Measuring multiple buying influences

Measuring multiple buying influences

Measuring Multiple Buying Influences Ajay K. Kohli Gerald Zaltman In this article, we review existing measures ofinfluence in buying centers, anddescr...

877KB Sizes 1 Downloads 164 Views

Measuring Multiple Buying Influences Ajay K. Kohli Gerald Zaltman In this article, we review existing measures ofinfluence in buying centers, anddescribe the development of a new measure of inJuence. Next, we report the psychometric properties of the new measure as assessed in a field study involving joint purchase decisions. The results suggest that the measure has very encouraging reliability and validity, and is likely to be very useful in future research on buying centers.

Purchase decisions in organizations are often made by a group of individuals referred to as a buying center [3 I]. The issue of why some members exert more influence than others in a buying center has proven to be of enduring interest [31, 5, 35, 37, 40, 411. Unfortunately, the reliability and validity of the types of measures of influence employed in studies to date are open to question [34]. As such, the development of an alternative measure of influence appears to be an important step towards developing and testing theories, and generating a better understanding of the influence phenomenon. This article first discusses the alternative definitions of the construct in the literature and indicates reasons for favoring the adoption of a particular definition. A review of previous approaches to measuring the influence construct follows. Next, there is a description of the devel-

Address correspondence to Professor Ajay K. Kohli, Department of Marketing, University of Texas, Austin, TX 78712-l 176.

Industrial Marketing Management 17, 197-204 (1988) 0 Elsevier Science Publishing Co., Inc., 1988 52 Vanderbilt Ave., New York, NY 10017

opment of a new influence measure and a report on its psychometric properties as assessed from a field investigation. In conclusion, suggestions are offered pertaining to the use of the measure in future research.

THE INFLUENCE CONSTRUCT The buying center literature was reviewed with a view to understanding the conceptualization of influence in previous research. Surprisingly, very few scholars offer explicit definitions of the influence construct. The occasional definitions that are offered in the literature tend to be at variance with each other. Influence has been defined as the ability of an individual to affect purchase decision making [3, 29, 361, and alternatively, as the actual effect on others [37, 391. Because an individual may have the potential to influence a decision but choose not to realize it, there need not be a high correlation between the ability to effect changes and actual changes [6, 8, 16, 23, 41. Although influence may be defined in either of the two ways noted above, the definitions do not appear to be equally useful. It is instructive to note here that the ability (or potential) to effect changes is universally termed “power” in a variety of literatures [ 1,4, 13-15, 17, 19, 301. On the other hand, power is not defined in any

197

Influence may mean different things to different individuals. literature we know of as the effects or changes caused by a source. As such, defining influence as actual changes rather than the potential to effect changes appears to be preferable for two reasons. First, it eliminates the obvious redundancy and confusion in having two constructs (power and influence) with identical meanings. Second, it offers a label for a phenomenon that is in need of one. Influence (of an individual in a buying center) is herein defined as the changes in purchase decision-related opinions and behaviors of buying center members as a consequence of the individual’s participation in the decision making. This definition taps the notion of how different the buying center members’ opinions and behaviors would have been if the individual had not been involved in the decision making. Because an individual can exert influence without actually trying [24], the definition provides for intentional as well as unintentional change resulting from the individual’s participation.

PREVIOUS MEASUREMENT APPROACHES An illustrative set of measurement approaches employed in previous research is provided in Table 1. In many instances, the exact wording of items is not reported; in these cases, the text of the published reports of the concerned studies was consulted. In general, the approach has been to ask seemingly straightforward questions about the influence of a person, position, or department. The measures generally appear to have some degree of face validity, although their reliability and va-

lidity is not systematically assessed in any of the studies reviewed. A closer examination, however, reveals that several characteristics of the measurement approaches suggest the need for an alternative operationalization. This is discussed in greater detail below.

