MacOS versus Microsoft Windows: A Study on the

Cybersecurity and Privacy User Perception of Two

Popular Operating Systems

Cem Topcuoglu∗ , Andrea Martinez† , Abbas Acar† , Selcuk Uluagac† , and Engin Kirda∗
∗ Northeastern University, † Florida International University

Abstract—Operating Systems (OSs) play a crucial role in 1990s until mid-2000 (e.g., there was no privilege separation
shaping user perceptions of security and privacy. Yet, the distinct and all operations ran under the administrator account), Win-
perception of different OS users received limited attention from dows has evolved to become a significantly more secure and
security researchers. The two most dominant operating systems reliable operating system today and is now much more difficult
today are MacOS and Microsoft Windows. Although both oper- to remotely compromise compared to its earlier versions [4].
ating systems contain advanced cybersecurity features that have
made it more difficult for attackers to launch their attacks and Discussions comparing the security and privacy of Win-
compromise users, the folk wisdom suggests that users regard dows and MacOS typically focus on two arguments [10], [36].
MacOS as being the more secure operating system among the
The first argument is that MacOS has a smaller user base
two. However, this common belief regarding the comparison of
these two operating systems, as well as the mental models behind and thus, fewer incentives to attack the OS [10], [36]. While
it, have not been studied yet. historically, the smaller deployment of MacOS did provide
its users a degree of protection against widespread threats,
In this paper, by conducting detailed surveys with a large with the increasing popularity of MacOS, attackers started to
number of MacOS and Windows users (n = 208) on Amazon focus more on this operating system and its users. Indeed, the
Mechanical Turk, we aim to understand the differences in number of malware instances for MacOS not only rose but
perception among MacOS and Windows users concerning the
cybersecurity and privacy of these operating systems. Our results
even outpaced Windows by detections per endpoint [17]. The
confirm the folk wisdom and show that many Windows and second argument is that MacOS, being a UNIX-based OS, is
MacOS users indeed perceive MacOS as a more secure and inherently more secure than Windows [36]. While this might
private operating system compared to Windows, basing their have held for earlier versions of Windows, but not necessarily
belief on reputation rather than technical decisions. Additionally, for the modern versions. Many modern and cutting-edge
we found that MacOS users often take fewer security measures, cybersecurity mechanisms have been integrated into Windows
influenced by a strong confidence in their system’s malware that were not contained in the original UNIX OS. Regardless
protection capabilities. Moreover, our analysis highlights the of security architectures and mechanisms inherent to each of
impact of the operating system’s reputation and the primary these OSs, the attack surface for both OSs is continuously
OS used on users’ perceptions of security and privacy. Finally, evolving and attackers are increasingly deploying OS-agnostic
our qualitative analysis revealed many misconceptions such as
being MacOS malware-proof. Overall, our findings suggest the
methods such as web-based and social engineering attacks,
need for more focused security training and OS improvements targeting users across both platforms. In this shifting threat
and show the shreds of evidence that the mental model of users landscape, user awareness consistently remains a crucial, yet
in this regard is a vital process to predict new attack surfaces often overlooked, defense point. A recent survey showed that
and propose usable solutions. many MacOS users harbor misconceptions and engage in risky
behaviors [21]. For instance, a significant portion mistakenly
believes malware does not exist for MacOS, while others reuse
I. I NTRODUCTION passwords or skip software updates. Therefore, it’s essential to
The two most popular desktop operating systems (OSs) understand the security perceptions of users for each OS.
today are Microsoft Windows and Apple MacOS, together
While prior research has extensively addressed technical
accounting for approximately 90% of all desktop users glob-
security aspects of both Windows and MacOS [14], [38], [11],
ally [27]. Historically, both operating systems have been an
[22], and some have probed users’ mental models concerning
attractive attack surface. Ransomware, droppers, and trojan
security tools and threats [31], [9], [34], a noticeable research
horses have been almost a permanent security problem and
gap exists in the comparative study of MacOS and Windows
have caused much suffering over the years to operating system
user perceptions. This is surprising given the dominance of
users. Although Microsoft did not sufficiently secure their
these two OSs in today’s desktop usage [28]. In this work, we
systems in the initial versions of the Windows family from the
fill this research gap and primarily focus on the user perception
of the current security and privacy posture of the two most
popular operating systems [28]. We provide an empirical
Symposium on Usable Security and Privacy (USEC) 2024 analysis of the determinants of the users’ perception and
26 February 2024, San Diego, CA, USA
ISBN 979-8-9894372-5-2 study how users evaluate their operating systems, highlighting
[Link] the importance of user education with respect to operating
[Link], [Link] systems security. Specifically, we aim to understand which
OS – MacOS or Windows – is perceived as more secure Windows users perceive cybersecurity and privacy;
and private by users and explore the implications of these and 2) how these users view the reputations of Apple
perceptions in terms of the security and privacy behaviors of and Microsoft. Conversely, we determined that demo-
the users. graphic features do not have any statistically signifi-
cant influence on how users perceive the cybersecurity
We conducted a series of surveys on Amazon Mechan- and privacy of their respective OSs.
ical Turk, focusing on the 18-49 age group who possessed
