Measuring Climate for Work Group Innovation: Development and Validation of the Team

Climate Inventory
Author(s): Neil R. Anderson and Michael A. West
Source: Journal of Organizational Behavior, Vol. 19, No. 3 (May, 1998), pp. 235-258
Published by: John Wiley & Sons
Stable URL: http://www.jstor.org/stable/3100170
Accessed: 07/06/2010 10:09
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=jwiley.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

John Wiley & Sons is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Organizational Behavior.

http://www.jstor.org

JOURNALOF ORGANIZATIONALBEHAVIOR,VOL. 19, 235-258 (1998)

Measuringclimatefor workgroupinnovation:
developmentand validationof the team
climateinventory
NEIL R. ANDERSONl* AND MICHAEL A. WEST2
Psychology Department, Goldsmiths College, Universityof London, London SE14 6NW, U.K.
2Institute of Work Psychology, Universityof Sheffield, Sheffield S10 2TN, U.K.

Summary

This paper reports the development and psychometric validation of a multi-dimensional

measure of facet-specific climate for innovation within groups at work: the Team
Climate Inventory (TCI). Brief reviews of the organizational climate and work group
innovation literatures are presented initially, and the need for measures of facet-specific
climate at the level of the proximal work group asserted. The four-factor theory of
facet-specific climate for innovation, which was derived from these reviews, is described,
and the procedures used to operationalize this model into the original version measure
described. Data attesting to underlying factor structure,internal homogeneity, predictive
validity and factor replicability across groups of the summarized measure are presented.
An initial sample of 155 individuals from 27 hospital management teams provided data
for the exploratory factor analysis of this measure. Responses from 121 further groups in
four occupations (35 primary health care teams, 42 social services teams, 20 psychiatric
teams and 24 oil company teams; total N = 971) were used to apply confirmatory factor
analysis techniques. This five-factor, 38-item summarized version demonstrates robust
psychometric properties, with acceptable levels of reliability and validity. Potential
applications of this measure are described and the implication of these findings for
the measurement of proximal work group climate are discussed. ? 1998 John Wiley &
Sons, Ltd.
J. Organiz. Behav. 19: 235-258 (1998)

Climate:DefinitionalIssues
The concept of climate has received considerable attention from applied psychologists and
organizational sociologists over the last three decades. Numerous empirical studies have been
conducted and regular reviews of the research findings have appeared (the major reviews are
* Addresseefor
Neil Anderson,GoldsmithsCollege, Universityof London, New Cross, London
correspondence:
SE146NW, U.K.
Wewishto expressour thanksto EdwardBuckandJanJacksonfor theirassistancein analyzingsectionsof thisdataset.
Ourthanksalso to BrendaPoulton,StephenKellett,MichaelSteadand TraceyHepplestonall of whomassistedwith
datacollection.WewouldacknowledgePaulJackson,TobyWallandPeterWarrfor theirhelpfulcommentson an earlier
draftof this paper.
CCC 0894-3796/98/030235-24$17.50
? 1998 John Wiley & Sons, Ltd.

Received 18 July 1995

Accepted 26 June 1996

236

N. R. ANDERSON AND M. A. WEST

Campbell, Dunnette, Lawler and Weick, 1970; Joyce and Slocum, 1984; Rentsch, 1990;
Rousseau, 1988; Schneider and Reichers, 1983; Schneider, 1990). Despite this growth in research
interest, climate researchhas been affected by two intractable and related difficulties:defining the
notion of climate, and measuring climate accurately at different levels of analysis.
Many definitions of climate have been put forward, but two approaches in particular
have received substantial patronage; the cognitive schema approach and the shared perceptions
approach. The former conceptualizes climate as individuals' constructive representations or
cognitive schema of their work environments, and has been operationalized principally
through attempts to uncover individuals' sense-making of their proximal work environment
(e.g. Ashforth, 1985; James and Jones, 1974; James and Sells, 1981; Schneider and Reichers,
1983). For instance, James and Sells (1981) define climate as 'individuals' cognitive representations of proximal environments ... expressed in terms of psychological meaning and significance
to the individual' (p. 276). Superordinate to this focus at the individual level, other authors have
emphasized the importance of shared perceptions as underpinning the notion of climate
(e.g. Koys and DeCottis, 1991; Payne, Fineman and Wall, 1976; Uttal, 1983). Thus, Reichers and
Schneider (1990) define organizational climate as '. .. the shared perception of the way things are
around here. More precisely, climate is shared perceptions of organizational policies, practices,
and procedures' (p. 22). The difficulty faced by researchersadopting this approach has been to
attain consensus over criteria for minimum levels of agreement sufficient to indicate that
perceptions are truly shared amongst members of an organization or organizational subunit
(Guion, 1973;Jackofsky and Slocum, 1988; Joyce and Slocum, 1984; Patterson, West and Payne,
1992; Payne, 1990). The cognitiveschema and the sharedperceptionsapproaches are, in principle,
compatible with one another and are thus not mutually exclusive. In this study we adopt the latter
approach, applying the concept of shared perceptions to the work group level of analysis to
develop a measure of proximal work group climate.
Attention has been directed by researchers adopting the shared perceptions approach in
defining cut-offs for consensus and interrateragreement between individuals' ratings of climate
dimensions indicative of shared perceptions of climate (Dansereau and Alutto, 1990). Despite
this increasing focus upon shared perceptions, there has been a paucity of research which has
addressed the work group or team as a level of analysis distinct from the wider organization or
the individual. Indeed, the concept of 'organization climate' has entered everyday usage, yet the
diversity and sheer size of many organizations would suggest a more micro-analytical examination of sharedness at the level of the work group, team or subunit, is warranted.
In this paper we argue that an appropriate level of analysis at which to examine shared
perceptions of climate is the proximal work group. Having established this focus, we describe the
development of a measure of proximal work group climate designed to elicit team members'
perceptions of climatic dimensions hypothesized to relate to one facet of group output-work
group innovation.

ProximalWorkGroupClimate
We define the proximal work group as either the permanent or semi-permanent team to which
individuals are assigned, whom they identify with, and whom they interact with regularly in
order to perform work-related tasks. Individuals are likely to identify most closely with their
proximal work group and to have commitment to its ongoing social structure, although our
? 1998John Wiley& Sons, Ltd.

J. Organiz.Behav. 19:235-258 (1998)

CLIMATEFOR WORKGROUPINNOVATION

237

conceptualization precludes purely social cliques in the workplace since the prerequisite of taskinterdependence is held to be an essential element of our definition. Of course, individuals will
commonly be members of more than one group at work, but our attention is specifically upon the
group within which their day-to-day tasks and activities at work are carried out. Further, we
assert that the proximal work group represents the primary medium through which shared
climates will evolve through active social construction and become embedded into the fabric of
the organization (Hosking and Anderson, 1992; West, 1995).
For sharedness to even be a possibility, we would argue, demands (a) that individuals interact
at work, at least on an infrequent basis, (b) that there exists some common goal or attainable
outcome which predisposes individuals toward collective action, and (c) that there is sufficient
task interdependence such that individuals need to develop shared understandings and expected
patterns of behaviour (see for instance, West, 1995). Importantly, these three criteria represent
necessary but not sufficient conditions for shared climate to exist; even if all are present in a
group, it does not necessarily follow that a shared climate will evolve (Schneider and Reichers,
1983). Individuals who identify with their proximal work group and who interact with colleagues
are likely to develop shared patterns of understanding and norms of behaviour, thereby allowing
the opportunity for a shared climate to evolve (Campion, Medsker and Higgs, 1993). Of course,
shared climates may also evolve through other means. Individuals may progress through similar
socialization processes and their common experiences lead to shared perceptions. Individuals at
different levels in the hierarchy may similarly be exposed to common experiences which lead to
shared perceptions of climate. The organizational elite may communicate downwards a common
message as to its vision, culture and strategies so powerfully that employees develop shared
perceptions to some extent. Thus, sharedness can arise at a variety of strata in the organization

and be influencedby a rangeof factors.Ourpoint is simplythat it is most likelythat sharedness

will evolve where individuals have the opportunity to interact and to co-construct perceptions
within their proximal or immediate work environment-their proximal work group (e.g. Hosking
and Anderson, 1992). Importantly, then, it becomes necessary to be able to measure such shared
climates within groups, and this indeed was the main objective of the current study-to develop
and validate a facet-specific measure of proximal work group climate.
It is unlikely, then, that shared climates exist at the overarching level of the organization in its
entirety, particularly where the organization is large, divisionalized and multilayeredin its formal

structure(seealso DansereauandAlutto, 1990).Wewouldarguethatit is morejustifiedto search

for shared climates within identifiable groups, teams, cliques and cohorts within an organization,
where each of the three criteria outlined above may conceivably be met. As agreed previously,
individuals are likely to identify with their proximal work group, and moreover, shared patterns
of understanding and norms of behaviour are most likely to develop at this level, allowing the
opportunity for a shared climate to evolve (Campion et al., 1993). For these reasons, this article
argues for the utility of the concept of proximal work group climate as an appropriate level of
focus and for the need to develop measures specifically at the group level of analysis (see also
Anderson and West, 1996).

