Review of Public Personnel Adm1nistration

spring 1999

vol. xix

- no. 2

Editorial

Scaff

PIIbIiaber
LDouaWDobeon use lnadrute olPublic Affirin EcIitor-la.cbW NkhoIu P. Lovrich,Jr. Wuhineron Srate Univeniry MaaaciDlEdftor SteVen W. Hays Univeniry of South Carolina

Seaior Editorial A'"'-Y

Board

JS. Bowmaa FIorid8 Scate Uniwntcy DouWICIiaper FloridaIncemadon8IUntv. DmdH. Ro8nbIoom AmericanUnMmicy O. GIeaa SC8bIAr\tnaton.VA 'I'I18IIJ. 'J'hoMp.oo SUNYIe Alban, EditorIal AdvI8ory Bo.rd C8dyaBMa UntveniryoiPlcabulah

Highliglw

E_a.-

UntvmiryofCenll1llFlodda

5
18

Editon
Richard C. Kearney E.tCarolina
Cbarlie

N.JoeepIa c.,. AriIonI~ UnIvenky CMhyC- Pon~Auchority

DeDak D8Iey NCSU MRaIeIan AIIIIIoDeNW Deaai8 DNe.ac RobercEllioa t.I8k ~ Rutam Uniwnitv UniftnicyofWl.anin Auburn Untwnity Univenicyof Connecdcuc

u.lna Performance Measurementin Human ResourceManagement

A SlII'Vey <iU.S. Counl:ies Evan Berman. Jonathan West

61 Profe8llional Notes

Potential Consequences of

Community.Oriented Policing for Civil Liability: Is There A Dark Side to Employee Empowerment?

Univeniry

B. T yet" olSouth Carolina

and XiaoHu Wang The Kirton AdaptationInnovation Inventory Validity Issues. Practical Questions Michael &hic. Eric Davis and RobeTtCunningham

John Worrall and Ricky Gutierrez

Universiry

Pro_ioaaI Notes Editor JackRabin Pc:rmsylv.nia State Universiry Book Review Editor Step~ Win
Boise State Univeniry Publicatiooa Dinoc:tor

Ge..ldG.bria NonhemIllino8Un1wnicy C. BllueGnhala use acColumbia

}oWy E. Guy FIodda Scare Univenicy

71

Precursors to Human Resource Planning in the Public Sector

Mario Martinez

PIIIridI"""" Syncu.e Univenicy J.I!dw8rdJCellouP UntvmiryofGeoqia

~M.

Whit6ekl

CamlLiaerSanM-. TX ~Manouo TOWIOIISraeeUniYenIty JobaN.a-fim Uniwniryofx-

Univeniry of South Carolina Subecriptioaa Freda AtkiNon Univeniry of South Carolina

s.n-~
Feh A. Nipo IJo,d Nipo

UniftnicyolArltar..
Athena, OA GeorgiaSraee Univenicy

32 50

Continuity and Change in Public Personnel Administration

Stephen Stehr and Ted Jcmes

77 Contributors 79
How to Join SPALR

JL Peny Indi8n8 Univenicy Gary Pokomy ElCcrrico,CA Notma lUcc:ucd SUNY ac Ahmy Gary RoI8t8 AbaSebeeia FairleiahDicldNoa Uniwni&y Calibn8SraeeUnivenity

SupervUon' Perceptions of the Performance of Cooperative Education Employees Working in Federal Agencies

J-ySWriu UntvmiryofPkabuJlh ADIDIIIoSia.oe UnivenicyollDinoia RoaaIcI D. s,t.Ia San.Joee Scare UnlYeniry r..- Selia \.JnMnityofM8louri JonMt- Weac UniwnlryolMiImI HoycN. Wheeler USCacColumbia

James Douglm and Gene Brewer

The loumal of the Secnon on Pe""mnel AdminiStration lahor Rel.non.

and

of the American Society (or PuNk AdmlniAtnttlon.

The Kirton Adaptation--Innovation Inventory

Validity Issues, Practical Questions
MICHAEL BaBIC, EMMANUEl COllEGE ERIC DAVIS, UNIVERSITY OF CALIFORNIA, RIvERSIDE ROBERT CUNNINGHAM, UNIVERSITY OF TENNESSEE, KNOXVILLE
The Kirton AdaplallOl1./nnowUon Irwtntory (KAI) is dcsipcd co "",as"", proptnsi" coin"""""......... pm" i" 10adapI, a pmonali" This ani& p",mu dimtnsian c/aimtd .ignificant.for wndmmding and buiIdint organitalianal G !niL! of ""lidiry chtdu for u.. KAI. r<porn two wmmpiloNJ findings, adapI..., '" 0 in""""IO" is It.. offtClM ~ ,{fWMtvU,

lhen It'"~ Rin""', hy~.is lha. d U4m compo.tJ of o~ rhan d ,,"'" balanad on rhis dimtnSion.

rganizations, public and private, face turbulent and uncertain environments. Downsizing, rightsizing, and contracting out threaten the jobs of public sector managers, supervisors, and front line workers. Restructuring, decentralization, and empowerment are recommended for meeting the challenges of fast-changing environments. However, adjusting blocks and lines on an organization chart, or creating task forces and cross-functional work groups, will not automatically transform an organization's behavior. Perhaps working with people, and allowingchange to emerge from the natural dispOsitionsof the group members,can bringorganizationaleffectiveness. The Kirton Adaptation-Innovation Inventory (KAI) purports to measure a natural disposition, an individual's propensiry to innovate. The 32-item measure was published in Kirton (1976) and has been tested and validated in several languages and cultures (Tullett & Kirton, 1995; Tullett, 1997).' Given the importance of Innovation to the public sector and the dearth of

alternative research instruments to measure the propensiry to innovate, the KAI merits investigation. If the KAI is a valid measure of Innovativeness, and the theory Issubstantiated that groups diverse as to KAI scores are particularly effective, then the KAI can be an important management tool for human resources managers In composing the makeup of their work force. This research reports a variery of validiry checks for KAI, and then tests the theory that tearns diverse on the KAI perform more effectively than teams domina red by either adaptors or Innovators. These two hurdles constitute a strenuous test for the KAl.
THEORETICAL FOUNDATION AND RATIONALE FOR THE KAI

