Marketing Intelligence & Planning

Emerald Article: Customer orientation: construction and validation of the CUSTOR scale Mahmood M. Hajjat

Article information:
To cite this document: Mahmood M. Hajjat, (2002),"Customer orientation: construction and validation of the CUSTOR scale", Marketing Intelligence & Planning, Vol. 20 Iss: 7 pp. 428 - 441 Permanent link to this document: http://dx.doi.org/10.1108/02634500210450864 Downloaded on: 12-06-2012 References: This document contains references to 32 other documents Citations: This document has been cited by 3 other documents To copy this document: [email protected] This document has been downloaded 2050 times since 2005. *

Users who downloaded this Article also downloaded: *

Khalid M. Dubas, Ian Brennan, (2002),"Marketing implications of Webcasting and extranets", Marketing Intelligence & Planning, Vol. 20 Iss: 4 pp. 223 - 228 http://dx.doi.org/10.1108/02634500210431612 Stewart Adam, Rajendra Mulye, Kenneth R. Deans, Dayananda Palihawadana, (2002),"E-marketing in perspective: a three country comparison of business use of the Internet", Marketing Intelligence & Planning, Vol. 20 Iss: 4 pp. 243 - 251 http://dx.doi.org/10.1108/02634500210431649 Qing Wang, Elisa Montaguti, (2002),"The R&D-marketing interface and new product entry strategy", Marketing Intelligence & Planning, Vol. 20 Iss: 2 pp. 82 - 85 http://dx.doi.org/10.1108/02634500210418491

Access to this document was granted through an Emerald subscription provided by UNIVERSITAS GADJAH MADA For Authors: If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service. Information about how to choose which publication to write for and submission guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information. About Emerald www.emeraldinsight.com With over forty years' experience, Emerald Group Publishing is a leading independent publisher of global research with impact in business, society, public policy and education. In total, Emerald publishes over 275 journals and more than 130 book series, as well as an extensive range of online products and services. Emerald is both COUNTER 3 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.

Customer orientation: construction and validation of the CUSTOR scale

Mahmood M. Hajjat Sultan Qaboos University, Muscat, Oman

Keywords

Customer orientation, Factor analysis, Customer satisfaction, Middle East

Introduction
The notion that creating and retaining a satisfied customer should be the primary purpose of business is central to strategic marketing thinking (Drucker, 1954; Deshpande et al., 1993). Drucker (1954) asserts that:
There is only one valid definition of business purpose: to create a customer F F F It is the customer who determines what the business is.

Abstract

The purpose of the present research was to develop and validate a multidimensional scale, CUSTOR, for the measurement of the construct of customer orientation. A pilot study generated 115 statements about what a customer-oriented company should do. The statements were refined, categorized, and used in a questionnaire to collect exploratory data to analyze and confirm the statements and their purported categories. In several confirmatory factor analyses, 17 items and four dimensions exhibited high reliabilities and were retained. Customer orientation is hypothesized to be a multidimensional construct composed of: customer intimacy, customer welfare, company transparency, and continuous improvement. Several other studies collected data for testing the latent structure of customer orientation and the reliability and construct validity of its individual components. The hypothesized relationships were supported and all hypotheses were accepted. Findings suggest that organizational performance can be improved, not only by focusing on customer orientation but rather by examining and emphasizing some of its elements. Research can investigate whether customer orientation encourages innovative, entrepreneur, or learning orientation.

Marketing Intelligence & Planning 20/7 [2002] 428441 # MCB UP Limited [ISSN 0263-4503] [DOI 10.1108/02634500210450864]

An increasing number of studies have recently focused on the concept of market orientation with the aim of understanding the impact of some of its components such as customer orientation on organizational performance. Some studies were interested in a direct causal relationship between market orientation and organizational performance, whereas others looked at the impact of intervening variables on the purported relationship. Research has indicated that market orientation has positive effects on profitability (Ruekert, 1992; Slater and Narver, 1994), employees' attitudes (Jaworsky and Kohli, 1993), sales force orientation (Siguaw et al., 1994), channel relations (Siguaw et al., 1998; Baker et al., 1999), and innovation (Han et al., 1998). In addition to customer orientation, a balanced view of market orientation identifies the components of competitor orientation, inter-functional coordination, and long-term profit focus (Kohli et al., 1993; Narver and Slater, 1990). Consequently, studies of the effect of market orientation on organizational performance measured the combined, but not the independent, effects of all these variables. However, some researchers argue that profit is a consequence of customer orientation (Bell and Emory, 1971) and some others assert that the best way to beat the competition is through customer orientation rather than competitor preoccupation (Perreault and McCarthy, 1997). Therefore, customer
The research register for this journal is available at http://www.emeraldinsight.com/researchregisters

orientation should take precedence over all other components of market orientation. Despite the great emphasis on customer orientation, extant research has not reformulated this concept to suit the study of marketing. Academics and practitioners may have little guidance as to how to actually examine and measure customer orientation as an important aspect of organizational behavior from a marketing perspective. Managers may actually be operating with a strong focus on competitors or products while thinking that they are successfully adopting a customer orientation approach (Cahill et al., 1994; Webster, 1988). The purpose of the present research, therefore, is to reformulate the construct of customer orientation by identifying its latent structure and to develop and validate a psychometrically rigorous scale for its measurement.

Literature review
Conceptualizations of customer orientation evolved from definitions of both marketing concept and market orientation. McKitterick (1957) eloquently pronounced the first widely known statement of customer orientation. From his own observations about success and failure in the marketplace, McKitterick concluded:

So the principal task of the marketing function in a management concept is F F F F F F . to be skillful in conceiving and then making the business do what suits the interests of the customer.