Facet Specificor GeneralizedClimate

Another controversial issue in climate research has been the meaning of the construct itself and
its operationalization in applied research. Although definitional specificity has proved to be an
( 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

238

N. R. ANDERSONAND M. A. WEST

elusive goal for climate researchers,much ground has been made recently through attempts to
deconstruct the notion of generalized climate into its constituent dimensions or subdomains.
These advances stem from Schneider and Reichers' (1983) assertion that it is meaningless to
apply the concept of climate without a particular referent (e.g. climate for change, climate for
quality, climate for innovation, etc.). Rousseau (1988) argues similarly, and advocates the study
of 'facet-specific climates', again referring to climate as a dynamic but intangible aspect of
organizational reality. Deconstructing climate as a generic term embracing multiple facets has
been a valuable way of clarifying some of the confusion over the precise meaning of the term.
Indeed, research into facet-specific climates continues to hold out promise as one route toward
overcoming the conceptual-definitional impasse, noted in many of the recent reviews as the
critical blockage to climate research (Glick, 1985; Rentsche, 1990; Rousseau, 1988; Schneider,
1985).
There has been growing interest in how particular types of climate (e.g. for innovation or
safety) lead to particular types of work group outcomes (e.g. innovativeness or accident
avoidance). However, as already noted, this requiresthat the existence of 'shared climates' can be
demonstrated in work groups, and necessarily therefore, that they can be measured with validity
and reliability. Moreover, it is as yet unclear whether certain dimensions of climate are predictive
of just one facet-specific outcome or numerous outcomes. For instance, high cohesion may be
simultaneously associated with greater innovativeness, lower frequency of accidents at work,
greater resistance to change, and so forth. Despite a proliferation in recent years of in-house and
proprietary climate measures, usually at the organizational level of analysis, there remains a
dearth of measures which have been properly validated to demonstrate both consensual and
discriminable validity within and across work groups and organizational subunits. Consensual
validity has been defined in terms of the shared perceptions approach, a measure having consensual validity if there is sufficient agreement within a team or organizational subunit over
perceived climate. Ironically, there is still little consensus amongst researchers over the precise
level at which the minimum cut-off for consensual validity should be set. Most measures of
climate take the organization as the unit of analysis (see for example, Patterson et al., 1992) but
there are real concerns about the extent to which agreement on climate perceptions can be
demonstrated across the entire organization, characterizedby quite disparate subcultures, departments, roles and hierarchical levels.
In terms of intragroup agreement, the James, Demaree and Wolf (1984) within-group interrater agreement statistic, rwg,was put forward as a measure of between-rateragreement. James
et al. (1984) propose the rwgstatistic as a measure of agreement for single item scales and a
derivative, the rwgj) statistic for multiple item scales. They argued for a criterion value of 0.70 and
above as being indicative of an agreement level sufficient to suggest sharedness. George (1990)
used this statistic to determine estimates of within-group interrater agreement for positive and
negative affect within 26 work groups. She found an average interrater agreement of 0.87,
concluding that it is meaningful to speak of an affective tone of work groups. The rwg(j)statistic
has provoked some debate, however. Kozlowski and Hattrup (1992) argue that rwg()should only
be used as an index of agreementand not of reliability,whereas Schmidt and Hunter (1989) have
criticized the James et al. method for not complying to tenets of classical measurement theory.
Nevertheless, James and his co-workers have responded constructively to these criticisms,
and guidelines for its interpretation
suggesting a slightly modified calculation procedure for rwgoj)
Demaree
As
James
et
al.
and
Wolf,
(1993) argue, there have been so few studies
(James,
1993).
it
is premature to draw definitive conclusions
their
in
that
statistic
using
organizational settings
over its application (see also Patterson et al., 1992).

? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

CLIMATEFOR WORKGROUPINNOVATION

239

Work GroupInnovation
In common with the literature on organizational climate, the last 30 years have witnessed a
burgeoning volume of research into innovation in organizations. This literatureis now as vast as

it is disparate,with contributionsfrommanagementscientists,appliedsociologists,and organizational psychologists (see for example, Hosking and Anderson, 1992; Kanter, 1983; Pettigrew,
1985; West and Farr, 1990). As an indication of the size of this literaturebase, reviews by Rogers
and Eveland (1978) and Kelly and Kranzberg (1978) cite 2400 and 4000 publications respectively
on organizational and technological innovation alone. Recent reviews provide relatively
integrated and structured overviews of this research (see Anderson, 1992; Anderson and King,
1993; King and Anderson, 1995; West and Farr, 1990), and so we confine our comments here to
briefly acknowledge work group innovation as being the facet-specific construct of interest in the
present study. West and Farr, (1989) define innovation as 'the intentional introduction and
application within a role, group or organization of ideas, processes, products or procedures, new
to the relevant unit of adoption, designed to significantly benefit role performance, the group, the
organization or the wider society' (p. 16). Comparatively few studies have focused at the level-ofanalysis of the work group. This is a notable shortcoming since it is often the case that an
innovation is originated and subsequently developed by a team into routinized practice within
organizations (West and Farr, 1990; Anderson and King, 1993; King and Anderson, 1995). For
example, a management team may initiate changes in organization procedures;a marketing team
may modify approaches to advertising product lines; and an assembly team may institute new
and improved methods of product manufacture. It is therefore important to address the topic of
work group innovation as an outcome in relation to proximal group climate.
To summarize, the present study had the objective of developing a multidimensional
measure of proximal work group climate for innovation for use in future research based upon
a priori deconstructions of group climate, and upon an hypothesized four-factor theory of
climate for innovation (West, 1990). The following sections describe the theoretical model and
the development and psychometric validation of this measure-the Team Climate Inventory
(TCI).

Developmentof the Team Climate

Inventory(TCI)
Thefour-factor theory
Previous reviews of research into both climate and innovation (e.g. West and Farr, 1990; West,
1990; Anderson and King, 1993; King and Anderson, 1995) informed the development of a fourfactor theory of climate for work group innovation. Reviews of this literature revealed a
consistent pattern of climate factors found across studies to be associated with team innovativeness. Summarizing these factors, West (1990) proposed a four-factor model of work group
innovation, hypothesizing that four major factors of climate are predictive of innovativeness (see
West and Anderson, 1996). This theory is described in detail elsewhere (West, 1990; West and
Anderson, 1996); but a brief description of the four factors-vision, participative safety, task
orientation, and supportfor innovation-is presented below.
? 1998John Wiley& Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

240

N. R. ANDERSONAND M. A. WEST

Vision
'Vision is an idea of a valued outcome which represents a higher order goal and a motivating
force at work' (West, 1990, p. 310). Work groups with clearly defined objectives are more likely to
develop new goal-appropriate methods of working since their efforts have focus and direction.
West (1990) asserts that work group vision has four component parts: clarity, visionary nature,
attainability, and sharedness. Clarity refers to the degree to which the vision is readily understandable. Visionary nature depicts the extent to which the vision has a valued outcome to
individuals in the group and thus engenders their commitment to group goals. Sharedness refers
to the extent to which the vision gains widespread acceptance by individuals within the team.
Visions should also be relatively attainable if they are to facilitate innovation, since if the goal

cannot be reached,it may eitherbe demotivatingor so abstractthat practicalstepstowardsits

achievement cannot realistically be envisaged.

Participative safety
'Participativeness and safety are characterized as a single psychological construct in which the
contingencies are such that involvement in decision-making is motivated and reinforced while
occurring in an environment which is perceived as interpersonally non-threatening' (West, 1990,
p. 311). West proposes that the more people participate in decision-making through having
influence, interacting, and sharing information, the more likely they are to invest in the outcomes
of those decisions and to offer ideas for new and improved ways of working. The construct of
participative safety hence relates to active involvement in group interactions wherein the
predominant interpersonal atmosphere is one of non-threatening trust and support. For example,

it is arguedthatparticipativesafetyexistswhereall membersof a workgroupfeel ableto propose

new ideas and problem solutions in a non-judgemental climate (see also Rogers, 1983).
Task orientation
'A shared concern with excellence of quality of task performance in relation to shared vision or
outcomes, characterized by evaluations, modifications, control systems and critical appraisals'
(West, 1990, p. 313). Within groups, the task orientation factor is evidenced by emphasis on
individual and team accountability; control systems for evaluating and modifying performance;
reflecting upon work methods and team performance; intra-team advice; feedback and cooperation; mutual monitoring; appraisal of performance and ideas; clear outcome criteria;exploration
of opposing opinions; constructive controversy (Tjosvold, 1982); and a concern to maximize
quality of task performance. This factor hence describes a general commitment to excellence in
task performance coupled with a climate which supports the adoption of improvements to
established policies, procedures, and methods.
Support for innovation
'... the expectation, approval and practical support of attempts to introduce new and improved
ways of doing things in the work environment' (West, 1990, p. 38).
Support for innovation varies across teams to the extent that it is both articulated and enacted.
West argues that articulated support, by implication, may be found in personnel documents,
policy statements, or conveyed by word of mouth. It is argued that a necessary condition for
group innovation is enacted support, as opposed to merely articulated support, whereby active
support is provided for innovatory behaviour. Daft (1986), for instance, found that resources
( 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

CLIMATEFOR WORKGROUPINNOVATION

241

needed to be made available to develop innovations, whilst Schroeder, Van de Ven, Scrudder
and Polley (1989) stressed the importance of support from the power elite for innovation
implementation.
To summarize, the four factor model was propounded by West (1990) as a facet-specific theory
of climate for work group innovativeness. This paper is primarily concerned, however, with the
initial phase of research into West's (1990) model-the development and psychometric validation of a measure of group climate based upon the four-factor theory. The following sections
describe the development and validation of this measure-the Team Climate Inventory (TCI)in some detail. The major predictive validity study testing West's model is presented in West and
Anderson (1996).

