The effects of online tactile information source for low-touch products on consumer responses

Online touch information

Previous literature has underlined the crucial role of the sense of touch (Major et al., 1990; Ackerman et al., 2010) that allows consumers to obtain information about the surface and other spatial information of commodities. Several scholars have analyzed the impact of physical touch in different marketing contexts, as interpersonal touch (Crusco and Wetzel, 1984; Hornik and Ellis, 1998) or product taste (Hornik, 1992) . Moreover, Peck and Childers (2003b) demonstrated that haptic orientation is an individual difference variable reflecting consumers’ NFT, which is defined as their “preference for the extraction and the utilization of information obtained through the haptic system” (p. 431) and has two underlying dimensions: instrumental, related to the need of touching products driven by a purchase goal, and autotelic, related to hedonic motivations and to touch products as an end in and of itself. Recently, Jha et al. (2019) showed that the instrumental and autotelic dimensions of NFT moderate the evaluation of haptic and nonhaptic products in different purchase environments. Other studies found a positive relationship between touch and both impulse purchasing (Kacen and Lee, 2002; Peck and Childers, 2006) and perceived ownership of the touched object (Peck and Shu, 2009; Maille et al., 2020). Moreover, Streicher and Estes (2015) demonstrated that haptic brand identities can facilitate recognition, consideration and brand choice, whereas Streicher and Estes (2016) showed that touching a given product can influence perception and choice of other seen products. Haptic cues also play a pivotal role in food perception, as taste (Krishna and Morrin, 2008) and perceived naturalness (Labbe et al., 2013). Finally, Yoganathan et al. (2019) found that tactile priming statement increases consumers’ willingness to pay for an ethical brand online.

Given that the absence of touch is one of the main deterrents of online shopping (Alba et al., 1997; Citrin et al., 2003), previous research has demonstrated that such an absence can be coped with providing tactile information, that is, information on the material properties of products, such as their feel, temperature and weight (Klatzky and Lederman, 1993). Such information can be offered through a written description of the tactile sensations deriving from the usage of the product and/or a visual representation (McCabe and Nowlis, 2003; Peck and Childers, 2003a) available on both company websites and web communities (Bickart and Schindler, 2001).

Company websites are formal sources of information, where firms offer information about products through impersonal one-to-many communication. Many traditional companies moved into e-commerce, still maintaining their physical stores (Lawson, 2001). These companies generally provide well-known brand names, local store delivery and return/exchange facilities through their websites (Nataraj and Lee, 2002); moreover, they offer trust, personal privacy, technical reliability and fast delivery (Schoenbachler and Gordon, 2002). Multichannel company websites are thus likely to reduce the levels of perceived risk even though consumers cannot evaluate products through the touch. However, other single-channel companies started operating only in online setting (Laudon and Traver, 2001). As these companies do not benefit from a physical store, the presented products are supposed to be perceived as riskier. As a consequence, the presence of haptic information is particularly relevant for consumers not having any previous experience with the products. In the present research, we focus on company websites offering their (unknown) brands only in the online setting.

Web communities are informal sources of information, where consumers, organized in groups engaged in computer-mediated interactions around a common interest (Bagozzi and Dholakia, 2002), spontaneously share opinions and experiences about products through interpersonal many-to-many communication based on posts, comments and reviews. The interactive nature of online communities allows consumers to exchange information about product experiences by giving opinions and advices on the quality and usage of products, thereby leading to electronic word of mouth (i.e. e-WOM; Hennig-Thurau et al., 2004) communication (Brown et al., 2007; Park and Cho, 2012). An extensive research stream on e-WOM has analyzed how it works (King et al., 2014; Babić Rosario, de Valck and Sotgiu, 2019) and its effects on consumer behavior (López and Sicilia, 2014; Chu and Kim, 2018) and product sales (Chevalier and Mayzlin, 2006; Babić Rosario et al., 2016), demonstrating that users comments and reviews are influential sources of information for other users (Godes et al., 2005; Pitta and Fowler, 2005).

