® Academy of ManagemenI loumal

1994. Vol. 37. No, 3. 580-607,

DETERMINANTS OF INNOVATIVE BEHAVIOR:

A PATH MODEL OF INDIVIDUAL INNOVATION
IN THE WORKPLACE
SUSANNE G. SCOTT
University of Colorado at Colorado Springs
REGINALD A. BRUCE
University of Louisville

The present study integrated a number of streams of research on the

antecedents of innovation to develop and test a model of individual
innovative behavior. Hypothesizing that leadership, individual prob-
lem-solving style, and work group relations affect innovative behavior
directly and indirectly through their influence on perceptions of the
climate for innovation, we used structural equation analysis to test the
parameters of the proposed model simultaneously and also explored
the moderating effect of task characteristics. The model explained ap-
proximately 37 percent of the variance in innovative behavior. Task
type moderated the relationship between leader role expectations and
innovative behavior.

The central role of innovation in the long-term survival of organizations

(Ancona & Caldwell, 1987] provokes continuing interest among social sci-
entists and practitioners alike. Since the foundation of innovation is ideas,
and it is people who "develop, carry, react to, and modify ideas" (Van de
Ven, 1986: 592), the study of what motivates or enables individual innova-
tive behavior is critical. However, West and Farr noted that "there has been
scant attention paid to innovation at the individual and group levels" (1989:
17). The present study integrated a number of independent streams of re-
search on the antecedents of creativity, innovation, and organizational cli-
mate to develop and test a theoretical model of individual innovative be-
havior.
Van de Ven (1986) noted that one of the central problems in the man-
agement of innovation is the management of attention. Managing attention is
difficult because individuals gradually adapt to their environments in such
a way that their awareness of need deteriorates and their action thresholds
reach a level at which only crisis can stimulate action. A number of theorists

M. Ann Welsh and Harold Angle provided thoughtful and insightful comments during both
the conceptual and data analysis stages of the study, Deborah Bruce was extremely helpful in
the crafting of the multiple drafts of this article. We would like to enthusiastically thank each
of them as well as this journal's reviewers for their constructive assistance. This research was
funded in part by grants from the University of Cincinnati Research Council.

580
19g4 Scott and Bruce 581

have suggested that climate may channel and direct both attention and ac-
tivities toward innovation (e.g., Amahile, 1988; Isaksen, 1987; Kanter, 1988).
Following James, Hater, Gent, and Bruni, we defined climate as individual
cognitive representations of the organizational setting "expressed in terms
that reflect psychologically meaningful interpretations of the situation"
(1978: 786). The model guiding this study draws on the social interactionist
approach and posits that leadership, work group relations and problem-
solving style affect individual innovative hehavior directly and indirectly
through perceptions of a "climate for innovation."
The study setting was a research and development subunit. The organ-
izational literature has tended to treat R&D as a special case with little
relevance to other types of functional areas within organizations. Because
the central tasks of R&D traditionally have involved unstructured problem
solving, and unstructured problem solving is becoming increasingly com-
mon throughout organizations (Walton, 1985), the study of R&D profession-
als may have substantial relevance for promoting innovation among all
organizational participants.
The present study also investigated whether the type of job or task an
individual is engaged in influences the posited relationships. Task routine-
ness and the amount of discretion granted individuals in task performance
have previously been reported to moderate the relationship between climate
and performance (Middlemist & Hitt, 1981). These same variables have also
been implicated as important boundary conditions in models of creativity
(e.g., Amabile, 1988). Therefore, we tested whether type of job assignment
moderated the relationship hetween innovative behavior and each of the
predictors in the model.

THEORETICAL BACKGROUND AND MODEL

Definition of Innovative Behavior
The terms creativity and innovation are often used interchangeably in
research studies, and the distinction between the two concepts may be more
one of emphasis than of substance (West & Farr, 1990). Nonetheless, some
agreement about the terms' definitions has emerged recently; creativity has
to do with tbe production of novel and useful ideas (Mumford & Gustafson,
1988), and innovation has to do with the production or adoption of useful
ideas and idea implementation [Kanter, 1988; Van de Ven, 1986). Although
creativity is often framed as "doing something for the first time anywhere or
creating new knowledge" (Woodman, Sawyer, & Griffin, 1993: 293], inno-
vation also encompasses the adaptation of products or processes from out-
side an organization. Finally, researchers exploring innovation have explic-
itly recognized that idea generation is only one stage of a multistage process
on which many social factors impinge (Kanter, 1988).
From this perspective, individual innovation begins with problem rec-
ognition and the generation of ideas or solutions, either novel or adopted.
582 Acodemy of Management Journal June

During the next stage of the process, an innovative individual seeks spon-
sorship for an idea and attempts to build a coalition of supporters for it.
Finally, during the third stage of the innovation process, the innovative
individual completes the idea hy producing "a prototype or model of the
innovation . . . that can be touched or experienced, that can now be diffused,
mass-produced, turned to productive use, or institutionalized" [Kanter,
1988: 191).
Thus, innovation is viewed here as a multistage process, with different
activities and different individual behaviors necessary at each stage. Since
innovation is actually characterized by discontinuous activities rather than
discrete, sequential stages (Schroeder, Van de Ven, Scudder, & Polley, 1989),
individuals can be expected to be involved in any combination of these
behaviors at any one time.
A Model of Individual Innovative Behavior
In the model tested here (Figure 1), we viewed individual innovative
behavior as the outcome of four interacting systems—individual, leader,
work group, and climate for innovation.
Climate and innovative behavior. Studies at both the organizational and
subunit level have offered empirical support for climate's effects on inno-
vation (Abbey & Dickson, 1983; Paolillo & Brown, 1978; Siegel & Kaemmerer,
1978). However, empirical study of climate's effects on individual innova-
tive behavior has been limited (Amabile and Gryskiewicz [1989] is an ex-
ception).
At the individual level, climate is a cognitive interpretation of an organ-
izational situation that has been labeled "psychological climate" James,
James, & Ashe, 1990). Proponents of psychological climate theory posit that
individuals respond primarily to cognitive representations of environments
"rather than to the environments per se" (James & Sells, 1981]. Climate
represents signals individuals receive concerning organizational expecta-
tions for behavior and potential outcomes of behavior. Individuals use this
information to formulate expectancies and instrumentalities [James, Hart-
man, Stehbins, & Jones, 1977). People also respond to these expectations by
regulating their own behavior in order to realize positive self-evaluative
consequences, such as self-satisfaction and self-pride (Bandura, 1988).
Schneider (1975) suggested that there are many types of climates, and
Schneider and Reichers wrote that "to speak of organizational climate per se,
without attaching a referent, is meaningless" (1983: 21). Not all of the di-
mensions contained within omnibus climate measures [e.g., Jones & James,
1979; Pritchard & Karasick, 1973] are relevant to the criteria of interest in a
specific research study. For example, in the often-cited Abbey and Dickson
(1983) study of innovative performance among R&D units, only two of the
ten generic work-climate dimensions examined, performance-reward depen-
dency and flexibility, were consistently correlated with measures of R&D
innovation. Abbey and Dickson concluded that the climate of innovative
1994 Scott and Bruce 583

FIGURE 1
Determining Innovative Behavior: A Hypothetical Model

Psychological
Cliniiih; for
Iniinvnlioii

( Siinonii
[or

liinov;ilive
"-I

2 a)

Sysk-iiiiilic
Probli!ni-S(ilving
SIvli;

R&D units is characterized by rewards given in recognition of excellent

performance and by organizational willingness to experiment witb innova-
tive ideas.
Others have noted that innovative organizations are characterized by an
orientation toward creativity and innovative change, support for tbeir mem-
bers in functioning independently in the pursuit of new ideas [Kanter, 1983;
Siegel & Kaemmerer, 1978). and a tolerance for diversity among their mem-
bers (Siegel & Kaemmerer, 1978). Finally, adequate supplies of such re-
sources as equipment, facilities, and time are critical to innovation (Am-
abile, 1988; Angle, 1989; Taylor, 1963), and the supply of sucb resources is
another manifestation of the organizational support for innovation.
Hence, given climate's conceptualization as a determinant of individual
behavior and tbe previous empirical support for climate's effect on organ-
izational and departmental innovation, we predicted that the degree to
which organization members perceived an organizational climate as sup-
portive of innovation would affect individual innovative behavior.
584 Academy of Management Journal June

