19* .10 .02 .17 .14 .25* .28* .21*
Self-confidence .20* - .10 .15 .18* .12 .04 .09
Sucorance - .23* .10 .08 .2* .16 .03 .05 N = 19. * p
< .05
This content downloaded from 36.84.1.157 on Sun, 28 Oct 2018 00:39:36 UTC
All use subject to https://about.jstor.org/terms 504 Academy of Management Journal
September absence may reflect the fact that the Adjective Check List does not contain
any assessment of obsessive-compulsive tendencies. Overall, the results sug- gest
that the dimensions underlying individual preferences for distinctive cultures are
meaningfully associated with characteristic differences in un- derlying personality
variables. Additional important questions are whether the OCP reflects meaning- ful
organizational dimensions and whether the individual and organiza- tional matrices
are similar. To address those issues, we had 826 respondents from the government
agency and six accounting firms (groups 4 and 5) profile the culture in their
organizational units. Each respondent sorted the 54 items into nine categories on the
basis of how much each attribute char- acterized the focal organization; that is,
individuals described their organi- zation's culture, not their personal preferences.
We performed a principal components analysis and varimax rotation using those
data, again retaining items with loadings greater than .40. Table 3 shows the factor
loadings for the 26 items that loaded cleanly on factors retained on the basis of a
scree test. In this instance, seven clearly defined factors emerged. An inspection of
the two factor analyses reveals that five of the eight factors shown in
Table 1 are replicated almost exactly-innovation, outcome orientation, aggres-
siveness, detail orientation, and team orientation. Sixteen of the 20 items shown in
Table 3 that are also in Table 1 load on the same factors. The nonreplication of the
other three individual preference dimensions (sup- portiveness, emphasis on
rewards, and decisiveness) seems to result from a lack of inclusion of the same
items. Although direct comparison of the factor structures could be misleading
because of the different stem questions, over- all there appears to be good
comparability between cultures as defined by individual preferences and actual
organizational descriptions. The results of the factor analyses suggest that the OCP
can provide a reasonable mapping of organizational culture. Person-Organization
Fit and Individual Outcomes The evidence suggests that the 54 statements
represent one possible approach to depicting culture, but an important question
remains: Is person- organization fit systematically related to relevant
organizational outcomes such as satisfaction, commitment, and turnover? Table
4 shows the correla- tions between person-organization fit and a set of outcome and
control vari- tions between person-organization fit and a set of outcome and control
vari- ables. The correlation between an individual's preferences and a firm's char-
ables. The correlation between an individual's preferences and a firm's char-
acteristics across the entire set of 54 items measures person-organization fit. Of
central interest are the correlations between person-organization fit and normative
commitment (r = .25, p < .01), overall job satisfaction (r = .35, p < .01), and intent to
leave an organization (r = - .37, p < .01). These relationships suggest clearly that
high person-organization fit as measured at time 1, when respondents
originally entered their firms, is associated with high positive affect and a low
intent to leave at time 2, a year later. Person-1991 O'Reilly, Chatman, and Caldwell 505

Inovation .76 - .08 .06 .07 .0 .12 .04


Oportunities .50 - .23 .09 .07 .01 .03 .2
Experimenting .76 - .03 .04 .15 .1 .06 Risk
taking .76 -.08 .02 13 .04 .07 Stability -
.09 .75 .08 .1 .01 .06 Respect for individual -
.05 .06 .75 04 12 .0 14 Achievement oriented .01 -
.08 07 .73 04 .0 03 Analytical -.04 13 10
02 .5 .06 01 Team oriented .14 -.06 .03 02
07 .69 .12 Colaboration .0 -
.2 .03 .01 .15 .58 .06 People
oriented .10 .03 .3 - .05 .17 .48 .1
Agresive .13 - .17 .24 .2 .02 -.24 .42
Competitive - .04 -.25 .19 .14 .06 .02 .49
Socialy responsible - .13 .03 .12 .08 .05 .2 -.63
Eigenvalues 6.75 4.14 2.20 2.12 1.84 1.69
1.54 acounted for .13 .08 .04 .03 a The data represent 826
acountants in seven firms. Boldface statistics loadings greater than .40 on that factor.
This content downloaded from 36.84.1.157 on Sun, 28 Oct 2018 00:39:36 UTC
All use subject to https://about.jstor.org/terms 506 Academy of Management Journal

September TABLE 4 Means, Standard Deviations, and Corelations Variables


Means.d. 1 2 3 4 5 6 7 8 1. Person-organization fit 0.23

0.19 2. Age 23.32 3.45 .14 3. Gender 1.53 0.50 -.08


10 4. Tenurea 19.0 27.0 .12 .04 -.01 5. Normative comitment 0.0
1.0 .25* -.08 .01 02 6. Instrumental comitment 0.0 1.0 .15 - .03 .01 .02 .0 7.
