The influence of virtual direct experience (VDE) on on-line ad message effectiveness.

The World Wide Web has created a new communication environment for advertising campaigns, thus initiating a new era of firm-consumer interaction (Rust and Oliver 1994). Firms use advertising messages and direct experience (DE) as two common sources of information to communicate with consumers

about products (Singh, Balasubramanian, and Chakraborty 2000). These two sources of information differ significantly in their ability to foster strongly held beliefs about search and experiential product attributes. Advertising has been found to be superior at communicating search attribute beliefs and DE has been found to be superior at fostering experiential attribute beliefs (Kempf and Laczniak 2001).

While advertising messages are capable of being easily communicated across a broad spectrum of media, DE has been somewhat limited, particularly in relation to experience products--products that are dominated by attributes that cannot be known until (limited) use of the product takes place (Klein 1998; Nelson 1974; Wright and Lynch 1995). This has been a substantial limitation for firms, as consumers use DE and advertising messages in a complementary fashion (Kempf and Laczniak 2001).

The Internet offers firms the unique opportunity to digitalize experiential attributes in multimedia formats (Alba et al. 1997; Burke 1997; Hoffman and Novak 1996). The Internet has expanded consumer access to information and provided firms an opportunity to provide consumers additional layers of information (Rust and Oliver 1994). One of the key advantages of on-line advertising over traditional advertising is that it can proximate key characteristics of DE when promoting experience products.

Although researchers and organizations differ in their estimates, most agree that on-line advertising is developing rapidly along with changes in the way people communicate and conduct business (Steven and Gangadharbatla 2001). The present study, accordingly, explores the integration of digitalized direct experience, that is, virtual direct experience (VDE), into on-line advertising. This issue is important for both academics and practitioners in that the ability to integrate the delivery of experiential product attributes into an advertising context suggests unique theoretical and performance implications. The message effectiveness of on-line ads, varying in VDE related to the degree of digitalization of experiential attributes, and the moderating effect of consumer product expertise are thus investigated.

PRIOR RESEARCH AND CONCEPTUAL FRAMEWORK

Prior research indicates that advertisements and direct experience work in a complementary fashion (Deighton and Schindler 1988; Hoch and Ha 1986; Kempf and Smith 1998). Furthermore, research indicates that for experience products, direct experience influences recall, attitudes, and conation to a greater degree than do traditional advertisements (Singh, Balasubramanian, and Chakraborty 2000). The conceptual framework developed in this research is anchored by both of these insights.

Direct Experience (DE)

Direct experience (DE, i.e., product trial) has received considerable attention as a result of its effectiveness in stimulating positive consumer responses for experience products (e.g., Kempf and Smith 1998). DE is defined as a consumer's first usage experience with a product (Singh, Balasubramanian, and Chakraborty 2000). Previous studies of the effects of DE indicate that it influences brand beliefs, attitudes, and purchase intentions (e.g., Kempf and Smith 1998; Smith 1993; Smith and Swinyard 1982; Wright and Lynch 1995). The effectiveness of DE is derived from its ability to allow consumers firsthand evaluation of product claims, thus increasing consumers' confidence in their evaluations. However, research has yet to determine whether the superiority of DE in a traditional ad format in terms of facilitating consumers' evaluations, attitudes, and conations also holds true in the on-line venue, where the direct experience is simulated in a virtual direct experience (i.e., a simulated product experience in an on-line environment).

The conceptual framework presented in this study pertains only to experience products (as evaluative attributes for search products are as fully accessible in the on-line advertising domain as they are in the traditional advertising domain). Experiential product attributes are attributes that can be accessed only through (limited) use of the product (Kempf and Smith 1998; Klein 1998; Nelson 1974; Wright and Lynch 1995). Products that are dominated by experiential attributes are referred to as experience products and are best evaluated by consumers through firsthand experience (Kempf and Smith 1998; Klein 1998; Nelson 1974; Smith and Swinyard 1982; Wright and Lynch 1995).

Virtual Direct Experience (VDE)

The Internet is fundamentally different from traditional media in that it provides for two-way interactivity (Cho 1999; Hoffman and Novak 1996). Interactivity and multimedia displays help to enhance consumer learning (Novak, Hoffman, and Yung 2000). Meeker (1997) describes the Internet as the only medium that allows consumers to interact with products, investigate further details, and immediately make purchases. For example, many on-line retailers provide interactivity through on-line product experiences (e.g., allowing a consumer to change clothes on a virtual model, rotate products), thus integrating the conveyance of experiential product attributes into on-line advertising.

Given the potential importance of on-line product experience in general, and on-line advertising in particular, the focus of this study is to investigate the effect of virtual direct experience (VDE). Virtual denotes concepts, activities, and organizations that are realized or carried out chiefly in an electronic medium, and tends to be used in reference to things that imitate their real-life equivalents. Thus, VDE is formally defined as the conveyance of experiential product attributes in an on-line simulation of a direct experience. It is important to note that similar to DE, VDE is employed to convey experiential product attributes to stimulate message effectiveness. Thus, similar to DE, VDE attempts to convey experiential product attributes. VDE only mimics DE, however, and is simulated by the computer via the Internet, that is, VDE is direct experience that is mediated by a virtual communication channel. Thus, the greatest difference between VDE and DE resides in the simulated conveyance of experiential product attributes. Hence, while VDE can approximate DE, its ability to provide an exact replication of a DE experience is influenced by the ability of the product's experiential attributes to be digitized and conveyed on-line. As such, VDE and DE diverge in effectiveness to the greatest extent when a product's complement of experiential attributes cannot be fully conveyed through the given channel.

Interaction Between Advertising and Direct Experience

Several researchers have studied the interaction between advertising and DE (Deighton 1984; Ha and Hoch 1989; Hoch and Ha 1986). These studies imply a two-stage model (as later summarized in Singh, Balasubramanian, and Chakraborty 2000). In the first stage, consumers exposed to an ad "may not believe the ad claims (because of the partisan nature of the source) but retain the claims as tentative hypotheses" (Singh, Balasubramanian, and Chakraborty 2000). In the second stage, consumers test the hypotheses based on the evidence about the advertised brand obtained subsequently from a DE. In the testing process, consumers are likely to show a bias toward confirming their hypotheses, particularly if the evidence is ambiguous and does not contradict the claims in the ad. Thus, beliefs and attitudes about an advertised brand are found to shift in the direction of the hypotheses after an ambiguous product experience (Hoch and Ha 1986). Subsequent research by Smith (1993) extended support for the two-stage model (Ha and Hoch 1989; Hoch and Ha 1986), even in situations that involve positive (and therefore unambiguous) experience outcomes. Consider an ad message and a positive trial experience denoted as "ad" and "+DE," respectively. Smith's experiment found no significant differences in belief strength, belief confidence, and total expectancy within each of the following two comparison sets: +DE and ad//+DE; +DE and +DE/lad (note that the sign "//" means "followed by" here).

