a comparison of organizational factors influencing new product success in internal and alliance based products

These and otherrequirements seen as essential to new product development success are difficult to accomplish within thecontext of alliances Bidault and Cummings 1994.Research QuestionsOu

“A Comparison of Organizational FactorsInfluencing New Product Successin Internal and Alliance Based Products”

Eugene SivadasF Robert DwyerEmerson CollegeISBM Report 4-1998

Institute for the Study of Business MarketsThe Pennsylvania State University402 Business Administration Building

University Park, PA 16802-3004(814) 863-2782 or (814) 863-0413 Fax

The Pennsylvania State University is committed to the policy that all persons shallhave equal access to programs, facilities, admission, and employment withoutqualifications as determined by University policy or by state or federal authorities.The Pennsylvania State University does not discriminate against any person becauseof age, ancestry, color, disability or handicap, national origin, race, religious creed,sex, sexual orientation, or veteran status.Direct all inquiries regarding thenondiscrimination policy to the Affirmative Action Director, the Pennsylvania StateUniversity, 201 Willard Building, University Park, PA 16802-2801; Tel (814) 865-4700/ V; (814)

U.Ed BUS 98-058

New Product Success in Internal andAlliance Based Processes

Eugene SivadasF Robert DwyerAddress Correspondence to:

Eugene SivadasAssistant Professor

School of Communication, Management, and Public PolicyEmerson College

100 Beacon StreetBoston, MA 02116

Phone: (617) Fax: (617) 824-8749;

Eugene Sivadas is an Assistant Professor of Integrated Marketing Communications, Emerson College,Boston, MA and F Robert Dwyer is the Joseph S Stem Professor of Marketing, Department of Marketing,University of Cincinnati, Cincinnati, OH The research support of the Institute for the Study of BusinessMarkets and the Direct Marketing Policy Center is gratefully acknowledged The authors thank KarenMachleit, Ann Welsh, and B.J for their comments on earlier drafts of this paper.

New Product Success in Internal andAlliance Based Processes

New products provide increased sales, profits, and competitive strength for most organizations However,nearly 50 percent of the new products that are introduced each year fail Organizations thus find themselvesin a double bind On the one hand they have to consistently innovate to remain competitive, but on the otherhand innovation is risky and expensive Many organizations are forming business alliances to quicken thepace of and reduce risks associated with innovation Yet, by some estimates, 70 percent of these alliancesfail Many of the prescriptions for successful alliance management clash with recommendations foreffective innovation management Using the model, this paper develops testablehypotheses by integrating the new products and alliance literature Internal new product development efforts are compared and contrasted with new product development efforts conducted underalliances The model is tested with data from a sample survey in the semiconductor manufacturing contextand replicated in the healthcare sector Results suggest that cooperative competency, clarity of agreement,fit, and clan-oriented administrative mechanisms contribute to new product success The importance ofvarious factors in the model on internal and external NPD is discussed.

New products provide increased sales, profits, and competitive strength for most organizations.Many successful corporations such as JVC (which pioneered the VHS format for home VCRs) and AppleComputer owe their fortunes to new products they developed (Cooper 1993) A study of more than 700Fortune companies indicated that new products would provide about a third of their profits over thenext five years (Booz, Allen, and Hamilton 1982).

On the other hand, nearly 50 percent of the new products that are introduced in the market placeeach year fail, causing considerable financial loss and embarrassment to their promoters (Business Week1993; and Maidique 1990) At the extreme, Ford Motor Co lost $250 million with the Edsel in 1958and RCA’s failed player launched in 1981 cost them $500 million 1984).

