Exploring the influence of sales management practices on the industrial salesperson: A multi-source hierarchical linear modeling approach

Transcription

1 Journal of Business Research 60 (2007) Exploring the influence of sales management practices on the industrial salesperson: A multi-source hierarchical linear modeling approach George J. Avlonitis, Nikolaos G. Panagopoulos 1 Athens University of Economics and Business, Department of Marketing & Communication, Evelpidon 47 & Leykados 33, Athens , Greece Received 1 April 2005; received in revised form 1 January 2007; accepted 1 February 2007 Abstract Despite the importance of sales management practices (e.g., compensation, organization design) as critical determinants of salespeople's role stress, attitudes and outcomes, remarkably very few studies are available in the area. Unless the effects of sales management practices are determined, sales practitioners may rely on less efficient and less effective practices. The vast majority of previous studies relies on single-source data and has overlooked the multilevel nature of organizations when modelling relationships of interest. To address these gaps, this study examines the behavioral and psychological effects of sales management practices by applying a multilevel analytic approach to data gathered from two different data sources, sales managers and salespeople, respectively Elsevier Inc. All rights reserved. Keywords: Sales management; Hierarchical linear modeling; Training; Compensation; Territory design; Salesperson performance; Sales organization; Role stress 1. Introduction Sales management practices (e.g., training, selection, evaluation) constitute a crucial aspect of the sales manager's work, as they exert influence on the salesperson's motivation, satisfaction and performance (Churchill et al., 2000). Indeed, the literature (e.g., Bartol, 1999) is replete with provocative speculations regarding the impact of sales management activities on the sales force attitudes and performance. Considering that the sales function is the source of revenue for the majority of industrial companies (Cespedes, 1990), its effective management becomes a relentless necessity. Paradoxically, this recognition comes with the realization that although the relevant literature seems voluminous, remarkably very few The authors thank David A. Hofmann for his earnest encouragement and valuable suggestions regarding hierarchical linear modeling issues. The authors thank Arch Woodside (editor-in-chief), Hans Mühlbacher (associate editor) as well as the two anonymous JBR reviewers for their many helpful comments. Corresponding author. Tel.: ; fax: addresses: avlonitis@aueb.gr (G.J. Avlonitis), npanag@aueb.gr (N.G. Panagopoulos). 1 Tel.: ; fax: empirical studies are available that directly address this important issue (e.g., Chonko, 1982). More analytically, three major turning points collectively underscore the importance of examining the consequences of sales management practices for industrial salespersons. First, in their groundbreaking model of motivation and performance, Walker et al. (1977, 1979) propose that organizational factors exert influence on the role stress levels as well as on the performance and attitudes of salespeople. Second, and perhaps somewhat ironically, Churchill et al. (1985) meta-analysis of the determinants of salesperson performance reports that the authors were able to trace only five studies that directly examine the impact of organizational/ environmental variables on performance. Third, Brown and Peterson's (1993) meta-analytic study of job satisfaction antecedents and consequences acknowledges the important effect of organizational variables on the job satisfaction of salespeople. However, one of their main findings was that very few studies had examined the relationship between organizational variables and salesperson outcomes, and these studies have considered primarily supervisory behaviors and job/task characteristics. For this reason, Brown and Peterson did not included organizational variables in their model because of the limited number of study effects available. Unfortunately, the /$ - see front matter 2007 Elsevier Inc. All rights reserved. doi: /j.jbusres

2 766 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) paucity and the need for further research on the area are still apparent (Babakus et al., 1996b; Grant et al., 2001). Unless the effects of sales management practices are determined, sales practitioners may rely on inefficient and ineffective practices. From a methodological standpoint, previous sales management studies suffer from two major drawbacks. First, studies utilizing different data sources to investigate relationships among the constructs of interest are extremely rare. As a result, much of the empirical research in marketing/sales suffers from common method bias, which may lead to the artificial inflation of the studied relationships (Summers, 2001). Second, past studies have overlooked the natural nesting of salespeople within sales units that is neither accidental nor ignorable (Hox, 2002). In most of the studies, researchers draw conclusions by pooling together salespeople from different companies into a single sample. However, such an approach overlooks the fact that the context (i.e. sales unit) exerts significant influence on salespeople as well as the potentially meaningful individual variance within each sales unit (Jedidi et al., 1997; Kohli, 1989); thus, this approach erroneously assumes that salespeople comprising each sales unit will feel and behave similarly (Glick and Roberts, 1984). Importantly, salespeople within the same sales unit are very likely to perceive and respond in a differential manner to sales management activities (Brown et al., 1998; Oliver and Anderson, 1994; Shoemaker, 2003; Vink and Verbeke, 1993). In addition, ignoring the multilevel nature of data collected from salespeople nested within sales units, may result in serious violations of the independence assumption (Hox, 2002), which constitutes a prerequisite for employing many of the statistical procedures applied to sales research (e.g., structural equation modeling, ordinary least squares regression). Failure to properly account for the nested structure of data may lead to small standard errors, large test statistics and biased estimated coefficients, as disturbance errors are highly correlated within groups (Hox, 2002). Relatively recently, scholars in the organizational behavior (e.g., Hofmann et al., 2000) and sales management literature (Vink and Verbeke, 1993; Yammarino, 1997) have underlined the importance of conducting multilevel research. This study contributes to sales management research in three important ways. First, by focusing on four important sales management practices (i.e., sales organization, territory design, training, and compensation), this study provides an in-depth investigation of their often overlooked consequences for salespeople's role stress, attitudes and outcomes, thereby offering new insights both to sales practitioners and researchers. Second, by utilizing two different data-sources, this study takes a critical step towards minimizing the common method bias of estimated relationships. Third, the present study introduces a new methodological approach, namely, hierarchical linear modeling (HLM) to sales research. This statistical method is developed elsewhere (Bryk and Raudenbush, 1992) but the method is useful for studying the sales management effects that this article explores. Specifically, the HLM approach appropriately models the multilevel nature of sales organizations, thereby offering more robust results pertaining to the consequences of sales management practices. 