Towards Understanding Connections between Security/Privacy Attitudes and Unlock Authentication

TL;DR

This study explores how user attitudes towards privacy and security influence the strength of Android's graphical unlock patterns, revealing some correlations but highlighting the need for better measurement methods.

Contribution

It provides an empirical analysis of the relationship between privacy/security attitudes and unlock authentication choices using a large online survey.

Findings

Attitudes towards data concern and unauthorized access are significant factors.

Demographic differences influence privacy/security responses.

Limited correlation between attitudes and authentication strength was observed.

Abstract

In this study, we examine the ways in which user attitudes towards privacy and security relating to mobile devices and the data stored thereon may impact the strength of unlock authentication, focusing on Android's graphical unlock patterns. We conducted an online study with Amazon Mechanical Turk ($N=750$) using self-reported unlock authentication choices, as well as Likert scale agreement/disagreement responses to a set of seven privacy/security prompts. We then analyzed the responses in multiple dimensions, including a straight average of the Likert responses as well as using Principle Component Analysis to expose latent factors. We found that responses to two of the seven questions proved relevant and significant. These two questions considered attitudes towards general concern for data stored on mobile devices, and attitudes towards concerns for unauthorized access by known actors. Unfortunately, larger conclusions cannot be drawn on the efficacy of the broader set of questions for exposing connections between unlock authentication strength (Pearson Rank $r=-0.08$, $p<0.1$). However, both of our factor solutions exposed differences in responses for demographics groups, including age, gender, and residence type. The findings of this study suggests that there is likely a link between perceptions of privacy/security on mobile devices and the perceived threats therein, but more research is needed, particularly on developing better survey and measurement techniques of privacy/security attitudes that relate to mobile devices specifically.