Later, King (1965) proposed in his definition of the marketing concept that the purpose of the total corporate effort should be helping consumers solve selected problems. Consistent with these early conceptualizations, marketing researchers considered customer orientation to be the focal element of the marketing concept and market orientation (Deshpande and Farley,
The current issue and full text archive of this journal is available at http://www.emeraldinsight.com/0263-4503.htm

[ 428 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

1996; McNamara, 1972; Narver and Slater, 1990; Webster, 1988). Deshpande and Farley (1996, p. 13), for example, have tailored their definition of market orientation to emphasize customer orientation. They define market orientation as:
The set of cross-functional processes and activities directed at creating and satisfying customers through continuous needsassessment.

Albeit the increased emphasis on customer orientation, little theoretical and research attention has been devoted to either its domain specification or its measurement (Dreher, 1993; Siguaw et al., 1998). Deshpande et al. (1993), for example, developed a measure that was initially called ``customer orientation'' but later was relabeled as ``market orientation'' (Deshpande and Farley, 1996). This scale was later used to measure ``perceived market orientation'' as it was conceived to be composed of, in addition to customer orientation, competitor orientation and interestingly product orientation as indicated by the items in the scale (Baker et al., 1999). Jaworski and Kohli (1993) had also developed a market orientation scale, ``MARKOR'', in which, presumably, customer orientation as an integral component of market orientation could be assessed by a number of statements (Kohli et al., 1993). In a study that utilized this scale, however, the majority of customer-related statements were dropped from the analysis for lack of significance (Siguaw et al., 1998). The apparent emphasis on the combined rather than the individual effects of the components of market orientation can be attributed to an underlying assumption that all these components contribute equally in constituting the construct of market orientation. However, many researchers and practitioners stress customer orientation in their works as being the most important component (Hall, 1999; Deshpande et al., 1993; Narver and Slater, 1990). Narver and Slater (1990) suggest that customer orientation may play the biggest role in market orientation, and hence suggest that:
Future studies should examine the impact of individual components [on corporate performance] in a given magnitude of market orientation.

underlying structure and relationships to other psychological constructs. Moreover, practitioners need to understand how customer orientation gives rise to superior customer value, and hence contributes to corporate performance. Otherwise, market orientation in general, and customer orientation in particular will remain just articles of faith (Day, 1994) and their parent construct, the marketing concept, will remain just a concept (Cahill et al., 1994; Webster, 1988).

Scale construction
Measurement scales that are currently available assess market orientation and are not directly relevant to the measurement of customer orientation (e.g. Deshpande et al., 1993; Jaworski and Kohli, 1993). The present research describes the process of developing and validating an instrument, called the CUSTOR scale, to assess organizations' customer orientation. The scale is intended to capture the ``general disposition of organizations to undertake in some consistent fashion a set of processes and activities that are aimed at acquiring and retaining customers through continuous needs-satisfaction rather than the actual activities and processes that are being planned and implemented at any single point in time''.

Pilot study
The purpose of the pilot study was to gain insights from managers' thoughts about customer orientation. In total 95 men and women managers from a Middle East country expressed their views of what a customeroriented company should do. Content analysis of managers' responses, along with insights from pertinent literature led to the generation of 115 statements. These statements were found to fall into one of two categories: customer-related and companyrelated statements. Hence, in addition to verifying related statements, a preliminary hypothesis that customer orientation is a higher order entity composed of two latent factors will be tested. From different universities, five business professors were requested to assign each of the 115 statements to one of two conceptual categories (i.e. customer factors and company factors). An a priori decision rule specified that a statement should not be retained unless at least four judges assign it to the same factor. Items were then carefully edited. Redundant items were removed,

In the same vein, Deshpande et al. (1993) suggest that the:

F F F fundamental question is whether customer orientation has a measurable impact on corporate performance.

From a theoretical standpoint, however, the construct of customer orientation remains a hypothetical construct with no empirical value if researchers do not examine its

[ 429 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

double-barreled items were split into single-idea items, and if this was not possible the item was eliminated entirely. Items laden with obvious ``socially acceptable'' responses were refined. As a result of rigorous editing, only 41 statements were retained. All of these statements have the desirable properties of being reworded in the common vernacular of ordinary managers and recast into positive and negative statements, in order to reduce positive and negative response tendencies.

improvement in item loadings in the fourfactor over the two-factor analyses for all items. Loading improvement ranges between 0.09 for the second item to 0.27 for item number 9 for an average of 0.16 (see Table I for item description).

Confirmatory factor analysis

The discrepancy between the preliminary hypothesis (i.e. two-factor model) and the results of the exploratory factor analysis (i.e. four-factor model) requires that the analysis proceeds in an exploratory rather than a confirmatory fashion. The purpose of this analysis at this stage is to examine the theoretical pertinence of the two- and fourfactor models. Because different factor extraction and factor rotation procedures in exploratory factor analysis inevitably yield different pictures about the underlying structure of the 17-item scale, it is very difficult to conclude that a particular hypothesis is validated. Therefore, structural equation modeling is used at this stage for parameter estimation and for model testing. Data were subjected to two independent confirmatory factor analyses using LISREL VII (Joreskog and Sorbom, 1989). In each analysis, the hypothetical model (i.e. two- or four-factor) was hypothesized to fit the data better than other models. For comparative purposes, the results of the two analyses are juxtaposed in Figure 2. The right side of Figure 2 shows parameter estimates (i.e. factor loadings and inter-factor correlations) from the four-factor analysis and the left side exhibits parameter estimates from the two-factor analysis. Unique variances from both analyses are shown and compared together at the bottom of Figure 2. Figure 2 shows that parameter estimates from the four-factor model are much better than their counterparts from the two-factor model. Although the improvement in factor loadings over variables X1 through X5 and X14 through X17 is small, the improvement in factor loadings over variables X6 through X13 is large (between 0.38 and 0.51) for an overall average improvement of 0.22. Unique variances are much smaller in the four-factor than in the two-factor model; the decline is up to 0.62 in variable X9. Additionally, interfactor correlation has improved from 0.17 in the two-factor model to 0.37 in the four-factor model.

