"Development of a tool for selecting mobile shopping site: A customer perspective"

Abstract:

While mobile technologies and applications are rapidly and widely utilized in electronic commerce, it is extremely important to better understand the evaluative criteria from the consumer's viewpoint in mobile shopping (m-shopping) site selection. This paper investigated the consumer-oriented criteria for m-shopping site selection. An initial criteria list was developed based on some previously validated instruments and then tested to prove its reliability and validity. The final criteria list includes three factors: assurance, merchandise, and enabling functions. This list provides multidimensional criteria for m-shopping site selection in business to consumer markets and provides great insight for managing and developing m-shopping sites. 2006 Elsevier B.V. All rights reserved.

1. Introduction:

With the tremendous advances in hand-held computing and communication capabilities, mobile commerce (mcommerce) is expected to be the next big wave in business. A number of m-commerce applications have been developed and are already in use, covering a wide range of business functions from advertising, banking, ticketing, games, to shopping. Today, the world of business is witnessing profound changes under the influence of wireless technology. The opportunity for m-commerce has opened. M-commerce broadly refers to any transactions with monetary value that is conducted over a wireless telecommunication network [1]. Market researchers have predicted that, by the end of the year 2005, nearly 500 million wireless device users willexist, generating more than $200 billion in revenue [27]. Dollars increase in wireless services may soar from $37 billion in 2001 to $74 billion in 2005 [21]. Mobile Internet access will become a primary tool for completing daily information transactions, e.g., e-mail, retail shopping, and receiving the news [8,22]. Among the innovations in m-commerce services for consumers, shopping via the mobile channel could have a great potential and opportunity [16] and would be a major business channel in the coming years [22]. Mobile shopping (mshopping) allows the consumers to order and pay for goods using a mobile phone regardless of time and place. However, m-shopping could disrupt existing retail models and threaten the established orthodoxies of online selling as many as those of entrenched physical retailers [29]. If shops equip with an understanding of what determines, encourages, and promotes m-shopping consumers, steps can be taken to meet the consumer's expectations and thereby increase the consumer and sales growth rate. However, researches addressing the consumer's perception of m-shopping sites are scarce. The main goal of this paper is to identify consumer evaluative criteria for the selection of m-shopping sites in business to consumer (B2C) markets that are multidimensional nature. The list of consumer evaluative criteria proposed in this study would be valuable to researchers and practitioners interested in implementing and managing m-shopping sites. The remainder of the paper is organized as follows. Section 2 introduces the m-shopping background and briefly reviews prior related research, including the characteristics and scenario for m-shopping and criteria for the electronic store selection. Section 3 presents the initial criteria list for selecting m-shopping sites. Section 4 describes the research methodology and the sampling techniques adopted for this study. Section 5 analyzes and discusses the results. The last section suggests several implications in administrating m-shopping sites and identifies future research directions that they suggest.

3. Selection criteria for m-shopping sites A three-phased approach was used in constructing the selection criteria list: Phase 1. Develop an initial measure list by taking results from a literature review and examining m-shopping characteristics and then determining whether it was complete clear by using it in interviews of domain experts of mobile commerce applications. Revise the list accordingly and use it in a pilot test with 30 experienced mobile shoppers. As a result, the initial survey instrument was extensively revised. Phase 2. The new instrument was then tested via a survey. Some tests were conducted to purify the instrument. Exploratory factor analysis (EFA) was then performed to initially estimate the factor structure. Phase 3. A new data set was gathered. The factor structure derived from EFA was tested and modified using the new data set. 3.1. Initial consumer-oriented criteria development This study aims to identify consumer evaluative criteria for the selection of m-shopping sites in B2C markets. To ensure a comprehensive list of criteria was included, a broad range of previous studies from the marketing, physical shopping, web-based shopping, and m-commerce areas were reviewed [1,2,9,12-14,17-20,24,25,27,28]. An initial list of 35 items based on the five consumer criteria dimensions, as indicated in Table 1, were selected and reworded for the m-shopping environment. These included criteria such as a light screen presentation, providing personalized shopping-information, providing more decision support functions to shorten browsing time and uncertainty (e.g., comparing shopping information and a function for a product-information search, recommendation, etc.). A five point Likert-type scale was used to determine individual reactions to the items, where: 1, not important at all; 2, unimportant; 3, moderately; 4, important; and 5, extremely important. Once the initial list was generated, a review process by the domain experts from the mobile service providers (MSPs) was conducted to verify the completeness, wording, and appropriateness of the items and to verify the content validity. The domain experts included three marketing managers and six professionals on mobile commerce applications. First, we explained the research purpose to them. Then, they were asked to provide feedbacks and comments about the measure. The review process was followed by a pilot study that involved sending thirty questionnaires to mobile shoppers that were recommended by the four MSPs in Taiwan. Feedbacks from the pilot study served as the basis for correcting, refining and enhancing the items. Finally, 27 items remained,