Method
Item generation
Initially, an extensive review of published measures of climate was conducted (see also Koys and
DeCottis, 1991). These measures were examined for their component subdimensions in relation
to the posited four-factor climate model. Only subscales or items appropriate to these factors
were retained. Measures were further screened against a level-of-analysis criterion, with those
measuring exclusively at the individual or organizational level being rejected. Relatively few items
or scales were retained so that most of the items comprising the original version measure were
self-generated. This original version comprised of 61 items grouped onto four scales of facetspecific climate. The composition of these subscales is described below.
Vision
Twelve items elicited information about team members' views on the clarity, sharedness,
attainability, and value of team objectives, (see Burningham and West, 1995). Respondents were
also asked to indicate the extent to which they felt their team colleagues were in agreement with,
and committed to, these objectives. Respondents were asked to indicate the extent to which
each statement was true of their team on a 7-point scale ranging, I=not at all to 7= completely,
(see Table 1 for a complete list of all items).
Participative safety
The construct of Participative Safety was subdivided into two components-team participation
and safety.
Teamparticipation was measured using 15 items to which respondents were asked to respond
on a 5-point strongly disagree to strongly agree scale. Following Wall and Lischeron (1978), the
measure was designed to tap three dimensions of participation: influence over decision making,
information sharing, and interaction frequency. Eight of these items were drawn from the
Tjosvold, Wedley and Field (1986) scale of constructive controversy and adapted for use in the
present study. Responses were on 5-point scales ranging from 1=strongly disagree to 5=strongly
agree.
Safety was measured by nine items asking respondents to indicate the extent of perceived safety
in their work group. Again, this was rated on a 5-point response scale ranging from l=a very
little extent to 5=a very great extent.
? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

242

N. R. ANDERSON AND M. A. WEST

Task orientation
This concept was also sudbivided into two components-climate for excellence and constructive
controversy.
Climatefor excellence was measured via a 10-item scale addressing the extent to which team
members interacted in order to promote excellence in the team's work (Burningham and West,
1995). The 7-point response scale ranged l=to a very little extent to 7= to a very great extent.
Tjosvold et al.'s (1986) concept of constructive controversywas used in the study as being
conceptually similar to the task orientation construct. Tjosvold et al. propose that when
controversy is productively discussed, it creates epistemic curiosity that leads to the 'exploration
of opposing positions, open-minded consideration and understanding of these positions, and a
willingness to integrate these ideas into a high quality, accepted solution' (p. 127). Seven items
from the constructive controversy scale were therefore included in the original version TCI. Items
were rated on a 5-point response scale from l=strongly disagree to 5=strongly agree.

Support for innovation

Four items were taken from Siegel and Kaemmerer's (1978) climate for innovation measure, and
four new items were developed to tap enacted support for innovation. Siegel and Kaemmerer's
measure was designed to assess organizational level attributes and so items were modified using
the word 'team' instead of 'organization'. The enacted support subscale assessed the extent
to which time, cooperation, practical support and resources were given by team members to
implement new ideas and proposals. Respondents were asked to indicate the extent to which each
statement was the true of their team on a 5-point scale ranging from l=strongly disagree to
5=strongly agree.

Procedurefor validation
This 61-item, four-scale original version Team Climate Inventory was piloted with 14 nursing
teams in a hospital setting and with two hospital management teams prior to use in this main
study in order to obtain reactions and comments on the measure (Anderson and Pineros, 1990).
No data were collected as part of this pilot; the purpose was simply to evaluate the face validity
and acceptability of the TCI to respondents.
Development of the TCI was undertaken as part of a wider-ranginglongitudinal study into the
relation between facet-specific climate for innovation and the innovativeness of management
teams within the British National Health Service (see West and Anderson (1992, 1996) and
Anderson, Hardy and West (1990) for detailed reports of this research). The procedure for
administering the scale to team members was as follows: the senior management teams of
35 major hospitals in three Regional Health Authorities across the U.K. were approached and
invited to participate in the study. Twenty-seven teams agreed to participate, giving a total
sample size of 243 individual subjects. Team size ranged between four and 19 members. A typical
team consisted of the Unit General Manager, Head of Nursing, Accountant, Personnel
Manager, Business Manager, and several Senior Medical Consultants. The 27 hospitals ranged in
size from 310 to 4000 employees, and had budgets ranging from ?6 million (approx $10 million)
per annum to ?47 million (approx $80 million) per annum. Each team was visited by a researcher
who outlined the research project and the commitment to the research process that would be
required from the management team. Batches of the original version measure were sent to all
Unit General Managers with a covering letter requestingthat they distribute the questionnaires to
( 1998 John Wiley & Sons, Ltd.

J. Organiz.Behav. 19:235-258 (1998)

CLIMATEFOR WORKGROUPINNOVATION

243

all members of their team. Individual respondents were provided with a pre-paid, addressed
envelope and were instructed to return their completed questionnaire direct to the researchers.
All questionnaires were completed anonymously and subjects were assured of the strict
confidentiality of responses to the questionnaire.

Results
Results are presented in order of the analyses undertaken: (i) exploratoryfactor analysis, the
sample being 155 managers from the 27 original hospital management teams; (ii) internal
homogeneity, alpha coefficients for this sample; (iii) predictive validity, variance accounted for
in independently rated team innovativeness by TCI results across the 27 hospital teams;
(iv) confirmatoryfactor analysis, based upon an independent sample of 121 teams from a variety
of organizations; and (v) consensual and discriminablevalidity measured by James et al. (1993)
rwg(j)and interclass correlations (ICC) across all sample groups in this study.

Exploratoryfactor analyses
A total of 155 questionnaires were returned from the 243 management team members surveyed,
giving an overall response rate of 63.7 per cent. Of this sample, all teams provided multiple
respondents with the number of respondents ranging between three of four for the smallest team
(i.e. 75 per cent response rate) to 11 of 19 for the largest team (i.e. 58 per cent response rate). The
latter team, in fact, provided the lowest within-team response across this sample. Initial analysis of
between-item correlations on the four scales relating to group climate (vision, participative safety,
task orientation and support for innovation) revealed positive and significant associations across
several items and therefore highlighted the possibility of an underlying simple structurewithin the
climate subscales. To examine this issue, a series of exploratory factor analyses were computed
using the Varimax procedure on SPSS-X. These analyses were computed at the individual level of
analysis in accordance with traditional approaches to item analysis and scale development. Items
were examined for skewness, kurtosis and intercorrelations at this level prior to running the
exploratory factor analysis (Anderson and West, 1996). This strategy of running item analyses on
individual level data first was intentional. As Anderson and West (1996) note, examining item
statistics at the individual level avoids additional problems of dealing with summed data at the
team level. Indeed, combined team-level data can obscure the psychometric characteristics of
items by collapsing-down distribution statistics to the team level. Additionally, of course,
retaining individual level data at this stage of development of the measure also maximized the
sample size, an important consideration in exploratory factor analysis as noted below.
Pre-analysis tests for the suitability of this data set for factor analysis were computed as
recommended by Comrey (1978). The Kaiser-Meyer-Olkin (KMO) measure of sampling
adequacy was 0.81, and the Bartlett test of sphericity was significant at p < 0.001, indicating the
suitability of this data for factor analytic procedures. One caveat should be noted at this point.
The ratio of cases was 155 : 61, or 2.5 : 1. Whilst it is acknowledged that this ratio is somewhat
lower than that recommended by some factor analysts (e.g. Gorsuch, 1983; Nunnally, 1978),
others have argued that smaller ratios are acceptable, and alternatively, that the absolute number
of cases should be considered as critical (Comrey, 1978; Ferguson, 1981; Kline, 1986). Kline
? 1998 John Wiley & Sons, Ltd.