The KAI Is based theoretically upon requisite variery, a principle spanning the physical, natural, and social sciences (Ashby, 1956). Requisite variery assumes that dlversiry within a system enhances the likelihood of system success in the face of external challenge. Research substantiating the requisite variery principle appears in the lit-

erature on communication (Thompson, 1991), social relations (Sirgy, 1990), alcohol rehabilitation (Sterman, 1990), education (Foden & Mason, 1986), leadership (Fisher, 1985), and organizations (Follett, 1918/1998; Miller, 1990; Weick, 1987, 1991). Follett (1918/1998) insists that the diversiry of neighborhoods strengthens both the effectiveness of their organizations as well as democracy. In organization theory, Miller (1990) adopts requisite variery principles In describing the Icarus Paradox, where the organization's strength, which generated the Initial success, eventually leads to the organization's downfall. The specialized organization filled with similarminded individualsremains effectiveonly so long as its abilities, skills,and products connect with demand in the environment. Without requisite variery, the organization lacks sensitiviry to environmental changes. When the stakeholders demand different services, an organization with focused, specializedskillsmaybe unable to meet the new challenge. The organization having requisite variery more easily senses and adjusts to the changing environment. The KAI connects requisite variety theory to the management of an organization's human resources. Everyone has a decision-making style. Decision-making sryle Is stipulated as an integral aspect of personaliry-stable over time, incident, and culture and uncorrelated with cognitive capaciry,cognitive techniques, or personaliry traits (Kirton, 1989,chap. 1;Clapp, 1993; Murdock, Isaksen & Lauer, 1993).One's declsion-making sryle ranges on a continuum from innovative to adaptive. The extreme innovator starts each issue on a fresh page as a new opportunity; the extreme adaptor depends heavily on precedent and past experience to craft an appropriate response to
KIRKTPN

the issue. To the layperson, creativity and Innovativeness are synonyms,both different from adaptiveness. Kirton (1978) offers a different perspective. Both innovators and adaprors are creative. Adaptors are creative within a narrow range, seeking minor improvements, initiating changes that lie near current organizational practices, and pushing boundaries incrementally. A manager with an adaptive sryle accepts established routines as appropriate, makes marginal changes within existing boundaries, and is seen as conforming, safe, and dependable. Innovators have a different mindset. They allow creativity freer rein and accept few givens. Innovators do things differently rather than necessarily better. They may even reconstruct the paradigmatic frame within which the organization operates. Fleenor and T.\ylor(1994) from the Center for Creative Leadership, analyzing a data base of 12,115 managers, report KAI scores positivelycorrelated with the MBTI Creativiry Index and the CPI Creativiry Scale. Similarly, in a smaller study, Jacobson (1993) found correlations between the KAIand the MBTI intuitive and perception dimensions. The manager with an innovative sryleregularlyquestions established routines and may be seen by adnptive manalotersas undisciplined and insensitive to others' consensually agreed ways (Kirton, 1976; 1989).
According to the theory, KAI facilitates

the construction of diverse teams. The ideal team or organization includes both employees desiring substantial change (innovators) and employees preferring minimal change (adaptors). If dominated by either adaptors or innovators, an organization risks diminished effectiveness (Kirton, 1978). The adaptor-dominated organization may offer a product or service which does not meet the customer's neWneeds. The innovatorAI1ArT'ON,lNNOV A T'ON

1M RI'\'IIIX' ('I' rl 'I'l.Ie rERSONNEI AOMINISTRATION

SrRIN<1IQ</9

C\INNINnHAM FTAI.. ,q

1i

..."

controlled organization may produce goods and services which are superior to alternatives, but clients are not yet ready to accept the new concepts. A cohesive dedslonmaking team Including both adaptors and innovators is more likely sensitive to environmental challenges.
KAI: TIiEORYANDOPERATIONAUZATION Assuming that personality II composed of abilities. traits, and styles, the KAI measures cognitive style--()ne's preferred or characteristic manner of processing information. Conceptually, KAI has three dimensions: I. Rule/Group Conformity (R) 2. Efficiency (E) 3. Sufficiency vs Proliferation of Originality (0) I Rule/group conformiry (R) is the degree [0 which one works within the accepted structures or rejects critical elements of such structures. Innovators may overlook or ignore pressures to conform to consensual views on what is needed and how to get it; adaptors are more likely to abide by system dictates (IZ items). Examples of conformity items include "A person who likes the protection of precise instructions" (Q29), and "A person who prefers colleagues who never rock the boat" (Q32). Efficiency (E) exemplifies Weber's notion of the legal-rational bureaucracy. The efficient manager emphasizes precision, reliability, and efficiency. Innovation is discontinuous to the existing system, thereby to be rejected as hindering efficiency (7 items). Sample items for efficiency include "A person who is thorough" (QI4), and "A pers<'n who is consistent (QI7). SuffICiency liS proliferation of originaliry

tend to lie within present practices and procedures (13 items). Representative Items are "Aperson who has originalideas" (Q21), and "A person who Is stimulating" (QI9). Validation studies have supported Kirton', factor structure (Bagozzi & Foxall, 1995; Murdock et aI., 1993; Clapp, 1993; FoxaU & Hackett, 1992). Individuals respond to items by indicating their level of agreement on a 1-5 scale, with scores in a theoretical range from 32 to 160. Empirical results show a mean score of95; with 67% of the scores (one standard deviation) lying betWeen 79 and 113 (KAI Report Back Form 1985). In multi-factor space, each of the 32 items reportedly loads on the appropriate dimension 83% of the time (Kirton, 1989, chapter 1).
TEST POPULATION