Hypothesis 1: The degree to which individuals perceive

dimensions of organizational climate as supportive of in-
novation is positiveiy reiated to their innovative behav-
ior.
Leadership and innovative behavior. Numerous writers have impli-
cated leadership as critical in the innovation process, but such accounts
have largely focused on the need for participative or collaborative leadership
styles (Kanter, 1983; Pelz & Andrews, 1966) or have provided lists of specific
activities that leaders should engage in to allow creativity to emerge (e.g.,
Amabile, 1988). Theoretical development in this area has been weak as
traditional leadership approaches are more relevant to the explanation and
prediction of productivity outcomes than to innovation outcomes (Waldman
SE Bass, 1991). In this study, we investigated two contemporary leadership
approaches. Leader-member exchange (LMX) theory [Dansereau, Graen, &
Haga, 1975; Graen & Scandura, 1987) was used as it has been previously tied
to innovation. We also investigated the effect of the role expectations of
leaders on innovative behavior because the consequences of the "Pygmalion
effect" (Livingston, 1969) on innovative behavior has yet to be tested. The
Pygmalion effect refers to tbe modification of a focal individual's behavior
based on tbe expectations for that behavior received from another (Eden,
1984).
LMX theory suggests that the quality of the relationship between a su-
pervisor and a subordinate is related to innovativeness (Graen & Scandura,
1987). In essence, theorists posit that supervisors and subordinates engage in
a role development process during which understandings are arrived at re-
garding the amount of decision latitude, influence, and autonomy the sub-
ordinates will be allowed (Graen & Gashman, 1975). Over time, some leader-
subordinate relationships develop from interactions that are formal and im-
personal (low-quality leader-member exchange) to mature interactions
characterized by trust, mutual liking, and respect (high-quality leader-
member exchange). In these latter relationships, subordinates are allowed
greater autonomy and decision latitude, both of which have been shown to
be essential to innovative behavior (Gotgrove & Box, 1970; Pelz & Andrews,
1966). Although research on the relationship between leader-member ex-
change and innovation is still in the nascent stages, preliminary results (e.g.,
Basu, 1991) have supported a positive relationship between leader-member
exchange and innovative behavior.
Hypothesis 2: The quality of Jeader-member exchange be-
tween an individual and his or her supervisor is posi-
tively related to the individual's innovative behavior.
Leader-member exchange was also hypothesized to influence innova-
tive behavior indirectly through its influence on the formation of climate
perceptions. Gontemporary theorists have focused on cognitive sense-
making to describe the formation of psychological climate and give primary
emphasis to the social influence processes that affect the sense-making pro-
1994 Scott and Bruce 585

cess (Ashforth, 1985; Glick. 1985,1988; Schneider & Reichers, 1983]. Central
to this approach is the notion that proximal others—those in close psycho-
logical proximity to a focal individual, including his or her co-workers and
leader—are likely to have a strong influence on the individual's perceptions
of psychological climate (Lewin, 1938).
In a recent integration of LMX theory and the extant research on climate,
Kozlowski and Doherty (1989) argued that because supervisors are the most
salient representatives of management actions, policies, and procedures,
subordinates tend to generalize their perceptions of supervisors to their
organization at large. Thus, subordinates successfully negotiating high-
quality relationships with their supervisors will perceive their organization
as providing greater autonomy, decision-making latitude, and supportive-
ness overall than will subordinates with low-quality relationships with their
supervisors. Several studies have reported empirical support for a positive
relationship between LMX quality and climate perceptions (Dunegan, Tier-
ney, & Duchon, 1992; Kozlowski & Doherty, 1989]. Thus,
Hypothesis 3: The quality o/leader-member exchange be-
tween an individual and his or her supervisor is posi-
tively related to the degree to which the individual per-
ceives dimensions of climate as supportive of innovation.
Managers may have expectations of subordinates that are not negotiated
through the role development process suggested by LMX theory. Suhordi-
nate roles may be rigidly prescribed hy an organization or by technological
constraints; for instance, their tasks may be routine or machine-driven. Fur-
ther, managers may have inflexible expectations for specific roles within
their domains or may lack interest in or enough imagination to negotiate
subordinates' roles with them (Craen & Scandura, 1987]. The expectations
that supervisors have for their subordinates are antecedents of the Pygma-
lion effect, and they have been suggested to shape the behavior of subordi-
nates (Livingston, 1969] by altering their self-expectancies and subsequent
motivations (Eden, 1984). Thus,
Hypothesis 4: The degree to which a supervisor expects a
subordinate to be innovative is positiveiy reJated to the
subordinate's innovative behavior.
Eden (1984) theorized that managers' expectations of subordinates' per-
formance are communicated to them through the managers' behaviors. As
with leader-member exchange, we expected that subordinates' perceptions
of their managers' behaviors would be generalized to their organizations at
large. More specifically, when managers expect subordinates to be innova-
tive, tbe subordinates will perceive the managers as encouraging and facil-
itating their innovative effort. These behaviors will be seen as representative
of their organizations at large, and therefore the organizations will be per-
ceived as supportive of innovation.
586 Academy o/Management/oumol June

Hypothesis 5: The degree to which a supervisor expects a

subordinate to be innovative is positively related to the
degree to which the subordinate perceives dimensions of
cJimate as supportive o/innovation.
Work groups and innovative behavior. Although idea generation and
evaluation within an organization may sometimes be a solitary activity,
more commonly work group members and peers influence individual inno-
vation. Researchers have typically studied work group effects at the group
level of analysis, using group outcomes or group innovation as the depen-
dent variable. The influence of work groups on individual innovative be-
havior has received minimal attention in the literature to date.
Rogers (1954) suggested that the cohesiveness of a work group deter-
mines the degree to which individuals believe that they can introduce ideas
without personal censure. Others bave suggested tbat collaborative ef-
fort among peers is crucial to idea generation (Amabile & Gryskiewicz,
1987; Sethia, 1991). Here, we tested how the quality of the working rela-
tionships between individuals and their work groups affected innovative
behavior.
Drawing from LMX theory. Seers (1989) suggested that individuals en-
gage in a role-making process with their work groups. This process may
result in high-quality team-member exchange (TMX) characterized by mu-
tual trust and respect and in cooperation and collaboration between a focal
individual and the work group. Alternatively, the role-making process may
result in low-quality team-member exchange, in which the focal individual
is not integrated into the work group and collaboration, trust, and respect
are low.
Thus, in conditions of high team-member exchange, individuals have
additional resources available to them in the form of idea sharing and feed-
back. We predicted that this availability would be positively related to in-
novative behavior.
Hypothesis 6: The quality of team-member exchange be-
tween an individuaJ and his or her work group is posi-
tively related to the individuaJ's innovotive behavior.
We also expected the quality of the relationship with a work group to
influence climate perceptions. Climate perceptions "emerge out of the in-
teractions that members of a work group have with each other" (Schneider
& Reichers, 1983: 30). Further, since work group members are typically more
similar to a focal individual than is the group's leader, social information
provided by the work group is likely to have a greater influence on individ-
ual meaning analysis (Festinger, 1954).
Thus, we suggest that when a work group supports an individual in
ways that allow innovation to emerge, offering, for example, cooperation and
collaboration, the individual is more likely to see the organization as a whole
as being supportive of innovation.
1994 Scott and Bruce 587