Thus, many organizations are entering business alliances to offset many of the pitfalls associatedwith the innovation process A business alliance is “an ongoing, formal, business relationship between twoor more independent organizations to achieve common goals” (Sheth and 1992, p 72).Organizations enter alliances to quicken the pace of innovation, overcome budgetary constraints, spread outrisks, and gain access to resources (e.g., technological, financial) not otherwise available to them (Bleekeand Ernst 1993, and Cunningham 1995) For example, Intel has entered into an alliance withHewlett-Packard to develop a single

computers (Wall Street Journal 1994).Nevertheless, many alliancesSome estimates put the failure rate of

computer chip capable of running software in both PCs and large

tend to be unstable and a large number of these fail (Gates 1993).alliances as high as 70 percent (Business Week 1986; 1993).The potential for conflict and a clash of interest among alliance partners is inherent Either party canopportunistically use the alliance to learn the other’s business or technological secrets andSengupta 1993) Furthermore, because alliances are largely self-governing; there are no practical higherauthorities who ensure that errant partners will be brought in line 1993).

uncertainties in alliance impair the new product development objectives Indeed, Bidault and Cummings( 1994) suggest that there is a fundamental clash between the “ logic of innovation” and the “ logic ofalliances Alliances succeed when goals and responsibilities of partners are clearly detailed (Hausler,Hohn, and Lutz 1994; and Roos 1992) Innovation, however, requires a measure of flexibility to begranted to those directly involved with the project Alliance partners seek joint control of a project anddepartures from prior agreements may involve renegotiation, and, thus may impede the flexibility requiredfor effective innovation Organizations enter into alliances to take advantage of partners’ knowledge andstrength but most alliances are characterized by lack of trust and Roos 1992) This restricts thefree flow of information, critical for new product success (e.g., Barclay 1992b) These and otherrequirements seen as essential to new product development success are difficult to accomplish within thecontext of alliances (Bidault and Cummings 1994).

Research Questions

Our paper aims to integrate these streams of research on product innovation and alliances in orderto compare internal new product development efforts (efforts conducted within an organization) withexternal new product development (efforts undertaken within alliances) This problem has received onlylimited attention from researchers and empirical work in this area has been exploratory in nature (cf Bidaultand Cummings 1994; Littler et 1995) By comparing and contrasting internal and external new productdevelopment efforts we hope to identify advantages and disadvantages of alliance new productdevelopment, specify conditions under which it might be more appropriate to conduct innovation internallyversus externally, and identify steps which could take to generate positive outcomes for external newproduct development efforts.

Conceptual Model and Hypotheses

The on is vast and varied in its focus (cf Barclay 1992; Calantone, Benedetto,and Divine 1992; Crawford 1991) Our research rests principally on the comprehensive model fromZirger and Maidique (1990) Stemming from a research program they initiated in the early 1980s (see

uncertainties in alliance impair the new product development objectives Indeed, Bidault and Cummings( 1994) suggest that there is a fundamental clash between the “logic of innovation” and the logic of“alliances.” Alliances succeed when goals and responsibilities of partners are clearly detailed (Hausler,Hohn, and Lutz 1994; and Roos 1992) Innovation, however, requires a measure of flexibility to begranted to those directly involved with the project Alliance partners seek joint control of a project anddepartures from prior agreements may involve renegotiation, and, thus may impede the flexibility requiredfor effective innovation Organizations enter into alliances to take advantage of partners’ knowledge andstrength but most alliances are characterized by lack of trust and Roos 1992) This restricts thefree flow of information, critical for new product success (e.g., Barclay 1992b) These and otherrequirements seen as essential to new product development success are difficult to accomplish within thecontext of alliances (Bidault and Cummings 1994).

Research

Our paper aims to integrate these streams of research on product innovation and alliances in orderto compare internal new product development efforts (efforts conducted within an organization) withexternal new product development (efforts undertaken within alliances).This problem has received onlylimited attention from researchers and empirical work in this area has been exploratory in nature (cf Bidaultand Cummings 1994; Littler et al 1995) By comparing and contrasting internal and external new productdevelopment efforts we hope to identify advantages and disadvantages of alliance new productdevelopment, specify conditions under which it might be more appropriate to conduct innovation internallyversus externally, and identify steps which firms could take to generate positive outcomes for external newproduct development efforts.