(0) taps the number and scope of innovative ideas the subject generates. The innovator regularly reinvents the wheel; the adaptor offers fewer ideas and these Ideas
20 RfVIFW PF 1'\JlIlIC PERSONNELADMINISTRATION

Our evidence to test KAI II drawn from three respondent-groups: middle-level state managers, international managers studying in the United States, and university students. Middle-level Slate managers (N=2OJ). From 1983 to 1990 Cunningham coordinated an annual three-week management development program spread over three months for 25-28 middle-level state managers. Participants filled in the KAI instrument prior to orientation and again three months later, on the next to last day of the program. KAlscores weredebriefed after the second test only, so the corrupting effects of test-retest carry-over should be minimal. In the fall of 1988, participants from the 1983 through 1987 management classes were mailed the KAIand asked to fillit in. Therefore, from this training program the 1983, 1984,and the 1988-1990classeswere tested twice, and classes 1985-1987were tested three times. Personal interviews with participants occurred from 1984-1989, and a
SPRING 1999 ,,-c,

quasi-experiment was conducted in 1985. lnumationalmanagers (N= 12Z).Since 1988 Cunningham has taught at training programs held in the United States for international managers. Participants in these programs are sponsored by the U.S. government or by foreign governments. Middlelevel public managersfrom around the world completed the KAI and were debriefed. StUdmu (N"262). From 1985 to 1988 various student classes filled In the KAI. Some were beginning university students, some were upper division undergraduates, and some were graduate students. This research project did not emerge from a pre-conceived design. Initially,information was collected so students and managers could learn about themselves and share their experiences with cohorts. The respondents constitute accidental populations. The evidence presented here testing the validityof the KAIconsists of fragments. No single piece of evidence can or would be convincing alone. The strength lies in the consllteney. PerhapssurpriJingly,all the frag- . ments point in the same direction. Triangulation,bringing multiple data sets to bear on a single theoretical problem, is applauded In evaluation research as a method for assuring validity of the findings (Campbell, 1979, Filstead, 1979; Trend, 1979).
TESTING KAI

evidence (Messick 1988). Validity can be categorized as construct, content, and criterion validity. J

CONSTRUCT VALIDITY The KAI concepts of originality, efficiency, and rule conformity are theoretically Independent of each other. Sufficiency of originality need not imply lesser or greater efficiency or rule conformity. Nor should one's high efficiency necessarily Imply lesser originality or lesser conformity.. The KAI is operationalized by asking respondents to indicate their ease or difficulty in presentIng themselves, consistently, over a long period, in various situations (the 32-item instrument; Kirton 1989, chap. 6). For the three dimensions (subscales) plus the overall KAI to meet Kirton's criteria for construct validity, the individual items should load most heavily on the predicted subscale, in the predicted direction, and the three subscales must be positively, but modestly, positively intercorrelated. Factor analysis of

the data describedaboveconfirmthese expectations. The results of factor analyses include: factors, which are estimations oflatent variables underlying a data set; factor loadings, which are correlations between the original variable and the underlying factor (Hair, Anderson & Tatham, 1987, p. 249); and measures of the variance explained by the factors extracted from a given data set. Factor loadings are considered significant if they are greater than :t .30 (Hair, Anderson & Tatham, 1987. p. 249), and a variable is associated with that factor whose "loading" isthe highest for that variable. For example, QZ's Rules Conformity loading is .35, its efficiencyloadingis .27, and its originality loading is .21.5 From this, we say that Q210ads on the firstfactor,whileQ2 I (loadings .04, -04, .70, respectively) loads on the
ClINNINOHAM FT At. 21

ValidityII defined as how well has been measured what was intended to be measured (Guion, 1980) and as the accuracy of inferences about test scores (Tenopyr, 1977). Inferences are based upon a theoretical/ truth component-the subjective judgment that the concept In question Isreasonably represented by the operational tests which measure it; and a pragmatic, utilitarian component-the scoresderived from tests fall into an Interpretable pattern based on empirical

KIRKT0N AOAPTION.INNOVA TION..

"

third factor, originality. Table 1 presents the factor loadings and the amount of variance explained by those facrors. All variables for a given factor load at greater than .30, and in every case, the next highest loading score is statistically significantly smaller than the score for the factor chosen as its primary loading dimension. Twenty-five of the thirty. two items (78%) load strongest on the predicted fac. tor, with three factors accounting for 33% of the variance. Such a result supports the ,'"nc"nli,'II, 1(11 IhI"\'('.dimt'nNiolllll NtrllCtllre. Since explained variance is a function of the degree to which the items are measuring the same thing, accounting for a high proportion of variance would suggest item redundancy. On the other hand, too little variance accounted for indicates the absence of a unique, clearly defined dimension. With each of the three factors accounting for a respecrable proportion of the variance, the three-factor solution is a reasonable inter. pretation. The subgroup personality mea. sures also correlate moderately with each other and with the KAI overall score, indi. eating that subscales are tapping a common underlying phenomenon; see Table 2. CONTENTVALIDITY Concl'T1lI!alidity checks whether all significant aspects of the domain are included in the measure. This is difficult to detennine, for there are not clearly defined behaviors to match adaptive and innovative attitudes, except perhaps at the extremes. Dimensions of Kirton's conceptual meaning were operationalized by choice-pairs of phrases to create an alternative measure (Altkirt); see Appendix. Respondents chose the one statement from a two.statement pair which hener describes himself or herself. With a Pearson's r of .65, KAI and Altkirt appear to he tapping the same dimension; see Table
H RI\"IFW or rl JR!Ie rFRSnNNH. ""MINISTRAnON

