JCMC 8(1) October 2002

Message Board

• Abstract
• Introduction
• Product Classification Theory
• Shopping Preferences
• Internet Retailer Attributes
• Method
• Research Instruments and Data Collection
• Second Stage Sample
• Second Stage Survey Instrument
• Analysis
• Results
• Discussion
• Conclusion
• References
• About the Authors

Abstract

Introduction

Despite the increase in use and popularity of the Internet over the last few years, the question of why consumers prefer to shop on the Internet for certain products and not for others still remains poorly understood. Empirical research on consumers' preference for shopping on the Internet has not been abundant. Until just a few years ago, the Internet had been relatively new to consumers as a shopping medium, and is still in a growth phase. Lack of familiarity with its use and the risk perceived by consumers in revealing personal information as a part of online purchasing has created uncertainty and wariness about untried e-tailers. In addition, the appeal and adoption of online shopping have been hindered by inferior Internet retail site design and functions. Finally, historical trends have not had sufficient time to accumulate to predict consumer shopping behavior (Peterson et al., 1997).

The euphoria of the early years of online shopping has been replaced by more realistic and cautious projections of e-commerce sales. While use of the Internet for the purposes of shopping, information search, communication, interaction, and entertainment has continued to increase, the actual figures for e-commerce sales have not increased as rapidly as expected. Taylor Nelson Sofres Interactive (TNSI) in June 2001 reported a fifty percent increase in the number of Internet users in thirty-six countries between June 2000 and 2001 (http://www.tnsofres.com/ger2001). However, the U.S. Census Bureau report for the first quarter of 2001 demonstrated that online retail sales in the U.S. have not increased as rapidly as predicted. E-commerce sales in the first quarter of 2001 accounted for 0.9 percent of total sales, compared to 0.7 percent of total sales in the first quarter of 2000.

As online retail sales continue to increase at a slower pace than expected, academicians and practitioners alike are searching for the product categories that consumers will shop for on the Internet. Consumers' preferences for shopping on the Internet may depend on the product type, which will in turn influence the need to obtain product information easily and cost-effectively, or to test or try products before purchasing. In addition, consumers' willingness to purchase on the Internet may vary depending on the attributes that Internet retailers offer for online-shopping (i.e., information and order services, privacy, quality of products, site quality, etc.). In the brick-and-mortar retailing and catalog shopping literature, the published research indicates that the importance of store/mail-order attributes varies by product category (Eastlick & Feinberg, 1999). Lynch, Kent and Srinivasan (2001) found that with respect to e-commerce "the impact of site quality on loyalty and purchase intentions depends on the particular product category" (p.7). Drawing upon the previous literature, the authors propose that product classifications have a significant impact on consumers' preference for shopping on the Internet, and the importance they assign to Internet retailers' attributes.

According to the Ernst and Young Global Online Retailing Report (Macintosh, 2001), there is a discrepancy between e-tailers and customers regarding why customers visit a site. Retailers were reported to believe that factors such as convenience, reputation/trust, and customer service were most important, while customers were reported to list merchandise assortment and competitive prices as the factors that mattered most to them (McIntosh, 2001). The reason for the discrepancy is perhaps that the type of product purchased is influential in determining which attributes are more important in choosing a retailer to patronize. In the present study, the authors attempt to clarify why e-commerce is not growing as fast as expected and why consumers prefer to purchase certain products online and not others. The findings of the study reported here may assist academic researchers in marketing, advertising, and communication to build paradigms related to e-commerce. It may also help Internet retailers understand which Internet retailer attributes are important to consumers for specific product types so that they can communicate to them with proper messages and convey the appropriate product-related information on their Web sites and in their advertising.