Both online information sources can provide consumers with tactile information about products before purchasing. However, web communities seem to be more effective in influencing consumer behaviors than company websites (Dhar and Chang, 2009; Mudambi and Schuff, 2010; Senecal and Nantel, 2004; Punj, 2013). Indeed, previous research shows that information exchanged within web communities through e-WOM reflects the knowledge and the personal experience of other users (Zhu and Zhang, 2010; Liu and Lopez, 2016) and is generally considered to be more reliable than the information provided by the company (Wathen and Burkell, 2002; Fan et al., 2013). Accordingly, some scholars have identified in the sincere concern for other consumers (Hennig-Thurau et al., 2004) and in the willingness to engage in helping behaviors for other consumers (Johnson et al., 2013) primary motivations that lead users to post their opinions about a product online. Being regarded as a more reliable source, it is likely that e-WOM communication has a greater impact on products evaluation of their members (Cheung et al., 2009; López and Sicilia, 2014) and, therefore, that tactile information provided by web communities is perceived as more reliable than the same information provided by a company website and may have a greater impact on consumer responses. However, previous research has not clarified for which type of product, in terms of touch diagnosticity, the provision of tactile information within a web community (vs. a company website) is more effective in influencing consumers’ evaluations. Such effect may also vary for consumers with different levels of NFT.

Diagnosticity of touch information: product type and need for touch

We maintain that the type of product moderates the relationship between online touch information sources and consumer attitude, by distinguishing between low- vs. high-touch products that differ in the extent to which they possess salient material properties, that is, in the importance of touch (Lederman and Klatzky, 1987; Peck and Childers, 2003a, 2003b). Dual-process theories of persuasive communication (e.g. elaboration likelihood model – Petty and Cacioppo, 1981; and heuristic-systematic model – Chaiken, 1980) provide the background for our predictions. These theories propose that consumers process information based on either a central route (that requires considerable effort) or a peripheral route (that implies using mental shortcuts). The central route implies high motivation and/or ability in processing the information and corresponds to those situations in which consumers think attentively and take a decision after a careful consideration of all given information. Conversely, the peripheral route implies low motivation and/or ability in processing the information and corresponds to those situations in which consumers do not think carefully and use shortcuts (i.e. heuristics) to make judgments.

For low-touch products, tactile information is not diagnostic during product evaluation. Consumers will follow a peripheral route because they are less motivated to have tactile information about the product and will, therefore, use the perceived reliability of the information source as a heuristic. Under such conditions, the source of information (web communities vs. company website) is more relevant than the information itself (i.e. the description of tactile characteristics). Therefore, we expect that, for low-touch products, tactile information provided by a web community (vs. a company website) will increase the perception of reliability that, in turn, will produce higher consumer attitude toward the product. For high-touch products, tactile information is diagnostic during product evaluation. Consumers will follow a central route because they are more motivated to have tactile information about the product and will, therefore, not use the perceived reliability of the information source as a heuristic. Under such conditions, the source of information is less relevant than the (diagnostic) tactile information itself. Therefore, we expect that, for high-touch products, there is no difference in the perceived reliability between the two sources of tactile information, and thus in terms of consumer attitude toward the product. Of importance, we hypothesize that such a differential effect of the online tactile information sources (web community vs. company website) for low rather than high-touch products occurs both as a direct effect of information source and product type on consumers’ attitude (H1) and through the effect of such two factors on perceived reliability (H2). Formally, the following hypotheses are postulated:

We also maintain that the effects of online tactile information sources on consumer responses may vary depending on consumers’ NFT. As already stated, NFT is an important personal trait related to diagnosticity of touch information that has been demonstrated to influence online shopping (González-Benito et al., 2015; Lee et al., 2017; Pino et al., 2019; Jha et al., 2019). In particular, previous research found that consumers high in NFT – namely, those consumers for which the absence of tactile experience generates a greater frustration and a lower trust in product evaluations compared to low-NFT counterparts (Peck and Childers, 2003b) – are more likely to rate the quality of high-touch products lower than consumers low in NFT (Rodrigues et al., 2017; San-Martín et al., 2017; Kühn et al., 2020). We expect to find similar effects also for low-touch products. In particular, we predict that the positive effect played by a web community (vs. a company website) in providing tactile (vs. generic) information for a low-touch product on consumers’ responses holds only when consumers present a low NFT. Those consumers are likely to follow a peripheral route of processing information. Conversely, for consumers that present a high NFT and are therefore very sensitive to tactile information, the provision of tactile information by a web community (vs. a company website) is not expected to influence their responses toward low-touch products (H3) as they still prefer the physical inspection of the product and consequently are more likely to follow a central route. Formally, the following hypothesis is postulated:

Studies overview

We tested our conceptual framework in three experimental studies. Study 1 was aimed at investigating H1 and H2 for a pair of low- and high-touch products. Study 2a was aimed at analyzing the effect of online sources of tactile information (web communities vs. company websites) for low-touch products using a control condition (i.e. the absence of tactile information) and investigating the moderating role of the NFT, thus testing H3. Moreover, Study 2a examined consumers’ willingness to buy (hereafter: WTB) as another DV. In Studies 1 and 2a, we operationalized the web community source considering a consumer forum. Finally, Study 2b was aimed at generalizing previous results for low-touch products, considering a more realistic scenario (i.e. consumer reviews available on an online store) and analyzing brand attitude as additional DV. In each study, we blocked the exposure time of the experimental scenarios for a minimum of 50 s to ensure that all relevant information contained in each scenario’s description has been read and understood by participants. The extended versions of the scenarios used in all the studies are available upon request.

Pretest

To select the products to use in our experimental studies, we firstly showed to two Web-retailing experts a list of 20 commodity products [1] and asked them to select two everyday products generally sold online for which the perceived importance of touch has a low or high level, respectively. Following our instructions, the experts converge in indicating the scarf and the keyring as high-/low-touch products, respectively.

After that, we selected a picture from Google images for each selected product and conducted a pretest where 30 subjects (13 females, M_age = 26.37, SD = 4.18) were asked to evaluate the two products, shown in a random order, in terms of the perceived importance of touch (1 = not at all important; 7 = very important). Results of a repeated measures ANOVA confirmed that the keyring received the lowest evaluations (M_keyring= 2.77; SD = 1.70) and the scarf the highest evaluations (M_scarf = 6.37; SD = 1.03). Moreover, the keyring and the scarf significantly differ in terms of the perceived importance of touch (t(29) = −10.26, p = 0.00).

Study 1

Study 1 was a lab study aimed at testing H1 and H2. We used a 2 (online tactile information source: web community vs. company website) × 2 (type of product: low- vs. high-touch products) between-subjects design. In total 133 students (72 males, M_age = 22.69, SD = 1.46) participated in the study in exchange of a bonus course credit. Participants were randomly assigned to one of four experimental scenarios: the description of a low-touch product (i.e. a keyring) given by the users of a fictitious web community (i.e. Look&Accessories) or by the company website of a fictitious brand (i.e. Class); the description of a high-touch product (i.e. a scarf) given by the users of a web community (i.e. Look&Accessories) or by the company website of a fictitious brand (i.e. Class). In each description, the same tactile information about the products (e.g. texture, weight, tactile pleasantness, feel) was provided. After that, attitude toward the product was measured. In particular, participants were asked to indicate how much they liked the product (1 = not at all, 7 = a lot), their general orientation toward it (1 = negative, 7 = positive) and to what extent they considered the product desirable (1 = not at all, 7 = a lot). The three items were averaged in an overall attitudinal score (α = 0.86; M_att = 4.04, SD_att = 1.18). Moreover, perceived reliability was measured by asking participants to what extent they perceived the description of the product on each scenario as reliable (“To which extent do you believe that the description of the “Class” keyring is reliable?” 1 = not at all, 7 = a lot). Finally, participants answered to two manipulation checks of information source (1= company; 7= users) and touch importance (1 = not at all important; 7 = very important), then they were asked demographic questions (i.e. age and gender), were debriefed and compensated.

