Adoption And Implementation Of Technological Innovations Within

Journal of Business Research 56 (2003) 681 – 686

Adoption and implementation of technological innovations within long-term relationships Angela Hausmana,*, James R. Stockb a

Management, Marketing, and International Business, University of Texas-Pan American, 1201 West University Drive, Edinburg, TX 78539, USA b University of South Florida, 4202 East Fowler Avenue, Tampa, FL 33620, USA Received 7 December 2000; accepted 6 September 2001

Abstract As businesses move toward long-term cooperative relationships, they face increasing needs to coordinate, especially with respect to the adoption of innovative technologies. Since effective adoption involves both adoption and implementation, both stages are critical. This study builds and tests models of adoption and implementation as a function of influence, dependence, and relational variables. Results of this study on electronic data interchange (EDI) adoption in hospital supply chains indicate social influence achieves higher adoption rates than either coercive or noncoercive influence efforts. In addition, communication and participative decision-making are critical implementation variables. D 2003 Elsevier Science Inc. All rights reserved. Keywords: Innovation; Electronic data interchange; Influence; Implementation; Statistical analysis

1. Introduction Although often considered a natural extension of adoption, implementation does not follow automatically and additional research into successful implementation is necessary (Rogers, 1995). In addition, an understanding of potential adopters as active decision-makers, rather than as passive units, is required (Windsor, 1995). In organizations, adoption and implementation processes might be more complex due to the web of relationships surrounding the adopter, such that each independent stakeholder is potentially affected by technological changes (Hausman, 1996). Thus, promoting cooperative adoption among relational partners may be critical to successful adoption (Hakansson and Johanson, 1988). Due to the complementary nature of cooperative adoption, the firm who desires to implement a particular innovation (the focal firm) may need to convince relational partners (recipient firms) to implement it as well. Extant

* Corresponding author. Tel.: +1-956-381-2826; fax: +1-956-3845065. E-mail address: [email protected] (A. Hausman).

literature is mute on the process most useful in encouraging this cooperation without damaging the partnership. Insights from relationship marketing do suggest that the type of influence exerted, as well as interorganizational variables, are drivers of other types of cooperation (Brown and Pattinson, 1995; Dwyer and Gassenheimer, 1992). Therefore, this study is directed towards answering the following questions: (1) what effects do influence efforts exerted by focal firms have on the technology adoption decisions of recipient firms?; (2) do these influence efforts affect the implementation of technological change?; and (3) to what extent do interorganizational variables, such as trust, communication, dependence, and participative decisionmaking between focal and recipient firms affect (a) adoption and (b) implementation of innovations?

2. Conceptual development The context of this study is adoption of electronic data interchange (EDI) by hospitals. EDI is actually five related software programs designed to facilitate the ordering, tracking, and payment of goods across a channel. By electronically processing orders, EDI eliminates mistakes, shortens

0148-2963/03/$ – see front matter D 2003 Elsevier Science Inc. All rights reserved. doi:10.1016/S0148-2963(01)00313-7

682

A. Hausman, J.R. Stock / Journal of Business Research 56 (2003) 681–686

lead times, and speeds payment. Allegiance Healthcare, an early and strong proponent of EDI in this channel, has encountered difficulties in adoption of EDI by downstream channel partners over the past 20 years (Emmelhainz, 1990). Allegiance has become increasingly aggressive in attempts to obtain cooperation from Hospital Material Managers— mainly through letter writing campaigns using coercive influence. Attempts to explicate the adoption process in this type of organizational setting are limited and influence efforts by relational partners are summarily overlooked (Hausman, 1996). Since the adoption process involves influence directed at a specific other, rather than a population in general, an appropriate model must incorporate micro-level variables acting in the firm interface (Gatignon and Robertson, 1989). Once adopted, the innovation must be implemented — it must be installed, personnel trained on its use, and it must be incorporated into the daily routines within the firm — or desired efficiencies will not be achieved (Rogers, 1995). Despite its importance, little evidence exists to facilitate our understanding of the factors affecting extent or speed of implementation (Rogers, 1995; Zaltman et al., 1973). 2.1. The role of influence in cooperative adoption Research supports a relationship between the use of interfirm power (influence) and technological change (Frost and Egri, 1990; Maute and Locander, 1994). In its least coercive forms, influence might promote adoption by providing information about the innovation. In its more coercive forms, influence may be detrimental to the longterm survival of the relationship. Evidence suggests coercive influence strategies are counterproductive in producing favorable adoption decisions, while noncoercive influence strategies are more effective (Brown and Pattinson, 1995; Rogers, 1995). Contradictory results suggest that a powerful member is necessary to drive the adoption process (Maute and Locander, 1994). However, this type of influence resembles championing more than coercive influence. Thus, the following hypotheses were incorporated into the model. Hypothesis 1. Coercive influence strategies are negatively correlated with adoption of technological innovations by recipient firms engaged in long-term relationships. Hypothesis 2. Noncoercive influence strategies are positively correlated with adoption of technological innovations by recipient firms engaged in long-term relationships. In addition to coercive and noncoercive influence efforts, firms might use social influence to encourage adoption (Cialdini, 1993; Frost and Egri, 1990). Social influence, operationalized as referent power, involves associative desires including respect for and friendship with counterparts. Several studies cite such social influences as