Hypothesis 7; The quality of team-member exchange be-

tween an individual and his or her work group is posi-
tively related to the degree to which individuals perceive
dimensions of climate as supportive of innovation.
Problem-solving style and innovative behavior. Recently, researchers
have given increased attention to specific dimensions of cognitive style as
antecedents of innovative behavior (e.g., Barron & Harrington, 1981; Jabri,
1991; Kirton, 1976). Kirton proposed that individuals .can be located on a
continuum ranging from those who have an ability to do things "better" to
those who have an ability to do things "differently" and reflecting the qual-
itatively different solutions they produce to seemingly similar problems. In
contrast, Jabri, drawing on Koestler's (1964) work on creative thinking, con-
ceptualized problem-solving style as composed of two independent modes
of thinking: associative and bisociative. Associative thinking is based on
habit, or following set routines, adherence to rules and disciplinary bound-
aries, and use of rationality and logic. It represents the systematic problem-
solving style. The systematic problem solver, working within established
methods or procedures, is likely to generate conventional solutions to prob-
lems. Bisociative thinking, in contrast, is characterized by overlapping sep-
arate domains of thought simultaneously, a lack of attention to existing rules
and disciplinary boundaries, and an emphasis on imagery and intuition. We
call this mode the intuitive problem-solving style. The intuitive problem
solver has a propensity to process information from different paradigms
simultaneously, and is therefore more likely to generate novel problem so-
lutions (Isaksen, 1987).
Neither style is considered preferable per se; it is the fit between
problem-solving style and a task and work environment that determines
individual task performance (Payne, Lane, & Jabri, 1990). Since the task
environment studied here involved R&D, and preliminary site interviews
suggested that the primary mission of the unit was the development of
novel problem solutions, we predicted that an intuitive problem-solving
style would be positively related to innovative behavior and that a sys-
tematic problem-solving style would be negatively related to innovative
behavior.
Hypothesis 8a; The degree to which an individual's prob-
lem-solving style is intuitive is positively related to his or
her innovative behavior.
Hypothesis 8b: The degree to which an individual's prob-
lem-solving style is systematic is negativaJy related to his
or her innovative behavior.
We also suggest here that problem-solving style is related indirectly to
innovative behavior through its effect on climate perceptions. Although
most climate research has treated differences in work group members' cli-
 588 Academy of Management /ourna) June

mate perceptions as error variance (e.g.. James et al., 1978), others have
argued that individual personalities, values, and cognitive characteristics
are important (e.g., James et al., 1990). This study tested the effect of prob-
lem-solving style on perceptions of the climate for innovation.
James and colleagues (1990) noted that individuals interpret environ-
mental phenomena by referencing personal values or internal standards.
Internal standards or values are also related to tbe concept of needs. Needs
in part determine the value of environmental phenomena to individuals, and
"based upon these values, the areas of climate that are likely to be high-
lighted in their perceptions" (James et al., 1978: 792). Thus, need states, such
as the need to be innovative, are likely to make certain aspects of an envi-
ronment—such as support for innovation—more salient. However, theo-
rists have given the effect of this increasing salience on individual interpre-
tations of environmental stimuli minimal attention. The results of two recent
empirical studies suggest that increasing salience results in higher internal
standards against which environmental conditions are judged (Eiter, 1991;
Isaksen & Kaufmann, 1990).
Given the lack of theoretical development in this area, we considered
the test of the relationship between problem-solving style and the climate for
innovation exploratory, and thus framed no specific hypotheses prior to
testing the model.
Covariates. We included several control variables that may influence
climate perceptions, innovative behavior, or both in testing the hypothe-
sized model. Although not of primary interest in this study, the relationship
between these variables and the mediating and dependent variables has been
well established. Previous work has shown both climate perceptions and
innovative behavior to be significantly related to a number of demographic
and position variables (James et al.. 1990; Mumford & Gustafson, 1988). In
this study, we included individual age, R&D tenure, task type, and education
as control variables in the prediction of both climate perceptions and inno-
vative behavior.

The Moderating Effect of Task Type

Despite evidence that task type and core technology moderate the rela-
tionship between climate perceptions and subunit effectiveness (e.g., Mid-
dlemist & Hitt, 1981), there has been little effort to study the moderating
effect of task at the individual level. When a task is routine or when indi-
vidual discretion is low, the relationship between climate and innovative
behavior is likely to be weaker than when the task is nonroutine and high
discretion is granted. Therefore, type of task may delineate one of the bound-
ary conditions within which the proposed model of individual innovation
applies. To test this possibility, we conducted an exploratory analysis of the
moderating effect of task type on the emergence of individual innovative
behavior.
1994 Scott and Bruce 589

METHODS
Respondents and Procedure
The respondents for this study included all engineers, scientists, and
technicians employed in a large, centralized R&D facility of a major U.S.
industrial corporation. The R&D center was engaged in applied research
within a specific technology area, and it was organized in three sections: (1)
product technology, (2) process technology, and (3) supporting technologies
(i.e., laboratory services such as analytical chemistry). Each area was headed
by a director who reported to the vice president of research at the site,
Initially, we conducted a series of interviews with the directors and vice
president of the R&D center to develop an understanding of how innovation
was viewed in the organization and to determine what specific behaviors
were seen as critical to innovation. We then conducted semistructured in-
terviews with a stratified sample [N ^ 22) of the R&D engineers, scientists,
and technicians to gain an understanding of how the employees viewed
innovation and to determine what organizational factors might play a part in
the innovative process. This information was used to offer some assurance
that the climate measure being used in the study was relevant in this setting.
Questionnaires were administered via company mail to study respon-
dents who completed tbem during normal working hours. We omitted the
responses of the 22 employees who were interviewed in tbe first stages of the
project from the subsequent analysis of the survey data. Participation was
voluntary for all employees, and confidentiality of responses was assured.
We received 189 questionnaires, a response rate of 85 percent. Incomplete
questionnaires reduced the usable responses to 172, of which 108 were from
engineers and scientists and 64 were from technicians. The average age of
respondents was 40.2 years and their average tenure in the R&D organization
was 14.4 years. Men comprised 91.6 percent ofthe group; 61.6 percent ofthe
respondents bad at least baccalaureate degrees, and 41.6 percent had post-
graduate degrees. Tests for nonresponse bias did not indicate any differences
between respondents and nonrespondents in terms of R&D tenure, level in
the hierarchy, job classification, education, or work group.
A second questionnaire was completed by all 26 managers at the re-
search site. These managers rated each of their subordinates on the criterion
variables, and they completed an item that assessed their own expectations
regarding tbe role of each subordinate (see the description of measures be-
low).

Measures
Innovative behavior consisted of six items completed by each of the
managers for eacb of their subordinates; the Appendix gives the scale. We
developed this measure specifically for use in this study, drawing on
Kanter's (1988) work on tbe stages of innovation and on our interviews with
the focal firm's directors and vice president. Responses were made on a
590 Academy of Management Journal June

five-point Likert-type scale ranging from "not at all" to "to an exceptional

degree." Gronbach's alpha on this scale was .89.
To provide some assurances as to the validity of the innovative behavior
scale, we obtained an objective measure of each respondent's innovative
history from the organization's archives. This measure consisted of the total
number of invention disclosures filed by an individual divided by his or her
organizational tenure in years. The correlation between the objective mea-
sure and the supervisors' ratings of innovative behavior was .33 (p < .001).
Problem-solving style was operationally defined by the two subscales of
Jabri's (1991) associative/bisociative index. Systematic problem-solving
style was measured with the ten-item associative scale, and intuitive prob-
lem-solving style with the nine-item bisociative scale. The response format
was a seven-point Likert scale ranging from "likely to enjoy" to "unlikely to
enjoy." All responses were reverse-coded, so a high score on the associative
scale indicated a preference for systematic problem solving, and a high score
on tbe bisociative scale indicated a preference for intuitive problem solving.
Gronbach's alphas for the associative and bisociative scales were .90 and .91,
respectively.
Jabri (1991) reported that qualitative testing conducted during the de-
velopment of the associative/bisociative index suggested the scales had good
content validity. In addition, high correlations between individuals' self-
ratings of problem-solving style and the supervisors' ratings of these indi-
viduals provided some evidence of concurrent validity (r's = .94, associa-
tive, and .69, bisociative). Further, the index is very similar in content to the
more widely accepted but less accessible Kirton Adaption-Innovation Mea-
sure (KAI; Kirton, 1976). For example, sample items from the associative/
bisociative index include "linking ideas which stem from more than one
area of investigation" (intuitive) and "being methodical and consistent in
tbe way I tackle problems" (systematic). Sample items from the KAI include
"copes with several new ideas and problems at the same time" and "is
methodical and systematic." Gomparison of the two scales provided further
evidence of the content validity of the associative/bisociative index.
Leader-member exchange quality was measured using the 14-item scale
developed by Graen, Novak, and Sommerkamp (1982). The measure was
administered to all engineers, scientists, and technicians. The response for-
mat was standardized using a five-point Likert scale ranging from "strongly
disagree" to "strongly agree." The scale measures the quality of the relation-
ship between manager and subordinate. Gronbach's alpha for this sample
was .90.
The role expectations of the leaders were measured by the following
single item: "Not all work roles require individuals to be innovative. In fact,
it could be argued that effective work groups have a blend of innovative
individuals and individuals whose role it is to support the innovation of
others. In this context, the role is a set of expectations of the position inde-
pendent of the person holding the position. Indicate the degree to which you
would describe the role for each of your subordinates as being either an
1994 Scott and Bruce 591

