Santa Clara UniversityScholar Commons

Management Leavey School of Business

3-1996

Cognitive Elements in the implementation of NewTechnology: Can Less information Provide MoreBenefits?Terri L. GriffithSanta Clara University, tgriffith@scu.edu

Gregory B. Northcraft

Follow this and additional works at: http://scholarcommons.scu.edu/mgmt

Part of the Business Administration, Management, and Operations Commons

Copyright © 1996 by the Management Information Systems Research Center (MISRC) of the University of Minnesota.

This Article is brought to you for free and open access by the Leavey School of Business at Scholar Commons. It has been accepted for inclusion inManagement by an authorized administrator of Scholar Commons. For more information, please contact rscroggin@scu.edu.

Recommended CitationGriffith, T.L., & Northcraft, G.B. (1996). Cognitive elements in the implementation of new technology: Can less information providemore benefits? MIS Quarterly, 20, 99-110.

Note: Implementing New Technology

Cognitive Elements inthe implementation ofNew Technology:Can Less informationProvide MoreBenefits?

By: Terri L. GriffithManagement and PolicyColiege of Business and Pubiic

Administrationl\icCleiiand Hall, Room 405University of ArizonaTucson, AZ 85721U.S.A.griffith@ccit.arizona.edu

Gregory B. NorthcraftDepartment of Business AdministrationCoiiege of Commerce and Business

AdministrationUniversity of iiiinoisChampaign, iL 61820U.S.A.northcra® uxi .cso.uiuc.edu

Keywords: IS implementation, user-analyst dif-ferences, IS impiementation approaches,user training, user-analyst interaction, userexpectations

iSRL Categories: FD, FD01, FD05, FD06,FD08, GB07

IntroductionOrganizations have come to rely on technologi-cal innovation as a central component of their

competitive strategy (Reddy, 1990). While newtechnologies hold tremendous promise forenhancing organizations' efficiency and effec-tiveness, much of this potential is never real-ized (e.g., Kwon and Zmud, 1987). One studyof 2,000 U.S. companies found that 40 percenthad not achieved the intended benefits fromimplementing an office technology (Bikson andGutek, 1984). Significantly, less than 10 per-cent of these implementation failures appearedto stem from technical problems; most occurredfor human and organizational reasons, such aspoor technology management (Bikson andGutek, 1984), including users' misunderstand-ing of the meaning and/or uses of the technolo-gy {e.g.. Griffith, 1993).

Griffith and Northcraft (1993) have proposed amodel of the cognitive determinants of technol-ogy implementation success. Their modelemphasizes that differences in cognitions (e.g.,thoughts, perceptions, and constructed under-standings) among users, designers, and imple-menters (e.g., Lind and Zmud, 1991) are criticaldeterminants of impiementation success. Priorresearchers have provided broader models ofimplementation (e.g.. Cooper and Zmud, 1990;Goodman and Griffith, 1991); the Griffith andNorthcraft (1993) model focuses on the prob-lematic human and organizational componentsof technology implementation success.

This paper explores the major mechanismswithin the Griffith and Northcraft (1993) cogni-tive model. This model offers a fine-grainedview of how user and implementer understand-ings influence implementation success. Whilebroader implementation models suggest struc-tural and process strategies for increasing thelikelihood of implementation success, thismodel describes user and implementer under-standing and can be used to design appropri-ate implementation strategies.

The Cognitive Framework

Past research has underemphasized the role ofcognitions in implementation, even though cog-nitions are known to be cnjcial to the adoptionprocess immediately preceding implementation(Sproull and Hofmeister, 1986). Griffith and

MIS Quarterly/March 1996 99

NotQ: implementing New Technoiogy

Northcraft (1993) have suggested an importantrole for implementer cognitive frames, in thepresentation of information during the imple-mentation of a new technology. Frames are theperceptual sets that direct an Individual's criti-cal cognitive processes (e.g., Pinkley, 1991;Pinkley and Northcraft, 1994), including direct-ing what information to attend to and how tointerpret that information. Frames invoke selec-tive perception (e.g., Dearborn and Simon,1958) and thus, influence how users come tounderstand a new technology in its organiza-tional setting (e.g., Louis, 1980). During imple-mentation, the frames of implementers (thoseresponsible for the introduction of the technolo-gy to prospective users) will limit both whatinformation implementers provide to users, aswell as influence how implementers interpretusers' comments or questions.