J. Organiz.Behav. 19:235-258 (1998)

244

N. R. ANDERSON AND M. A. WEST

(1986), for instance, argues that a minimum of 100 cases is required. The exploratory factor
analysis was thus undertaken at the individual level of analysis, rather than at the group level in
order to retain the maximum number of cases.
An initial factor analysis without iteration was computed. The first factor extracted accounted
for 34.2 per cent of the variance, and application of the Scree Test (Cattell, 1966) indicated a fivefactor solution. All factors held eigenvalues of greater than unity, whilst this five-factor solution
accounted for 57.4 per cent of total variance. A further analysis was thus computed limiting the
number of factors to five with iteration and Varimax (orthogonal rotation). This solution
extracted 61.7 per cent of total variance. The factor loadings, cross-loadings, eigenvalues, and
variance statistics are presented in Table 1.
The factor loadings are relatively unambiguous, although some cross-loadings are evident, as
discussed below. A total of 38 items load at 0.50 or above onto the five-factor simple structure.
Factor I, which accounts for 37.8 per cent of the variance, loads exclusively onto scale items
relating to vision (e.g. How worthwhiledo you think these objectivesare to you?; How worthwhile
do you think these objectivesare to your team?;To what extent do you think membersof your team
are committedto these objectives?;To what extent do you think these objectivesare realistic and can
be attained?). This factor is conceptually clear, with 11 items loading from the original vision
subscale. Factor I is therefore labelled vision.
Factor II comprises eight items from the original participation subscale loading at 0.50 or
above. However, four items from this original subscale load onto factor V, indicating a division
between these items arising from orthogonal rotation of the factor matrix. The items loading
onto factor II all relate to perceptions of a participative climate within the team; perceptions of
the degree of information sharing between team members; perceptions of influence over others in
decision making; and of interpersonal safety and trust within the team (e.g. There are real
attempts to share informationthroughoutthe team; We all influenceeach other; Peoplefeel understood and accepted by each other). Conversely, factor V comprises four items relating to the
frequency of interaction both formally and informally, between team members (e.g. We keep in
touch with each other as a team; We keep in regularcontact with each other; Members of the team
meetfrequently to talk bothformally and informally).Consequently, factor V can be conceived as a
scale of interaction frequency, whilst factor II is comprised of conceptually distinct items relating
to participativeness and interpersonal safety. Factor II is thus titled participative safety, and
factor V is provisionally labelled interactionfrequency.
Factors III and IV display unambiguous patterns of item loadings, with factor III comprising
exclusively items measuring support for innovation (e.g. People in this team are always searching
for fresh, new ways of looking at problems; This team is always moving toward the developmentof
new answers; Team membersprovidepractical supportfor new ideas and their application).A total
of eight items from the original support for innovation subscale load onto Factor III which
accounts for a further 6.0 per cent of total variance.
Factor IV is comprised of seven items, all from the original task orientation subscale and adds
a further 4.6 per cent of the variance (e.g. Does your team critically appraisepotential weaknesses
in what it is doing in order to achieve the best possible outcome?;Does the team continuallymonitor
its ownperformancein order to achieve the highest standards?;Do you and your colleagues monitor
each other so as to maintain a high standardof work?).As items loading onto both factor III and
factor IV are drawn exclusively from distinct subscales of the original version measure, the
subscale titles are retained from these factors: factor III as supportfor innovation,and factor IV as
task orientation.
To summarize this five-factor solution, it is apparent that the underlying simple structure
displays a fairly unambiguous pattern of item loadings, mostly in line with the postulated model
? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

CLIMATEFOR WORKGROUPINNOVATION

245

of climate for innovation. Overviewing Table 1, four items display cross-leadings of 0.40 or
above. These are items 9 (How worthwhiledo you think these objectivesare to the widersociety?),
38 (People in the team co-operate in order to help develop and apply new ideas), 47 (Does the team
have clear criteria which members try to meet in order to achieve excellence as a team?) and 58
(We keep in touch with each other as a team). In addition, it should be noted that on all of these
items, primary loadings were relatively high (0.59, 0.62, 0.53 and 0.74, respectively). We therefore
determined a threshold of 0.50 and above to retain items instead of the more common 0.40 or
even 0.30. One caveat regards the original participative safety subscale which, following Varimax
rotation of the matrix, is more appropriately represented as two discrete factors-participative
safety and interaction frequency, extracted as factors II and V in this solution. The other three
factors (factor I: vision, factor II: support for innovation, and factor IV: task orientation)
complete this solution which accounts for 61.7 per cent of the total variance. Note that although
the original scales contained negatively connotated items, none of these loaded in this factor
solution.

Internal homogeneity
More detailedanalysesof this solutionwereundertakento examinethe internalconsistencyof
the factors and factor independence. Table 2 presents means, standard deviations, intercorrelations and Cronbach's alpha coefficients for the five factors as composite scales on this sample of
hospital management teams.
Alpha coefficients range between 0.84 and 0.94 indicating acceptable levels of internal
homogeneity and reliability for all five factors. This table displays some intriguing results. All
scales are significantly and positively intercorrelated(p < 0.01). Scale correlations range between
0.35 (interaction frequency with objectives) and 0.62 (task orientation with support for innovation), indicating a strong halo effect across scales. Whilst these correlations are not high enough
to give rise to concerns over multicollinearity, they do highlight the possibility that perceptions of
climate may be prone to a halo effect. Further research focusing on work groups in different
environments is called for to examine this finding in other contexts.

Predictive validity
As previously noted, the development of the Team Climate Inventory was undertaken as part of a
longitudinal study testing the validity of the four-factor theory of facet-specific climate for
innovation (West and Anderson, 1996). Reports of innovations implemented by the management
teams in the 27 hospitals in the 6 months succeeding the administration of the TCI were judged
by expert and naive raters on a number of dimensions, including overall innovativeness, number
of innovations, radicalness, magnitude, novelty and administrative effectiveness (see West and
Anderson, 1996; Anderson and West, 1994). Combined, team level scores on the TCI were used
to predict team scores on the dimensions of innovativeness. Support for innovation emerged as
the only significant predictor of overall innovation, accounting for a substantial 46 per cent of
the variance; and the only predictor of innovation novelty. Participative safety emerged as the
best predictor of the number of innovations and team self-reports of innovativeness, while task
orientation predicted administrative effectiveness. For a detailed report of these findings and
examples of the types of innovations implemented by the management teams see West and
? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

O?
I'.0
oo
0

Table 1. Varimax-rotated loadings on the five-factor solution

Team Climate Inventory (TCI)* items

I
Vision

0r
3

C
pCD
oi
o

.I0
tr
C_
0

0o
s.

%0
OC

0.85
1. How clear are you about what your teams objective are? (r)
0.85
2. To what extent do you think they are useful and appropriate
objectives? (r)
0.81
3. How far are you in agreement with these objectives? (r)
0.77
4. To what extent do you think other team members agree with
these objectives? (r)
0.75
5. To what extent do you think your team's objectives are clearly
understood by other members of the team? (r)
0.74
6. To what extent do you think your team's objectives can actually
be achieved? (r)
0.74
7. How worthwhile do you think these objectives are to you? (r)
0.70
8. How worthwhile do you think these objectives are to the
organization? (r)
0.59
9. How worthwhile do you think these objectives are to the
wider society? (r)
0.55
10. To what extent do you think these objectives are realistic and
can be attained? (r)
0.54
11. To what extent do you think members of your team are
committed to these objectives? (r)
0.08
12. How worthwhile do you think these objectives are to your team?
0.14
13. We share information generally in the team rather than keeping it
to ourselves (r)
0.10
14. We have a 'we are in it together' attitude (r)
0.32
15. We all influence each other (r)
0.13
16. People keep each other informed about work-related issues
in the team (r)
0.15
17. People feel understood and accepted by each other (r)
0.01
18. Everyone's view is listened to even if it is in a minority (r)
19. There are real attempts to share information throughout the team (r)
0.06
0.08
20. There is a lot of give and take (r)
21. Disagreeing with another's idea is not a rejection of that person
0.26
22. People try to control each other [R]
-0.24
23. We try to blame each other [R]
-0.26
0.61
24. How friendly or easy to approach are the people in your team?
25. To what extent are the members of your team critical of new ideas? [R] -0.01

II
Participation
safety

III
Support for
innovation

0.04
0.25

0.12
0.37

0.17
-0.01

0.10
0.06

-0.09

0.23

0.10

0.17

0.26
0.29

0.24
0.27

-0.15

0.10

0.32

0.30

0.31

0.27

0.84
0.72

0.21
0.19

0.69
0.66
0.66

0.24
0.06
0.14

0.65
0.64
0.62
0.58
0.06
0.05
-0.05
0.04
-0.21

0.11
0.06
0.18
0.08
0.16
0.14
0.01
-0.11
-0.06

00
o
~:
cn
?
y'