Scale refinement
The remaining 41 items were randomly ordered as seven-point Likert-type statements in a questionnaire that was used to collect data for scale refinement. The questionnaire included a cover letter stating the purpose of the study and highlighting the importance of respondents' contribution to scientific progress in marketing theory and research. Useable data was provided by 213 managers. An exploratory factor analysis was performed in order to verify the postulated factors and related statements. An a priori decision rule specified that an item will not be retained unless its loading is greater than 0.7 from at least one factor. Unexpectedly, four factors and 17 items were extracted. Eigenvalues of these factors are 5.19, 3.54, 2.33, and 2.00. The variance accounted for by these factors is 13.07 for a proportion of 0.77 of the total variance. The discrepancy between the preliminary hypothesis and results of the exploratory factor analysis requires a closer look. Data were subjected to a separate exploratory factor analysis in which the number of factors was set to two. Eigenvalues of these two factors are 5.19 and 3.54. The variance accounted for by these two factors is 8.74 for a proportion of 0.51 of the total variance. For comparative purposes, the results of the two exploratory analyses are juxtaposed in both Table I and Figure 1. Table I shows the remaining statements and their final commonality estimates (reliabilities) and Figure 1 shows the rotated factor loadings (i.e. correlations) and inter-factor correlations from both analyses. Table I shows that there is a big improvement in statement reliabilities in the four-factor over the two-factor analyses for all items. Improvement ranges between 0.14 for statement number 4 (i.e. customers are given information that helps in developing realistic expectations) to 0.41 for statement number 9 (i.e. policies and procedures that do not create value for customers are eliminated) for an average improvement of 0.25. Figure 1 shows that there also is a big

Testing the model fit

In addition to parameter estimates, LISREL VII calculates the chi-square statistic (2 ) of

[ 430 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

Table I The 17-statement CUSTOR scale and their final commonality estimates Variable code X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15 X16 X17 Average Statement Customers play a consultative role in the selling process Customers are not promised more than can be delivered Employees understand what product attributes customers value most Customers are given information that helps in developing realistic expectations Employees go beyond normal call of duty to please customers Company has a system for monitoring, analyzing, and solving customer problems Customers can raise complaints very easily Information is continually sought and collected from customers Policies and procedures that do not create value for customers are eliminated Employees understand the company mission and objectives Customer needs take precedence over company internal needs Employees are given adequate resources to meet customer needs Customers' input and feedback are used into the product development process Company continually improves technologies, processes, and products Company continually reduces research and development cycle time Company continually invests in developing new ideas Management considers constant innovation a key to obtaining and sustaining competitive advantage Commonality estimates (reliabilities) Four factors Two factors Improvement 0.77 0.75 0.75 0.74 0.76 0.75 0.77 0.77 0.79 0.77 0.76 0.77 0.78 0.76 0.79 0.78 0.79 0.77 0.62 0.60 0.58 0.60 0.60 0.38 0.40 0.37 0.38 0.53 0.49 0.51 0.55 0.52 0.51 0.55 0.54 0.52 0.15 0.15 0.17 0.14 0.16 0.37 0.37 0.40 0.41 0.24 0.27 0.26 0.23 0.25 0.28 0.23 0.25 0.25

the model fit. The 2 statistic formally tests the null and alternative hypotheses where the null hypothesis is that the hypothesized model fits the data and the alternative hypothesis is that some model other than the one hypothesized fits the data. A significant 2 statistic indicates a misfit between the model and the sample data, hence, rejecting the null hypothesis. Alternatively, an insignificant 2 statistic indicates that the model and the sample data are consistent, hence, not rejecting the null hypothesis. However, showing that a theoretical model and sample data are consistent does not prove that the model is correct. The sample data may be consistent with many other competing models. In fact, we can only show that there is no discrepancy between the model and sample data. Additionally, the 2 statistic is highly sensitive to sample size. For a large sample size, a small difference

between the sample covariance matrix and the estimated covariance matrix will produce a statistically significant 2 statistic, and hence a model is rejected. Conversely, for a small sample size, the resulting 2 statistic will be normally insignificant, and therefore a model is accepted unless it is profoundly inappropriate. Consequently, a researcher has to use some other methods for evaluating the fit of the model to the data that are relatively independent of sample size (Bearden et al., 1982). Past research has suggested several goodness-of-fit indices for evaluating model fit (see Marsh et al. (1988) for a good review of these indices). Three of these indices will be used in the present research in addition to the 2 statistic. Bentler and Bonnet (1980) have proposed an index that they termed delta statistic (). Delta statistic refers to the proportion of covariance among all observed