3.2. Purify measure:

Subjects for this study were users engaged in m-shopping in Taiwan. The Taiwan Government launched its National Mobile Infrastructure Project in 2002 and claimed that by year 2006 Taiwan would be a mobile island. However, currently, m-shopping in Taiwan is still in the early implementation stage. Only few well-known stores have actually implemented or partially implemented m-shopping. Due to the small amount of m-shopping samples, four main MSPs in Taiwan are good base to distribute questionnaires because they provide the mobile channel and have the transaction records. We endeavored to find a specific local contact person for each targeted MSP that was placed in charge of distributing the questionnaires and the follow-up activities. A questionnaire was mailed to 1100 random customers that possessed mobile phones with micro-browser features, via the marketing departments of four main MSPs in Taiwan. The respondents were asked to assess the importance each of item regarding to selecting an m-shopping site. We received 183 usable responses, for a response rate of 16.6%.We assessed potential non-response bias by comparing the early versus late respondents that were compared on several demographic characteristics as shown in Table 3. The results indicated that there are no statistically significant differences across early and late respondents. These suggested that non-response bias was not a serious concern. Prior to conducting formal data analysis, the internal consistency (a coefficient) was examined to ensure that the measures were unidimensional and to eliminate ''garbage items'' [3]. The results showed that the 27-item instrument had an acceptable reliability of 0.89. Correlations of each item with the sum of scores on all items were computed. Items were eliminated if their correlations with the sum of scores were less than 0.4 [3]. Thus, the item no. 12 was eliminated. The remaining 26 items had a reliability of 0.938. Bartlett's test of sphericity (p = 0.000) indicated that correlations among these items existed. A Kaiser-Meyer-Olkin measure of sampling adequacy yielded a score of 0.92, indicating high shared-variance and relatively low uniqueness. These test results suggested that factor analysis was worth pursuing.

3.3. Exploratory factor analysis EFA was used to validate the various constructs underlying the data set. The principal components and maximum likelihood methods with varimax rotation were used to examine the data. To derive a stable factor structure, two rules were applied to eliminate items: (1) loadings of less than 0.35 on all factors; and (2) loadings greater than 0.35 on two or more factors [23]. Three iterations yielded a stable set of three factors and left 11 items, as shown in Table 4. The first factor drew from items related to assurance. The second contained items related to the merchandise, such as providing products that I need, providing attractive products, providing a function for a product information search, and providing a function for a product preview. Three items were loaded on a new factor called ''enabling functions'' created by combining service, promotion, and convenience. This can be the reason that service and convenience are highly related and service always embraces convenience. Therefore, these factors were labeled ''Assurance'', ''Merchandise'', and ''Enabling Functions'' and explained 68.8% of the variance in the data set. The 11-item measure had an acceptable reliability of 0.86 and the a coefficients for the three derived factors were all above 0.70.

