14
Marketing Controls and Employee Responses: The Moderating Role of Task Characteristics Sanjeev Agarwal Iowa State University Sridhar N. Ramaswami Iowa State University A key concern in implementing organizational controls is that little is known about when controls lead to "negative" employee responses. Previous research has suggested that lower levels of negative responses will be observed only if the controls that are being employed "fit" the characteris- tics of the tasks being controlled. The two task character- istics usually referred to include performance (outcome) documentation and procedural (cause-effect) knowledge. Unlike previous studies, however, this study assumes that the two task characteristics should have a joint, rather than independent, influence on employee responses to controls. The reason is that knowledge of how well one is doing is not sufficient for expecting lower levels of negative responses if information on procedures that may lead to better perfor- mance is not available. Knowledge of procedures, likewise, is not sufficient for lower levels of negative responses if performance documentation is not available. Results of an empirical study provide general support for the above line of reasoning. Most utilitarian organizations use a variety of organiza- tional control forms (e.g., output, process, and informal) to ensure that employees' efforts can be channeled toward ac- tivities that facilitate achievement of organizational objec- Journal of the Academy of Marketing Science Volume 21, Number 4, pages 293-306. Copyright 1993 by Academy of Marketing Science. All rights of reproduction in any form reserved. ISSN 0092-0703. tives (Flamholtz, Das, and Tsui 1985; Jaworski 1988; Ouchi 1979). While the intention of most conscientious organiza- tions in instituting controls is to constructively monitor, direct, evaluate, and reward employees, use of controls, nevertheless, has been found to lead to numerous "nega- tive" employee responses (Ouchi 1979; Merchant 1982; Jaworski and Maclnnis 1989). For example, if organizations set sales targets and measure performance of their salesper- sons against the targets, salespersons tend to underreport market potential to their supervisors in order to obtain lower sales targets. Salespersons may respond this way despite having precise information about the market potential in their respective sales areas. Also, they may concentrate their efforts on existing accounts at the expense of identify- ing and developing new prospects. This may lead to achievement of stated sales targets but may hamper the longer-term performance of organizations. Because such "negative" responses ultimately impact organizational per- formance, a critical role of managers should be to design control systems that help reduce the intensity of such re- sponses, if not completely eliminate them. Although substantial knowledge exists about how organi- zations may choose a particular control form, very little is currently known about when controls produce negative em- ployee responses. The choice of a particular control form, for example, is believed to be a function of two task charac- teristics, namely, the extent to which forms of documenta- tion are available to measure an employee's performance (labeled "performance documentation"), and the extent to which activities an employee should perform to achieve desired outcomes can be specified (labeled "procedural knowledge"). In a marketing setting, performance docu- mentation would imply that organizations develop measure- ment scales such as sales volume, productivity, profit mar- JAMS 293 FALL. 1993

Marketing controls and employee responses: The moderating role of task characteristics

Embed Size (px)

Citation preview

Marketing Controls and Employee Responses: The Moderating Role of Task Characteristics

Sanjeev Agarwal Iowa State University

Sridhar N. Ramaswami Iowa State University

A key concern in implementing organizational controls is that little is known about when controls lead to "negative" employee responses. Previous research has suggested that lower levels of negative responses will be observed only if the controls that are being employed "fit" the characteris- tics of the tasks being controlled. The two task character- istics usually referred to include performance (outcome) documentation and procedural (cause-effect) knowledge. Unlike previous studies, however, this study assumes that the two task characteristics should have a joint, rather than independent, influence on employee responses to controls. The reason is that knowledge of how well one is doing is not sufficient for expecting lower levels of negative responses if information on procedures that may lead to better perfor- mance is not available. Knowledge of procedures, likewise, is not sufficient for lower levels of negative responses if performance documentation is not available. Results of an empirical study provide general support for the above line of reasoning.

Most utilitarian organizations use a variety of organiza- tional control forms (e.g., output, process, and informal) to ensure that employees' efforts can be channeled toward ac- tivities that facilitate achievement of organizational objec-

Journal of the Academy of Marketing Science Volume 21, Number 4, pages 293-306. Copyright �9 1993 by Academy of Marketing Science. All rights of reproduction in any form reserved. ISSN 0092-0703.

tives (Flamholtz, Das, and Tsui 1985; Jaworski 1988; Ouchi 1979). While the intention of most conscientious organiza- tions in instituting controls is to constructively monitor, direct, evaluate, and reward employees, use of controls, nevertheless, has been found to lead to numerous "nega- tive" employee responses (Ouchi 1979; Merchant 1982; Jaworski and Maclnnis 1989). For example, if organizations set sales targets and measure performance of their salesper- sons against the targets, salespersons tend to underreport market potential to their supervisors in order to obtain lower sales targets. Salespersons may respond this way despite having precise information about the market potential in their respective sales areas. Also, they may concentrate their efforts on existing accounts at the expense of identify- ing and developing new prospects. This may lead to achievement of stated sales targets but may hamper the longer-term performance of organizations. Because such "negative" responses ultimately impact organizational per- formance, a critical role of managers should be to design control systems that help reduce the intensity of such re- sponses, if not completely eliminate them.

Although substantial knowledge exists about how organi- zations may choose a particular control form, very little is currently known about when controls produce negative em- ployee responses. The choice of a particular control form, for example, is believed to be a function of two task charac- teristics, namely, the extent to which forms of documenta- tion are available to measure an employee's performance (labeled "performance documentation"), and the extent to which activities an employee should perform to achieve desired outcomes can be specified (labeled "procedural knowledge"). In a marketing setting, performance docu- mentation would imply that organizations develop measure- ment scales such as sales volume, productivity, profit mar-

JAMS 293 FALL. 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

gins, and so forth, whereas procedural knowledge would imply that managers (and their subordinates), for example, have knowledge regarding how sales calls should be distrib- uted between existing and new customers for balancing short-term and long-term performance; or, what factors such as routing, ability to close a sale, and so forth, contrib- ute to maximizing call efficiency. Using these two task char- acteristics, Ouchi (1979) proposed that organizations should use output control when performance documentation is high, process control when procedural knowledge is high, and informal controls when neither performance documen- tation nor procedural knowledge is h i g h . .

In order to assess when controls produce negative em- ployee responses, Jaworski and Maclnnis (1989) proposed a contingency model in which they extended Ouchi's logic and suggested that organizations should be able to minimize the negative effects on employees by using controls that "fit" (as implied in Ouchi's model) the nature of the tasks that are being controlled. In other words, they hypothesized that the use of output control for tasks characterized by high performance documentation, and the use of process control for tasks characterized by high procedural knowledge, should result in lower levels of employee job tension, dys- functional behavior, and information asymmetry. However, despite the appeal of these hypotheses, Jaworski and Macln- nis did not find empirical support. Close scrutiny of how Jaworski and Maclnnis derived the contingency hypotheses from Ouchi's framework reveals two critical assumptions that may have contributed to their nonfindings. In this arti- cle, we discuss these assumptions and present alternative hypotheses.