[ 431 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

Figure 1 Variable loadings from four-factor (right side) and two-factor (left side) exploratory analyses

variables that a given model can account for. A value of at least 0.90 is recommended by Bentler and Bonnet for a researcher in order to have confidence in the model fit. A second index for evaluating the model fit is the goodness-of-fit index (GFI). This index provides a summary measure of the residual matrix (i.e. the difference between the sample and the estimated covariance matrices, RES = s ^). Notice that when s ^ (i.e. RES = 0), GFI = 1 and the model fits the data perfectly. Alternatively, when s T S^ (i.e. RES > 0), GFI equals less than 1 and the model does not fit the sample data perfectly. A rule of thumb is that GFI should be greater than 0.90 for good-fitting models. Finally, an index termed root-mean-square residual (RMSR) is used for evaluating the model fit (see Bone et al. (1989); Maiti and Mukherjee (1990) for an evaluation of different model fit indices). This index is the square root of the average of

the square of the residuals. The larger the RMSR, the less is the fit between the model and the sample data and vice versa. It is recommended that RMSR should be less than 0.10 in order to have confidence in the model. For the two-factor model, the 2 produced by LISREL VII in the confirmatory factor analysis is highly significant (2 118 = 981, p = 0.00), Goodness-of-fit index is low (GFI = 0.60), root-mean-square-residual is high (RMSR = 0.16), and delta statistic is low ( = 0.39). All these tests clearly indicate that the two-factor model is inconsistent with the sample data and that another model may represent the underlying structure of the scale items better. Alternatively, for the four factor model, the 2 produced by LISREL VII in the confirmatory factor analysis is insignificant (2 114 = 89, p = 0.96), Goodness-offit index is high (GFI = 0.96), root-meansquare-residual is low (RMSR = 0.02), and

[ 432 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

Figure 2 Variable loadings from four-factor (right side) and two-factor (left side) confirmatory analyses

delta statistic is high ( = 0.94). All these statistics demonstrate that the four-factor model is more consistent with the sample data and represents the underlying structure of the scale items better than the two-factor model. However, complete reliance on statistical measures may produce a statistically significant model but it is unlikely to lead to a theoretically meaningful model. A better

approach is to examine the scale items and to look for similarities and dissimilarities in what the scale items might be assessing, and to construct and test the model in light of this examination. Earlier, statements generated were classified into customer-related and company-related items; however, items within the same category may be measuring different lower-level constructs. In the present situation, customer-related and company

[ 433 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

related items appear to fall into two subcategories each. The first five statements (X1-X5 in Table I) appear to assess how intimate the company tries to get to customers whereas the next four statements (X6-X9) seem to assess how much the company cares about customer welfare. Consequently, the first group of statements is termed in the fourfactor model as customer intimacy and the second group is termed as customer welfare. Likewise, items numbered X10-X13 appear to measure how transparent is the company to its employees whereas the last cluster of items (X14-X17 in Table I) appear to assess how much improvement the company endeavors. Therefore, the third group of statements is termed in the four-factor model as company transparency and the last group is termed as continuous improvement. Although the four-factor model was more consistent with the sample data than the twofactor model, a number of other different models may be as consistent or even better. Because the four-factor model was developed and tested in an exploratory rather than a confirmatory fashion, it would be very advantageous to test this model on a fresh set of data. A second-order confirmatory factor analysis was performed on the matrix of intercorrelations between the 17 scale-items from another data set that was collected later from 167 subjects. The resulting model fit indices reached the pre-specified cutoff values (2 114 = 96, p = 0.89; GFI = 0.94; RMSR = 0.03; and  = 0.91) and indicated that this model is consistent with the new set of data as well.

Reliability and construct validity of CUSTOR scale

Figure 3 depicts the nomological structure of the construct of customer orientation. Customer orientation is viewed as a multifaceted construct composed of four components. Two of these components are called customer intimacy and customer welfare and pertain to the company's customers (i.e. external factors). The other two components are called company transparency and continuous improvement and pertain to the company's employees (i.e. internal factors). It is hypothesized that external factors are reflected in the company's economic performance, which can be assessed by measures such as sales volume, return on investment, and market share whereas the internal factors are reflected in job performance, which can be measured by jobrelated measures such as job satisfaction and self-worth. The analyses, therefore, will be aimed at securing enough evidence to validate the model portrayed in Figure 3.

Three types of construct validity will be examined in the present research. These are: convergent, predictive, and discriminant validity. First, convergent validity or the internal consistency of the scale items will be analyzed. This will be a test of the hypothesis that the scale items that are purported to measure a certain component of customer orientation have in fact a common, single underlying factor. Second, predictive or criterion validity will be examined in order to test whether a component of customer orientation as measured by particular scale items is related to other constructs in theoretically predicted ways. Finally, discriminant validity of the scale items will be examined. This test will attempt to answer two questions. First, are the scale items measuring a component of customer orientation reliably different from those items measuring another component? Second, do the scale items measuring a component predict a certain construct better than scale items measuring another component? In sum, the following hypotheses will be tested: H1. CUSTOR scale items converge into four latent constructs. In order of item appearance in Table I, items 1-5 assess the construct of customer intimacy, items 6-9 assess the construct of customer welfare, items 10-13 assess the construct of company transparency, and items 14-17 assess the construct of continuous improvement. H2. External factors (i.e. customer intimacy and customer welfare) predict the company's economic performance better than internal factors (i.e. company transparency and continuous improvement) whereas internal factors predict job performance better than external factors. H3. The four components of customer orientation as measured by their respective scale items are significantly different from each other. H4. The LISREL model depicted in Figure 3 as composed of a measurement and a structural equation models fits the sample data.