Single-Item Measures Scales used in previous research uniformly rely on a single item or question to measure influence (see Table 1). Several scholars have argued that single-item measures are likely to have low reliability and validity [8, 21, 271. Jacoby [21] makes the point particularly well: “How comfortable would we feel having our intelligence assessed on the basis of our response to a single question?” Single-item influence measures are especially undesirable in the present context because, as Jackson, Keith, and Burdick [20] point out, “influence” (and other similar terms) may mean different things to different individuals. Some may consider giving the final approval to be a very influential act; others may interpret influence primarily as the effect on the determination of acceptable product specifications. Further, as Silk and Kalwani [34] note, single-item measures place enormous demands on informants by asking them to provide summary judgments about complex decision processes, thus leading to measurement error. They argue for the development of multiple-item measures that tap the processes through which influence is manifested. In line with their recommendations, we focus both on the influence and the processes that lead to influence, as will become clear later.

Confounding Power With Influence AJAY K. KOHLI is Assistant Professor of Marketing versity of Texas, Austin.

at the Uni-

GERALD ZALTMAN is the Albert Wesley Frey Professor of Marketing at the Joseph M. Katz School of Business, University of Pittsburgh.

198

A somewhat different concern with measures of influence used in previous research is that they have the potential to confound the ability to effect changes with actual changes. Consider the following measures: “ . . . how much say . . . the purchasing agent has” in the different stages of decision making, or “How much influ-

TABLE 1 Previous Measures (In Chronological Order)

Weigand

I. Informants were ” asked to rate (their) concern with six elements of the product.”

[40]

2. Informants were asked L‘ how they perceived the responsibilities of the purchasing agent.”

2. “Regardless of who had the final authority, who, would you say, had the most influence on the choice of which particular (product) to get?” (Note: influence was not measured; rather the most influential individual was identified.) Grashof and Thomas

Cooley,

Salancik,

how much say do you feel the purchasing agent has with respect to the following phases of the decision making process?”

Stlk and Kalwani [34]

I. I‘.. how likely would each of the following individuals be to suggest the need for a change?” (Change initiation phase.)

Jackson,

[ 181

and Ostrom [JO]

Pfeffer, and Kelly [33]

Bellizzi [3]

2. “How much influence would you expect each of the following individuals to have in evaluating alternative brands?” (Alternative evaluation phase.)

I. “Regardless of who had the final authority, who, would you say, had the most influence on the decision to get a (product).”

Patchen [28]

3. “How much influence would you expect each of the following individuals to have in making the final decision as to what brand to change to?” (Final decision phase) Thomas [37]

The change in the opinions of subjects as to the relative importance of the features of a service (experimental design).

Jackson,

Informants were asked “ to indicate the relative influence of the buying center members for the purchasing situation. (by allocating). a fixed total of 100 points among members that participate in the decision-making process. ”

Keith, and Burdick [20]

Informants “ were asked to rank order the persons holding (certain) positions in their firms with respect to their importance of influence over each of the eleven different aspects of the buying process.” Each informant was asked to “ allocate a fixed sum, in this case 100% of the total power in a decision, between the decision makers (purchasing, production, engineering, or other functions) based upon how power was distributed.” (Note: influence and power are explicitly treated as being synonymous.) Each informant was “ asked to rank a11 participants, including himself, in order of their importance in making the final decision.” Informants reported “ their perception of the buying influence of the six participant categories on

‘L

Spekman and Stem [35]

Each informant “ was asked to list the members of the decision team from the most important or influential to least important or influential in the selection of the brand of equipment purchased.”

Kelly [22]

. ”

each of the nine buying stages.

Naumann, Lincoln, McWilliams [25]

and

Each informant was “ asked to allocate 100 points among the eight identified potential buying center members . according to the informant’s perception of each member’s relative influence in the purchasing decision.”

Robles [32]

Each informant was asked to “ allocate 100 points among the participants according to their influence in each of five stages of the purchase process.”

Thomas 1381

Each subject rank ordered other buying center members depending upon “. .how ‘influential’ . (they). would be in getting the informant to change his or her position. . to be in agreement with the other person.”