VDE: Degree of Digitalization

As mentioned earlier, VDE is similar to DE in that it conveys experiential product attributes; it differs from DE, however, in that its conveyance of experiential product attributes is mediated by the ability to digitize experiential product attributes for on-line conveyance. Current technology facilitates conveyance of some experiential attributes (e.g., sound), whereas the conveyance of other experiential attributes (e.g., taste and touch) is constrained. In some cases, all of a product's experiential attributes are capable of being fully digitized (e.g., music), whereas in others (e.g., apparel) only a portion of the product's overall experiential attributes are capable of being digitized. Furthermore, firms currently differ in the degree of digitalization of experiential attributes within the same product category, that is, some businesses choose to fully convey experiential attributes capable of being digitized, whereas other firms choose to only convey a portion of the product's experiential attributes, or choose not to convey experiential attributes on-line at all. The fundamental rationale for differences in VDE employment across firms may relate to the fact that there are still questions about whether VDE can provide similar response outcomes as DE, especially given its costs. This research aims to address these questions by examining (1) whether, when integrated into on-line ads, VDE for an experience product with a high degree of experiential product attributes capable of being digitized will be more effective than VDE for an experience product with a low degree of experiential product attributes capable of being digitized, and (2) whether the degree of digitalization of VDE for a single experience product category will influence the effectiveness of on-line ads.

EXPERIMENT 1

Hypotheses

Direct experience provides firsthand experience with a product's experiential attributes, increasing a consumer's knowledge structures relating to the product (Smith and Swinyard 1982; Wright and Lynch 1995). The development of knowledge structures decreases perceived risk (Alba and Hutchinson 1987; Moorthy, Ratchford, and Talukdar 1997) and increases a consumer's ability to evaluate the product (Alba and Hutchinson 1987; Marks and Olson 1981; Rao and Monroe 1988). Research also indicates direct links between DE and a positive affective response (Kempf and Smith 1998; Mano and Oliver 1993), and between affect and conation (Fishbein 1967; Sheppard, Hartwick, and Warshaw 1988).

Previous researchers have compared the effectiveness of different types of advertising (e.g., print, television, and radio) with DE (Deighton and Schindler 1988; Hoch and Deighton 1989; Smith and Swinyard 1982). These studies suggest that DE has a greater influence on recall, attitudes, and conation when compared with existing ad formats. Extending this insight to the on-line environment, it can be argued that introducing VDE might enhance the effectiveness of an on-line ad message, that is, VDE will increase a consumer's knowledge structures relating to the product. As mentioned earlier, VDE, as a rich presentation on-line, is more effective than a less rich presentation, such as on-line advertising based on text information alone. Consistent with prior research (e.g., Smith and Swinyard 1982), the focus of the effectiveness of the ad message in this study is restricted to the following four message effectiveness dimensions: perceived risk, evaluation, affect, and conation. Therefore, extending prior research, we hypothesize the following:

Hl: Compared with on-line ads without VDE, on-line ads with VDE will stimulate:

a. Lower perceived risk relating to the purchase of the product

b. Higher evaluation of the product

c. Higher affect

d. Higher conation

As indicated previously, VDE requires conversion of experiential attributes to a suitable format for computer processing. The ability to convey experiential attributes through the Internet is dependent on the technology available to digitize (i.e., reduce to bytes) experiential attributes. As such, two products can vary greatly in the degree of digitizable experiential attributes that can be integrated into on-line advertising via VDE. When all, or a large portion, of an experience product's experiential attributes can be conveyed to consumers on-line via VDE, consumer knowledge structures increase, thus influencing ad message effectiveness (i.e., reduced perceived risk, increased attitude, product evaluation, and conation). Alternatively, when only a small portion of an experience product's experiential attributes can be digitized and conveyed on-line, only a partial VDE can be conveyed, thus limiting overall ad message effectiveness. Therefore, we propose the second hypothesis:

   H2: Compared with on-line ads with VDE for experience
   products with a low degree of digitalization of experiential
   attributes, on-line ads with VDE for experience products with
   a high degree of digitalization of experiential attributes will
   stimulate:

   a. Lower perceived risk relating to the purchase of the product

   b. Higher evaluation of the product

   c. Higher affect

   d. Higher conation

Participants

To test the hypotheses, 111 undergraduate students in a major state university were recruited from a marketing department subject pool. The sample consisted of 35 female and 76 male students. Eighty-eight percent of the subjects were between the ages of 20 and 25, with the balance between the ages of 26 and 35. Approximately 26% of subjects had purchased products on-line. Subjects averaged 6.17 hours on the World Wide Web per week.

Experimental Design and Stimuli

Two principal considerations guided the selection of the products used in the on-line ads in this experiment: (1) the products should vary in the proportion of digitizable experiential attributes, with one having a high proportion of digitizable experiential attributes ([D.sub.H]) and a second one having a lower proportion of digitizable experiential attributes ([D.sub.L]), and (2) the products should be appropriate for use as stimuli for the intended subject pool (i.e., students).

To facilitate the selection of the products, a committee of on-line experts (i.e., academics and practitioners involved in e-commerce) generated a list of experience products varying in proportion of digitizable experiential attributes. Employing this list, five products with a high proportion of digitizable experiential attributes (music, newspapers, movies, books, multimedia library) and five products with a low proportion of digitizable experiential attributes (apparel, sunglasses, athletic shoes, watches, and food) were pretested for use in this experiment. Pretest subjects (n = 34) (who were similar to the subjects used in the final experiment in terms of their demographic profile) rated movies and music the highest in involvement and product category knowledge among the products [D.sub.H]. Sunglasses and apparel were rated highest among the products [D.sub.L]. A second pretest (n = 28) was then conducted to assess the perceived degree of experiential attributes of both movies and sunglasses (see Perdue and Summers 1986). Murray's (1991) three-item seven-point scale was used

([D.sub.H][alpha] = .67, [D.sub.L][alpha] = .65). No differences were observed across products ([D.sub.H][bar.X] = 5.25, [D.sub.L][bar.X] = 5.33, df = 28, t = .50, p = .620). Thus, movies ([D.sub.H]) and sunglasses ([D.sub.L]) were deemed appropriate categories to explore.

The free-elicitation task recommended by Fishbein and Ajzen (1975) was then used to identify salient attributes for each product category (n = 28). Subjects were asked to list all the attributes that would be of importance to them when making a purchase decision. The results indicated that the product categories suggested a diversity of salient attributes with the primary attribute for movies being type (action, drama, etc., mentioned by 93% of subjects), and purpose (athletic, daily wear, etc., mentioned by 78% of subjects) for sunglasses. Other movie experiential attributes consisted of story line (86%), cast (64%), and so forth. Additional salient attributes identified for sunglasses were darkness (71%), comfort of frames (46%), UV protection (32%), and so forth. These attributes were therefore employed in the subsequent design of the treatments.

To enhance comparability across product categories, care was taken to maintain consistency in the presentation of the experiential attributes of each product across on-line promotion sites. Individual on-line promotion sites, identical in terms of structure, were created for movies and sunglasses. Each on-line promotion site consisted of a homepage including a fictitious company name (i.e., Native Star Pictures for the movie category; Star Light Sunglasses for the sunglasses category), an inlaid graphical image (movie posters and models wearing the sunglasses) and logo, as well as two text links to individual product promotion pages (i.e., each on-line promotion site contained ads for two separate products). Based on the pretest results (i.e., the primary salient attribute), the on-line promotion movie site contained links to a drama movie ad and an action movie ad, while the sunglasses site had links to active and daily wear sunglasses ads. Each individual product page (two movie: action/drama; two sunglasses: active/ daily wear) included a link to an advertisement providing a 50-word product description. Each 50-word product description provided product arguments relating to the three primary salient attributes for each product category determined during pretesting, as well as a photo of the product. Pretesting (n = 42) of stimuli for content equivalence indicated no significant differences in perceived product quality, employing Dodds, Monroe, and Grewals's (1991) five-item seven-point scale, across product categories ([D.sub.H][bar.X] = 5.48, [D.sub.L][bar.X] = 5.53, df = 41, t = .51,p = .357) and within product categories in the on-line promotion ([D.sub.H1][bar.X] = 5.48, [D.sub.H2][bar.X] = 5.57, df = 40, t = .98, p = .331;[D.sub.L1][bar.X]= 5.50, [D.sub.L2][bar.X] = 5.45, df= 41, t = .51, p = .613; [D.sub.H][alpha] = .88, [D.sub.H][alpha] = .86, [D.sub.H1][alpha] = .78, [D.sub.H2][alpha] = .72, [D.sub.Ll][alpha] = .67, [D.sub.L2][alpha] = .68).