41

d'
N
g
to
ht
oo
Ioo

-0.25
26. How threatening do you find putting forward new ideas
to the team? [R]
0.63
27. How supportive are the other members of your team?
0.34
28. To what extent is there feeling of trust between members of
your team?
0.32
29. To what extent are persons in your team willing to listen to
your problems?
0.39
30. To what extent do others foster an atmosphere of non-threatening
co-operation amongst members of the team?
0.42
31. To what extent do you feel at ease with the members of your team?
0.51
32. Do other team members have a genuine concern over your
personal well-being?
33. This team is always moving toward the development of new answers (r) 0.30
0.30
34. Assistance in developing new ideas is readily available (r)
0.16
35. This team is open and responsive to change (r)
0.20
36. People in this team are always searching for fresh, new ways of
looking at problems (r)
0.35
37. In this team we take the time needed to develop new ideas (r)
0.40
38. People in the team co-operate in order to help develop and apply
new ideas (r)
0.25
39. Members of the team provide and share resources to help in the
application of new ideas (r)
0.21
40. Team members provide practical support for new ideas and their
application (r)
0.17
41. Do your team colleagues provide useful ideas and practical help to
enable you to do the job to the best of your ability? (r)
0.37
42. Do you and your colleagues monitor each other so as to maintain
a higher standard of work? (r)
0.31
43. Are team members prepared to question the basis of what the
team is doing? (r)
0.31
44. Does the team critically appraise potential weaknesses in what it is
doing in order to achieve the best possible outcome? (r)
0.05
45. Do members of the team build on each other's ideas in order to
achieve the best possible outcome? (r)
0.22
46. Is there a real concern among team members that the team should
achieve the highest standards of performance? (r)
0.27
47. Does the team have clear criteria which members try to meet in
order to achieve excellence as a team? (r)

0.06

0.01

0.24
0.21

0.19
0.20

0.24

0.11

0.19

0.12

0.11
0.24

-0.02
0.25

0.14
0.28
0.18
0.14

0.73
0.72
0.67
0.63

0.32
0.13

0.62
0.62

0.36

0.60

0.30

0.56

0.22

0.06

0.21

0.20

0.35

0.10

0.03

0.23

0.35

0.09

0.16

0.13

0.41

0.28

?
0
0o

0.

pp
0o

00
-

-4
OO

oN

SP
CD
?
k-

to
Eo
00

Table 1. Continued
Team Climate Inventory (TCI)* items

48. Do your team colleagues provide helpful advice and constructive

feedback in order to encourage you to do the job to the best
of your ability?
49. Does the team continually monitor its own performance in order to
achieve the highest standards?
50. Does the team continuously evaluate its work in order to improve
its effectiveness?
51. People express their own views fully
52. We first try to understand the problem fully
53. People try to win by pushing and keeping their own original views
54. We understand the problem before we seek a solution
55. We seek a solution good and acceptable to all
56. Opposing views aid in the full consideration of all the issues
57. All ideas are expressed before we begin to evaluate them
58. We keep in touch with each other as a team (r)
59. We keep in regular contact with each other (r)
60. Members of the team meet frequently to talk both formally and
informally (r)
61. We interact frequently (r)
Eigenvalue
Percentage of variance

I
Vision

II
Participation
safety

III
Support for
innovation

0.41

0.33

0.29

0.18

0.24

0.75

0.04

0.29

0.67

0.39
0.30
-0.19
0.11
0.21
0.13
0.36
0.17
0.08
0.06

0.21
0.18
-0.07
0.29
0.35
0.12
0.09
0.49
0.35
0.29

0.14
0.25
-0.23
0.38
-0.04
0.38
0.44
0.23
0.24
0.18

0.01

0.28

0.16

4.05
9.00%

2.71
6.00%

17.02
37.80%

TotalVarianceAccountedfor = 61.7 per cent.

Itemsloadingat or above0.50 are shownin bold for clarity.
(r), itemretainedfor the short-formversionTCI.
[R], reversescoreditem.
* ?Copyright, Andersonand West/ASE(1994).See also Anderson,N. R. and West,M. A. (1994). TheTeamClimateInventory,Asses
NFER-Nelson-Nelson,DarvilleHouse, 2 OxfordRoad East, Windsor,Berkshire,SL4 IDF, U.K. Researchusage of the TCI is pe
consultancypurposesis governedby worldcopyrightheld by ASE, NFER-Nelson.

00

?p
o0

S.b
r

Table 2. Descriptive statistics, reliabilities, and intercorrelation matrix for the five-factor solution
Factor
n
I
II
III
IV
V
?-

0
0(-

I
CD

00

00

Vision
Participative safety
Support for innovation
Task orientation
Interaction frequency

*p < 0.01.

Descriptives
n-items loading
at 0.50

149
154
153
153
154

11
8
8
7
4

Mean
score

S.D.

59.31
30.32
38.42
42.57
14.97

10.74
4.87
7.28
7.73
2.84

Intercorrelation matrix
Coefficient Objectives Participativ
safety
alpha
0.94
0.89
0.92
0.92
0.84

0.46*
0.60*
0.59*
0.35*

0.60*
0.60*
0.49*

250

N. R. ANDERSONAND M. A. WEST

Anderson (1996). Follow-up studies into the predictive validity of this measure further support
its utility as a facet-specific scale (e.g. Agrell and Gustafson, 1994; Forrester, 1995). These results
suggest the predictive validity of the TCI and provide some support for the theory of work group
innovation.

Confirmatoryfactor analysis
To ensure the robustness of this five-factor solution, a confirmatory factor analysis (CFA) using
AMOS version 3.51 (Arbuckle, 1995) was performed. AMOS is a comprehensive text- and
graphics-based structural equation modelling and CFA program similar in design to LISREL
and EQS. This analysis utilized an independent sample who completed the 38-item, short-form
Team Climate Inventory. The questionnaire was administered to three additional samples of
work groups: 273 members of 35 primary health care teams; 220 members of 24 management
teams in an international oil company, 360 members of 42 social series teams and 118 members of
20 community psychiatric care teams. This confirmatory sample therefore comprised a total of
121 teams (total N individuals = 971).
Initially, individual responses to items were summed within teams to create a group level sum
for each item. An intercorrelation matrix was then calculated for these 38 group level variables
and this intercorrelation matrix utilized as input data within AMOS. The issue of sample size
remains an active debate within the structural equation modelling literature (for detailed recent
commentaries, see Brannick, 1995; Kelloway, 1995; Williams, 1995). With smaller samples
(n < 150) there is a danger of obtaining nonconvergent solutions, even for more highly specified
models (Boomsma, 1982; Anderson and Gorbing, 1984). With larger samples (n > 400), trivial
discrepanciescan lead to rejection of a satisfactory model since absolute indices of fit (see below)
are prone to influence by sample size (Bollen, 1989; Loehlin, 1992). Two main suggestions and
'rules of thumb' have been proposed in response. First, as Loehlin (1992) states 'As a rough rule
of thumb, one would probably do well to be modest in one's statistical claims if N is less than
100'. The sample size for the present study, once data had been combined into team level
responses, was 121. Second, nonconvergent solutions have been found to be frequent in CFA
samples of less than 150 with only two observedvariablesper factor, (Boomsma, 1982; Anderson
and Gorbling, 1984). Here, as described below, we modelled between four and twenty items per
factor, and one of our computed models failed to reach convergence. Therefore, although the
sample size was reduced substantially by combining individual responses into team level data, the
present data set met both of these recommended criteria.
The following procedurewas adopted to examine the relationship between different theoretical
models (Bentler and Bonnet, 1980; Breckler, 1990; Byrne, 1989; Loehlin, 1992). Firstly, a one
factor model with all items loading onto a single factor was run. Next, two, two factor models
were run. In each model the first factor contained all the Participation and Support items
hypothesized by West (1990) to relate to the quantity of innovations generated by teams. The
second factor contained all the Task Orientation and Vision items hypothesized by West (1990)
to relate to the quality of innovations developed by teams. These two models differed in that the
first did not allow for any factor inter-correlation(Twofactor (2a)), whereas the second did allow
for factor inter-correlation (Twofactor (2b)). Next, two, four factor models were run. In these
models, the first factor contained all the Vision items, the second all the Participative Safety
items, the third all the Support for Innovation items, and the fourth all the Task Orientation
items. In these models, the four Interaction Frequency items revealed in the exploratory factor
analysis were incorporated into the Participation scale in order to test this factor structure as
? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

CLIMATEFOR WORKGROUPINNOVATION

251

Table 3. Overall fit indices for the team climate inventory scales

X2
Null model
One factor
Two uncorrelated factors (2a)
Two correlated factors (2b)
Four uncorrelated factors (4a)
Four correlated factors (4b)
Five uncorrelated factors (5a)
Five correlated factors (5b)

1966.74
1799.68
1714.04
1650.50
1406.82
1569.94
1286.93

Absolute
indices
df
X2/df
703
665
665
664
665
659
665
655

2.96
2.71
2.58
2.48
2.13
2.36
1.96

Relative
Parsimony
index
indices
CFI
PNFI
TLI
NFI

AX2

167.06
85.64
63.54
243.68
(163.12)
283.01

0.91
0.92
0.93
0.93
0.95
0.94
0.96

0.88
0.89
0.90
0.90
0.92
0.91
0.92

0.92
0.93
0.93
0.94
0.95
0.94
0.96

0.79
0.80
0.81
0.81
0.81
0.81
0.82

N, 121teamsfor all models.

