Content uploaded by Jing Chen
Author content
All content in this area was uploaded by Jing Chen on Sep 22, 2023
Content may be subject to copyright.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 1
Framing Security under Time Pressure: Brand Familiarity Matters for Mobile
Application Choices
Jing Chen1 & Cody Parker2
1Department of Psychological Sciences, Rice University
2Department of Psychology, Old Dominion University
Author Note
The authors have no known conflict of interest to disclose. The authors thank Jeremiah
Still and Abby Braitman for their helpful comments on an earlier version of this paper.
Correspondence should be addressed to Jing Chen, jingchen@rice.edu, Department of
Psychological Sciences, Rice University, 6100 Main Street (MS-25), Houston, TX 77005.
https://orcid.org/0000-0003-0394-0375
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 2
Abstract
The current study examined the effects of security score framing, time pressure, and brand
familiarity on mobile application choices. Past research has found the framing of safety versus
risk scores affects how potential risks for mobile apps is communicated to users. Both time
pressure and brand familiarity have been shown to affect consumers’ purchase behaviors but not
yet for app-selection decisions. The current study examined the effects of time pressure and
brand familiarity on the effectiveness of risk displays (framed as safety or risk) for mobile apps.
Participants were shown screenshots of various apps with these factors manipulated, and they
were to choose one out of six apps. Our findings indicate that users rely heavily on brand
familiarity when choosing apps, which could lead to insecure decisions. Additionally, security
scores guided app choices towards more secure apps when framed as safety than when framed as
risk, although this advantage was only evident without time pressure and disappeared under time
pressure. The design implications call for more careful screening and user education about the
potential risks associated familiar apps, as well as the need of new security design solutions to
help users under time pressure.
Keywords: mobile application; security framing; time pressure; brand familiarity
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 1
Framing Security under Time Pressure: Brand Familiarity Matters for Mobile
Application Choices
The ubiquity of smartphones and their contained applications (apps) cannot be disputed,
with an estimated 6.26 billion smartphone users worldwide as of 2021 (Statista Research
Department, 2022) as well as an estimated 3.5 million android apps and 2.2 million for iOS
available as of August 2022 (Ceci, 2022). With this large numbers of mobile users and apps, it is
critical that users choose the apps that not only function as expected and provide satisfactory user
experience (Liu et al., 2021; Park et al., 2013) but also are secure to use. Unfortunately, there are
apps that gather user data through illegal or unethical practices, enabled in part by imperfect
screening of apps and a lack of consumer scrutiny (Price, 2018). On the one hand, app stores
should strive to better screen apps; on the other hands, users can be informed to make secure
decisions regarding the apps they choose to download onto their devices. These types of
decisions can be fostered through effective communication of the potential risks associated with
mobile apps to the users (Chen, 2020; Chen et al., 2015; Kang et al., 2015).
Users’ decision making can be influenced by various factors. We identified factors that
are potentially influential and directly manipulable in mobile apps, which are closely relevant to
the majority of the population. The current study aimed to examine these factors for mobile app
selections to inform risk display designs, with a goal to bolster secure user decision making.
These factors include how information is framed (Kahneman & Tversky, 1979; Tversky &
Kahneman, 1981), the presence of time pressure (Young et al., 2012; Saqib & Chan, 2015), and
misplaced trust in brand familiarity (Baker et al., 1986).
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 2
Risk Communication for Mobile Apps
The number of mobile apps downloaded by users has steadily increased over the past few
years, reaching 255 billion worldwide in 2022 (Ceci, 2023). There is an average of 2,453 new
apps released per day in the Google Play Store and 1,277 in the Apple App Store in 2023. Users
have different motivation while using their mobile devices (42matters, 2023a, 2033b). For apps
to function properly, they may need access to various sensors and folders of information within a
device, such as location, contacts, and the camera (Felt et al., 2012). While certain permissions,
such as location for GPS navigation apps, are legitimate when required for the app to function,
others may be unrelated to the app’s function. For instance, there are apps that may use the data
for ulterior motives or sell data to third parties (Vidas et al., 2011), or gather such information
when the app is not actively in use (Nakashima, 2018).
Given the gravity of such data abuse, it is concerning that smartphone users tend not to
investigate or fully understand permissions requested by the apps when downloading new apps
(Chin et al., 2012; Felt et al., 2012; Kelley et al., 2012; Benton et al., 2013). Although both
Android and iOS provide some safeguards in the format of app permissions, smartphone users
often ignore or are unaware of permission settings (Baarslag et al., 2016; Kang et al., 2015; Tay
et al., 2021), and very few are fully aware of the data collected by the apps they install on their
mobile devices (Almuhimedi et al., 2015). One way to educate and empower users is to provide
them with simple, explicit displays of risks for mobile apps (Chen et al., 2015; Choe et al., 2013;
Chen et al., 2014; Tay et al., 2021). These studies have proposed designs for security scores that
summarize the safety or risk associated with apps, based on their permission requests. This
approach is expected to reduce risky, uninformed app choices and increase safer choices.
However, the framing of such information is of vital importance: whether the system should
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 3
utilize a safety score (the greater the score, the safer the app) or a risk score (the greater the
score, the more dangerous the app) (Chen et al., 2015; Chong et al., 2018).
Past research on the risk displays for mobile apps has suggested that the framing of the
risk information can influence users’ app-selection behaviors (Chen et al., 2015; Chong et al.,
2018; Rajivan & Camp, 2016). For example, Chen et al. (2015) included a summary risk/safety
rating for each app and showed that the rating influenced app selection. They also found that
compared to the risk framing, the safety framing was more effective, in terms of eliciting more
choices of safer apps. Rajivan and Camp (2016) and Chong et al. (2018) also found that, beyond
positively framing a security system with safety scores, users can be influenced to make more
secure decisions when primed to think about cyber security prior to choosing an app. In addition,
Rajivan and Camp (2016) tested the effectiveness of different iconography on promoting safe
decision making, and they found that images of locks resulted in the safest choices, likely due to
the familiarity of the icon and existing mental models for lock icons used in web browsers to
denote security.
Effects of Time Pressure on Decision Making
Time pressure can affect the way people make decisions by inducing a sense of urgency
when attempting a task (Klapproth, 2008; Liu et al., 2016; Young et al., 2012). The general
decision-making literature has examined the effect of time pressure on risk taking for decades,
yet with a plethora of competing findings. For example, Ben Zur and Breznitz (1981) showed
that, under time pressure, participants were risk averse with hypothetical gambles and focused on
the negative aspects of each gamble. Similarly, El Haji and colleagues (2016) showed that,
compared to no time pressure, people are less likely to bid on a lottery (i.e., more risk averse)
under time pressure. On the other hand, Chandler and Pronin (2012) found that, after being
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 4
prompted to read sentences at a fast pace, participants were more risk-seeking when completing
the Balloon Analogue Risk Task (BART) than their slow-paced counterparts. Similarly, Madan
and colleagues (2015) showed that participants under time pressure were more risk-seeking in a
gambling task. Given these mixed results, further research is needed to understand the influence
of time pressure on risky decisions for decisions made under gain and loss frames.
In consumer decision-making literature, Wright’s (1974) foundational research showed
that, under time pressure, consumers tend to emphasize negative traits of a product. Since then,
other research in various shopping contexts suggests that manipulations of time pressure, such as
via scarcity of products (Soliman, 2017; Devlin et al., 2007) or length of sale (Aggarwal &
Vaidyanathan, 2003), can dictate the strategy with which consumers approach purchases
(Chowdhury et al., 2009; Vlašić et al., 2011) and their acceptance of risk (Shehryar, 2008).