VDE stimuli were modified from preexisting ads for use in the experiment. A 35-second VDE was embedded in the on-line ad for each product. In the case of the movie ads, modified movie "trailers" were used. The movies selected were made by independent film studios to minimize preexposure bias (prior experience with each movie was measured during the experiment; no subjects were familiar with the movies). Two 35-second VDEs consisting of viewing objects "with" and "without" the sunglasses were generated by modifying an existing sunglass infomercial (prior experience with the infomercial was measured during the experiment; no subjects were familiar with the infomercial). To ensure comparability of VDE across products within and across products [D.sub.H] and [D.sub.L], modifications were made to each treatment condition until no differences were observed in pretesting (n = 42) in product quality, as measured by Dodds, Monroe, and Grewal's (1991) five-item seven-point scale ([D.sub.H][bar.X] = 5.62, [D.sub.L][bar.X] = 5.55, df = 41, t = .61, p = .548), and within product categories in the on-line promotion ([D.sub.H1][bar.X] = 5.64, [D.sub.H2][bar.X] = 5.61, df = 40, t = 1.14, p = .263;[D.sub.L1][bar.X] = 5.51,[D.sub.L2][bar.X] = 5.56, df = 41, t = .48, p = .633; [D.sub.H][alpha] = .84, [D.sub.H][alpha] = .81, [D.sub.H1][alpha] = .74, [D.sub.H2][alpha] = .75, [D.sub.L1][alpha] = .70, [D.sub.L2][alpha] = .71), or argument strength, as measured using Beltramini and Evans's (1985) ten-item semantic differential advertising believability scale ([D.sub.H][bar.X]= 5.55, [D.sub.L][bar.X] = 5.53, df = 41, t = .20, p = .844), and within product categories in the on-line promotion ([D.sub.H1][bar.X] = 5.56, [D.sub.H2][bar.X] = 5.51, df = 41, t = .39, p = .701;[D.sub.L1][bar.X] = 5.52,[D.sub.L2][bar.X] = 5.57, df = 41, t = .40, p = .694; [D.sub.H][alpha] = .83, [D.sub.H][alpha] = .79, [D.sub.H1][alpha] = .73, [D.sub.H2][alpha] = .70, [D.sub.L1][alpha] = .69, [D.sub.L2][alpha] = .71).

Finally, price is a critical factor in product evaluation (Rao and Monroe 1988) and in the determination of the probability of perceived risk (Bauer 1960). To maintain consistency in the level of perceived financial risk in the intended subject pool, identical product prices were established (i.e., all products were priced at $19.95). A three-item seven-point scale derived from Biswas and Burton (1993) was used to assess pricing appropriateness across products (D.sub.H][alpha] = .77, [D.sub.L][alpha] = .81). The scale assessed the appropriateness of the $19.95 price for each product. Pretesting (n = 42) indicated that pricing appropriateness was comparable across product categories (D.sub.H][bar.X] = 5.14, [D.sub.L][bar.X] = 4.90, df = 41,t = 1.53,p = .133).

Procedure

A 2 X 2 experiment with a between-subjects factor (VDE versus no VDE) and a within-subject factor (sunglasses-[D.sub.L] versus movie-[D.sub.H]) was employed. Subjects were randomly assigned to either the VDE (n = 56) or no VDE (n = 55) treatment condition, and were exposed to both product categories (order of product category exposure was randomized across groups).

To the extent that on-line ads have the capacity to combine the characteristics of both ad messages and VDEs, they are analogous to the combination of ad messages and +DE. Research suggests that the exposure sequence of these two parts (whether it is an ad message followed by positive direct experience or positive direct experience followed by ad message) does not materially affect performance outcomes (e.g., Smith 1993). Therefore, in the on-line environment, where the exposure sequence of ad message and VDE is likely to be controlled by the consumer and not the firm, exposure sequence of ad message and VDE should not influence the message effectiveness of on-line promotion. Hence, subjects in the VDE treatment condition were given freedom to explore the stimuli without an imposed exposure sequence.

Independent administrators read instructions from a script describing the procedures. Administrators indicated that subjects would be asked to evaluate their preferred products from two different on-line promotion sites. Subjects were first asked to complete a series of preexposure questions. Next, subjects were directed to the first on-line promotion site. After viewing the first on-line promotion site, they were asked to complete a portion of the survey pertaining to their preferred product (of the two products viewed). After completing the relevant survey portion, subjects were directed to the second on-line promotion site. After viewing the second on-line promotion site, they were asked to complete the remaining survey pertaining to their preferred product from the second on-line promotion site (of the two products viewed). A debriefing was conducted after all questionnaires were collected.

Measures

Perceived Risk

Perceived risk was captured using a five-item seven-point scale anchored by "not likely" to "very likely," similar to that in Murray and Schlacter (1990). This scale measured the probability of loss (financial, performance, physical, psychological, and social) that a subject perceived to be associated with the product purchase (Cunningham 1967) ([D.sub.H][alpha] = .80, [D.sub.L][alpha] = .86).

Evaluation

Product evaluation was assessed using a four-item seven-point semantic differential scale similar to the scale proposed by Petty, Cacioppo, and Schumann (1983). Subjects were asked to rate their overall impression of the product from bad/good, unsatisfactory/satisfactory, unfavorable/favorable, and not carefully produced/carefully produced ([D.sub.H][alpha] = .86, [D.sub.L][alpha] = .77).

Affect

Affect toward the product advertised was measured using Munch, Boiler, and Swasy's (1993) attitude scale. The scale consisted of two 7-point semantic differentials, from "I don't like it" to "I like it," and "negative" to "positive." The correlation for the attitude scale relating to the products in the [D.sub.H] and [D.sub.L] treatments were .95 and .96, respectively.

Conation

Conation was assessed using two, three-item seven-point scales (ranging from "not likely" to "very likely"), similar to that in Baker and Churchill (1977). Conation toward the product in the [D.sub.H] treatment (i.e., movie) consisted of the subject's intention to (1) buy the movie from the Web site, (2) buy the movie if he or she saw it in a store, and (3) actively seek out the movie in a store to purchase it (a = .73). Conation toward the product in the [D.sub.L] treatment (i.e., sunglasses) consisted of the subject's intention to (1) buy the sunglasses from the Web site, (2) buy the sunglasses if he or she saw them in a store, and (3) actively seek out the sunglasses in a store to purchase them ([alpha] = .69).

