Logistics Performance, Customer Satisfaction, and Share of Business ...

Journal of Business Logistics, 2012, 33(3): 210–226 © Council of Supply Chain Management Professionals

Logistics Performance, Customer Satisfaction, and Share of Business: A Comparison of Primary and Secondary Suppliers Rudolf Leuschner1, Douglas M. Lambert2, and A. Michael Knemeyer2 1 2

Rutgers University The Ohio State University

he linkage between logistics performance and overall firm performance has received attention in the literature for more than 30 years. However, researchers have not investigated if differences in performance between primary and secondary suppliers affect customer satisfaction and the percentage of business allocated to suppliers. In this research, primary suppliers received more than four times as much business as the secondary suppliers. We investigated the impact of the Marketing Mix on customer satisfaction and share of business for primary suppliers and secondary suppliers, and identified differences between the two groups, using multigroup structural equation modeling. The results indicate that perceived performance on logistics attributes significantly affects customer satisfaction and the percentage of business that is allocated to primary and secondary suppliers, which is key information for developing competitive strategies. Our research findings challenge the practice of providing service levels to customers based on current revenue or profitability which does little to convince customers, who are using a company as secondary source, to make it the primary supplier.

T

Keywords: logistics customer service; Marketing Mix; customer satisfaction; share of business; primary versus secondary suppliers

INTRODUCTION There is evidence that strong performance on logistics attributes improves business outcomes (Daugherty et al. 1998; Stank et al. 1999, 2003; Davis-Sramek et al. 2008). In addition, in some studies, the effect of logistics service on share of business was greater than other factors commonly ascribed to the marketing function (Sterling and Lambert 1987; Emerson and Grimm 1998). Similar to share of wallet, share of business is the percentage of sales that is assigned to a specific supplier (Mägi 2003). In many industries, business is distributed among several suppliers, and multiple sourcing is often the preferred arrangement (Williamson 1991; Richardson and Roumasset 1995). But, the issue of how primary and secondary suppliers are perceived and how this information can be used by suppliers to differentiate themselves with logistics has not been investigated. Customers are more dependent on primary suppliers, select them more carefully (Provan and Skinner 1989), and there is some evidence that they perform better, but the link between better performance and share of business has not been investigated (Gassenheimer et al. 1995). Over 80% of the respondents in our sample used multiple suppliers, which is consistent with other industries (Gassenheimer et al. 1995). These suppliers are competing for a share of the customers’ purchases (Jackson 1985). In the research reported in this article, primary suppliers received more than four times the amount of business as secondary suppliers and therefore understanding how to become the primary supplier or maintain the primary supplier status is critical information for management. With primary suppliers, the growth may be limited to the growth of the customer, but if they are large profitable customers, it is important to retain them and growth may be substantial. With secondary suppliers, the opportunity is twofold, by taking share

Corresponding author: Rudolf Leuschner, Rutgers University, 1 Washington Park, Newark, NJ 07065, USA; E-mail: [email protected]

from other suppliers and by growing with the customer. To focus a supplier’s scarce resources, management must understand these differences. Our first goal for this research was to determine if the impact of the Marketing Mix on customer satisfaction and share of business varies between primary suppliers and secondary suppliers. Potential differences have not been explored in this context, which represents a significant gap in the literature. If researchers do not identify how primary and secondary suppliers are perceived, it may lead to incorrect conclusions about how resources should be allocated. For example, if all customers of a supplier are considered as a sample (Emerson and Grimm 1998; DavisSramek et al. 2008), a significant relationship may be found between performance and satisfaction (and loyalty), but this may be caused by undetected differences between primary and secondary suppliers, and the researchers and management would not know this. The second goal was to compare the influence of logistics and marketing attributes in achieving customer satisfaction and share of business to explore the potential value-added role of logistics. An industry’s competitive forces influence what factors contribute to customer satisfaction and the decision to award more or less business to a supplier, and include: (1) threat of new competition, (2) threat of substitute products or services, (3) bargaining power of customers, (4) bargaining power of suppliers, and (5) intensity of competitive rivalry (Porter 1979). These five forces are different from one industry to another, and it is not safe to assume that they stay the same over time. Consequently, it is reasonable to expect that research results will vary from industry to industry and over time. Replication of research studies is necessary to determine if the results obtained in one industry can be generalized to others (Hunter 2001). Our research is consistent with calls for more replication (Hunter 2001; Goldsby and Autry 2011). Establishing the impact of logistics performance on business outcomes provides evidence of the revenue-generating capabilities of logistics, which has been portrayed as “our discipline’s equivalent to finding a cure for cancer” (Bowersox 1999).

