Chapter 4: Location Based Services with Push Information Delivery Mechanisms and
5.3 Research Model and Hypothesis Development
As depicted in Figure 16, our research model examines the effects of high message interactivity (MI) and platform self-disclosure (PSD) on user disclosure propensity (H1/H2) as well as the role of MI in moderating the effect of PSD on user disclosure propensity (H3). Thus, we intend to investigate the isolated and combined effects of our chosen social cues.
Figure 16: Research Model.
5.3.1 The Effect of Message Interactivity on User Disclosure Propensity
As described earlier, we intend to investigate what will happen when the requests are low in message interactivity, so that all questions are presented at once at the beginning like in a classic computer form, in contrast to a high message interactive condition, when the questions are presented stepwise and conversational like in a dynamic dialogue. Social response theory (Nass & Moon, 2000) suggests that the more social cues are present, the more will a user perceive a CA as a social actor (Nass et al., 1994), making the user respond more socially. Thus, the conversational turn-taking in a high message interactive condition may improve the perception of the chatbot as a social actor in contrast to a low message interactive condition, as one more essential nonverbal social cue is included in the former.
Indeed, research has shown that interactivity is related to the perception of social presence which has been found in studies on CAs as well. For instance, Skalski and Tamborini (2007) demonstrated that perceived interactivity can influence social presence, information processing, and persuasion. Regarding social presence, research on embodied CAs revealed that it directly influences trusting beliefs, perceived enjoyment, and ultimately usage intentions (e.g., Hess et al., 2009; Qiu & Benbasat, 2009a). Trusting beliefs, furthermore, were shown to influence privacy concerns as well as to increase user disclosure propensity (e.g., Smith, Dinev, & Xu, 2011b; Taddei & Contena, 2013). Consequently, based on previous research on the positive effects of interactivity on business-oriented outcomes and related research on CAs that linked these effects to other outcomes on user behavior and intentions, we hypothesize that high (vs. low) message interactivity can increase user disclosure propensity.
5.3.2 The Effect of Platform Self-Disclosure on User Disclosure Propensity
A considerable amount of research has used social exchange theory to explain the reciprocation of favorable and unfavorable behaviors between parties (Cropanzano & Mitchell, 2005) and found disclosure reciprocity as a meaningful social norm in many social exchange contexts (e.g., Cropanzano & Mitchell, 2005; Sprecher et al., 2013). When two individuals encounter each other, the ability to build rapport is contingent on both parties to reciprocate in a dialogue (Collins & Miller, 1994; Sprecher et al., 2013). Normally, adhering to social norms improves the relationship, while violating hurts it (e.g., Collins & Miller, 1994; Sprecher et al., 2013). Consequently, if a party fails to reciprocate, the relationship will less likely have a positive development (Sprecher et al., 2013).
Applied to our experiment, social exchange theory suggests that if a platform gives away a piece of information, the user tends to respond by providing a piece of information of similar value to adhere to social norms. Indeed, past studies on website disclosure (e.g., “unreasoned dyadic relationships” defined as the platform discloses information first before asking for similar information) (e.g., Zimmer et al., 2010) have already indicated this reaction, in that a user may perceive an appropriate and non-manipulative self-disclosure as a rewarding outcome and a cue to build trust (Collins & Miller, 1994), hence appreciating the action (Emerson, 1976) and tending to mimic the behavior (Chartrand & Bargh, 1999). Actually, reciprocal self- disclosure may even pose such a strong social norm that even information disclosure by a computer may be considered a verbal social cue and can, thus, create the perception of a social actor (Nass et al., 1994). Consequently, platform self-disclosures may create feelings of imbalance in users that are usually only created in human-human-interactions. As a result, a user desires to restore equality in the relationship (Sprecher et al., 2013) and reestablish an equilibrium with the computer (Homans, 1958). Thus, we expect that the self-disclosure of the platform in a disembodied CA will cause the user to self-disclose information more likely.
H2: Platform self-disclosure will positively affect user disclosure propensity.
5.3.3 The Moderating Role of Message Interactivity on the Effect of
Platform Self-Disclosure on User Disclosure Propensity
Previous research has shown that social cues may surprisingly interact with each other, increasing the perception of social presence and related dimensions (e.g., Seeger et al., 2018). Regarding the effects of our investigated cues, the high message interactivity condition with its sequential turn-taking as a nonverbal cue, also known as prerequisite of the “conversational ideal” (Sundar et al., 2016), may be so essential that other cues can develop their potentials
more effectively in its presence. The verbal social cue self-disclosure may be a specifically intriguing candidate, as both cues are fundamental in common human-human interactions where information is exchanged and revealed turn by turn and one after another: Whereas high message interactivity is defined as one message is contingent on and only revealed after another message, reciprocal self-disclosure is built on the concept that one party starts to self-disclose so that the other party can socially respond by self-disclosing as well. Therefore, the perception of a give-and-take information exchange may flourish better when a user perceives a sequential turn-taking in form of high message interactivity, so that the user reasons that his or her self- disclosure has consequences on the conversation and, thus, on the relationship and following interaction between the user and the chatbot. Consequently, we believe that when both cues are presented together, they increase the chances that users will disclose information, in that high message interactivity enhances the effect of platform self-disclosure.
H3: High message interactivity will moderate the effect of platform self-disclosures so that high message interactivity will enhance the effect of platform self-disclosures on user disclosure propensity.