Job satisfaction 5.10 1.31 .35* .03 - .01 .09 .39* .62* 8. Intento leaveb
0.0 1.0 .37* -.1 .12 .0 34* 53* 75* 9. Turnover 1.27 0.45 -
.16* .13 06 .05 .10 30* .24* a Tenure was measured in days. b This variable had a
value of 1 for intento stay and 2 leave.
This content downloaded from 36.84.1.157 on Sun, 28 Oct 2018 00:39:36 UTC
All use subject to https://about.jstor.org/terms 1991 O'Reilly, Chatman, and Caldwell
507 TABLE 5 Results of Regression Analysisa Normative Instrumental Job Intent to
Variables Commitment Commitment Satisfaction Leave Person- organization fit .28** -
.07 .36** .36** Age -.11 -.06 -.05 -.04 Gender -.04 -.11 -.01 .12 Tenure .01 .06 .08 .01 Adjusted
Rf2 .06 .10 .12 F 2.62* n.s. 4.31** 5.04** a Entries represent standardized regression
coefficients. *p < .05 ** p < .01 organization fit is not significantly related to either age
or gender; however, in order to insure that these variables were not affecting the
bivariate rela- tionships, we included them as controls in the regression equations.
Again, the results, reported in Table 5, are straightforward. Person-organization fit is
a significant predictor of normative commitment, job satisfaction, and intentions to
leave, independent of age, gender, and tenure. Approximately 12 months after the
collection of satisfaction and com- mitment measures, or 24 months after we
assessed person-organization fit, we collected data on the employment status of all
respondents. To assess whether person-organization fit would predict actual
turnover 24 months later, we used survival analysis (Kalbfleish & Prentice, 1980).
Several recent papers (e.g., Fichman, 1988; Morita, Lee, & Mowday, 1989; Peters &
Sheri- dan, 1988) have demonstrated the appropriateness of survival analysis for
turnover research. This technique takes explicit account of time in the anal- ysis of
turnover and corrects for right censoring in the data. Table 6 presents
these findings. To determine whether person-organization fit has an effect on staying
with a firm, the log likelihood of the base equation (model 1) was
compared to the log likelihood of model 2. The chi-square of this difference is
tically significant (X2 = 8.69, p < .01), indicating that
statis-
person-organization fit positively predicts the probability of a person's
staying with a firm. Al- though the form of the relationship between person-
organization fit and staying is nonlinear, an individual with a perfect fit score
would be pre- dicted to stay approximately twice as long as would be predicted
using the base model alone. Similarly, an individual with a very high negative fit
score would be predicted to stay about half as long as would be predicted
with the base model.2 2 For any respondent in this group, the predicted time of
staying can be calculated by 2 For any respondent in this group, the predicted time
of staying can be calculated by multiplying the exponential of the coefficient for person-
organization fit (.64 in this group) by multiplying the exponential of the coefficient for person-
organization fit (.64 in this group) by each respondent's person-organization fit score. Thus,
for a person who has a fit score of 1.0, the (continued)508 Academy of Management Journal
September TABLE 6 Results of Survival Analysis Independent Model 1 Model 2 Variables
Parameter Estimatesa s.e. Parameter Estimatesa s.e. Person-organization fit .64* .29 Age -
.02 .01 - .02 .01 Gender .05 .10 .05 .10 Scale parameter .34 .34 Logarithmic likelihood -
97.70 - 93.36 a Estimates are unstandardized. * p < .05 for the chi-square value. Figure
1 is a graph of the survival curve of the effects of person-organization fit on the
likelihood of staying with an organization. We esti-mated the survivor function using
Kaplan and Meier's nonparametric ap-proach, as recommended by Tuma and
Hannan (1984). Because it makes no assumptions about the functional form of the
curve, the Kaplan and Meier estimator is one of the most common methods for
assessing relationships when data are right-censored. Each of the descending legs
indicates one or more instances of leaving. We divided the data into quartiles based
on per-son-organization fit scores; Figure 1 shows the survival curves of the top and
bottom quartiles. As shown, the curve for individuals with low person-organization fit
scores has a steeper descent than the curve for high scorers-that is, low scorers are
more likely to leave an organization than high scorers. Taken together, the data
shown in Table 6 and Figure 1 offer compelling evidence for the positive effect
of person-organization fit on an individual's staying with an organization.