2. Background and hypotheses development 2.1. Theoretical rationale Walker et al. (1977, 1979) model identifies a set of organizational variables (e.g., sales management practices) that may influence a salesperson's role stress, attitudes and performance levels. In the present investigation we concentrated on the impact of four sales management practices on the salesperson's role stress, attitudes, and outcomes: (1) sales organization structure, (2) sales territory design, (3) sales training, and (4) sales compensation. The selection of these variables was based on the following compelling reasons: (a) numerous recent studies have underscored that sales organization (e.g., Babakus et al., 1996b; Piercy et al., 1998), sales territory design (e.g., Babakus et al., 1996a; Baldauf et al., 2001; Zoltners and Lorimer, 2000), sales training (e.g., El- Ansary, 1993; Johnston et al., 1989; Morris et al., 1994; Piercy et al., 1998), and sales compensation (e.g., Anderson and Oliver, 1987; Churchill and Pecotich, 1982; Churchill et al., 2000, p. 411) are among the most important managerial activities, (b) empirical research on the area is strikingly limited (Babakus et al., 1996a; Churchill et al., 2000, p. 403), (c) companies spend an overwhelming amount of dollars in training, allocating, and compensating salespeople each year (Leach and Liu, 2003; Zoltners and Lorimer, 2000), and (d) world-class consulting firms have conducted studies that further elevate the importance of these practices for enhancing the effectiveness of the sales organizations (e.g., Forum Corporation, 1998). The present study focuses on the consequences of sales management practices for salespeople's role stress (role ambiguity and role conflict), attitudes (job satisfaction and organizational commitment) and outcomes (job performance, and turnover intentions). Brown and Peterson (1993) elevate the importance of studying these criteria in their meta-analytic study. Following Brown and Peterson, several scholars examine these variables in their research endeavors (e.g., MacKenzie et al., 1998; Singh et al., 1996). Consequently, an explicit examination of these variables could add to the existing body of knowledge, by investigating a series of different antecedent factors, and could allow for a direct comparison of the results of the present study with the results of previous research The influence of horizontal and vertical sales organization structure The design of sales organizations involves two crucial decisions pertaining to horizontal and vertical structure, respectively (Churchill et al., 2000). With respect to the horizontal structure, sales managers have to decide on how they should allocate the selling effort. Sales scholars propose the following basic forms of organizing the sales force: a) geographic organization, b) product organization, c) customer/ market organization, and d) combination of the three aforementioned forms of organization. Paradoxically, the impact of the various forms of horizontal structure on salespeople constitutes

3 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) one of the most neglected issues in the sales management literature. Notwithstanding the shortage of empirical studies, however, conventional logic pinpoints the relative advantage of organizing the sales force by customer/market type (Churchill et al., 2000, p. 108; Leigh and Marshall, 2001). This type of organizing the selling effort resembles both the marketing concept and the strategy of market segmentation, as specialized sales forces are assigned to servicing a particular market segment (Churchill et al., 2000). By focusing on a specific type of customers, salespeople become more knowledgeable of both the customers' needs and the ways to satisfy those needs. Thus, the expectation is that salespeople within companies employing this form of sales organization will experience greater job satisfaction and commitment and will perform better. In addition, in the cases whereby salespeople specialize in specific customer segments the possibility of conflicting expectations is decreasing while the role clarity is increasing. On the basis of the above discussion, we hypothesize the following: H1. Customer/market organization outperforms the other types of sales organization (geographical, product and combination) in terms of achieving greater levels of organizational commitment, job satisfaction and job performance, and lower levels of role ambiguity, role conflict and turnover intentions among salespeople. Sales executives also have the important task of designing the appropriate vertical structure, which primarily involves the determination of the managerial span of control. Surprisingly, however, empirical research concerning its consequences for industrial salespersons is scarce, at best. In fact, researchers often disagree with regard to the most appropriate span of control (Churchill et al., 2000, p. 119). However, Chonko (1982) conducted a study that provides evidence in favor of a small span of control. In particular, his findings show that a smaller span of control is related to lower levels of role ambiguity and role conflict. Chonko concluded that a wider span of control could hinder the effective supervision and, consequently, the communication between the sales manager and the salesperson. As a result, salespersons will receive insufficient information regarding either the performing sales task or management's expectations. Conversely, under close supervision (i.e., narrow span of control) salespeople experience less role conflict as their sales manager provides them with work structure and directions for dealing with the various parties that they are involved with. Furthermore, sales managers that favor a narrower span of control and a more closely supervised sales force will invest more time in developing salespersons' capabilities and in discussing promotion opportunities with them. The sales force may perceive this situation as beneficial that may, in turn, increase the sales force's commitment, job satisfaction and job performance (Churchill et al., 1976; Dubinsky et al., 1994). The discussion leads to the following hypothesis, H2. The smaller the span of control, the higher the salesperson's organizational commitment, job satisfaction and job performance, and the lower the salesperson's role ambiguity, role conflict and turnover intentions The influence of satisfaction with territory design Sales manager's satisfaction with sales territory design, indicating the manager's assessment of the appropriateness of structure (Babakus et al., 1996a), concerns sales scholars with alarming regularity. Indeed, the sales territory decision constitutes one of the most important yet overlooked factors for improving sales force performance (Zoltners and Lorimer, 2000). As a consequence, a limited but continually growing number of studies demonstrates that salespeople working in well-designed sales territories will experience less role stress, they will feel more motivated and satisfied, and they will perform better and tend to remain in the organization (e.g., Babakus et al., 1996a; Churchill et al., 2000,p.173;Grant et al., 2001; Piercy et al., 1998, 2004; Zoltners and Lorimer, 2000). Drawing on the above discussion, we propose the following hypothesis, H3. Sales manager's satisfaction with the sales territory design positively influences salesperson's organizational commitment, job satisfaction and job performance, and negatively influences salesperson's role ambiguity, role conflict and turnover intentions The influence of sales training Sales researchers frequently assert that sales training can increase salesperson's performance and motivation and decrease his/her role stress (Walker et al., 1977). The reasoning for this is that training improves the salesperson's knowledge base and skill level that, in turn, improve performance (Churchill et al., 2000). In spite of the positive benefits attached to sales training, however, empirical evidence in this area remains scarce (Attia et al., 2002; Jantan et al., 2004; Román et al., 2002). Six studies offer empirical support for the positive relationship between training and sales force performance (El-Ansary, 1993; Christiansen et al., 1996; Farrell and Hakstian, 2001; Jantan et al., 2004; Pelham, 2002; Román et al., 2002). Moreover, Ingram et al. (1992) have shown that sales training is a key factor for addressing salesperson's failure. According to the results of the study of Piercy et al. (1998), sales managers rated sales training as one of the most important factors for improving sales force performance. Babakus et al. (1996b) found that training is positively influencing the satisfaction and motivation of salespeople, while Russ et al. (1998) found that training reduces role stress and increases job satisfaction; however, none of the two studies found support for the positive impact of training on performance. Notwithstanding these inconclusive results, however, one can expect that sales training will have beneficial effects for salespersons. Therefore, we propose the following hypothesis, H4. Sales training positively influences salesperson's organizational commitment, job satisfaction and job performance, and negatively influences salesperson's role ambiguity, role conflict and turnover intentions.