The first assumption made by Jaworski and Maclnnis (1989) is that "fit" between a particular task characteristic, performance documentation or procedural knowledge, and a given form of control can be achieved independently of the other task characteristic. In other words, fit is consid- ered high when process control is used for tasks charac- terized by high procedural knowledge, regardless of the level of performance documentation. Similarly, fit is con- sidered high when output control is used for tasks charac- terized by high performance documentation, regardless of the level of procedural knowledge. This notion of fit is, however, difficult to defend, especially for tasks charac- terized by both high performance documentation and high procedural knowledge. In this task condition, managers can use either output or process control. Researchers have sug- gested that when both process and output controls are feas- ible options, managers may have a preference for output control because 1) it is generally more convenient to imple- ment (Rockness and Shields 1984), 2) it reduces the need for direct planning and management for a wide variety of contingencies (Turcotte 1974) and hence is more efficient (Ouchi 1979), and 3) it is perceived as more complete and objective by employees (Hirst 1981; Luthans 1977). Thus, even if procedural knowledge is high, organizations may use output control instead of process control when perfor- mance documentation is also high. The notion that a given form of control can be achieved independently of the other task characteristic also fails to explain the use of informal controls. The Ouchi model implies that at low levels of

procedural knowledge, use of informal controls is nega- tively associated with performance documentation, whereas at high levels of procedural knowledge, use of informal controls is unrelated to performance documentation. These phenomena can only be captured if the joint effects of the two task characteristics are included in predicting choice of a particular control form.

The second assumption implicit in Jaworski and Macln- nis' study is that the notion of task-control fit that underlies Ouchi's model is consistent with the notion of task-control fit that is necessary for studying employee responses. It should be noted that Ouchi's concept of fit is based on whether a particular task condition justifies the use of a particular control type. The bases for justification usually include criteria such as efficiency, ease of implementation, and convenience. Jaworski and Maclnnis (1989), for exam- ple, suggest that the use of output control when perfor- mance documentation is high indicates presence of fit. The meaning of this fit is that managers find output control to be convenient and easy to implement for tasks characterized by high perfo(mance documentation. But can we expect this fit to lead to lower levels of negative employee responses? The answer is, probably not (Hirst 1983; Porter, Lawler, and Hackman 1975). When employees view a control form, they are probably less concerned with its ease of implemen- tation for managers than how the control provides relevant information for performing their own jobs better. Using this logic, this study suggests that high procedural knowledge needs to be present for the "performance documentation- output control" fit to lead to reduced levels of negative employee responses. High procedural knowledge would en- able employees to know how behaviors need to be changed to narrow gaps in performance (that are identified by the use of output control). If procedural knowledge is low, employ- ees may know how they are performing but have little knowledge about why they are performing well or poorly. These arguments imply that while fit (as viewed by the organization) may be achieved when performance docu- mentation is high, output control may not lead to a reduc- tion in negative employee responses unless procedural knowledge is also high.

Alternatively, fit, as viewed from the implementation per- spective, may not be achieved, even though employee re- sponses may be positive. For example, while informal con- trois are not appropriate or necessary for tasks characterized by high procedural knowledge and high performance docu- mentation, employees, however, may show lower levels of job tension and dysfunctional behavior with informal con- trois because these controls do not demand explicit monitor- ing of their behaviors or outcomes. At the least, informal controls will not increase negative employee responses in this task condition. These examples indicate that there is a fundamental problem in positing that (lack of) fit, as viewed by the organization, will lead to less (more) nega- tive employee responses.

The above arguments can be best illustrated with an ex- ample from a typical salesforce situation. Assume that a salesperson has been unable to achieve an assigned quar- terly sales target. The supervisor may compare actual per- formance to the target and inform the salesperson that

JAMS 294 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

he/she has to improve his/her performance. The means by which this may be carried out is left to the discretion of the salesperson. In this situation, a salesperson who has infor- mation about the behavioral changes needed to overcome the performance deficiency will experience lower job ten- sion than one who does not have this information. This suggests that employee responses are contingent on the two task characteristics jointly. In other words, although the organization may be justified in using output control for tasks characterized by high performance documentation, employees may feel tense when procedural knowledge is low because of the difficulties associated ,with identifying specific behavioral changes needed for performance im- provements. The overall implication is that, from the em- ployee's perspective, fit is not achieved.

In summary, the major objective of this study is to pro- vide alternative contingency effect hypotheses to the ones specified in Jaworski and Maclnnis (1989) by incorporating the previously mentioned issues. These hypotheses examine the joint effect of performance documentation and pro- cedural knowledge on the use of controls and the joint mod- erator effect of these two task characteristics on the relation- ships of each of the four control forms (output, process, self, and professional) with each of the three employee responses (job tension, dysfunctional behavior, and infor- mation asymmetry).

HYPOTHESES

Task Characteristics and Control Forms

Ouchi (1979) posited that control can be achieved by monitoring employee behaviors (process control) when pro- cedural knowledge is high and by monitoring employee performance (output control) when performance documen- tation is high. The rationale for using process control when procedural knowledge is high is that the outcome of work behavior is ensured (even without output measurement) if employee behavior conforms to the optimal way of per- forming tasks. The rationale for using output control when performance documentation is high is that by pointing out the ultimate performance gaps, employees will change their behaviors suitably. When there are high levels of both per- formance documentation and procedural knowledge, an or- ganization could use either output or process control. Thompson (1967) and Ouchi and Maguire (1975) suggested that these two forms of control may be substitutable in this task condition. Rockness and Shields (1984) pointed out that output control may be favored over process control when managers are provided an option between the two. One reason is that use of output control reduces the need for direct planning and management for a wide variety of con- tingencies (Turcotte 1974). But, at the same time, it is possi- ble that the use of output control may be somewhat reduced when both performance documentation and procedural knowledge are high as compared to when performance doc- umentation alone is high, and the use of process control may be somewhat reduced when both performance docu- mentation and procedural knowledge are high relative to when procedural knowledge alone is high.

Finally, when both performance documentation and pro- cedural knowledge are low, neither process nor output con- trol is feasible, and informal controls, such as professional and self, may be necessary. These relationships can be rep- resented in the form of the following two main effect and three contingency effect hypotheses.

HI: The greater the performance documentation, the greater the use of output control.

H2: The greater the procedural knowledge, the great- er the use of process control.

H3: The association between performance documen- tation and the use of output control is higher when procedural knowledge is high rather than when it is low.

H4: The association between procedural knowledge and the use of process control is lower when performance documentation is high rather than when it is low.

H5a: The lower the performance documentation and procedural knowledge, the greater the use of self control.

H5b: The lower the performance documentation and procedural knowledge, the greater the use of professional control.

Controls and Employee Responses

Output Control It was mentioned previously that when performance doc-

umentation is high, output control is the most appropriate control form to be used. This control form enables manag- ers to provide employees with feedback regarding whether their performance conforms to expected levels. For this rea- son, Jaworski and Maclnnis (1989) suggested that output control should be associated with lower levels of negative employee responses.

H6: Output control is more likely to decrease job ten- sion, dysfunctional behavior, and information asymmetry when performance documentation is high than when it is low.

However, from an employee's perspective, a perfor- mance system should not only show that something is going wrong; it should also point out why this happened and what is necessary to correct the situation (Luthans 1977). If the relevant procedures are not pointed out by management, employees may have to do with self-help. Self-help, how- ever, is only possible when procedural knowledge is high. This means that the use of output control should result in lower levels of negative responses only when both perfor- mance documentation and procedural knowledge are high.

On the other hand, the use of output control may be

JAMS 295 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

associated with higher levels of negative responses when either (or both) performance documentation or procedural knowledge is low. When performance documentation is low, employees will definitely question the objectivity and completeness of output measures used for comparing with standards; however, even when performance documentation is high, questions about the completeness of the output measures may arise, particularly when procedural knowl- edge is low (Hirst 1981). Hirst (1983) found a higher level of job tension with the use of output control in situations involving high performance documentation but only when high performance documentation was associated with low procedural knowledge. Thus, we present an alternative hy- pothesis to H6.