Although it is not reported in detail, in addition to the measurement of person-
organization fit, we obtained an assessment of person-job fit. Follow-ing the
approach of Caldwell and O'Reilly (1990), we conducted a job anal-ysis for the
position held by all the entry-level accountants (group 1). We obtained a profile
consisting of 60 competencies using six subject-matter experts from each of the
eight firms. The reliability of this job profile across raters and within firms, assessed
using a variation of the Spearman-Brown general prophecy formula, was quite high
(.97). Entering accountants also provided a self-assessment using the same 60
competencies, and a person-job fit score was computed (average r = .27, s.d. = 14).
We then included this score in the analyses shown in Table 5 and as a control
variable in the calculation would be el 64>X1) = 1.89. Conversely, for a person with the
lowest possible person-organization fit score (- 1), the calculation would be e( 64X -1) - .53.
The interpretation is that with a perfect score on person-organization fit, an individual is likely
to stay twice as long as we would have predicted without having information about their
score. Conversely, a person with the lowest possible person-organization fit score will stay
approximately half the time we would have predicted without knowing their score.
All use subject to https://about.jstor.org/terms 1991 O'Reilly, Chatman, and Caldwell
509 FIGURE 1 Survival Curve 1~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1| __
~~~~Group _ ~ ~ ~ ~ ~ ~ ~ Gro~~~~~rup co -0 co cn 0 1,125 Time to leaving in days
Group 0 had the lowest 25 percent of scores on the person-organization fit measure. Group
1 had the highest 25 percent of scores on the person-organization fit measure. survival
analysis. Although person-job fit did have some independent effects on job
satisfaction and intent to leave, it had no impact on the person-culture fit
relationships reported. The simple correlation between person-job and person-
organization fit was .16 (n.s.). Thus, person-organization fit has quite independent
effects on commitment, job satisfaction, and turnover. DISCUSSION Overall, the
results presented here offer strong support for the validity of assessment of person-
organization fit on the basis of value congruency. The results suggest that the
Organizational Culture Profile shows reasonable re-liability and convergent-
discriminant validity. As results shown in Table 2 indicate, individual variations in
preferences for different organizational cultures are associated with interpretable
differences in personality charac-teristics. The structure underlying individual
preferences (Table 1) also ap-pears comparable to the structure underlying the
culture in a selection of firms (Table 3). The comparatively large number of values
examined here also provides a fine-grained evaluation of organizations' cultures. As
Chatman (1988) dem-onstrated, even organizations that appear highly similar, such
as accounting firms as a group, may vary widely in their underlying value structures.
Although there may be characteristic similarities in cultures within indus-tries (e.g.,
Burns & Stalker, 1961), use of a Q-sort approach allows for the identification of
contrasts within and across organizations. The factor analytic results provide a basis
for future research. In view of
This content downloaded from 36.84.1.157 on Sun, 28 Oct 2018 00:39:36 UTC
All use subject to https://about.jstor.org/terms 510 Academy of Management Journal
September those results, one general hypothesis might be that organization's cultures
tend to be similar when the organizations are in relatively homogeneous industries
and thus have similar sizes, structures, levels of technological maturity, personnel
configurations, regulatory demands, and orientations. Conversely, the cultures of
firms in heterogeneous industries may be less similar. Our factor analytic results
would be especially useful for addressing this hypothesis because they allow
assessment of differences in both factor patterns across industries and the mean
values of particular firms on each relevant factor. For example, there may be greater
variation on the innova-tion factor within a high-technology industry than within the
public ac-counting industry. Previous researchers have conducted a similar kind of
analysis of cultural factors across nations (Hofstede, Neuijen, Ohayv, & Sanders,
1990). The results reported here also suggest that person-organization fit may
provide meaningful insights into individuals' adjustments to organizations (e.g.,
Holland, 1985; Louis, 1980). The predictive validity shown in Tables 4 through 6 is
evidence of this potential. Measures of person-organization fit at time 1 were
significantly related to individual commitment and satisfaction approximately 12
months later. Interestingly, person-culture fit is related to normative, value-based
commitment but not to instrumental, compliance-based commitment. Caldwell and
colleagues (1990) showed that normative commitment is often associated with firms
with strong cultures. Researchers have often suggested that high commitment and
satisfaction are outcomes of high person-organization fit (e.g., Kilmann et al., 1986;
Ouchi & Wilkins, 1985), but little empirical evidence of this association has been
available. The lack of a significant correlation between congruence and instrumental
commitment here is not surprising, given that our measures of congruence are
predicated on fit between individual preferences and organizational val-ues rather
than on specific attributes of extrinsic rewards (Meyer & Allen, 1984). Of
perhaps more practical importance is the association between per-son-
organization fit and turnover almost 24 months later (Table 6). Recall that new
accountants completed the OCP during their first few days with their organizations.