4 768 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) The influence of compensation decisions Designing a sales compensation plan involves the following two decisions: (a) the determination of the pay level (i.e., the total remuneration received) and (b) the determination of the compensation mix (i.e., the percentage of salary in the total compensation mix). On a theoretical level, several scholars acknowledge the impact of compensation on salespeople (e.g., Anderson and Oliver, 1987; Bartol, 1999). Surprisingly, our empirical knowledge regarding compensation effects on salespersons remains scarce, as most of the studies have focused on the antecedents of sales compensation plans (e.g., Cespedes, 1990). In one early study, Churchill and Pecotich (1982) reported that pay level is negatively related to salespeople's pay satisfaction; the authors attributed this counterintuitive finding to the inequities salespeople realized in the company surveyed. Chonko et al. (1992) reported a significant preference for pay rises among all other rewards types in a sample of industrial salespeople. However, Cravens et al. (1993) reported a moderate correlation between the percentage of salary compensation and outcome/information performance, while they did not find any empirical support for the relationship between salary and the other behavioral dimensions of performance. Oliver and Anderson (1994) demonstrated that salary is positively related to outcome performance and job satisfaction, but they found no significant relationship with organizational commitment, motivation and behavioral performance. Babakus et al. (1996b) and Piercy et al. (1998) found no support for the relationship between compensation and motivation and performance. Flaherty and Pappas (2002) found that fixed salary leads to higher job satisfaction and lower turnover intentions for salespersons in the exploration stage of their career; however, a greater amount of incentive pay seems to increase job satisfaction and lower turnover intentions for salespeople in the establishment career stage. Finally, Piercy et al. (2004) reported that salary compensation is negatively related to sales force performance in India and Greece, and positively related to behavior performance in Malaysia. In spite of these inconclusive results, however, conventional wisdom suggests that the larger the total compensation received and the larger the proportion of fixed salary in the total compensation, the more positive the consequences for salespeople. Thus, the discussion leads to the following hypotheses, H5. The greater (a) the total compensation and (b) the fixed salary as a percentage of total compensation, the greater the organizational commitment, job satisfaction and job performance, and the lower the role ambiguity, role conflict and turnover intentions of the salesperson. 3. Research method 3.1. Sample and data collection procedure Our research objective was to secure responses from a sufficient number of sales managers and salespersons working in different sales organizations in order (a) to maximize between-sales units-variance, thus allowing the application of multilevel modeling (cf. Hofmann et al., 2000), and (b) to allow for greater generalizability. Consequently, we used a sample of multiple Greek companies operating in three different industrial sectors that reflect the three basic selling situations (cf. Churchill et al., 2000): the ethical-pharmaceutical industry (missionary sales forces), the food & beverage industry (trade sales forces), and the electrical-machinery industry (technical sales forces). From the initial sampling frame of 1894 companies identified in these three sectors, we selected those employing more than 100 employees (442 in total). We contacted all these companies, and 134 agreed to participate in the study (30% response rate). The district sales managers of these companies provided responses regarding the sales management practices in-use, by means of a personal interview with a structured questionnaire. Next, managers distributed via company mail a second anonymous (self-administering) questionnaire to all salespersons working in each of the 134 participating companies (3285 in total). This questionnaire solicited answers regarding salespersons' role stress, attitudes and outcomes. A covering letter accompanying the questionnaire explicitly instructed salespersons to mail back the questionnaire to the research team, by using a prepaid universitystamped envelope. This procedure resulted in 878 useful responses or a 27% response rate, which is comparable to that obtained in similar studies (e.g., Flaherty and Pappas, 2002). Moreover, the average number of responding salespersons per sales unit was 8, while the average salesperson response rate per sales unit was approximately 52%. To evaluate non-response bias in the company sample, we compared the composition of responding companies to that of non-responding companies (as appeared in the rosters of their respective professional associations) by net profits and by number of employees. We found no significant differences (based on t-tests of means), thereby leading us to conclude that the resulting sample is representative. To test for non-response bias in the salesperson sample, we assumed that response nonresponse differences might be manifested to some degree between early and late responses (Armstrong and Overton, 1977). Late responses were those that resulted from our follow-up efforts at every participating company. Specifically, 550 responses were early and 328 were late. By comparing the demographic profile (using chi-square tests) and the mean values of all the remaining study variables (using t-tests) between early and late respondents, we found no significant differences at the 0.05 level (largest chi-square value = 3.279, p =.512; largest t-value = 1.182, p=.238). Based on the results of these tests, we conclude that non-response bias is an unlikely problem Research instruments and measures To minimize the effect of common method bias, we employed a series of procedural remedies in developing the two questionnaires (Podsakoff et al., 2003), besides collecting data from two different data sources. First, we used a multitude of different response formats in developing the two questionnaires. Second, we explicitly instructed respon-