Indeed, time pressure has been shown to impact the ability of consumers to investigate product
information (Kardes et al., 2006; do Prado & Lopes, 2016) and reduces the amount of time they
spend browsing unfamiliar products (Liu et al., 2017). A recent study using eye-tracking
measures also showed that participants made faster decisions with fewer eye fixations when
making purchases under time pressure, and five-star rating products were chosen more often
under time pressure (Ammons et al., 2022).
For mobile app downloads specifically, there has been limited research or report on the
scenarios in which users install apps under time pressure or how time pressure affects users’ app
choices. Time pressure was manipulated in an MTurk study, where users were displayed with a
free app with advertisements and a paid app without advertisements (Dinsmore et al., 2021). The
researchers manipulated time pressure through instructions conveying a limited amount of time
and a visible clock ticking off 30 seconds while they viewed descriptions of the two apps. They
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 5
found that under time pressure, participants had a stronger preference of the paid app when it is
presented first than when it is presented second, and this effect was affected by perceived risks
associated with the decision. A study on manipulative design features in video games for
children found that time pressure was used induced in-app purchases or prolonged gameplay in
about 28% of the game apps studied (Radesky et al., 2022). The researchers found that apps use
visual indicators to convey scarcity of time, using countdown indicators or messages such as
“Limited time only!” This finding is consistent with recent study that conducted interviews on
users in emerging countries and identified time-limited offers to be one of the main drivers of in-
app purchase (Buzulukova & Kobets, 2022).
Effects of Brand Familiarity on Consumer Behavior
As previously mentioned, threats to mobile users can come in a variety of forms, from a
variety of sources. While some lesser-known developers have been caught writing malware into
their apps (Price, 2018), more familiar developers may also choose to collect and sell user data to
third parties (Wong, 2019). Both forms of exploitation are cause for concern but may be
represented differently in the minds of users, with more familiar apps seeming more innocuous
(Harris et al., 2016). Indeed, brand familiarity both increases brand satisfaction, and drives
purchase behaviors (Baker et al., 1986; Ha & Perks, 2005; Menon & Kahn, 2002). Brand
familiarity also reduces the time that consumers spend when shopping indicates a reduction in
the need for information search (Biswas, 1992). Stocchi et al., (2019) studied the motives behind
using branded mobile apps, and found that branded apps will have greater usage if they are
viewed as protecting users’ privacy and supporting what they do. Pasaribu et al. (2013)
conducted a survey and found consumers’ attitude toward online advertising and brand
recognition on social media affected purchase intension. A recent study by Kumar and Tuli
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 6
(2021) explored how and why users engage with branded mobile apps. They found that privacy
and security, as well as brand-related aspects (e.g., popularity and reputation) of an app were
among the main reasons for users’ cognitive and affective engagement with the app. However,
this finding was based on subjective reports in user interviews.
To date, though, no studies have examined the effect of brand familiarity on the summary
security scores proposed by previous studies (Chen et al., 2015; Choe et al., 2013; Chen et al.,
2014). However, previous literature shows a strong effect of brand familiarity on purchase
intentions in more general consumer domains (Laroche et al., 1996; Park & Stoel, 2005).
Therefore, because of the effect brand familiarity has on purchase intentions, it is expected that
brand familiarity may be influential in guiding app choice.
Current Study
The current study aimed to further investigate the security score display proposed by
Chen et al. (2015), the role that time pressure plays in the decision-making process, and the
degree to which brand familiarity predicts app selection preferences. All the previously listed
literature on mobile app decisions (Chen et al., 2015; Chong et al., 2018; Gates et al., 2014;
Rajivan & Camp, 2016) controlled for brand familiarity by removing the top search results from
inclusion in the experimental stimuli, and had participants perform the task using as much time
as they needed. Under typical condition, it is reasonable to assume that users install application
due to time pressure and that a major determinant may be brand familiarity. However, it is
realistic that people install apps under time pressure and that they may be influenced by brand
familiarity. For example, a user may be in great time pressure to download a cash transfer app
(e.g., Venmo) when they need to pay a contractor waiting in their house but have no cash at
hand. This critical decision to download this app is certainly under time pressure. Users may also
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 7
mistrust apps of familiar brands (Baker et al., 1986; Ha & Perks, 2005). To the best of our
knowledge, this study is the first to consider time pressure and brand familiarity in mobile app
decisions. If time pressure or brand familiarity impacts the decision-making strategies of mobile
users when choosing apps like they do in other domains, the design of the previously proposed
security scores may need to be reconsidered.
The Pilot
To obtain brand familiarity ratings for the mobile apps used in the main experiment, a
pilot study was first conducted by having a total of 287 participants (217 female, 69 male, one
declined to identify; age M = 21.72, SD = 4.86) rate their familiarity with a series of apps,
among other questions. They were recruited from Old Dominion University’s online SONA
system and granted partial course credit for their participation. A total of 25 app-function
categories with 12 apps for each function were chosen (banking, bowser, dating, drawing, e-
reader, fitness, food delivery, games, housing, language, local business reviews, navigation,
messaging, money transfer, music, news, notes, PDF scanner, photo editing, ride sharing,
shopping, social media, travel, video streaming, weather). Participants rated the statement of
“This app is familiar” on a 7-point Likert scale, with a score of 1 representing Strongly
Disagree, 7 representing Strongly agree, and 4 being Neutral. Three app-function category,
flashlight, calendar, and clock that were not planned to be used in the main experiment, were
used to serve as attentional catch trials, on which participants were instructed to always rate
“disagree.” A total of 171 participants correctly answered at least two of the three catch trials
(75% of them answered all three correctly), and participants missed the three catch trials roughly
equally. This suggests that participants missed the attention checks not due to one of them being
particularly confusing. Data from these participants were used to select the apps. For the
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 8
remaining 24 app-function categories, the apps with the three lowest scores (M = 1.77, SD =
0.30) and those with the three highest scores (M = 4.83, SD = 0.30) were chosen from each
function for use in the main experiment. A one-way ANOVA comparing the lowest versus the
highest scores for each app indicated a significant difference between the two app groups,
F(1,148) = 423.88, p < .001,
p2 = .74, MSE = 2.86.
The Main Experiment
During the main experiment, participants were presented with multiple choice screens,
each of which contained an assortment of apps including the three least and three most familiar
apps for each app-function category identified in the pilot. The independent variables were
security framing (between-subjects), security score (within-subjects), time pressure (between-
subjects), and brand familiarity (within-subjects). Security scores were framed as either safety
using closed locks or risk using open locks, consisting of one, two, three, four, or five locks;
safety scores indicated higher levels of safety (less risk) with increasing locks, while risk scores
indicated higher levels of risk (less safety) with increasing locks. The brand familiarity scores
were determined before the experiment by the pilot study, which were not shown to the
participants. The dependent variable was app choice. Based on prior studies, the main hypotheses
of the current study were generated as follows. While other main effects and interaction effects
could be interesting, we did not have specific hypotheses regarding them, but included them in
the analyses for exploration.
Hypothesis 1: For the main effect of security scores on choice, increases in safety scores
would positively predict app choice, such that, with each additional lock, a participant would be
more likely to choose an app; conversely, increases in risk scores would negatively predict
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 9
choice, such that, with each additional lock, the app would be less likely to be chosen (Chen et
al., 2015; Chong et al., 2018; Rajivan & Camp, 2016).