Product Classification Theory

Although it is often speculated that the type of product purchased will have a significant impact on Internet retail patronage, little published research exists investigating the impact of the type of product-purchase on Internet shopping preferences. Several research studies acknowledge that consumers' buying behavior characteristics vary noticeably across product categories. Porter (1974) suggests that consumers may base their purchase decisions on product attributes such as brand image, reliability, styling, and availability of servicing. Porter explains that retailers control some of the attributes which consumers may want in the product. For instance, the reputation and image of a retailer may be reflected in the quality of the product or image of the brand. Recognizing that consumer buying characteristics vary by product type, Porter points out the shortcomings of classifying goods into only two categories: convenience and shopping goods. He argues that the factors such as unit price and purchase frequency do not necessarily distinguish buying behavior between the two product classes.

Although not investigated empirically, the Bloch and Richins (1983) classification of goods theory suggests that consumers' shopping efforts vary with respect to type of product. While Copeland (1923) identifies goods in separate categories such as convenience, shopping, and specialty goods, Aspinwall (1968) and Holton (1958) propose that products reflect shopping effort more appropriately if they are placed along a continuum. Klein (1998) examines the Internet's influence on information search and proposes a consumer information search model using the principles of information economics and a goods classification model based on search, experience, and credence paradigms. She demonstrates how search goods can become experience goods by three routes. Similarly, using search, experience, and credence product classification along a continuum, Brucks, Zeithaml, and Naylor (2000) develop a typology of quality dimensions for durable goods. They draw their model from Nelson (1970), who distinguishes between two categories of products, search and experience, and from Darby and Karni (1973), who add a third product category to Nelson's classification called credence goods.

The first attempts to define the term search, experience and credence were triggered by consumers' skepticism about advertisers' exaggerated claims and consumers' efforts to verify the truthfulness of those claims (Darby & Karni, 1973; Nelson, 1970). Deriving from Stigler's (1961) explanation of the "search" phenomena and the theory of economics of information, Nelson (1970) originally describes the qualities of search and experience goods in the advertising context. Nelson (1974) defines a search good as one whose qualities and suitability a consumer can determine by inspection prior to purchase of the brand. More specifically, a good is a search good when full information for dominant product attributes can be known prior to purchase, whereas an experience good is one whose qualities cannot be determined prior to purchase (Klin, 1998). Nelson (1970, 1974) classifies experience goods as experience durable (low frequency of purchase goods) and experience nondurable (high frequency of purchase goods) and tests for significant differences in the advertising sales ratios for search, and the two experience good classifications. Nelson finds significant differences among the means of advertising sales ratio for the three classifications.

Nelson defines an experience good as one whose qualities a consumer cannot determine prior to purchase. Based on Nelson's work (1974), Kline (1998) provides two criteria for classifying a good as an experience good. A good is an experience good when either (1) full information on dominant attributes cannot be known without direct experience, or (2) information search for dominant attributes is more costly/difficult than direct product experience (Kline, 1998, p. 199). Wright and Lynch (1995) broaden Nelson's definition of experience goods to include "after using" rather than "after purchasing" because of the fact that consumers might receive or test free samples in a store without purchasing the product.

Darby and Karni (1973) originated the definition of a credence good: a good such that the average consumer can never verify the level of quality of an attribute possessed by a brand or even their level of need for the quality supplied by the brand. That is, consumers will have great difficulty in evaluating the quality level of a product such as vitamins with confidence, or similarly a service such as termite fumigation or surgery, "even after purchase or consumption" (Asch, 2001; Brucks, Zeithalm, & Naylor, 2000). Ford, Smith, and Swasy (1988) characterize credence goods as those which for average consumers are mostly taken on trust (Asch, 2001). Credence qualities are primarily found in professional contexts, such as medical services and pension plans, because consumers do not usually have the knowledge to evaluate them (Asch 2001).

Shopping Preferences

Drawing from the previously published research on product typology and using Nelson's (1974) definition of search goods, Kline's (1988) two-fold classifications of experience goods, and Darby and Karni's (1973) definition of credence, the authors suggest that type of product (search, experience-1, experience-2, and credence) will influence consumers' purchase preferences as well as the importance they attach to Internet shopping-related attributes. More recently, Ford, Smith, and Swasy (1988) provided an operational definition of search, experience, and credence qualities in the context of a consumer's effort to verify advertising claims. Ford, Smith, and Swasy (1990) continued with the examination of differences in consumer skepticism for search, experience and credence advertising claims. They reported that consumers are more skeptical of experience attribute claims than search attribute claims and more skeptical of subjective claims than objective claims.