3. Altkirt does not dispute a content validity claim.

CRITERION VALIDITY

TABLE 1. Kirton Adaptation Inventory, Rotated Factor Matrix (N-587) RULES (R)
EFFICIENCY(E)

ORIOINALITY (0)

CriterionI!alidity can be established by showing KAI scores appropriately relating to other measures, either as predictors or as measures of similar content. Here, KAI scores will be checked for (a) consistency with a content analysisof leadership stories told by managers, (b) stability over time (to test the assumption that KAIscores are perNonlllity mCIISllreswhich do not (hRnlle eM'

ily), and (c) consistency when managers carry out a task within a group fairlyhomogeneous with respect to KAI scores. stories. In Content analysisof leadership the 32.item KAI, Kirton sets the agenda by defining the dimensions and constructing the items. Respondents assess their com. fort level with Kirton's items. The instruc. tions read: "How easyor difficultdo youfind it to present yourself, consistently, over a long period of time as " In 1985 each middle manager was in. terviewed privately during the management development programand asked to tell leadership stories about situations in which he or she was involved. The interviews, which lasted from 20 minutes to over an hour, were audiotaped and transcribed. These stories are assumed to represent typical ways that the manager presents self. Because the stories which managers tell are self.generated and personal, the respondent is obviously describinga management stylewhich iscomfortable to him or her.The stories bridge the cognitive stylelbehavior gap, for the stories are self.reports of a manager's behavior. The manager's comfort level is reflected by the customaty behavior present in the story. The coder's challenge is to determine whether the manager's comfort level lies in adapting or in innovating.
SPRING 19Q9

Q24 Q29 Q32 Q27 Q30 QIO Q20 Q8 Q28 QIJ Q7 Q2 Q14 QI7 Q25 Q22 Q4 Q33 QI5 Q9 Q6 Q21 QI9 QIl Q23 QI8 Q26 Q5 Q12 Q31 Q3 Facror 1 2 3

0 R R R R R R R E 0 R R E E E E E R E R R 0 0 0 0 0 0 0 0 0 0
Eigenvalue

.71 .64 .60 .53 .50 .48 .44 .42 .41 .40 .39 .35 <.05 <.08 .30 .19 .09 .18 .18 .15 .08 .04 -.05 .13 .12 .26 .34 .01 .04 .10 .25

-.05 .17 .00 .22 .21 -.02 .14 .37 .24 .11 .27 .27 .67 .65 .58 .54 .53 .49 .47 .47 .33 -.04 -.03 -.1\ -.\2 <.07 .17 .15 .35 .24 -.25 Cum Pct of Variance 17.7 27.4 H.2

.05 .\2 .10 .08 .28 .25 .00 .03 .15 .1/1 .15 .21 -.12 -.04 .06 -.16 -.05 .31 .09 .24 -.04 .70 .70 .5/1 .57 .51 .48 .48 .42 .42 .39

5.66 3.11 1.85

Nou. In theKAI.itemI I, 0.dummy..totementwhichi, not.cored;,coredItem,Are2throughJJ.

KIRKTON AnAI'TION.INNOVAT'\N... ('\INNINO!1t\M FTAI.. ZJ

TABLE Z. Pearson's r among Subscale. (N-S87)

RULES R 0 KAI ---...ORIOINALm EmCIENCY

.-1

.49 .88 .77

.44 .13 .61

i.~

TABLE 3. Pearson's r: Altkirt with KAland Subscalel, Student Sample

RULES
ORIGINAUTY
EFFICIENCY

KAI
.65

Alrkirt N=R4

.62

.50

.40

TABLE 4. KAI Mean: Pre- and Post-Training by Year, State Managen 1985 1986 N=23 104.8 104.9 1987 N=23 105.7 108.2 1988 N=27 98.6 96.7 1989 N=25 97.6 96.2 1990 N"'27 99.8 101.6

courage empowerment, risk-taking, and innovation. This intensive residential program took state managers away from their work sltel Into a campus environment where they worked with practitioners and academIcs In an experiential program incorporatIng both Individual and group projects. If the program Increased propensity to Innovate. participants should have gravitated toward the Innovative pole of the KAI In the three-month Interval betWeenorientation and program conclusion. Enthusiastic evaluation of the program by participants suggests that their KAI scores should have moved toward the innovative end of the spectrum; yet aggregate KAI scores did not budge. see Table 4. The slight changes are random. This evidence supports the argument that training cannot easily change KAI scores. Perhaps the messageof the training program takes longer to soak in. In that case, scores should rise in one to three years after the management development experience. Some participants filled In KAI protocols one to three years after the program was concluded. The results are found In Table 5. Ai8l'egate scores remain stable across time, which supports the originalfindIng that training programs do not change KAI scores. However, racial differences emerge. Table 6 contrasts the mean KAI score before and after training. Because of the .mall number of black respondents for whom information from three time points is available, the data are pooled for all respondents from 1985 through 1987. Black managers before training are lessinnovative than their white peen. At the end of the training program, KAIscores for black managers remain unchanged. However, after one to three years, innovation propensity among black respondents hu Increased to the point that