199

ence would you expect (an individual) to have in . . . ?” Notice that the questions can be interpreted as asking for ratings of power (ability to effect changes) or influence (changes). The confounding primarily results from the use of the term influence as a noun which can be interpreted both as capacity and actual effect. This potential for confounding highlights the need for developing a scale that unambiguously measures the changes induced rather than the capacity to do so.

Global Measures Several studies employ global measures of influence which ask informants for overall assessments of an individual’s influence across several purchase stages [20, 22, 281. Because purchase decisions often evolve over long periods of time [ 121, global measures may place enormous demands on informants that they may be unable or unwilling to meet, thereby leading to measurement error [34]. Further, different purchase stages have qualitatively different characteristics, which suggest that different types of measures are needed for different stages (this point is discussed in detail later). There appears to be a need for developing stage-specific measures that would be more relevant to the phenomenon under investigation, as well as less demanding of informants.

Influence of a Person, Position, Department Several scales attempt to measure the influ-,nce of a position (e.g., purchasing manager) or a department (e.g., finance). Whereas the measurement of a person’s influence is straightforward (at least conceptually), questions pertaining to the influence of a position (or department) may be more difficult to interpret unambiguously because of the interrelationship between the power ascribed to a position and the personal power of an individual occupying the position. The influence of a position may be interpreted to mean the influence a person occupying the position exerts because of the power vested in him by virtue of the position, or because of his personal power (e.g., expertise, charisma), or both. Similarly, the influence of a department may refer to the influence exerted by members of a department because of their personal qualities, or because of formalized rules and procedures, or both. It therefore appears preferable to measure influence of specific persons rather than positions or departments. As Nicosia and Wind observe, “Just as the job title of an individual may not be informative as to the role played in the buying process, so may the label of a department be uninformative and possibly misleading in our attempts to understand the buying process [26].”

Scoring Formats Multiple Decisions Some measures expect informants to assess a buying center member’s influence across multiple purchase decisions. This requires informants to consider the possibly varying levels of the member’s influence across individual purchase decisions, which places even greater demands on informants than the simple global measures discussed above. Although the content of specific questions should naturally be determined by the research questions of interest, we make this observation here primarily to highlight the benefits to be gained from asking informants about a single rather than multiple decisions.

A final point relates to the scoring scheme adopted in certain approaches. Specifically, many scales ask informants to rank buying center participants based on their influence. Rank ordering individuals forces discrimination between two or more members who may have had equal influence on a decision, thus resulting in inevitable errors. Additionally, rank ordering does not permit an assessment of the mugnitude of influence. In this regard, the use of interval level response formats would be more informative. The limitations of the influence measures employed in previous research discussed above suggest that there exists a need for a reliable and valid measure of influence.

Influence processes are different in each procurement stage. 200

We turn now to a discussion of the effort undertaken to develop a measure that addresses the above noted concerns. The procedure adopted relies heavily on the approach recommended by Churchill @I.

SCALE DEVELOPMENT The first decision confronting us was the selection of the specific stage of the decision making for which the measure would be designed. The nature of decision making differs across stages of the decision making process, which suggests that influence processes are different in each procurement stage, and require correspondingly different measurement approaches. For example, the information search phase is qualitatively different from the final evaluation and selection phase. Information search involves obtaining information from a variety of sources, such as suppliers, salespeople, trade shows, and promotional literature. Influence in this stage refers to the quantity and quality of information obtained by an individual for consideration by buying center members. In contrast, influence in the final evaluation and selection phase concerns the compliance obtained by an individual from other buying center members with regard to the evaluation of alternatives. As such, it is important to design measures of influence to correspond to the specific stages of interest. For the present research, the final evaluation and selection phase was selected, in part because the social dynamics in this phase are probably the richest. The manner in which influence is manifested in the final evaluation and selection phase was examined, and items designed to capture the essence of influence in this phase. The measure was designed to assess the influence exerted by an individual in a specific purchase decision, rather than in purchase decisions in general, so as to reduce the burden on informants and obtain more accurate measurement, as discussed earlier. One indicator of an individual’s influence is the consistency between the views of the individual and the final purchase decision. Accordingly, an item was designed to assess the extent to which the final decision reflected the individual’s views (Item 9). This indicator, however, is not an adequate measure because the consistency may simply be a coincidence. That is, the buying center members may have reached the final decision quite independently of the views expressed by the individual concerned. Accordingly, additional items were devel-