Based on the research findings about consumers' skepticism for search, experience and credence advertising claims, the authors of the present study speculate d that because of the differences in consumers' information needs for different product types, their preference for shopping online will vary across product categories. Particularly, given that the credence products are the hardest to evaluate even after purchase or consumption, consumers' desire to shop online for credence products may be lower than that their desire to shop for search or experience products. Similarly, the consumer's need to test or try out the experience products such as clothing and perfume (experience-1) or cellular phone and television (experience-2) will be higher than the need to experience search products such as books and personal computers. As a result, their desire to shop online for search products will be greater than for experience products. Based on Kline's (1988) classifications of two types of experience goods (experience-1 and experience-2), information search for experience-2 products is more costly and difficult than for experience-1 products. As mentioned earlier, experience-1 products necessitate direct experience compared to experience-2 products. Therefore, the authors speculate that consumers will be more likely to shop online for experience-1 products than experience-2 products. Generally, the type of product is expected to influence consumers' preferences for shopping with an Internet retailer significantly.

  More specifically,

    H1: Consumers' willingness to shop from an Internet retailer for search products will be significantly greater than their willingness to shop for experience-1 products.

    H2: Consumers' willingness to shop from an Internet retailer for experience-1 products will be significantly greater than their willingness to shop for, experience-2 products.

    H3: Consumers willingness to shop from an Internet retailer for experience-2 products will be significantly greater than their willingness to shop for credence products.

Internet Retailer Attributes

Several published studies in retailing as well as non-store retailing suggest that consumers' patronage decisions are influenced by the importance of store or non-store attributes such as perceived value and quality of products, responsiveness, convenience, company reputation, customer service, information and order services provided, merchandise assortment, salesperson interaction, shopping from home, and economic utility (Eastlick & Feinberg, 1999; Hansen & Deutscher, 1977-78). The importance of attributes in general is well-established in retailing as well as non-retailing in the context of catalog shopping. Eastlick and Feinberg (1999) investigated consumers' functional and nonfunctional shopping motivations in the context of print-catalog shopping, using sporting goods as the moderate purchase frequency product and clothing as the high purchase frequency product. They suggested perceived value, order services, and convenience as functional motives, and company responsiveness and reputation as nonfunctional motives influencing consumers' preference for catalog shopping. The findings of various studies suggest that larger merchandise assortment (Reynolds, 1974), lower prices (Korgaonkar, 1984; Reynolds, 1974), unique merchandise offerings and convenience (Gillet, 1970; Jasper & Lan, 1992; Korgaonkar, 1984) are the significant motives for catalog shopping. Consistent with the Bellenger and Korgaonkar (1980) classification of recreational shoppers, Westbrook and Black (1985) identified the most prominent motives of highly involved shoppers in the context of retail stores to be economic role enactment, choice optimization, negotiation, affiliation, and sensory stimulation.

In line with the previously published research in retailing, researchers have attempted to assess the importance of various e-tailer attributes, often with mixed and inconclusive results. For instance, Jarvenpaa and Todd (1996-1997) suggested that the most important perceived benefit of Internet shopping was convenience, while poor customer service, poorly designed Websites, and perceived risk were cited by online shoppers as negative factors. Their findings have suggested that consumers' shopping experiences on the Internet were both enjoyable and frustrating. Consumers acknowledged the savings of time and effort compared to traditional shopping, but were not satisfied with online customer service. Further, consumers perceived goods and services on the Internet to be intangible and involve risk.