There are two types of information that imple-menters might present to introduce a newtechnology — descriptive and operational —and either type of information can be positiveor negative. Thus, descriptive informationrefers to the positive and/or negative uses ofthe technology, i.e.. the benefits and costs thatcan result from the technology's use. Similarly,operational information describes how to use(positive operational infonnation) and/or not touse (negative operational information) thetechnology.

ty with the technology; even complex opera-tional issues will have become second natureto implementers and thus, are not salient whenimplementers present information to prospec-tive users (e.g., Sproull and Hofmeister, 1986).Subconsciously, implementers may emphasizethe benefits (positive descriptions) of a technol-ogy in order to insure users' initial interest or torationalize their imptementer role (e.g.,Festinger, 1958).

Users, in contrast, have a high need to reduceuncertainty (Lester, 1986) and gain control overthe technology (Baronas and Louis, 1988;Falcione and Wilson, 1988) during implementa-tion. Appeasing these needs should lead usersto want to know more than just the benefits ofthe technology; users should want an under-standing of the operational facets of the tech-nology, as well as an understanding (forewarn-ing) of any negative features. This is the para-dox of positive value: By focusing only on thebenefits of the technoiogy, implementers seemdestined to disappoint users — not fuifiil users'informationai needs — and thus, increase thelikelihood of implementation failure (e.g.,Sproull and Hofmeister, 1986). It is not thatimplementers wish to deceive users; imple-menters, because of the cognitive frame theybring to implementation, simply are unable toadequately empathize with users' informationneeds.

Paradox of positive value

A problem arises when there is a discrepancybetween the cognitive frames that imple-menters bring to implementation — and conse-quently the information that impiementers pre-sent to prospective users — and users' infor-mational needs (Griffith and Northcraft, 1993).This problem, known as the paradox of positivevalue (Baier, et al., 1982), occurs when imple-menters present predominantly positivedescriptive information about a technology.Implementers may emphasize positive descrip-tive information because that is the cognitiveframe they bring to implementation — a frameof strong belief in the benefits of the technolo-gy. Implementers may unintentionally disregardoperational concerns because of their famitiari-

Paradox of negative experience

The problem presented by the paradox of posi-tive value is that users who are underpreparedby a positively biased introduction to a technol-ogy will encounter negative surprises (Louis,1980) — operational difficulties and unantici-pated costs — and that these negative surpris-es doom implementation to failure. Griffith andNorthcraft (1993) have suggested, however,that within this problem there is a surprisingopportunity — the paradox of negative experi-ence. Users should be discouraged by a techrnology's negative surprises only when thosesurprises are costly. If the discovery of negativesurprises is not costly to users, negative sur-prises offer opportunities for trial-and-errorlearning that instill in users the prospect that

100 MIS Quarterly/March 1996

Note: Implementirjg New Technology

there is more to learn. The paradox of positivevalue is that an impiementer's positively biasedpresentation of a technology makes negativesurprises inevitable; the paradox of negativeexperience is that these negative surprises, ifmanaged well, become valuable positive learn-ing experiences for users.

The paradox of negative experience is built onthe idea of exploration-based (rather thaninstruction-based) learning. Exploration-basedlearning entails providing novices only enoughunderstanding of something (e.g., a new tech-nology) to begin using it and to begin discover-ing the limitations of that understanding {e.g.,Davis and Bostrom, 1993). Implementers pro-vide users with an initial (positively biased)understanding of the technology; users' initialexperiences with the technology help usersbegin to restructure and adapt their under-standing of the technology (beyond that provid-ed by the implementer). These steps are relat-ed to the concepts of "mapping via training"and "mapping via usage," respectively(Bostrom, et al., 1990, p.1O3).

There are two important qualifications to theparadox ol negative experience. The first is thatnegative surprises will only be more valuablerather than discouraging if they are not person-ally costly to users. In organizations, users'early experiences with a technology can bethought of as either on-the-job or free (Griffithand Norihcraft, 1993). Initial experiences thatare on-the-job require users to complete work(and have that work evaluated!) while alsolearning to use the technology. Under these cir-cumstances, negative surprises will be person-ally costly to users; users will incur the cost o!not finishing the required work or finishing itpoorly (and suffering commensurate perfor-mance evaluations). The alternative — freetraining — refers to time off-line provided forusers to explore the limits of their understand-ing of the technology when organizational workis not required or not evaluated. Under thesecircumstances, the errors of trial-and-errorlearning can be relatively costless.

Free training also has important implicationsfor user satisfaction. Discrepancy theories ofjob satisfaction (e.g., Katzell, 1964; Locke,1976) suggest that negative surprises will cre-

ate user dissatisfaction with the technology.However, this dissatisfaction should be moder-ated by the costliness of the negative surpris-es. If negative surprises are encountered dur-ing evaluated on-the-job performance, thecosts will be greater for users and thereforedissatisfaction with the technology moreextreme, than if those negative surprises areencountered during free training.