Results

Descriptive statistics are presented in Table 1. To test the hypotheses, first, the four dependent variables were entered into a multivariate analysis of variance (MANOVA) with VDE/ no VDE and [D.sub.H]/[D.sub.L] as the independent variables. As indicated in Table 2, results of the MANOVA indicate significant main effects for VDE (VDE/non-VDE: Wilks's)[lambda]. = .755, F = 16.77, df = 4/207,p < .001) and degree of digitalization ([D.sub.H]/[D.sub.L]: Wilks's [lambda] = .897, F = 5.91, df = 4/207, p < .001). The multivariate effect sizes for VDE and degree of digitalization ([n.sup.2] of .245 and .103, respectively) were large enough to be significant both practically and statistically.

Furthermore, the univariate F test indicates that the VDE main effect was attributable to all four dependent variables: perceived risk relating to the purchase of the product (Hla: F = 17.97, df = 1/210, p < .001), evaluation of the product (H1b: F = 16.50, df = 1/210, p < .001), affect (Hlc: F = 21.94, df= 1/210, p < .001), and conation (Hid: F = 21.25, df= 1/210, p < .001). The degree of digitalization main effect was attributable to perceived risk relating to the purchase of the product (H2a: F = 3.51, df = 1/210, p = .062), and conation (H2d: F = 13.99, df= 1/210, p < .001). The degree of digitalization, however, did not significantly influence affect (H1b: F = .003, df = 1/210, p = .958) or evaluation (H1c: F = 1.22, df = 1/210,p = .271).

Results presented in Table 2 further indicate that while VDE did stimulate key ad message effectiveness variables compared with on-line ads alone, differing response patterns across products were evident. Thus, whereas Hl was fully supported by the result, H2 was only partially supported in this experiment. However, a significant interaction effect was observed between the VDE/no VDE and [D.sub.H]/[D.sub.L] treatments (Wilks's)[lambda] = .847, F = 9.34, df = 4/207,p < .001,[[eta].sup.2] = .153).

Discussion

Results from Experiment 1 support the hypothesis that on-line ads employing VDE are more effective than on-line ads alone. Furthermore, the results indicate that there are differences between the effects of VDE based upon the proportion of digitizable experiential attributes. The inability of on-line ads for products with a lower degree of experiential attribute digitalization to influence ad message effectiveness may be directly related to the inability to digitize a large portion of the product's experiential attributes. This suggests that an on-line ad for products whose experiential attributes are easily digitizable could more fully utilize the Internet for experiential attribute conveyance. Several concerns, however, emerged from Experiment 1.

As two different products were used to explore differences in the degree of digitizable experiential attributes, product category might be a potential confound, as the products used differed on a number of other dimensions in addition to their degree of digitalization. Furthermore, the design of the experiment, that is, VDE/no VDE, may have biased the results by introducing additional salient information within the VDE treatment that was not included in the no-VDE treatment. Also, the results from the experiment indicated significant interaction between the VDE and the degree of digitalization, which interfered with our interpretation of the main effects. To address these concerns and further explore the issues that may influence ad message effectiveness in relation to VDE, a second experiment was conducted. The primary goals for Experiment 2 were to further investigate the main effect of the degree of digitalization within a single product category (employing three degrees of digitalization: low, medium, and high) and to identify the moderating effect of consumer product expertise (i.e., the degree to which consumers feel knowledgeable about/familiar with a certain product category).

EXPERIMENT 2

Product Selection

The second experiment examined the effect of the degree of digitalization within a single product category. Apparel was selected as the product category, based on several considerations: (1) it permits a meaningful VDE in a laboratory setting since it would present an experience product with limited digitizable experiential product attributes, and (2) it is appropriate for use as stimuli for the intended subject pool (i.e., undergraduate students) as indicated in initial pretesting.

Salient attributes play a significant role in determining how consumers evaluate products (Deighton 1984; Klein 1998; Smith 1993; Wright and Lynch 1995). Salient attributes for apparel were obtained in a pretest (n = 35) using the free-elicitation technique recommended by Fishbein and Ajzen (1975). Color emerged as the most salient attribute, listed by 89% of the respondents, followed by style (86%), fit (63%), and fabric (63%). These four salient attributes were used in Experiment 2 to check the perceived diagnosticity of the VDEs.

Hypotheses

As indicated previously, current technology allows easy digitalization of some experiential product attributes (such as sound and sight), whereas some other experiential product attributes (such as taste or touch) are not easily digitalized. Nevertheless, firms often attempt to provide VDE for those experience products low in overall proportion of digitizable experiential attributes. Although the advantage of VDE is apparent, firms are uncertain about how much effort needs to be devoted to its development. One of the reasons for their reservations is that they are skeptical about the effects of the degree of the digitalization, that is, whether a high degree of digitalization of experiential product attributes significantly influences ad effectiveness when promoting an experience product. We therefore theorize:

   H3: Compared with on-line ads integrated with VDE with
   medium-and low-degree digitalization, on-line ads integrated
   with VDE with a high degree of digitalization will produce:

   a. Lower perceived risk relating to the purchase of the product

   b. Higher evaluation of the product

   c. Higher affect

   d. Higher conation

   H4: Compared with on-line ads integrated with VDE with a
   low degree of digitalization, on-line ads integrated with VDE
   with medium-degree digitalization will produce:

   a. Lower perceived risk relating to the purchase of the product

   b. Higher evaluation of the product

   c. Higher affect

   d. Higher conation

Furthermore, research indicates that consumer product expertise influences trial experience and product evaluation (e.g., Kempf and Smith 1998). Consumers unfamiliar with a product, or product category, are more likely to use extrinsic attributes to evaluate the product. As a consumer's product expertise increases, his or her ability to assess products based on their intrinsic knowledge improves (Dowling and Staelin 1994; Rao and Monroe 1998). Degrees of digitalization consist of different levels of extrinsic attributes. Therefore, in VDE with a high degree of digitalization treatment, low product expertise consumers are likely to be more positive about their VDE experience than are high product expertise consumers. By comparison, in VDE with a low degree of digitalization treatment, low product expertise consumers are likely to be less positive about their VDE experience than are high product expertise consumers, since the extrinsic attributes they rely on for product evaluation are less available to them. Therefore, we propose the following hypothesis:

   H5: Consumer product expertise moderates the main effect of
   the degree of digitalization on perceived risk, product evaluation,
   affect, and conation.

Participants

One hundred students (48 male and 52 female) in undergraduate marketing courses in a major state university participated in the second experiment. Ninety-two percent of the subjects were between the ages of 20 and 25, with the balance between the ages of 26 and 35. Seventy-seven percent of subjects had purchased products on-line. Among them, nearly half of the subjects had made purchases on-line once a month or more frequently (42%), with 24% routinely purchasing clothes on-line. Subjects spent an average of 11 hours on the World Wide Web per week. Subjects were assigned randomly to three treatment conditions, that is, proportion of experiential attributes digitized (low, [D.sub.L]; medium, [D.sub.M]; high, [D.sub.H]), resulting in ceil sizes of 33 to 35 people.

Procedure

The researchers identified a retailer employing VDE to be used in this study by reviewing on-line apparel vendors.

Internet retailer Lands' End was chosen as the context for this study.

Subjects were asked to sign up for the experiment. They were then randomly assigned to three treatments ([D.sub.L][D.sub.M], [D.sub.H]) and asked to fill out a short survey about their body features (needed to create virtual models for the high digitization treatment). Although only the body-feature information from the subjects assigned to high digitalization treatment was needed, all participants were asked to fill out the short survey to minimize between-subject bias. A virtual model was then built for each subject in the [D.sub.H] treatment. Virtual models were personalized to share the same body features as the subjects. Three on-line ads integrated with VDE elements with three degrees of digitalization were created based on www.landsend.com. Each on-line ad consisted of the company name and logo for Lands' End and an advertisement providing a 50-word product description. Each 50-word product description provided product arguments relating to the four primary salient attributes (color, style, fit, and fabric), determined during pretesting, as well as a photo of the product.