4. Data analysis and discussion:

4.1. Confirmatory analysis

According to Doll et al. [5], the research cycle for developing a standardized instrument should have two steps: (1) an exploratory study that develops a hypothesized measurement model via empirical data analysis; and (2) confirmatory studies that test the hypothesized measure model. Confirmatory factor analysis (CFA) provides a more rigorous and systematic factor structure test, thus CFA was conducted in this study. The data gathering method in confirmatory analysis was identical to those used at the exploratory stage. The 11-item measure questionnaire was mailed to the remaining 917 customers that made no response to the exploratory survey via the contact persons by the four main MSPs in Taiwan. We received 180 usable responses, a response rate of 19.6%. This sample size exceeds the minimum of 100 recommended by Joreskog and Sorbom [10]. The demographic characteristics of the sample are summarized in Table 5. A three first-order factor model was formulated based on the EFA results (see Fig. 2). This model hypothesized that the three first-order factors were correlated with each other. The LISREL 8.30 program was then used to test the model fit.

4.1.1. Comparing model-data fit Because not one statistic was universally accepted as an index of model adequacy, several measures were used. The goodness-of-fit indexes for the hypothesized measure model are summarized in Table 6. Although the v2 statistic is a global test of a model's ability to reproduce the sample variance/covariance matrix, the v2 statistic is sensitive to the sample size [5]. We therefore utilized the ratio of v2 to the degrees of freedom (i.e., v2 /df) to examine the hypothesized model fit. Researchers have recommended using ratios of greater than 2 and less than 5 to indicate a reasonable fit [15]. Goodness-of-fit (GFI), root mean square residual (RMR), adjusted goodness-of-fit index (AGFI), normed fit index (NFI), non-normed fit index (NNFI), comparative fit index (CFI), and incremental fit index (IFI) were used to evaluate the model. The following criteria of the goodnessof-fit indices were used to assess model fit: GFI > 0.85, RMR < 0.05, AGFI > 0.80, NFI > 0.80, NNFI > 0.80, and CFI and IFI approach 1 [6,7]. The value of CFI and IFI varies between zero and 1.0, with higher values indicating greater model parsimony. As shown in Table 6, the hypothesized model exhibits good levels of fit and thus provides a satisfactory representation of the underlying structure of the measure for mshopping site selection.