The second qualification to the paradox of neg-ative experience is that users are most likely tolearn when their experiences disconfirm theexpectations (schema) provided them by imple-menters (Louis, 1980; Louis and Sutton, 1991).Discovery of discrepancies between expecta-tions and perceived reality pushes users intoactive thinking, and away from habits of mind(e.g., Louis and Sutton, 1991) where learningdoes not take place. If implementers provideenough information for users to use the tech-nology without encountering negative surprises(Louis, 1980) during free training, then usersmay only confirm their expectations (Klaymanand Ha, 1987). The more information usershave during free training, the less likely thatthey will learn to adapt in this period where mis-takes are relatively costless. Thus, a little fail-ure is not only good but necessary for success-ful leaming and adaptation, as long as it can bemade relatively costless (March, 1976). Costlytraining, (e.g., on-the-job) where individuals donot have time to make mistakes, cannot makemistakes without cost to company, customer, orself, or where mistakes result in embarrass-ment dramatically limit individuals' opportunitiesfor learning. Instead, users may learn only toavoid mistakes, and so never adapt or explorethe technology.

Cognition and implementation:hypotheses

To summarize, users provided with positivelybiased introductions to a new technology, suchas implementers tend to provide (Griffith andNorthcraft, 1991), will encounter negative sur-prises during their initial use of the technology.Implementation success depends on whetherthose surprises are costly to the user. Costlysurprises (In which users' work is lost or not

MIS QuartBtiy/March 1996 101

Note: Implementing New Technology

completed due to probiems with the technolo-gy) likely will decrease user satisfaction withand interest in the technology. Relatively cost-less negative surprises, on the other hand, pro-vide users with the knowledge that there ismore to learn about the technology, but withoutdamaging the users' reputation or work. Thus,the following hypotheses are examined:

H1: Users provided with informationbiased toward positive descriptionand no chance for costiess discoverywilt have lower satisfaction with thetechnoiogy than users provided witheither more balanced information(positive and negative operationai anddescriptive information) or thoseaiiowed free training (costless preper-formance opportunities to discoverthe technoiogy).

i-i2: Users provided with informationbiased toward positive descriptionbut aiiowed free training wiii be moresuccessfui in their utilization of thetechnoiogy than users provided withmore balanced information or thosenot aiiowed a chance for costiessdiscovery.

Users who are provided with and confirm a rel-atively balanced understanding of the technolo-gy (e.g., Klayman and Ha, 1987; Louis, 1980)are likely to conclude that there is little more tolearn. If the presentation has truly been realisticand balanced, then this prediction applies toboth users provided with free training and thoseonly given on-the-job experience. Users provid-ed with a fuller, balanced spectrum of informa-tion should be abte to perform the basic taskstaught during implementation, but they willhave a shallow understanding of the technolo-gy and may be less prepared to adapt for long-run implementation success.

H3: Users provided with balanced informa-tion wiil have lower perceived need tolearn than users provided with posi-tively biased information.

Method

Subjects and task

One hundred twenty-nine upper-division univer-sity students enrolled in an organizationalbehavior course volunteered to use a presenta-tion software technology to create presentationmaterials for assigned group projects. Theseprojects entailed grades for professionalism ofpresentation, a large component of which wasthe quality of the presentation materiais creat-ed. Presentation grades were not a componentof the study, and students were not required touse the materials they created with the technol-ogy, although they were required to create pre-sentation overhead materials for their groupprojects by some method. Volunteers receivedextra course credit for agreeing to learn to usethe software.

The context of the study was thus more fieldthan laboratory (Mawinney, 1986). Subjectswere members of the organization (the class)that would utilize the product of their work, andthis work had to be completed whether or notthe study took place. Subjects' use of the tech-nology was directly related to their real worldneed to complete presentation materials (over-head transparencies) for their projects. As sug-gested by Campbell (1986), the constructs test-ed in this research were comparable to thoseextant in a field setting. Although subjects' par-ticipation was relatively brief, parallel tasks in anorganizational setting would be similarly limited(e.g., a project team creating a presentation toreport on its work). Therefore, the constructsexamined here are expected to operate as theywould with similar tasks and technologies; dif-ferences in effects should be in level rather thandirection.

Design

A 2x2 (Balanced/Positive-Only Information byFree-Time Training/On-the-Job Performance)between-subjects design was employed. TheInformation manipulations were provided bothwithin the classroom introduction to the soft-ware and during actual use of the softwafe. The

102 MIS Ouarterly/March 1996

Note: implementing New Technology

Free-Time Training condition was created byproviding subjects with preperfonnance time toexperiment with the software (versus providingonly the three hours of on-the-job performancetime allocated to create the presentations).