Method
The 17 items in Table I were randomly ordered as Likert-type statements carrying seven-points response formats in a questionnaire that was mailed to managers in a Middle East country. The questionnaire included a cover letter explaining the purpose of the study and the importance of

[ 434 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

Figure 3 The full customer orientation LISREL model

subjects' contribution to marketing research. At the end, managers were requested to answer some questions about themselves and their companies and two questions about their feelings of their job-satisfaction and self-worth. A self-addressed, stamped envelope was included in the mailer in order to enhance the response rate. Three weeks later, a reminder card was mailed to subjects encouraging them to respond if they had not. Over a period of eight weeks, 219 questionnaires were received of which only 197 were useable. Table II provides a summary of respondents' characteristics.

Analysis
Convergent validity
Several confirmatory factor analyses using LISREL VII were performed to test each of the research hypotheses. The first hypothesis requires evidence that the 17 items in the CUSTOR scale converge into four latent

constructs (i.e. theoretical convergence) but not in any other number such as 17 (i.e. no convergence), one (i.e. full convergence), or two latent constructs (i.e. partial convergence). The convergent validity analysis was performed in four stages. In the first stage, a model of 17 independent, though related, latent constructs measured by the 17 items of the CUSTOR scale (i.e. no convergence) was tested. Test statistics showed that this model was inconsistent with sample data. Chisquared was highly significant, goodness-offit index (GFI) was very low, root-meansquare-residual (RMSR) was very high, and delta () was extremely low (2 103 = 1,400, p = 0.000; GFI = 0.620; RMSR = 0.599; and  = 0.102). In the second stage, a model of one latent construct measured by all seventeen items of the CUSTOR scale (i.e. full convergence) was tested. Similarly, none of the test statistics rendered any support to this hypothesized model and the model was judged to be inconsistent with sample data

[ 435 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

Table II A summary of respondents' characteristics Characteristic 1. Respondent's gender Male Female 2. Respondent's age Under 35 35 to 49 50 and above 3. Respondent's education Up to high school Undergraduate degree (two to four years) Graduate degree (MA or PhD) 4. Respondent's managerial position Top (CEO, vice president) Middle (department head) Low (section head, supervisor) 5. Respondent's major product line Automobiles and industrial machines Electronics and electrical appliances Furniture and office equipment 6. Company experience Little (ten years or less) Moderate (11-20 years) Big (more than 20 years) 7. Company size Small (50 employees or less) Medium (51-100 employees) Large (more than 100 employees) 8. Last year's return on investment Small (up to 6 per cent) Medium (above 6 per cent and up to 12 per cent) Large (above 12 per cent) 9. Last year's market share Small (up to 5 per cent) Medium (6-14 per cent) Large (above 15 per cent) 10. Last year's dollar sales volume Small (up to $5million) Medium (above $5million and up to $10million) Big (above $10million)
(2 118 = 1,559, p = 0.000; GFI = 0.427; RMSR = 0.511; and  = 0.306). The third stage of the convergent validity analysis was concerned with testing the two-latent constructs model measured by the seventeen items of the CUSTOR scale (i.e. partial convergence). Once again, enough evidence not to reject this model was not obtained and the model was considered to be inconsistent with sample data (2 103 = 1,004, p = 0.000; GFI = 0.544; RMSR = 0.355; and  = 0.155). Finally, the analysis was concerned with testing a four-latent constructs model as

Number of respondents 137 60 58 107 32 32 137 28 65 103 29 53 95 49 28 130 39 76 108 13 111 76 10

Per cent of sample 69.5 30.5 29.4 54.3 16.3 16.2 69.6 14.2 33.0 52.3 14.7 26.9 48.2 24.9 14.2 66.0 19.8 38.6 54.8 06.6 56.3 38.6 05.1

113 59 25 118 62 17

57.4 29.9 12.7 59.9 31.5 08.6

hypothesized in H1 (i.e. theoretical convergence). The resulting test statistics were all at the specified levels and it was appropriate to conclude with a reasonable degree of confidence that this model was consistent with sample data (2 113 = 131, p = 0.112; GFI = 0.930; RMSR = 0.039; and  = 0.916).

Predictive validity

The purpose of the predictive validity analysis is to examine whether the four latent constructs of customer orientation are

[ 436 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

related to the constructs of economic performance and job performance in theoretically predicted ways. The second hypothesis postulates that the constructs of customer intimacy and customer welfare predict the company's economic performance better that what would the other two constructs, company transparency and continuous improvement, do. It also postulates that the latter two constructs predict the company's job performance better than the first two constructs would. Six separate LISREL analyses were performed in order to secure evidence in support of this hypothesis. In the first analysis, customer orientation constructs as measured by the 17-item CUSTOR scale were allowed to load freely on a single, common construct called economic performance that was measured by three indicators, sales volume, ROI, and market share. Test statistics showed that this model was consistent with sample data (2 159 = 184, p = 0.082; GFI = 0.921; RMSR = 0.040; and  = 0.909). Gamma values (i.e. the loadings of the different customer orientation constructs on the construct of economic performance) were consistent with the predicted direction and magnitude ( 11 = 0.243; 12 = 0.230; 13 = 0.062; 14 = 0.043). In the second analysis, the same model that was specified for the first analysis was tested with the assumption that the first two gamma values were not different from zero (i.e. 11 = 0.0; 12 = 0.0).