Conceptual Model and Hypotheses

The literature on NPD is vast and varied in its focus (cf Barclay 1992; Calantone, Benedetto,and Divine 1992; Crawford 1991) Our research rests principally on the comprehensive model from and Maidique (1990) Stemming from a research program they initiated in the early 1980s (see

research, and exploratory interviews involving 330 new products in the electronics industry Their modelidentifies the critical factors in both the internal and external environment of the organization, as well askey activities for successful NPD.

There are two points we want to underscore before we present our adaptation of their model.First, because the model is phenomenon centered, we should not be surprised that it uses perspectivesfrom many promising conceptual vantage points, including strategic management theory, resourcedependence theory, economics, and more Their orientation is a good reference point for the approach inthe subject study aims are not to test a single theoretical prescription or reconcile two theories, butto use theory as a lens to examine the vexing phenomena of internal and alliance-based

Second, as you will see presently, we have consolidated a number of model factors forparsimony This enables us to link conceptually and examine empirically the strategic alliances issues wesee in We only briefly treat our modest integrations because they are task oriented simplificationsrather than critical disputations.

Successful NPD

rely on strategic management theory in their propositions that strong R&D and manufacturing prowess and coordination are critical for NPD success Our adaptation brings together allthree functions Indeed, many companies fail to bridge the barriers between functional areas andinformation critical to a product’s formation and function gets lost Besides communication difficulties,lack of familiarity with another unit’s procedures and personnel can impede efficiency leaving sometasks undone and others redone Lastly, we look for a spirit of candor, teamwork, and reliance betweenmembers of different units (cf., Song and Xie 1995; Souder 1981) Formally, we have:

hl: success derives from competence in and coordination among R&D, marketing, andmanufacturing.

verify research by Cooper and Romanelli (1985) and others that productsare more likely to be successful if they build upon a firm’s existing technologies and market strengths.

New systems, technologies, personnel and customers can disrupt patterns of coordination and impair thespeed of the project At the same time, firms enhance their survivability by doing what they do best inthe ecosystem that favors them We have:

H2: NPD success is positively related to activities that fit and build upon core competencies hypothesize that top management support is critical to the process lean onrudimentary leadership theory in their proposition that general management can provide the overcoming implicit barriers between functions, the requisite organizational resources, and a spirit ofcommitment to NPD (cf., Booz Allen Hamilton 1968) If we consider their logic more closely,however, it seems clear that top management support is not so much expected to affect NPD directly, butindirectly by influencing the level of effort and cooperation between units We hypothesize:h3: Competence and coordination among units in the NPD process derive from top

management support of the effort.

The model also recognizes the criticality of customer knowledge and market standards in Consistent with contemporary market-based management (Best 1997; reason thatsuccessful will derive from products that provide value (Buzzell and Gale 1987).Similarly, successful results from products that offer and technical superiority overcompeting products We appreciate the practical utility of these variables as antecedents to successin the model because they sharpen the customer focus of efforts Nevertheless, in our own pilotresearch and in designing our cross-sectional sample survey, we encountered difficulty separating theseantecedents from the item domain of the criterion variable itself, success Thus, customer value andfunctional/technical superiority are not exogenous variables in our adaptation of the model; they areindicators of success.

Finally, the model regards a focus on large and growing markets as a factor of success.A resource dependence perspective (cf Aldrich 1979; Dwyer and Oh 1987) provides the conceptual

model because our efforts to provide a more “sophisticated” analysis failed Briefly, we thoughtdimensions of predictability and uncertainty in the environment promised more intrigue than

munificence variable, but our efforts to obtain good measures of environmental perceptions frominformants failed.’

Successful Alliances

The study of alliances is relatively recent, yet a body of empirical understanding has begun toaccumulate (cf Varadarajan and Cunningham 1995) Although alliances have been formed fordistribution, product bundling, technology sharing and assorted other purposes, we will presume for thetime being that the factors of alliance success generalize to NPD goals Thus, we will try to link theveritable “checklist” of factors for alliance success to the model of new product success.

Trust and Communication: Organizations in alliances become vulnerable to the actions of partners whosebehavior is not under their control 1993) Participants in an alliance may not share one paramountgoal Because partners enter into an alliance to maximize their own gains, it may be advantageous to seekgains at the expense of the other partner 1993; Williamson 1985) of trust is inherent and thenumber one concern in alliances (Wolff 1994).