As indicated by Ariely (2000), information control can have both advantages and disadvantages. For example, under some circumstances, controlling the information flow could increase consumers' decision quality, whereas under other circumstances, it could have detrimental effects on consumers' ability to utilize information. Therefore, whereas in Experiment 1, subjects were given the full freedom to explore the on-line ads, the information flow in Experiment 2 was controlled. Specifically, in the [D.sub.L] treatment, the color plate and choices of fabrics were embedded as the VDE element; in the [D.sub.M] treatment, the color plate and fabric choices plus a generic body model that the subject could use to "try on" the apparel were embedded as the VDE element; in the [D.sub.H] treatment, the color plate and fabric choices plus the subject's own personalized virtual model that the subject could use to try on the apparel were embedded as the VDE element. The experiment restrained the subjects from accessing any other information. Care was taken to make sure that the male and female subjects would be exposed to apparel that was comparable in terms of price, style, color, and fabric.

All experimental sessions were conducted in a computer laboratory in groups ranging from 8 to 12 participants. Male and female subjects were assigned to different sessions to avoid confounding resulting from the opposite gender's presence. The administrator's gender also matched the subjects' gender in each session. Administrators read instructions from a script describing the procedures. Subjects were first asked to complete the questions inquiring about their preexposure to the brand and their product expertise. Next, subjects were directed to the terminals with on-line ads preloaded accordingly for each treatment. After viewing the on-line ads, subjects were asked to complete the rest of the questionnaire where their postexposure brand attitude was measured together with the manipulation measures and dependent measures. A debriefing was conducted after all questionnaires were collected.

Measures

The dependent measures administered in Experiment 1 were employed for assessment: perceived risk ([alpha] = .72), product evaluation (a = .89), affect ([gamma] = .94), and conation ([alpha] = .92). An additional check for the digitalization manipulation was collected via perceived diagnosticity and perceived trial validity.

Perceived diagnosticity is defined as the degree to which the consumer believes that the on-line product direct experience (trial) is useful in evaluating the product's attributes (Kempf and Smith 1998). Since the degree of digitalization is related to the concept of VDE realism, a positive relationship should exist between the degree of digitalization and perceived diagnosticity. Perceived diagnosticity was measured at both product and attribute levels (Kempf and Smith 1998). Product-level diagnosticity was assessed, using a single-item scale, by asking respondents, "Overall, how helpful would you rate the trial experience you just had in judging the quality and performance of the software?" Responses were recorded on a one-to-seven scale, with the endpoints labeled "not helpful at all" and "extremely helpful" (cf. Kempf and Smith 1998). In addition, four attribute-level diagnosticity items (one for each of the four salient product attributes) were used, which asked, "To what extent did your trial experience with the apparel enable you to directly judge whether the product (possessed attribute X)?" The responses were recorded on a one-to-seven scale, with the endpoints labeled "trial did not enable me to judge this attribute" and "trial fully enabled me to judge this attribute" (cf. Kempf and Smith 1998). The correlation for the perceived diagnosticity scale (attribute and overall level) was .80.

Trial validity was defined as how credible and representative of the product's true performance a trial episode is perceived to be (Kempf and Smith 1998). Trial validity was measured using the following two items ([gamma] = .82) adopted from Kempf and Smith (1998): "Do you feel that this trial experience was a fair test of the (apparel)?" and "Do you feel that this trial experience was a valid test of the (apparel)?" The response scale ranged from one to seven, with endpoints labeled "unfair" and "fair" and "completely invalid" and "completely valid," respectively (correlation = .82).

Consumer product expertise was measured by adopting Mishra, Umesh, and Stem's (1993) scale. Subjects were asked to rate their product expertise on a seven-point semantic differential scale consisting of "know very little about clothes" to "know very much about clothes," "not experienced in buying clothes" to "experienced in buying clothes," "uninformed about fashion" to "informed about fashion," and "novice about clothes" to "expert on clothes" (correlation = .92).

As the stimulus adopted in this research was associated with the brand name Lands' End, consumer pre- and posttrial attitude toward the brand was measured by a three-item semantic differential scale: bad/good, unpleasant/pleasant, and dislike/like (Kempf and Smith 1998; MacKenzie and Lutz 1989; Smith 1993) ([alpha] = .87). Before the experiment, subjects familiar with Lands' End were asked to rate their pretrial attitude toward the brand based on the scale, while subjects not familiar with Lands' End had the option of skipping the evaluation. The same scale was again administrated after the VDE. The purpose of this pre-and posttrial attitude toward the brand measure was (1) to assess change in pre-and posttrial brand attitude, especially the magnitude of the change across the degree of digitalization, and (2) to assess interaction between consumers' brand exposure and the degree of digitalization.

Results and Analysis

Table 3 summarizes the cell means and analysis of variance (ANOVA) results for the measures.

Digitalization Manipulation

Results indicated a higher (F = 22.91, df = 2/96, p < .001) level of perceived diagnosticity in the [D.sub.H] treatment ([bar.X] = 5.68) than in the [D.sub.M] treatment ([bar.X] = 4.18) and the [D.sub.L] treatment ([bar.X] = 4.01). Results also indicated a higher level of perceived trial validity in the [D.sub.H] treatment ([bar.X] = 5.28) than in the [D.sub.M] treatment ([bar.X] = 4.52) and the [D.sub.L] treatment ([bar.X] = 4.09) (F = 7.70, df= 2/96, p < .001). Furthermore, the ANOVA test indicates that the [D.sub.L] and [D.sub.H] treatments and the [D.sub.M] and [D.sub.H] are significantly different at a level of .001 for both perceived diagnosticity and perceived trial validity, whereas the [D.sub.L] and [D.sub.M] treatments were not significantly different.

Main Effect of Degree of Digitalization

To test the hypotheses, first, the four dependent variables were entered into a MANOVA, with degree of digitalization as the independent variable. Significant main effects were found (Wilks's [lambda] = .207, F = 27.55, df = 4/184,p < .001, [[eta].sup.2] = .549). As indicated in Table 3, the main effect of the degree of digitalization was attributable to evaluation (F = 20.72, df = 2/96,p < .001), affect (F = 21.65, df = 2/96,p < .001), and conation (F = 15.71, df = 2/96, p < .001), with the exception of perceived risk (F = .43, df = 2/97,p = .654). Next, planned multiple comparisons were conducted using the Tukey's Post Hoc Test to examine the mean differences among low, medium, and high degrees of digitalization for each of the dependent variables. Specifically, subjects exposed to the [D.sub.H] treatment in comparison with the [D.sub.L] treatment consistently evaluated the apparel product more positively (H3b: [D.sub.H][bar.X] = 4.51, [D.sub.L][bar.X] = 3.00, p < .001), reported greater affect (H3c: [D.sub.H][bar.X] = 5.44, [D.sub.L][bar.X] = 3.43, p <.001), and showed higher conation (H3d: [D.sub.H][bar.X] = 5.48, [D.sub.L][bar.X] = 3.71, p < .001). Subjects exposed to the [D.sub.H] treatment also reported less perceived risk (H3a: [D.sub.H][bar.X] = 4.05, [D.sub.L][bar.X] = 4.21, p = .770) than subjects exposed to the [D.sub.L] treatment; this difference, however, is not statistically significant. In addition, subjects exposed to the [D.sub.H] treatment in comparison with the [D.sub.M] treatment rated the apparel product significantly more positively (H3b: [D.sub.H][bar.X]= 4.51,[D.sub.M][bar.X]= 3.31,p < .001), and had greater affect (H3c: [D.sub.H][bar.X] = 5.44,[D.sub.M][bar.X] = 4.11,p < .001). Subjects exposed to the [D.sub.H] treatment also reported less perceived risk (H3a: [D.sub.H][bar.X]= 4.05, [D.sub.M][bar.X] = 4.25,p = .668) and higher conation (H3d: [D.sub.H][bar.X]= 5.53, [D.sub.M][bar.X] = 5.06,p = .357) than subjects exposed to the [D.sub.M] treatment; these differences, however, are not statistically significant. Thus, H3b and H3c are fully supported, whereas H3d is partially supported and H3a is not supported.