innovator or being a supporter of innovation." The supervisors rated each

suhordinate using a five-point Likert scale ranging from "role requires an
innovator" to "role requires a supporter." The item was reverse-coded so
that a high value indicated an innovative role and a low one, a supportive
role. A second administration of this measure was conducted 14 months
after the first (N = 142), and test-retest reliability was .87.
Team-member exchange quality was measured using the 12-item scale
developed by Seers (1989). The scale was administered to all engineers,
technicians, and scientists in the study. The response format was a five-
point Likert-type scale ranging from "strongly disagree" to "strongly agree."
The scale measures the quality of the working relationship developed be-
tween a focal individual and a work group. Cronbach's alpha in this study
was .84,
The climate for innovation measure was completed by all study partic-
ipants and contained 26 items. The measure was a modification and exten-
sion of the innovative climate measure developed by Siegel and Kaemmerer
(1978). The original measure contained three subscales: (1) support for cre-
ativity, (2) tolerance of differences, and (3) personal commitment. We did
not use the personal commitment subscale in this study because the con-
struct failed to distinguish between innovative and noninnovative organiza-
tions in the original Siegel and Kaemmerer (1978) study. Further, it seems
likely that commitment is an outcome rather than a dimension of climate.
We examined the published factor structure of the subscales measuring
support for creativity and tolerance for differences. Items relating specifi-
cally to supervisors were not used in order to minimize conceptual overlap
with the LMX measure, thus reducing method variance resulting from com-
mon source. We examined the content of the remaining items to assess how
well they represented dimensions suggested to be important to innovative
performance during the interviews at the facility and selected 16 items, 8
from the support-for-creativity subscale and 8 from the toierance-for-
differences subscale.
We wrote four additional items to assess perceptions of reward-
innovation dependency in this environment. These items tapped the degree
to which rewards were based on innovative performance, and this subscale
overlapped conceptually with performance-reward dependency measures
frequently used in the extant literature (e.g., Pritchard & Karasick, 1973).
Finally, we wrote six items to assess the degree to which respondents be-
lieved resources were adequate for accomplishing the task of innovation.
The response scale for the final 26-item measure was a five-point Likert scale
ranging from "strongly disagree" to "strongly agree."
The data were submitted to a factor analysis using principal compo-
nents extraction and varimax rotation. A four-factor solution resulted, and
our interpretation of the scree plot suggested that only factor 1 be main-
tained. However, an examination of the loadings on each factor indicated
that the resource items primarily loaded on factors 2, 3, and 4, but most of
the items relating to rewards, support for creativity, and tolerance for dif-
592 Academy of Management /ournal June

ferences loaded on factor 1. Although factor 1 accounted for only 33.5 per-
cent of the variance, the other three factors accounted for an additional 18.6
percent, and each had an eigenvalue greater than 1.0. In view of this pattern,
we conducted a second factor analysis, forcing the items to load on two
factors.
In the two-factor solution, two items failed to load over .40 on either
factor, and two items loaded over .40 on both factors. We dropped these four
items from further analysis. Table 1 shows the final results. Factor 1 (16
items) was named support for innovation; it measures the degree to which
individuals viewed the organization as open to change, supportive of new
ideas from members, and tolerant of member diversity. Factor 2 (6 items),
resource suppJy, measures the degree to which resources (i.e., personnel,
funding, time) were perceived as adequate in the organization. We treated
these factors as separate dimensions of the climate for innovation in the
model. Cronbach's alpha for the support for innovation subscale was .92. For
tbe resource supply subscale, it was .77.
Job type was used as a proxy for task type. Interviews with employees
and managers at the facility indicated that the technicians' jobs were more
structured and routine than the jobs of the engineers and scientists and that
the technicians were granted less personal discretion and autonomy in their
work. Tbus, we dummy-coded technicians as 0 and engineers and scientists
as 1, obtaining data from current organizational records.
An individual's career stage was measured in terms of two highly cor-
related, time-based demographic variables, age and R&D tenure, that have
been shown to influence innovative behavior (Mumford & Custafson, 1988).
The data were self-reported by study respondents. Since the correlation
between the two variables was .80, we calculated a score by standardizing
individuals' responses to each of the variables and averaging them. The
higher the score, the further along the individual was in his or her career.
Data on education level, which has also been suggested to be important
to innovation (Mumford & Gustafson, 1988), were obtained by self-report
from respondents and coded as follows: high school, 1; some college, 2;
associate degree, 3; bachelor's degree, 4; master's degree, 5; and Ph.D. de-
gree, 6.
Assessment of Common Method Variance
A number of the subjective measures used in this study were gathered
from the same source in the same questionnaire, which introduced the ques-
tion of common method variance as a potential explanation for the findings.
However, it is common in climate research to assess hoth perceptions of
climate and perceptions of the antecedents of climate in the same question-
naire (e.g., Kozlowski & Doherty, 1989).
Harman's one-factor test (Scbriesheim, 1979) was used to empirically
address the common method variance issue. If common method variance
were a serious problem in tbe study, we would expect a single factor to
emerge from a factor analysis or one general factor to account for most of the
1994 Scott and Bruce 593

TABLE 1
Factor Structure of the Climate for Innovation Measure^
Loadings
Hems Factor 1 Factor 2
1. Creativity is encouraged here. .66 .23
2. Our ability to function creatively is respected by the leadership. .65 ,34
3. Around here, people are allowed to try to solve the same
problems in different ways. .52 .39
4. The main function of memhers in this organization is to follow
orders wbicb come down through channels.'' .73 •Ql
5. Around here, a person can get in a lot of trouble by being
different.'' .89 .18
6. This organization can be described as flexible and continually
adapting to change. .58 .32
7. A person can't do things that are too different around here
without provoking anger.** .88 .28
8. The hest way to get along in tbis organization is to think the way
the rest of the group does.*" .66 .25
9. People around here are expected to deal with problems in the
same way.** .89 .22
10. This organization is open and responsive to change.
11. Tbe people in charge around here usually get credit for others'
m
.53 .03
12. In this organization, we tend to stick to tried and true ways.*" .55 .36
13. Tbis place seems to be more concerned with the status quo than
with change.'' .70 .34
14. Assistance in developing new ideas is readily available. .25 .62
15. There are adequate resources devoted to innovation in this
organization. .18 .70
16. There is adequate time available to pursue creative ideas here. .12 .80
17. Lack of funding to investigate creative ideas is a problem in this
organization.^ .08 .53
18. Personnel shortages inhibit innovation in this organization.'' .10 .65
19. This organization gives me free time to pursue creative ideas
during the workday. M m
20. The reward system here encourages innovation.
21. This organization publicly recognizes those who are innovative. .59 .07
22. The reward system here benefits mainly those who don't rock the
boat.'' .68 .21
Eigenvalue 6.97 3.46
Percentage of variance explained 31.67 15.74
^ Boldface indicates loadings over .40. Associated items were retained in the subscates of
the climate for innovation measure.
'' Item was reverse-coded.

covariance in the independent and criterion variables (Podsakoff & Organ,

1986). We performed a principal components factor analysis on items in the
climate measures and the items in the four subjective independent variable
measures (leader-member exchange, team-member exchange, and intuitive
and systematic problem-solving style), extracting 16 factors, with factor 1
accounting for only 18 percent of the variance. No general factor was appar-
594 Academy of Management Joumoi June

ent in the unrotated factor structure. A varimax rotation failed to converge.