N=24
rr".rraininl:
P",r-rraining

100.8 99.9

Kirton provides 14 characteristics of adaprors. and their corresponding innovator characteristics (Kirton.1989.pp. 8-9). Respondenrs' stories were read and coded the-

matically by Bobic and Davis separately. Each theme dimension was allocated one point. so each respondent's score ranged from I to 14. The coders then discussed and justified their assessments of each respondent's score on each of the 14dimensions" On the initial pass. 80 percent agreement was achieved between the coders. Respondents who received different scores lay close to the midpoint. Consultation resolved those disputes. Changeooer time. Kirton (1989, chap. i) insists that KAI is a measure of personality.therefore stable over time and not easily changed. Van der Molen (1989) and Clapp (1993) offer supporting evidence. Oold14 Rf\'lfW Of M.'III.I(' PERSONNel AOMINISTRATION

smith (1989) reports that training programs appear to have no impact on KAI scores. KAIstability has important Implications for organizations which Invest In training programs to teach people to be innovative. If training programs can teach innovative behavior.then not only does KAItheory err in assuming the adaptation-Innovation dimension as an Integral aspect of personalIty. but organizations are spending money wiselyas they train people to innovate, assuming that their organization lacks Innovativeness. However,ifKirton and others are correct on this issue. organizations seeking to change their culture should consider recruitment, replacement, or building diverseteams rather than training to achieve the needed balance of managerialstylesOver

thewhole cognitive stylerange.

SPRING 1999

State manager training sought to en-

they are as innovative as their white colleagues, who remain unchanged. This is a small data set (7 black managers and 43 white managers) from which to extract a firm conclusion. but the findIngs offer an Idea to explore. Perhaps minorities have a narrower range of behavIors acceptable to majorities. Successful minority managers are sensitive to the parameters of group norms, and stay within them. The training program encouraged managers to go beyond the boundaries. The reinforcement of superiors, peers. and instructors that innovative behavior 15acceptable. even desirable. may have fallen on a fertile field among minorities, who broke barriers to achieve these management positions and who may have constrained their natural Innovative inclinations in order not to risk damaging their careers. Work environment may Influenceone'. Innovative or adaptive attitude. The Influence of work environment Is reported significant In Holland. Bowskill & Bailey (1991), who found a regression to the mean .where new employees' KAI scores differed from the mean of the organization, but no change where new employees' KAI scores did not differ from the mean. We tested the influence of work environment on KAI scores. The governor in office fromJanuary 1979 until January 1987 qot only encouraged training to Improve management, he also preached to cabinet members and other top level officials the philosophy of empowerment and devolution of authority. Cabinet members were expected to work with their own line and staff to create more effective state government. The commissionermade the decisions in the department; the governor's staff members were subordinate. and stayed out of department matters unless Invited In. Informal conversations with middle managers during
CIINNINGHAM ET At. n

KIRKTON AOAmON.INNOVATION...

'j' ..<

i.

TABLE S. Kirton Scores: Pre-training, Post-training and Follow-up.

TEST AOMINISTEREO

i~

1985 N=16 98.H 99.56 102.75

1986 N=18 103.28 105.11 107.61

1987 N=16 106.56 108.31 106.81

,IB<,

Pre.rraining
POM -!raining

I 10}-var follow-up

"Indu,te, onlv tho", ".10 manaRen who have ocarel for allthr.. observations.

_d___.
TABLE 6. Mean Kirton Scores and Standard Error Over Time by.Race. State Managers Only
PRE White Standard Error mack Standard Error 103.4 2.5 98.9 4.7 POST 105.3 2.2 98.9 4.7
FOLLOW-UP

N 43 7

105.8 2.5 104.6 3.9

the summer training programsindicated that participants understood the governor's messa~e and appreciated the opportunities for innovation offered by this managerial philosophy. The next governor adopted a contrary managerial philosophy. Policy decisions were made by the governor's personal sraff, and cabinet members were subordinate to staff on policy matters. Deviation from existing policy had to be cleared by commissioners through the governor's staff. Commissioner and middle manager innovation was not encouraged. The managerial environment emphasized adaptation rather than innovation. Several managers expressed their discomfort at having decisionsimposed upon them by the governor's staff.One cabinet member stated privately that he sometimes learned from the newspaper about decisions affecting his department. This management style differs sharply from the previous governor. Note from Table 4 that KAI pre-trainIt> RI'\'IFW t)F rlll\UC rERS0NNEt AOMINIS11lAllON

ing scores from 1983-1987 (first governor) are at or above 100. Pre-training scores for 1988-1990 (second governor) do not reach 100. The differences betWeenthe tWogovernors are consistent over the data set, and support a speculation that while training may have no effect on KAI score, the culture of the workplace appears to have an impact. An innovative workplace culture fosters higher KAIscores; an adaptive workplace culture encourages adaptation. Emphasizing the influence of the environment is compatible with the Deming argument (Bowman, 1994) that the workplace environment and task structure are significant sources for organization effectiveness. The inference from this finding is that to effect change, the organization executive should restructure the environment or redesign the job rather than flre employees. Sending people to training programs may teach specmc work behaviors, but does not appear to generate an innovative orientation. However, a suppressed innovative attitude can
srRINO 1m