We then compared individual preferences for a partic-ular organizational culture
to an organizational profile provided by a set of experienced
organizational members. The degree to which individual pref-erences
matched organizational realities was predictive of turnover two years later. Although
some authors have questioned whether the strength of an organization's culture can
be measured (Saffold, 1988), others have ar-gued that strong agreement among
organizational members about a few cen-tral norms or values defines an
organizational culture (e.g., Enz, 1988; O'Reilly, 1989; Rousseau, 1990). These
results seem to support that view. In this regard, the OCP appears useful. First, as
recent advances in the study of person-situation interaction have suggested, a Q-
sort approach pro-vides an idiographic assessment of the unique patterning of a
large number of defining attributes for individuals and organizations. In the case of
organ-
This content downloaded from 36.84.1.157 on Sun, 28 Oct 2018 00:39:36 UTC
All use subject to https://about.jstor.org/terms 1991 O'Reilly, Chatman, and Caldwell
511 izational culture, such a broad assessment would allow organizations to screen
candidates against the organizational attributes that are most relevant to their
personal values, not against some general rating of a few values that may not be
personally important. In this respect, the profile comparison approach is simply a
formalization of the logic of fit we use in everyday life. We ask "Will person X fit in
situation Y?" This is genuinely a question about person-situation congruence, not a
main effect. We are not interested in how the person will do in situations A and B or
whether person Z will fit in situation Y. To answer such questions, we must know
what a situation demands and how an individual's competencies and preferences fit
a spe-cific situation. This is precisely the information the profile comparison pro-cess
yields. Most models of fit assess individual and environmental charac-teristics
separately and then combine the two independent variables to pre-dict behavior and
attitudes (e.g., Joyce, Slocum, & Von Glinow, 1982); the profile comparison process,
however, provides a direct evaluation of the simultaneous effects of person and
situation. The process assesses the rela-tive salience and configuration of
characteristics within individuals and then permits a comparison across situations.
By evaluating the individual and situation on items that are relevant to both, the
process is similar to what Wright and Mischel (1987) referred to as the "competency
demand hypothesis." This method also permits individuals to hold values that may, in
a broad sense, be conflicting. Both Ashforth and Mael (1989) and Swidler (1986)
noted that value conflicts are common in organizations and are not measurable with
more restrictive approaches. In earlier studies, the idea of fit, although important, has
remained elu-sive. As Rynes and Gerhart (1990) pointed out, most
discussions of fit have implied something more than a simple match of an
individual to a job's requirements and have frequently invoked notions of "chemistry"
or finding the "right type" of person. In a study of recruiting M.B.A. graduates, they
showed that firm-specific characteristics had significant impacts on recruit-ers'
judgments beyond general employability attributes such as grade point averages and
previous accomplishments. The results of the present study are consistent with
those findings and suggest the fit of firm-specific and indi-vidual values may underlie
earlier discussions of chemistry. Controlling person-job fit did not affect the results
for person-culture fit as assessed here. Both are relevant. For an individual to be
satisfied and attached to an organ-ization, the person may need both task
competency and a value system congruent with the central values of the
organization. As for the organiza-tion, it needs to select people who fit a given
situation, which is likely to include some combination of task and cultural
requirements. Failure to fit on either dimension may reduce employees'
satisfaction and commitment and increase the likelihood of their leaving. The
results of this study can contribute to future research in a number of
ways. First, as organizational researchers continue to debate the impor-
tance of congruence between individuals and situations, these results dem-onstrate
that a widely used tool for assessing personality can be adapted to
This content downloaded from 36.84.1.157 on Sun, 28 Oct 2018 00:39:36 UTC
All use subject to https://about.jstor.org/terms 512 Academy of Management Journal
September provide comparable measures of persons and situations. The development of
such methods can allow for research in a number of areas where fit to a job, occupation, or
organization is conceived as either an important causal or outcome variable. Second, and
more substantive, this research provides an empirically based definition of the pattern of
values that define organiza-tional culture. Although further validation of this approach to
culture is necessary, the relatively consistent patterns of the individual preferences for
values and the observation of those values in organizations suggests that the pattern
defining organizational culture is relatively robust. Third, and per-haps of greatest
importance, these results demonstrate that the fit between an individual's preference for a
particular culture and the culture of the organ-ization the person joins is related to
commitment, satisfaction, and turnover. This study and others like it can help clarify both
the nature of organ-izational culture and the impact of cultures on individuals. REFERENCES

You might also like