Trust exists when “ one party has confidence in an exchange partner’s reliability and integrity”(Morgan and Hunt 1994, p 23) Predictability, dependability, and faith are three key components of trust(Andaleeb 1992) In a study of vertical partnerships between manufacturers and dealers, Mohr andSpekman (1994) found that more successful partnerships were characterized by greater levels of trust.Kanter (1994) studied 37 companies from 11 counties and likewise found trust to be a key element inalliance success Conversely, Sherman (1992) found lack of trust to be a major cause of alliance failure.

Effective communication between partners is essential for alliance success Communication refersto “the formal as-well as informal sharing of meaningful and timely information between (Andersonand Narus 1990, p 44) It enables goal adjustment, task coordination, and interfii learning Mohr and

information sharing.

In the process, trust seems to be a critical ingredient for interfunctional cooperation.Although we sense that the relationship between trust, communication, and coordination is worthy ofconsiderable study in its own right, for our purposes, we need a more global construct We will use amidrange variable comprised of three interacting facets Logically, coordination without communicationis fantasy, communication without trust is babble, and trust without coordination is empty Similarly,trust results from successful coordination and enables further coordination; communication arises fromand enables trust We don’t claim that these three variables completely define a new abstract construct,but will regard them in our study as three fairly compelling facets of an organizational syndrome we willcall cooperative

If we use the cooperative competency syndrome to capture the essence of criticalantecedent to success, the competence and coordination among units, we can make a pivotalconnection between the NPD and alliance literatures This prompts us to reword hl:

success derives from the cooperative competency among the functional and corporateunits involved.

We will now revisit one more hypothesis from the ZM model and develop five new hypothesesinvolving antecedents to cooperative competency.

Resources Alliances can be thwarted by the changing strategic priorities of partners New priorities mayresult in their not wanting to commit adequate resources to the alliance And without adequate human,financial, and other resources alliances will disintegrate or collapse and Roos 1992) Gates (1993)and Wolff (1994) suggest that the amount of resources each partner is to expected to commit should beclearly laid out and understood by the other partner(s).

The commitment of resources to an alliance is akin to general management support of the NPDeffort in the model Resource commitments are a manifestation of top management support of the

operations of top management support, but the logic for now rephrased is the same:H3: Cooperative competency among the functional and corporate units involved derives from

top management support of the effort.

Governance Administrative Strategic alliances are fundamentally aninterorganizational system applied to the process of NPD Thus, as a governance mode, we shouldobserve alliance effects on NPD success primarily through cooperative competency Followinginstitutional economics (cf Williamson 1985) and generalizing from work (cf Boyle etal internal NPD should outperform alliances on such criteria as inter-unit communication,auditability, goal cohesion, and attitudinal solidarity In terms of our model, we have:

H4: Cooperative competency is affected by governance structures in that internally conductedinnovation process provide higher levels of cooperative competency than in the NPD effortsof alliances.

Multifunction teams and alliances need administrative mechanisms that provide the parties witha measure of certainty regarding roles and procedures, for making decisions, and for determining thescope of participants providing input Formalization, the use of explicit rules in the relationship, hasbeen identified as an impediment to the spontaneity and flexibility needed for internal innovation(Bidault and Cummings but between firms tends to enhance effectiveness (cf., Dahlstrom, Dwyerand 1995) Centralization, the concentration of decision-making authority, typicallyimpairs effectiveness as it increases perceptions of bureaucratic structuring which decreases favorabilityof participants’ attitudes towards the project and results in increased opportunism (cf John 1984) A clansystem is governed by shared values and norms This common ground limits the needs for monitoringand other bureaucratic devices and should enhance the abilities of the parties to work cooperatively.

H5: Cooperative competency is positively affected by administrative mechanisms that are (a)formalized, (2) decentralized, and (3) clannish.

Fit The success of strategic alliances depends critically on the strategic fit between the partners’ products,

and lack of trust.