4.1.2. Assessment of reliability and validity The reliability and validity of the measurement model were assessed using several tests. On the first-order CFA measurement models, the standard factor loading of an observed variable (item) on its specified latent variable (factor) is the estimate of the observed variable validity. The larger the factor loading, compared with their standard error and expressed by the corresponding t value, the stronger the evidence that the measured variable (factor) represents the underlying construct [5]. In general, if the t values are greater than |1.96| or |2.576|, they are considered significant at the 0.05 and 0.01 levels, respectively [11]. Our examinations of the t values (Fig. 3) indicated that each item exceeded the critical value for the 0.01 significance level. That is, all items were significantly related to their specified constructs, verifying the posited relationships among items and factors. All construct reliabilities (i.e., 0.93, 0.76 and 0.72) exceeded the recommended level of 0.70. For the average variance-extracted (AVE) measures, the estimates were 0.76, 0.45, and 0.46, respectively. Although factor 2 (merchandise) and factor 3 (enabling functions) fell somewhat short of the recommended level of 0.5, the overall reliability of the measurement was acceptable. Discriminant validity was assessed by developing a confidence interval of w 2re for each pair of factors and examining whether 1 was included. The w notation means that the correlation between two factors and thee notation means standard error between two factors. If 1 is not included, it will provide discriminant validity evidence [11]. All three confidence intervals (0.48 and 0.72), (0.54 and 0.82), and (0.4 and 0.68) do not include the value of 1. This provides evidences to support discriminant validity. 4.2. Results and discussion The empirical results enhance our understanding of the nature and dimensionality of m-shopping site selection from the consumer's perspective. The consumer-oriented criteria provide multidimensional constructs (i.e., assurance, merchandise, and enabling functions) for the selection of m-shopping sites. All of these factors receive considerable consumer support. The three factors are interwoven, and one must not focus exclusive on any single factor. The positive correlations among the three factors imply that they may also be interpreted hierarchically with the three first-order factors being aspects of the higher second-order factor of ''consumer's site image''. Table 7 summarizes the mean perceived importance score for the 11 criteria. The scores for all items are above 3.88. This means that these eleven items are important tothe subjects. More specifically, the ''assurance'' factor scores (i.e., items 21, 22, 23, and 24) are higher than the ''merchandise'' and ''enabling function'' factor scores. Since m-shopping occurs at a distance via a limited screen interface rather than face-to-face, the shoppers may perceive that uncertainty, unsafety, and uneasiness are all higher than in the traditional shopping or electronic shopping context. This may imply that security, privacy, product return, and refund guarantees are the important issues for m-shopping site selection. As in electronic commerce [4,26], the ''assurance'' factor strongly influences the decision of whether to adopt m-shopping and select a mobile store. The ''merchandise'' and ''enabling function'' are also important factors to shoppers in choosing an m-shopping site because not all products that consumers need are available in an m-shopping site. This may cause un-satisfaction to the shoppers and suggest that the m-shopping site needs to provide and modify product portfolio to satisfy their buyers' tastes or needs. Further, the features of product description information and product preview are important in m-shopping because consumers cannot touch or feel products. The functions of product search, shopping information comparison, and on-line help in the m-shopping site can enable consumers to reduce search time and cost. These functions can help consumers find the products they seek effectively and efficiently. Promotion is another important issue to motivate site visits and sales to the shoppers, which may involve sales, advertising, and appetizer features that attract shoppers. Good promotion may not only attract customers to revisit but also increase the enjoyment of m-shopping.

5. Conclusion and implications With the growing penetration of mobile device usage and the widespread adoption of the Net, stores need to investigate their customers' perceptions and fit their actual needs. This research presents an initial effort in developing a standard criteria list for selecting m-shopping sites from the consumer perspective. The three-factor consumer-oriented criteria further provide great insights for managing and developing m-shopping sites: M-shopping sites should provide the right products (that attract consumers and fit the consumer's needs) and promotion activity. M-shopping sites should provide functions that help consumers search, view, compare, and purchase merchandises easily and securely. M-shopping sites should provide refunds, flawed-product return guarantees and have a commitment to privacy protection. The findings have several implications for practitioners and researchers. Understanding the consumer is critical for successful management and development of m-shopping sites. The criteria list gives practitioners the insight they need to define what is extremely important to their customers. Once the researchers and practitioners know what is important, they can use the criteria list to compare each store's performance on those items. Comparing the m-shopping site with similar competitors' sites generates ideas about how to improve the store's design. Surveys help the practitioners identify which stores may need to improve so they can provide consumers with a better shopping experience. Results of statistical analysis help management decide how and where to better allocate store resources, such as training and technology. This research provides a better understanding of the design space in which the mobile stores operate. While this study has produced some interesting results, they should be interpreted with caution. Although the surveyed buyers represent a wide diversity of subjects, the sample total of 183 and 180 respondents may not be large enough to generalize the entire mobile shopper population. M-shopping is still in its infancy and rare relative to Internet and physical shopping. Thus, this instrument was developed and tested on a consumer sample that possessed mobile phones and had Internet access experience via mobile phones. This study provided a relatively small test of consumer involvement. The disproportionate sample size among the ages could also lower the significance of the study results. There are several directions for follow-up research. For instance, the criteria could be examined again to confirm the hypothesized model, assess the stability of the instrument and develop standards for evaluating specific applications. More attention could be directed toward understanding the antecedents and consequents of the evaluative criteria for m-shopping site selection and shopping behavior.

Q.No.1 Briefly explain Business to Consumer (B2C) in perspective of electronic Commerce (Only from assigned material).