The results of this test offer some indication that common method variance
was not a problem in this study.

RESULTS

Correlations

Table 2 presents the summary statistics, zero-order correlations, and

covariances among the constructs. The bivariate relationships indicate that
all the independent variables were significantly related to innovative behav-
ior with the exception of team-member exchange and resource supply. As
can be seen, the study variables most highly related to innovative behavior
were leader role expectations (r = .33, p < .001) and systematic problem-
solving style (r = - .29, p < .001). The study variable most highly related to
dimensions of climate—support for innovation and resource supply—^was
leader-member exchange (r = .53, p < .001 and r = .33, p < .001, respec-
tively).

Analytic Strategy for Assessing the Model

The analytic strategy used LISREL VI (Joreskog & Sorbom, 1986). Al-
though LISREL VI provides the capability to assess goodness of fit for over-
identified models, it also provides the capability to assess a traditional path
model by estimating paths simultaneously. In the current study, the hypoth-
esized model was "just-identified" since each exogeneous variable was hy-
pothesized to directly influence each endogeneous variable. Because the
model was just-identified, there were no degrees of freedom available with
which to calculate a chi-square goodness-of-fit index. Thus, in the first stage
of testing the model, we used LISREL VI in order to obtain maximum like-
lihood path estimates but not to obtain overall fit measures. In the second
stage, we revised the hypothesized model by removing nonsignificant paths.
The use of this procedure provided additional degrees of freedom so it was
possible to obtain a goodness-of-fit index for the model.
Further, in the present study each latent construct was indicated by only
one manifest variable (either a single variable or a composite measure). How-
ever, unlike many researchers using only single indicators of latent variables
(cf. Fornell, 1983), we did not assume perfect measurement of each variable
or scale. Instead, the diagonal entries in tbe lambda matrix (the loadings
from indicator to latent construct) were calculated as the square root of the
coefficient-alpha internal consistency estimate for each manifest scale, and
the error terms (estimates of random measurement error) were fixed to equal
1.0 minus the value of alpha. This approach draws more traditional methods
of assessing reliability into the structural equation modeling arena and fol-
lows the procedures recommended by Kenny (1979), James, Mulaik, and
Brett (1982), and Williams and Hazer (1986). Netemeyer, Johnston, and Bur-
ton (1990) demonstrated that thus combining indicator variables into com-
1994 Scott and Bruce 595

r - p p O M p C - j p O ' t C ^

M O f-" O CO !-• CO

I I

I I

I I

I I

a.
u

W I I
CO

O O O O T-" O • -

p. .ii

.13 to
_c I-"

M c
(13
•1
co

3-5 a
fe ^ S ><
o 3 3 iu g OJ -t^ 5 3
H ^ fl*i y*^ n fli ^ U
j==^ H «
596 Academy of Managemenl Journal Jqne

posite scales led to path estimates that were virtually identical to the esti-
mates generated by using multiple single-variable indicators.
Four measures in the current study were single-item measures. For the
measure of role expectations, the reliability was set at .85, consistent with
the 14-month test-retest correlation (r ^ .87) reported previously. The mea-
sure of career stage, a composite of age and tenure, and the measure of
education were set at .90 and .85, respectively. Finally, the reliability of the
task-type measure was set at .90.
The exogeneous variables were allowed to covary in the estimation of
the model. That is, we assumed that relationships existed among leader-
member exchange, leader role expectations, team-member exchange, intu-
itive problem-solving style, and systematic problem-solving style.
Evaluating the Hypothesized Model
Table 3 presents the structural parameter estimates for the hypothesized
model. Figure 2 presents the final model with nonsignificant paths removed.
For the equation predicting innovative behavior, all but two hypothesized
parameters were significant. These were the paths from intuitive problem-
solving style to innovative behavior and from team-member exchange to
innovative behavior. There were significant paths between innovative be-
havior and each of the other predictors—leader-member exchange, role ex-
pectations, systematic problem-solving style, support for innovation, and
resource supply—with the covariates entered in the model.
In terms of goodness-of-fit indicators, the model accounted for 37 per-
cent of the variance in innovative behavior. Furthermore, leader-member
exchange accounted for 39 percent of the variance in support for innovation,
and that variable and intuitive problem-solving style accounted for 29 per-
cent of the variance in resource supply. Further, the assessment of tbe good-
ness of fit of the revised model (Figure 2) revealed a quite good fit to the data
(X^ = 23.99, df = 16, p = .462). The following values of additional fit
indexes also indicated a good fit: goodness-of-fit index, .98, adjusted good-
ness-of-fit index, .94, and root-mean-square residual, .036.
For the equations predicting the climate dimensions, support for inno-
vation and resource supply, the structural path from leader-member ex-
cbange was significant in both cases. In addition, there was a significant
relationship between intuitive problem-solving style and resource supply.
No support was found for the relationships between role expectations and
either climate dimension or for the relationship between team-member ex-
change and either climate dimension. Finally, there was a significant rela-
tionship between the unaccounted variances of the two climate measures
(^2,3 = 421, s.e. = .074, p < .001), indicating that some unmeasured vari-
able or set of variables similarly influences perceptions of both climate di-
mensions.
In examining the direction of the significant parameters, we found one
relationship that was contrary to hypothesis. Hypothesis 1 predicts a posi-
tive relationship between the dimensions of the perceived climate for inno-
1994 Scott and Bruce 597

TABLE 3
Standardized Path Estimates
Dependent Standardized
Variables Paths Path Estimates
Innovative
behavior Support—* Innovative behavior 30" .14
Resource supply—»Innovative behavior 31* .15
Leader-member exchange—* Innovative behavior 20* .12
Role expectations—* Innovative behavior 28"' .08
Team-member exchange^ Innovative behavior 04 .09
Systematic problem-solving style —• Innovative behavior 33**'' .09
Intuitive problem-solving style —• Innovative behavior 03 .09
Job type—* Innovative behavior 07 .14
Career stage -» Innovative behavior 19* .10
Education —• Innovative behavior 06 .13
Support for
innovation Leader-member exchange—* Support 61"* .08
Role expectations —• Support 08 .11
Team-member exchange -> Support 06 .09
Systematic problem-solving style—* Support 05 .07
Intuitive problem-solving style—* Support 13 .07
Job type —• Support 17 .11
Career stage —* Support 06 .08
Education-* Support 03 .11
Resource
supply Leader-member exchange —• Resources .45*** .09
Role expectations —* Resources 09 .09
Team-member exchange —* Resources 02 .10
Systematic problem-solving style —* Resources 06 .09
Intuitive problem-solving style—* Resources 25'* .11
Joh type —• Resources ,17 .12
Career stage —* Resources ,21* .10
Education —» Resources ,11 .14

* p < .05
*• p < .01
• " p < .001

vation and innovative behavior. The structural path from support for inno-
vation to innovative behavior offered support for this hypothesis. However,
the structural path from resource supply to innovative behavior was nega-
tive. The high correlation between the two climate dimensions [r = .59, p <
.001). accompanied by the significant correlation between innovative bebav-
ior and support for innovation (r ^ .15, p < .01) and tbe lack of a significant
correlation between resource supply and innovative behavior (r = —.02,
n.s.), suggest suppression was occurring (cf. Cohen & Cohen, 1983; Lord &
Novick, 1968). It is likely that resource supply was suppressing some of the
variance in support for innovation that was irrelevant to innovative behav-
ior. When this error variance was partialed out, or suppressed, the remaining
variance in support for innovation was more strongly related to innovative
598 Academy of Monagement Journal June

FIGURE 2
A Path Model of Individual Innovative Behavior'