TLI, TuckerLewis Index;NFI, Normed Fit Index;CFI, Normed NoncentralityFit Index; PNFI, Parsimonious
Normed-FitIndex.

originally hypothesized by West (1990). Again, these models differed with regard to factor
intercorrelations;the first did not allow for factor intercorrelation (four-factor (4a)) whereas the
second did (four-factor (4b)). Finally, two, five-factor models were run extracting the interaction
frequency scale as a separate fifth scale, again one allowing the factors to intercorrelate and the
other not. See Bollen (1989); Byrne (1989), and Loehlin (1992) for general introductions and
overviews of these procedures. Table 3 displays the CFA results and reports absolute, relative and
parsimonious indices of fit to comprehensively evaluate the fit of the different models computed.
The change in chi-square illustrates the incremental improvement achieved by each successive
model (Bentler and Bonnett, 1980; ollen, 1989). The ratio of chi-square to degrees of freedom
(X2/df) is also given, with ratios of less than 2.0 indicating a good fit. However, since absolute
indices can be adversely affected by sample size (Byrne, 1989; Loehlin, 1992), three relative
indices (TLI, NFI, and CFI) together with the Parsimonious Normed-Fit Index were computed
and provide a more robust evaluation of model fit (e.g. Tucker and Lewis, 1973; Byrne, 1989).
For the TLI, NFI and CFI, coefficients close to unity indicate a good fit, with acceptable levels of
fit being above 0.90 (Marsh, Balla and MacDonald, 1988; Mulaik, James, Van Alstine, Bennett,
Lind and Stilwell, 1989). The PNFI takes the degrees of freedom for a specified model into
account and so, for models involving larger numbers of observed and unobserved variables, the
PNFI value will often be lower than those obtained for relative indices (Mulaik et al., 1989;
Loehlin, 1992).
The results indicate that the five correlated factors model (model 5b) has the most parsimonious fit, with the TLI suggesting that this model accounts for 96 per cent of the variance
(TLI = 0.96; X2/df = 1.96). Further, the x2 to degrees of freedom ratio is below 2.0, again
suggesting acceptable fit for this model. The NFI and CFI are also above 0.90 for this model.
Although this five-factor correlated model possesses the most robust fit statistics of all models
tested, it should be noted that for only two models, the one factor model and the two
uncorrelated factors model, the NFI fell marginally below acceptable levels (NFI = 0.88 and
0.89, respectively). However, the change in chi-square figures indicate an improvement for each
successive model computed, apart from the five uncorrelated factors model (model 5a) over the
four-correlated factors model (model 4b) where the chi-square increased rather than decreased.
Indeed, the four-correlated factors model is only marginally less parsimonious than the fivecorrelated factors model, although the chi-square to degrees of freedom ratio would suggest the
latter to possess better fit (model 4b: X2/df = 2.13; model 5b; x2/df = 1.96). Both models are
( 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

252

N. R. ANDERSONAND M. A. WEST

acceptable, however, and the differences in fit too small to permit conclusive distinctions being
drawn between these two correlated factors models. This point is considered further in the
Discussion. However, in order to maximize the predictive utility of the measure we elected to
employ the five-factor solution in further analyses.

Consensualand discriminableclimates
We noted earlier the issue of whether shared climates can be claimed to exist within organizations
and that previous researchers have drawn attention to the problem of determining criteria for
minimum levels of agreement. Using the James et al. (1984) formula we calculated the rwgj) and
inter-class correlation (ICC) statistics for the groups in the five samples (National Health Service
hospital management teams, oil company teams, community psychiatric teams, primary health
care teams and social services teams) across the five scales. For four of the five scales distributions
revealed no significant skew. However, the vision scale was negatively skewed and so the James
et al. (1984) procedure for employing an expected variance to take account of this skew was
followed. Table 4 shows that the average rwg()within each of the five samples across the five Team
Climate Inventory scales ranged from 0.67 through to 0.98 with only one out of the 25 values
falling below the 0.7 level used by George (1990) and by Nunally (1978) as an acceptable level for
internal consistency. These results suggest that the measure is consistently tapping shared climate
perceptions, rather than aggregating radically diverse individual perceptions.
However, it is also important to demonstrate differences between groups to determine the
discriminant power of any climate instrument (Rousseau, 1988). One-way ANOVAS on the
aggregate variables were therefore performed within each of the five samples. Hays (1981)
suggests that minimum evidence for differences across groups would be an F ratio from an
ANOVA greater than 1.00. The results in Table 4 show that on all but one of the scale scores
across all of the samples, the ANOVAS produce F ratios greater than unity. Moreover, in all but
three out of 25 cases, the F value is statistically significant (p < 0.05). This indicates that the
measure distinguished between different groups and thus possessed adequate discriminable, as
well as consensual, validity.

Discussion
This paper has described the development of the Team Climate Inventory for measuring an
important aspect of facet-specific work group climate-climate for innovation. Evidence of the
factor structure of the measure has been provided, based both on exploratory and confirmatory
factor analyses. Evidence is also given about the internal homogeneity and reliability of the scales
and their predictive validity. Finally, consensus amongst teams in relation to team climate is
demonstrated, both within and across samples in relation to the five dimensions of the TCI. These
analyses provide initial support for the utility of the measure as a self-report measure of facetspecific climate within work groups and organizational teams.
Rousseau (1988) has argued for the development of facet-specificmeasures of climate and our
data show that it does indeed make sense to develop measures of climate which are facet-specific.
Our data also show that the level of agreement about climate is considerably greater than that for
? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

O0
00
O
C-

_.

I
go

rL

Table 4. Team scores and mean indices of concensus on the five climate factors
Vision

0
p

to
i
N
CD

tO
I

hA
oo
I-

o
<-

27 NHS
5.84
teams
35 primary 4.84
care teams
42 social
5.10
services teams
20 psychi- 5.27
atric teams
24 oil co.
5.33
teams
148 teams 5.26
overall
*p <0.05.

S.D. ICC F

rwg(j) X

Participative safety

S.D. ICC

1.06 0.040 1.11 0.84 3.79 0.62 0.003 0.40

Support for innovation

rwg(j) X

S.D. ICC

rwg(j) X

Task orientation

S.D. ICC

rwg(

0.91 4.81 0.91 0.035 1.80* 0.70 4.51 1.09 0.026 1.64* 0.82

1.13 0.027 1.96* 0.90 3.59 0.69 0.04 2.50* 0.89 3.95 0.71 0.038 0.28* 0.92 4.20 1.34 0.022 1.78* 0.81
1.12 0.017 1.71 0.85 3.71 0.65 0.04 2.28* 0.91 3.57 0.71 0.05 3.29* 0.93 4.79 1.28 0.02 2.00* 0.75
1.05 0.120 3.68* 0.67 3.93 0.65 0.100 3.34* 0.98 3.62 0.63 0.130 3.95* 0.92 5.77 1.28 0.060 2.35* 0.85
0.90 0.047 2.19* 0.97 3.84 1.37 0.160 5.57* 0.92 4.40 1.03 0.126 4.47* 0.72 4.75 0.94 0.099 3.65* 0.70
1.05 0.050 2.13* 0.85 3.77 0.80 0.060 2.82* 0.92 4.07 0.80 0.060 3.18* 0.84 4.80

