JMIR Publications

ABSTRACT

Background:

eHealth technologies aim to change their users’ health-related behavior. Persuasive design and system features can make an eHealth technology more motivating, engaging or supportive to its users. The Persuasive Systems Design (PSD) model poses software features that have the possibility to increase such persuasiveness of technologies. However, the effects of specific PSD software features on the effectiveness of an intervention are still largely unknown. In order gain insight into the working mechanisms of persuasive technologies, the Perceived Persuasiveness Questionnaire (PPQ) was developed. While the PPQ seems to be a suitable way for measuring subjective persuasiveness, it needs to be further evaluated in order to determine how suitable it is for measuring perceived persuasiveness among the general public.

Objective:

Objectives of the current paper are to evaluate the face and construct validity of the PPQ; to identify points of improvement; and to provide suggestions for further development of the PPQ.

Methods:

In this study, an online closed-ended card sort study was performed, wherein participants grouped existing PPQ items under existing PPQ constructs. Participants were invited via a Massive Open Online Course (MOOC) on eHealth. A total of 398 people (average age 44.15 years, 63% male) completed the card sort. Face validity was evaluated by determining item level agreement of the original PPQ constructs. Construct validity was evaluated by determining (a) in what construct each item was placed most often, regardless of the original placement and (b) how often two items were (regardless of the constructs) paired together and what inter-item correlations were according to a cluster analysis.

Results:

Four PPQ constructs scored a relatively high face validity: Perceived Social Support, Use Continuance, Perceived Credibility and Perceived Effort. Item level agreement on the other constructs was relatively low. Item level agreement for almost all constructs, except for Perceived Effort and Perceived Effectiveness, would increase if items would be grouped differently. Finally, a cluster analysis of the PPQ indicated that the strengths of the newly identified nine clusters varies strongly. Unchanged strong clusters are only found for Perceived Credibility Support, Perceived Social Support and Use Continuance. The placement of the other items was much more spread out over the other constructs, suggesting overlap between them.

Conclusions:

The findings of the current study provide a solid starting point towards a re-designed PPQ that is a true asset to the field of persuasiveness research. To be able to achieve this, we advocate that the re-designed PPQ should stick more closely to what persuasiveness is according to the PSD model, and to the mental models of potential end-users of technology. The revised PPQ should for example enquire if the user thinks anything is done to provide task support, but not how this is done exactly.