5 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) dents that no right or wrong answer exists. Third, salespersons returned their anonymous questionnaires directly to the researchers. Fourth, through a pretesting process of the questionnaires with 6 sales managers and 17 salespersons, we secured that were not any ambiguous and confusing items. Finally, we placed the constructs inside the questionnaires in a random manner, so that predictor variables were not preceding criterion variables. Regarding the measurement of the constructs of the study, we employed and used, where available, scale items from the relevant literature (see Appendix for a description of the scales). Previous research, however, has shown that individual-level variables such as age, tenure and experience might affect the relationships between organizational practices and salesperson's role stress, attitudes and outcomes (cf. Churchill et al., 1976; Churchill et al., 1979; Flaherty and Pappas, 2002). In addition, one might expect that the industry (i.e., selling situation) in which a salesperson operates will affect the relationships of interest. Thus, we used two dummy variables for missionary selling and trade selling (with observations from the technical selling situation as reference) and three dummies referring to the salesperson's tenure, age, and experience, respectively, as control variables in all analyses (see Table 1 for a description of these variables). 4. Analysis and results 4.1. Measurement validation We subjected the two multiple-item scales measuring sales training and sales territory design (contained in the sales manager's questionnaire) to confirmatory factor analysis (CFA), which yielded the following fit-statistics: χ 2 =207.25, df=132, p b.01; NFI=.91; NNFI=.96; CFI=.96; GFI=.85; SRMR =.064; and RMSEA (90% CI) =.065 (.048 to.082). Next, we performed a second CFA on the scales from the salesperson's questionnaire (i.e., role stress, job satisfaction, organizational commitment, job performance, and turnover intentions). Because of poor construct loadings (maximum likelihood estimate b.30), and strong cross-loads on a different construct than that hypothesized (modification index N 5.0 and standardized expected parameter estimate N.20), we dropped six items and we re-estimated the model. More analytically, we dropped items 4, 7, 9, and 12 from the organizational commitment scale (Mowday et al., 1979) and items 1 (measuring role ambiguity) and item 8 (measuring role conflict) from the role stress scale (Rizzo et al., 1970). The measurement model produced the following fit-statistics: χ 2 = , df=614, pb.01; NFI =.95; NNFI =.96; CFI =.96; GFI=.89; SRMR =.058; RMSEA (90% CI)=.058 (.055 to.060). In both CFAs, the fit statistics fell within the range suggested in the literature, unequivocally suggesting that the hypothesized measurement models constitute a reasonably good representation of the variance covariance matrices of study measures. Moreover, all composite reliabilities exceeded the.70 cut-off point. Overall, this procedure resulted in unidimensional, reliable and valid measures. Table 1 Descriptive statistics for study measures and examination of the between-units variance for individual-level variables Variable Mean Standard deviation 4.2. Hierarchical linear modeling procedure F-value from ANOVA a ICC Role ambiguity Role conflict Organizational commitment Satisfaction with supervisor Satisfaction with the job Satisfaction with company policy-support Satisfaction with promotion policy Pay satisfaction Peer satisfaction Satisfaction with customers Turnover intentions Job performance Training for product knowledge Training for sales skills Satisfaction with territory design Average total annual 19, compensation (Euros) % of fixed salary in total compensation Span of control Geographical organization b Customer/market organization b Product organization b Geographical/customer organization b Geographical/product organization b Customer/product organization b Age c Job tenure d Selling experience e Missionary selling situation f Trade selling situation f a All F-values are significant at the.01 level. This is the ANOVA procedure suggested by Marascuilo and Serlin (1988). b Dummy-coded variable, where 1 = employed and 0 = not employed. c Age consists of five categories that range from b30 years old to N60 years old. d Job tenure consists of six categories that range from b1 year to N21 years. e Selling experience consists of six categories that range from b1 year to N21 years. f Dummy-coded variable, where 1 = yes and 0 = no. We estimated the effects of sales management practices on the salesperson's role stress, attitudes and job outcomes by employing a two-level (main-effects) hierarchical linear model (cf. Hofmann et al., 2000) using HLM 5.04 software. For all subsequent analyses, we included only 90 sales units that returned at least 5 salesperson's questionnaires (781 salespersons, in total). Because hierarchical linear modeling assumes

6 770 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) sufficient between-groups variation on individual-level variables, we performed a one-way analysis of variance with group membership as the independent variable and the constructs measured at the salesperson level as the dependent variables (see Table 1). Because group sizes varied considerably, we employed the ANOVA procedure described by Marascuilo and Serlin (1988). This procedure takes into account the group size and produces an approximate F distribution with adjusted denominator degrees of freedom. Based on the results of the ANOVA procedure, we computed the intraclass correlation coefficients (ICC) for each dependent variable. In all cases, we observed a significant F-value, while the lowest ICC value was 6.20%, thereby indicating a substantial amount of between-units variance in the dependent variables. Importantly, these statistics indicate that individual-level outcomes violate the assumption of independence, thus making necessary the use of HLM. Conceptually, HLM is a simultaneous two-step approach to modelling multilevel relationships. Step 1 involves estimating a separate regression for each group while step 2 models the variance in the individual-level intercepts and slopes using the group-level variable. Consequently, to test for the effects of the set of sales management practices on each dependent variable, we estimated twelve sets of equations. These equations, (intercepts-as-outcomes model) take the following form (see footnote in Table 2 for a description of the abbreviations used): Level 1: Y ij ¼ b 0j þ b 1j ðageþþb 