Logistics Performance, Customer Satisfaction, and Share of Business

211

In the remainder of this article, we first develop our hypotheses. Second, we describe the research methodology, including sample selection, parceling of scale items, and scale development. Data from two surveys were used for this research, one to develop the scales and the other to identify the differences between primary and secondary suppliers via multigroup structural equation modeling (SEM). Third, the results of the analysis are presented. Finally, we provide our conclusions, including limitations, research implications, opportunities for future research, and managerial implications.

ter products than the competition is a source of competitive advantage (Ittner and Larcker 1997). The technical service provided for the product is another aspect that contributes to customer satisfaction (Cronin and Taylor 1992; Lin et al. 2010). However, given Porter’s (1979) five forces model, it is reasonable to expect that the relative importance of product could vary by industry. For example, in this research, all suppliers that are evaluated by the respondents had products that met a minimum level of quality because the reagents had to be approved by the Food and Drug Administration. Thus, it was hypothesized that:

HYPOTHESIS DEVELOPMENT Traditionally, logistics has been viewed as a cost center in companies with the function’s primary contributions to financial performance being in the areas of cost and asset reductions (Novack et al. 1994, 1996; Andraski and Novack 1996; Tracey 1998). Logistics managers point to service improvements as evidence that they add value to the company; however, convincing upper management is difficult because these improvements are not easily translated into financial performance (Novack et al. 1994, 1996). If superior logistics performance produces desirable business outcomes like increased customer satisfaction and a higher share of a customer’s purchases, the notion that logistics only contributes cost and asset reductions can be challenged. In addition, if superior logistics performance leads to more revenue from satisfied customers, it could change how logistics is perceived within the organization, and could offer opportunities to use logistics performance to achieve competitive differentiation. The link between marketing and logistics elements of customer service has been documented in several studies (Sterling and Lambert 1987, 1989; Lambert and Harrington 1989; Innis and La Londe 1994; Emerson and Grimm 1996, 1998). The “Marketing Mix” (Borden 1964) has been conceptualized as the “Four P’s” of marketing (McCarthy 1960) and they include product, price, promotion, and place. Once the channels of distribution have been established, “place” generally occurs in the logistics function as time and place utilities are created, and for this reason, it is often synonymous with logistics service (Coyle et al. 1992; Stock and Lambert 2001). Innis and La Londe (1994) propose a model of customer service where the Marketing Mix is conceptualized as product, price, promotion, and logistics activities as the place component. This conceptualization was chosen because it allows comparison of the impact of logistics and marketing on firm performance. The product construct contains attributes related to quality, performance, new product development, and the technical support for the product. The attributes were designed to provide a holistic assessment of all important product aspects of reagents. Product quality has been shown to positively impact customer satisfaction (Selnes 1993; Chen and Chuang 2008; Lin et al. 2010). There is evidence that better-performing products lead to increased customer satisfaction (Swan and Jones Combs 1976; Maddox 1981; Johnson et al. 1995). Also, research has shown the link between product development performance and increased customer satisfaction (Yadav and Goel 2008). Because customers take a long-term perspective when evaluating suppliers (Ganesan 1994), the ability of an organization to consistently develop bet-

H1: Performance on product attributes has a positive impact on customer satisfaction. Price is a determinant in supplier selection (Hoyer et al. 2002; Herrmann et al. 2004). The attributes that make up the price construct address billing, competitiveness of price, discounting, and price level. Billing procedures can be a source of customer satisfaction (Meuter et al. 2000; Lee et al. 2001). Competitiveness of pricing has been shown to significantly impact satisfaction (Morganoski 1988; Voss et al. 1998; Hoyer et al. 2002; Herrmann et al. 2004). The price sets expectations about product and service performance and if those expectations are not met, it can have a negative effect on satisfaction (Hoyer et al. 2002; Herrmann et al. 2004). This view is called relative pricing because price is evaluated considering other factors. Although the importance of price may vary by industry, based on previous findings, it was hypothesized that: H2: Performance on price attributes has a positive impact on customer satisfaction. While advertising is an important part of promotion in business-to-consumer relationships, business-to-business relationships are heavily weighted in terms of actions by salespeople (Mudambi 2002). Sales representatives promote the product during their meetings with customers. In addition, sales forces are trained to deliver the branding message of the company (Lynch and Chernatony 2004). Previous research has shown that good salespeople can help increase satisfaction (Grewal and Sharma 1991; Goof et al. 1997; Johnson et al. 2001; Liu and Leach 2001). Salespeople set customers’ expectations regarding the Marketing Mix. If those expectations are not met, negative consequences are directed toward sales people and toward the components of the Marketing Mix that are over-promised. Achieving uniform performance from individual members of a sales force is difficult (Barker 1999). This is further complicated by the relative density of customers in an industry. For example, if there are less than a hundred large customers, it may be possible to assign top sales people to the most profitable customers. However, in industries where business customers might number in the thousands, sales territories will include customers of various size and profitability. Thus, it was hypothesized that: H3: Performance on promotion attributes has a positive impact on customer satisfaction. The three constructs that are controlled by the marketing function (Product, Price, and Promotion) have received attention