be encouraged when the executive communicates by words and deeds that risk-taking and innovation are encouraged. The findings regarding the impacts of race and work environment on KAI scores were not predicted; they appeared as serendipitous findings in the data analysis. The speculations we offer that (a) a member of minority identity group or (b) working in a controlling environment constitute conditions which depress one's propensity to innovate constitute reasonable hypotheses, but demand further testing. These findings may seem at odds with the overarchlng theme of a stable, unchangin~ personality factor which affects how each person responds to the work environment. However, paradox rather than consistency is a reality both in life and in the work place. None of us responds identically to the same stimulus both at home and at work, nor on different days of the week. Quinn (t 988) argues that competent managers must be inconsistent, must respond paradoxicallyto the situations they face. One has a basic orientation, yet deviates from that consistency in specific' situations. There is no isomorphism betWeen personality And response to Adecision situation. Decisiongroups. Homogeneous groups of adaptors and innovAtorsshould arrive at decisions which reflect the modal characterisdcs of their KAIscore. A group of adaptors should arrive at an adaptive decision; a group of innovators should reach an Innovative decision. If the KAI measures adaptadon and innovation accurately, and if the theory that homogeneous groups of adaptors and irmovators will arrive at less effective decisions, then the following quasi-experimental research task should support that hypothesis. The 1985 management training class was split into three groups, and KAI score

(in addition to gender and department distribution) was a major criterion in dividing the participants.7 High KAIscorers and low KAI scorers were assembled into homogeneous groups. A third group wascomposed of scorers falling near, and on either side of, the KAImean of98. The three groups were charged with the same task: over the next three months you are to develop a proposal for addressing the disposal of nonhazardous solid waste for the state. Speakers were brought in to address the topic; participants visited several solidwaste disposal sites, And each group spent many hours In Indlvlduul Rnd I(roup research. Because this report would be presented In both oral and written form to the state's Department of Health and Environment, participants were committed to an outstanding product. The commissioner of the department would attend the oral presentation. The groups worked on their proposals all summer. During the last week of training, each group in a "dress rehearsal" presented its recommendations to a panel of four solid waste experts. The experts commented on each of the three plans, and
evaluAted the plans 910nl(several dimensions

usinga 1-10scoringscheme. Innovativeness WAS meAsuredby the dimension "departure from standard practice." An adaptive score approaches the "standard practice" pole; an innovative score approaches the "sharp departure" pole. The project teams scored as predicted by the KAI theory. Aggregating a mean among the scoresof the four experts, the innovator team achieved a mean score of 7, the adaptor team scored 3, and the middle team scored 5. Again, the validity of the KAI measure is confirmed. When innovators get together, the decision willbe innovative; when adaptors get together, the decision will be adaptive. Also in line with diversity theory, the experts evaluated the
CliNNINGUAM n AI.. 27

KIRKTON APAI'TION.INNOVA'IION...

TABLE 7. KAI Group Score and Innovative Group Behavior, State Managen
INNlW A'I\)RS GROUP

MllmLES GROUP

ADAPTORS GROUP

Mean KAI score *Departure from OI:mJard practice

I2Z 7

97 5

79 3

.Composj.e expen score (I =srandard practice. 10=sharp departure).

middle group as offering the best project plan. These three tests of criterion validity support the KAI as a valid measure of decision-making style, corroborating previous evidence for construct and content validity. Scores remain stable over time, and Kirton's hypothesis that balanced teams outperform teams nf homogeneous adaptors or innovatOrsis confirmed.
POLICY IMPLICATIONS

main a factor influe~cing a subordinate's propensity to innovate.

SUMMARY

Organizations face the paradoxical challenge of maintaining the satisfaction level of their current stakeholder base while seeking new opportunities and more effective waysof developing, producing, and delivering the product or service.Becauseindividuals differ in their propensity to welcome or seek change, the work group which incorporates and celebrates a wide spectrum of employeeorientations to change willbe prepared !>othto hold onto the best of the old and to introduce the new. Byassessingindivi,llIals "n Ihe a,IAptlltinn.innovlltion di. mensi"", KAI can assist management in building effective work teams. However, one's KAI score may also be influenced by one's work environment or minority social status. Individuals take cues from management and peers. KAI can be useful in enhancing innovative problem solutions, but an executive's demonstrated behaviors reltI RF\'IE\\' ("If rlJRue: PERSONNELADMINISTRATION

Based upon multiple data sets and multiple methods, this research finds the KAI to be a valid measure of the adaptation-innovation dimension of managerial decision style. Factor analysis supports tri.dimenslonality for the overall KAI, and confirms the Internal structure of subscales.Triangulation byemploying an alternative method of testing adaptation-innovation (Altklrt), a content analysis of manager stories, and expert evaluation of three groups' project proposals argue that KAImeasures the innovationadaptation continuum. KAI appears to be tapping a stable personality dimension. A significant training experience did not significantly elevate (or lower) innovation scores. While the data indicate that one's KAIscore remains stable over time, minority group status or the organization culture communicated by top manaKement likely nudge. scores to come in line with the executive's predisposition. In each test, Kirton's theory Is confirmed. In none of the tests does the KAIor Kirton's theory fail. Seeking organizational effectiveness through work-group diversity along the adaptation-innovation dimension receivessupport and iscompatible with prior research showing that organizational effec-