H6a: Output control is more likely to decrease job tension, dysfunctional behavior, and informa- tion asymmetry when performance documenta- tion and procedural knowledge are high than when performance documentation and/or pro- cedural knowledge are/is low.

Process Control Jaworski and Maclnnis (1989) suggested that when pro-

cedural knowledge is high, process control is the most ap- propriate control form to be used. Procedural knowledge allows managers to take an active role in monitoring em- ployee behaviors and recommending changes to them whenever they are needed. Moreover, employees are more likely to accept the recommended changes when procedural knowledge is high than when procedural knowledge is low. The reason is that employees will have greater faith in su- pervisory feedback when they can clearly see the cause- effect relationships. And, since there is adequate fit, there should be a reduction in negative employee responses.

H7: Process control is more likely to decrease job tension, dysfunctional behavior, and information asymmetry when procedural knowledge is high than when it is low.

However, employees (and supervisors) can check wheth- er the procedures they have incorporated are appropriate in fulfilling the output goals only if performance measures are available. Because employees can trace changes in overall performance to changes in task behaviors, the use of pro- cess control should result in lower levels of negative re- sponses when both performance documentation and pro- cedural knowledge are high. Use of process control may, however, be associated with higher levels of negative re- sponses when either performance documentation or pro- cedural knowledge is low. When procedural knowledge is low, employees may question the objectivity and the com- pleteness of the behavioral changes suggested; when perfor- mance documentation is low, employees may not be aware of the consequences of the behavioral changes. Thus, we present an alternative hypothesis to H7.

H7a: Process control is more likely to decrease job tension, dysfunctional behavior, and informa- tion asymmetry when performance documenta-

tion and procedural knowledge are high rather than when procedural knowledge and/or perfor- mance documentation are/is low.

Informal Controls Control theory considers self and professional controls to

be appropriate and the only choice (compared to formal controls) when both performance documentation and pro- cedural knowledge are low (Oucbi 1979). Studies that have explored the effect of self-reinforcement on motivation have, however, shown that control systems that allow em- ployees to keep a record of their own performance to use as a continuous feedback system and for reinforcement pur- poses (i.e., self control) help them increase their motivation (Bandura and Cervone 1983). This is particularly so when the goals can be specified in sufficiently detailed behavioral terms (i.e., when procedural knowledge is high) to provide adequate guidance for the actions that must be taken to obtain desired outcomes (i.e., when performance documen- tation is high) (Kerr 1990). The underlying logic is that people are very good at modifying and correcting their work behavior on their own if information about how they are doing is made available and if they possess the knowledge about how they can change their behavior.

Additionally, studies have also shown that control achieved through interaction with group members alleviates bureaucratic pressure and enhances feelings of belonging to a group (Argyris 1953). These benefits may also be rein- forced by the perception that only fellow professionals know enough about the work to evaluate it competently. Such interactions, for tasks characterized by high role clari- ty, have been shown to facilitate group effectiveness and development of positive employee attitudes (Sarbin and Al- len 1968; Collins and Guetzkow 1964). The implication, therefore, is that negative responses will be lower with the use of self and professional controls when both procedural knowledge and performance documentation are high.

Use of informal controls when either procedural knowl- edge or performance documentation is low (or both are low), on the other hand, should not pose any disadvantages from the attitudinal perspective of employees. In fact, in- creased challenge faced by employees when both perfor- mance documentation and procedural knowledge are low, when matched by self-control opportunities, increases per- ceived self-competency and reduces mental strain (Karasek 1979). Although informal controls may be unnecessary in other situations (as viewed from the implementation per- spective because of the possibility of using formal controls), there is no reason to believe that employees may react nega- tively to such controls in other situations. This is because employees (individually or as a group) prefer to exercise some control over their own destiny but they often receive a greater share of the reinforcing consequences when they are successful (Frederiksen 1982). Thus, it is hypothesized that:

H8: Task characteristics do not moderate the relation- ships between self control and job tension, dys- functional behavior, and information asymmetry.

H9: Task characteristics do not moderate the relation- ships between professional control and job ten-

JAMS 296 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

sion, dysfunctional behavior, and information asymmetry.

Because Jaworski and Maclnnis did not consider the above relationships, these hypotheses are not stated as alternative hypotheses.

RESEARCH METHOD

Sample

The sampling frame for the mail survey was comprised of 1,298 marketing professionals drawn randomly from the 1990 AMA roster. Forty-eight of these individuals had changed their positions and hence could not be contacted, which reduced the sample base to 1,250. Of these, 335 responded to the questionnaire mailed, resulting in a re- sponse rate of 26.8 percent. The responses of high-level executives such as marketing directors and presidents were removed from the sample (29 responses out of 91 mailed) because these individuals were considered to be the makers of, rather than responders to, control systems in organiza- tions. In addition, six responses were excluded because of missing data. This reduced the sample base to 1,159 and the responses to 300, resulting in a response rate of 25.9 per- cent.

The sample covered a wide variety of marketing profes- sionals who differed in the type of function performed and the organizational level. The sample, for example, included product managers, marketing managers, general managers, marketing research analysts, marketing planning and devel- opment personnel, and sales and marketing personnel. Age- wise, 3.7 percent of the respondents were from the age category of less than 25 years, 72.6 percent were from the age category of 25-39, 20.9 percent were from the age category of 40-55, and 2.8 percent were from the age cate- gory of 56-70 years. Gender-wise, the sample consisted of 49.1 percent female and 50.9 percent male respondents. Education-wise, the sample consisted of 32.0 percent with high school diplomas, 56.5 percent with bachelor degrees, 9.3 percent with master degrees, and 2.2 percent with doc- toral degrees.

To evaluate nonresponse bias, an analysis of early versus late respondents was carried out by determining the date of receipt of the response and regressing important constructs used in the study against time taken to respond. This meth- od is appropriate for testing nonresponse bias because late respondents are expected to resemble nonrespondents (Armstrong and Overton 1977). Results of the regression analysis showed no significant relationship with time for any of the constructs, suggesting that nonresponse bias may not be a significant issue.

Measurement Scales

The survey questionnaire was designed primarily on the basis of measurement scales used by Jaworski and MacInnis (1989). Modifications were made to four scales that were identified by them as having less-than-desirable scale prop- erties. These included procedural knowledge, performance

documentation, self control, and job tension. A complete description of the scales used in the study is provided in the Appendix.

All scales were assessed for internal and external con- sistency by using correlation analysis, factor analysis, and Cronbach's alpha. The external consistency assessment was performed independently for three sets of constructs--task characteristics, controls, and employee responses. Results showed that a two-factor model, a four-factor model, and a three-factor model were better than a single-factor model (as indicated by changes in chi-square values) for each of the above sets of constructs, respectively. This finding sug- gests, for example, that the measures of the four controls tap four distinct dimensions and that each of the four mea- sures is unidimensional. Similar implications apply for the two task characteristics and the three types of employee responses. With the exception of self control, all constructs satisfied the Nunnally (1978) reliability criterion of .70.