In the alliance literature one concrete question looms large in the discussion of fit: Are thepartners competitors? Competitors are apt to hold many common understandings of the market andproduct development and production making for a good “fit.” But concerns about zero sumopportunities, long-run partner motives, and the vulnerability of proprietary know how may seriouslyimpair “fit” (cf Sheth and Parvatiyar 1992).

H6: Cooperative Competency in a NPD alliance is higher when the partners are noncompetitorsthan when they are competitors.

A second aspect of fit is the very nature of the innovation sought in the alliance Radical new tothe world innovations may tax existing systems of communication and patterns of collaboration morethan incremental innovation Such innovations require greater outlay of resources and are riskier (Kotler1997) Radical innovations are inherently more unpredictable opening up possibilities of conflictingcultures (both interdepartmental and interorganizational) The “gate-keeping” approach in which newproduct development occurs in clearly defined stages is more difficult to accomplish in radicalinnovation projects (cf Song and Xie 1995) Thus:

H7: Cooperative Competency in NPD efforts is higher when the innovation is incremental thanwhen the innovation is radical.

Mutual Dependence Alliances which are dominated by a single partner also typically have a very highrate of failure (Cateora 1993; Gates 1993) The dominated partner typically stands to loose from sucharrangements and hence the high failure rate Partners should safeguard their core competencies and Roos 1992) in order to discourage partners from breaking the rules governing the alliance But as

Bazerrnan and Fader (1986) explain, asymmetrical dependence interferes with joint problemsolving because the weaker is guarding against exploitation while the stronger tends to probe theboundaries of exploitation or guard against the appearance of intentions to exploit Of course, allianceswith low mutual dependence are not likely to show much cooperative competency, because exit orneglect are real, viable options Hence:

mutually dependent than otherwise i.e., when dependence is skewed or minimal.

Figure 1 summarizes our discussion and serves as a reference point for the eight formalhypotheses NPD success is our ultimate criterion variable in the model, and its principal antecedents (inovals) derive from Zirger and Maidique’s model The boxed variables come from the alliance literatureand are hypothesized to impact NPD success through the variable cooperative competency, a midrangeconstruct and node in our model that has essential elements identified in both the alliance and NPDliteratures.

The Contexts for Theory TestingSemiconductor Industry

We chose the semiconductor industry as the primary context for our research The semiconductorindustry (SIC code 3674) includes about 400 including Digital Equipment Corporation, Intel,Motorola, and Texas Instruments This industry tends to be innovative and enters into a variety ofpartnership agreements and Weiss 1994) Technological innovation was identified as a top concernfor the 1990s by of US electronics industry and 50 percent of those surveyed indicated that they werelooking for research partnerships (Rayner 1991).

In one year alone, 130 alliance partnerships were reported in the semiconductor industry (U.S.Industrial Outlook, 1993) For example, Advanced Micro Devices and Fujitsu have entered into an allianceto develop a that can replace the hard disk drive in computers (New York Times 1992).

Demands from telecommunications, computer networking, automobile electronics, high-definitionTV, smart credit cards, and the personal computer industry make this industry competitive, innovative andhigh-growth (Standard and Poor 1994) The Semiconductor Chip Protection Act of 1984 protects asemiconductor’s design for up to 10 years, thus providing an added incentive for innovation.

The Healthcare Sector

was used a second context to replicate the hypotheses relating specifically to the introduction ofnew services via alliances About 30 percent of U.S hospitals are members of alliances (Zuckerman and 1990) Resource crunch, need for cost containment, need to be innovative, increasedinterdependency between research and clinical service and changing disease patterns are promptinghospitals to engage in alliances (Kaluzny and Sheps 1992).

Latour, and Calder (1994, p.53) point out that “the belief that hospitals are fundamentallydifferent in function and structure from for-profit, product producing companies has existed for some time.”