oped to assess the extent to which the individual was responsible for the consistency between his views and the decision, In line with the recommendations of Silk and Kalwani [34], these additional items relate to “ . . . accommodation and other processes through which influence is manifested. ” An individual can influence the final decision by influencing the criteria used for selection, or the evaluation of the different alternatives on relevant criteria, or by obtaining compliance with his preferences regardless of the content of the evaluation process, or some combination thereof. The items constructed tap the effects of these varied processes that represent antecedents of influence in a buying center. To ensure the face validity of the measure, the term influence, as well as its synonyms-impact, change, effect-were included in several items. Unlike previous measurement approaches, however, the word “influence” was used as a verb rather than as a noun. Influence as a noun is open to alternative interpretations as the ability to effect change or actual change (e.g., “How much influence did Steve have on John?“). Influence as a verb, however, clarifies that the question pertains to actual change (e.g., “To what extent did Steve influence John?“).

ITEM PURIFICATION

AND DATA COLLECTION

A panel of academic judges was asked to critique the structure and content of the initial set of items designed to measure influence. The items were revised based on the comments received from the judges. Next, the revised items were pretested in personal interviews with managers involved in joint purchase decisions. The managers were asked to think of a purchase decision with which they were very familiar and select a person involved in the final evaluation and selection phase of the decision making process. Next, they were asked to respond to certain questions (measure items) about this person. They were encouraged to point out any ambiguity or difficulty concerning to the questions asked, and share with the interviewer any observations that would help improve the instructions or the items. Based on the comments received, the items were revised to improve their clarity and applicability across a wide range of purchase situations. The revised items were pretested again in personal interviews, revised one more time, and pretested again to ensure that their content, structure, and comprehensibility were of very high quality. A total of 14 personal

201

interviews were conducted in course of the pretesting. The final set of items is reported in the Appendix. Next, a survey questionnaire (which incorporated the influence scale as well as other scales designed to assess its construct validity) was mailed to 500 randomly selected members of the National Association of Purchasing Management. Two reminders were sent pursuant to the initial mailing. A total of 25 1 usable responses were obtained for an effective response rate of approximately 55% (42 informants either could not be reached or indicated they had little experience with joint purchases). An analysis of early versus late responses revealed no significant differences in the types of buying centers or situational characteristics, suggesting the absence of nonresponse bias [2]. The following section is an assessment of the reliability and validity of the influence measure using data from the national mail survey.

SCALE PROPERTIES Reliability The measure’s reliability or internal consistency was assessed by calculating its Cronbach’s alpha. A coefficient alpha of 0.93 was obtained. The high reliability estimate is very encouraging, especially in light of the fact that alphas of 0.5 to 0.6 are considered acceptable in the initial phases of measure development [27]. We now turn to the assessment of the measure’s dimensionality.

Dimensionality To assess the dimensionality of the measure items, as suggested by Churchill [8], a factor analysis was performed. Only one factor with an Eigen-value greater than 1 was extracted. This factor had an Eigen-value of 5.48, and accounted for over 60% of the variance in the data. These findings are very encouraging, because they suggest that the measure is unidimensional and that this dimension captures a large portion of the variance in the measure items. Assured of the measure’s reliability and unidimensionality, we proceeded to investigate the measure’s face validity, content validity, and aspects of con-

The measures 202

struct validity (namely convergent validity, discriminant validity, and nomological validity).