Analyses

For examining H I through 1-15, regression analyses of each of the output (OC), process (PC), self (SC), and pro- fessional (PRC) controls against performance documen- tation (PD), procedural knowledge (PK), and their in- teractions were performed. The contingency hypotheses specified in 1-16 and tt6a were tested using three hierarchi- cal regression models. The first model (Model A) captured the main effects of all four control types on a particular negative employee response, l The second model (Model B) included the main effects and a two-way interaction term reflecting the interaction between output control and perfor- mance documentation. 2 The third model (Model C) in- cludes the effects in the second model and a three-way interaction effect reflecting the joint influence of out- put control, performance documentation, and procedural knowledge) A comparison of the variance explained by Models A and B provides a test for H6, and that for Models B and C provides a test for H6a (Chow 1960). Similarly, to test 1-17 and 117a, the variance explained by Model A is compared with that of Models D and E, respectively. 4,5

It should be noted that if the coefficients d 6 and f6 for the three-way interactions in Models C and E are statistically significant, the implication is that performance documenta- tion and procedural knowledge together moderate the rela- tion between output control and negative response and be- tween process control and negative response, respectively. If the two coefficients are also negative, they provide sup- port for H6a and H7a, respectively. If the coefficients for the three-way interaction terms d 6 and f6 are not significant, but the coefficients of the two-way interaction terms c 5 and e 5 alone are significant, then support is provided for the contingency hypotheses H6 and H7 proposed by Jaworski and MacInnis.

To test tt8 and 119, Model A is compared with Models F and G that include, in addition to the main effect terms, the three-way interaction terms (SC*PD*PK) and (PRC*PD*PK), respectively. 6,7 The two-way interaction terms (PRC*PD, PRC*PK, SC*PD, and SC*PK) were not in the models due to the lack of any theoretical basis for their inclusion.

JAMS 297 FALL, 1993

M A R K E T I N G C O N T R O L S A N D E M P L O Y E E R E S P O N S E S A G A R W A L A N D R A M A S W A M I

TABLE 1 Regression Analysis Results For Relationships Between Task Characteristics and Controls

Independent Variable

Performance Procedural Documentation Knowledge

Dependent Variable (PD) (PK) PD*PK R 2 F-Value

Outpu t control .59 a .12 b - . 0 5 b .26 36.67 a P rocess cont ro l .43 a .21 a - .05 b .19 24.04~ S e l f cont ro l .22 ~ .06 - . 0 2 .04 4 .42 a P ro fes s iona l cont ro l .16 b .20 ~ - .02 .08 9 .13 ~

,,p < .01. bp < .05.

An important complaint against the use of multiplicative terms in regression analysis focuses on the issue of collin- earity (Blalock 1979). Multiplicative terms usually exhibit strong correlations with their component parts and inflate the standard errors for the regression coefficients. This issue was addressed in two ways. First, the model proposed does not include procedural knowledge and performance docu- mentation as independent explanatory variables. This was done in accordance with the Jaworski and Maclnnis control model that specifically excludes such direct effects. Second, the main effect variables were mean-centered and stan- dardized prior to forming the multiplicative term. Such a transformation reduces the correlations between the product term and the component parts of the term. Collinearity is also possible between the two-way and three-way interac- tion terms in some of the models. Because this could not be entirely avoided, a test of the variance inflation factor (VIF) for each interaction term was computed. The highest VIF value was 1.75, much below the cut-off criterion of 10 recommended by Neter, Wasserman, and Kutner (1985). This finding suggests that collinearity is not a likely threat to the substantive conclusions drawn from the parameter estimates.

Using standardized scores, however, can lead to a prob- lem in interpreting interactions. For example, a low/low combination of any two standardized variables would yield a large positive term, making it indistinguishable from the high/high combination of the same two variables. An inter- action term is interpretable only if the low/low combination is represented by a smaller number relative to the high/high combination. We have used a transformation suggested by Cooper and Nakanishi (1983) of the original variables to calculate the interaction term in order to rectify this problem. 8

RESULTS

Output and Process Control

The results pertaining to relationships hypothesized be- tween task characteristics and control forms are reported in Table 1. The use of output control is positively (b = 0.59, p < .01; Table I) associated with performance documentation

(as hypothesized in HI), and the use of process control is positively (b = 0.21, p = .01; Table 1) associated with procedural knowledge (as hypothesized in H2). The interac- tion term in the equation relating output control and perfor- mance documentation is negative (b = - . 05 , p < .05; Table 1), implying that the presence of procedural knowl- edge weakens the association between output control and performance documentation. This result does not support H3. On the other hand, the negative coefficient (b = - . 05 , p < .05; Table 1) for the interaction term in the equation relating process control and procedural knowledge supports H4. Finally, the interactive effects of the two task charac- teristics on the use of self and professional controls are not significant, thus failing to support H5a and H5b, respec- tively.

Controls and Employee Responses The results pertaining to the relationships hypothesized

between control forms and employee responses are reported in Table 2, 3, 4, and 5. The use of output control is not associated with job tension and dysfunctional behavior but is associated with information asymmetry (b = - . 35 , p < .01; Table 2). The use of process control is associated with job tension (b = .15, p < .01; Table 3), dysfunctional behavior (b = .19; p < .01; Table 3), and information asymmetry (b = - . 14, p < .05; Table 3). The use of self control is associated with dysfunctional behavior (b = - . 29 , p < .01; Table 4) but not with job tension and infor- mation asymmetry. Finally, the use of professional control is associated with job tension (b = - . 20 , p < .01; Table 5), dysfunctional behavior (b = - . 19, p < .01; Table 5), and information asymmetry (b = - . 29 , p < .05; Table 5). These main effects results are consistent with the main ef- fects findings of Jaworski and Maclnnis (1989).

Next, the contingency effects (corresponding to H6 and H7) were evaluated. The results show that performance documentation does not moderate the relationships between output control and job tension, dysfunctional behavior, and information asymmetry, thus not supporting H6. The differ- ence in R-square between the model that included main effects and two-way interaction effects (Model B) and the model that included only main effects (Model A) is not significant for the three employee responses, namely, job

JAMS 298 F A L L , 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

TABLE 2 Regression Analysis Results For Contingent Effects of Task Characteristics on Relationship

Between Output Control and Employee Responses

Independent Variable

Output Process Self Dependent Control Control Control Variable (OC) (PC) (SC)

Professional Control Cumulative Incremental Incremental (PRC) OC*PD c OC*PD*PK a R 2 F-Value R 2 F-Value

Job tension Model A e .03 .15" .08 Model BY .00 .14 ~ .08 Model Cg .130 .14" .08

Dysfunctional behavior

Model A .05 .19 ~ - . 2 9 a Model B .04 .19 ~ - . 2 9 ~ Model C .01 .19 a - . 3 0 a

Information asymmetry

Model A - . 3 5 a - . 1 4 b - . 0 6 Model B - . 3 8 - - . 1 5 b - . 0 6 Model C - . 3 9 a - . 1 5 b - . 0 7

- . 2 0 ~ .09 7.61 ~ - . 2 0 ~ .02 .10 6.49 ~ .01 1.90

, - . 2 0 4 .00 .02 .10 5.394 .00 .33

- . 1 9 ~ .17 15.03 ~ - . 1 9 ~ .01 .17 12.03 ~ .00 .21 - . 1 8 ~ .04 - . 0 1 b .18 11.00 a .01 5.07 b

- . 2 9 a .26 26.03 a - . 2 9 ~ .02 .26 21.00~ .00 .80 - . 2 9 a .03 .00 .26 17.50 ~ .00 .40

ap < .01. bp < .05. cPerformance documentation. dprocedural knowledge. "Model A includes main effects only. /Model B includes main and two-way interaction effects. sModel C includes main, two-way, and three-way interaction effects.