Szymanski and Hise (2000) investigated the role of online convenience, merchandising (product offerings and product information), site design, and transaction security on consumers' satisfaction online. They found that convenience, product information, site design, and transaction security had a statistically significant influence on satisfaction with online shopping. Keeney (1999) studied the positive and negative aspects of Internet shopping experiences, and concluded that different customers would have different priorities for Internet shopping. Bakos (1997) asserted that the Internet lowers the search cost to acquire information about seller prices and product offerings, and reduces inefficiencies caused by the buyer's search cost. Phau and Poon (2000) found that consumers were more likely to purchase, via the Internet, products and services that have a low outlay, are frequently purchased, have intangible value proposition, and are relatively high on differentiation. Vijayasarathy and Jones (2000) examined the factors that affected consumers' attitudes and intentions to shop using print and Internet catalogs. They found that consumers thought that differences between Internet and print catalog media had to do with differences in reliability, tangibility, and consumer risk. Further, they suggested that factors such as product value, pre-order information, post-selection information, shopping experience, and risk to consumers influenced attitudes and intentions to shop using print and Internet catalogs.

Consumers' online shopping behavior and its characteristics still remain a conceptual domain that demands attention. Vellido, Lisboa, and Meehan (2000) proposed a framework to characterize Internet users' opinions on Web vendors and on-line shopping. They confirmed that consumer risk perception is the main discriminator between Internet shoppers and Internet non-shoppers. They further reported that variables such as age, household income, and Web-usage patterns do not predict Internet purchasing behavior. However, Donthu and Garcia (1999) foundd that Internet shoppers were older and earned higher income than Internet non-shoppers. Moreover, Li, Kuo and Russell (1999) found that education, convenience orientation, experience orientation, channel knowledge, perceived distribution utility, and perceived accessibility are strong predictors of online buying status such as frequent online buyer, occasional online buyer, or non-online buyer.

Rowley (1996) articulated the challenges facing the Internet retailer and shopper. The challenges include locating shops on the Internet, time involved in comparison shopping, security related to financial transactions, the customer base and profile, the nature of the shopping experience, and legal or marketplace control or lack thereof. Rowley pointed out that the Internet has not yet accommodated to the cultural and social issues associated with shopping. Reichheld and Schefter (2000) argued that price is not an important factor for customer loyalty for e-tailers, but trust is the determining factor that is built based on "the delivery of a consistently superior customer experience."

What has not been investigated extensively is the role that product classification plays in determining the importance of the Internet retailers' attributes. The research findings by Lynch, Kent and Srinivasan (2001) indicated that impact of Internet retailers' attributes such as trust, affect (entertainment), and site quality vary across different product categories. Given the lack of "physical exposure and contact," Lynch, Kent and Srinivasan (2001) pointed out that trust as an attribute may affect shoppers' willingness to purchase online. More specifically, they anticipated that "trust may be more important in online buying of high-touch (experience) products" (p.17). The results of their study also indicated that site quality explains loyalty or purchase intentions for high-touch goods such as t-shirts (experience products) but not for low-touch goods such as CD players (search products).

Thus, drawing from the literature on consumers' motivations towards store and non-store attributes and their preference for type of products, the authors speculate that product categories will have a significant impact on the importance consumers attach to Internet retailers' attributes, Similarly, consumers' emphasis on Internet retailer attributes will significantly influence their preferences for purchasing online across different product categories. The hypothesized importance of different Internet retailer's attributes and product categories is exhibited in Table 1.

    More specifically

    H4: For search products, attributes such as convenience, home shopping, order services, and economic utility will be more important than the other Internet retailer attributes.

    H5: For experience, products Internet attributes such as merchandise assortment, customer service, and perceived value/ responsiveness will be more important than the other Internet retailer attributes.

    H6: For credence products, Internet attributes such as reputation, customer service, perceived value/responsiveness, and information services will be more important than the other Internet retailer attributes.

    H7: Security/privacy will be equally important across the different product categories.

    H8: The importance consumers assign to Internet retailer attribute will significantly influence their preference for purchasing online across different product categories.