Mintzberg ( 1979) refers to hospitals as professional bureaucracies where exist a core of professionals witha great degree of autonomy and power Control in such organizations is based on bureaucracy and a systemof shared values Thus, a replication in the hospital sector is a reasonable test of the generality of the theory.Measures

Measures were developed and refined based on the guidelines provided by Churchill ( 1979) andGerbing and Anderson ( 1988) The relationship between alliance partners (for external productdevelopment) and departments (for internal new product development) are the units of analysis for thestudy The Appendix presents the measures used in the study.

Product success was measured using an eight-item scale that evaluates the new product on quality, financialsuccess, time-taken, market share, speed to market, and meeting of target costs Cooper (1993) was used as aguide to help specify the domain of this construct.

is defined as the confidence an organization (or department) has in the ability and motivationof the alliance partner (or other departments) to produce positive outcomes for the organization The statusof measure development of this construct remains unsatisfactory (Andaleeb 1992; Sheth and 1992) Our six-item measure of trust builds upon Mohr and Spekman (1994) and Morgan and Hunt but includes an ability item and a reliability item.

critical, and proprietary information among alliance partners We use a set of items modified from Mohrand Spekman’s (1994) instrument The communication measure is a five-item, five point Likert-type scalethat reflects communication quality (timeliness, and adequacy of information) and information sharing(willingness to exchange critical, “ proprietary” information).

“ Coordination is often discussed but seldom measured” (Price and Mueller 186, p 108).Coordination is conceptualized as the extent to which activities, people, routines, and assignments worktogether to accomplish overall objectives We adapted Georgopoulos and Mann’s (1962) definition of coordination, where coordination refers to the extent to which the alliance members (ordepartments) function each according to the needs and requirements of the other parts and of the totalsystem This definition is also consistent with and Spekman’s (1994, p 138) definition thatcoordination “ reflects the set of tasks each party expects the other to perform.” A five-item measureborrowed from Georgopoulos and Mann’s measure was modified to make it suitable for assessingcoordination in both interorganizational and interdepartmental settings.

Top management support was measured on a four-item five point scale with two items each forclarity of agreement (the extent of clear-cut understanding about

side was expected to contribute) and lack of resistance (extent toplayers in both organizations).

financial resources and manpower eachwhich there was no resistance from key

Three types of control mechanisms are examined Centralization (the extent of concentration ofdecision making); formulization (the extent to which explicit rules and procedures govern decision making); (the degree to which governance is conducted by shared values) 1980) were measuredusing five-point Likert type scales (cf Dwyer and Welsh 1985).

A competitor alliance is one where parties to the alliance are in either direct (e.g., an alliancebetween two auto makers) or indirect competition (e.g., an alliance between a steel and plasticmanufacturer) outside the-relationship A non-competitor alliance is one where parties to the alliance do notvie for the same customers outside the relationship (e.g., an alliance between a manufacturer and a supplier

on these dimensions Respondents were asked to indicate whether their partner was a competitor in thesame or different industry, a supplier, or a customer.

All innovations lie on a continuum of newness We adopt the dichotomous classification ofinnovations presented in the literature and classify all new products as incremental or radical Radicalinnovations are new-to-the-world pioneering products representing “technological breakthroughs.”Incremental innovations refer to improvements and revisions to existing products, and additions to existingproduct lines that supplement a company’s existing product lines (cf Booz, Allen, and Hamilton 1982) Afour-item measure building upon Cooper (1993) is employed to classify innovations under one of these twocategories An innovation was classified as radical if respondents indicated that the innovation waspioneering, and did not directly build upon existing technology.

dependence is conceptualized from the Emersonian (1962) perspective i.e., the power of Aover B derives from B’s dependence on A Our focus is on an alliance partner s relative power-dependence’which is the difference between the dependency of the focal and its partner on the alliance We employa modified version (modified to make it suitable for assessing dependence in alliances) of the instrumentsused by Anderson and Narus (1990) and Boyle and Dwyer (1995) Two sets of four-item five point type scales (both completed by the same informant) were used to measure the perceived relative dependenceof the alliance partners Partnerships where both parties depended highly on the other and exhibitedmarginal to no asymmetry in dependence (no difference in dependency of focal firm and its partner on thealliance), were categorized as high mutually dependent relationships Alliances exhibiting asymmetry independence and those balanced at an insignificant level (both partners had low levels of power-dependence)were classified otherwise.