2j ðexpþþb 3j ðjobtenþþr ij Level 2: b 0j ¼ g 00 þ g 01 ðmisþþg 02 ðtradeþþg 03 ðspanþ þ g 04 ðgeoþþg 05 ðcust=markþþg 06 ðprodþ þ g 07 ðgeo=custþþg 08 ðgeo=proþ þ g 09 ðpro=custþþg 010 ðterrþ þ g 011 ðtrain1þþg 012 ðtrain2þ þ g 013 ðsalþþg 014 ðpayþþu 0j b qj ¼ g q0 þ U qj for q ¼ 1; 2; 3 where, Y ij =dependent variable measured at the individual-level for individual i in group j, β 0j =intercept value for group j, γ 00 =intercept value for group level variable, γ 01 = slope term for group level variable, r ij =random individual error, and U 0j = group level residual. Therefore, in order to test the present research's hypotheses, that is the direct effects of sales management practices on salesperson's attitudes and outcomes (Y ij ), we estimated the coefficients γ 03 γ 14 for each regression equation. In addition, a significant estimated parameter γ q0 in the level-2 equations indicates whether the control variables have significant relationships with the dependent variables. Table 2 displays the results of this procedure Hypotheses testing ð1þ ð2þ ð3þ Overall, the study findings support only two of the study's hypotheses. More specifically, our findings reveal that customer/market organization does not outperform the other types of horizontal structure; thus H1 is not supported. In sharp contrast with H2, a larger span of control, rather than a narrower one, positively relates to salespeople's attitudes and outcomes. One may posture, however, that the level of sales manager that we employed for the purposes of the present study (i.e., district sales manager) may have influenced this result. To evaluate the stability of the results, we calculated the field sales managers' span of control by using the following ratio: (Number of field sales managers that each district manager reported to supervising)/(number of salespeople each district manager reported to supervising). Subsequently, we re-ran the HLM analysis, by using this new measure; the results did not indicate any significant differences from those that we obtained by using the district sales manager's span of control. As predicted by H3, satisfaction with the territory design has a significant (and positive) effect on salespersons. The effect of sales training on salespeople is insignificant, thereby not providing support for H4. The total amount of pay that salespeople receive has strong and positive consequences for the salespeople's attitudes and behaviors, in support of H5a. Finally, the empirical analysis did not confirm the positive link between the percentage of straight salary in the total compensation and salesperson's role stress, attitudes and outcomes, thereby not lending support to H5b. With respect to the effect of control variables, we found that both the type of the selling situation, as well as the individuallevel demographic variables, exert significance influence on the salesperson's role stress, attitudes and outcomes. In particular, missionary salespeople experience lower levels of role stress and higher levels of job satisfaction and organizational commitment than their counterparts in the other two industrial sectors. However, they tend to be less satisfied with their customers. A plausible explanation of this finding is that pharmaceutical salespeople call on physicians, who are highly demanding customers, given the ethical nature of their jobs, and who also are not directly involved in buying the drug, thus making the selling process quite complex. With regard to the impact of the demographic characteristics, our findings reveal that older and more experienced salespeople perform better, presumably because they have developed either better knowledge structures or better relationships with their customers. In addition, more experienced salespeople feel more committed to the organization and more satisfied with their customers. Interestingly, as job tenure increases, so does the level of role conflict that salespeople experience, perhaps because salespeople expect to have greater operational freedom as a result of their seniority within the company. Finally, salespeople that have been working for a long time in the same company, feel less satisfied with the promotion opportunities that they are experiencing. 5. Discussion and implications Our study provided new insights into the often-overlooked effects of sales management practices on salesperson's role stress, attitudes and outcomes. Unlike most previous studies, we relied on data that are rarely used in marketing research, in that

7 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) Table 2 Direct effects of sales management practices: hierarchical linear modeling results Independent variables Dependent variables OC RA RC SATSUP SATJOB SATCOMP SATPROM SATPAY SATPEERS SATCUST JP TI Hypothesis testing result GEO H1: not supported CUST/MAR PROD GEO/CUST GEO/PRO PRO/CUST SPAN H2: not supported TERR H3: supported TRAIN H4: not supported TRAIN SAL H5b: not supported PAY H5a: supported AGE EXP JOBTEN MIS TRADE (1) All coefficients (γ 10, γ 20, γ 30, and γ 03 γ 014 ) are unstandardized coefficients estimated by restricted maximum likelihood method. (2) Independent variables are uncentered, while dependent variables are grand mean centered. OC = organizational commitment, RA = role ambiguity, RC = role conflict, SATSUP = satisfaction with supervisor, SATJOB = satisfaction with the job, SATCOMP = satisfaction with the company policy/support, SATPROM = satisfaction with promotion, SATPAY = satisfaction with the pay, SATPEERS = satisfaction with peers, SATCUST = satisfaction with customers, JP = job performance, TI = turnover intentions, GEO = geographical organization, CUST/MAR = customer/market organization, PROD = product organization, GEO/CUST = combination of geographical/customer organization, GEO/PRO = combination of geographical/ product, PRO/CUST = combination of product/customer organization, SPAN = span of control, AGE = salesperson's age, EXP = salesperson's selling experience, JOBTEN = salesperson's job tenure, MIS = missionary selling situation, TRADE = trade selling situation, TERR = satisfaction with territory design, TRAIN1 = training for selling skills, TRAIN2 = training for product knowledge, SAL = % of straight salary in total compensation, PAY = average total annual compensation. Significant coefficients at p-value=0.01 based on t-test. Significant coefficients at p-value=0.05 based on t-test. Significant coefficients at p-value=0.10 based on t-test.