factors behind certain consumer behaviors (Hallen et al., 1991) and business relationship success (Frazier et al., 1988), but have never looked at social influence in the context of adoption or implementation of innovations. Thus, we propose the following: Hypothesis 3. Social influence strategy is positively correlated with cooperative adoption of technological innovations by recipient firms engaged in long-term relationships. The dependence of one firm on the other may exert a certain amount of implicit influence. As such, perceptions of relative dependence are linked with other attitudes about the focal firm that may encourage cooperation, including trust, commitment, relational satisfaction, and relational behavior thereby increasing the likelihood of adoption (Kumar et al., 1995; Lewis and Lambert, 1991; Lusch and Brown, 1996). However, other studies find a negative relationship between dependence and joint action in relationships (Dwyer and Gassenheimer, 1992; Hart and Saunders, 1997). A negative relationship between dependence and adoption more closely fits our model, which emphasizes limited use of power and increased use of more cooperative means to achieve adoption. An additional consideration involves the relative independence of hospitals and suppliers in an industry composed of multiple suppliers and low switching costs. Thus, the following hypothesis is proposed: Hypothesis 4. Perceptions of dependence on the focal firm held by recipient firms will be negatively correlated with adoption of technological innovations by recipient firms in the context of long-term relationships. 2.2. The role of behavioral variables in cooperative adoption Critical behavioral variables linked with effective channel relationships are trust and commitment (Morgan and Hunt, 1994). The linkage between high levels of trust and both cooperation and willingness to allocate resources to joint action has been well established and provides a conceptual basis for proposing a similar effect on cooperative adoption (Morgan and Hunt, 1994). Hypothesis 5. High levels of trust between focal and recipient firms are positively correlated with cooperative adoption of technological innovations by recipient firms. The role of commitment to proper channel functionality is also well documented. Empirical evidence supports the role of commitment in acquiescence to joint actions (cf. Kumar et al., 1995; Morgan and Hunt, 1994). Therefore, the following hypothesis is presented: Hypothesis 6. High levels of commitment between focal and recipient firms are positively correlated with cooperative adoption of technological innovations by recipient firms.

A. Hausman, J.R. Stock / Journal of Business Research 56 (2003) 681–686

2.3. Factors affecting implementation

3. Methodology

A significant consideration at the implementation stage of the adoption process is the likelihood of rejection (Rogers, 1995). In EDI implementation, it is understandable that implementation might pose a significant problem, since fears over loss of purchasing records are recognized as a major impediment to EDI utilization (Emmelhainz, 1990). Further, the relative costs of implementation, such as retraining employees and building a necessary interface between the software and the current materials management system, dwarf the cost of adoption. Finally, the psychological start-up costs of EDI are high, including confusion, frustration, and potential conflict with employees (Emmelhainz, 1990). Studies of strategy implementation find a negative relationship exists between the use of power and implementation (Conners and Romberg, 1991). Since little guidance exists relative to implementation of technological innovations, this study will utilize the influence variables proposed earlier to explore their effect on the implementation stage of technological innovation. This leads to the following hypotheses:

3.1. Measures

Hypothesis 7. Coercive influence strategies are negatively correlated with implementation of technological innovations by recipient firms engaged in long-term relationships. Hypothesis 8. Noncoercive influence strategies are positively correlated with implementation of technological innovations by recipient firms engaged in long-term relationships. Hypothesis 9. Social influence strategy is positively correlated with implementation of technological innovations by recipient firms engaged in long-term relationships. Other factors uniquely contributing to implementation have not been identified, since systematic studies focusing on overcoming this resistance have not been conducted. In studies of corporate strategy, input from individuals at the operational level was identified as a precursor to successful implementation (Strahl et al., 1996). Depth interviews with practitioners who have recently been involved in the cooperative adoption process report reliance on communication and participative decision-making by members of implementation teams as a primary means for overcoming internal resistance to innovation (Stockley et al., 1996). This suggests the following hypotheses: Hypothesis 10. Participative decision-making by employees of recipient firms is positively related to implementation of technological innovations by recipient firms engaged in long-term channel relationships. Hypothesis 11. Open communication between employees of focal and recipient firms is positively related to implementation of technological innovations by recipient firms engaged in long-term relationships.

683

Influence strategies were measured using modifications of the scales developed by Frazier and Rudy (1991). Hypotheses addressed only coercive and noncoercive influence, hence, a subset of the items comprising these constructs was used (see Boyle et al., 1992; Simpson and Mayo, 1997; Venkatesh et al., 1995 for support for this technique). Social influence was measured by modifying a scale for referent power originally developed by Brown et al. (1995). Participative decision-making was measured utilizing a fiveitem scale developed by Mohr and Spekman (1994). Open communication was assessed using a five-item scale that was modified to reflect interorganizational, rather than intraorganizational openness (Kitchell, 1995). Commitment and trust were measured using scales adapted from Morgan and Hunt (1994). All of the above scales were measured using seven-point Likert-type items, both positively and negatively worded. Adoption was measured using a dichotomous item assessing adoption/nonadoption of EDI. Communication frequency was measured using a formative scale modified from Mohr and Sohi (1995). The implementation scale was similarly a formative scale that assessed the percentage of transactions completed electronically across the various options commonly available on EDI software packages and time necessary for this implementation level to be reached. 3.2. Survey procedures The preliminary survey instrument was distributed to a group of practitioners, researchers, and EDI experts for feedback regarding clarity and readability of the instrument. After multiple iterations and subsequent modifications, the questionnaire was mailed to 300 randomly selected hospital material managers from a membership list. Based on pretest results, slight changes were made to the instrument before mailing it to the available population of hospital material managers (4700) obtained from Allegiance Healthcare, after elimination of managers who participated in the pretest. A smaller second mailing (1000) was sent to nonrespondents 3 weeks after the initial mailing. Each mailing contained a letter from the researchers, the survey instrument, an incentive form for a cash lottery, and a return envelope.

4. Results 4.1. Sample characteristics MANOVA analysis supported the decision to combine data received from the two mailings, but exclusion of the

684

A. Hausman, J.R. Stock / Journal of Business Research 56 (2003) 681–686

Table 1 Demographics of hospital respondents Characteristic

Frequency

Adopted EDI Yes No

220 61

73.3 21.7

38 4 18

13.5 1.4 6.4

191 69

68.0 24.6

Planning to adopt Yes No Uncertain Purchasing type Centralized Decentralized

Percentage

Decision-making involvement (more than one answer possible) Made the decision 102 36.3 Influenced the decision 87 31.0 Part of a committee made decision 70 24.9 Not involved in the decision 20 7.1 Only involved in the implementation 20 7.1 of the decision Type of software utilized for EDI Off the shelf Vendor customized Internally developed

64 154 30

22.8 54.8 10.7

Supplier who influenced adoption Allegiance Healthcare Abbott Labs Owens and Minor General Medical Bergen Brunswig Other