Furthermore, subjects exposed to the [D.sub.M] treatment had significantly more positive conation than those in the [D.sub.L] treatment (H4d: [D.sub.M][bar.X] = 5.06, [D.sub.L][bar.X] = 3.71, p < .001). However, no statistically significant differences were found between the [D.sub.M] treatment group and the [D.sub.L] treatment group in relation to perceived risk (H4a: [D.sub.M][bar.X] = 2.75, [D.sub.L][bar.X] = 2.79,p = .979), evaluation (H4b: [D.sub.M][bar.X] = 3.31, [D.sub.L][bar.X] - 3.00, p = .334), and affect (H4c: [D.sub.M][bar.X] = 4.11, [D.sub.L][bar.X] = 3.43,p = . 125). Thus, H4d is fully supported, whereas H4a, H4b, and H4c are not supported.

Moderating Effect of Consumer Product Expertise

To test H5, the interaction between the degree of digitalization and consumer product expertise was examined via a multivariate general linear model. The scale midpoint was used as a cutting point to group subjects as either high (n = 62) or low (n = 38).

The test of the moderating effect of consumer product expertise yielded one significant interaction between the degree of digitalization and consumer product expertise for affect (F = 3.12, df = 2/95, p < .05). Specifically, for the [D.sub.H] treatment, consumers with low product expertise ([bar.X] = 6.03) responded more favorably than consumers with high product expertise ([bar.X] = 4.94) toward the on-line ad. Alternatively, for the [D.sub.L] treatment, consumers with low product expertise ([bar.X] = 3.19) responded less favorably than consumers with high product expertise (X = 3.55) toward the on-line ad. Interestingly, in the [D.sub.M] treatment, high/low product expertise consumers' preferences resembled those in the [D.sub.H] treatment. Figure 1 plots the significant interaction. As such, the results provide partial support for H5, although the overall interaction is not significant in the multivariate test (Wilks's [lambda] = .924, F = .86, df= 8/172,p = .550,[[eta].sup.2] = .039).

[FIGURE 1 OMITTED]

Attitude Toward the Brand

Of the 100 participants, 65 were unfamiliar with Lands' End prior to the experiment. Among the 35 subjects that had previous exposure to the brand, 14 of them were in the [D.sub.H] treatment, 8 were in the [D.sub.M] treatment, and 13 were in the [D.sub.L] treatment.

Descriptive statistics showed that there were changes in subjects' pre- and posttrial brand attitude in all three treatments. Specifically, subjects in the [D.sub.H] treatment showed a positive change of attitude toward the brand ([delta]X = 1.19), whereas subjects exposed to the [D.sub.M] treatment ([delta]X = -.48) and [D.sub.L] treatment ([delta]X = -.36) showed a slightly negative change of attitude toward the brand. A multivariate general linear model was then constructed to examine whether an interaction effect was evident between the pre- and posttrial brand attitude change and the degree of digitalization. The interaction was not statistically significant.

Discussion

The results demonstrate that on-line ads integrated with VDE incorporating a high degree of digitalization are more effective than on-line ads incorporating VDE with medium or low degrees of digitalization. The results provide further evidence of the importance of VDE, and more specifically, on the degree of digitalization, on ad message effectiveness. Indeed, the results suggest that even with an experience product with a low proportion of digitizable experiential attributes, such as apparel, a firm can still benefit from the use of VDE.

Second, the results indicate the significance of the moderating role of consumer product expertise. For instance, this research suggests that consumers with greater product knowledge are more likely to investigate the product based on their own expertise rather than through the assistance of VDE. Alternatively, consumers who are less confident about their product knowledge will rely more heavily on VDE.

GENERAL DISCUSSION AND CONCLUSIONS

Findings of this research suggest that on-line ads that employ VDE are more effective than on-line ads alone. Although the VDE effects were found for products with both a lower and greater proportion of digitizable experiential attributes, VDE effects were especially pronounced for products with a greater proportion of digitizable experiential attributes. The findings lend some of the first empirical support to the importance of product characteristics in relation to the effective employment of specific on-line marketing tactics (e.g., Peterson, Balasubramanian, and Bronnenberg 1997).

Furthermore, consistent with prior literature on DE (e.g., Kempf and Smith 1998; Wright and Lynch 1995), product level analysis indicated that for experience product [D.sub.H], an on-line firm can effectively stimulate ad message effectiveness by integrating VDE into on-line ads. This finding confirms the effectiveness of integrating VDE into on-line ads where experiential attributes are capable of being fully digitized (thus extending the current research). For experience products [D.sub.L], however, integrating VDE into on-line ads was only effective in influencing affect and product evaluation. The inability of on-line ads for this type of product to reduce perceived risk or stimulate conation may result from the fact that a lower proportion of the product's experiential attributes were able to be digitized and conveyed to consumers. This is important, as it suggests a new contextual effect, thus providing a foundation for theoretical development.

Managerially, the findings suggest that compared with traditional ad formats, the Internet provides firms with unprecedented opportunities for integrating VDE into their ads. This integration can equip on-line firms with the capability of communicating with their consumers more effectively. Despite the growth of the Internet and e-commerce, this advertising media has not yet been fully utilized. Elkin and Neff(2002) note that the on-line venue has not yet been effectively used in the larger mix in advertising campaigns. They indicate that most marketers only spend 2% to 3%, or less, of their media budgets to advertise to consumers on the Internet, despite the fact that the Internet represents 10% to 15% of total media consumption. The present study therefore highlights the necessity of exploring the Internet as an advertising media in general, and as an advertising media for experience products in particular.

This study also suggests that the effectiveness of VDE might vary for on-line ads across different experience product categories. Porter's (2001) analysis of strategic use of the Internet cautions that differentiation from competitors on-line is much more difficult, as critical points of distinction, such as experiential elements of touch and feel, are lacking. The findings of this study suggest that on-line firms focus on creatively applying technology to integrate VDE into on-line ads for experience products. For example, the effectiveness of VDE in a retail setting may minimize the competitive advantage that "brick and mortar" retailers have over on-line retailers when offering experience products. By providing VDE-laden ads of experience products, on-line retailers can enhance consumer information processing and conation.

