Display of major risk categories for android apps.

Abstract

Communicating cybersecurity risks to mobile-device users is essential. However, existing means of conveying the risks through detailed permission lists are ineffective. Risk indexes that convey overall risk are effective at influencing app-selection decisions, but many users want more information. We examined how users assess the risks associated with downloading applications on the Android platform by comparing various graphical formats of intermediate-level risk displays containing three risk categories: personal privacy; monetary loss; device stability. Bar-graph and table formats were compared, as were vertical and horizontal displays. Participants performed app risk-rating (Experiment 1) and app-selection (Experiments 2 and 3) tasks for hypothetical apps with risk scores on each of the categories. They also specified which risk category was of most concern to them. Increased risk scores in each category led to higher rated risk and lower app-selection rate which matched with self-reported risk concerns. Bar-graphs were more time-efficient and yielded higher risk-ratings than tables, although the two formats did not differ in the app-selection task. Moreover, horizontal bar-graphs yielded faster responses than vertical bar-graphs. The results indicate that the intermediate-level risk display was effective in conveying risk-category information, especially with horizontal bar-graphs, and personalized design of this display based on users' risk concerns is potentially useful. (PsycINFO Database Record

Comparing estimates for decision-making: Numerical processing and preferences for underestimates versus overestimates.

“…People also increasingly use screens and computer aided display in areas like healthcare (Kunar et al, 2017), geo-location (Hegarty et al, 2016), driving (Rizzardo et al, 2013), and learning (Johnson & Mayer, 2012). The proliferation of smartphones has made it easier to access numerical information at any time and any place (Chen et al, 2018), and simulations have made it possible to use complex data for estimations (Brunyé et al, 2014). In these technology-enriched contexts, people receive estimates in both point value and interval form (Padilla et al, 2015), and later compare estimates to observed values.…”

Psychological Effects and Their Role in Online Privacy Interactions: A Review

“…A ''high risk'' display was investigated in A6 [51]. Through three online experiments, this research aimed to evaluate the effectiveness of different representations of major risk categories -the intermediate level of risk representations (personal privacy, monetary loss, and device stability) in the context of mobile app installation.…”

“…In the context of mobile app installation, A6 [51] established that the ''low risk rating'' cue influenced application selection. The results showed that applications with lower risk ratings were more likely to be chosen for installation.…”

The Effects of Security Framing, Time Pressure, and Brand Familiarity on Risky Mobile Application Downloads

“…Furthermore, Rajivan and Camp tested the effectiveness of different iconography on promoting safe decision making and found that locks resulted in the safest choices, likely due to the familiarity of the icon and associated mental models with web browsers. Alternatively, Chen et al (2018) investigated intermediate-level risk displays containing three major risk categories as opposed to the summary risk/safety ratings in the research by Chen et al (2015), Rajivan and Camp (2016), and Chong et al (2018), and found that displaying these risk categories promoted secure app selections and allowed for potential personalization of risk displays based on the user's risk concerns. Finally, a study by Shuster et al (2015) tested users' preference for risk presentation between automated reports and human reviews.…”

“…Based on the ICC calculated for the current study's data, a higherorder analysis, such as HLM (Lindley & Smith, 1972) was deemed unnecessary. Rather, a generalized linear mixed-effects regression with the random intercept effects for participants and app function was used in R to analyze choice data (Baayen, 2008;Chen et al, 2018). Participants and app function served as random factors to account for variability not explained by brand familiarity, security score, time pressure, or security framing.…”