tiveness is produced by the requisite variety principle (Miller, 1993; Watson, Kuman & Michaelsen, 1993;Weick, 1979). The practical management strategy for organizational effectiveness is to recruit or build diverse decision teams. If the adaptation.lnnovation dimension Issignificantfor supplyingneeded variety for an organization, and i( the theory of requisite variety applies to organizations, then (other (acton being equal) organizations which use managers' KAI scores for build. ing diverse work teams willbe more successful than organizations which ignore such information about their employees when building teams. Notes
'More .han II0reoeRrch articinmakllllluoeo(the KAI are reponed In the P8ycINFO 1996.lmdatabue. I TIle II.. of Items comprill"8 each luboeale can be found In Klnon (1976). 'Oulon (1980) and Mellick (1988) Illte that validity II more a unitary than a lrinltlry concept. TIle catel/Ory oeparatlon8 are. uaed here for orlfllnlzR.fonal PUrpmel. . Conceptually, the orlllnRllty. efficiency and rule con(ormlty luboeale. are orrholOnal (uncorrellted) In multidimensional Ipace. and the overall KAlil linear In two-dimensional space. In rwo.dlmen.lonallpace the Iteml form a .1"8le me..ure; 10 delplte the orthotronalltruc.ure of the three component (acton. .he"" .ub.scorel will have modes. pooluve In.ercorrelatlon8. By dellnltlon, orthogonal facton are uncorrelated. Howeve.. when lpeciAc Item. are .ummed to create lubocalel. the (ac.or loadl Rnd effectl of non luboeale items are removed, .herebv Rllowlnll (or the I'OIIIbility of either pool.lve or nelflltlve correl"lon. alllOtll lubocalel. In thll Cloe. al expected, the lu,,"cole. lhawed modest pooltlve IntercorreIA.ionl. Thll mean. thaI the .uboeales are Independent. ~. related to the umbrellA dimension of adap'ation.lnnovotlon. , The dlffiirence berween the "'0 larBeIt loadlnlll I. .09. (the smalln. difference In our do.a) and 18"""IlIcAnt at ,,-.0202 (t- 2.304). . In .h.l. own penonal KAI lenre, nohic InJ DRvl. leore approximately one ..anJArJ d.vlRllon from .he mean In oppotite dlrectlonl. Thll dlvenlry be!Ween the coden reduce. the likelihood that a .Igniflcant theme II overlooked. and the con venation. (literally vehemenl argumenu) between the codell after their InlUal Independent jud(ll1len18lncreRaed the likelihood o( a valid Intetpfttation of themel from the srotie.. The different penpectiVei of Bobk: and Davll Ind .hat Impacl on lhe leorl"8 points up the Importlnc. of seeklllll dlverllty In oelectlng code... Validity II enhRnced by a thorough dlacuaion amotll coden with dUI'eri", perspective.. Scorillll

done by coden wl.h .imil.r ocore. on the dlmen.lons .elled mAYachieve reliability, vel .acriAce validity. , Only .he pr"ll'am coordlnAton knew of thl. 1110111' "'AIiAcation.

References
Alhby, w. R. (1956). In.mJ..crion10c,btmtric< New York: John Wiley. Balloul. R. &. Foxall, 0. (1995). Conltruc. validity and lenerall.abillty o( the Klnon AdAptation.Innova.'on Invenlnry. Ewro/Jtan }OIIma'of l'erslllld/ir,. 9 (3), 185. 206. Bowman.J. (1994). At IA".Analternotive to performance apprAI..1: Total quality mAnallement. Pwblic AdminiJrra.imI RevitUl, H (2). 129.116.
CAmpbell. DonAld. (1979). ~lIIeel o((reedom And .he CO"" Itudy. In T. Cook &. C. ReichArt (Edl.~, Quantirarille and qua/jra.IIIe mtlhodJ in tIIal /{m rt!tarch (pp. 47. 69). Beverly Hili.. CA: SOlie. Clapp. R. G. (1993). Stobtllty of cOllflltive .tyle In odul.. and 8Ome ImplicAtion., A longitudinal llUdy of the Kirton Adapta.lon.lnnovation Inventory. P.",ho~lgical Rtpan" 73 (}), 1235.1245. Ftlltead. W. (1979). Quolitative methodl.ln T. Cook &.c. ReichArt (Ed..), Quan.jratille and qwaUra.wt -thods in tIIal imI rmarch (pp. 33.48). Beverlv HIIII, CA: SAge. Fllher. A. (1985). Leadenhlp 01 medium: Treo'ing enmplexl'y In II"'UP communication re.earch. S""," GrowpB.havior, 16 (2), 167.196. Fleenor. J. &.Taylor. S. (1994). Conmuct validity ofthree lel(.report meo.ure, o( c.eativlty. Edwcarlonal and P.,cho~'gical M.alwrtmenl, 54 (2),46<1.470. Foden. L. &. MalOn, H. (1986). Who. price partlclpAtionl EdwcarJanal and Child l'J,choIoO, 3 ()), 230.236. Foxoll, O. &. HRckert, P. (1992). Th. (.ctor strucrure and c"",rruc. vAlidity of.he Kirton Adoptatlon .Innovatlon Inventory. PnsonaUryand Individ 1DifftrtnCtS,/3 (9). 967.975. Follen. M. P. (1918/1998). The ntwsra... University Pork. PA: PennlylvaniA S,... Unlv.,,11)' Pres.. G,ldsmlth. R. (1989). CreAtive style And pe""nality .heory. 'n M.Kirton(Ed.). Ada;tors and inn""".OTS (chap. 2). London: Rou.ledge, . Oulon, R. (1980). On trlnharlan doctrinel o( validity. Pmfmiorlllll'J,choloo, I I. (3), 385.398. Hair. J.. Anderton. R. E, &.Tatham, R. (1987). MwlriV<Jriare dala anal,,;, (2nd ed.). New York: MacmillAn Pubilihen. Holland. P.A.. Bowsklll, I., Bolley, A. (1991). Adap.on And
Innovarnn: Selection veroo. R,ptlrU, 1\11(I), 12/1 1.1290. IndOlc.l"n. Ps,chological

':,:

Jacoh.on, C. (1<191). C"lInhlve 'Ivle. of creallvl.v: Relollonl o( score. on .he Kirton Adap...lon . Innovallon Inventory and the Myero.Brlglls type Indicator among manaBers In the USA. P"choIoRical R./IOfU, 72, IIJ 1.11 36. KAI Repon Back Form. (1965). Kirton. M. (1989). Ada".ors and Innavarors London: Routledge. Kirton. M. (1976). Hove adaptors And Innovators equAl levell o( creatlvityl P':1"h%gical R."OTU, 42.695.698. Kirton. M. (1976). Adap.onand Innova.or., A description

SPRING 1m

klRKTON ADAmoN.INNOV

ATlON...