Search	Experience	Credence
Convenience Home Shopping Order Services Economic Utility Security/Privacy	Merchandise Assortment Customer Service Perceived Value/ Internet Retailer Responsiveness Security/Privacy	Internet Retailer Reputation Customer Service Perceived Value/ Internet Retailer Responsiveness Information Services Security/Privacy

Table 1. Hypothesized importance of Internet retailer attributes across product categories.

Method

The first product class description was for a search product/service whose relevant attribute information could be easily obtained prior to use or purchase. They would be confident of making the purchase decision without using/sampling the product/service prior to its use or purchase.

The second product class description was for an experience-1 product/service whose relevant attribute information could not be known until the product was used. They would not be confident of making the purchase decision without using/sampling the product/service prior to its purchase.

The third product class description was for an experience-2 product/service for which it was more difficult/costly to get the relevant attribute information than actual product/service experience prior to its purchase. They would not be confident of making the purchase decision without using/sampling the product/service prior to its purchase.

Finally, the fourth product class description was for a credence product/service for which relevant attribute information was not available prior to as well as after the use of the product/service. They would not be confident of their purchase decision even after using/sampling the product/service.

A total of 64 products were listed by the respondents under the search product category, 70 products were listed under the experience-1 category, 87 products were listed under the experience-2 category, and 69 products were listed under the credence product category. Based on the examination of the product list provided by the respondents, the authors selected the two most representative products for each product class (see Table 2) to be used in the second stage of the study. We also focused on those products that are more readily available on the Internet so as to make the second stage of the study more relevant to the respondents.

Second Stage Sample

In the second phase, the data for the study was collected in two major metropolitan areas with a population of 3.9 million residents in the Southeastern United States. The respondents were contacted during different days of the week and different times of the day. The survey was administered only to those who were at least 18 years of age and had used the Internet regularly in the past. The demographic profile of the respondents is shown in Table 3 & 4. In comparing our sample to the Census 2000 of the local area we found that our sample was skewed towards the more highly educated and higher income respondent, and to skilled as well as managerial types of occupations. This was expected as we surveyed only those who were regular users of the Internet. However, the sample profile is similar to the national profile of the Internet users as reflected in the last GVU survey.

Demographic Characteristics	Search Products	Experience-1 Products	Experience-2 Products	Credence Products	Chi-Sq. Statistics (sig. level)
Gender   Male   Female	58 73	54 86	70 84	62 68	2.530 (0.470
Occupation   Unskilled   Office/Sales   Skilled   Supervisory Non-Technical   Supervisory Technical   Professional   Managerial	31 35 21 2 7 16 5	38 30 29 5 6 10 4	52 26 31 5 4 15 3	49 24 16 2 2 16 4	21.076 (0.276)
Education   Attended high school   High school graduate   Attended trade school   Trade school graduate   Attended college   College graduate   Postgraduate/professional	4 6 0 5 68 31 17	6 15 2 4 70 28 15	3 16 2 4 76 39 14	4 12 1 1 80 29 4	20.621 (0.299)
Household Income   Under 20K   20-40K   40,001-60K   60,001-80K   80,001-100K   Over 100K	31 29 30 18 11 13	29 39 33 19 10 9	24 47 32 16 13 19	37 34 21 15 11 10	14.076 (0.520)

Table 4. Sample demographic characteristics (continued). The demographic profile of sample respondents of four product categories
(actual count).