212

in the marketing literature, but only a few studies have evaluated constructs from both marketing and logistics (see, e.g., Sterling and Lambert 1987; Lambert and Harrington 1989; Emerson and Grimm 1998). The attributes that make up the place (logistics) category are focused on delivery issues such as problem and complaint handling, responsiveness and delivery flexibility, lead time, and information accuracy. The items are similar to the dimensions of the logistics service quality scale developed by Mentzer et al. (1999). Effective problem and complaint handling is important to prevent customer dissatisfaction (Wagner 1994). The degree to which a supplier can be responsive and flexible to changes in deliveries can affect the value a supplier delivers to their customers (Naim et al. 2010). Lead time has been shown to have a positive effect on customer satisfaction (Mentzer et al. 2001; Rafiq and Jafaar 2007). The quality of information has been part of previous conceptualizations of logistics service and has been found to indirectly impact customer satisfaction (Mentzer et al. 2001). Previous research has shown that logistics service performance has an impact on customer satisfaction (Stank et al. 1999; Mentzer et al. 2001; Davis-Sramek et al. 2008). However, often these studies did not include the other components of the Marketing Mix: product, price, and promotion. When other constructs like price were added to the model, the impact of logistics was inconsistent, and this suggests the need for further research (Stank et al. 2003). Thus, it was hypothesized that: H4: Performance on place (logistics) attributes has a positive impact on customer satisfaction. Share of business, the percentage of a customer’s annual purchases given to a specific supplier, is an indicator of financial success within a business-to-business relationship and is used as an outcome variable in this research. Usually, it is more profitable to increase business with existing customers than to attract new ones (Brown et al. 2005). If the goal of management is to increase sales with existing customers, then share of business is an important measure. If satisfied customers award a larger share of business to a supplier, then there should be a focus on activities that increase customer satisfaction. The link between customer satisfaction and financial outcomes has been addressed in the Service-Profit Chain (Heskett et al. 1994) and the Satisfaction-Profit Chain (Anderson and Mittal 2000). For most managers, satisfaction is an abstract concept, but if it can be linked to tangible business outcomes, then it is more meaningful (Anderson and Mittal 2000). In most studies, a positive link between satisfaction and financial outcomes has been established (Rust et al. 1995; Stank et al. 2003). If customer satisfaction does not influence share of business, significant switching costs may exist (Liu et al. 2005), which are highlighted in the “Always-a-share Model” (Jackson 1985). Most customers use more than one supplier and suppliers compete for a larger share of each customer’s business. Thus, the following hypothesis was proposed: H5: Customer satisfaction has a positive impact on share of business. The five previously described hypotheses were explicitly tested in this research. The path diagram in Figure 1 shows the

R. Leuschner et al.

Figure 1: Conceptual model with hypotheses.