Results. Manipulation checks were successful for both information sources (M_community = 4.12, SD = 1.16; M_website = 2.29, SD = 1.53; F(1, 131) = 26.69, p < 0.01; η_p² = 0.99) and touch importance (M_scarf = 5.18, SD = 1.18; M_keyring = 3.16, SD = 1.36; F(1, 131) = 80.28, p < 0.01; η_p² = 1.00). A two-way ANOVA on the previously computed attitudinal score revealed a significant direct effect of both the information source (M_community = 4.27, SD = 1.08; M_website = 3.74, SD = 1.25; F(1, 129) = 6.43, p = 0.01; η_p² = 0.05) and the product type (M_scarf = 4.56, SD = 0.09; M_keyring = 3.64, SD = 1.20; F(1, 129) = 26.30, p < 0.001; η_p² = 0.17). More importantly, the main effects are qualified by a significant interaction effect between the two independent variables (F(1, 129) = 4.05; p = 0.05; η_p² = 0.03). In line with H1, planned comparisons indicated that, for low-touch products, consumer attitude is higher when tactile information is provided by a web community than a company website (M_{community_keyring} = 4.02, SD = 1.17; M_{website_keyring} = 3.18, SD = 1.08; F(1,129) = 12.10, p = 0.001; η_p² = 0.09), whereas for high-touch products there is no significant difference in terms of consumer attitude toward the product (M_{community_scarf} = 4.60, SD = 0.08; M_{website_scarf} = 4.51, SD = 1.06; F(1,129) = 0.12, p = ns; η_p² = 0.001).

Then, to test H2, we conducted a mean centered moderated mediation analysis (Model 7 of PROCESS macro; Hayes, 2017), considering information source as independent variable (website = 0, web community = 1), type of product as moderator (low-touch = 0, high-touch =1), the perceived reliability as mediator and the overall attitude score as dependent variable. Indeed, we expect that, for low-touch products, web communities providing tactile information are perceived to be more reliable than the company website, whereas for high-touch products the description of tactile characteristics does not influence perceived reliability irrespective of the source providing it (Figure 1). Results showed that both information source (b = 0.48, SE = 0.22; t = 2.13; p = 0.04; r² = 0.03) and type of product (b = 0.92, SE = 0.26; t = 3.60, p < 0.001; r² = 0.09) have a significant effect on perceived reliability. Moreover, the interaction between information source and type of product is not significant (b = −0.49, SE = 0.34, t = −1.43, p = 0.15; r² = 0.02). However, in line with our conceptualization, simple slopes analysis showed that, for low-touch products, web communities are perceived as more reliable than the company website (b = 0.25, p < 0.05), whereas for high-touch products information source does not affect the perception of reliability (b = −0.01, p = ns). Considering attitude score as dependent variable, both reliability (b = 0.52, SE = 0.09; t = 5.87; p < 0.001; r² = 0.21) and information source (b = 0.39, SE = 0.18, t = 2.16; p = 0.03; r² = 0.03) have a significant effect. Moreover, there is a significant indirect conditional effect of information source on attitude for low-touch products equal to 0.25 [95% CI: 0.02, 0.55]. Conversely, for high-touch products the indirect conditional effect is not significant [95% CI: −0.28, 0.26]. We can thus confirm our conceptualization that, for low-touch products, a web community is perceived as more reliable than the company website, thus increasing consumers’ attitude. Instead, for high-touch products, there is no difference in terms of perceived reliability and thus on consumers’ attitude between the two information sources.

Our results did not take into account the level of product involvement. As a consequence, one may infer that consumers have a relatively higher involvement for high-touch products (i.e. scarf) compared to low-touch products (i.e. keyring).

To rule out this alternative explanation and therefore to show that our results are driven by touch importance and not by product involvement, we conducted a post-test where 52 undergraduate students (14 females, M_age = 20.84, SD = 0.64) were asked to evaluate the two products used in Study 1 in terms of their perceived involvement. Product involvement was measured asking participants to which extent they evaluated the product as “attractive,” “important,” “interesting” and “relevant” (1 = not at all; 7 = a lot). Results of a repeated measures ANOVA confirmed that the two products (M_Keyring = 3.91, SD = 0.15; M_Scarf = 4.48, SD = 0.18) did not differ in terms of perceived involvement (ΔM_{keyring-Scarf} = −0.57, SE = 0.24; p = 0.21).