In addition, findings of this research suggest that even for products with a lower proportion of digitizable experiential attributes, VDE with a higher degree of digitalization can stimulate message effectiveness to a greater extent than can VDE with a lower degree of digitalization. To date, although the advantage of VDE is apparent, firms are still conservative about the amount of resources allocated to the development of VDE. One of the reasons for this reservation is that firms have been skeptical about the effectiveness of the degree of digitalization, that is, whether a high degree of digitalization really makes a difference compared with a low degree of digitalization. If a higher degree of digitalization is not significantly more effective than a lower degree of digitalization, investment in VDE would not be worthwhile. Findings from this research shed light on this important issue.

Findings of the present study also suggest that while VDE with a high degree of digitalization might work well for consumers in general, it might be particularly beneficial to consumers with low product expertise. This finding therefore prompts firms to fully utilize the interactivity of the Internet that is feasible via VDE to customize on-line ads catering to different consumers' needs. This could be critical as firms expand to new markets where consumer product expertise is low. This suggests, for example, that retailers may be able to increase store patronage by offering VDE for new products to low product expertise consumers.

LIMITATIONS AND FUTURE RESEARCH DIRECTIONS

Although this study provides significantly new insights, it is not without its limitations. One limitation of the current study is that only a limited number of experience product categories were tested in the two experiments. Future research could expand on the current findings by using online ads for a variety of products within a broader range of product categories, thus extending the generalizability of the work.

Second, there is a possibility that the VDE effect is confounded by an information effect (e.g., difference in content embedded in the VDE stimuli). In Experiment 1, subjects in the VDE treatment were potentially exposed to more information elements than were subjects in the no-VDE treatment. In Experiment 2, subjects using the high and medium degree of digitalization were exposed to additional VDE elements (the virtual model). Although Experiment 2 reduced the potential of information flow to control effects, the participants in the different VDE conditions were exposed to different information (i.e., no model, generic model, virtual model), thus creating a possible confound of an information effect. Future research might attempt to further control the information effect to prevent the possible interaction it has with VDE and the degree of digitalization effect. One way to control the information effect might be to provide exactly the same amount of information through a VDE condition versus through another condition, such as a verbal description, so that these two conditions would differ only in terms of their modality of conveying information rather than the amount of information conveyed.

In conclusion, while many are concerned about the developing area of on-line advertising, there has been little empirical research examining how VDE embedded in on-line ads could influence ad message effectiveness. Specifically, this study demonstrated the importance of the degree of digitalization of experiential attributes for experience products in stimulating ad message effectiveness and, more important, highlighted the unique advantage of on-line ads in their ability to integrate VDE. These findings suggest that VDE is an important area of advertising inquiry. As such, this study provides a starting point for the development of more elaborate on-line VDE advertising models.

TABLE 1
Descriptive Statistics: Experiment 1

                       Perceived    Product              Purchase
                         risk      evaluation   Affect   intention

Treatment conditions

[D.sub.H] offering
No VDE                   3.17         3.97       3.24       3.03
                        (1.16)       (1.00)     (1.30)     (1.48)
VDE                      2.10         4.73       4.80       4.53
                         (.75)        (.94)     (1.06)     (1.24)

[D.sub.L] offering
No VDE                   3.07         4.36       3.96       3.01
                        (1.44)        (.89)     (1.41)     (1.43)
VDE                      2.79         4.65       4.05       3.21
                        (1.15)        (.83)     (1.35)     (1.14)

Notes: Standard deviations are in parentheses. Scores are presented on
a 1-7 scale to enhance comparability.

TABLE 2
Results of MANOVA: Experiment 1

                        Multivariate

         Wilks's    Effect
Effect   [lambda]    size     df     F value   Signif.

V          .755      .245    4/207   16.768     .000
D          .897      .103    4/207    5.914     .000
V x D      .847      .153    4/207    9.342     .000

                        Univariate

           Perceived
             risk         Evaluation        Affect         Conation
Effect   (df = 1/2 10)   (df = 1/2 10)   (df = 1/2 10)   (df = 1/2 10)

V         17.972 ***      16.502 ***      21.936 ***      21.247 ***
D          3.512 *         1.220            .003          13.991 ***
V x D      6.311 **         2.853 *       18.297 ***      12.452 **

Notes: MANOVA = multivariate analysis of variance; V = VDE/No VDE;
D = degree of digitalization ([D.sub.H] vs. [D.sub.L]).

* p < .10.
** p < .05.
*** p < .01.

TABLE 3
F Test of Three Degrees of Digitalization: Experiment 2

                           Degrees of digitalization

                                     Low        Medium
                            df     (n = 35)    (n = 31)

Perceived diagnosticity    2/96   4.01 (b)     4.18 (c)
Perceived trial validity   2/97   4.09 (b)     4.52 (c)
Perceived risk             2/97   2.80         2.73
Evaluation                 2/97   3.00 (b)     3.31 (c)
Affect                     2/96   3.43 (b)     4.11 (c)
Conation                   2/96   3.71 (a,b)   5.06 (a)

                           Degrees of digitalization

                              High
                            (n = 34)    F value   p value

Perceived diagnosticity    5.68 (b,c)     22.91     0.000
Perceived trial validity   5.28 (b,c)      7.70     0.001
Perceived risk             2.93             .43     0.654
Evaluation                 4.51 (b,c)     20.72     0.000
Affect                     5.44 (b,c)     21.65     0.000
Conation                   5.48 (b)       15.71      .000

(a) Difference between low-degree and medium-degree digitalization
is significant at p = .05 level.

(b) Difference between low-degree and high-degree digitalization
is significant at P = .05 level.

(c) Difference between medium-degree and high-degree digitalization
is significant at p = .05 level.

REFERENCES

Alba, Joseph W., and Wesley Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, 13 (4), 411-454.

--, John Lynch, Barton Weitz, Chris Janiszewski, Richard Lutz, Alan Sawyer, and Stacy Wood (1997), "Interactive Home Shopping: Consumer, Retailer, and Manufacturer Incentives to Participate in Electronic Marketplaces," Journal of Marketing, 61 (3), 38-53.

Ariely, Dan (2000), "Controlling the Information Flow: Effects on Consumers' Decision Making and Preferences," Journal of Consumer Research, 27 (3), 233-248.

Baker, Michael J., and Gilbert A. Churchill, Jr. (1977), "The Impact of Physically Attractive Models on Advertising Evaluations," Journal of Marketing Research, 14 (4), 538-555. Bauer, Raymond A. (1960), "Consumer Behavior as Risk Taking," in proceedings of the American Marketing Association, Dynamic Marketing for a Changing World, R. S. Hancock, ed., Chicago: American Marketing Association, 389-398.

Beltramini, Richard E, and Kenneth R. Evans (1985), "Perceived Believability of Research Results Information in Advertising," Journal of Advertising, 14 (3), 18-24, 31.

Biswas, Abhijit, and Scot Burton (1993), "Consumer Perceptions of Tensile Price Claims in Advertisements: An Assessment of Claim Types Across Different Discount Levels," Journal of the Academy of Marketing Science, 21 (Summer), 217-229.

Burke, Raymond R. (1997), "Do You See What I See? The Future of Virtual Shopping," Journal of the Academy of Marketing Science, 25 (4), 352-360.

Cho, Chang-Hoan (1999), "How Advertising Works on the WWW: Modified Elaboration Likelihood Model, "Journal of Current Issues and Research in Advertising, 21 (Spring), 33-50. Cunningham, Scott M. (1967), "Major Dimensions of Perceived Risk," in Information Handling in Consumer Behavior, Donald F. Cox, ed., Boston: Harvard University, 82-108.