78 12 65 24 13 22

27.8 4.3 23.1 8.5 4.6 7.8

Percentage purchases from this supplier Mean Median

57.68 60.00

Number of beds Mean Median Maximuma Minimum a

pretest data. A total usable sample of 281 responses was received from the two mailings. Survey data show 70% of respondents were either facing adoption of EDI or had adopted EDI within the previous 2 years; suggesting recall adequacy. As shown in Table 1, almost 86% of respondents were actively involved in the adoption decision and all were involved in the implementation process. Thus, respondents satisfied the criterion for knowledgeable key informants, as established by Heide and John (1990). The number of nondeliverables obtained suggested a response rate of 12%, which, while low, compares favorably with the 11.1% response rate obtained in a study of hospital administrators (Naidu et al., 1999). Nonresponse bias was tested using two different methods. First, respondents to the first mailing were compared with respondents from the second mailing (Armstrong and Overton, 1977), then respondents from a shortened version of the mailing were compared with respondents who answered the complete survey. No statistically significant differences were noted in either test of nonresponse bias. As shown in Table 1, data represented a reasonable crosssection of existing hospitals. Perhaps surprisingly, especially with respect to size, adoption and implementation did not correlate with any of the demographic variables. 4.2. Psychometric analysis of existing and new scales All scales utilized in hypothesis testing performed as expected in terms of both dimensionality and item loadings. These scales also demonstrated adequate reliability based on Cronbach’s a (Table 2). Confirmatory Factor Analyses results (see Table 2) supported the validity of individual scales used in this study and demonstrated acceptable fit for the overall confirmatory model. 4.3. Hypothesis testing

259.49 150 10,000 5

Implementation still lagged far behind the adoption decision. After selecting only those hospitals that had adopted EDI, the average percentage of transactions completed electronically was only 66% for the most frequently

This involved a multiunit hospital chain.

Table 2 Fit statistics — Confirmatory Factor Analysis for constructs Statistic a c2 P value GFI AGFI RMSEA NFI CFI Means S.D. Mode

Model 975.16 .000 .79 .73 .079 .89 .89

Coercive

Noncoercive

Social

Dependence

Trust

Commitment

Communication

Decision-making

.8800 133.32 .000 .92 .87 .09 .97 .97 2.59 1.07 1.63

.8310 28.84 .000 .98 .92 .08 .99 .99 4.17 1.12 4

.8805 25.08 .000 .94 .89 .08 .99 .99 4.40 0.98 4

.7346 5.29 .071 .96 .89 .07 .98 .99 3.89 1.20 4

.9422 36.04 .000 .97 .92 .07 .99 .99 5.21 1.32 6.33

.7534 .620 .733 .98 .93 .08 .99 .99 4.38 1.09 4

.8025 32.59 .000 .97 .93 .09 .99 .99 4.92 1.37 5.60

.8563 31.07 .000 .95 .90 .09 .99 .99 3.69 1.37 4.20

A. Hausman, J.R. Stock / Journal of Business Research 56 (2003) 681–686

implemented component of EDI. Implementation levels for two other components of EDI were less than 20% and implementation for the other components was negligible. The amount of time necessary to achieve this level of implementation was, on average, over 4 months (4.28 months). The high standard deviation (4.09) showed a wide variation in the ability of firms to implement EDI. PRELIS was used to calculate polychoric correlations as input for structural equation modeling, since the adoption variable was dichotomous (which also required that adoption and implementation processes be analyzed separately) (Jo¨reskog and So¨rbom, 1993). The resulting transformed data were analyzed using sequential fit processing via LISREL to achieve both statistical fit and substantive meaning (Jo¨reskog and So¨rbom, 1993). Tests conducted on the implementation half of the model contained only those hospitals which had already made affirmative adoption decisions (n = 158). Table 3 contains the fit statistics associated with both models. Table 3 also summarizes supported hypotheses and associated fit statistics. Hypotheses 1 – 4 postulated the effects of both explicit and implicit influence on the recipient’s EDI adoption decision. Hypotheses 1 and 2 were not supported, while Hypothesis 3 was supported. Thus, social influence affected adoption, but coercive and noncoercive influence did not. Hypothesis 4 was supported; hence, dependence was negatively related to adoption. Hypothesis 5, was supported, however, Hypothesis 6 was not, demonstrating a positive relationship between trust and adoption, but not commitment. No support was obtained for the influence variables tested in Hypotheses 7– 9 on implementation. The model does support Hypotheses 10 and 11, since open communication and participative decision-making were positively related to timely implementation of EDI. Table 3 Structural equation results associated with supported hypotheses Hypothesis number

Relationship tested

Hypothesis Hypothesis Hypothesis Hypothesis

Social influence – adoption Dependence – adoption Trust – adoption Participative decisionmaking – implementation Communication – implementation