Second Stage Survey Instrument

The survey instrument had four sections. In section one, we asked each respondent to indicate his or her preference for purchasing from the Internet each of the eight products selected from stage one of the study. The preference for purchasing each product from an Internet Retailer was measured on a five point scale of (1) "May Never Buy" to (5) "May Prefer Buying" for each of the selected eight products. In the second section, we asked the respondents to rate how important each Internet Retailer attribute would be for purchasing the two products from one of the four product classifications on a scale of (1) "Not Important at All" to (5) "Extremely Important." For example, one group of respondents was asked hoe, assuming they wanted to purchase a product such as a book or personal computers (search) from an Internet retailer, they would rate the importance of various features in choosing a specific Internet retailer to purchase from. The subjects were given a list of 50 statements designed to capture 11 different dimensions of internet retailers. The eleven dimensions that fifty statements measured were Perceived Value (6 items); Convenience (6 items); Economic Utility (6 items); Home Shopping (3 items); Merchandise Assortment (4 items); Order Services (4 items); Company Clientele (4 items); Information Services (7 items); Customer Service (4 items); Security/Privacy (3 items), and Internet Retailer Reputation (3 items). The items were chosen from an exhaustive search of the literature in the area of Internet retailing as well as direct marketing (e.g., Eastlick & Feinberg, 1999; Vijaysarathy & Jones, 2000). A separate set of respondents was asked for the same information but assuming they were purchasing products such as clothing and perfume (experience-1). A third group of respondents was asked for the same information but assuming they were purchasing products such as cellular phones and televisions (experience-2). The fourth group of respondents was asked for the same information assuming they were purchasing products such as vitamins and water purifiers (credence). The administration of the survey instruments was randomized to prevent a response bias. There were no statistically significant differences in the demographic profiles of the four groups of respondents. A total of 559 valid surveys was obtained. The breakdown of the sample size in each product category is as follows: 132 for books and personal computers, 142 for clothing and perfume, 153 for cellular phone and television, and 131 for vitamins and water purifier. The third section of the questionnaire pertained to a variety of shopping orientation statements measured on a Likert scale of (1) "Strongly Agree" to (5) "Strongly Disagree." Finally, the fourth section pertained to demographics.

Analysis

A first step in the analysis was aimed at ensuring that the survey instrument captured all the attributes of Internet retailer. Hence, Principal Component Analysis with Varimax rotation was performed on the fifty importance of Internet retailer attribute items to examine their discriminant and convergent validity. The analysis produced a clean factor structure with items loading on the appropriate components Table 5. Ten dimensions were obtained with Eigenvalues greater than 1, and 66 percent of the cumulative variance was explained. Only seven items did not load on the underlying dimensions. Among those, two items, "The Internet retailer is well known" and "The Internet retailer is in business for a long time," that were originally expected to measure IR Reputation loaded on Company Clientele. These two Internet Retailer (IR) Reputation and Company Clientele items were combined under IR Reputation because they seemed more suitable to measure that component. The items, "The Internet retailer allows me to comparison shop" and "Third party evaluations about the Internet retailer's business practices are easily available," loaded on Information Service rather than Economic Utility and Security/Privacy, respectively. These two items were retained where they loaded because they seemed to be relevant to the respondents' information search; therefore, they were suitable items to measure the Information Service component.

The items, "The Internet retailer provides information" and "The Internet retailer Web site provides close-up product images," loaded on both Security/Privacy (slightly higher) and Information Service. Since the two items did not measure either component rigorously, they were eliminated. The item, "Using the Internet retailer, I feel like a careful shopper," loaded on Merchandise Assortment rather than Economic Utility. The term "careful" may have caused ambiguous interpretation among the respondents. Therefore, it was eliminated. A total of three items out of fifty were deleted. The rest of the items that loaded on the appropriate components produced ten dimensions with high Chronbach Alphas in the range of 0.76 through 0.89. Thus, our analysis confirmed the presence of 10 attributes labeled as follows: 1) Perceived Value, 2) Internet Retailer Reputation, 3) Convenience, 4) Information Services, 5) Customer Service, 6) Security/Privacy, 7) Home Shopping, 8) Order Services, 9) Merchandise Assortment, and 10) Economic Utility. A scale for each attribute was created by summing up the responses to the items loading on the corresponding factor.