1.19 0.050 2.28* 0.79

254

N. R. ANDERSONAND M. A. WEST

measures which are less facet-specific (for example, see Patterson et al., 1992). This conclusion is
amply illustrated by the finding that broad measures of team level cohesion are not particularly
useful for predicting team outcomes, since cohesion can serve a variety of purposes for groups
(Schroder, 1990). Our examination of climates for innovation indicates that by focusing on a
specific aspect of climate and its relationship to specific aspects of group level outcomes, greater
predictive accuracy is achieved. These findings noted, we would also acknowledge that the
dimensions comprising the TCI may correlate with other types of group level outcome. For
instance, higher scale scores on task orientation may correlate with overall group productivity
since team members are constantly appraising each others' performance; higher scores on
participative safety may predict strong group cohesiveness which, in turn, may be suggestive of
higher levels of resistance to change imposed from above upon the group (King and Anderson,
1995); and, group scores on the vision scale may be predictive of double-loop learning in the
group since team members are continuously questioning the relevance of their goals and
objectives. The point is that the facet-specific nature of any multidimensional climate scale is
likely to be mediated by a wider nomological net of relations between climate dimensions and
various group level outputs. The TCI was originally operationalized and developed as a facetspecific measure of climate for innovations; it is likely in addition to be useful in measuring
climate dimensions predictive of other types of group output, but further researchis called for to
examine this issue.
Another reason why the TCI appears to produce high levels of consensus amongst team
members is because of its focus on the proximal work group. Most previous measures of climate
have evaluated organizations as a whole where there may be considerably more variation in
perceptions of the work environment, and where, as Payne (1990) has argued, there is less
likelihood of social interaction leading to shared meanings. As noted earlier in this paper, it may
well be that the proximal work group represents one of several examples within organizations
where consensual climates can be readily identified. Other possibilities include those occupying
similar roles at identical hierarchiallevels within the organization, individuals having progressed
through a standardized socialization procedure into the organization, or the downward
communication of organizational vision which is accepted by those lower down the organization
hierarchy.
One issue which arose from the analyses undertaken in this study concerns the factor structure
of the TCI, and specifically, whether a four- or a five-factor structurerepresentsthe measure most
parsimoniously. West (1990) proposed a four-factor theory as his model of facet-specific climate
in work groups. Agrell and Gustafson (1994), in a study using a Swedish translated version of the
original measure, conducted an exploratory factor analysis with Varimax rotation on data for 124
respondents comprising 16 work teams. This indicated a four-factor solution which accounted
for 54 per cent of the total variance. Conversely, in the present study, the exploratory factor
analysis conducted on the data from 27 hospital management teams suggested a five-factor
solution; the difference being the separation out of four items from the original participation
scale relating specifically to interaction frequency within the team. The confirmatory factor
analysis subsequently undertaken on a different sample of 121 teams suggested both the fourfactor and the five-factor solutions were acceptable in terms of goodness of fit statistics, but that
the five-factor intercorrelated model demonstrated the most parsimonious fit. As the items
comprising the fifth factor are conceptually distinct, this scale is likely to have incremental value
either as a subscale within the participation scale, or as a separate factor. Further research into
this aspect of the psychometric structure of the TCI is thus called for.
The extent to which it is possible to assume the existence of climates within teams was also
examined in this study. The data suggest that discriminable climates do exist within teams, since
? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

CLIMATEFOR WORKGROUPINNOVATION

255

there is evidence of greater variation between, than within, teams. However, there is a need to
more precisely address the level of agreement within teams, as well as examining the differences
between them. Our findings indicate that there is a good deal of agreement about the dimensions
of climate within teams although, perhaps not surprisingly, there is variation in level of
agreement both across teams within samples and within teams across particular dimensions. In
the current study, the extent of agreement within teams varied across dimensions, the evidence
suggesting that there was most agreement in relation to the participative safety scale and least
agreement in relation to the task orientation scale (see Table 4). Moreover, in comparing across
teams within any one sample, it is clear that agreement on any particular climate dimension can
vary considerably and in ways which are not immediately predictable. Clearly, agreement on
dimensions of climate can itself be used as an important defining characteristicof groups. Future
research might usefully examine the question of why particular dimensions produce more or less
agreement, and whether these variations occur in consistent ways across teams.
At a more pragmatic level, the 38-item TCI provides an accessible and easily administered
measure of team climate for innovation (Anderson and West, 1994). Its potential for use in
settings such as organizational climate surveys, team building and development, selection of new
members into groups, and group development over time, should be noted. Practitioners and
consultants in the field now face a bewildering choice of psychometric measures, but few have
been conceived of, and validated, specifically at the group level of analysis. In conclusion, the TCI
holds promise as a measure of group climate in organizations, and for team building and
organization development interventions. Further researchis needed to investigate the potential of
the TCI as a measure of team climate in such a variety of contexts and applications.

References
Academy
Abbey,A. and Dickson,J. W. (1983).'R & D work:Climateand innovationsin semiconductors',
of ManagementJournal,26, 362-368.
Agrell,A. and Gustafson,R. (1994).'TheTeamClimateInventoryand groupinnovation:A psychometric
and Organizational
test on a Swedishsampleof work groups',Journalof Occupational
Psychology,67,
143-151.
Anderson,J. and Gorbing,D. W. (1984).'Theeffectsof samplingerroron convergence,impropersolutions
and goodness-of-fitindicesfor maximumlikelihoodconfirmatoryfactor analysis',Psychometrics,49,
155-173.

Anderson,N. R. (1992). 'Workgroup innovation:A state-of-the-artreview'.In: Hosking, D. M. and

andPracticesin
Anderson,N. R. (Eds)Organizational
ChangeandInnovation:
PsychologicalPerspectives
Europe,Routledge,London,pp. 127-138.
In: Cooper,C. L. and Robertson,I. T.
Anderson,N. R. and King, N. (1993).Innovationin Organizations.
Reviewof IndustrialOrganizational
(Eds) International
Psychology,Vol. 8, Wiley,Chichester,pp. 1-33.
Anderson,N. R. and Pineros,C. (1990). 'Innovationin nursingteams: SAPU Memo Number 1004',
Unpublished manuscript, MRC/ESRC Social & Applied Psychology Unit, University of Sheffield, U.K.

Anderson,N. R. and West,M. A. (1994). The TeamClimateInventory.Manualand Users'Guide,Assessment Services for Employment, NFER-Nelson, Windsor, U.K.

Anderson,N. R. and West,M. A. (1996).'TheTeamClimateInventory:The developmentof the TCI and

its applicationsin teambuildingfor innovativeness',EuropeanJournalof Workand Organizational
Psychology, 5, 53-66.

Anderson,N. R., Hardy,G. and West, M. A. (1990, September).'Innovativeteamsat work', Personnel

Management, 48-53.

Arbuckle,J. L. (1995).AMOS 3.5for Windows,SmallWatersCorporation,Chicago,II.

? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

256

N. R. ANDERSONAND M. A. WEST

Ashforth,B. E. (1985). 'Climateformation:Issuesand extensions',Academyof ManagementReview,4,

837-847.

Bentler,P. and Bonnett,D. (1980). 'Significancetests and goodnessof fit in the analysisof covariance
structures', Psychological Bulletin, 88, 588-606.
Bollen, K. A. (1989). StructuralEquations with Latent Variables,Wiley, New York.

Boomsma,A. (1982).'The robustnessof LISTRELagainstsmall samplesizes in factoranalysismodels.

In: Joreskog, K. G. and Wold, H. (Eds) Systems Under Indirect Observation, Part I, North-Holland,
Amsterdam, pp. 149-173.
Brannick, M. T. (1995). 'Critical comments on applying covariance structure modelling', Journal of
OrganizationalBehavior, 16, 201-213.

Breckler,S. (1990). 'Applicationsof covariancestructuremodellingin psychology:cause for concern?'

Psychological Bulletin, 107, 260-273.

Burninghamand West(1995).'Individualclimate,and groupinteractionprocessesas predictorsof work

team innovation', Small Group Research, 26, 106-117.

Byrne, B. (1989). A Primer of LISREL: Basic Applications and Programmingfor ConfirmatoryFactor
Analytic Models, Springer-Verlag,Berlin.
Campbell, J. P., Dunnette, M. D., Lawler, E. F. and Weick, K. E. (1970). Managerial Behaviour,
Performanceand Effectiveness,McGraw Hill, New York.
Campion, M. A., Medsker, G. J. and Higgs, A. C. (1993). 'Relations between work group characteristics and effectiveness: Implications for designing effective work groups', Personnel Psychology, 46,
823-850.

Behavioral
Cattell,R. B. (1966).'Thescreetestforthenumberof factors',Multivariate
Research,1, 140-161.
Comrey, A. (1978). 'Common methodological problems in factor analytical studies', Journal of Clinical &
ConsultingPsychology, 46, 648-659.

Daft, R. L. (1986).'A dual-coremodelof organizationinnovation',Academyof ManagementJournal,21,

193-210.

Dansereau, F. and Alutto, J. A. (1990). 'Level-of-analysisissues in climate and culture research',

In: Schreider, B. (Ed.) OrganizationalClimate and Culture, Jossey Bass, San Francisco, pp. 193-236.
Ferguson, G. A. (1981). Statistical Analysis in Psychology and Education, McGraw Hill, New York.
Forrester, R. H. (1991). 'Team climate and innovation: An empirical study'. Paper presented at the Fourth
European Congress of Psychology, Athens, July 1995.
George, J. (1990). 'Personality, affect and behavior in groups'. Journal of Applied Psychology, 75, 107-116.
Glick, W. H. (1985). 'Conceptualising and measuring organizational and psychological climate: Pitfalls in
multi-level research', Academy of Management Review, 10, 601-616.
Gorsuch, R. (1983). Factor Analysis, Lawrence Erlbaum Associates, London.
Guion, R. M. (1973). 'A note on organizational climate', OrganizationalBehaviorand Human Performance,
9, 120-125.
Hays, W. L. (1981). Statistics. Rinehart and Winston, New York.
Hosking, D. M. and Anderson, N. R. (Eds) (1992). OrganizationalChange and Innovation:Psychological
Perspectivesand Practices in Europe, Routledge, London.
Jackofsky, E. F. and Slocum Jr, J. W. (1988). 'A longitudinal study of climate', Journal of Organizational
Behavior, 8, 319-334.