3 4 5 10

Hypothesis 11

Path estimate

T value

0.39 0.20 0.39 0.30

3.62 2.40 3.62 2.04

0.20

0.88

Fit statistic

Adoption model

Implementation model

c2 Degrees of freedom P value GFI AGFI RMSEA NFI CFI R2

524.43 143 .000 .82 .76 .098 .86 .89 .12

150.67 62 .000 .96 .90 .071 .98 .99 .27

685

5. Discussion 5.1. Implications The factors affecting adoption of technological innovations and those affecting implementation appear to be entirely different. Specifically, a number of interfirm relational variables affect the adoption stage, while implementation appears to require more coordination and input from various individuals. As expected, the correlation between adoption and implementation is not perfect (.5169). Findings offer empirical support for open interorganizational communication and participative decisionmaking in overcoming implementation resistance. By the implementation stage, efforts to influence the speed or degree of implementation through interorganizational influence are ineffective. Surprisingly, neither coercive nor noncoercive influence attempts are effective in influencing adoption decisions made by hospitals. While this result might seem strange, studies argue the relative impotence of overt influence strategies in affecting behavioral change in interorganizational relationships (Simpson and Mayo, 1997; Venkatesh et al., 1995). Another possible explanation may lie in the relative power symmetry between members of the channel. As proposed, social influence is related positively to cooperative adoption. Dependence is negatively related to adoption; possibly because recipient firms are reluctant to make idiosyncratic investments that increase risk should the relationship dissolve. This study provides some guidance for supply firms who need to gain cooperation from their customers for the adoption of technological innovations and, potentially, other types of joint action. The study suggests that when both parties are allowed input into the implementation process the outcome is likely to be more favorable in terms of both the extent of implementation and its speed. The study also suggests the positive impact of strong, personal relationships between boundary spanners. These boundary spanners generate social capital that can be expended to speed the adoption process. Additionally, interpersonal communication appears more memorable, and potentially more influential than nonpersonal communication, as managers recall receiving few letters despite Allegiance’s claim to use them frequently. Not only is interpersonal communication important in achieving support for the adoption decision among material managers, it also helps speed the implementation process. 5.2. Limitations and future research The most severe limitation of this study, however, is that it tested the model in a single context —hospital supply chains —and extending results to similar contexts appears appropriate. Low response rate is a concern in this study, although extensive tests detecting negligible nonresponse

686

A. Hausman, J.R. Stock / Journal of Business Research 56 (2003) 681–686

bias suggest the generalizability of the data to the population. Further, despite the low response rate, a sufficient sample size was obtained to allow statistical significance in hypothesis testing. However, results only reflect the perceptions of hospital material managers and similar perceptions of the suppliers are limited. Several additional studies are suggested above, namely, validation of the model and extension of the results to other industries. The effects of interfirm influence strategy on other types of interfirm cooperation should also be studied to determine if the same positive relationship between joint action and social influence occurs in these situations as well. It would also be interesting to investigate whether prior experience with technological innovations increases the likelihood of future adoptions of these innovations. In addition, several other studies are suggested by these results. First, the potential to test the model in other cultures appears appealing since several of the constructs identified in this study may be culturally bound. Specifically, variables such as cooperativeness, decision-making, and relationship to authority, which differ by culture, might induce the models to perform dramatically differently in other cultures.

References Armstrong JS, Overton TS. Estimating nonresponse bias in mail surveys. J Mark Res 1977;14:396 – 402 (August). Boyle B, Dwyer FR, Robicheaux RA, Simpson JT. Influence strategies in marketing channels: measures and use in different relationship structures. J Mark Res 1992;29(3):462 – 73. Brown L, Pattinson H. Information technology and telecommunications: impact on strategic alliance formation and management. Decision 1995;33(4):41 – 51. Brown JR, Lusch RF, Nicholson CY. Power and relationship commitment: their impact on marketing channel member performance. J Retailing 1995;71(4):363 – 92. Cialdini RB. Influence: the psychology of persuasion. 2nd ed. New York: Quill, 1993. Conners JL, Romberg TA. Middle management and quality control: strategies for obstructionism. Human Organization 1991;50:61 – 5. Dwyer R, Gassenheimer J. Relational roles and triangle dramas: effects on power play and sentiments in industrial channels. Mark Lett 1992; 3(2):187 – 200. Emmelhainz MA. Electronic data interchange: a total management perspective New York: Van Nostrand-Reinhold, 1990. Frazier GL, Rudy RC. The use of influence strategies in interfirm relationships in industrial product channels. J Mark 1991;55(1):52 – 69. Frazier GL, Spekman RE, O’Neal CR. Just-in-time exchange relationships in industrial markets. J Mark 1988;52:52 – 67 (October). Frost PJ, Egri CP. Influence of political action on innovation. Leadership and Organizational Development Journal 1990;11(1):17 – 25.

