We posted on SSRN a working paper co-authored by Heng Xu and Nan Zhang, “From Contextualizing to Context-Theorizing: Assessing Context Effects in Privacy Research”.

People’s privacy attitudes and concerns are often diverse, dynamic, and situation specific. Although the contextual sensitivity of privacy is theoretically established, the assertion deserves substantial qualification from a methodological perspective. In this research, we take a first step towards examining the multiplicity of contexts and their effects on privacy attitudes. Specifically, we develop a two-dimensional framework for explicating how two pronounced effects of context, range restriction and uncertainty shift, alter the grounds on which people ascribe meanings to “privacy concerns”. We use this insight – implicit in some of the privacy literature, but neither widely recognized nor empirically understood – as the starting point for the theoretical substance and methodological demonstration of each dimension. We operationalize this two-dimensional framework by introducing a recent methodological advance in survey data analysis, the covariates in a uniform and shifted binomial mixture (CUB) model, to quantitatively assess the magnitude of heterogeneity and uncertainty occasioned by a given context. We illustrate the usage of our framework with a public survey dataset, and conclude with a set of recommended practices for future privacy research.

This research makes two important contributions to the privacy literature. First, we quantitatively demonstrated how contextual factors can substantially shift the distributional properties of people’s stated privacy concerns. The surprisingly wide dispersion of context effects substantiated the need for researchers to pay special attention to how they contextualize a privacy study, especially when examining the generalizability of their research findings across contexts. Second, our findings provided a diagnostic device for comparing and contrasting the effects of different contexts, so as to unpack the distributional properties of the self-reported privacy concerns, and impute from them two important dimensions of context-contingent shifts. It introduces an emerging method for survey data analysis with an easy-to-use visualization tool that provides important hints as to how context shapes people’s self-reported privacy concerns and how it governs the conditioning of privacy concern-behavior relationships.

Here is the link to the working paper.