Items	Perceived Value/ Company Responsiveness α = .89	Company Clientele/ IR Reputation α = .85	Convenience α = .88	Info. Service α = .84	Cust. Service α = .85	Security /Privacy α = .80	Home Shopping α = .88	Order Services α = .80	Merch. Assort. α = .76	Econ. Utility α = .77
Return for credit Easy exchange Dependable products Stands behind Value for money Quality products	.787 .778 .731 .730 .663 .655
Friends like IR Friends know Friends recommend IR well-known Long-time People like me shop		.865 .864 .864 .612 .501 .472
Saves time Saves effort Allows shop whenever Find what I want Downloads fast Easy to surf			.814 .807 .689 .509 .469 .464
Comparison guides Product availability Search function Reviews/evaluation Comparison shop Business practices Trace Close-up Provides info				.640 .569 .541 .535 .456 .436 .429 .343* .462*		(.529) (.496)
Access to a salesperson Responds quickly Talk with salesperson Knowledgeable					.793 .755 .752 .654
Keeps info. confidential Keeps promises Safe to buy						.666 .627 .625
Privacy of home Safety of home Comfort of home							.805 .770 .753
Place orders Allows credit card Cancel orders 1-800 freecall								.721 .713 .558 .557
Many brands Wide selection Latest styles Rare products Careful shopper*									.731 .632 .613 .497 .439*
Free of sales tax Free shipping Real bargains Competitive prices										.689 .618 .565 .498

Table 5: Factor analysis of the Internet retailer attribute ratings

Results

Hypothesis 4 through 7

To test hypotheses 4 through 7, ten separate one-way Analyses of Variance were performed in which product category was the independent variable and the summated score of each importance of Internet retailer attribute component was the dependent variable. Additionally, mean values of each of the 10 Internet retailer attributes for each product class were compared to determine the rank order order of importance for the Internet retailer attributes for each of the product categories.

The results of the ten separate Analyses of Variance to test hypotheses 4 through 7 indicated that product category produced significant differences in two Internet retailer attributes on a summated scale. The two components were Perceived Value (F = 2.87, p < 0.05) and Merchandise Assortment (F = 2.561, p < 0.05). The comparison of the mean values of each Internet retailer attribute importance scores across all product types indicated that Perceived Value was the most important attribute followed by Information Services, followed by Convenience and then IR Reputation Table 7. These attributes were followed by Order Services, Economic Utility, Customer Service, Merchandise Assortment, Security/Privacy, and Shopping from Home. For Search and Credence product types, Information Services had the highest importance, followed by Perceived Value. However, for the Experience-1 and Experience-2 product classes, Perceived Value had the highest importance followed by Information Services. Convenience, IR Reputation, and Order Services ranked third, forth, and fifth respectively for all product types. The remaining rankings of relative importance are shown in Table 7.

Internet retailer attributes Convenience, Order Services, and Economic Utility for the search products, Perceived Value/ Responsiveness, Merchandise Assortment, and Customer Service for the experience-1 and experience-2 products, and Information Services, Perceived Value/Responsiveness, Reputation, and Customer Service for credence products were rated as important by the respondents. Therefore, hypotheses 4, 5, and 6 were partially supported. Because Security/Privacy was an equally important attribute across all product categories, hypothesis 7 was also supported.

	Search	Rank Order	Exp.-1	Rank Order	Exp.-2	Rank Order	Credence	Rank Order	Sig.
Perceived Value	26.42	2	26.93	1	26.86	1	25.60	1	0.04**
Info. Services	26.56	1	26.21	2	26.84	2	25.87	3	0.39
Convenience	24.57	3	25.17	3	25.03	3	24.17	4	0.24
IR Reputation	19.98	4	20.60	4	20.47	4	19.38	4	0.23
Order Services	16.61	5	16.67	5	16.37	5	15.85	5	0.15
Economic Utility	15.64	6	15.67	6	15.67	6	14.98	6	0.23
Customer Service	15.15	7	15.16	7	15.57	7	14.78	7	0.36
Merchandise Assort.	14.95	8	15.27	8	15.82	8	14.94	8	0.05**
Security/Privacy	13.44	9	13.56	9	13.50	9	13.03	9	0.18
Home Shopping	11.33	10	11.62	10	11.48	10	10.93	10	0.26

Table 7. Importance ratings of ten Internet retailer attribute components by product categories.
** Significant at 0.05 level based on the ten ANOVA tests.