 = 0.927). However, gamma values (i.e. the loadings of the different customer orientation constructs on the construct of economic performance) were consistent with the predicted magnitude but not with the predicted direction ( 11 = 0.028; 12 = 0.044; 13 = 0.283; 14 = 0.262). In the fifth analysis, the same model that was specified for the first analysis was tested with the assumption that the second two gamma values were not different from zero (i.e. 13 = 0.0; 14 = 0.0). Two out of four test statistics showed that this assumption was not supported. Chi-squared was significant and root-mean-squareresidual was high (2 144 = 178, p = 0.025; GFI = 0.918; RMSR = 0.139; and  = 0.911). In the final analysis, the model was tested again with the assumption that the first two gamma values were not different from zero (i.e. 11 = 0.0; 12 = 0.0). Test statistics provided enough evidence to support this assumption (2 144 = 147, p = 0.393; GFI = 0.931; RMSR = 0.040; and  = 0.927). Additionally, gamma values became consistent with the predicted direction. The third hypothesis called for evidence to support the proposition that the constructs measured by the CUSTOR scale are statistically different at a significant level. Previously, it was shown that different components of customer orientation predict different job-performance constructs. It was also shown that customer-related constructs (i.e. customer intimacy and customer welfare) predict the construct of economic performance better than company-related constructs (i.e. company transparency and continuous improvement). Alternatively, company-related constructs were shown to predict the construct of job performance better than what customer-related constructs. Additional analysis to secure such evidence supporting the third hypothesis proceeded in four stages. First, a model of four independent, though related, latent constructs measured by the 17 items in the CUSTOR scale was tested (i.e. a test of the theoretical discrimination). Test statistics showed that this model was consistent with sample data. Chi-squared statistic was insignificant, goodness-of-fit index was high, root-mean-squared-residual was low, and delta statistic was high (2 113 = 131, p = 0.112; GFI = 0.930; RMSR = 0.039; and  = 0.916). The analysis resulted in relatively high correlation coefficients between customer intimacy and customer welfare ( = 0.311) and between company transparency and continuous improvement

Discriminant validity

` ... gamma values (i.e. the loadings of the different customer orientation constructs on the construct of economic performance) were consistent with the predicted magnitude but not with the predicted direction... '
Three out of four test statistics showed that this assumption was not supported. Chisquared was significant, root-mean-squareresidual was high, and delta was low (2 161 = 210, p = 0.005; GFI = 0.911; RMSR = 0.153; and  = 0.895). In the third analysis, the model was tested again with the assumption that the latter two gamma values were not different from zero (i.e. 13 = 0.0; 14 = 0.0). Test statistics provided evidence to support this assumption (2 113 = 161, p = 0.088; GFI = 0.921; RMSR = 0.050; and  = 0.908). In the fourth analysis, customer orientation constructs as measured by the 17-item CUSTOR scale were allowed to load freely on a single, common construct called job performance that was measured by two indicators, job satisfaction and self worth. Test statistics showed that this model was consistent with sample data (2 142 = 147, p = 0.362; GFI = 0.931; RMSR = 0.036; and

[ 437 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

( = 0.248). Other correlation coefficients were noticeably low and they ranged between 0.117 and 0.135. Second, the same four-construct model was tested while assuming that all four constructs are not different from each other by forcing all inter-factor correlations to equal unity (i.e. zero discrimination). Test statistics showed clearly that this assumption couldn't be accepted. Chisquared statistic was highly significant, goodness-of-fit index was extremely low, root-mean-squared-residual was very high, and delta was extremely low (2 119 = 1,959, p = 0.000; GFI = 0.427; RMSR = 0.511; and  = 0.0001). In the third analysis, the four-factor model was tested while assuming that all four factors were completely unrelated by forcing all inter-factor correlations to equal zero (i.e. full discrimination). Three out of four test statistics showed that this assumption could not be accepted. Chi-squared statistic was significant, root-mean-squared residual was high, and delta was below the cutoff value of 0.9 (2 119 = 166, p = 0.003; GFI = 0.911; RMSR = 0.252; and  = 0.893). In the final analysis, the correlation between customer intimacy and customer welfare constructs and the correlation between company transparency and continuous improvement constructs were forced to unity (i.e. partial discrimination) test statistics showed that this assumption was not supported. Chi-squared statistic was significant, goodness-of-fit index was low, root-meansquared residual was high, and delta was low (2 115 = 1,258, p = 0.000; GFI = 0.547; RMSR = 0.352; and  = 0.193).

latent constructs. In the course of doing so, variances will be described and assigned to their respective variables. Data was subjected to a LISREL analysis using LISREL VII software. Relationships between observed and latent variables were specified in the analysis exactly as they are described in Figure 3. Three test statistics out of four indicated that this model achieved convergent, predictive, and discriminant validity, concurrently, and that hypothesis 4 could not be rejected (2 198 = 224, p = 0.094; GFI = 0.916; and  = 0.9). The only test statistic that did not reach the rule-ofthumb cutoff value of 0.10 for its class was root-mean-square-residual (RMSR = 0.117). Figure 4 is a replica of Figure 3 but with all parameter estimates produced by the LISREL analysis shown. Of course, all factor loadings are correlations because all variables were standardized. Reliability estimates are shown in the table at the bottom of Figure 4.