Facility and software

The presentation software was Installed on nine386-level PCs, each located in a separateroom. The software provided capabilities fortext, drawing, clip-art, and data charts.Pretesting had revealed that only one student inthe course was familiar with this particular soft-ware package. That student did not participate,so all subjects were new to the software.

Materials and measures

Training materials included a scripted introduc-tion to the software {for use by the two imple-menters), overhead transparencies explainingthe software, and detailed instruction sheets forcreating presentation materials using the tech-nology. The introductions, overheads, andinstruction sheets provided the Informationmanipulations. Balanced Information materialsincluded positive and negative statementsabout the software and operational instructionsabout what to do and not do using the software.Positive Description-Biased materials providedonly the most necessary operational steps tousing the software and neither negative com-ments about the software nor any instructionsabout what not to do. Each computer room wassupplied with the software's summary referencemanual and the full reference guide (over 400pages). Students supplied their own sketchesand ideas for their work time. Subjects wereprovided five sample slides created using thesoftware. These slides were text only and usedthe default background. These sample slidesillustrated the most basic use of the software.

Subjects" utilization of the technology was mea-sured against the basic format provided in thesamples. Text only and the default backgroundwere used as the baseline for measurement ofsubject performance. Elaborations from base-

line (e.g, the use of clip art or a custom back-ground) demonstrated the subject's skill atusing the technology and represented bothknowledge and use. Two outcome variableswere thus created: demonstrated skill (DemSk;mean of coders' skill ratings — ranging from -1to 6) and a more basic measure (coded as 0 or1) of whether the subject was able to create apresentation and print file (File). Two codersassessed subjects' demonstrated skill with 99percent agreement.

A post-experimental questionnaire providedmanipulation checks for the Information pre-sented (positive and negative description andoperational/how-to information). Enough Time(a 1 to 7 rating scale anchored by. "I felt I didnot have enough time to practice with the sys-tem before I had to create my group's presenta-tion," and "t felt I had enough time to...") servedas the manipulation check for Free Time versusOn-the-Job training. Attitudinal outcome mea-sures also were assessed: Satisfaction with thesystem and training (Satisfaction: 14 semanticdifferential items adapted from Baroudi andOrlikowski, 1988); and subject's perception ofthere being more to leam, a two-item measure(More2Leam).

Three control measures were used:Innovativeness, Computer Graphics Experience,and Academic Skill. Innovativeness measuredsubjects' fiexibiiity and willingness to challengeparadigms (Kirton, 1976) using an adaptation(Marcic, 1992) of Kirton's A-l scale (1976). Theoriginal scale has been widely validated and hashigh internal reliability (Cronbach's Alpha andKR-20 in the range of 0.80 to 0.90) over a vari-ety of samples (Taylor, 1989). Innovativenesscontrolled for subjects' propensity to push thesoftware to its limits or to try new approaches.Computer Graphics Experience was measuredusing an open-ended item asking about sub-jects' experience with presentations. Responseswere coded 0/1, subjects coded 1 having somecomputer graphics experience. ComputerGraphics experience controlled for subjects'general skill in creating the presentation materi-als. Both Innovativenss and Computer GraphicsExperience had been assessed earlier in thesemester in preparation tor the group project.The Academic Skill measure was each subject'scurrent score (out of 515) from exams and other

MiS Quarterly/March 1996 103

Note: Implementing New Technology

course work. Academic skill controlled for ability,and/or motivation to perform well on the project.These three measures controlled for individualdifferences, which have been found to influencethe successful implementation of technology(e.g.. Alavi and Joachimsthaler, 1992; Bostrom,etal., 1990).

Procedures

The manipulations were provided over thecourse of two meetings of the course's discus-sion sections (12 sections with 30 students persection, meeting once per week with a teachingassistant to discuss course material) and a lateron-the-job working session. During the firstweek, experimenters presented a brief PositiveInformation handout; if the discussion sectionhad been assigned to a Balanced Informationcondition, subjects also received a BalancedInformation overhead presentation of the soft-ware system. (The additional information neces-sary in the Balanced Information conditions waspresented via overhead transparencies in orderto reduce the possibility for contamination ofPositive Description-Biased conditions.Overhead presentations of the additional infor-mation left no hardcopy that might highlight dif-ferences among experimental conditions.)During the second week, experimentersreturned to the discussion sections to reiterateand reinforce the Information manipulations.The presentations lasted five to 10 minutes dur-ing each discussion section and were fullyscripted to insure consistency.