constructs and the hypotheses that were evaluated. In addition, the differences between primary and secondary suppliers were tested using multigroup analysis. Comparing primary and secondary suppliers In past research, the potential differences between primary and secondary suppliers have not been explored. In many cases, the data collected would not have allowed this type of analysis. For example, in some studies, researchers only asked respondents to evaluate their primary supplier (e.g., Stank et al. 1999, 2003; Rafiq and Jafaar 2007). In these studies, it was not possible to investigate the differences between primary and secondary suppliers. In other studies, the research was supported by a manufacturer and the survey’s respondents were only asked to evaluate the sponsoring manufacturer (e.g., Emerson and Grimm 1998; Davis-Sramek et al. 2008). Thus, the results provided an average view of a supplier, as respondents may be using the manufacturer as their primary supplier, their secondary supplier, or a supplier to whom they give a relatively small amount of business. And, it would be reasonable for customers in these three roles to evaluate the supplier quite differently, but it was not measured or considered by the researchers. For example, logisticians are taught to ABC service levels (Stock and Lambert 2001; Fawcett et al. 2007; Coyle et al. 2008; Christopher 2010). If product is in short supply, it is common to give preferential service to key customers. In a few studies, respondents were instructed to evaluate multiple suppliers, but comparisons between primary and secondary suppliers were not reported (e.g., Sterling and Lambert 1987; Lambert and Harrington 1989; Innis and La Londe 1994). In a number of articles, the authors did not report which supplier the respondent was asked to assess (e.g., Daugherty et al. 1998; Ellinger et al. 1999). It is possible that respondents were asked to think about a typical supplier or their primary supplier. The issue of whether customers evaluate primary and secondary suppliers differently remains unexplored. Given the time period when previous research was completed, the researchers may not have had the analytical tools to conduct such analysis. Also, only few researchers had the sample size necessary to conduct a multigroup SEM analysis. Consequently, this research addresses a significant gap in the literature.


213

Being a primary supplier or a secondary supplier has the potential to moderate the impact of the Marketing Mix components directly on customer satisfaction and indirectly on share of business as shown in Figure 1. Thus, the following hypotheses were proposed:

organization was responsible for purchasing reagents before the mail survey was sent. To improve the overall response rate, their cooperation in the research was solicited and a summary report of the findings was provided (Allen et al. 1980; Hornik 1982). The respondents were primarily supervisors, section heads, and chief medical technologists, and the majority stated that they had the authority to select reagent suppliers. For both samples, three separate mailings of the 12-page questionnaire, spaced at threeto four-week intervals, were conducted. In the second mailing, all respondents contacted in the first survey were polled a second time. The final, third mailing was restricted to individuals representing those institutions from which questionnaires had not been received in the first two mailings. These mailings were followed by a two-page version of the questionnaire to a sample of nonrespondents to examine any potential nonresponse bias (Lambert and Harrington 1990). For the coagulation sample, a total of 1,005 hospitals in the United States having on-site coagulation/hematology laboratories were surveyed. A total of 299 responses to the full survey were received as well as 147 to the short version. This represented a response rate of 29.75% (299/1,005). After removing responses with missing data, a total of 435 data points for the coagulation sample were obtained because some respondents evaluated more than one supplier. For the blood banking reagents sample, a total of 2,015 hospitals with on-site blood banks were surveyed. All of the 1,000 largest hospitals in the United States and a random sample of the remaining hospitals (with more than 100 beds) from the American Hospital Association membership directory were surveyed. Hospitals with 100 or more beds were believed to be the smallest-sized hospital to have an on-site blood bank. A total of 753 of the 12-page surveys were returned, representing a response rate of 37.37% (753/2,015). In addition, a sample of 100 nonrespondents was surveyed with a two-page questionnaire and 55 responses were received. In the blood banking sample, primary suppliers are compared with secondary suppliers. For this reason, we eliminated respondents who only evaluated one supplier as well as respondents with significant amounts of missing data. Respondents with two or more suppliers were analyzed. Hospitals with a single blood banking reagent supplier tended to be smaller hospitals in terms of number of beds and annual purchases of blood banking reagents. There were 181 respondents with two suppliers and 307 respondents with three or more suppliers. Overall, 488 respondents evaluated both primary and secondary suppliers resulting in 976 data points. The sample for the coagulation reagents was smaller than the sample for the blood banking reagents because fewer hospitals perform the coagulation tests in-house. The sample characteristics of the blood banking sample are shown in Table 1. There are two common methods of determining nonresponse bias. In one, early and late respondents are compared (Armstrong and Overton 1977) and in the other, a subset of questions is sent to nonrespondents and the responses are compared for the same questions on the full version of the questionnaire (Lambert and Harrington 1990). Early and late respondents were compared and no differences were found. In addition, a two-page version of each survey was sent to nonrespondents to test for nonresponse bias. An analysis of variance was performed on each sample to

H6a: The effect of the Marketing Mix components on customer satisfaction differs between primary and secondary suppliers. H6b: The effect of customer satisfaction on share of business differs between primary and secondary suppliers.