Study 2a

Study 1 did not include a control condition. We therefore conducted an additional experimental study in which we manipulated the presence (vs. absence) of tactile information focusing on low-touch products. Moreover, according to H3, we tested the moderating role of the NFT. Finally, to show that our findings go beyond to just attitudinal changes, we examined WTB as DV.

We used a 2 (online tactile information source: web community vs. company website) × 2 (tactile information: present vs. absent) lab experiment in which 127 students (53 males, M_age = 22.63, SD = 1.15) participated in exchange of one bonus course credit. Participants were randomly assigned to one of four experimental scenarios in which the same low-touch product used in Study 1 (i.e. keyring) was described with (vs. without) tactile information by the users of a fictitious web community (i.e. Look&Accessories) or by the company website of a fictitious brand (i.e. Class). After that, participants were asked for their WTB the product that was measured through three seven-point Likert items (i.e. “If available I would buy the Class keyring,” “If available I could consider to buy the Class keyring,” “The likelihood of buying the Class keyring, if available, is high”; α = 0.96; M_WTB = 2.58, SD_WTB = 1.36) adapted from Dodds et al. (1991). Then, their personal NFT (α = 0.90; Peck and Childers, 2003b) was measured. Finally, participants answered to two manipulation checks of information source (1 = company; 7 = users) and touch information (1 = absence of tactile information; 7 = presence of tactile information), then they were asked demographic questions (i.e. age and gender), were debriefed and thanked for their participation.

Results. Manipulation checks were successful for both information sources (M_community = 5.71, SD = 1.67; M_website = 3.47, SD = 2.03; F(1, 125) = 46.45, p < 0.001; η_p² = 1.00) and touch information (M_touch = 5.57, SD = 1.52; M_{no_touch} = 3.37, SD = 1.62; F(1, 125) = 62.15, p < 0.001; η_p² = 1.00). Then, we conducted a three-way mean centered moderation analysis (Model 3 of PROCESS macro; Hayes, 2017), considering information source as independent variable (website = 0, web community = 1), tactile information (absent = 0, present =1) and the NFT score as moderators, and the WTB as DV. Indeed, we expect that product information provided by web communities and containing tactile details increases consumers’ WTB but only in case of low NFT. Results showed a significant and positive interaction between information source and tactile information (b = 4.98, SE = 1.86, t = 2.68, p < 0.001; r² = 0.05), a significant and positive interaction between information source and NFT (b = 0.68, SE = 0.28, t = 2.41, p = 0.02; r² = 0.04), a significant and positive interaction between touch information and NFT (b = 0.78, SE = 0.29, t = 2.68, p < 0.01; r² = 0.05) and more importantly a significant and negative three-way interaction between source information, touch information and NFT (b = −1.22, SE = 0.41, t = −2.99, p < 0.001; r² = 0.07). The Johnson-Neyman “floodlight” analysis (Spiller et al., 2013) revealed a significantly positive effect of tactile information provided by web communities on WTB for NFT values lower than 2.50 (b_JN = 1.93, SE = 0.90, 95% CI: 0.14, 3.72). This evidence allows to confirm our expectations that for customers low in NFT the presence of tactile information in web communities for low-touch products is more effective, as it leads to a higher WTB. However, for values of NFT higher than 4.9 (b_JN = −0.99, SE = 0.50, 95% CI: −1.98, −0.12), the effect on WTB becomes significantly negative. Then, looking at the conditional effects of X on Y at values of moderator (NFT), we found that, for values of NFT bigger than 4.40, the absence of tactile information in web- communities generates a higher WTB. A possible explanation of this surprising result could be that for people high in NFT the absence of haptic experience in online setting is so frustrating that making this absence salient (through the provision of tactile information) generates an opposite effect in terms of WTB. Future studies could shed light on these relationships.