8 772 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) we used the perceptions of a large number of both sales managers and salespersons to investigate relationships of interest. We also modeled the hierarchical nature of the data by applying a random coefficient model. On the basis of the relevant literature, we postured that the organization of the selling effort by customer/market type will outperform the other types of organization, in terms of decreasing the role stress and the turnover intentions of salespeople, while at the same time, increasing positive feelings and job performance. The empirical evidence of our study show that not only this is not the case but, on the contrary, this type of structuring tends to decrease the salespeople's satisfaction with the company policy/support and promotion opportunities as well as with the customers. The most logical explanation of this theoretically unexpected result is that under such a type of organization, salespeople work and built close relationships with their customers. This may, in turn, impose a burden of being knowledgeable about the firm's total product line, which in many instances can be quite diverse or technically complex (Hite and Johnston, 1998, p. 85) and thus may lead to lower satisfaction. Alternatively, a customer/ market organization necessitates that salespeople work together with their customers to help them solve their buying problems. However, if the selling organization does not provide salespeople with adequate training or with adequate support, salespeople may feel frustrated and dissatisfied, as they are unable to help customers. Interestingly, the type of organization that involves the more negative consequences for salespeople is the geographical organization. This type of structuring leads to high levels of role stress and low levels of organizational commitment and job satisfaction. This finding implies that companies should avoid this specific type of organization, as modern marketplaces require salespeople to build relationships with customers. Indeed, such a structure weeds out the opportunity to focus on specific segments in the market and to specialize on specific product lines and specific selling activities (Churchill et al., 2000, p. 104). Furthermore, product organization as well as its combination with customer/market organization tends to have deleterious effects on salespeople, presumably because the duplication of effort that characterizes this type of organization creates confusion among both buyers and salespeople. The results of the present study suggest that the most effective type of organization is the geographical organization in combination with either customer/market organization or product organization. Presumably, these combinations of organizational forms blend the positive effects of all types, thus leading to positive consequences for salespeople. Therefore, sales managers should avoid organizing sales forces on the basis of a single dimension (i.e., geography or markets) and pursue the adoption of a combination of geography with the other two types. The current study provides additional support to the remarkably limited studies examining the consequences of span of control. Contrary to the findings of previous studies (Chonko, 1982), a narrower span of control is not straightforwardly leading to positive work outcomes. Specifically, as the number of salespeople that a sales manager supervises increases, salespeople feel more committed to the company, more satisfied with their supervisor as well as with their customers, and they tend to perform better. Apparently, salespeople perceive a narrow span of control to affect their autonomy and flexibility in handling their sales tasks. This may, in turn, lead to lower organizational commitment and satisfaction. Our findings reveal that in sales units in which the sales manager is satisfied with the design of the sales territories, salespeople tend to experience lower levels of role ambiguity and role conflict and demonstrate increased levels of satisfaction with the job. This result is in accordance with that of the study of Grant et al. (2001), and clearly suggests the critical role of effectively assigning salespersons in balanced and welldesigned sales territories. Therefore, actions should be taken to effectively allocate salespeople to sales territories, by matching each salesperson's unique characteristics, skills and abilities to the territory that s/he can serve best. A rather surprising result of the present investigation involves the insignificant impact of both training for product knowledge and training for sales skills on salesperson's attitudes and job outcomes. In fact, the only significant (positive) relationship is between the provision of training for selling skills and satisfaction with company policy/support. Presumably, salespeople working for companies that systematically provide training for developing sales skills feel that the company views the sales force as an important capital by investing in salespeople's capabilities. However, we did not find support for the rest of the relationships between sales training and the salesperson's role stress and outcomes. A possible explanation for the absence of association between sales training and salesperson's attitudes and outcomes may be that in the present study we measured the extent of providing training programs and not the effectiveness of the sales training. This means that merely providing sales training programs is neither a sufficient nor an adequate condition for gaining performance improvements. Consequently, sales managers should be very cautious when they simply assume that more training is always better. Rather, they should strive to evaluate the quality and the effectiveness of the training programs on the basis of both qualitative and quantitative criteria (Attia et al., 2002). A second explanation may involve the fact that the sample companies were using a variety of training methods, thus making immensely difficult the detection of significant relationships. Importantly, however, the insignificant effect of sales training on salespersons constitutes a critical issue for sales organizations that has to be explained and further investigated. Perhaps the greatest impact among all the managerial practices that we examined in the present study stems from the total remuneration that salespersons receive. Specifically, the study findings reveal that the higher the total remuneration the higher the levels of organizational commitment, satisfaction with company policy/support, pay, promotion, and customers, and the lower the levels of role conflict and turnover intentions. This finding resonates with those of previous studies (e.g., Chonko et al., 1992; Churchill et al., 1979; Ford et al., 1985)