Hypothesis 2: For the main effect of brand familiarity on app choice, brand familiarity
would positively predict choice, with less familiar apps having a lower likelihood of being
chosen, due to a greater sense of trust born out of familiarity (Ha & Perks, 2005).
Hypothesis 3: For the interaction between security framing and security score on choice,
the security scores under the safety frame would be more impactful on choice (i.e., safer apps
would be chosen more often and less safe apps would be chosen less), while security scores
under the risk frame were expected to less clearly guide decision making, possibly due to a
confusion of the score’s meaning (Chen et al., 2015).
Hypothesis 4: For the interaction between brand familiarity and time pressure, compared
to participants without time pressure, the association between brand familiarity and likelihood of
being chosen would be stronger for those under time pressure. This result was expected due to
browsing behaviors shown by Liu et al. (2017), wherein participants under time pressure focused
more on familiar brands with greater observation durations and counts than on their competitors.
Hypothesis 5: There would be an interaction between security scores and brand
familiarity on app choice. Based on the strong effect of brand familiarity (Ha & Perks, 2005), it
was expected that the effect of security score on likelihood of being chosen would be stronger for
apps with lower brand familiarity than for those with higher brand familiarity. In other words,
participants would be more reliant upon safety (risk) scores for apps with lower brand familiarity
ratings than the more familiar apps, if they chose apps with low brand familiarity scores.
However, the lack of literature on this interaction means that this hypothesis was exploratory.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 10
Hypothesis 6: There would be an interaction between security score and time pressure.
Because there has been no direct research on how time pressure would affect the effect of
security scores, this hypothesis was based on Ammons et al.’s (2022) finding that five-star rating
products were chosen more often under time pressure. An assumption was that the security
scores would work similarly to user ratings when the latter was controlled during the experiment.
It was expected that participants would choose apps with higher security scores more often with
time pressure.
Method
Participants
A total of 128 participants (51 females, 77 males; age M = 40.96, SD = 12.25) were
recruited via Amazon’s Mechanical Turk (MTurk) and were compensated $1 each for their
participation. The MTurk participants were required to live in the United States of America and
to have a 95% approval rate of their prior Human Intelligent Tasks (HITs) in order to ensure
quality data collection (see Peer et al., 2014 for recommended qualification requirements). This
study was approved by the Institutional Review Board of Old Dominion University.
Materials
The study was hosted and accessed via Qualtrics. Devices were controlled such that only
laptop or desktop computers could be used; this ensured the proper display of the stimuli. For the
app download screens, the design replicated the desktop version of the Google Play Store at the
time of the study (July, 2020) with additional manipulations for the purpose of this study.
Participants were instructed to choose apps as if they were for their own device. On each trial,
participants were shown six apps displayed on one screen. The function of the apps varied from
trial to trial, from social media to note-taking to weather apps, but the apps on each trial had the
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 11
same function. Participants were informed that the app information displayed in the study were
hypothetical. Each app had a user rating (controlled at four stars for all apps) and a security score
(randomized across apps), framed as risk (negative) or safety (gain, shown). The apps were
displayed in two rows of three apps, and the position of the apps on the screen was randomized
across the six positions.
At the end of the experiment, an exit survey examined participants’ subjective reasoning
for app choice and thoughts regarding the experiment. The survey consisted of rationale
questions for app selection, the CyberDOSPERT (Kharlamov et al., 2018), a question that asked
the meaning of the locks, a color-blindness question, Likert-scale questions regarding
cybersecurity expertise, and an open-ended prompt for additional feedback.
Design of experiment
The independent variables included brand familiarity (low, high), security framing (risk,
safety), security score (one, two, three, four, or five locks), and time pressure (present, absent).
Security framing and time pressure were between-subjects, and security score and familiarity
score were within-subjects. This design led to four experimental groups, with 32 participants in
each group. Brand familiarity was determined based on the data obtained in the pilot study, and
the three apps with the lowest scores and three with highest scores for each of the 25 app-
function categories used in the pilot were chosen to represent apps with low or high familiarity.
Among the 25 categories, “drawing” was randomly chosen to be used on an attentional check
trial, and the remaining on experimental trials. The attentional check trial intended to consist of
five apps with low brand familiarity and low (high) safety (risk) scores and one app with high
brand familiarity and a high (low) safety (risk) score, and participants were expected to choose
the one familiar, secure app among the less familiar, riskier options. However, due to a
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 12
programming error, the safety/risk scores were distributed as other apps (randomly assigned as 1,
2, 3, 3, 4, and 5 for each of the 6 apps), so this attentional check did not work as expected (see
results).
The security scores were framed as risk or safety and consisted of colored locks (see
Figure 1). The safety score was represented by the number of teal, locked locks, and the more
locks represented more safety; the risk score was represented by pink, unlocked locks, and the
more locks represented more risks. The use of locks was inspired by the design of Rajivan and
Camp’s (2016) study. The risk and safety scores were inversions of each other (e.g., a risk score
of two would be a safety score of four). Security framing was manipulated by presenting either
the safety or risk scores using locks (see Figure 1). Those in the safety frame saw scores
represented by one to five closed, teal locks (with more locks meaning safer apps), and those in
the risk frame saw scores represented by one to five open, pink locks (with more locks meaning
greater risk). Teal and pink coloring were chosen because they are not only discernible for those
with red-green color blindness, but also similar to green and red, which have associated “go and
stop” meanings (Bergum & Bergum, 1981). Among the six apps on each trial, five possibilities
of the scores were presented during each trial and the middle scores (three locks) represented
twice each.
Time pressure was manipulated similarly to the Dinsmore and colleagues’ (2021) study
mentioned in Introduction. A countdown timer was included above the app options for those in
the time-pressure condition, whereas this countdown timer was absent for those in the non-time-
pressure condition. In addition, in the instructions, those in the time-pressure condition were told
to make decisions quickly within the time provided on the countdown timer, whereas those in the
non-time-pressure condition was told to take as long as necessary to make their decisions. The
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 13
time allotted for each trial in the time-pressure condition was equal to the mean decision time of
that trial in the non-time pressure conditions (see Chen & Proctor, 2017). Finally, user ratings
were controlled by assigning a rating of four stars across all apps to minimize their effect on app
choices.
Figure 1
A safety score of 5 represented by teal, locked locks (top row), and a risk score of 5 represented
by pink, unlocked locks (bottom row). Safety/risk scores ranged from 1 to 5 in the experiment;
more locks represented more safety/risks.
Because risk taking behaviors were the focus of this study, the main dependent variable
was whether an app was chosen. Choices on all six apps were recorded, five of which had a
dependent variable value of zero (not chosen), while one was coded as a one (chosen). We also
measured decision times for the purpose of validating the time-pressure manipulation. The
decision time on a trial was recorded from the beginning of the trial until participants clicked a
button to advance the page.
Procedure
In terms of timing of the participants, the first half of the participants were assigned to the
non-time pressure condition and the second half to the time pressure condition, because the
countdown times for the time pressure condition were based on data from the non-time pressure
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 14
condition. Data for both conditions were collected within the same week. For both the time-
pressure and non-time-pressure groups, each participant was randomly assigned to one of the two
security framing conditions (safety, risk). At the beginning of the experiment, participants were
shown an instruction screen containing the different elements on the screen as well as
corresponding explanations. Different from those in the non-time pressure condition, participants
in the time pressure condition were instructed to make their decisions within the time provided in
the countdown timer.