Discussion and conclusions

It is rather clear that the development of valid and reliable scales is essential for improving marketing theory and research. Pioneering scholars of marketing thought have always stressed that the development of marketing as a science seems to be virtually impossible unless theoretical constructs can be specified and adequately measured (Bartels, 1951; Converse, 1945; Hunt, 1976). The present research took the initiative and identified the domain of an important construct, customer orientation, and developed and validated a multidimensional scale for its measurement. Customer orientation was found to be a first-order entity composed of four second-order constructs. These are customer intimacy, customer welfare, company transparency, and continuous improvement. A psychometrically rigorous scale composed of 17 items was empirically developed and tested. The scale was found to possess all desirable measurement properties. Item reliabilities ranged between 0.71 and 0.80 and LISREL tests produced strong evidence supporting the convergent, predictive, and discriminant validity of the structure of customer orientation (see Table III for a summary of test results). The primary limitation of the present research is that all studies were carried out in Middle East countries where these countries virtually have the same culture (i.e. majority of people are Muslim Arabs). An English version of the questionnaire was sometimes used and at other times an Arabic

Validity and reliability: a comprehensive analysis

In the previous analyses, parts of the model depicted in Figure 3 were tested independently for convergent, predictive, and discriminant validity. Evidence was obtained in support of all H1-3 and neither could be rejected. H4 is concerned with testing the whole model depicted in Figure 3 at once for convergent, predictive, and discriminant validity of both observed and latent variables concurrently. The analysis will be a test of how much the components of customer orientation depend on the CUSTOR scale items and how much are they indicated by economic and job performance measures. In other words, the analysis will be a test of the measurement properties of all observed variables (i.e. reliabilities and validities) and purported causal relationships between

[ 438 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

Figure 4 Parameter estimates of the full customer orientation LISREL model

translation was used, depending on the subjects' English proficiency. No East-West validation of the model was possible. This opportunity is left to future research. Another limitation is that although

predictive validity analysis required the incorporation of other constructs in the model (i.e. economic performance and job performance), the measurement of these constructs was not exhaustive. Only three

[ 439 ]

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

Likert-type items were used to tap the company's economic performance and two to tap its job performance. This explains some of the weak results in the comprehensive validity and reliability analysis in the last section. Structural equation (i.e. economic and job performance equations) reliabilities 2 were low (2 1 = 0.318; r2 = 0.324). The relatively low economic and job performance reliabilities indicate that these constructs require additional research attention from marketing scholars. The major contribution of the present research is that it explicates the latent structure of a psychological construct that has always been of major importance to marketing theory and research. This has several research and managerial implications. The primary focus of customer orientation research has been through studying the combined effects of market orientation components (Lucas and Ferrell, 2000). The preliminary findings of the predictive analyses suggest that the company's economic and job performance can be improved, not only by focusing on customer orientation but rather through examining and emphasizing certain elements. This signifies the importance of certain components of customer orientation to achieving varying performance outcomes under different conditions. Researchers and practitioners can examine the effect of various elements of customer orientation on

specific strategies and orientations. For example, research can address the question of whether customer orientation encourages innovative, entrepreneur, or learning orientation. It also can incorporate customer orientation in more comprehensive nomological networks, such as the market orientation network, and reexamine its role in achieving certain positional advantages and performance outcomes. Longitudinal research is also needed in order to examine causal relationships between customer orientation and other related behavioral constructs such as superior customer value, buyerseller relations, disconfirmation of expectations, customer satisfaction, brand loyalty, and brand attitudes. The CUSTOR scale can be reworded and administered to customers to assess their perception of the desired level of customer orientation the company should exercise. More important, different levels of customer orientation can be examined in order to find the one accepted by customers and which lead to the desired performance outcomes. The relationship between customer-related factors and companyrelated factors is of special interest to marketers. The magnitude and direction of impact should be investigated in future research.

Table III A summary of validity test statistics Model Convergent validity test Zero convergence Full convergence Partial convergence Theoretical convergence Predictive validity test: predicting economic performance Constrained ( 11 and 12 = 0) Constrained ( 13 and 14 = 0) Theoretical prediction Predictive validity test: predicting job performance Constrained ( 13 and 14 = 0) Constrained ( 11 and 12 = 0) Theoretical prediction Discriminant validity test Zero discrimination (all s = 1) Full discrimination (all s = 0) Partial discrimination (all s T 1) Theoretical discrimination (21 and 43 are free, all other s = 0) Comprehensive validity test Theoretical validity (preceding theoretical models altogether) Notes: number of customer orientation constructs; constructs on performance constructs [ 440 ]
a b

Model specifications NKa = 17 NK = 1 NK = 2 NK = 4 NK = 4, NE = 1, GA = 2 NK = 4, NE = 1, GA = 2 NK = 4, NEb = 1, GAc = 4 NK = 4, NE = 1, GA = 2 NK = 4, NE = 1, GA = 2 NK = 4, NE = 1, GA = 4 NK = 4 NK = 4 NK = 2 NK = 4 NK = 4, NE = 2

d.f. 103 118 103 113 161 161 159 144 144 142 119 119 115 113 198
c

2 1,400 1,559 1,004 131 210 185 184 178 147 147 1,959 166 1,258 131 224

p 0.000 0.000 0.000 0.112 0.005 0.088 0.082 0.025 0.393 0.362 0.000 0.003 0.000 0.112 0.094

GFI 0.620 0.427 0.544 0.930 0.911 0.921 0.921 0.918 0.931 0.931 0.427 0.911 0.547 0.930 0.916

RMSR 0.599 0.511 0.355 0.039 0.153 0.050 0.040 0.139 0.040 0.036 0.511 0.252 0.352 0.039 0.117

Delta 0.102 0.306 0.155 0.916 0.896 0.908 0.909 0.911 0.927 0.927 0.000 0.893 0.193 0.916 0.900

number of performance constructs;

number of loadings from customer orientation

Mahmood M. Hajjat Customer orientation: construction and validation of the CUSTOR scale Marketing Intelligence & Planning 20/7 [2002] 428441