9 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) and signifies the importance of financial compensation for money-motivated individuals such as business-to-business salespeople (Smyth, 1968). Finally, we did not find any direct relationship between the percentage of fixed salary in the total compensation and salespeople's role stress, attitudes, and outcomes. By juxtaposing the two findings that involve the design of compensation systems, one may posture that, regardless of the type of compensation used (i.e., straight salary vis-à-vis bonus/commission), what matters is the total remuneration package that salespeople receive. Consequently, sales managers should strive to increase the total amount paid to salespeople, although this may be restricted by company (i.e., resources, size) as well as by market factors (salespeople availability, competitive practices). investigate the moderating, rather than the direct, effects of the various sales management activities on role stress processes. Seventh, research is needed to unravel the interaction of control systems with span of control and its effects on business-tobusiness salespersons. Finally, future research should also investigate whether softer sales management issues, such as leadership styles and sales control systems, influence the feelings and outcomes of salespeople to a larger extent than objective sales management practices. In conclusion, our contention is that future researchers should focus their attention in conducting more programmatic research, as knowledge regarding the positive as well as the negative consequences of specific sales management practices for salespeople is severely limited by the lack of empirical investigations. 6. Limitations and future research directions Several limitations of our study are noteworthy and can serve as a stepping-stone for future research. Firstly, and most importantly, the present study constitutes an initial exploration; thus, we focused our attention on the direct effects of several sales management practices on the industrial salesperson. However, we did not include some possible intervening and moderating variables, which may affect the hypothesized relationships and might well explain why many of the hypotheses did not receive empirical support. For instance, the salesperson's career stage may have affected the relationship between the span of control and organizational commitment and/or job satisfaction. What could be true for salespersons at early stages in their career could be false for salespeople at later stages (because senior salespersons are probably more positively responsive to empowerment than to close supervision). We strongly believe that future investigators should strive to develop a comprehensive conceptual framework that delineates how sales managers' practices influence their subordinates and that would take into account the possible moderating effects of variables such as career stage and gender. The application of multilevel modeling, as shown in this study, could immensely help researchers to reach this goal. Secondly, while we focused our attention on four important managerial practices, future research should examine the impact of other practices, such as evaluation, recruiting, promotion, and control mechanisms. Thirdly, although we examined a very important aspect of the sales training process (i.e., training content), future investigations should enrich the sales training construct by examining other aspects, such as the frequency of the training sessions, the quality of the trainers etc. Fourthly, although we examined a set of dependent variables of primary importance to sales force research (Brown and Peterson, 1993) future investigations should examine other interesting variables, such as organizational citizenship behaviors, and customer orientation behaviors. A fifth limitation involves the selection of a single geographical context (Greece) as the sampling frame. Thus, further replication is needed to ensure confidence in the stability of the findings and ultimate generalizability, by conducting research in other countries such as the UK and US especially. Sixth, a fruitful opportunity for further investigation would be to Appendix A. Scale items, sources and measurement properties Scale (source) No. of items Scale format Horizontal structure a (new scale) 4 2-point (0 = not Which of the following organizational forms have you employed in your sales department: 1. Geographical organization 2. Customer/market organization 3. Product organization 4. Combination organization Vertical structure (new scale) 1 employed, 1 = employed) Ratio scale What is the total number of salespeople that you are supervising? Sales training programs (new scale) 7 7-point (1 = not at To what extent are the following training programs being offered to your sales force: Training for product knowledge (α=.88, CR=.90) 1. Characteristics of company's products 2. Characteristics of competitive products Training for Sales Skills (α=83., CR=.83) 1. Interpersonal communication skills 2. Customer characteristics/needs 3. Company policies/procedures 4. Sales techniques 5. Technology usage Satisfaction with the territory design (Babakus et al., 1996a; α=.89, CR=.90) b Sales compensation (adapted from Cravens et al., 1993) 1. What is the % of fixed salary component in the total compensation received by your sales force? 2. What is the average total annual compensation of a salesperson in the sales organization? all, 7 = to a large extent) 11 7-point (1 = not at all, 7 = very satisfied) 2 Ratio scale Role stress (Rizzo et al., 1970) b 14 7-point Role ambiguity (α=.74, CR=.74) Role conflict (α=.80, CR=.81) (1 = never happens, 7 = happens all the time) (continued on next page)

10 774 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) Appendix A (continued) Scale (source) No. of items Scale format Organizational commitment (Mowday et al., 1979; α=.88, CR=.90) b 15 5-point (1 = completely disagree, 5 = completely agree) Job satisfaction (Comer et al., 1989) b 29 5-point (1 = completely Satisfaction with supervisor (α=.86, CR=.88) Satisfaction with the job (α=.86, CR=.86) Satisfaction with company policy-support (α=.86, CR=.87) Satisfaction with promotion policy (α=.70, CR=.71) Pay satisfaction (α=.81, CR=.81) Peer satisfaction (α=.78, CR=.80) Satisfaction with customers (α=.85, CR=.87) Job performance (adapted from Sohi, 1996; α=.78, CR=.79) How well have you performed during the last 12 months relative to sales quotas concerning 1. Sales volume 2. Profitability 3. New account development 4. Servicing existing accounts Turnover intentions (Bluedorn, 1982; α=.92, CR=.92) What is the possibility that you will stay in the sales organization within the next 3, 6, and 12 months c disagree, 5 = completely agree) 4 5-point (1 = much worse, 5 = much better) 3 5-point (1 = very unlikely, 5 = very likely) a Respondents were instructed to provide multiple-responses in the case they were employing more than one organizational form. b These constructs were operationalized exactly as in their original scales, and thus we do not report on the specific items. c Items in this scale were reversed scored. References Anderson E, Oliver RL. Perspectives on behavior-based versus outcome-based salesforce control systems. J Mark 1987;51:76 88 [October]. Armstrong SJ, Overton TS. Estimating nonresponse bias in mail surveys. J Mark Res 1977;14(3): Attia A, Honeycutt Jr ED, Attia MM. The difficulties of evaluating sales training. Ind Mark Manage 2002;31(3): Babakus E, Cravens DW, Grant K, Ingram TN, LaForge R. Investigating the relationships among sales management control, sales territory design, salesperson performance, and sales organization effectiveness. Int J Res Mark 1996a;13(4): Babakus E, Cravens DW, Johnston M, Moncrief WC. Examining the role of organizational variables in the salesperson job satisfaction model. J Pers Sell Sales Manage 1996b;16:33 46 [Summer]. Baldauf A, Cravens DW, Piercy NF. Examining business strategy, sales management, and salesperson antecedents of sales organization effectiveness. J Pers Sell Sales Manage 2001;21: [Spring]. Bartol KM. Reframing salesforce compensation systems: an agency theorybased performance management perspective. J Pers Sell Sales Manage 1999;19:1 16 [Summer]. Bluedorn AC. A unified model of turnover from organizations. Human Relat 1982;35(2): Brown SP, Peterson RA. Antecedents and consequences of salesperson job satisfaction: a meta-analysis and assessment of