Persuasive Technology: An Investigation of Efficacy Using Process Tracing

Persuasive technology involves the influence of people's attitudes and beliefs by a computerized system (Fogg, 2003). Models of persuasive technology have been proposed in the literature, but efforts to validate or evaluate the effects of these models has been lacking. The current work attempts to address this lack of empirical investigation by drawing from similarities between the persuasive and decision making literatures. The decision making literature includes information grids, a type of process tracing method, to capture the extent of systematic processing during decision making. The effects of persuasive technology on decision making were evaluated using process tracing methods within a game. Study one investigated whether exposure to persuasive technology influences how individuals acquire information to make a decision. It also investigated whether differences in the focus (e.g., consider an attribute or consider an alternative) and the inclusivity (e.g., consider all alternatives or consider a specific alternative) of the persuasive technology suggestion influences processing differently. Results suggested that although both alternative-focused and attribute-focused persuasive technology manipulations influence the decision outcome, only alternative-focused persuasive technology influenced systematic processing. Persuasive technology that provided an exclusive recommendation influenced systematic processing. Exclusive manipulations did not influence systematic processing. Study two replicated study one but also investigated the persistence of the persuasive technology by manipulating how many times participants were exposed to the persuasive elements. In general, findings were consistent with study one. PT influenced the decision made. Exclusive PT influenced systematic processing but inclusive PT did not. Manipulation of the amount of exposure to the persuasive technology influenced the effect of the exclusive condition, but there was no effect on the inclusive condition. Implications are discussed.