The Effect of Message Persistence and Disclosure on Liking in Computer-Mediated Communication

Abstract

In computer-mediated communication (CMC) systems, the messages that a user types usually persist on the screen for some time, a feature that distinguishes CMC from face-to-face interaction. Persistence may activate psychological self-perception, leading communicators to infer from their persistent messaging how they feel about the subject more so than if messages did not persist. This study applies persistence and self-perception to the relationships between self-disclosure and liking. It identifies which among several disclosure or liking relationships may be most susceptible to self-perception effects. An experiment found that message persistence interacts with a conversational partner’s responses to self-disclosure and intensifies liking toward the partner. Suggestions follow for future research further exploring the mechanisms of persistence, and reconceptualizing self-perception factors in interactive media settings.

Notes

1. In some CMC literature, the term “persistence” refers to the affordance being able to store, retrieve, and review messages in many forms of CMC. Although this feature also distinguishes CMC from other communication modalities and the impacts of this affordance are also potentially important (for review, see Treem & Leonardi, 2012), the present study focuses only on the more limited visual persistence of messages on one’s screen as they are created.

2. Given that most findings about disclosure in CMC arose in conversations among strangers, it is unclear whether these dynamics also occur among friends. Traditional research suggests that disclosure plays an important role in both developing and developed relationships (Altman & Taylor, 1973) and leads to psychological adjustment (Jourard, 1964). Recent CMC research shows that disclosure among friends online has even stronger effects on psychological adjustment than does disclosure among strangers (Valkenburg & Peter, 2007). Although a great deal of mediated disclosure now takes place among friends in social network sites (Bazarova, 2012), there are numerous CMC venues in which strangers disclose as well, including online date-seeking, chatting in online games, and describing one’s concerns in online support groups (see, resp., Fiore & Donath, 2004; Pena & Hancock, 2006; Turner, Grube, & Meyers, 2001). In response, this study originally employed dyads comprised of two friends and dyads composed of two strangers. Recruitment instructions required each participant to bring a friend to the study (a close acquaintance who spent time with the participants beyond having the same classes, and was not an intimate or romantic partner; Dunne & Ng, 1994). When two dyads arrived at the lab simultaneously, assistants split each dyad and matched a participant from one dyad with one of the other dyad members, composing pairs of strangers. When only one dyad arrived at the lab, that pair of friends participated. As a result, approximately half of the dyads in this study were friends and other half was strangers. Experimental treatments were randomized across friend versus stranger dyads. Preliminary analyses did not detect any interaction effects between the factor for friend versus stranger with any of the hypothesized variables (although some main effects of friend vs. stranger appeared on dimensions of liking, unsurprisingly). Therefore, no further analyses or discussion follows with respect to this factor in the remainder of this work, the findings of which appear to generalize across both friends and strangers.

3. In the original experiment, a third response condition was also employed by some of the askers: complimenting the discloser. Analyses reported elsewhere indicate that compliments affected liking similarly to reciprocal disclosures. Because this form of response does not contribute conceptually to the current formulation, for simplicity and to enhance the power of the analyses, this response mode was not included in the current study. For more information, see Dai et al. (2014).