TABLE 3 Regression Analysis Results For Contingent Effects of Task Characteristics on Relationship

Between Process Control and Employee Responses

Independent Variable

Dependent Variable

Output Process Self Professional Control Control Control Control Cumulative Incremental Incremental

(OC) (PC) (SC) (PRC) PC*PK e OC*PK*PD a R e F-Value R e F-Value

Job tension Model A/ .03 .15 a .08 - . 2 0 ~ .09 7.61 ~ Model Ds .03 .16 a .08 - . 1 9 ~ - . 0 1 .09 6.15 ~ .00 .33 Model E h .02 .18 a .08 - . 1 9 4 - . 0 3 .01 b .11 6.68 a .02 3.97 b

Dysfunctional behavior

Model A .05 .194 - . 2 9 ~ - . 1 9 a .17 15.03 ~ Model D .05 .24 a - . 2 9 ~ - . 1 8 a - . 0 2 b .18 13.38 ~ .01 5.82 ~ Model E .04 .25 a - . 2 9 a - . 1 7 4 - . 0 3 c .00 .19 11.18 ~ .01 .36

Information asymmetry

Model A - . 3 5 " - . 14 t' - .06 - . 2 9 " .26 26.03" Model D - . 3 5 " - . 1 t - . 0 6 - . 2 8 " - . 0 1 .26 21.01 ~ .00 .80 Model E - . 3 5 " - . 1 0 - . 0 6 - . 2 8 " - . 0 4 .01 .26 17.66" .00 1.21

~p < .01. bp < .05.

"p < .10. aperformance documentation. eProcedural knowledge. :Model A includes main effects only. gModel D includes main and two-way interaction effects. hModel E includes main, two-way, and three-way interaction effects.

JAMS 299 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

TABLE 4 Regression Analysis Results For Contingent Effects of Task Characteristics on Relationship

Between Self Control and Employee Responses

Dependent Variable

Independent Variable

Output Process Self Professional Control Control Control Control Cumulative Incremental Incremental

(OC) (PC) (SC) (PRC) SC*PDc*PK d R 2 F-Value R z F-Value

Job tension Model A e .03 .15 ~ .08 - .20`" .09 7.61 ~ Model FY .03 .15 ~ .08 - . 2 0 ~ .00 .09 6.07 ~ .00 .01

Dysfunctional behavior

Model A .05 .19 ~ - . 2 9 a - . 1 9 ~ .17 15.03" Model F .05 .20 a - . 2 6 ~ - . 1 8 ~ - . 0 1 b .18 13.19- .01 5.08 b

Information asymmetry

Model A - . 3 5 a - . 1 4 b - . 0 6 - .29`" .26 26.03" Model F - .35`" - . 1 4 b - . 0 5 - . 2 9 a .00 .26 20.82 ~ .00 ,24

op < .01.

bp < ,05. cPerformance documentation. dProcedural knowledge. eModel A includes main effects only. /Model F includes main and three-way interaction effects.

tension, dysfunctional behavior, and information asymmet- ry (F-values = 1.90, 0.21, and 0.80, respectively; Table 2). Procedural knowledge, on the other hand, was found to only moderate the relationship between process control and

dysfunctional behavior (F-value = 5.82, p < .01; based on a comparison of Models A and D in Table 3). The coeffi- cient estimate for the interaction term PC*PK was negative suggesting that at higher levels of procedural knowledge,

TABLE 5 Regression Analysis Results For Contingent Effects of Task Characteristics on Relationship

Between Professional Control and Employee Responses

Independent Variable

Output Process Self Professional Dependent Control Control Control Control Variable (OC) (PC) (SC) (PRC) PRC*pDd*pK e

Cumulative Incremental Incremental R 2 F-Value R 2 F-Value

Job tension Model AI .03 .15 a .08 - .20 a .09 7.61 ~ Model Gg .03 .15" .08 - .20`" .00 .09 6.07`" .00 ,01

Dysfunctional behavior

Model A .05 .19 ~ - . 2 9 a - . 1 9 a .17 15.03`" Model G .06 .20 ~ - . 2 8 a - . 1 6 ~ - . 0 1 b .19 13.48 ~ .02 6.20 b

Information asymmetry

Model A - . 3 5 a - . 14 b - . 0 6 - . 2 9 a '.26 26.03 ~ Model G - . 3 4 a - , 1 4 c - . 0 6 - . 2 8 " .00 .26 20.79 ~ .00 .13

`'p < .01. bp < .05. cp < .10. aperformance documentation. eProcedural knowledge. /Model A includes main effects only. sModel G includes main and three-way interaction effects.

JAMS 300 FALL, 1993

M A R K E T I N G CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

use of process control was associated with lower dysfunc- tional behavior. The other two moderator relationships in- volving procedural knowledge (corresponding to job ten- sion and information asymmetry) were, however, not supported (F-values = .33 and .80, respectively; Table 3). There is thus limited support for the presence of a two-way interaction between procedural knowledge and the use of process control (i.e., in one out of three cases only).

The next set of analyses tests whether performance docu- mentation and procedural knowledge jointly moderate the relationships between controls and employee responses. This involves testing the increase in R-squaa'e for statistical significance between Models B and C to test H6a, between D and E to test HTa, between A and F to test 1-18, and between A and G to test H9. The results provide mixed evidence because five out of a possible twelve three-way interaction terms were as hypothesized, and three more were significant but in the opposite direction. These results are discussed in detail below.

H6a. This hypothesis is supported for the relationship between output control and dysfunctional behavior, but not supported for the relationships between output control and job tension and output control and information asymmetry. The incremental R-square in the three cases, job tension, dysfunctional behavior, and information asymmetry, were .00, .01, and .00 (corresponding F-values were 0.33, 5.07, and 0.40; based on a comparison of Models B and C in Table 2), respectively. The negative coefficient (b = - . 0 1 , p < .05; Table 2) for the significant three-way interaction term, OC*PD*PK, suggests that the use of output control reduces dysfunctional behavior only at high levels of both performance documentation and procedural knowledge.

H7a. This hypothesis is not supported. Although a sig- nificant three-way interaction (incremental R-square = .02, p < 0.05; based on a comparison of Models D and E in Table 3) was found in one out of three cases (for the rela- tionship between process control and job tension), the coef- ficient estimate was positive rather than negative. This coef- ficient value implies that at higher levels of performance documentation and procedural knowledge, the use of pro- cess control increases felt job tension of employees.

H8. This hypothesis is partially supported. The hypoth- esized null effect was observed in all cases except one. The results show that while performance documentation and procedural knowledge do not moderate the relationships between self control and the dependent variables, job ten- sion and information asymmetry, they moderate (b = - . 0 1 ; p < .05; Table 4) the relationship between self control and dysfunctional behavior. The incremental R-square value (.01 based on a comparison of Models A and F) is signifi- cant at the p < 0.05 level.

H9. This hypothesis is also partially supported. The hypothesized null effect was observed in all cases except one. The results show that while performance documenta- tion and procedural knowledge do not moderate the rela- tionships between professional control and the dependent variables, job tension and information asymmetry, they moderate (b = - . 0 1 , p < .05; Table 5) the relationship between professional control and dysfunctional behavior. The incremental R-square value (.02 based on a comparison of Models A and G) is significant at the p < 0.05 level.