Face and Content Validity Face validity refers to the extent to which, on the face of it, a measure’s items appear to tap the construct of interest [8]. As noted earlier, several items of the measure explicitly pertain to the influence of a buying center member in the decision making process, as well as on the final decision. As such, the measure can be said to have face validity. Other items incorporate synonyms of influence (e.g., impact, change, effect) and get at similar notions, thereby providing support for the face validity of the measure. Content validity refers to the extent to which the domain of a construct is represented by a measure’s items [8]. The measure items tap the extent to which a person influences a decision, as well as the process by which influence is exerted (e.g., changing the thought processes of buying center members, altering the selection criteria, rating of alternatives), thus capturing the domain of the influence construct as explicated earlier in this paper. As such, the measure appears to have content validity.

Convergent and Discriminant Validity Convergent validity requires that the measure of a construct correlate strongly with other maximally different measures of the construct [6]. The assessment of convergent validity posed a rather unique problem, in that it was difficult to identify an acceptable measure of influence (if one were available, the alternative offered in this paper would not be required). Accordingly, a second measure of influence was devised as follows. With a view to using dissimilar formats for the two measures, a forced dichotomy response format as opposed to interval scaling was used for a second measure. Specifically, one half of the mail survey informants were asked to select a member (excluding themselves) “ . . . who had less impact than others on the decision reached eventually (that is, the decision finally reached was more a reflection of other members’ views). ” The other half were asked to select who had more impact than others on the amember“...

must be reliable

and valid.

decision eventually reached (that is, the decision finally reached was more in line with his views). ” The two sets of members were assigned scores of 1 and 2 respectively. The Pearson correlation between this measure and the new 9-item measure was 0.66 (p < O.OOO), which suggests a high degree of convergent validity. Discriminant validity requires that the measure of a construct not correlate strongly with measures of other constructs [6]. Though it was possible to employ a variety of constructs to assess the measure’s discriminant validity, we selected constructs that were theoretically not related to influence because these provide more accurate tests of discriminant validity. Measures of buying center size, familiarity viscidity, risk, and time pressure were employed in this analysis because these constructs do not have theoretical linkages with influence (measures of these constructs are available from the authors). The size of the correlations between measures of influence and these five constructs ranged from 0.01 to 0.15, and four of these were not statistically significant at the 0.05 level. These findings strongly suggest that inferences of discriminant validity for the new 9-item measure of influence are appropriate. As an additional check on the measure’s convergent and discriminant validity, the correlations among each of the measure’s items and the items designed to measure other constructs (noted above) were examined. In every instance, intercorrelations among the influence measure items were higher than their correlations with items of other constructs, the latter typically being not significant at the 0.05 level. These results lend further support to the convergent and discriminant validity of the measure. Nomological

Validity

A favorable assessment of nomological validity requires that the measure of a construct correlate with the measure of another construct that is theoretically related to the first construct [ 11,421. Findings from several studies suggest that the expertise, reward power, and coercive power of an individual are predominant determinants of his or her influence [28, 36, 381. Accordingly, the correlations between the influence measure and measures of the expertise, reward power, and coercive power of the individual concerned were computed (measures of these constructs are available from the authors). The correlations were 0.45,0.34, and 0.25 respectively, and all were statistically significant beyond the 0.000 level. These findings are consistent with prior expectations on theo-

retical grounds and provide strong support for the nomological validity of the influence measure.

CONCLUSION Results from the field investigation reported above are very encouraging, and suggest that the measure meets the important criteria of reliability and validity. The multiple-item measure taps influence and the processes through which it is manifested in buying centers. It is important, however, to recognize that the measure is primarily designed to assess influence in the final evaluation and selection phase of decision making. For studying influence in other phases, we recommend that either the measure be appropriately adapted, or a new measure be designed to suit the nature of the phase of interest. Further, while the new measure appears to be very valid and useful for future theoretical investigations, theory development and measure validation go hand in hand. Just as the measure is likely to facilitate further theory development, so too must the measure’s properties be assessed in future theoretical investigations.