CtJNNINOHAM ET Al.

19

anJ ",,'a""c. Journal..f AppheJ rs~,h"I"I<.f. 6/. 6ZZ.6Z9. ~uhn. 1. (I<nO). TIle IIruc'ure of ",enufi, revo/u'ion! U'Kago: Unive"ity of ChICago Pre". Me""k. ~. ( I'mil). The once anJ futute issues of validity: A',e",ng the meanong and .nsequences of mea,,"ement. In H. Wainer & H. Braun (Eds.), Tell ..,/"1,,, (1'1'. }}.45)' A"""ares. Hill,Jale. NJ: Lawrence Erlhaum

~1III<-r.n. (1'19(\). The IWTU' pMaJo,. New York Harper Bu'one". ~1",I,I.~. (l'l()t>\. ~"""" A.I theory: hidence ""aringon ,I.,' "d,. 1< 1 :m,lt.."", ""ml~"i'ion issoes. IIri'i."

./",m",I..{ I'",'h"/,,,,':I. H7 (2), 241.254.

~1""I",k. ~1. h..k','n. S. & Lauer. K. (1993). Creativiry consisrency of Inventory. rr.""'IIg .m,I,he ,t..",hty anJ inrernal till' ~"ton AJartlOn.lnnova,ion P"oI",I"g,,',d

Rel'rlS. 12 (J). II ZJ. II 30.

011inll. R. (I'll'S). Bev"IIJ wtlO""/ maougemen!. San h.u"",o lo"ey.B:.... ~"I[\. I. (I'NC'). 'I;.ward a ,heory o( "",al relatio",: The "'g".,,,on analog lIelw"low/ Sdena, 350), 197.206, ~'<lm.lII. C. (1990). Arc vou the product o( my m"undeNanJong! Or the n.le o( ,orting mechanisms ,11\,1tI... "a,i< human program' on ,he ITeatment of :ol<,.hol"m. In C. St,'rmall (EJ.). Neuro.lmguilli, l'r",~,"nll1l11gm ab'/v'/15111treatmenl (1'1'. 125.140). New Y"rk. Haworth. '\en. '1'\'1.M. (1'177). Content 'C' ""ITuet confusion. Personnel P,," ""/"1::" 3(1. 47.54. rhollll""II.T (1991). P".I,'cl'cs, communication anJ nl'II "'" (or I II'allvIlV. Jo'm",lo}< :rf"'flW Be/uwior, 15 (11.4 \. 51.

Trend, M.G. (1979). On the reconciliation of qualitative and qUAntitative analyses, In T. Cook & c. Reichart (Eds.). Qu,mtilativeand qualillltivemelhod.. in evaluation rmareh (pp. 68-116).Beverly Hills, CA: Sa~e. Tulle\!, A, (1997). Co~nitive style: Not cuhure's wnsequence. E"TfJpean P"eholagist.1 0). Z58-Z67, Tullert, A, & Kirton, M. (1995). Further evidence for the independence of adaptive.innovative (A.I) cognitive style (rom na'ional culture. PemmalilJ and Individual Differenas, 190),393.396. Van deT Molen, P. (1989). Adaptalion.innovation and d,,"'ges in sodlll ""octlll<: On ,he anAtomy o( ca'Astrophe. In M, Kinon (EJ.). AdaplOTS and inll//\l<JtOTS (PI" chap. 7). LonJon: Ro,;tledge. WatSon.W, Kuman. K, & Michaelsen, L. (1993). Cultural diversi'y's impact on intetaction process anJ performance: Comparin~ homogeneous and diverse 'ask groups. Academ, of Manllgernenr Journal, 36 0). 590.602. Sy.tem:An analy.i.of Weick, K. (1991),The vulnerahle the Tenerife air disaster. In P.Frost, L.Moore, M, Loul., e. Lundherg & I. MArtin. (Ed..). Reframing organi.atumal w/lure (pp. 117.!}0). Newhury Park: Sage, Weick, K, (IIJII7). Organizational culture as a source of high reliahility. Califomw MaMRemen'Review,19 (2),
II

Appendix
ALTKIRT Check the one from each of the following pairs which best describes you. A 0 0 1. Thinking characterized by precision, reliability, efficiency, prudence, discipline 2. Thinking characterized by lack of discipline. linking of unrelated ideas. unusual thought patterns 1. Interested in finding problems to solve 2. Interested in solving problems 1. If rules don't fit. bend them a bit 2. Prefer to work within established rules Solutions sought by tried and true methods 2. Use unproven ideas in seeking solutions 1. Can maintain high accuracy for long periods of work
2. Work best for short bursts of high intensity

B
C

0 0 0 0

DOl. 0

E F G H

Z.I27-

Weick. K. (1979). The ,ocial ""ch%o York:Ranl!om House.

of organizingNew

0
0 0

1. Bending the rules for one person is unfair to the rest

2. Bendingthe rules if necessarymakesbureaucracyhuman 1. Impractical,unpredictable.change-oriented type 2. Practical,predictable, take-care-of-businesstype

of specialized km}wledge 2. Command of general knowledge 1. When involved in a project. I forget that other people are involved and probably should be consulted 2. When involved in a project, I am still considerate of others

0
0
0 0

0 1. Command