On each trial, participants were instructed to choose an app as if selecting an app for their
personal mobile device. If a participant under time pressure took longer to make a decision than
the time provided by the timer, the timer ended and showed feedback asking them to respond
faster (about 14% of total trials). Participants’ choice was still recorded in this case. This setting
allowed for full data collection and was expected to still induce time pressure despite the lack of
consequence for slow responses. Participants performed 24 experimental trials, this number of
trials was modelled after that was used by Schuster and colleagues (2015), though the current
study added one attentional check trial. After completing all 25 trials, participants completed the
exit survey, described in Materials, which then concluded the study, and participants were
compensated.
Results
App Choice Data
Two participants’ choice data were excluded from the analysis due to incomplete data
acquisition in the demographics portion of the experiment (1.6% of total participant data). Due to
the programming error in the attentional check trial, no data were excluded based on this trial,
with the caveats discussed in the limitations of this study. Additionally, the trial containing the
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 15
weather apps failed to collect choice data for 19 participants (0.63% of the decision data); all
other data for these 19 participants were included in the analysis.
A generalized linear mixed-effects regression (GLMER) with the random intercept
effects for participants and app function was used in R to analyze the choice data (Baayen et al.,
2008; Chen et al., 2018). Participants and app function were included as random factors in an
initial analysis, but only app function served as a random factor in the reported analysis due to
the random factor variance for participants being zero. Demographic information was included in
the model as predictor variables to ensure there were no confounding variables that predicted app
choice. None of the demographic information significantly predicted choice, ps >.05.
The GLMER revealed the slopes of each predictor and likelihood ratio tests (LRT) were
conducted for each of the model’s terms to test the significance of the main effects and
interactions of the predictors (see Table 1). Therefore, while the GLMER’s coefficients provide
information regarding the differences in behavior between groups, the LRT provides information
about the strengths of the predictors themselves rather than in relation to a specific group. Due to
the substantial amount of data analyzed in the GLMER (and thus the large degree of freedom),
the subsequent LRT α level was set at .001 to ensure that the detected effects are not merely due
to chance (Chen et al., 2018; Miles & Shevlin, 2001).
Of importance to the study at hand, security scores were a significant predictor of choice
(see Figure 2), supporting Hypothesis 1, χ2(1) = 53.04, p < .001; the coefficient was 0.13, with a
95% confidence interval (CI) of [0.10, 0.17]. The mean percentage of an app with a security
score of 1, 2, 3, 4, or 5 being chosen was 14.8%, 14.0%, 13.4%, 15.7%, and 28.7%, respectively.
When transformed from a log likelihood into an odds ratio of 1.14 with a 95% CI of [1.10, 1.18],
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 16
the model predicted that, with each additional (reduced) safety (risk) lock, users are on average
1.14 times more likely to choose a given app.
Table 1
The generalized linear mixed-effects regression (GLMER) and likelihood ratio tests LRT Results
General Linear Mixed Effects Regression Likelihood Ratio
Test
Predictors Coefficient Lower CI Upper CI Z Value Odds
Ratio
Odds
Ratio
Lower CI
Odds
Ratio
Upper CI
χ2
Value p Value
Intercept -1.74 -1.80 -1.67 -26.50 0.18 0.16 0.19 - -
Frame -0.18 -0.25 -0.11 -2.68 0.83 0.78 0.89 3.57 0.059
Time Pressure -0.02 -0.08 0.05 -0.26 0.98 0.92 1.05 3.55 0.059
Security Score 0.13 0.10 0.17 3.74 1.14 1.10 1.18 53.04 < .001
Brand Familiarity 0.34 0.32 0.37 14.18 1.41 1.38 1.45 55.76 < .001
Frame × Time Pressure 0.16 0.07 0.26 1.76 1.18 1.07 1.29 4.22 0.040
Frame × Security Score 0.32 0.27 0.37 6.26 1.38 1.31 1.45 3.79 0.052
Frame × Brand Familiarity 0.12 0.09 0.15 2.41 1.13 1.09 1.17 11.45 < .001
Time Pressure × Securit
y
Score 0.12 0.07 0.17 3.57 1.13 1.07 1.19 8.15 0.004
Time Pressure × Brand Familiarity -0.01 -0.05 0.02 -0.41 0.99 0.95 1.02 0.00 1.000
Security Score × Brand Familiarity 0.004 -0.01 0.02 0.21 1.00 0.99 1.02 9.77 0.002
Frame × Time Pressure × Security
Score -0.44 -0.51 -0.37 -6.19 0.64 0.60 0.69 67.80 < .001
Frame × Time Pressure × Brand
Familiarity -0.06 -0.11 -0.01 -1.28 0.94 0.90 0.99 54.93 < .001
Frame × Security Score × Brand
Familiarity -0.06 -0.09 -0.04 -2.51 0.94 0.91 0.96 0.00 1.000
Time Pressure × Security Score ×
Brand Familiarity -0.03 -0.06 -0.01 -1.24 0.97 0.94 0.99 0.30 0.586
Frame × Time Pressure × Security
Score × Brand Familiarity 0.10 0.06 0.13 2.74 1.10 1.06 1.14 3.73 0.054
Random
Variance SD
App Function 0.05 0.23
Note. Bold texts mark significant effects at an alpha level of .001.
Brand familiarity was a significant predictor of app choice (see Figure 2), χ2(1) = 55.76, p
< .001; the coefficient was 0.34 with a 95% CI of [0.32, 0.37], with more familiar brands being
chosen more often, supporting Hypothesis 2. Transformed into an odds ratio, this means that,
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 17
with each increasing unit of brand familiarity score (out of seven total), users were 1.41 times
more likely to choose a more familiar app, with a 95% CI of [1.38, 1.45].
Figure 2
The effect of brand familiarity on app choice
Note. The main effect of brand familiarity was significant; the interaction between brand
familiarity and time pressure was not significant.
Conversely, the interaction between security scores and framing did not significantly
predict app choice (see Figure 3), not supporting Hypothesis 3, χ2(1) = 3.79, p = .052; the
coefficient was 0.32, with a 95% CI of [0.27, 0.37]. Note that the result pattern was in the same
trajectory as prior studies (Chen et al., 2015; Chong et al., 2018), with safety framing tending to
have a stronger effect (indicated by a steeper slope of the safety line) than risk framing on
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
1234567
MeanPercentofTimethatAppwasChosen
(%)
BrandFamiliarityScore
TimePressureAbsent
TimePressurePresent
Linear(TimePressure
Absent)
Linear(TimePressure
Present)
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 18
choice. One reason for the nonsignificant finding for the interaction between security scores and
framing could be because this analysis averaged data across the time pressure conditions.
Figure 3
The effect of the interaction between security scores and frame on app choice
Note. This interaction effect was not significant. Error bars represent 95% between-subjects
confidence intervals. Security Score 1 = Safety Score 1 or Risk Score 5, Security Score 2 =
Safety Score 2 or Risk Score 4, Security Score 3 = Safety Score 3 or Risk Score 3, Security
Score 4 = Safety Score 4 or Risk Score 2, Security Score 5 = Safety Score 5 or Risk Score 1.