Measure Purification

Depth interviews with key informants in four organizations and a jury of five advanced doctoralstudents indicated that the questions were clear and relevant, except for one original trust item, which was

many measures, us confidence going to the pilot test.The Sampling Frame: The Semiconductor Context

A list of semiconductor firms with more than 20 employees was purchased from a leadingindependent list compiler The list contained names of 718 semiconductor firms, similar to numbersreported by many other list compilers, but far greater than those reported in Manufacturing USA (457semiconductor companies) and Ward’s Business Directory (346 semiconductor companies) Aftereliminating what appeared to be multiple sites of the same company, sales and small sub-contractorsthere were 600 companies on the list These 600 companies (the universe of semiconductor comprised the sampling frame.

The Pilot Test

We selected 150 companies on a 4th name basis for the pretest These companies were contacted bytelephone to a) obtain the name of the key informant (i.e., head of the research and developmentdepartment), so that the survey could be addressed to that individual; b) this key informant about _this survey and solicit participation; and c) verify the mailing address of the firm Many of the smallercompanies did not have a formal R&D department, the function being assumed by the engineeringdepartment In these cases the contact person was either the President, CEO, or head of the engineeringdepartment.

All questionnaires were mailed along with a cover letter and a $2 bill as incentive, and a first-classpostage paid return envelope The cover letter explained the purpose of the survey and contained a non-disclosure agreement indicating that the responses would be treated confidentially Following Yu andCooper (1983) we used appeals that highlighted the importance of each response and the research We alsooffered to share the results in summary form if the informants so desired A reminder/thank you card wasmailed 10 days after the questionnaire was mailed.

After four weeks, 40 completed surveys were received This 28% response rate is comparable tothat reported in other studies on strategic alliances (e.g., Mohr and Spekman 1994; 1993).

Gerbing and Anderson’s (1988) paradigm all measures survived an exploratory factor analysis In theconfirmatory factor analysis, although the chi-squares were significant, based on the maximum likelihoodfactor loadings, and normalized residuals we appear to have unidimensional, internallyconsistent, and reliable measures.

Overall, our measures performed well and we made only minor modifications to some of themeasures for the main study TO make the E-page instrument less burdensome, we trimmed one item eachfrom scales (for trust and communication) with alpha greater than 0.90 (Devellis Wesubstantially revised the measure for radical/incremental innovation to enhance its content validity (themeasures retained for the main study are reported in the Appendix).

The Main Study

After using, 150 of the 600 semiconductor for the pilot study, we drew 350 for the mainstudy Our sampling procedure attempted to include all publicly traded (Compustat PC+ a Standard Poor database that lists all publicly traded was cross-referenced to identify publicly tradedsemiconductor then in the rest on a random basis The protocol for the main study was similarto that of the pilot study beginning with telephone name verification and prenotification of the surveys.

The prenotification phase eliminated 68 for one of the following reasons: a) firms were notsemiconductor manufacturers; b) disconnected telephone number; company was no longer in business; d)company was a sales or purchasing outfit only; e) R&D facilities were located in a foreign country; company had a strict no-survey policy.

We sent 282 questionnaires using a protocol similar to the one used in the pilot study The pilotstudy suggested that many may not have entered into an alliance Hence, the cover letter was changedto encourage firms that had not entered into alliances to also respond to the survey About four weeks afterthe initial mailing, non-respondents were mailed a second copy of the questionnaire, thanking them if theyhad already completed the survey and urging them to do so if they had not.

participate in the survey; b) the company was not in the semiconductor business; or c) was in thesemiconductor business but did not manufacture semiconductors In the end 95 completed responses werereceived Our 37% response rate compares favorably to those reported in other studies (e.g., 1993).Confirmatory Factor Analyses: Confirmatory Factor Analyses provided similar fit values to those reportedin the pilot data Initially, separate single-factor models were evaluated for each of the constructs’ measures.A separate four factor CFA model of trust, communication, coordination, and success was run forinternal innovation measures It was not possible to run CFA on measures in the study simultaneouslydue to sample size constraints, because the ratio of sample size to the number of free parameters did notapproach the minimum 5: 1 requirements (cf and Baumgartner 1994).