Deighton, John (1984), "The Interaction of Advertising and Evidence," Journal of Consumer Research, 11 (December), 763770.

--, and Robert M. Schindler (1988), "Can Advertising Influence Experience," Psychology and Marketing, 5 (Summer), 103-115.

Dodds, William B., Kent B. Monroe, and Dhruv Grewal (1991), "The Effects of Price, Brand and Store Information on Buyers' Product Evaluations," Journal of Marketing Research, 28 (August), 307-319.

Dowling, Grahame R., and Richard Staelin (1994), "A Model of Perceived Risk and Intended Risk-Handling Activity," Journal of Consumer Research, 21 (1), 119-134.

Elkin, Tobi, and Jack Neff(2002), "Study Documents Branding Impact of Online Ads--Analyzed a Six-Week Dove Soap Campaign," AdAge.com (February 20).

Fishbein, Martin A. (1967), "Attitude and Prediction of Behavior," in Readings in Attitude Theory and Measurement, Martin Fishbein, ed., New York: Wiley, 477-497.

--, and Icek Ajzen (1975), Beliefs, Attitudes, Intention and Behavior: An Introduction to Theory and Research, Reading, MA: Addison-Wesley.

Ha, Young-Won, and Stephen J. Hoch (1989), "Ambiguity, Processing Strategy, and Advertising-Evidence Interactions," Journal of Consumer Research, 16, 354-360.

Hoch, Stephen J., and John Deighton (1989), "Managing What Consumers Learn from Experience," Journal of Marketing, 53 (April), 1-20.

--, and Young-Won Ha (1986), "Consumer Learning: Advertising and the Ambiguity of Product Experience," Journal of Consumer Research, 13 (September), 221-233.

Hoffman, Donna L., and Thomas P. Novak (1996), "Marketing in Hypermedia Computer-Mediated Environments: Conceptual Foundations," Journal of Marketing, 60 (3), 50-68.

Kempf, Deanna S., and Russell N. Laczniak (2001), "Advertising's Influence on Subsequent Product Trial Processing," Journal of Advertising, 30 (3), 27.

--, and Robert E. Smith (1998), "Consumer Processing of Product Trial and the Influence of Prior Advertising: A Structural Modeling Approach," Journal of Marketing Research, 31 (3), 325-338.

Klein, Lisa R. (1998), "Evaluating the Potential of Interactive Media Through a New Lens: Search Versus Experience Goods," Journal of Business Research, 41 (3), 195-203.

MacKenzie, Scott B., and Richard J. Lutz (1989), "An Empirical Examination of the Structural Antecedents of Attitude Toward the Ad in an Advertising Pretesting Context, "Journal of Marketing, 53 (April), 48-65.

Mano, Haim, and Richard L. Oliver (1993), "Assessing the Dimensionality and Structure of the Consumption Experience: Evaluation, Feeling, and Satisfaction," Journal of Consumer Research, 20 (4), 451-466.

Marks, Lawrence J., and Jerry C. Olson (1981), "Towards a Cognitive Structure Conceptualization of Product Familiarity," in Advances in Consumer Research, vol. 8, Kent B. Monroe, ed., Ann Arbor, MI: Association for Consumer Research, 145-150.

Meeker, Mary (1997), The Internet Advertising Report, New York: Harper Business.

Moorthy, Sridhar, Brian T. Ratchford, and Debabrata Talukdar (1997), "Consumer Information Search Revisited: Theory and Empirical Analysis," Journal of Consumer Research, 23 (4), 263-277.

Munch, James, Gregory W. Boiler, and John L. Swasy (1993), "The Effects of Argument Structure and Affective Tagging on Product Attitude Formation," Journal of Consumer Research, 20 (2), 294-302.

Murray, Keith B. (1991), "A Test of Services Marketing Theory: Consumer Information Acquisition Activities," Journal of Marketing, 55 (1), 10-25.

--, and John L. Schlacter (1990), "The Impact of Services Versus Goods on Consumers' Assessment of Perceived Risk and Variability," Journal of the Academy of Marketing Science, 18 (1), 51-65.

Nelson, Philip (1974), "Advertising as Information," Journal of Political Economy, 83 (July/August), 729-754.

Novak, Thomas P., Donna L. Hoffman, and Yiu-Fai Yung (2000), "Measuring the Customer Experience in Online Environments: A Structural Modeling Approach," Marketing Science, 19(1),22-42.

Perdue, Barbara C., and John O. Summers (1986), "Checking the Success of Manipulations in Marketing Experiments," Journal of Marketing Research, 23 (4), 317-326.

Peterson, Robert A., Sridhar Balasubramanian, and Bart J. Bronnenberg (1997), "Exploring the Implications of the Internet for Consumer Marketing," Journal of the Academy of Marketing Science, 25 (4), 329-346.

Petty, Richard E., John T. Cacioppo, and David Schumann (1983), "Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement," Journal of Consumer Research, 10(2), 135-146.

Porter, Michael E. (2001), "Strategy and the Internet," Harvard Business Review (March), 63-78.

Rao, Akshay R., and Kent B. Monroe (1988), "The Moderating Effect of Prior Knowledge on Cue Utilization in Product Evaluations," Journal of Consumer Research, 15 (2), 253-264.

Rust, Roland T., and Richard W. Oliver (1994), "The Death of Advertising," Journal of Advertising, 23 (4), 71-78.

Sheppard, Blair H., Jon Hartwick, and Paul R. Warshaw (1988), "The Theory of Reasoned Action: A Meta-Analysis of Past Research with Recommendations for Modifications and Future Research," Journal of Consumer Research, 15 (3), 325-343.

Singh, Mandeep, Siva K. Balasubramanian, and Goutam Chakraborty (2000), "A Comparative Analysis of Three Communication Formats: Advertising, Infomercial, and Direct Experience," Journal of Advertising, 29 (4), 59-72.

Smith, Robert E. (1993), "Integrating Information from Advertising and Product Trial: Processes and Effects on Consumer Response to Product Information," Journal of Marketing Research, 30 (2), 204-219.

--, and William R. Swinyard (1982), "Information Response Models: An Integrated Approach," Journal of Marketing, 46 (4), 81-93.

Steven, Edwards M., and Harshavardhan Gangadharbatla (2001), "The Novelty of 3D Product Presentations Online," Journal of Interactive Advertising, 2 (1).

Wright, Alice Ann, and John G. Lynch, Jr. (1995), "Communication Effects of Advertising Versus Direct Experience When Both Search and Experience Attributes Are Present," Journal of Consumer Research, 21 (4), 708-718.

David A. Griffith (Ph.D., Kent State University) is an assistant professor of marketing, Michigan State University, The Eli Broad Graduate School of Management.

Qimei Chen (Ph.D., University of Minnesota) is an assistant professor of marketing, University of Hawaii at Manoa, College of Business Administration.

The authors contributed to this paper equally; the order of the authorship was decided randomly. The authors thank the University of Hawaii at Manoa and the Michael E Price College of Business at the University of Oklahoma for providing support for this project, as well as Michael Y. Hu, Richard Kolbe, and William D. Wells for their insightful comments on previous versions of this paper. The authors also thank the University of Hawaii New Business Information Technology Center for providing facilities for this study. Finally, the authors thank the Editor, Ronald R. Faber, and the anonymous reviewers for constructive comments on previous versions of this manuscript.