Leader Role .28

Expectations

Leader-Member
Excbange

Intuitive
Problem-Solving
Style

Career Stage

-.33
Systematic
Problem-Solving
Style

' For clarity, only significant paths are sbown

behavior. Our interpretation of the pattern found here is that resource supply
was a suppressor variable and was not significantly related to innovative
behavior.
In summary, results partially supported Hypothesis 1 in that support for
innovation was positively related to innovative behavior and resource sup-
ply was not. The significant, positive paths between leader-member ex-
change and innovative behavior, between leader-member exchange and each
of the climate dimensions, and between role expectations and innovative
behavior fully supported Hypotheses 2, 3, and 4, respectively. The signifi-
cant, negative path between systematic problem-solving style and innova-
tive behavior supported Hypothesis 8b. No support was found for Hypoth-
esis 5 (the relationship between role expectations and climate dimensions).
Hypothesis 6 (the relationship between team-member exchange and inno-
vative behavior). Hypothesis 7 {the relationship between team-member ex-
change and climate dimensions), and Hypothesis 8a (the relationship be-
tween intuitive problem-solving style and innovative behavior).
1994 Scott and Bruce 599

The Moderating Effect of Task Type

A single hierarchical moderated regression analysis would typically be
used to test for interactive effects between task type and each of the six
independent study variables in predicting innovative behavior. However,
the inclusion of six multiplicative interactions used substantial degrees of
freedom, resulting in relatively low power with wbich to test for moderator
effects. Therefore, we conducted a series of six hierarchical regressions,
testing each interaction term in a separate analysis. In each analysis, we
entered the covariates, education and career stage, and a single independent
variable in step 1. In step 2, the interaction term, job type by the independent
variable, was entered. To determine whether moderation was occurring, we
tested the change in R^ from step 1 to step 2 for significance. The results
showed that task type moderated only the relationship between role expec-
tations and innovative bebavior (AF4 i^s " 3.84, p < .05).
To further explore the interaction between task type and innovative
behavior, we performed additional analyses by dividing the data set into two
subgroups, one composed of engineers and scientists (N = 108) and tbe
otber composed of technicians (N = 64). All of the study predictors were
then regressed on innovative behavior for each subgroup. Table 4 shows
results. Tests of tbe significance of the difference between tbe betas in the
two independent subgroups were conducted. Of the six study variables, only
the beta for role expectations was significantly different between the two
groups. Although the innovative behavior of the technicians was positively
related to managers' role expectations (p = .37, p < .001), this relationship
was nonsignificant for the engineers and scientists (P = .05, n.s.).

TABLE 4
Results of Regression Analyses for Subgroups'
Job Type
Variables Technicians Engineers and Scientists
Support for innovation .26 .26*
Resource supply -.30 -.13
Leader-member exchange -.07 .23*
Role expectations'" .37-** .05
Team-member exchange .17 .02
Systematic problem-solving style -.31" - .29'*
Intuitive problem-solving style -.06 -.05
Career stage -.10 -A7
Education -.10 M
Adjusted R^ .24 .23
F 3.90** 3.67**
° Column entries are standardized regression coefficients. N = 64. technicians; N = 108,
engineers and scientists.
'' There is a significant difference in the betas between the two subgroups for the variable.
* p < .05; **p < .01; ***p < .001.
600 Academy of Management /ournal June

DISCUSSION

In this study, we developed and tested a model in vi^hich leadership,

work group relations, and individual attributes were hypothesized to affect
individual innovative behavior directly and indirectly, through climate per-
ceptions. We found leadership, support for innovation, managerial role ex-
pectations, career stage, and systematic problem-solving style to be signifi-
cantly related to individual innovative behavior, and the hypothesized
model explained almost 37 percent of the variance in innovative behavior.
Although our cross-sectional study design precluded making inferences
about causality, this is a substantial finding given the limited theory devel-
opment in this area to date.
The study provides evidence that innovative behavior is related to the
quality of the supervisor-subordinate relationship, as Basu (1991) reported
in a recent study of blue-collar workers. High-quality dyadic relationships
may give subordinates the levels of autonomy and discretion necessary for
innovation to emerge (Graen & Scandura, 1987). In addition, the findings
support the hypothesis that individuals generalize their supervisor-
subordinate relationships to their organizations. In this study, subordinates
who reported having relationships with their supervisors characterized by
high levels of support, trust, and autonomy also reported the organization to
be supportive of innovation and judged the resource supply to be high. The
positive relationship found here between leader-member exchange and cli-
mate perceptions replicates prior work by Kozlowski and Doherty (1989)
and others, and it does so in the context of innovative climate, fulfilling
Schneider's (1975) charge that climate studies be anchored within a specific
domain of inquiry.
We also found that the role expectations of a supervisor influenced
individual innovative behavior, providing support for the Pygmalion effect
(Livingston, 1969) within the context of innovation. However, this effect was
only operative for the technicians in this study. The innovative behavior of
the engineers and scientists was not affected by their managers' role expec-
tations. One explanation for the engineers' and scientists' apparent lack of
receptivity to leader role expectations may be their high levels of education
and high independence or their having status equality with the managers in
this setting (cf. Bass, 1985; Jussim, 1986).
The significant relationship between role expectations and innovative
behavior could be the result of common source bias as the managers pro-
vided assessments of both variables. However, if method variance resulting
from common source and format were a problem, we would expect to find
the two variables significantly related to each other in both the engineer-
scientist and the technician subgroups. The different patterns of relationship
between the subgroups provides some evidence that there was no substantial
common source or format similarity bias.
Considering individual attributes, the findings suggest that individuals
1994 Scolf and Bruce 601

do not need to be highly intuitive problem solvers to be innovative, but being

systematic problem solvers appears to inhibit high levels of innovative be-
havior. As hypothesized, systematic problem-solving style had a direct neg-
ative effect on innovative hehavior. This was true botb for those engaged in
routine tasks (the technicians] and those engaged in nonroutine tasks (the
engineers and scientists).
We treated the systematic and intuitive problem-solving styles indepen-
dently, as prior theory suggests (Jabri, 1991). However, it is likely that the
same individuals use systematic and intuitive problem solving at different
times and on different tasks. Perhaps the true innovators are people who can
use a style that is appropriate to the stage of the innovation cycle in which
they are involved. Further study on the implications of the various combi-
nations of these two styles on innovative behavior is needed.
Several findings of the current study were contrary to hypothesis and
deserve comment. First, although psychological climate perceptions of sup-
port for innovation were positively related to innovative behavior, the coef-
ficient between resource supply and innovative behavior was negative.
Given that the zero-order correlation between these two was nonsignificant,
it appears that a suppression effect was operating and that there was no
relationship between resource supply and innovative behavior. This finding
is surprising in that resources have previously been theorized to be critical
to innovation (e.g., Lawrence & Dyer, 1983).
One explanation for this finding emerges from the form of the relation-
ships between the climate dimensions and innovative behavior and the na-
ture of the variables themselves. In the case of resources, a threshold effect
may operate, whereby perceived increases above some point have no further
effect in facilitating innovative behavior. Since the data in this study were
from an R&D laboratory with the espoused mission of innovation, resource
levels are likely to have been consistently above sucb a threshold. Thus, no
relationship was found between resources and innovative behavior in this
study. Cross-organizational research at the individual level is needed to
broaden the range of the resource supply variable and test for threshold
effects.
In contrast to the resource supply variable, support for innovation was
positively related to innovative behavior. If a threshold effect exists for the
support variable, this organization may have operated below the threshold
level and thus, a positive linear relationship was found. Alternatively, it may
be that the support variable is very different from the resource variable.
Support for innovation, as defined here, measures abstract c o n c e p t s -
flexibility, encouragement, tolerance for change—that may, in fact, be lin-
early related to behavior across the entire range. In other words, more sup-
port may always be better than less. Further, it is likely that the dimensions
of climate exist on a hierarchy of need. Support for innovation may only
influence innovative behavior once the need for some threshold level of
resources is met.
Finally, in this study innovative climate perceptions only mediated be-
602 Academy of Management Journal