Given that measures performed similarly in the pilot and final studies, we pooled the data forhypotheses testing The coefficient alpha reliabilities for the pooled data along with confirmatory factoranalysis are reported in

The factor loadings arethan 12.01

Table These results provide evidence of unidimensionality of the constructs.significant, indicating convergent validity The residuals were less

[Insert Table 1 About Here]Healthcare

As indicated the healthcare context was used as a second context to test the hypotheses tomake possible greater generalizability of the theory Resource constraints allowed us to administer only atruncated version of the semiconductor questionnaire for the healthcare context Questions pertained onlyto experiences with alliance innovation only; no questions were asked about internal innovation processesfor healthcare.

Using the same protocol as used in the semiconductor industry, 250 questionnaires were mailed outto Chief Officers in hospitals We received 117 responses, of which 52 were from those who hadpartaken in alliances The measures performed very well in the healthcare context also with all measuresshowing unidimensionality and having coefficient alpha greater than 0.70.

A multiple regression analysis with new product success as the dependent variable was performedto test hypotheses H 1 and H2 (please refer Table 2) As hypothesized in new product developmentsuccess is positively associated with cooperative competency both in internally andexternally p conducted projects in the semiconductor industry These results also hold truefor the healthcare sector where also external NPD success is positively associated with cooperativecompetency 10).

Hypothesis is supported in the semiconductor context, thus suggesting that NPD success isassociated with projects that build upon partners’ core competencies for the semiconductorsector In the health care industry results were not statistically significant.

[Insert Table 2 About Here]

Hypotheses H4 and H7 examined whether cooperative competency is impacted by governancestructure (innovation is conducted internally versus externally) and by the type of innovation (innovation isradical or incremental) A 2 x 2 (governance structure x innovation type) revealed significantmain effects for governance structure indicating support for H4 The absence of innovationtype main effects 12) and interaction effects implies that H7 was not supported.Thus a higher level of cooperative competency is observed in internal ( as opposed to external( 19) NPD projects No statistically significant differences in cooperative competency are observedin radical ( as opposed to incremental innovation projects ( Internal NPD executionsexhibited higher levels of cooperative competency than external NPD efforts for both radical ( 1.09 vs.10.61) as well as incremental ( vs 10.01) innovation projects.

Table 3 provides the estimates of a multiple regression model of antecedents to competency in alliances, in both research contexts.

supported (please refer Table 3) General management support is reflected in the quality of resources andpersonnel allocated and priority accorded to the project by top management We generalmanagement support by examining clarity of agreement about resources to be provided to the project as wellas examining the (lack of) resistance of senior management to the project Clarity of agreement about inputs 1, significantly enhances cooperative competency in the semiconductor industry However,lack of resistance from key players did not significantly contribute to cooperative competency in thesemiconductor sector In the healthcare sector both independent variables-clarity of agreement

and lack of resistance from key players strongly influence cooperative competencies.Thus overall H3 is supported in both contexts.

[Insert Table 3 About Here]

We attend to the same regression model used to test H3 in order to test our hypothesized effects oncooperative competency from administrative mechanisms partner type and mutual dependence on cooperative competency As can be seen from Table 3, in the semiconductor industry, cooperativecompetency is influenced by clan-oriented and formalized relationships

Though not statistically significant, cooperative competencies are negatively associated with high levels ofcentralization In the sector also cooperative competencies are positively associated with clanoriented relationships but not with formalized or decentralized administrative mechanisms.Thus, overall is partially supported.

Hypotheses H6 was not statistically significant in the semiconductor sector and could not be testedin the healthcare sector since only 2 of the alliances were characterized by respondents as competitoralliances Thus, we cannot conclude that cooperative competency is affected by partner status as competitoror non-competitor.