In addition, the interaction between brand familiarity and time pressure did not significantly
predict choice (see Figure 2), χ2(1) < 0.01, p = 1.000; the coefficient was -0.01 with a 95% CI of
[-0.05, 0.02], not supporting Hypothesis 4. This result may be due to a ceiling effect, wherein
0%
5%
10%
15%
20%
25%
30%
35%
12345
MeanPercentofTimethatAppwasChosen(%)
SecurityScore
RiskFrame
SafetyFrame
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 19
familiar brands were already chosen so often under no time pressure that there was little room for
participants to choose them even more frequently under time pressure.
The interaction between security scores and brand familiarity did not quite attain the .001
α level (see Figure 4), χ2(1) = 9.77, p = .002; the coefficient was 0.004 with a 95% CI of [-0.01,
0.02]. In addition, the trend was not in the direction that Hypothesis 5 predicted, but rather
indicates that, with increasing brand familiarity, participants tended to be more likely to use the
security scores in their decision making. This difference between security score effectiveness as
brand familiarity increased could be due to the overall low rate at which low-familiarity apps
were chosen, but suggests that security scores could, in fact, guide consumer purchases with
familiar apps.
Figure 4
The effect of the interaction between security scores and brand familiarity on app choice
Note. This interaction effect was not significant.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1234567
MeanPercentofTimethatAppwasChosen
(%)
BrandFamiliarityScore
SecurityScore=1
SecurityScore=2
SecurityScore=3
SecurityScore=4
SecurityScore=5
Linear(SecurityScore=1)
Linear(SecurityScore=2)
Linear(SecurityScore=3)
Linear(SecurityScore=4)
Linear(SecurityScore=5)
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 20
Similarly, the interaction between security scores and time pressure neared statistical
significance, though it fell short of the defined α level (see Figure 5), χ2(1) = 8.15, p = .004; the
coefficient was 0.12 with a 95% CI of [0.07, 0.17]. That said, the results were not in the direction
anticipated by Hypothesis 6; under time pressure, participants tended to be less likely to select
apps with higher security scores than participants who did not experience time pressure.
Figure 5
The effect of the interaction between security scores and time pressure on app choice
Note. This interaction effect was not significant. The error bars represent 95% between-subjects
confidence intervals. Security Score 1 = Safety Score 1 or Risk Score 5, Security Score 2 =
Safety Score 2 or Risk Score 4, Security Score 3 = Safety Score 3 or Risk Score 3, Security
Score 4 = Safety Score 4 or Risk Score 2, Security Score 5 = Safety Score 5 or Risk Score
0%
5%
10%
15%
20%
25%
30%
35%
40%
12345
MeanPercentofTimethatAppwasChosen(%)
SecurityScore
TimePressureAbsent
TimePressurePresent
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 21
Interestingly, while neither the 2-way interaction between security scores and framing nor
that between security scores and time pressure were significant on their own, the 3-way
interaction between time pressure, security framing, and security scores was a significant
predictor of app choice (see Figure 6, χ
2
(1) = 67.80, p < .001; the coefficient was -0.44, with a
95% CI of [-0.51, -0.37]. To determine what was driving the significance of the three-way
interaction, a post-hoc pairwise comparison with a Bonferroni correction was conducted using
estimated marginal means for the interaction between security score and frame for the presence
and absence of time pressure, respectively. Without time pressure, the log contrast estimate for
the interaction between security score and frame was 0.18, z = 2.67, SE = 0.06, p < .001.
However, with time pressure, the log contrast estimate was 0.01, z = 0.27, SE = 0.06, p = .785.
Therefore, participants that were not under time pressure relied more heavily on safety scores
than risk scores (See Figure 6, left panel), whereas those under time pressure did not utilize the
safety scores differently from the risk scores (see Figure 6, bottom panel).
Figure 6
The effect of the interaction between security scores, security frame, and time pressure on app
choice
Note. The three-way interaction was significant. The error bars represent 95% between-subjects
confidence intervals. Security Score 1 = Safety Score 1 or Risk Score 5, Security Score 2 =
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 22
Safety Score 2 or Risk Score 4, Security Score 3 = Safety Score 3 or Risk Score 3, Security
Score 4 = Safety Score 4 or Risk Score 2, Security Score 5 = Safety Score 5 or Risk Score
Decision Time Data
A two-way ANOVA was conducted on decision time, with time pressure (present,
absent) and security framing (risk, safety) as between-subjects factors. Distribution of the
decision times was not normal, but skewed right and leptokurtic. This means that participants
generally made decisions quickly, though there were a few very slow decision times. To account
for this lack of normality and extremely slow times, decision time data were winsorized (Wilcox,
2005) at the fifth and ninety-fifth percentiles; that is, data points outside these percentiles were
transformed to be equal to these percentiles, reducing the number of extreme decision times. A
total of 6.25% of the decision time data were Winsorized to fit within the defined range.
Decision times were then log-transformed (natural log) for analysis; note that the reported means
and standard deviations are Winsorized decision times (in seconds) rather than log-transformed
values for the purpose of easier understanding.
The ANOVA revealed a significant main effect of time pressure, F(1, 3196) = 487.32, p
< .001,
p2 = .13, with participants under time pressure making faster decisions (M = 4.87 s, SD
= 3.07 s) than those without time pressure (M = 8.35 s, SD = 7.06 s), indicating the validity of the
time-pressure manipulation in the experiment. Additionally, the main effect of security framing
was significant, F(1, 3196) = 7.80, p = .005,
p2 = .002, with those in the safety framing
condition making faster decisions (M = 6.31 s, SD = 4.48 s) than those in the risk framing
condition (M = 6.99 s, SD = 6.88 s). Note that the faster decision time for safety-framed locks, as
a proxy for ease of use, mirrors the pattern of the interaction between security scores and security
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 23
framing for the app choice data. This combination of both faster and safer choices for the safety
framing suggests a strong effect of the safety framing on guiding decision making with reduced
cognitive effort. The interaction between time pressure and framing was not significant, F(1,
3196) = 1.38, p = .240,
p2 < .01.
Rationale and Perception Data
Participants’ post-experiment rationale for app choice was analyzed using Cochran’s
(1950) Q test, because participants could choose multiple reasons. There was a significant
difference in the frequencies of the various reasons being chosen by participants, χ2(3) = 46.61, p
< .001. Brand familiarity and security rating were selected as the most influential factor of app
choice with 45% and 44% of participants, respectively; user ratings was chosen by 38% of
participants, and 8% of participants indicated that icon look and feel was important, while no
participants selected the other option. A pairwise post-hoc Dunn test with Bonferroni corrections
was conducted to further investigate the differences between these options; only icon look and
feel differed significantly from the other choices (brand familiarity, security ratings, user ratings;
ps < .001), whereas all other pairwise comparisons were not significant ps > .100.
Privacy concerns were measured via an open-ended question, Do you have any privacy
concerns regarding mobile applications?. Over half (66) of the participants expressed specific
concerns (e.g., misuse of personal data, unnecessary permissions, tracking location, accessing
camera), 6 participants expressed general concerns of privacy (responses such as yes), 28
participants expressed no concerns, and 28 participants either did not respond or did not respond
in a meaningful manner. Finally, perceptions of the security locks were also measured via an
open-ended question by asking What did the security scores mean?. The feedback was generally
positive, with several responses complimenting the design of the safety-framed locks overall.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 24
Notably, a few of those assigned to the risk frame expressed potential confusion with the design,
such that more locks could be misconstrued as greater safety. Among those, one participant’s
feedback hinted at the stimulus-stimulus compatibility principle (De Houwer, 2003; Kornblum et
al., 1990), “I think it’s counter-intuitive and confusing. You should instead [use] closed locks
and have more locks [equal] better…You have more stars [equal] better and right underneath,
more locks [equal] worse.” Conversely, those in the safety frame described the locks as
“intuitive” and a few participants expressed that the design helped them make decisions quickly.