ACKNOWLEDGMENT The authors thank members of the National Association of Purchasing Management who made this research possible.

APPENDIX Influence Measure Items 1. How much weight did the committee members give to his opinions? 2. How much impact did he have on the thinking of the other members? 3. To what extent did he influence the criteria used for making the final decision? 4. How much effect did his involvement in the purchase committee have on how the various options were rated? 5. To what extent did he influence others into adopting certain positions about the various options? 6. How much change did he induce in the preferences of other members? 7. To what extent did others go along with his suggestions? 8. To what extent did his participation influence the decision eventually reached? 9. To what extent did the final decision reflect his views? Items were scored on a Spoint Large.”

scale ranging from “Very Small” to “Very

203

with Implications for Marketing 2, 420-434 (October 1974).

REFERENCES 1. El-Ansary, Adel and Stem, Louis W., Power Measurement in Distribution Channels, Journal of Marketing Research 9, 47-52 (February 1972). 2. Armstrong, J. Scott, and Overton, Terry, Estimating Nonresponse Bias in Mail Surveys, Journal of Marketing Research 14, 396-402 (August 1977). 3. Bellizzi, Joseph A., Product Type and the Relative Influence of Buyers in Commerical Construction, Industrial Marketing Management 8, 2 I3220 (June 1979). 4. Bierstedt, Robert, An Analysis of Social Power, American Sociological Review 15, 730-738, (December 1950). 5. Bonoma, Thomas V., Major Sales: Who Really Does the Buying? Harvard Business Review, pp. 11 l-i 19 (May-June 1982). 6. Campbell, Donald R., and Fiske, Donald W., Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix, Psychological Bulletin 56, 81-105 (1959). 7. Cartwright, Dorwin, Influence, Leadership, Control, in Handbook of Urganizations, March, James Cl., ed., Rand McNally, Chicago, l-47, 1965. 8. Churchill, Jr., Gilbert A., A Paradigm for Developing Better Measures of Marketing Constructs, Journal of Marketing Research 16,6&73 (February 1979). 9. Coleman, James S., Foundations for a Theory of Collective Decisions, The American Journal of Sociology 71, 615-627 (May 1966). 10. Cooley, James R., Jackson, Donald W., Jr., and Ostrom, Lonnie L., Analyzing the Relative Power of Participants in Industrial Buying Decisions, in Educators’ Conference Proceedings, Series No. 41.. American Marketing Association, Chicago, 243-246, 1977. Il.

Cronbach, Lee J., and Meehl, Paul E., Construct Validity in Psychological Tests, Psychological Bulletin 52(4), 281-302 (1955).

12. Cyert, Richard M., Simon, Herbert A., and Trow, Donald B., Observation of a Business Decision, Journal of Business 29, 237-248 (October 1956). 13. Dahl, Robert A., The Concept of Power, Behavioral Science 2, 203-215 (July 1957). 14. Emerson, Richard, Power-Dependence Review 27, 31-41 (February 1962).

Relations, American Sociological

Strategy,

Journal of Business Research

23. Martin, Roderick, The Concept of Power: A Critical Journal of Sociology 22, 240-257 (September 1971).

Defense,

British

24. Nagel, Jack H., The Descriptive Analysis qfPower. Yale University Press, New Haven, 1975. 25. Naumann, Earl, Lincoln, Douglas J., and McWilliams, Robert D., The Purchase of Components: Functional Areas of Influence, Industrial Marketing Management 13, 113-122 (1984). 26. Nicosia, F., and Wind, Yoram, Behavioral Models of Organizational Buying Behavior, in Behavioral Models of Market Analysis: Foundations for Marketing Action, Nicosia, F. M., and Wind, Y., eds., Dryden Press, Hinsdale, 1977. 27. Nunnally, Jum C., Psychometric Theory. McGraw-Hill,

New York, 1978.