James, L. R. and Jones, A. P. (1974). 'Organizationalclimate:A review of theory and research',

Psychological Bulletin, 81, 1096-1112.

James, L. R. and Sells, S. B. (1981). 'Psychological climate: Theoretical perspectives and empirical
research'. In: Magnusson, D. (Ed.) Toward a Psychology of Situations. An International Perspective,
Erlbaum, Hillsdale, NJ, pp. 275-295.
James, L. R., Demaree, R. G. and Wolf, G. (1984). 'Estimating within-group interraterreliability with and
without response bias', Journal of Applied Psychology, 69, 85-98.
James, L. R., Demaree, R. G. and Wolf, G. (1993). 'rwg: An assessment of within-group interrater
agreement', Journal of Applied Psychology, 78, 306-309.

Joyce, W. F. and Slocum, D. (1984). 'Collectiveclimate:Agreementas a basis for definingaggregate

climates in organizations', Academy of Management Journal, 27, 721-742.
Jorsekog, K. and Sorbom, D. (1984). LISREL VI: Analysis of Linear StructuralRelationshipsby Maximum
Likelihood, Instrumental Variables and Leasts Squares Methods, 3rd edn, Department of Statistics,

Universityof Uppsala,Sweden.

Kanter, R. M. (1983). The Change Masters: CorporateEntrepreneursat Work. Allen & Unwin, London.
( 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

CLIMATEFOR WORKGROUPINNOVATION

257

Behavior,
Kelloway,E. K. (1995).'Structuralequationmodellingin perspective',Journalof Organizational
16, 215-224.
Kelly, P. and Kranzberg, M. (Eds) (1978). TechnologicalInnovation:A Critical Review of Current Knowledge, San Francisco Press, San Francisco.
King, N. and Anderson, N. R. (1995). Innovationand Change in Organizations, Routledge, London.
Kline, P. (1986). A Handbook of Test Construction:Introductionto PsychometricDesign, Methuen, London.

Koys, D. J. and DeCottis (1991). 'Inductivemeasuresof psychologicalclimate',HumanRelations,44,

265-285.
Kozlowski, S. W. J. and Hattrup, K. (1992). 'A disagreement about within-group agreement:Disentangling
issues of consistency versus consensus', Journal of Applied Psychology, 77, 161-167.
Loehlin, J. (1992). Latent VariablesModels, Erlbaum, Hillside, N.J.
Long, J. (1983). ConfirmatoryFactor Analysis, Sage, Beverley Hills, California.

Marsh,H., Balla,J. and MacDonald,R. (1988).'Goodness-of-fitin confirmatoryfactoranalysis:Theeffect

of sample size', Psychological Bulletin, 103, 391-410.

Mulaik, S., James, L., Van Alstine, J., Bennett, N., Lind, S. and Stilwell,C. (1989). 'Evaluationof
goodness-of-fit indices for structural equation models', Psychological Bulletin, 105, 430-445.
Nunnally, J. C. (1978). Psychometric Theory, McGraw Hill, New York.
Patterson, M., West, M. A. and Payne, R. L. (1992). 'Collective climates: A test of their socio-psychological
significance', Paper presented at the Academy of Management Conference, Los Angeles, 1992.

Payne,R. L. (1990).'Madnessin ourmethod:A commenton Jackofskyand Slocum'spaper'A longitudinal

study of climates'. Journal of OrganizationalBehavior, 11, 77-80.

Payne, R. L., Fineman, S. and Wall, T. D. (1976). 'Organizationalclimate and job satisfaction:

A conceptual synthesis', OrganizationalBehaviourand Human Performance', 16, 45-62.

Pettigrew, A. (1985). The AwakeningGiant: Continuityand Change in ICI, Blackwell, Oxford.
Pritchard, R. D. and Karasick, B. W. (1973). The effects of organizational climate on managerial job
performance and job satisfaction', OrganizationalBehaviorand Human Performance, 9, 126-146.

Reichers,A. E. and Schneider,B. (1990).'Climateand culture:An evolutionof constructs'.In: Schneider,

B. (Ed.) Organizational Climate and Culture, Jossey-Bass, San Francisco.

Rentsch,J. (1990).'Climateandculture:interactionandqualitativedifferencesin organizational

meanings',
Journal of Applied Psychology, 75, 668-681.
Rogers, E. M. (1983). Diffusion of Innovations,The Free Press, New York.
Rogers, E. M. and Eveland, J. D. (1978). 'Diffusion of innovation perspectives on national
R&D assessment: Communication and innovation in organizations'. In: Kelly, P. and Kranzberg,
M. (Eds) Technological Innovation: A Critical Review of Current Knowledge, San Francisco Press,
San Francisco.

Rousseau,D. M. (1988). 'The constructionof climatein organizationresearch'.In: Cooper,C. L. and

Robertson, I. T. (Eds) International Review of Industrial and OrganizationalPsychology, Vol. 3, Wiley,

Chichester, pp. 139-159.
Schmidt, F. L. and Hunter, J. E. (1989). 'Interraterreliability coefficients cannot be computed when only
one stimulus is rated', Journal of Applied Psychology, 74, 368-370.
Schneider, B. (Ed.) (1985). 'Organizational behavior', Annual Review of Psychology, 36, 573-611.
Schneider, B. (1990). 'The climate for service: An application of the climate construct'. In: Schneider, B.
(Ed.) Organizational Climate and Culture, Jossey Bass, San Francisco, pp. 383-412.
Schneider, B. and Reichers, A. E. (1983). 'On the etiology of climates', Personnel Psychology, 36, 19-39.
Schroder, B. (Ed.) (1990). Organizational Climate and Culture, Jossey Bass, San Francisco.
Schroeder, R. G., Van de Ven, A. H., Scrudder, G. D. and Polley, D. (1989). 'The development of
innovation ideas'. In: Van de Ven, A. H., Angle, H. L. and Poole, M. S. (Eds) Research on the
Management of Innovation: The Minnesota Studies, Harper & Row, New York, pp. 107-134.
Siegel, S. M. and Kaemmerer, W. F. (1978). 'Measuring the perceived support for innovation in
organizations', Journal of Applied Psychology, 63, 553-562.
Tjosvold, D. (1982). 'Effects of approach to controversy on superiors' incorporation of subordinates'
information in decision making', Journal of Applied Psychology, 67, 189-193.
Tjosvold, D., Wedley, W. C. and Field, R. H. G. (1986). 'Constructive controversy, the Vroom-Yetton
Model, and managerial decision-making', Journal of OccupationalBehaviour,7, 125-138.
Tucker, L. and Lewis, C. (1973). 'A reliability coefficient for maximum likelihood of factor analysis',
Psychometrika, 38, 1-10.
Uttal, B. (1983). 'The corporate culture vultures', Fortune Magazine.
? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)

258

N. R. ANDERSONAND M. A. WEST

Wall, T. D. and Lischeron, J. H. (1978). Worker Participation: A Critique of the Literature and Fresh
Evidence, McGraw-Hill, Maidenhead.
West, M. A. (1987). 'A measure of role innovation at work', British Journal of Social Psychology, 6, 83-85.
West, M. A. (1989). 'Role innovation in the world of work', BritishJournalof Social Psychology, 8, 305-315.
West, M. A. (1990). 'The social psychology of innovation in groups'. In: West, M. A. and Farr, J. L. (Eds)
Innovation and Creativity at Work: Psychological and Organizational Strategies, Wiley, Chichester,
pp. 4-36.
West, M. A. (1995). Effective Teamwork,British Psychological Society, Leicester.
West, M. A. and Anderson, N. R. (1992). 'Innovation, cultural values and the management of change in
British hospitals', Workand Stress, 6, 193-210.
West, M. A. and Anderson N. R. (1996). 'Innovation in top management teams'. Journal of Applied
Psychology, 81, 680-693.

West,M. A. and Farr,J. L. (1989).'Innovationat work:psychologicalperspectives',SocialBehaviour,4,

15-30.
West, M. A. and Farr, J. L. (Eds) (1990). Innovation and Creativity at Work: Psychological and
OrganizationalStrategies, Wiley, Chichester.
Williams, L. J. (1995). 'Covariance structure modelling in organizational research: Problems with the
method versus applications of the method', Journal of OrganizationalBedavior, 16, 225-233.

? 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 235-258 (1998)