At the end of the second week's discussion sec-tion presentation, students were asked to volun-teer to use the technology to create their classpresentation. (Students who had not been pre-sent for both information presentations wereallowed to use the technology, but not includedin the study.) From the volunteer list, eight sub-jects from each discussion section were ran-domly assigned to the Free Training condition.These subjects were taken from the discussionsection to the computer center. All volunteerswere candidly and truthfully informed that onlyeight subjects were selected from each discus-sion section because only eight computers wereavailable.

In the Free Training condition, subjects weregiven 40 minutes to work through the scriptedinstructions for creating a sample presentation.Once in the computer center, subjects wereeach handed a step-by-step guide to creating apresentation (commensurate with theirInformation condition). Subjects were told thateach room contained a quick reference guide, afuli reference manual, and the sampie slides.They were told to do their best to recreate twoof the sample slides.

The subjects worked on the sample slides untilthe end of their scheduled class period. At theend of the class session, all volunteers (boththose selected for Free Training and those whoremained in class) were offered the opportunityto schedule a three-hour on-the-job working ses-sion. This was an opportunity for the students tocreate presentation materials for their group pro-ject and was not billed as an experiment.

Thus, the Free-Training condition operationai-ized costless training by providing regular classtime for users to initially experiment and leamabout the technology. This training time wascompletely separate from the time slot providedfor actual production. The On-the-JobPerformance condition operationalized costlytraining by allowing the users to interact withand learn about the technology only during theirlimited time allotted for making presentationslides. Mistakes made during this period woulddetract from subjects' opportunity to create theactual presentation.

Three-hour time slots were available for the on-the-job working sessions during the two weeksprior to the due date of the class presentations.Subjects were greeted at the computer centerby the experimenter and asked if they had withthem sketches for prospective slides. (Subjectshad been told during the earlier presentationsthat they were required to have sketches beforethey came to use the software. This require-ment was made to insure that their use of thecomputer was a serious component of theirclasswork.) Subjects were handed the instruc-tion materials appropriate for their Infomiationcondition, told to follow the instructions verycarefully, and reminded of the manuals avail-able in the rooms for their use. Each was thensent to an assigned room.

104 MIS Quarterly/March 1996

Note: Implementing New Technology

if subjects said they were finished before theend of the three hour on-the-job working ses-sion, the experimenter prompted them to see ifthere was anything else they would like to tryusing the software. If not, their files werechecked by the experimenter, and they werethen given the questionnaire. Subjects were toldthe questionnaire was needed for feedbackabout whether to make this software availablefor future classes. The experimenter printed outeach of the files the students had created andmade these available to the students for theirproject presentations. Debriefing was conduct-ed during a regular session of the course.

Results

Manipulation ohecks

A comparison of the sum of the responses tothe four Information manipulation check itemsrevealed that subjects in the BalancedInformation condition reported that they wereprovided with significantly more informationthan subjects in the Positive Description-Biased

condition (M(Ba1anced)-14.42, W(PosDescB)=''2.52,

Subjects in the Free-Training condition did notprovide significantiy different responses fromsubjects in the On-The-Job Performance condi-tion for the Enough Time measure (M,F,ee,=4.84,M(o.j.p,=4.28, r=-1.42. p< .16), although thetrend was in the expected direction. SinceEnough Time is a state measure of how costlysubjects would perceive setbacks encounteredin their use of the technology (lower scoresmeaning setbacks were more costly), subjects'individual responses to the measure were usedin the analyses rather than condition assign-ment. Responses to the Enough Time measureare the result of condition assignment and indi-vidual differences in perception. The implica-tions are discussed below.

Analysis

Three dimensions of implementation successwere addressed in this analysis: (1) basic uti-lization of the technology (File) and demonstrat-ed skill level (DemSk), (2) Satisfaction, and (3)perceived need to learn more (More2Learn).Table 1 provides means, standard deviations,and correlations for all variables. Table 2 pro-vides the ordinary least squares analysis of thecontinuous dependent variables and a logitanalysis of the dichotomous variable. File.

Hypothesis 1 predicted that users provided withPositive Description-Biased information aboutthe technology and only On-the-job time toadjust to the technology would have lower satis-faction with the technology. This hypothesiswas not supported (f = -0.35, p < .73). Lowpower (.06) is of some concern here. Over3,000 observations would be required for thissize effect to result in significant differences.The data suggest the subjects' perceptions thatthey had Enough Time, regardless of informa-tion provided during implementation, has thestrongest positive relationship with satisfaction(f= 3.51, p<.001).