RESEARCH METHODOLOGY The research was conducted in the health care industry where purchasers of coagulation reagents and purchasers of blood banking reagents were surveyed. The coagulation sample (299) was used to build the scales and the blood banking sample (753) was used to validate the scales. Then, we eliminated the responses from the blood banking sample where the hospital used a single supplier (116 smaller hospitals) since we wanted to compare evaluations of primary and secondary suppliers. We took respondents from the remaining hospitals in the blood banking sample and grouped their evaluations of primary and secondary suppliers. For the hypothesis testing, primary and secondary supplier groups from the blood banking sample were analyzed. The data used in this study provide a cross-sectional perspective of all major suppliers in the industry. The research methodology section is divided into three parts. First, the sample selection is described. Next, the parceling of scale items is detailed. Then, the scales are developed and assessed. Sample selection Two samples were used in this research, one focused on the evaluation of suppliers of coagulation reagents and the other on blood banking reagents. Coagulation reagents are used to determine the rate at which a patient’s blood coagulates. Blood banking reagents are used to analyze antigen and antibody reactions. For example, they are used in hospitals to test blood samples before patients are transfused or undergo surgery. The surveys were developed separately; however, the methodology used to develop the surveys was the same. The products were similar and the attributes used in both surveys were the same. There were two parts to the data collection procedure: indepth, personal interviews and a mail survey (Sterling and Lambert 1989). For each questionnaire, interviews were conducted with key decision makers in the sponsoring manufacturer organization and in approximately 25 customer organizations. The questionnaire included all attributes that were mentioned during the interviews because the goal was to gain a comprehensive understanding of the attributes that customers used to select, evaluate, and retain suppliers. The interviews were continued until a saturation point was reached and no new attributes could be identified (Bowen 2008). In the second part, potential respondents identified from industry databases were contacted over the telephone to determine if they or some other individual in their

n/a n/a $2,715.50

test for differences. There was no evidence in either of the tests to indicate that nonresponse bias should be a concern. Parceling of scale items

Notes: A: supplier with the highest share of business; B: supplier with the second highest share of business; C: supplier with the third highest share of business.

n/a $5,030.15 $6,003.06 $17,645.00 $20,911.85 $26,593.56 n/a n/a 7.69% 100.00% 80.61% 75.31% 116 181 307 1 2 3

271 323 417

$17,645.00 $25,942.00 $35,312.12

n/a 19.39% 17.00%

B A C B A

Mean share of business to supplier

Mean annual purchases Mean number of beds Number of respondents Number of suppliers

Table 1: Blood banking sample characteristics

C

R. Leuschner et al.

Mean annual purchases from supplier

214

Parceling or partial disaggregation is a technique in which factors are estimated using composites of items instead of individual items (Garver and Mentzer 1999; Little et al. 2004). Instead of estimating the constructs with the ratings of single questions, the scores of several questions are combined into a composite that is used to estimate the construct. The first use of the parceling technique appeared in the Psychology literature (Cattell 1956) and this technique has been used in other areas such as Education, Psychology, and Marketing (Bandalos and Finney 2001). When the respondents answered the question about an attribute in the questionnaire, they were required to perform two tasks: (1) indicate the importance of that attribute in selecting and evaluating suppliers, and (2) evaluate the performance of each major supplier (up to three) on that attribute. Parcel scores take into account importance and performance ratings, so that more weight is given to performance of the more important items. When the attributes were aggregated, a large number of them loaded significantly onto constructs representing the Marketing Mix. The items were combined based on content (Landis et al. 2000). Parcels were used to reduce the number of predictors for each latent variable. Ideally, three to five manifest variables should be used to estimate a latent variable (Bollen 1989). When several items are combined into parcels, fewer indicators can be used to estimate the construct (Garver and Mentzer 1999; Bandalos 2002). The parcels in this research (see Appendix) are built as importance-weighted averages of the performance scores (Innis and La Londe 1994). Prior research on satisfaction has used this approach (Ajzen and Fishbein 1980; Hanan and Karp 1989). Weighted composites were used as they are preferred to unweighted or equal-weighted composites (Rozeboom 1979; Bollen and Lennox 1991). Scale development and assessment The statistical analysis was performed using SEM (Anderson and Gerbing 1988; Bollen 1989) in AMOS 17 (Arbukle 2008). The multi-item scales were developed on the coagulation reagents sample and validated on the blood banking reagents sample (Churchill 1979; Gerbing and Anderson 1988; Bagozzi et al. 1991). The analysis was performed in three steps. In the first step, the measurement model was established with a confirmatory factor analysis (CFA) on the coagulation data. In the second step, the measurement model was validated with another CFA on the blood banking data. In the third step, hypotheses were tested with the structural model using the blood banking data. A necessary condition for building scales estimating constructs is that the measures must be unidimensional (Gerbing and Anderson 1988). That is, each set of indicators has only one underlying trait or construct in common. Multicollinearity, or high correlations among endogenous variables, can lead to Type II errors (Grewal et al. 2004). “When multicollinearity is between .4 and .5, Type II errors tend to be quite small, except when reliability is weak (.7 or below), R2 is low (.25), and sample size is small (ratio of 3:1)” (Grewal et al.