28. Patchen, Martin, The Locus and Basis of Influence in Organizational Decisions, Organizational Behavior and Human Performance 11, 195221 (April 1974). 29. Pettigrew, Andrew M., Information ciology 6, 187-204 (1972).

Control as a Power Resource,

So-

30. Raven, Bertram H., Social Influence and Power, in Current Studies in Social Psychology, Steiner, Ivan D., and Fishbein, Martin, eds., Holt, New York, 371-382, 1965. 31. Robinson, Patrick J., Faris, Charles W., and Wind, Yoram, Industrial Buying and Creative Marketing. Allyn and Bacon, Boston, 1967. 32. Robles, Fernando, Buying in a Matrix Organization, Management, 13, 201-208 (1984).

Industrial Marketing

33. Salancik, Gerald R., Pfeffer, Jeffrey, and Kelly, J. Patrick, AContingency Model of Influence in Organizational Decision-Making, Pacific Sociological Review 21(2), 239-255 (1978). 34. Silk, Alvin J., and Kalwani, Manohar U., Measuring Influence in Organizational Purchase Decisions, Journal ofMarketing Research 19, 165181 (May 1982). 35. Spekman, Robert E. and Louis W. Stem, Environmental Uncertainty and Buying Group Structure: An Empirical Investigation, Journal of Marketing 43, 54-64 (Spring 1979).

TheActive Society. The Free Press, New York, 1968.

36. Spekman, Robert E., Influence and Information: An Exploratory Investigation of the Boundary Role Person’s Basis of Power, Academy of Management Journal 22(l), 104-l 17 (1979).

16. French, ‘John R. P., Jr. and Raven, Bertram H., The Bases of Social Power, in Studies in Social Power, Dorwin Cartwright, ed., University of Michigan Press, Ann Arbor, 1959.

37. Thomas, Robert J., Correlates of Interpersonal Purchase Influence in Organizations, Journal of Consumer Research 9, 171-182 (September 1982).

17. Gaski, John F., The Theory of Power and Conflict in Channels of Distribution, Journal of Marketing 48, 9-29 (Summer 1984).

38. Thomas, Robert J., Bases of Power in Organizational Buying Decisions, Industrial Marketing Management 13, 209-217 (1984).

18. Grashof, John F., and Thomas, Gloria P., Industrial Buying Center Responsibilities: Self Versus Other Member Evaluations of Importance, in Educators’ Conference Proceedings, Series No. 39.) American Marketing Association, Chicago, 344-349, 1976.

39. Thomas, Robert J , An Approach for Conceptually Defining Interpersonal Influence Process in Organizational Buying Behavior, in 1984AMA Winter Educators’ Conference: Scientific Method in Marketing, Anderson, Paul F., and Ryan, Michael J., eds., American Marketing Association, Chicago, 150-154 1984.

15. Etzioni, Amitai W.,

19. Hunt, Shelby D., and Nevin, John R., Power in Channels of Distribution: Sources and Consequences, Journal of Marketing Research 11, 186- I93 (May 1974). 20

40. Weigand, Robert E., Identifying Industrial Buying Responsibility, nal of Marketing Research 3, 81-84 (February 1966).

Jour-

Jackson, Donald W., Jr., Keith, Janet E., and Burdick, Richard K., Purchasing Agents’ Perceptions of Industrial Buying Center Influence: A Situational Approach, Journal of Marketing 48, 75-83 (Fall 1984).

41. Zaltman, Gerald, and Bonoma, Thomas V., Organizational Buying Behavior: Hypotheses and Directions, Industrial Marketing Management 6, 53-60 (1977).

21. Jacoby, Jacob, Consumer Research: A State of the Art Review, Journal ofMarketing 42, 87-96 (April 1978).

42. Zaltman, Gerald, Pinson, Christian R. A., and Angelmar, Reinhard, Metatheory in Consumer Research. Holt, Reinhart and Winston, New York, 1973.

22. Kelly, J. Patrick, Functions Performed in Industrial Puchasing

204

Decisions