Hypothesis 2 predicted that users provided withPositive Description-Biased information aboutthe technology, and Free Time to adjust to thetechnology, would be most successful in theiruse of the technology. The predicted interac-tions between the Infomiation variable and theEnough Time measure were significant for boththe File and DemSk measures of implementa-tion success (x̂ Fiie = 4.02, p < .05; t^^^y, = 2.06,p < .05). No main effect was found for EnoughTime on either measure. Perception of enoughtime did not. alone, influence performance.There was also a significant positive main effectfor Balanced information (xSpn̂ = 4.03, p < .05;'DemSk = 2.25, p < ,05). Figure 1 shows the pre-dicted DemSk scores for low and high respon-dents on Enough Time by Balanced versusPosDescB information conditions. High levels ofperformance seem to result from eitherBalanced information, or Positive Description-Biased information combined with EnoughTime. The overall effectiveness of the differentstrategies is discussed beiow.

MIS Ouarteriy/March 1996 105

Note: Implementing New Technology

Table 1. Correlation Coefficients for Independent Variables

PosDescB

Enough Time

Academic Skill

Computer GraphicsExperience

Innovativeness

Satisfaction

File

DemSk

More2Leam

PosDescB

M=.548d=.5O

1.00

-.01(127)

-.15(129)

-.21*(129)

.24*(127)

-.02(121)

.06(129)

-.12(129)

-.01(128)

EnoughTime

M=4.47sd=2.14

1.00

.02(127)

.01(127)

.14(126)

.41*(119)

-.18*(127)

.02(127)

-.30*(126)

AcademicSkill

M»«39.gBsd=2B.86

1.00

-.12(129)

.02(127)

-.09(121)

-.10(129)

.18'(129)

.01(128)

ComputerGraphics

Experience

M=.228d=.42

1.00

.02(127)

.12(121)

-.18'(129)

.19*(129)

-.08(128)

Innov.

M=4.418d=1.79

1.00

.05(119)

.11(127)

.09(127)

-.05(126)

Satlsfac.

M=61.188d=1S.37

1.00

-.04(121)

.04(121)

-.46*(120)

Hie DemSk More2Learn

M=.16 M=1.61 M=6.G6sd=.36 8d=1.43 sd=2.56

1.00

-.54* 1.00(129)

-.06 .17 1.00(128) (128)

•p <.O5, two tailed, number in parenthesis = N. PosDescB was dummy coded 1=Positive Description-Biased, O=FuIi Information.

Table 2. Analysis of Dependent Measures for Study 2

InterceptPosDescBEnough TimePosDescBEnough Time*Academic SkillComputer GraphicsExperienceInnovativeness

Satisfaction/=(6,111) = 4.15

p< .0009R2=:.18

p66.46***

1.783.13***

-0.44-0.04

3.35-0.05

Fiie«x2(3,123)=8.67

p< .03R2 = .08

P2.19*

-2.30*-0.07

0.50*

DemSk

p<.02

P-2.24-1.32'-0.13

0.24*O.or

0.570.07

More2LearnF(6,118) = 2.29

p<.04

P9.09**

-1.25-0.50'**

0.270.0002

-0.44-0.03

•Logit Analysis. Estimates were unstable when control variables were included. Larger x^ (twice the- LogLikehoods) in this analysis indicates better model fit (JMP®User's Guide, 1989, p. 312).*p<.05. " p < . 0 1 . * * ' p < . 0 0 1 .

106 MIS Quarterly/March 1996

Note: implementing New Technology

PredictedDemSk

--Full Info.—PosDesc

LoTime EnoughTime

Controlling for Academic Skill, Computer Graphics Experience, and Innovativeness

PredictedResidualDemSk

EnoughTime

PosDesc

•Note: A median split was used to create the two Enough Time groups.

Figure 1. Predicted Scores for DemSk: Full Versus Positive Description-Biasedinformation for Low Versus High* Respondents on Enough Time

Hypothesis 3 predicted that users provided withBalanced information wouid have the lowestperceived need to learn. This hypothesis wasnot supported (t ~ -1.20, p < .23). The power forthis effect is low (.22), though within the rangegenerally reported for small effects (Upsey,1990). A larger sample (e.g., over 300 observa-tions) might have produced a significant result.Enough Time had the only significant effect onthe perception that there was more to learn{t= -3.22, p< .01) and was negatively related.

Discussion

The concept of the paradox of positive value(Baier, et al., 1982; Sproull and Hofmeister,

1986) was combined with the idea of the para-dox of negative experience (Griffith andNorthcraft, 1993) to understand implementationsuccess. For attitudes, users' perceptions ofhaving enough time to adjust to the new tech-nology produced a main effect; both satisfactionand feelings of expertise were positively relatedto users' perceptions of having enough time.Balanced information significantly affected per-formance; however, limited information (positivedescription bias) — combined with perceptionsof enough time to work with the technology —also yielded high performance. As is discussedbelow, there may be reasons beyond pure per-formance that will lead us to choose particularimplementation strategies in the fieid.