References

Baker, T., Simpson, P. and Siguaw, J. (1999), ``The impact of suppliers' perceptions of reseller market orientation on key relationship constructs'', Journal of the Academy of Marketing Science, Vol. 27 No. 1, pp. 50-7. Bartels, R. (1951), ``Can marketing be a science?'', Journal of Marketing, 15 January, pp. 319-28. Bearden, W., Sharma, S. and Teel, J. (1982), ``Sample size effects on chi square and other statistics used in evaluating causal models'', Journal of Marketing Research, Vol. 19, November, pp. 425-30. Bell, M. and Emory, W. (1971), ``The faltering marketing concept'', Journal of Marketing, Vol. 35, October, pp. 37-42. Bentler, P. and Bonnet, D. (1980), ``Significance tests and goodness of fit in the analysis of covariance structures'', Psychological Bulletin, Vol. 88, pp. 588-606. Bone, P., Sharma, S. and Shimp, T. (1989), ``A bootstrap procedure for evaluating the goodness-of-fit indices of structural equations and confirmatory factor models'', Journal of Marketing Research, February, pp. 105-11. Cahill, D., Thach, S. and Warhshawsky, R. (1994), ``The marketing concept and new high-tech products: is there a fit?'', Journal of Product Innovation Management, Vol. 11, pp. 336-43. Converse, P. (1945), ``The development of a science in marketing'', Journal of marketing, Vol. 10, July, pp. 14-23. Day, J. (1994), ``The capabilities of market-driven organizations'', Journal of Marketing, Vol. 58, October, pp. 37-52. Deshpande, R. and Farley, J. (1996), ``Understanding market orientation: a prospectively designed meta-analysis of three market orientation scales'', working paper, No. 96, Marketing Science Institute, Cambridge, MA p. 125. Deshpande, R., Farley, J. and Webster, F. (1993), ``Corporate culture, customer orientation, and innovativeness in Japanese firms: a quadrad analysis'', Journal of Marketing, Vol. 57, January, pp. 23-7. Dreher, A. (1993), ``Marketing orientation: how to grasp the phenomenon'', in Chias, J. and Sureda, J. (Eds), Marketing for the New Europe: Dealing with Complexity, Vol. 1, European Marketing Academy, Barcelona, pp. 375-93. Drucker, P. (1954), The Practice of Management, Harper and Row Publishers, New York, NY. Hall, R. (1999), ``Customer-centric: from slogan to reality'', Bank Marketing, February, p. 12. Han, J., Kim, N. and Srivastava, R. (1998), ``Market orientation and organizational performance: is innovation a missing link?'', Journal of Marketing, Vol. 62, October, pp. 30-45.

Hunt, S. (1976), ``The nature and scope of marketing'', Journal of Marketing, Vol. 40, July, pp. 17-28. Jaworski, B. and Kohli, A. (1993), ``Market orientation: antecedents and consequences'', Journal of Marketing, Vol. 57, July, pp. 53-70. Joreskog, K. and Sorbom, D. (1989), LISREL VII User's Reference Guide, Scientific Software, Inc., Mooresville, IN. King, R. (1965), ``The marketing concept'', in Schwartz, G. (Ed.), Science in Marketing, John Wiley & Sons, New York, NY, pp. 70-97. Kohli, A., Jaworski, B. and Kumar, A. (1993), ``MARKOR: a measure of market orientation'', Journal of Marketing Research, Vol. 30, November, pp. 467-77. Lucas, B. and Ferrell, O. (2000), ``The effect of market orientation on product innovation'', Journal of the Academy of Marketing Science, Vol. 28 No. 2, pp. 239-47. McKitterick, J. (1957), What Is the Marketing Thought and Action?, American Marketing Association, Chicago, IL, pp. 71-82. McNamara, C. (1972), ``The present status of the marketing concept,'' Journal of Marketing, Vol. 36, January, pp. 50-7. Maiti, S. and Mukherjee, B. (1990), ``A note on distributional properties of the Joreskog and Sorbom fit indices'', Psychometrika, Vol. 55, December, pp. 721-6. Marsh, H., Balla, J. and McDonald, R. (1988), ``Goodness-of-fit indexes in confirmatory factor analysis: the effects of sample size'', Psychological Bulletin, Vol. 103, pp. 391-410. Narver, J. and Slater, S. (1990), ``The effect of a market orientation on business profitability'', Journal of Marketing, Vol. 54, October, pp. 2035. Perreault, W. and McCarthy, J. (1997), Essentials of Marketing, Irwin, Chicago, IL. Ruekert, R. (1992), ``Developing a marketing orientation: an organizational strategy perspective'', International Journal of Research in Marketing, Vol. 9, pp. 225-45. Siguaw, J., Brown, G. and Widing, R. (1994), ``The influence of the market orientation of the firm on sales force behavior and attitudes'', Journal of Marketing Research, February, pp. 106-16. Siguaw, J., Simpson, P. and Baker, T. (1998), ``Effects of supplier market orientation on distribution market orientation and the channel relationship: the distributor perspective'', Journal of Marketing, Vol. 62, July, pp.99-111. Slater, S. and Narver, J. (1994), ``Does competitive environment moderate the market orientation-performance relationship?'', Journal of Marketing, Vol. 58, January, pp. 46-55. Webster, F. (1988), ``Rediscovering the marketing concept'', Business Horizons, Vol. 31, May-June, pp. 29-39.

[ 441 ]