=..,

C o (J

r

O

DISCUSSION

This study was designed to examine the joint effect of performance documentation and procedural knowledge on the use of controls and the joint moderator effect of these two task characteristics on the relationships between each of four control forms (output, process, self, and professional) and three employee responses (job tension, dysfunctional behavior, and information asymmetry). A major postulate of this study was that the contingency effect of task charac- teristics on the relationships between controls and employee responses should be examined from an employee's perspec- tive rather than the organizational implementation perspec- tive.

Task Characteristics and Control Forms

Output and Process Controls The results provide clear evidence that a richer under-

standing of organizational use of controls can be obtained by studying the joint, in addition to the main, effects of the two task characteristics. For example, while the main effect results suggest that increasing levels of performance docu- mentation or procedural knowledge can lead to greater use of output control, the interaction effect could weaken this effect. To facilitate a finer interpretation of the interaction term, we computed the predicted values of output control at different levels of the two task characteristics using regres- sion equations. A visual inspection (see Figure 1) suggests that high procedural knowledge strengthens the use of out- put control at low to moderate levels of performance docu- mentation but weakens its use at high levels of performance documentation. What this means is that higher cause-effect

FIGURE 1 Predicted Values of Output Control at Selected

Levels of Performance Documentation (PD) and Procedural Knowledge (PK)

1

-2

f I I I I I

- 2 -1 0 1 2

Performance Documentation

[] P K = -2 . P K = 0 = P K = + 2

Note: All effects are in standardized units.

JAMS 301 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

O

r

o

knowledge provided by procedural knowledge facilitates the use of output control in conditions where its use may not be totally justifiable (i.e., at low to moderate levels of per- formance documentation). On the other hand, higher cause- effect knowledge weakens the use of output control in con- ditions where its use may be better justified (i.e., at high levels of performance documentation). These results sug- gest that managers do not choose output control solely on the basis of availability of performance documentation. At higher levels of performance documentation, increases in procedural knowledge may open up the possibility of using other forms of control. '

Similarly, the results also show complex relationships between procedural knowledge and the use of process con- trol at different levels of performance documentation. As hypothesized, the strength of this relationship was weak- ened when performance documentation was high. The slope characterizing this relationship was greater when perfor- mance documentation was low to moderate than when it was high (see Figure 2). Similar to the explanation offered for output control, it appears that process control is being substituted by other forms of control when both task condi- tions are high. A lower level of use of process control may also be a natural reaction on the part of managers to employ- ees' perception that it curtails their freedom and initiative, particularly when output measures are well defined. Porter, Lawler, and Hackman (1975) indicated that this is espe- cially likely to occur when the control system deprives em- ployees of certain discretionary decision making about how they are to carry out their job duties.

Finally, we find that while performance documentation,

r

O t .

FIGURE 2 Predicted Values of Process Control at

Selected Levels of Procedural Knowledge (PK) and Performance Documentation (PD) a

1

-1

-2

Y ! I I I I

- 2 -1 0 1 2

Procedural Knowledge

. a - - - PD = - 2 . P D = 0

= P D = + 2

Note: All effects are in standardized units.

as expected, had a stronger relationship with output control than with process control (b = .59 versus .43; Table 1) and that procedural knowledge, likewise, had a stronger rela- tionship with process control than with output control (b = .21 versus. 12; Table 1), the finding that process control is more strongly related to performance documentation than to procedural knowledge (b = .43 versus .21; Table 1) came as a surprise. No previous study has either hypothesized or empirically investigated this relationship. One possible ex- planation for this result is that implementation of process control requires information not only on the manner in" which behaviors need to be changed to reduce performance gaps, but also on who needs behavioral counseling and when. This kind of information, it should be noted, is only available through proper performance documentation.

Self and Professional Controls The hypothesis that use of self and professional controls

will be the greatest at low levels of performance documenta- tion and procedural knowledge was rejected. Contrary to expectations, use of these controls appears to be dependent on availability of performance documentation. This finding implies that organizational use of informal controls may not conform entirely to the postulates of the Ouchi (1979) con- trol model.

In summary, it appears that managers do not make control choices solely on the basis of fit with task characteristics (termed as efficiency basis by some researchers). It is possi- ble that managers may have experienced difficulty trading efficiency against motivation and morale benefits derived from using a particular form of control and hence may have focused instead on potential benefits. This difficulty, in turn, may have been instrumental in choosing informal con- trois even for tasks that were relatively more certain.

Controls and Employee Responses

The study findings also support the relevance of examin- ing the joint moderating effects of the two task characteris- tics on employee responses to controls. Significant mod- erating effects were found for the relationships between output control and dysfunctional behavior, process control and job tension, and informal controls and dysfunctional behavior. These results are unique because they represent the first empirical evidence in the marketing control lit- erature for the contingent role played by the two task char- acteristics, performance documentation and procedural knowledge. In light of the common view among researchers that finding two-way moderating effects in field settings is extremely difficult (Sackett, Harris, and Orr 1986), the sig- nificant results obtained for four out of twelve possible three-way effects is encouraging. The implications of the joint effect findings as well as nonfindings are discussed below.

Output and Process Controls Organizations may typically use either output or process

control when both performance documentation and pro- cedural knowledge are high. Both forms of control are easy to implement from the viewpoint of the organization. Em- ployees, however, appear to experience higher job tension

JAMS 302 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

when subject to process control (note that the three-way interaction term in the equation relating process control and job tension is positive and significant). On the other hand, the use of output control appears to lead to lower dysfunc- tional behavior in the high performance documentation, high procedural knowledge condition. Both of these results imply that employees perceive output control, rather than process control, as playing a facilitating role in the goal fulfillment process when performance documentation and procedural knowledge are high. By providing feedback, output control gives a sense of progression and achievement to employees. The outcome feedback itself is accepted by employees because high procedural knowledge implies that performance documentation may be perceived as objective and complete, and employees can decide for themselves the changes in behaviors needed to maximize goal fulfillment. Employees may, therefore, perceive less need to withhold information or to distort its presentation to their supervisors. Withholding information may also not be an option because supervisors may possess knowledge about what can and cannot be achieved (Hirst 1981). With a clearly defined set of behaviors and the ability to measure and verify the re- suits, falsification of information would be risky (Birnberg, Turopolec, and Young 1983).

The results provide legitimacy for nonreliance on process control by managers in any task condition. Future research might investigate the more fine-grained aspects of reactions evoked by process control. In particular, information relat- ing to aspects of process control that are liked or disliked in different task conditions would be extremely useful to prac- titioners.

Self and Professional Controls The use of self or professional control was expected to

uniformly reduce negative employee responses under all task conditions. However, the negative coefficients found for two of the interaction terms (SC*PD*PK and PRC*PD*PK) indicate that this reduction is much more likely under certain conditions. Specifically, when organi- zations allow the use of professional and self controls in conditions where they are deemed to be unnecessary (i.e., when performance documentation and procedural knowl- edge are both high), employees reciprocate by exhibiting fewer dysfunctional behaviors.

The managerial implication of the above result is that any organizational process encouraging self and professional initiative and feedback should have a positive potential pay- off. Managers should, therefore, continuously attempt to induce employees to set their own performance goals in explicit terms and also provide flexibility for procedures used in achieving these goals. This flexibility, coupled with continuous monitoring of performance, can reduce the inci- dence of negative employee attitudes and behaviors that may eventually lead to dissatisfaction and turnover.