Gatignon H, Robertson RS. Technology diffusion: an empirical test of competitive effects. J Mark 1989;53:35 – 49 (January). Hakansson H, Johanson J. Formal and informal cooperation strategies in international industrial networks. In: Contractor FJ, Lorange P, editors. Cooperative strategies in international business. Lexington (MA): Lexington, 1988. p. 369 – 79. Hallen L, Johanson J, Seyed-Mohamed N. Interfirm adaptation in business relationships. J Mark 1991;55:29 – 37 (April). Hart P, Saunders C. Power and trust: critical factors in the adoption and use of electronic data interchange. Organ Sci 1997;8(1):23 – 42. Hausman A. Diffusion of innovations in relational networks: a theoretical perspective. In: Calatone R, Dro¨ge C, editors. AMA Summer Educators Conference Proceedings, vol. 7. Chicago: AMA, 1996. p. 223. Heide JB, John G. Alliances in industrial purchasing: the determinants of joint action in buyer – seller relationships. J Mark Res 1990;27(1): 24 – 36. Jo¨reskog K, So¨rbom D. Lisrel 8: structural equation modeling with SIMPLIS command language Hillsdale (NJ): Lawrence Erlbaum Associates, 1993. Kitchell S. Corporate culture, environmental adaptation, and innovation adoption: a qualitative/quantitative analysis. J Acad Mark Sci 1995; 23(3):195 – 205. Kumar N, Scheer LK, Steenkamp JBEM. The effects of perceived interdependence on dealer attitudes. J Mark Res 1995;32:348 – 56 (August). Lewis CM, Lambert DM. A model of channel member performance dependence, and satisfaction. J Retailing 1991;67(2):205 – 25. Lusch RF, Brown JR. Interdependency, contracting, and relational behavior in marketing channels. J Mark 1996;60(4):19 – 38. Maute MF, Locander WB. Innovation as a socio-political process: an empirical analysis of influence behavior among new product managers. J Bus Res 1994;30(3):161 – 74. Mohr J, Sohi R. Communication flows in distribution channels: impact on assessments of communication quality and satisfaction. J Retailing 1995;71(4):393 – 412. Mohr J, Spekman R. Characteristics of partnerships success: partnership attributes, communication behavior, and conflict resolution techniques. Strategic Mark J 1994;15(2):135 – 52. Morgan RM, Hunt SD. The commitment – trust theory of relationship marketing. J Mark 1994;58:20 – 38 (July). Naidu GM, Parvatiyar A, Sheth JN, Westgate L. Does relationship marketing pay? An empirical investigation of relationship marketing practices in hospitals. J Bus Res 1999;46(3):207 – 18. Rogers EM. Diffusion of innovation. 4th ed. New York: Free Press, 1995. Simpson JT, Mayo DT. Relationship management: a call for fewer influence attempts. J Bus Res 1997;39(3):209 – 18. Stockley T, Maude R, Parish M. Bass Brewers international logistics joint venture to create the new world class distribution system. Presented at the 1996 Council of Logistics Management Conference. Orlando, FL, 1996. Strahl WM, Spiro RL, Acito F. Marketing and sales: strategic alignment and functional implementation. J Pers Sell Sales Manage 1996;16:1 – 18. Venkatesh R, Kohli AK, Zaltman G. Influence strategies in buying centers. J Mark 1995;59:71 – 82 (October). Windsor RD. Marketing under conditions of chaos: percolation metaphors and models. J Bus Res 1995;34(2):181 – 9. Zaltman G, Duncan R, Holbek J. Innovations and organizations. New York: Wiley, 1973.