Some participants discussed how they would use the security system, “I do like the security lock
designs. Even when dealing with brands I knew, it helped remind me of the flaws [inherent] in
the brands security. Facebook was a prime example of that.” Other participants seemed less
interested in the security scores, “App locks do not [affect] my app choices or concerns. If I like
it, I keep it. If I don’t like the app, I delete it.”
Discussion
The current study focused on the effect of security framing, time pressure, and brand
familiarity on mobile app choice. These factors have been shown to individually affect purchase
behaviors of other products but had not yet been investigated for mobile apps. As such, the
current study was the first to investigate the combination of the framing effect of security scores
for mobile apps, time pressure, and the effect of brand familiarity.
Past research on mobile app security scores has shown that such a system can assist
mobile device users and that certain design considerations are more beneficial than others (Chen
et al., 2015; Chong et al., 2018; Rajivan & Camp, 2016). The current study advanced this line of
research by examining external factors such as brand familiarity (Baker et al., 1986; Harris et al.,
2016) and time pressure (Young et al., 2012; Saqib & Chan, 2015), as well as introducing color-
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 25
coded locks (Rajivan & Camp, 2016). Indeed, the current study provides further support for the
security scores proposed by Chen et al. (2015), with safety framing resulting in significantly
faster and qualitatively better decisions than the risk framing. It is worth noting that apps with
security score of 5 were chosen almost twice as often as those with lower security scores. This
was also true for the apps with security score of 3, which were presented twice on each trial. This
result indicates that only the highest security rating had a positive impact on users’ selection of
apps. In addition to the behavioral measures, the subjective reports from participants also
supported the use of the safety-framed locks over the risk-framed locks. This could be due to the
confusing nature of the unlocked locks, as locks typically represent a mental model of safety
(Rajivan & Camp, 2016). While the unlocked locks were designed to be as equitable to the
locked locks as possible, they may have confounded the understanding of the risk scores
themselves.
By introducing the time pressure condition, which may be experienced by mobile users in
the real world, the current study advances the research of mobile app security. In addition to
making faster decisions under time pressure (Madan et al., 2015), mobile users appeared to
change the approach by which they make decisions under time pressure. Without time pressure,
participants under the safety frame made more secure decisions (i.e., choosing higher security-
score apps more often) than those under the risk frame, consistent with prior research
demonstrating the effectiveness of the safety frame (Chen et al., 2015; Chong et al., 2018;
Rajivan & Camp, 2016). However, when there was time pressure, the difference between the
safety and risk frames disappeared and both elicited less safer decision by choosing apps with
lower safety ratings, indicated by the significant three-way interaction between security frame,
security score, and time pressure. These results indicate that the previously proposed safety
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 26
frame of the security scores may not work as effectively under time pressure, which calls for
more research and new design solutions promote secure decision making under time pressure.
Brand familiarity strongly predicted app choice, consistent with literature on brand
familiarity for other products (Baker et al., 1986; Harris et al., 2016), but this effect was not
significantly moderated by time pressure or security score. This result indicates that mobile users
are likely to choose familiar apps regardless of time pressure, along with any potential dangers
associated with such apps. This result may be due to correlation between familiarity and
perceived trustworthiness (Ha & Perks, 2005). Therefore, in real-world usage, we can expect that
users will gravitate toward familiar apps. The design implications of this finding include the
necessity for app stores to better screen apps that users are more familiar with, as well as the
need to warn users about the potential risks of familiar apps.
We can expect that in real app stores, familiar apps are likely to be chosen far more than
unfamiliar apps. While an obvious point, it is important to note because participants in the study
tended to be more discriminatory among the more familiar apps along the security-score
spectrum. Therefore, the use of the security score in a real app store environment could help
users choose an app once they have narrowed their search down to a handful of familiar
alternatives. Another aspect of this finding is that when security score was 5, the app was chosen
much often than others across the range of the brand familiarity score, as shown by the blue solid
line in Figure 4. Apps should still strive for preserving the highest user security, which can be
appreciated by the user. This design may further serve users if the brand familiarity scores were
then used to compile the most familiar apps for direct comparison along security.
Beyond measuring the behavioral responses to the experiment, the current study also
gathered data on participants’ subjective report on their rationale of choosing apps, as well as
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 27
more general attitudes and perceptions regarding cybersecurity and the security score system in
question. A substantial number of participants expressed concern for their data and the
permissions that apps may request, indicating a potential increase in consumer awareness of the
potential threats associated with mobile apps compared to past research (Chin et al., 2012; Felt et
al., 2012; Kelley et al., 2012; Benton et al., 2013). In addition, participants’ suggestions and
feedback regarding the security scores supported the behavioral measures of the effectiveness of
the safety frame. These findings should be considered for future research and design for the
security display in question.
While the current study is the first to examine the effects of brand familiarity and time
pressure on mobile app choices with security scores, there are a few limitations. The pilot study
was conducted with college students, whereas the main experiment was with MTurkers. It is
likely that the familiarity scores obtained in the pilot may not fully reflect the MTurkers’
familiarity of the apps. The attentional check in the main experiment contained a programming
error that led it to be invalid. This led the data to be analyzed without data quality control in this
experiment. The results need to be validated in future research with valid catch trials. The risk
levels defined in this study have been shown effective in previous studies and followed the
general format of widely used user reviewers (i.e., levels one through five). However, it is a
simplification of potential app risks and does not specify the types of risks (e.g., privacy
violation). A personalized risk display that specifies the risk types that are of most concern to the
user (see Chen et al., 2022; Jorgenson et al., 2015) can be incorporated in future research.
Additionally, the use of locks was well accepted by the participants, and the closed locks
well represented the mental model of security (Rajivan & Camp, 2016). The current use of
unlocked locks in the risk frame may have caused confusion because it used similar iconography.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 28
Instead, other iconography that better represents danger can be used, such as crossbones, or X’s.
Future research should examine these other symbols for the risk conditions to better bolster the
impact of the negative frame so that the comparison against the lock image for the safety frame
can be fairer. In addition, given the strong effect of brand familiarity and its interaction with the
security score, further research is needed to examine the effectiveness and further design of the
security scores when users are interacting exclusively with highly familiar apps. Finally, other
measures beyond familiarity may affect users’ decisions, such as the reputation of the product or
company, and the category of the app functions. The reputation of a product or company may not
always be consistent with their familiarity and security ratings (e.g., users may be very familiar
with a company who has bad reputation). People may also weigh security differently if the app
has financial functions (e.g., banking, cash payment) than others. It would be interesting to
examine how these factors interact and affect users’ decisions.