CONCLUSION

This study was designed to investigate the interactive (in addition to the independent) influence of two task char- acteristics, performance documentation and procedural

knowledge, on employees' responses to organizational choice of controls. In doing so, it extended Jaworski and Maclnnis's efforts to examine the contingency model of Ouchi (1979). The results showed that control-employee response relationships are in fact moderated by the charac- teristics of tasks.

The inferences drawn from these results are, however, subject to several limitations. First, caution must be exer- cised in drawing cause-effect inferences because of the cross-sectional research design employed. Second, like most studies dependent on self-report data, the associations among constructs may be inflated as a result of common methods variance. In the absence of multiple measures for each construct, an effort was made to control for this meth- od bias qualitatively. This involved disguising the objec- tives of the research from respondents, negatively wording some of the items, randomly placing items within the ques- tionnaire, and assuring complete anonymity. Third, mea- surement scales of some of the study constructs, especially self and professional controls, need further refinement; the reliability of self-control, in particular, was lower than what is traditionally considered appropriate. The professional control construct was not completely defined because while it focused on control exerted by peer employees, it ignored the control exerted by professional codes of ethics.

On the other hand, the research design used in this study had several positive features. For example, the study used a heterogeneous sample comprised of marketing employees at all levels, rather than just senior-level marketing managers. This sample selection is consistent with one of the recom- mendations of Jaworski and Maclnnis that a sample should span multiple organization levels so that variance on the task characteristics can be enhanced (1989, p. 416). The study also provided partial solutions to measurement limita- tions acknowledged in their study by improving measure- ment reliability for the procedural knowledge, performance documentation, and job tension constructs.

Overall the findings of the study provide two important implications. First, they indicate that Ouchi's control model may not capture the reality of how managers make control decisions. Ouchi's model is predicated on choosing a con- trol system on the basis of feasibility or ease of implementa- tion for a given task situation. Managers, however, as the study suggests, exercise caution in choosing controls solely on the basis of feasibility or ease of implementation. They may also consider the impact that controls would have on employee responses. In other words, managers seem to rec- ognize that they cannot expect employees to endorse a par- ticular control form because it is easy to implement.

Second, the use of informal controls was found to reduce dysfunctional behavior in task conditions that were deemed inappropriate from an organizational implementation per- spective. Informal controls may thus have a stronger con- trolling value than recognized within the Ouchi model. Al- though such an extreme position was not taken in the hypotheses, it makes sense to conside: the following postu- late. If employees consider informal controls to be useful and appropriate in the most uncertain task condition (low performance documentation and low procedural knowl- edge), then their application in somewhat more certain task conditions ought to be perceived as even more useful and

JAMS 303 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

appropriate. Future research is needed that replicates this study and tests the above postulate, particularly in situations where informal controls may clearly not be the most effi- cient control form.

In drawing the above conclusions, it should be empha- sized that although this study assumed that output, process, self, and professional controls constitute distinct control forms, they may not be mutually exclusive and hence may be implemented together. Combining their relative strengths may resolve competing requirements of efficiency and cre- ativity. Future research should investigate how task charac- teristics moderate the choice of different combinations of controls and how different combinations of controls may influence employee responses. Future research should also investigate the effects of other relevant organizational fac- tors such as environmental uncertainty, competitive inten- sity, organizational structure, control system characteristics, and so forth and individual differences such as need for independence, self-efficacy, competence, and locus of con- trol on employee response to controls.

ACKNOWLEDGMENT

Both authors contributed equally to this article and are responsible for content and remaining errors. The authors would like to thank Bernard J. Jaworski, V. Kumar, James E. McElroy, Paula Morrow, G. Premkumar, Rajendra K. Srivastava, Roy Teas, and the anonymous reviewers for their comments and suggestions on earlier drafts.

APPENDIX: MEASUREMENT SCALES

Performance Documentation (5-point, strongly disagree- agree)

1. Documents exist to measure my performance after ac- tivities are complete.

2. The documents that exist can adequately assess my performance.

3. Documents exist to assess my performance on most of my activities, a

4. Information about how my performance will be evalu- ated has been directly communicated to me. a

Procedural Knowledge (5-point, strongly disagree-agree)

1. There exists a clearly defined body of knowledge or subject matter that can guide me during my work.

2. It is possible to rely upon existing procedures and prac- tices to do my work.

3. On my job, there exist objective procedures that tell me exactly how to respond in all situations, a

4. On my job, I have to actively search for solutions be- yond normal procedures many times, a

Output Control (5-point, strongly disagree-agree)

1. Specific performance goals are established for my job. 2. My immediate boss monitors the extent to which I at-

tain my performance goals. 3. If my performance goals are not met, I would be re-

quired to explain why. 4. I receive feedback from my immediate supervisor con-

cerning the extent to which I achieve my goals. 5. My pay increases are based upon how my performance

compares with my goals.

Process Control (5-point, strongly disagree-agree)

1. My immediate boss monitors the extent to which I fol- low established procedures.

2. My immediate boss evaluates the procedures I use to accomplish a given task.

3. My immediate boss modifies my procedures when de- sired results are not obtained.

4. I receive feedback on how I accomplish my perfor- mance goals.

Self-control (5-point, strongly disagree-agree)

1. The major satisfaction in my life comes from my job. 2. The work I do on this job is very meaningful to me. 3. I feel that I should take the credit or blame for the

results of my work. 4. I like to do more than my share of the work at my job. a 5. I feel bad when I do a poor job. a,b

Professional Control (5-point strongly disagree-agree)

1. The division encourages cooperation between market- ing professionals.

2. Most of the marketing professionals in my division are familiar with each other's productivity.

3. The division fosters an environment where marketing professionals respect each other's work.

4. The division encourages job-related discussions be- tween marketing professionals.

5. Most marketing professionals in my division are able to provide accurate appraisals of each other's work.

Job Tension (5-point, strongly disagree-agree)

1. I experience tension in my job. 2. I experience job tension during performance evalua-

tions, c 3. If I don' t attain rfiy performance goals, I feel tense. 4. On my job, I feel tense if I have unfinished work. a 5. I feel tense if I don' t get adequate time to complete my

assignments to my satisfaction, a

Dysfunctional Behavior (5-point, strongly disagree-agree)

1. I tend to ignore certain job-related activities simply because they are not monitored by the division.

2. I work on unimportant activities simply because they are evaluated by upper management.

JAMS 304 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

3. Even if my productivity is inconsistent, I still try to make it appear consistent.

4. I have adjusted marketing data to make my perfor- mance appear more in line with division goals.

5. When presenting data to upper management, I try to emphasize data that reflect favorably upon me.

6. When presenting data to upper management, 1 try to avoid being the bearer of bad news.

Information Asymmetry (5 point, strongly disagree-agree)

1. I know how to accomplish the work I riormally encoun- ter.

2. I am intimately familiar with the day-to-day decisions related to my work.

3. I have developed an excellent working knowledge of my job.

4. I can quantitatively assess my performance soon after I complete my activities.

5. I am able to adequately assess my performance after I complete my activities.

6. I can specify the most important variables to monitor in my work.

7. I could specify performance objectives to cover the range of activities I perform.

The seven items below sponding items above to items.

are subtracted from the corre- form seven "difference score"

1. I know how to accomplish the work I normally encoun- ter.

2. I am intimately familiar with the day-to-day decisions related to my work.

3. I have developed an excellent working knowledge of my job.

4. I can quantitatively assess my performance soon after I complete my activities.

5. I am able to adequately assess my performance after I complete my activities.

6. I can specify the most important variables to monitor in my work.

7. I could specify performance objectives to cover the range of activities I perform.

N o t e : " These i tems were added to increase the number o f items

measuring a construct.

b These items were omitted following internal consistency analysis.

