The Effects of Interactivity on Information Processing and Attitude Change: Implications for Mental Health Stigma

Abstract

Interactive media such as the Web have become a popular and important vehicle for communicating health information. However, little attention has been given to theorizing and empirically testing the effects of interactive media and the theoretical construct of interactivity. In this paper, we clearly identify and define the nature of interactivity examined. We then develop and test a theoretical model of website interactivity on information processing, involvement with communication, and attitude change in the context of stigma of mental illness. The results of an experiment revealed that interactivity of the website had positive main and moderating effects on dependent variables, while involvement with communication played a significant role in explaining the effects of interactivity. Implications for future research and for health communication campaigns for mental illness stigma are discussed.

ACKNOWLEDGMENTS

The study reported here is part of the first author's dissertation research. The first author thanks the members of her dissertation committee for their insightful comments and support. She also thanks the anonymous reviewers for their constructive comments.

Notes

¹Some researchers have pointed out potential negative effects of interactivity due to users' increased cognitive load and disorientation in an interactive communication environment. They viewed interactivity as a technological feature of an interactive medium and attributed users' cognitive load and disorientation to overstimulation by multimodality and other interactive features of the medium (Bucy, 2004; Sundar, 2007). In this study, we predict positive effects of interactivity on cognitive and attitudinal outcomes as we conceptualize and design interactivity as a central communication property (vs. a peripheral medium feature) that facilitates message processing.

²Homeless people were chosen as another social group to manipulate (low) interactivity, particularly for the dimension of message relevance and coherence. Previous research showed that people with schizophrenia and homeless people are two distinct social groups and yet they are perceived to share common characteristics and stigmatized in similar ways (Breakey, Fischer, Nestadt, & Romanoski, 1992). The results of manipulation checks and additional analyses on homeless people in this study supported this view, as the use of homeless people created perceptions of low message relevance and coherence, but perceptions of one social group (e.g., people with schizophrenia) influenced stigmatizing attitudes toward the other social group (e.g., homeless people).

³All participants were asked about their preexisting stereotyped perceptions of the two social groups during the experiment. However, depending on the interactivity condition, preexisting stereotyped perceptions of one social group were measured in the premanipulation questionnaire, while preexisting stereotyped perceptions of the other social group were measured during the exposure to the stimulus website. The study was designed in this manner in order to facilitate the interactivity manipulation (see the Interactivity Manipulation section), while ensuring that all participants respond to the same questions over the course of the experiment.

⁴The pretest of the interactivity manipulation showed that participants in the high interactivity condition viewed the website fairly randomly. In the main study, therefore, participants in the low interactivity condition viewed the stimulus webpages in a random order (vs. a predetermined linear sequence) in the attempt to approximate the high interactivity group's viewing sequence and to control for any order effects.

⁵Samples of the stimulus webpages can be obtained by contacting the lead author.

⁶Selective scanning of information as a result of high user control over information sequence has been found to confound the effects of interactivity in previous research (Tremayne & Dunwoody, 2001). In this study, selective scanning was prevented in both interactivity conditions, while all participants were exposed to the same stimulus content in the end. Multiple exposures to the same content were also prevented in attempt to control the “dose” of the stimulus exposure. For instance, the “Go Back” function of the Web browser was disabled and the links to the visited webpages were deactivated. Participants were also asked to view each webpage only once.

⁷Considering this study's focus on the moderating effects of interactivity on general stereotyped perceptions of people with schizophrenia rather than subdimensions of stereotyped perceptions, the 13 items were combined to constitute a single independent variable (preexisting stereotyped perceptions of people with schizophrenia).

⁸Full statistical information for the factor analysis can be obtained by contacting the lead author.

⁹Correct answers were scored as +1, incorrect as −1 (to compensate for guessing behavior), and “don't know” as zero (Mick, 1992). Even though questions were based on the content of the story, participants could have known about people with schizophrenia and answered the questions correctly without having reviewed the stimulus website. Therefore, a level of difficulty for each question was assessed by calculating the mean of each question and subtracting it from one (Rimal & Kim, 2002). Each correct response was then weighted in proportion to its level of difficulty for use in subsequent analyses. The aggregated weighted responses ranged from −4.55 to 8.04 (M = 2.01, SD = 2.96).

¹⁰It was possible that the low interactivity manipulation (the lack of user control and irrelevance of information) would result in participants' frustration and unfavorable responses to dependent variables. In order to examine the potential confounding effects of the low interactivity manipulation, attitudes toward the stimulus website were compared between the two interactivity conditions. Participants indicated the extent of agreement with four items developed for this research (e.g., “It is a good website,” “It was difficult to concentrate and follow the story”) (α = .81). A t-test showed that the low interactivity group's attitudes toward the website (M = 6.01, SD = 1.44) was not significantly different from the high interactivity group (M = 6.19, SD = 1.49, t(111) = .65, p > .52).

¹¹Full statistical information for the hierarchical regression models can be obtained by contacting the lead author.

¹²A covariate was either retained or removed by the stepwise method at the initial step of the hierarchical regression analysis. For some models, all covariates were removed by the stepwise method, in which case two independent variables were entered at the first step.

¹³Although continuous independent variables are sometimes centered before creating a product term, the continuous independent variable of this study (i.e., preexisting stereotyped perceptions of people with schizophrenia) was not centered because the product term involved a categorical variable (i.e., dummy-coded interactivity) and multicollinearity was not detected (Aiken & West, 1991; Cohen, Cohen, West, & Aiken, 2003).