215

2004, 527). This was not the case in this research and we concluded that multicollinearity was unlikely to be an issue. The results of the first CFA are shown in Table 2. The overall model fit the data well (Hu and Bentler 1999) as evidenced by the fit measures: v2(113) = 290.32, v2/df = 2.57, p < .001; comparative fit index (CFI) = .961; root mean square error of approximation (RMSEA) = .058 and RMSEA confidence interval (CI) (.052; .069). All standardized factor loadings are above the minimum recommended values for scale development (Chin 1998). The scales exhibited acceptable reliability as shown by the internal consistency method as measured by Cronbach’s alpha (Cronbach 1951; Nunnally 1978). Additionally, composite reliability (CR) scores and average variance extracted (AVE) were calculated (Bagozzi and Yi 1988). All constructs passed the recommended thresholds. Next, the scales were validated on the blood banking reagent sample. Another CFA was performed for primary and secondary suppliers individually and combined. For the multigroup SEM analysis, the procedure proposed by Zimmer-Gembeck et al. (2005) was used. It is suggested that invariance be tested at three levels: (1) configural invariance, (2) measurement invariance, and (3) structural invariance (Bollen 1989; Baumgartner and Steenkamp 1998; Cheung and Rensvold 2002; Byrne 2010). Configural invariance is a test for similar pattern factor loadings. That is, the factor loadings estimated for each group should not differ significantly from each other. A CFA is applied to the two groups independently and then by restricting the loadings to be equal across the groups (Bollen 1989; Baumgartner and Steenkamp 1998). Measurement invariance is a test for equal measurement models across groups (Bollen 1989; Byrne 2010). It is measured by comparing the unrestricted model, where estimates are produced for each group, to a nested model where the paths to one construct are restricted to be equal across both groups. Structural invariance is a test for equal structural paths, where a structural model is estimated and each structural path is restricted to be equal across both groups. To test for invariance, the difference in model fit was evaluated on several indices. The most common method to test for invariance is to use the difference in v2 values (Δv2) between the unconstrained model and the model with parameters restricted to being equal between groups (Koufteros and Marcoulides 2006; Germain et al. 2008). If the fit of the model does not deteriorate significantly between the unconstrained and the restricted model, then the groups are equivalent (Byrne 2010). In addition to the Δv2, which is known to be sensitive to sample size (Cheung and Rensvold 2002), we used fit indices like the CFI (Bentler 1990) and RMSEA (Steiger and Lind 1980) that are more robust to thoroughly assess invariance across the two groups. First, the configural invariance test was performed and the unconstrained two-group CFA showed good fit: v2(242) = 952.13, v2/df = 3.93, p < .001; CFI = .950; RMSEA = .049 and RMSEA CI (.045; .061). The factor loadings were all significant (see Table 3). Each group produced factor loadings and correlations that were significant and showed only minimal differences between the groups. Therefore, the structure of the model supported configural invariance. The constructs exhibited good reliability, as evidenced by highly reliable measures (Cronbach 1951; Nunnally 1978; Bagozzi and Yi 1988). In addition, all fac-

Table 2: Measurement model results for the coagulation sample Standardized loading

t-Value

Product (a = .82, CR = .80, AVE = .51) Prd1 .83 –* Prd2 .66 14.78 Prd3 .55 11.66 Prd4 .78 18.32 Price (a = .84, CR = .85, AVE = .58) Pri1 .67 –* Pri2 .77 13.36 Pri3 .77 13.36 Pri4 .83 13.00 Promotion (a = .87, CR = .85, AVE = .65) Prm1 .85 –* Prm2 .74 15.79 Prm3 .82 18.18 Place (Logistics) (a = .88, CR = .87, AVE = .62) Pla1 .68 –* Pla2 .85 20.62 Pla3 .78 18.52 Pla4 .83 15.10 Satisfaction (r = .49, CR = .70, AVE = .55) Sat1 .87 –* Sat2 .58 9.36

Significance –*