MIS Quarteny/March 1996 107

Note: impiementing New Technology

The relatively small effect sizes preclude furtheranalysis of this data, but may be explained bythe field nature of the study. As in any fieldresearch, the incentives for performing mayhave varied from subject to subject. The incen-tive structure was based on subjects' need tocreate presentations for their group projects.Additionally, the complexity of the software mayhave ieft many users below the level of exper-tise they would need to begin the discoveryprocess in the limited time subjects had to work.(Only 23 percent of subjects had any previouscomputer graphics experience.) Futureresearch could use subjects with broader expe-rience, longer performance periods, and controlfor subjects' incentives to fully utilize the tech-nology. It is encouraging that the hypothesizedperformance effects were visible even in thisrelatively uncontrolled field setting. However,studies of longer-term implementations of com-plex technologies may provide additionalinsights into the effects demonstrated here.Organizational or technological complexitycould increase the cost of negative surprises; ifso, free training may need to address organiza-tional issues as well as technological ones. Thekey is to anticipate problems — whether tech-nological or organizational — and provide newusers with costless ways to team to solve them.

Acknowledgements

We would like to thank Jay Nunamaker andDoug Vogel for their support during the earlystages of this work.

References

Alavi, M. and Joachimsthaler, E.A. "RevisitingDSS Implementation Research: A Meta-Analysis of the Literature and Suggestionsfor Researchers , " MiS Quarterly {'\6:^),March 1992, pp. 95-117.

Baier, V.E., March, J.G., and Saetren, H.implementation as a Doubtfui Metaphor,unpublished manuscript, Stanford UniversityBusiness School, Stanford, CA, 1982.

Baronas, A.K. and Louis, M.R. "Restoring aSense of Control During Implementation:

How User Involvement Leads to SystenAcceptance." MiS Quarteriy {^2^), Marci.1988, pp. 111-124.

Baroudt, J.J. and Orlikowski, W.J. "A ShortForm Measure of User InformationSatisfaction: A Psychometric Evaluation andNotes on Use," Journai of ManagementInformation Systems (4:4), Spring 1988, pp.44-59.

Bikson, T. and Gutek, B. "Implementation ofOffice Automation," Rand Corporation, SantaMonica, CA, 1984.

Bostrom, R.P., Olfman, L, and Sein, M. "TheImportance of Learning Style in End-UserTraining," MiS Quarterly (14:1), March 1990,pp. 101-119.

Campbell, J.P. "Labs, Fields, and StrawIssues," in Generalizing from Laboratory toFieid Settings, E.A. Locke (ed.), LexingtonBooks, Lexington, MA, 1986. pp. 269-279.

Cooper, R.B. and Zmud, R.W. "InformationTechnology Implementation Research: ATechnological Diffusion Approach,"Management Science (36:2), February 1990,pp. 123-139.

Davis, S.A. and Bostrom, R.P. "Training EndUsers: An Experimental Investigation of theRoles of the Computer Interface andTraining Methods," MiS Quarterly {"n-A),March 1993, pp. 61-85.

Dearborn, D. and Simon, H.A. "SelectivePerception: A Note on the DepartmentalIdentification of Executives," Sociometry(21:2), June 1958, pp. 140-144.

Falcione, R.L. and Wilson, C.E. "SocializationProcesses in Organizations," in Handbook ofOrganizational Communication, G. M.Goldhaber and G. A. Barnett (eds.), AblexPublishing Corp., Norwood, NJ, 1988, pp.151-169.

Festinger, L. The Motivating Effect of CognitiveDissonance," in Assessment of HumanMotives, G. Lindzey (ed.). Holt, Rinehart &Winston, New York, NY, 1958, pp. 69-85.

Goodman, P.S. and Griffith, T.L. "A ProcessApproach to the Implementation of NewTechnology," Journai of Engineering andTechnotogy Management (8:3/4), 1991. pp.261-285.

Griffith, T.L. "Monitoring and Performance: AComparison of Computer and Supervisor

108 MIS Quarterly/March 1996

Note: Implementing New Technology

Monitoring," Journal of Applied SocialPsychology (23:7), April 1993, pp. 549-572.

Griffith, T.L. and Northcraft, G.B. "CognitiveElements in the Implementation of NewTechnology: Testing the Paradox of Value,"presented at the Westem Decision SciencesMeeting, Kauai, HI, 1991.