Study 2b

So far, we demonstrated that tactile information provided by web communities is more effective compared to company websites, but only for low-touch products and for individuals low in NFT. Focusing on this specific case, Study 2b was aimed at generalizing previous results by comparing the presence (vs. absence) of tactile information provided by a web community for low-touch products and adopting a more realistic context, namely, consumer reviews available on an online store. Moreover, we analyzed brand attitude as additional DV.

Study 2b used a single factor (tactile information: present vs. absent) between-subject design, conducted among 91 Prolific Academic UK workers (29 males; M_age = 30.76; SD = 9.07). Participants were firstly asked to imagine that they had to buy a keyring of a fictitious brand (i.e. Class), and then they were exposed to a fictitious e-commerce website containing consumers’ reviews of the keyring. Depending on the condition, the reviews provided (vs. did not provide) tactile information (i.e. light metal material, nice to touch), the latter condition containing only generic information (i.e. metal material, easy to use) about the product. In each description, the same amount of text was provided (Appendix).

Then, brand attitude was measured by asking participants to indicate to which extent they agree or disagree (1 = totally disagree; 7 = totally agree) to the brand attitude scale items adapted from Spears and Singh (2004; “The Class brand is good,” “My attitude toward the Class brand is positive,” “I react favorably to the Class brand,” “I like the Class brand,” “I find the Class brand appealing”). The five items were averaged in an overall brand attitude score (α = 0.94, M_{att_brand} = 4.73, SD_att = 1.01). Moreover, as in Study 2a, we measured NFT (α = 0.92; Peck and Childers, 2003b) and a manipulation check of touch information (1 = absence of tactile information; 7 = presence of tactile information) was administered. Additionally, as past literature has extensively demonstrated that brand attitude is influenced by perceived product quality (Pappu et al., 2005; Johnson et al., 2006), we decided to measure also this variable using a single item seven-point Likert scale (“I think that the quality of the Class keyring is extremely high,” 1 = strongly disagree; 7 = strongly agree). Then participants were asked demographic questions (i.e. age and gender), were debriefed and compensated for their participation.

Results. Manipulation check for touch information was successful (M_touch = 5.40, SD = 1.31; M_{no_touch} = 3.98, SD = 1.45; F(1, 89) = 24.18, p < 0.001; η_p² = 0.21). Then, we conducted a mean centered moderation analysis (Model 1 of PROCESS macro; Hayes, 2017) with tactile information as independent variable (absent = 0, present = 1), the NFT score as moderator, the brand attitude mean score as DV and the perceived quality as a control variable. Indeed, we expect that reviews containing tactile information increase brand attitude but only for consumers low in NFT (Peck and Childers, 2003b). Results showed that both tactile information (b = 1.41, SE = 0.56, t = 2.51, p = 0.01; r² =0.07) and perceived quality (b = 0.53, SE = 0.07, t = 8.00, p < 0.001; r² = 0.42) positively influenced brand attitude, whereas NFT did not (b = 0.12, SE = 0.09, t = 1.29, p = ns; r² = 0.02). More importantly, as in Study 2a, the interaction between tactile information and personal NFT is significant and negative (b = −0.29, SE = 0.13, t = −2.15, p = 0.03; r² = 0.05). Consistent with our expectations, the Johnson–Neyman “floodlight” analysis (Spiller et al., 2013) revealed a significantly positive effect of tactile information on consumers’ brand attitude for NFT values lower than 3.78 (b_JN = 0.33, SE = 0.17, 95% CI: 0.00, 0.67). This evidence allows to confirm our previous findings, namely, that for consumers low in NFT the presence of tactile information for low-touch products leads to a higher brand attitude. For values of NFT higher than 3.91 (b_JN = 0.29, SE = 0.16, 95% CI: −0.03, 0.62), the effect on brand attitude becomes not significant. As a consequence, compared to generic information, user-generated tactile information about low-touch products positively influences consumers evaluations, but only if they are low in NFT.