Conclusion
The current study builds on the prior literature on mobile app security communication
(Chen et al., 2015; Chong et al., 2018; Rajivan & Camp, 2016) and supports the adoption of
positive framing of security scores for ease of use. However, the current study shows that
additional security precautions need to be taken when users are under time pressure or faced with
familiar apps. In addition, the use of locks resulted in positive subjective evaluations by the
participants and is recommended for future designs.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 29
References
42matters (2023a). Google play statistics and trends 2023. Retrieved July 23, 2023, from
https://42matters.com/google-play-statistics-and-trends#apps-released-per-day
42matters (2023b). iOS Apple App Store Statistics and Trends 2023. Retrieved July 23, 2023,
from https://42matters.com/ios-apple-app-store-statistics-and-trends
Ammons, J. G., Parker, C., & Chen, J. (2022). Time Pressure and User Ratings on Consumers’
Choice and Eye Fixations. International Journal of Human–Computer Interaction, 38,
993-1003.
Aggarwal, P., & Vaidyanathan, R. (2003). Use it or lose it: Purchase acceleration effects of time-
limited promotions. Journal of Consumer Behaviour: An International Research Review,
2, 393-403.
Almuhimedi, H., Schaub, F., Sadeh, N., Adjerid, I., Acquisti, A., Gluck, J., Cranor, L. F., &
Agarwal, Y. (2015). Your Location Has Been Shared 5,398 Times! A Field Study on
Mobile App Privacy Nudging. In Proceedings of the 33rd Annual ACM Conference on
Human Factors in Computing Systems (pp. 787–796). Association for Computing
Machinery.
Baarslag, T., Alan, A. T., Gomer, R. C., Liccardi, I., Marreiros, H., Gerding, E. H., & schraefel,
m.c. (2016). Negotiation as an Interaction Mechanism for Deciding App Permissions. In
Proceedings of the 2016 CHI Conference Extended Abstracts on Human Factors in
Computing Systems (pp. 2012–2019). Association for Computing Machinery.
Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with crossed
random effects for subjects and items. Journal of Memory and Language, 59, 390-412.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 30
Baker, W., Hutchinson, J., Moore, D., & Nedungadi, P. (1986). Brand Familiarity and
Advertising: Effects on the Evoked Set and Brand Preference. Advances in Consumer
Research, 13, 637-642.
Ben Zur, H., & Breznitz, S. J. (1981). The effect of time pressure on risky choice behavior. Acta
Psychologica, 47, 89-104.
Benton, K., Camp, L. J., & Garg, V. (2013). Studying the effectiveness of android application
permissions requests. In 2013 IEEE International Conference on Pervasive Computing
and Communications Workshops (PERCOM Workshops), 291-296.
Bergum, B. O., & Bergum, J. E. (1981). Population stereotypes: An attempt to measure and
define. In Proceedings of the Human Factors Society Annual Meeting, 25, 662-665.
Biswas, A. (1992). The moderating role of brand familiarity in reference price perceptions.
Journal of Business Research, 15(3), 251-262.
Buzulukova, E., & Kobets, A. (2022). Drivers for making in-app purchases in mobile games by
users in emerging countries. BRICS Journal of Economics, 3(3), 183-201.
Ceci, L. (2022). Number of apps available in leading app stores as of 2nd quarter 2022.
Retrieved September 23, 2022, from https://www.statista.com/statistics/276623/number-
of-apps-available-in-leading-app-stores
Ceci, L. (2023). Annual number of global mobile app downloads 2016-2022. Retrieved July 23,
2023, from https://www.statista.com/statistics/271644/worldwide-free-and-paid-mobile-
app-store-downloads/
Chandler, J. J., & Pronin, E. (2012). Fast thought speed induces risk taking. Psychological
Science, 23, 370-374.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 31
Chen, J. (2020). Risk communication in cyberspace: A brief review of the information-
processing and mental models approaches. Current Opinion in Psychology, 36, 135-140.
Chen, J., Ge, H., Li, N., & Proctor, R. W. (2022). What I say means what I did: Risk concerns
and mobile application-selection behaviors. Human Factors, 64(8),1331-1350.
Chen, J., & Proctor, R. W. (2017). Role of accentuation in the selection/rejection task framing
effect. Journal of Experimental Psychology: General, 146, 543-568.
Chen, J., Gates, C. S., Li, N., & Proctor, R. W. (2015). Influence of risk/safety information
framing on android app-installation decisions. Journal of Cognitive Engineering and
Decision Making, 9, 149-168.
Chen, J., Gates, C. S., Proctor, R. W., & Li, N. (2014). Framing of summary risk/safety
information and app selection. In Proceedings of the Human Factors and Ergonomics
Society Annual Meeting, 58, 1461-1465.
Chen, J., Ge, H., Moore, S., Yang, W., Li, N., & Proctor, R. W. (2018). Display of major risk
categories for android apps. Journal of Experimental Psychology: Applied, 24, 306-330.
Chin, E., Felt, A. P., Sekar, V., & Wagner, D. (2012). Measuring user confidence in smartphone
security and privacy. In Proceedings of the Eighth Symposium on Usable Privacy and
Security, 1-16.
Choe, E. K., Jung, J., Lee, B., & Fisher, K. (2013). Nudging people away from privacy-invasive
mobile apps through visual framing. In IFIP Conference on Human-Computer
Interaction (pp. 74-91). Springer, Berlin, Heidelberg.
Chong, I., Ge, H., Li, N., & Proctor, R. W. (2018). Influence of privacy priming and security
framing on mobile app selection. Computers & Security, 78, 143-154.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 32
Chowdhury, T. G., Ratneshwar, S., & Mohanty, P. (2009). The time-harried shopper: Exploring
the differences between maximizers and satisficers. Marketing Letters, 20, 155-167.
Cochran, W. G. (1950). The comparison of percentages in matched samples. Biometrika, 37,
256-266.
De Houwer, J. (2003). On the role of stimulus-response and stimulus-stimulus compatibility in
the Stroop effect. Memory & Cognition, 31, 353-359.
Devlin, J., Ennew, C., McKechnie, S., & Smith, A. (2007). A study of time limited price
promotions. Journal of Product & Brand Management, 16, 280-285.
Dinsmore, J. B., Wright, S. A., & Plotkina, D. (2021). When time pressure counters the zero
price effect. Journal of Consumer Marketing, 38(3), 399-350.
do Prado, R. A. D. P., & Lopes, E. L. (2016). Tick tock, tick tock! An experimental study on the
time pressure effect on omission neglect. Journal of International Consumer Marketing,
28, 332-346.
El Haji, A., Krawczyk, M., Sylwestrzak, M., & Zawojska, E. (2016). Time pressure and risk
taking in auctions: A field experiment. Journal of Behavioral and Experimental
Economics, 78, 68-79.
Felt, A. P., Ha, E., Egelman, S., Haney, A., Chin, E., & Wagner, D. (2012, July). Android
permissions: User attention, comprehension, and behavior. In Proceedings of the Eighth
Symposium on Usable Privacy and Security, 1-14.
Gates, C., Chen, J., Li, N., & Proctor, R. W. (2014). Effective risk communication for Android
Apps. IEEE Transactions on Dependable and Secure Computing, 11, 252-265.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 33
Ha, H. Y., & Perks, H. (2005). Effects of consumer perceptions of brand experience on the web:
Brand familiarity, satisfaction and brand trust. Journal of Consumer Behaviour: An
International Research Review, 4, 438-452.
Harris, M. A., Brookshire, R., & Chin, A. G. (2016). Identifying factors influencing consumers’
intent to install mobile applications. International Journal of Information Management,
36, 441-450.
Jorgenson, Z., Chen, J., Gates, C., Li, N., Proctor, R. W., & Yu, T. (2015). Dimensions of risk in
mobile applications: A user study. In Proceedings of Fifth ACM Conference on Data and
Application Security and Privacy. New York, NY: ACM.