Hypothesis 8

Hypothesis 8 was tested through a two-tailed correlation analysis to determine whether significant associations or correlations exist between the importance of Internet retailer attribute components and online purchase preference for each product class.

The results of the two-tailed correlation analysis indicated that product purchase preference from an Internet retailer had significant associations with Internet retailer attributes table 1. Consumers' online purchase preference for a search product category had significant correlations with Perceived Value, Convenience, Information Services, Customer Service, Home Shopping, Order Services, Merchandise Assortment, and Economic Utility at the 0.01 level, and IR Reputation at 0.05 level. Privacy/Security did not have a significant association with online purchase preference for the search products.

With respect to consumers' online purchase preference for the experience-1 product class, Perceived Value, Information Services, Order Services, Merchandise Assortment and Economic Utility had a significant influence at (p < 0.01), and Convenience, Privacy/Security, and Home Shopping at the 0.05 level. IR Reputation and Customer Service did not have a significant association with the online purchase preference for experience-1 products.

For experience-2 products, IR Reputation, Home Shopping, and Order Services had a significant influence on online purchase preference at the 0.05 level, while Customer Service and Merchandise Assortment were significant at the 0.10 level. Perceived Value, Convenience, Information Services, Privacy/Security, and Economic Utility had no significant effects on online purchase preference for the experience-2 product class.

For the credence products, IR Reputation, Convenience, Information Services, Home Shopping, Order Services, Merchandise Assortment and Economic Utility had a significant influence on online purchase preference at the 0.05 level, and Privacy/Security at the 0.10 level. Perceived Value and Customer Service had no significant effects on online purchase preference for credence products.

In order to ensure that the results were not an artifact of the demographic variables of gender, occupation, education, and income the results were confirmed by running correlations while partialling out the effects of the four demographic variables. The results remain almost identical except for change in the significance of six correlations out of a total of 40. Those changes were primarily the artifact of lower sample sizes as the missing values for the four demographic variables lowered the sample size without changing the correlation coefficients in substance. Thus, the authors found support for hypothesis 8 without the moderating effects of the demographic variables.

Components	Search N=132	Experience-1 N=142	Experience-2 N=142	Credence N=131
Perceived Value	0.202a (0.01)*	0.251 (0.01)*	-0.027 (0.185)	0.043 (0.16)
IR Reputation	0.115 (0.05)**	0.065 (0.11)	-0.119 (0.04)**	0.142 (0.03)**
Convenience	0.219 (0.01)*	0.110 (0.05)**	0.007 (0.23)	0.156 (0.02)**
Info. Services	0.275 (0.01)*	0.182 (0.01)*	0.050 (0.13)	0.116 (0.05)**
Customer Services	0.297 (0.01)*	0.002 (0.26)	0.082 (0.08)***	0.006 (0.24)
Privacy/ Security	0.070 (0.11)	0.124 (0.05)**	0.051 (0.14)	0.085 (0.084)***
Home Shopping	0.176 (0.01)*	0.160 (0.014)**	0.107 (0.05)**	0.118 (0.05)**
Order Services	0.240 (0.01)*	0.256 (0.01)*	0.140 (0.02)**	0.138 (0.03)**
Merchandise Assortment	0.222 (0.01)*	0.250 (0.01)*	0.099 (0.056)***	0.121 (0.05)**
Economic Utility	0.335 (0.01)*	0.192 (0.01)*	0.049 (0.14)	0.144 (0.03)**

Table 8: Correlation coefficients of product preference and importance ratings of ten components across product categories.
Pearson correlation coefficients
*Significant at 0.01 level; ** Significant at 0.05 level; *** Significant at 0.10 level

Discussion

Conclusion

References

About the Authors