causal effects. J Mark Res 1993;30:63 77 [February]. Brown SP, Cron WL, Slocum Jr JW. Effects of trait competitiveness and perceived intraorganizational competition on salesperson goal setting and performance. J Mark 1998;62:88 98 [October]. Bryk AS, Raudenbush SW. Hierarchical linear models. Newbury Park, CA: Sage Publications; Cespedes FV. A preface to payment: designing a sales compensation plan. Sloan Manage Rev 1990;32(1): Chonko LB. The relationship of span of control to sales representatives' experienced role conflict and role ambiguity. Acad Manage J 1982;25 (2): Chonko LB, Tanner Jr JF, Weeks WA. Selling and sales management in action: reward preferences of salespeople. J Pers Sell Sales Manage 1992;12:67 75 [Summer]. Christiansen T, Evans KR, Schlacter JL, Wolfe WG. Training differences between services and good firms: impact on performance, satisfaction and commitment. J Prof Serv Mark 1996;15(1): Churchill Jr GA, Pecotich A. A structural equation investigation of the pay satisfaction-valence relationship among salespeople. J Mark 1982; 46: [Fall]. Churchill Jr GA, Ford NM, Walker Jr OC. Organizational climate and job satisfaction in the salesforce. J Mark Res 1976;13: [November]. Churchill Jr GA, Ford NM, Walker Jr OC. Personal characteristics and the attractiveness of alternative rewards. J Bus Res 1979;7:25 49 [April]. Churchill Jr GA, Ford NM, Hartley SW, Walker Jr OC. The determinants of salesperson performance: a meta-analysis. J Mark Res 1985;22: [May]. Churchill Jr GA, Ford NM, Walker Jr OC, Johnston MW, Tanner Jr JF. Sales force management. 6th ed. Illinois: Irwin Press; Comer JM, Machleit KA, Lagace RR. Psychometric assessment of a reduced version of INDSALES. J Bus Res 1989;18(4): Cravens DW, Ingram TN, LaForge RW, Young CE. Behavior-based and outcome-based control systems. J Mark 1993;57(4): Dubinsky AJ, Yammarino F, Jolson M. Closeness of supervision and salesperson work outcomes: an alternate perspective. J Bus Res 1994;29(3): El-Ansary AI. Sales force effectiveness research reveals new insights and reward-penalty patterns in sales force training. J Pers Sell Sales Manage 1993;13(2): Farrell S, Hakstian AR. Improving salesforce performance: a meta-analytic investigation of the effectiveness and utility of personnel selection procedures and training interventions. Psychol Mark 2001;18(3): Flaherty KE, Pappas JM. The influence of career stage on job attitudes: toward a contingency perspective. J Pers Sell Sales Manage 2002;22(3): Ford NM, Churchill Jr GA, Walker Jr OC. Differences in the attractiveness of alternative rewards among industrial salespeople: additional evidence. J Bus Res 1985;13(2): Forum Corporation. Sales effectiveness in world-class organizations: identifying sales force and sales management capabilities. Mark Res Rep Glick WH, Roberts KH. Hypothesized interdependence, assumed independence. Acad Manage Rev 1984;9(4): Grant K, Cravens DW, Low GS, Moncrief WC. The role of satisfaction with territory design on the motivation, attitudes and work outcomes of salespeople. J Acad Mark Sci 2001;29(2): Hite RE, Johnston WJ. Managing salespeople: a relationship approach. Cincinnati, OH: South-Western College Publishing; Hofmann DA, Griffin MA, Gavin MB. The application of hierarchical linear modeling to organizational research. In: Klein KJ, Kozlowski SW, editors. Multilevel theory, research, and methods in organizations. San Francisco: Jossey-Bass; p Hox J. Multilevel analysis. Mahwah, NJ: Lawrence Erlbaum; Ingram TN, Schwepker CH, Hutson D. Why salespeople fail. Ind Mark Manage 1992;21(3): Jantan AM, Honeycutt ED, Thelen ST, Attia AM. Managerial perceptions of sales training and performance. Ind Mark Manage 2004;33(7): Jedidi K, Jagpal HS, DeSarbo WS. Finite-mixture structural equation models for response-based segmentation and unobserved heterogeneity. Mark Sci 1997;16(1): Johnston MW, Hair Jr JF, Boles J. Why do salespeople fail? J Pers Sell Sales Manage 1989;9:53 8 [Fall].

11 G.J. Avlonitis, N.G. Panagopoulos / Journal of Business Research 60 (2007) Kohli AK. Effects of supervisory behavior: the role of individual differences among salespeople. J Mark 1989;53:40 50 [October]. Leach MP, Liu AH. Investigating interrelationships among sales training evaluation models. J Pers Sell Sales Manage 2003;23: [Fall]. Leigh TW, Marshall GW. Research priorities in sales strategy and performance. J Pers Sell Sales Manage 2001;21:89 93 [Spring]. MacKenzie SB, Podsakoff PM, Ahearne M. Some possible antecedents and consequences of in-role and extra-role salesperson performance. J Mark 1998;62:87 98 [July]. Marascuilo LA, Serlin RC. Statistical methods for the social and behavioral sciences. New York: Freeman; Morris MH, LaForge RW, Allen JA. Salesperson failure: definition, determinants, and outcomes. J Pers Sell Sales Manage 1994;14:1 15 [Winter]. Mowday RT, Steers RM, Porter LW. The measurement of organizational commitment. J Vocat Behav 1979;14: [April]. Oliver LR, Anderson E. An empirical test of the consequences of behavior- and outcome-based sales control systems. J Mark 1994;58:53 67 [October]. Pelham AM. An exploratory model and initial test of the influence of firm-level consulting oriented sales force programs on sales force performance. J Pers Sell Sales Manage 2002;22(2): Piercy NF, Cravens DW, Morgan NA. Salesforce performance and behaviorbased management processes in business-to-business sales organizations. Eur J Mark 1998;32(1/2): Piercy NF, Low GS, Cravens DW. Examining the effectiveness of sales management control practices in developing countries. J World Bus 2004;39 (3): Podsakoff PM, MacKenzie SB, Lee JY, Podsakoff NP. Common method biases in behavioral research: a critical review of the literature and recommended remedies. J Appl Psychol 2003;88: [October]. Rizzo JR, House RJ, Lirtzman SI. Role conflict and ambiguity in complex organizations. Adm Sci Q 1970;15(2): Román S, Ruiz J, Munuera S. The effects of sales training on sales force activity. Eur J Mark 2002;36(11/12): Russ F, McNeilly KK, Comer J, Light T. Exploring the impact of critical sales events. J Pers Sell Sales Manage 1998;18(2): Shoemaker ME. Leadership behaviors in sales managers: a levels analysis. J Mark Theory Pract 2003;11(2): Singh J, Verbeke W, Rhoads GK. Do organizational practices matter in role stress processes? A study of direct and moderating effects for marketingoriented boundary spanners. J Mark 1996;60:69 86 [July]. Smyth RC. Financial incentives for salesmen. Harvard Bus Rev 1968;46 (1): Sohi RS. The effects of environmental dynamism and heterogeneity on salespeople's role perceptions, performance and job satisfaction. Eur J Mark 1996;30(7): Summers JO. Guidelines for conducting research and publishing in marketing: from conceptualization through the review process. J Acad Mark Sci 2001;29 (4): Vink J, Verbeke W. Adaptive selling and organizational characteristics: suggestions for future research. J Pers Sell Sales Manage 1993;13:15 24 [Winter]. Walker OC, Churchill Jr GA, Ford NM. Motivation and performance in industrial selling: present knowledge and needed research. J Mark Res 1977;14(2): Walker OC, Churchill Jr GA, Ford NM. Where do we go from here? Selected conceptual and empirical issues concerning the motivation and performance of the industrial sales force. In: Albaum G, Churchill GA, editors. Critical issues in sales management: state-of-the-art and future research needs. Eugene, OR: University of Oregon Press; Yammarino FJ. Models of leadership for sales management. J Pers Sell Sales Manage 1997;17:43 56 [Spring]. Zoltners AA, Lorimer SE. Sales territory alignment: an overlooked productivity tool. J Pers Sell Sales Manage 2000;20: [Summer].