tween leader-member exchange and innovative behavior. Although the find-

ings relative to leader-member exchange have significance for management
practice, the general lack of support for climate as a mediator in the deter-
mination of innovative behavior was surprising. The role of climate as a
mediator may be overstated in the literature, at least as it relates to innova-
tive behavior. We were not able to identify individual or work group cbar-
acteristics that engendered differences in climate perceptions in this study;
neither problem-solving style nor team-member exchange were related to
innovative climate perceptions here. Indeed, the findings suggest that it
would be more effective in future research to search for additional direct
antecedents to innovative bebavior rather than antecedents whose effects are
mediated by innovative climate perceptions.
Surprisingly, team-member exchange was not related to innovative be-
havior or to climate perceptions in this study. A possible explanation for
these seemingly implausible findings is that intragroup task interdepen-
dence may mediate the relationship between team-member exchange and
botb affective and behavioral responses. Where task interdependence and
work-group-member interaction is low, the relationship between measures
of work group cooperation and collaboration is likely to be weaker than it
will be when task interdependence and member interaction is high; indeed,
the relationship may be nonexistent. Future studies of work group effects on
innovative behavior sbould include measures of task interdependence.
The findings in this study are subject to a number of caveats. First, the
cross-sectional research design limits the ability to determine causation. In
fact, previous researchers have conceptualized the relationship between
leader-member exchange and performance outcomes and between leader
role expectations and performance outcomes as reciprocal [Eden, 1984;
Graen & Scandura, 1987). Future studies should include longitudinal de-
signs to further explore the reciprocality of many of the relationships posited
in this study.
The second caveat concerns the generalizability of these findings to
other types of work organizations and to other functional areas of organiza-
tions. Although a number of the hypothesized relationships replicate the
work of others in very different settings, the test of the overall model should
be replicated in other settings. This study focused on self-managing profes-
sionals in R&D; future studies should examine a broader range of tasks and
technologies as potential boundary conditions of the model.
Monomethod bias was minimized as much as possible in this study, but
it must be noted that it remains an issue. Although we took care to assure
that items in the measures of the predictors of climate tapped separate con-
ceptual domains, all the measures used similar response formats and were
completed by the same sources. However, we generally avoided the same-
source responses in assessing the determinants of innovative behavior. The
criterion was a supervisory assessment of innovative behavior, but measures
of the independent variables were assessed by subordinates or from ar-
chives, with the exception of role expectations.
1994 Scott and Bruce 603

The current study provides a first attempt at modeling a complex phe-

nomenon—individual innovative behavior. There has been no scarcity in
the literature of suggested antecedents to individual creativity and innova-
tion; we attempted to integrate some of these disaggregated findings into a
cohesive whole and tested them in the natural work context of an R&D
facility. Our study thus provides some understanding of the complex rela-
tionships among and causal paths between a number of antecedents often
cited in previous work. Although the findings reported here provide some
guidance to practicing managers, they also pose a whole new set of questions
for researchers. What is the joint effect of systematic and intuitive problem-
solving styles? Is there a threshold level at which additional resources no
longer improve innovative behavior? Is this threshold effect common to all
types of resources—time as well as equipment and facilities? Will addi-
tional support for innovation continue to improve innovative behavior, and
to what level? How do task type and technology moderate the relationships
reported here? The answers to these questions await further study.
Studying individual innovative behavior in a natural work context is a
complex and difficult task because the criterion is often difficult to validate,
and researchers are often limited to the use of perceptual measures. But as
organizations face increasingly turbulent environments and innovation be-
comes part of every employees' job description, the need for this kind of
research is ever increasing.

REFERENCES
Abbey, A., & Dickson, ]. 1983. R&D work climate and innovation in semiconductors. Academy
of Management Journal, 26: 362-36B.
Amabile. T. 1988. A model of creativity and innovation in organizations. In B. M. Staw & L. L.
Cummings [Eds.]. Researcb in organizational bebavior, vol. 10:123-167. Greenwicb.CT:
JAI Press.
Amabile. T., & Gryskiewicz, S. 1987. Creativity in the R&D iaboratory. Tecbnical report no. 30,
Center for Creative Leadersbip, Greensboro, NC.
Amabile, T., & Gryskiewicz. N. 1989. The creative environment scales; The Work Environment
Inventory. Creativity Researcb foumal, 2: 231-254.
Ancona. D,, & Caldwell, D. 1987. Management issues facing new product teani,s in bigb tecb-
nology companies. In D. Lewin, D. Lipsky, & D. Sokel (Eds.), Advances in industrial and
labor relations, vol. 4: 191-221. Greenwicb, CT: JAI Press.
Angle, H. 1989. Psychology and organizational innovation. In A. Van de Ven. H, Angle, & M.
Poole (Eds.), Researcb on tbe management of innovation: Tbe Minnesota studies: 135-
170. New York: Harper & Row.
Asbfortb, B. 1985. Glimate formation; Issues and extensions. Academy of Management Re-
view, 10; 837-847.
Bandura, A. 1988. Self-regulation of motivation and action tbrough goal systems. In V. Hamil-
ton, F. H. Bower, & N. H. Frijda (Eds.), Cognitive perspectives on emotion and motivation:
37-61. Dordrecht, Netherlands: Lkuwer Academic Publishers.
Barron, F.. & Harrington, D. 1981. Creativity, intelligence, and personality. In M. R. Rosenzweig
& L. W, Porter (Eds.), AMWual review of psycbology, 32: 439-476. Palo Alto, CA; Annuai
Reviews.
604 Acodemy of Management journal June

Bass, B. 1985. Leadership: Good, hetter, best. Organizational Dynamics, 14(3]: 26-40.
Basu. R. 1991. An empirical examination of leader-member exchange and transformational
leadership as predictors of innovative behavior. Unpublished doctoral dissertation, Pur-
due University, West Lafayette, IN.
Cohen, J., & Cohen, P. 1983. Applied multiple regression/correlation analysis for the behav-
ioral sciences. Hillsdale, NJ: Erlhaum.
Cotgrove. S., & Box, S. 1970. Science, industry, and society: Studies in the sociology of
science. London: George Allen & Unwin.
Dansereau, F., Graen, G., & Haga, W. 1975. A vertical dyad linkage approach to leadership
within formal organizations: A longitudinal investigation of the role-making process.
Organizational Rehavior and Human Performance, 13: 46-78.
Dunegan, K,, Tiemey, P., & Duchon, D. 1992. Toward an understanding of innovative climate:
Explaining variance in perceptions by divisional affiliation, work group interactions, and
subordinate-manager exchanges. IEEE Transactions on Engineering Management, 39
(special issue): 227-236.
Eden, D. 1984. Self-fulfilling prophesy as a management tool: Harnessing Pygmalion. Academy
of Management Review, 91: 64-73.
EJter, M. 1991. The influence of cognitive schemas on perceptions of organizational climate.
Paper presented at the 51st Annual Meeting of the Academy of Management, Miami.
Festinger, L. 1954. A theory of social comparison processes. Human Relations, 7: 114-140.
Fornell, C. 1983. Issues in the application of covariance structure analysis: A comment. Journal
of Consumer Research, 9: 443-448.
Glick, W. 1985. Conceptualizing and measuring organizational and psychological climate: Pit-
falls in multilevel research. Academy of Management Review, 10: 601 -616.
Glick, W. 1988. Response: Organizations are not central tendencies: Shadowboxing in the dark,
round 2. Academy of Management Review, 13: 133-137.
Graen, G., & Cashman, J. 1975. A role-making model of leadership in formal organizations: A
developmental approach. In }. Hunt & L. Larson (Eds.), Leadership frontiers: 309-357.
Kent. OH: Kent State University Press.
Graen, G., Novak, M., & Sommerkamp, P. 1982. The effects of leader-member exchange and job
design on productivity and job satisfaction: Testing a dual attachment model. Organiza-
tional Rehavior and Human Performance, 30: 109-131-
Graen. G., & Scandura. T. 1987. Toward a psychology of dyadic organizing. In L. L. Cummings
& B. M. Staw (Eds.}, Research in organizational behavior, vol. 9: 175-208. Greenwich,
CT: JAI Press.
Isaksen, S. 1987. An orientation to the frontiers of creativity research. In S. Isaksen (Ed.),
Frontiers of creativity research: Reyond the basics: 1-31. Buffalo: Bearly Limited.
Isaksen, S., & Kaufmann, G. 1990. Adaptors and innovators: Different perspectives of the psy-
chological climate for creativity. Studia Psychologica, 32: 129-140.
Jabri, M. 1991. The development of conceptually independent subscales in the measurement of
modes of problem solving. Educational and Psychological Measurement, 51: 975-983.
James, L., Hater, J., Gent, M., & Bruni. J. 1978. Psychological climate: Implications from cogni-
tive social learning theory and interactional psychology. Personnel Psychology, 31: 783-
813.
James. L., Hartman, E., Stebbins. M.. & Jones, A. 1977. An examination of the relationship
between psychological climate and a VIE model for work motivation. Personnel Psychol-
ogy, 30: 229-254.
1994 Scott and Bruce 605