Griffith, T.L. and Northcraft, G.B. "PromisesPitfalls, and Paradox: Cognitive Elements inthe Implementation of New Technology,"Journai of Managerial Issues (5:4), Winter1993, pp. 465-482.

JMP® User's Guide, SAS Institute Inc., Cary,NC, 1989.

Katzell, R.A. "Persona! Values, Job Satisfaction,and Job Behavior," in Man in a World atWork, H. Borow (ed.), Houghton-Mifflin,Boston, 1964, pp. 341-363.

Kirton, M.J. "Adaptors and Innovators: ADescription and Measure," Journal ofAppiied Psychology (61:5), October 1976,pp. 622-629.

Klayman, J. and Ha, Y.W. "Confirmation,Disconfirmation, and Information inHypothesis Testing," Psychology Review(94:2), April 1987, pp. 211-228.

Kwon, T.H. and Zmud, R.W., "Unifying theFragmented Models of Information SystemsImplementation," in Criticai Issues inInformation Systems Research, Boland andHirschheim (eds.), John Wiley, New York,1987.

Lester, R.E. "Organizational Culture,Uncertainty Reduction, and the Socializationof New Organizational Members," inCommunication and information Science:Studies in Communication, T. Sari (ed.),Ablex Publishing Corp., Norwood, NJ, 1986.

Und, M.R. and Zmud, R.W. The Influence of aConvergence in Understanding BetweenTechnology Providers and Users onInformation Technology Innovativeness."Organization Science (2:2), May 1991, pp.195-217.

Lipsey, M.W. Design Sensitivity, Sage,Newbury Park, CA, 1990

Locke, E.A. "The Nature and Causes of JobSatisfaction," in Handbook of industriai andOrganizational Psychology, M.D. Dunnett(ed.). Rand McNally, Chicago, IL, 1976.

Louis, M.R. "Surprises and Sense Making:What Newcomers Experience in Entering

Unfamiliar Organizational Settings,"Administrative Science Quarteriy (25:2).June1980, pp. 226-251.

Louis, M.R. and Sutton, R.I. "SwitchingCognitive Gears: From Habits of Mind toActive Thinking," Human Relations (44:1),January 1991, pp. 55-76.

March. J.G. The Technology of Foolishness,"in Ambiguity and Choice in Organizations,J.G. March and J.P. Olsen (eds.).Universitetsforiaget, Bergen, Nonway, 1976,pp. 69-81.

Marcic, D. Organizationai Behavior:Experiences and Cases (3rd ed.). West, St.Paul, MN, 1992.

Mawhinney, T.C. "Reinforcement ScheduleStretching Effects," in Generalizing fromLaboratory to Field Settings, E.A. Locke(ed.), Lexington Books, Lexington, MA,1986, pp. 181-186.

Pinkley. R.L. "Dimensions of Conflict Frame:Disputant Interpretations of Conflict," Joumalof Applied Psychoiogy (75.2), April 1991, pp.117-126.

Pinkley, R.L. and Northcraft, G.B. "ConflictFrames of Reference: Implications forDispute Processes and Outcomes,"Academy of Management Journal (37:1),1994, pp. 193-205.

Reddy, R. "Technology and Organizations: ATechnological Perspective," in Technologyand Organizations, P.S. Goodman, L.S.Sproull and Associates (eds.), Jossey-Bass,San Francisco, CA, 1990, pp. 232-253.

Sproull, L.S. and Hofmeister, K.R. Thinkingabout Implementation," Journai ofManagement i12-A), Spring 1986, pp. 43-60.

Taylor, W.G.K. "The Kirton Adaption —innovation inventory: A Re-examination ofthe Factor Structure," Journal ofQrganizational Behavior {^0:A), October1989, pp. 297-307.

About the Authors

Terri L. Griffith is assistant professor of man--agement and policy at the University of ArizonaCollege of Business and Public Administration.She received her Ph.D. from Carnegie MellonUniversity's Graduate School of IndustrialAdministration in organizational psychology andtheory. Her research interests include the imple-

MIS Quarterly/March 1996 109

Note: ImplQmenting New Technology

mentation and effective use of new technolo- Administration. He received his Ph.D. fromgies, inciuding group support systems and com- Stanford University in sociai psychoiogy. Hisputer monitoring. research interests include negotiation and con-

flict management, managerial decision making,Gregory B. Northcraft is professor of business and employee motivation and the design ofadministration at the University of illinois feedback delivery systems, particularly in highCollege ot Commerce and Business technology manufacturing settings.

110 MIS Ouarterly/March 1996