NOTES

1. (A)

2. (B)

3. (C)

4. (D)

Job Tension = b o + b I * OC + b z * PC + b~ * PRC + b 4 . SC

Job Tension = c o + c I * OC + c 2 * P C + c 3 * PRC + c 4 . SC + c 5 . OC * PD

Job Tension = do + dt * OC + d e * PC + d 3 * PRC + d o* SC + d~ * OC * PD + de * OC * PD * PK

Job Tension = e o + e / * O C + e 2 * PC + e 3 * PRC + e 4 . SC + e 5 * PC * PK

5. (E) JobTension = f o + f t * O C + f e * P C + f s * P R C + f 4 * S C + fs * PC * PK + fe * PC * PD * PK

6. (F) Job Tension = ho + hi * OC + h e * PC + h 3 * PRC + h 4 * SC + h 5 * SC * PD * PK

7. (G) Job Tension = i o + i I * OC + i 2 * PC + i s * PRC + i 4 . S C + i 5 * P R C * P D * P K

8. The transformation is carried out as follows:

p s i = 1 + ~ i f Z i ) ~ 0 = 1 / (1 + Zi) ifZe < 0

where Zij is the standardized response of respondent i on attribute j. This transformation provides interaction terms that have a theoreti- cal lower bound of zero and an upper bound of infinity.

REFERENCES

Argyris, Chris. 1953. "Human Problems With Budgets." Harvard Busi- ness Review 31 (January/February): 97-110.

Armstrong, J. Scott and Terry Overton. 1977. "Estimating Non-Response Bias in Mail Surveys." Journal of Marketing Research 14 (August): 396-402.

Bandura, Albert and Daniel Cervone. 1983. "Self-Evaluative and Self Efficacy Mechanisms Governing the Motivational Effects of Goal Sys- tems." Journal o f Personality and Social Psychology 45 (November): 1017-1028.

Birnberg, Jacob G., Lawrence Turopolec, and S. Mark Young. 1983. "The Organizational Context of Accounting." Accounting, Organizations, and Society 8 (2/3): 111-129.

Blalock, Hubert M. 1979. Social Statistics. New York: McGraw-Hill. Chow, Gregory C. 1960. "Test of Equality Between Sets of Coefficients in

Two Linear Regressions." Econometrika 28 (July): 591-605. Collins, Barry E. and Harold S. Guetzkow. 1964. A Social Psychology o f

Group Processes for Decision-Making. New York: John Wiley and Sons. Cooper, Lee G. and Masao Nakanishi. 1983. "Standardizing Variables in

Multiplicative Choice Models." Journal o f Consumer Research 10 (June): 96-108.

Flamholtz, Eric G., T.K. Das, and Anne S. Tsui. 1985. "'Toward an Integrative Framework of Organizational Control." Accounting, Organi- zations, and Society 10 (I): 35-50.

Frederiksen, Lee W. 1982. Handbook of Organizational Behavior Man- agement. New York: John Wiley and Sons.

Hirst, Mark K. 1981. "Accounting Information and the Evaluation of Subordinate Performance: A Situational Approach." Accounting Review 56 (October): 771-784.

- - . 1983. "Reliance on Accounting Performance Measures, Task Un- certainty, and Dysfunctional Behavior: Some Extensions." Journal o f Accounting Research 21 (Autumn): 596-605.

Jaworski, Bernard J. 1988. "'Toward a Theory of Marketing Control: Envi- ronmental Context, Control Types, and Consequences." Journal o f Mar- keting 52 (July): 23-39.

Jaworski, Bernard J. and Deborah J. Maclnnis. 1989. "Marketing Jobs and Management Controls: Toward a Framework." Journal o f Marketing Research 26 (November): 406-419.

Karasek, Robert A., Jr. 1979. "Job Demands, Job Decision Latitude, and Mental Strain: Implications for Job Redesign." Administrative Science Quarterly 24 (June): 285-308.

Kerr, Steven. 1990. "Substitutes for Leadership: Some Implications for Organizational Design." In Organizational Behavior and Management. Ed. Henry L. Tosi. Boston. MA: PWS-Kent Publishing, pp. 507-517.

Luthans, Fred. 1977. Organizational Behavior. New York: McGraw-Hill. Merchant, Kenneth. 1982. "'On the Incidence and Causes of Dysfunctional

Side Effects of Control Systems." Unpublished paper presented at the American Accounting Association Annual Meeting, Reno, NV.

Neter, John, William Wasserman, and Michael H. Kutner. 1985. Applied Linear Statistical Models: Regression, Analysis o f Variance, and Experi- mental Designs. Second Edition. Homewood, IL: Richard D. Irwin.

Nunnally, Jum C. 1978. Psychometric Theory. New York: McGraw-Hill. Ouchi, William G. 1979. "'A Conceptual Framework for the Design of

Organizational Control Mechanisms." Management Science 25 (Sep- tember): 833-847.

Ouchi, William G. and Margaret A. Maguire. 1975. "Organizational Con-

JAMS 305 FALL, 1993

MARKETING CONTROLS AND EMPLOYEE RESPONSES AGARWAL AND RAMASWAMI

troh Two Functions." Administrative Science Quarterly 20 (December): 559-569.

Porter, Lyman W., Edward E. Lawler III, and Richard J. Hackman. 1975. Behavior in Organizations. New York: McGraw-Hill.

Rockness, Howard O. and Michael D. Shields. 1984. "Organizational Control Systems in Research and Development." Accounting, Organiza- tions, and Society 9 (2): 165-177.

Sackett, Paul R., Michael M. Harris, and John M. Orr. 1986. "On Seeking Moderator Variables in Meta-Analysis of Correlational Data: A Monte Carlo Simulation Investigation of Statistical Power and Resistance to Type I Error." Journal of Applied Psychology 71 (May): 302-310.

Sarbin, Theodore R. and Vernon L. Allen. 1968. "Increasing Participation in a Natural Group Setting: A Preliminary Repot." The Psychological Record 18 (January): 1-7.

Thompson, James D. 1967. Organizations in Action. New York: McGraw- Hill.

Turcotte, William E. t974. "Control Systems, Performance, and Satisfac- tion in Two State Agencies." Administrative Science Quarterly 19 (March): 60-73.

ABOUT THE AUTHORS

San j eev A g a r w a l is Assistant Professor o f market ing at Iowa State University. He obtained his Ph.D. at the Ohio

State University. His research interests include the areas of international marketing and sales management. He has pre- viously published in Advances in International Marketing, International Trade Journal, Journal of the Academy of Marketing Science, Journal of lnternational Business Stud- ies, Journal of Personal Selling and Sales Management, and contributed to several national and international conference proceedings.

S r i d h a r N. R a m a s w a m i is Associate Professor of market- ing at Iowa State University. He obtained his Ph.D. at the Universi ty o f Texas at Austin. His research interests include sales management , marketing strategy, and international marketing. He has previously published in International Journal of Research in Marketing, International Trade Jour- nal, Journal of Economic Behavior and Organization, Jour- nal of the Academy of Marketing Science, Journal of Inter- national Business Studies, Journal of Personal Selling and Sales Management, Technology Forecasting and Social Change, and contributed to numerous conference proceed- ings.

JAMS 306 FALL, 1993