James, L., James, L., & Ashe, D. 1990. The meaning of organizations: The role of cognition and
values. In B, Schneider [Ed.j, Organizatianal climate and culture: 40-84. San Francisco:
Jossey-Bass.
James, L., Mulaik, S., & Brett, J. 1982. Causa/ analysis: Assumptions, models, and data. Bev-
erly Hills: Sage.
James, L., & Sells, S. 1981. Psychological climate: Tbeoretical perspectives and empirical re-
searcb. In D. Magnussen (Ed,), Toward a psychology of situations: An interactional per-
spective: 275-295. Hillsdale. NJ: Erlbaum.
Jones, A., & James, L. 1979. Psychological climate: Dimensions and relationships of individual
and aggregated work environment perceptions. Organizational Behavior and Human Per-
formance, 23: 210-233.
Joreskog, K., & Sorbom, D. 1986, LISREL VI: Analysis of linear structural relationships hy
maximum likelihood, instrumental variables, and least squares methods. Department of
Statistics, University of Uppsala, Uppsala, Sweden,
Jussim, L, 1986. Self-fulfilling propbesies: A tbeoretical and integrative review. Psychological
RevJeiv, 93: 429-445.
Kanter, R. 1983. The change masters. New York; Simon & Schuster.
Kanter. R, 1988, Wben a thousand flowers bloom: Structural, collective, and social conditions
for innovation in organizations. In B. M. Staw & L. L. Cummings (Eds,), Research in organ-
izational behavior, vol 10: 169-211. Greenwich, CT: JAI Press.
Kenny, D. 1979. Correlation and causality. New York: Wiley-Interscience.
Kirton. M. 1976. Adaptors and innovators: A description and measure. Journal of Applied
Psychology. 61: 622-629.
Koestler. A. 1964. The act of creation. London: Hutchinson.
Kozlowski, S,. & Doberty, M. 1989, Integration of climate and leadersbip: Examination of a
neglected topic. Journal of Applied Psychology, 74: 546-553,
Lawrence, P., & Dyer, P, 1983. Renewing American industry. New York: Free Press.
Lewin. K. 1938. The conceptual representation of the measurement of psychological farces.
Durham, NC: Duke University Press.
Livingston, J. 1969. Pygmalion in management. Harvard Business Review, 47(4): 81-89.
Lord. R,. & Novick. M. 1968, Statistical theories of mental test scores. Reading, MA: Addison-
Wesley.
Middlemist. R.. & Hitt, M. 1981. Tecbnology as a moderator of the relationship between per-
ceived work environment and subunit effectiveness. Human Relations, 34: 517-532,
Mumford. M,, & Gustafson. S. 1988. Creativity syndrome: Integration, application, and innova-
tion. Psychological Bulletin, 103: 27-43.
Netemeyer, R,. Jobnston. M,, & Burton, S. 1990. Analysis of role conflict and role ambiguity in
a structural equations framework. Journal of Applied Psychology, 75: 148-157.
Paolillo. J.. & Brown, W, 1978. How organizational factors affect R&D innovation. Research
Management, 21: 12-15.
Payne. R.. Lane. D,, & Jabri, M. 1990. A two dimensional person-environment fit analysis of the
performance, effort, and satisfaction of researcb scientists. British Journal of Management,
1: 45-57.
Pelz, D., & Andrews. F. 1966. Autonomy, coordination, and stimulation in relation to scientific
achievement. Behavioral Science, 11: 89-97.
Podsakoff. P.. & Organ, D. 1986. Self-reports in organizational researcb: Problems and prospects.
Journal of Management, 12: 531-544.
606 Academy 0/ Management Journoi June

Pritchard, R.. & Karasick, B. 1973. The effects of organizational climate on managerial job
performance and satisfaction. Organizational Bebavior and Human Performance, 9:
126-146.
Rogers, C. 1954. Toward a theory of creativity. fleWeiv of Genera! Semantics, 2: 249-260.
Schneider, B. 1975. Organizational climates: An essay. Personnel Psychology, 28: 447-479.
Schneider, B., & Reichers, A. 1983, On the etiology of climates, Personnel Psycbology, 36:
19-39.
Schriesbeim, C. 1979. The similarity of individual directed and group directed leader bebavior
descriptions. Academy of Management foumal, 22; 345-355.
Scbroeder, R., Van de Ven, A., Scudder, G., & Polley, D. 1989. Tbe development of innovation
ideas. In A. Van de Ven, H. Angle, & M. Poole (Eds.), Research on tbe management of
innovation: The Minnesota studies: 107-134. New York; Harper & Row.
Seers. A. 1989. Team-member excbange quality; A new construct for role-making researcb.
Organizational Bebavior and Human Decision Processes, 43: 118-135.
Sethia, N. 1991. Tbe evocation of creativity tbrough collaboration. Paper presented at tbe
annual meeting of tbe Academy of Management. Miami.
Siegel, S.. & Kaemmerer, W. 1978. Measuring tbe perceived support for innovation in organ-
izations, foumal of Applied Psychology, 63: 553-562.
Taylor, C. 1963. Variables related to creativity and productivity in men in two researcb labo-
ratories. In C. Taylor & R. Barron (Eds.), Scientific creativity: Its recognition and devel-
opment: 513-597. New York: Wiley.
Van de Ven. A. 1966. Central problems in the management of innovation. Management Sci-
ence, 32: 590-607.
Waldman, D., & Bass, B. 1991. Transformational leadersbip at different pbases of the innovation
process, foumal of High Technology Management Research, 2: 169-180.
Walton, R. 1985. Transformation of workforce strategies in tbe United States. In K. Clark, R.
Hayes, & C. Lorenz (Eds.), The uneasy alliance: Managing the productivity'technology
dilemma: 237-265. Boston; Harvard Business School.
West, M., & Farr, J. 1989. Innovation at work: Psychological perspectives. Social Behavior, 4;
15-30.
West, M., & Farr, |, 1990. Innovation at work. In M. West & J. Farr (Eds.). Innovation and
creativity at work: Psychological and organizational strategies: 3-13. New York: Wiley.
Williams, L., & Hazer, J. 1986. Antecedents and consequences of satisfaction and commitment
in turnover models; A reanalysis using latent variable structural equation metbods. foumal
of Applied Psycbology, 71: 219-231.
Woodman. R., Sawyer, J., & Griffin, R. 1993. Toward a tbeory of organizational creativity.
Academy of Management Review, 18; 293-321.

APPENDIX
Innovative Behavior Measure
Instructions to respondents were as follows:
"Innovation is a process involving both tbe generation and implementation of ideas. As
sucb, it requires a wide variety of specific bebaviors on tbe part of individuals. While some
people might be expected to exhibit all tbe behaviors involved in innovation, others may exhibit
only one or a few types of behavior. Please rate eacb of your subordinates on tbe extent to wbich
he or sbe:
1994 Scott and Bruce 607

1. Searches out new tecbnologies, processes, techniques, and/or product ideas.

2. Generates creative ideas.
3, Promotes and champions ideas to others,
4, Investigates and secures funds needed to implement new ideas.
5, Develops adequate plans and schedules for tbe implementation of new ideas.
6. Is innovative."

Susanne G. Scott received her Pb,D, degree in management from tbe University of
Cincinnati. She is an assistant professor of management at the University of Colorado
at Colorado Springs. Her current research interests include processes of change and
innovation at tbe individual, group, and organizational levels, empowerment, team-
work, workforce diversity, and decision making.
Reginald A. Bruce received his Pb.D. degree in organizational psychology from tbe
University of Michigan, He is an assistant professor of management in the College of
Business and Public Administration at the University of Louisville. His current re-
searcb interests include individual adjustment to role and career changes, the transition
from unemployment to employment, decision-making styles, and innovation.