Technology use and norm change in online privacy: experimental evidence from vignette studies

ABSTRACT

We suggest that explaining privacy behaviors requires understanding not only individual attitudes, but also norms and trust. We propose: (1) the popularity of a potentially privacy-violating technology leads individuals to expect that others approve of privacy violations and simultaneously increases their trust in the technology provider; (2) the frequency of privacy violations by other similar providers leads individuals to expect that a specific provider will engage in privacy-violating behaviors and decreases trust in that provider; (3) trust in a specific provider and expectations that others approve of the provider violating users’ privacy increase, and expectations that other similar providers are likely to violate privacy decrease, willingness to use a technology. We test our propositions using two vignette experiments in the context of a household energy app. Our results are generally consistent with our hypotheses. Our findings have implications for understanding privacy norms and highlight the potential consequences of major technology roll-outs.

KEYWORDS:

• Descriptive norms
• trust
• smart grid
• vignette experiment
• normative expectations

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 To determine the necessary size of the representative sample, we conducted a power analysis based on the results of the convenience sample study. We conducted a power analysis for a two-sample mean test based on the means and standard deviations in the respective treatment groups with no and with frequent occurrences of privacy violations among providers of similar apps (the factor that yielded the smaller treatment effect). Assuming a significance level of 5% for two-sided tests and a power of 80%, we obtain an estimated sample size of 151 participants per condition. However, this calculation was based on standard deviations estimated on the rather homogeneous online panel and hence likely too low. We therefore increased the standard deviations by one third to obtain an estimated sample size of 257 per condition. Finally, to be on the safe side, we rounded up the sample size to 300 participants per condition or 1200 participants in total.

2 Before conducting the studies reported here, we collected pilot data in which we used different vignette wording. The pilot data show that if people have reason to think that the technology provider is already violating users’ privacy, then popularity of the app does not affect normative expectations. That is, popularity matters when people know it is possible for the technology provider to violate privacy, not when there is evidence that the provider has already done so.

Additional information

Funding

This work was supported by National Science Foundation grant numbers 1441357, 1555980.

Notes on contributors

Christine Horne

Christine Horne is professor of sociology at Washington State University. She studies the emergence, enforcement, and application of social norms. Her work has been published in the American Sociological Review, Annual Review of Sociology, Social Psychology Quarterly, and Energy Policy.

Wojtek Przepiorka

Wojtek Przepiorka is assistant professor of sociology at Utrecht University. His research interests are analytical and economic sociology, game theory, organizational behavior and quantitative methodology. His work has been published in the American Journal of Political Science, American Sociological Review, and Social Forces.