Kahneman, D. & Tversky A. (1979). Prospect theory: an analysis of decision under risk.
Econometrica, 47, 263-291.
Kardes, F. R., Posavac, S. S., Silvera, D., Cronley, M. L., Sanbonmatsu, D. M., Schertzer, S., ...
& Chandrashekaran, M. (2006). Debiasing omission neglect. Journal of Business
Research, 59, 786-792.
Kelley, P. G., Consolvo, S., Cranor, L. F., Jung, J., Sadeh, N., & Wetherall, D. (2012). A
conundrum of permissions: installing applications on an android smartphone. In
International Conference on Financial Cryptography and Data Security, 68-79. Springer,
Berlin, Heidelberg.
Kharlamov, A., Jaiswal, A., Parry, G., & Pogrebna, G. (2018). A cyber domain-specific risk
attitudes scale to address security issues in the digital space. mimeo. https://doi.
org/10.13140/RG. 2.2. 31408.05122/2.
Klapproth, F. (2008). Time and decision making in humans. Cognitive, Affective, & Behavioral
Neuroscience, 8, 509-524.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 34
Kornblum, S., Hasbroucq, T., & Osman, A. (1990). Dimensional overlap: cognitive basis for
stimulus-response compatibility--a model and taxonomy. Psychological Review, 97, 253-
270.
Kumar, H., & Tuli, N. (2021). Decoding the branded app engagement process: a grounded theory
approach. Asia Pacific Journal of Information Systems, 31(4), 582-605.
Laroche, M., Kim, C., & Zhou, L. (1996). Brand familiarity and confidence as determinants of
purchase intention: An empirical test in a multiple brand context. Journal of Business
Research, 37, 115-120.
Liu, C. W., Hsieh, A. Y., Lo, S. K., & Hwang, Y. (2017). What consumers see when time is
running out: Consumers’ browsing behaviors on online shopping websites when under
time pressure. Computers in Human Behavior, 70, 391-397.
Liu, D., Peterson, T., Vincenzi, D., & Doherty, S. (2016). Effect of time pressure and target
uncertainty on human operator performance and workload for autonomous unmanned
aerial system. International Journal of Industrial Ergonomics, 51, 52-58.
Liu, W., Cao, Y., & Proctor, R. W. (2021). How do app icon color and border shape influence
visual search efficiency and user experience? Evidence from an eye-tracking
study. International Journal of Industrial Ergonomics, 84, 103160.
Madan, C. R., Spetch, M. L., & Ludvig, E. A. (2015). Rapid makes risky: Time pressure
increases risk seeking in decisions from experience. Journal of Cognitive Psychology, 27,
921-928.
Menon, S., & Kahn, B. (2003). Cross-category effects of induced and pleasure on the internet
shopping experience. Journal of Retailing, 78(1), 31–40.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 35
Miles, J., & Shevlin, M. (2001). Applying regression and correlation: A guide for students and
researchers. London, UK: SAGE.
Nakashima, R. (2018). AP Exclusive: Google tracks your movements, like it or not. Retrieved
September 23, 2022, from https://apnews.com/828aefab64d4411bac257a07c1af0ecb/AP-
Exclusive:-Google-tracks-your-movements,-like-it-or-not.
Park, J., & Stoel, L. (2005). Effect of brand familiarity, experience and information on online
apparel purchase. International Journal of Retail & Distribution Management, 33, 148-
160.
Park, J., Han, S. H., Kim, H. K., Oh, S., & Moon, H. (2013). Modeling user experience: A case
study on a mobile device. International Journal of Industrial Ergonomics, 43(2), 187-
196.
Pasaribu, A., Purwati, Y., & Jie, F. (2013). The influence of attitude and brand recognition
toward purchase intention of online advertising on social networking sites.
In Proceedings of the18th Asia Pacific DSI Conference (pp. 1-14). International Decision
Sciences Institute.
Peer, E., Vosgerau, J., & Acquisti, A. (2014). Reputation as a sufficient condition for data quality
on Amazon Mechanical Turk. Behavior Research Methods, 46, 1023-1031.
Price, E. (2018). More than 500,000 people downloaded these malware-infected android apps.
Retrieved September 23, 2022 from http://fortune.com/2018/11/26/google-play-malware-
apps/
Rajivan, P., & Camp, J. (2016). Influence of privacy attitude and privacy cue framing on android
app choices. In Twelfth Symposium on Usable Privacy and Security.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 36
Radesky, J., Hiniker, A., McLaren, C., Akgun, E., Schaller, A., Weeks, H. M., ... & Gearhardt,
A. N. (2022). Prevalence and characteristics of manipulative design in mobile
applications used by children. JAMA Network Open, 5(6), e2217641-e2217641.
Saqib, N. U., & Chan, E. Y. (2015). Time pressure reverses risk preferences. Organizational
Behavior and Human Decision Processes, 130, 58-68.
Schuster, D., Still, M. L., Still, J. D., Lim, J. J., Feria, C. S., & Rohrer, C. P. (2015). Opinions or
Algorithms: An Investigation of Trust in People Versus Automation in App Store
Security. In International Conference on Human Aspects of Information Security,
Privacy, and Trust (pp. 415-425). Springer, Cham.
Shehryar, O. (2008). The effect of buyer's gender, risk-proneness, and time remaining in an
internet auction on the decision to bid or buy-it-now. Journal of Product & Brand
Management, 17, 356-365.
Soliman, M. A. (2017). The impact of scarcity message on impulsive purchase intention among
smartphone shoppers (Doctoral dissertation, Trident University International).
Statista Research Department. (2022). Number of smartphone subscriptions worldwide from
2016 to 2021, with forecasts from 2022 to 2027. Retrieved September 23, 2022 from
https://www.statista.com/statistics/330695/number-of-smartphone-users-worldwide/
Stocchi, L., Michaelidou, N., & Micevski, M. (2019). Drivers and outcomes of branded mobile
app usage intention. Journal of Product & Brand Management, 28(1), 28-49.
Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice.
Science, 211, 453-458.
Vidas, T., Christin, N., & Cranor, L. (2011). Curbing android permission creep. In Proceedings
of the Web, 2, 91-96.
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 37
Vlašić, G., Janković, M., & Kramo-Čaluk, A. (2011). Information hunt: the impact of product
type and time pressure on choice of information source for purchase decisions.
Management: Journal of Contemporary Management Issues, 16, 87-103.
Wilcox, R. (2005). Trimming and Winsorization. In P. Armitage, & T. Colton (Eds.),
Encyclopedia of Biostatistics. Chichester, UK: John Wiley & Sons Ltd.
Wong, Q. (2019). Facebook will reportedly be fined a record $5 billion over privacy mishaps.
Retrieved September 23, 2022 from https://www.cnet.com/news/facebook-will-
reportedly-be-fined-a-record-5-billion-over-privacy-mishaps/
Wright, P. (1974). The harassed decision maker: Time pressures, distractions, and the use of
evidence. Journal of Applied Psychology, 59, 555-561.
Young, D. L., Goodie, A. S., Hall, D. B., & Wu, E. (2012). Decision making under time
pressure, modeled in a prospect theory framework. Organizational Behavior and Human
Decision
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 38
APPENDIX A. LIST OF APPLICATION FUNCTIONS
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 39
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 40
Authors' Accepted Manuscript
FRAMING SECURITY UNDER TIME PRESSURE 41