at least some college-level education. According to existing • We found a statistically significant relationship be-
literature [25], the results of security and privacy surveys tween being an IT worker and the perceived cyber-
conducted on this group have been shown to be representative security and privacy comparison of MacOS and Win-
of the U.S. population within the same group. We recruited dows, as well as security habits such as performing
103 MacOS and 105 Windows users and applied rigorous backups, covering the camera lens of the computer,
statistical analysis to gain a deeper understanding of how and using the private mode of the browser.
users perceive the cybersecurity and privacy of MacOS and
• We also observed that there is a significant correlation
Windows when the operating systems are compared. We asked
between a user’s primary OS and their proactive
how they compare these two operating systems in terms of
cybersecurity practices such as utilizing antivirus (AV)
cybersecurity, privacy, and reputation. We also investigated if
software, performing backups, and covering the lens
these different operating systems users demonstrate different
of the computer camera. This perception may lead to
behavior patterns (i.e., if there is a statistically significant
a more relaxed assumption about necessary security
difference between the security best practices such as antivirus
risks among MacOS users, indicating the need for
usage). Understanding these differences can shed light on
targeted cybersecurity education.
the user mental model and what the users expect from their
operating system, which can help us build operating system • Through qualitative analysis, we also uncovered the
features that match users’ expectations. Furthermore, we can reasons behind these perceptions and the common
determine trends, predict potential new attack surfaces, and misconceptions held by users.
propose actionable recommendations.
The rest of the paper is organized as follows: Section II
In summary, our research confirms that the perception of describes related work. Section III discusses our methodology.
MacOS as a more secure and private operating system is Section IV presents and discusses the results. Section VI dis-
widespread, although the nature of attacks and operating sys- cusses the limitations of our work. Section V summarizes our
tem security has significantly evolved. Despite all the progress findings and actionable results. Finally, Section VII concludes
made in the security of the Windows operating system, many the paper.
users, including Windows users, still believe that MacOS is a
much more secure operating system. A large number of users II. R ELATED W ORK
based their choice to the reputation and technology of these
OSes and have biased beliefs such as “MacOS is malware A. Security and Privacy of Desktop OSs
proof”, and “It is rare to hear about a MacOS device having a In the literature, a huge body of research is devoted to
virus or malware on it”. One of the participants stated that “It Windows computers [15]. Some research papers, though they
has nothing to do with the company, but simply the fact that did not specify the target malware, worked on PE malware
most hackers design programs to disrupt Windows than Mac”, that only runs on Windows computers [33], [37] or studied
similar to the current Internet discussions. This image of OS vulnerabilities that exploit Windows computers [24]. On the
security can have an impact on less sophisticated users. That other hand, researchers also investigated malware targeting
is, these users are left with the illusion that their operating MacOS [14], [38], [11], [22]. Lindorfer et al. built a high
system is responsible for every security and privacy task. This interaction honeypot capable of automatically downloading
is also reflected in the security behavior of users; we found that OS X binaries [14] while other works targeted the MacOS
users of the two operating systems show statistically significant kernel [38], [11]. Finally, Adam J. O’dOnnell examined why
differences in their antivirus usage, data backup habits, and malware attacks might occur on MacOS, and introduced a
behaviors related to covering the camera lens. This work also model on game theory to predict when the malware attacks
serves to provide insights into the importance of user education will increase on MacOS [22]. However, none of these studies
and awareness of newer forms of attacks that are agnostic to investigates the user perception of different OS users.
the type of operating system and can cause irreversible damage
to users on almost any platform. B. User Perception on Security and Privacy
The main contributions of this study are as follows: There has been research that has investigated the perception
• We conducted the first study that analyzes the user of users with respect to security and privacy issues from
perception with respect to the security and privacy of various angles. The first category focuses on the web domain.
the two most popular desktop operating systems. We Turner et al. investigated factors that affect the perception of
analyzed which operating system provides perceived security and privacy in e-commerce websites [31]. The paper
cybersecurity and privacy with a series of surveys on reports that consumers are mainly interested in company rep-
Amazon Mechanical Turk. utation, past experiences, and security recommendations from
third parties. Moreover, Flinn et al. investigated the security
• Our quantitative analysis reveals statistically signif- and privacy perception on the web [9]. The second category
icant differences in two areas: 1) how MacOS and focused on the security tools such as anti-virus scanners. For

2
example, Rick Wash presents an analysis of the mental models quality. To make sure our survey is reliable, we used a two-
of users about attackers and security technologies such as anti- fold survey structure. In the first survey, we collected demo-
virus scanners [34]. The paper reports that every MacOS user graphics and primarily used OS information. By collecting
believes that their system is immune to viruses and hacking this information from the user, we were able to verify if
problems and that some users do not use security software there were discrepancies between the first and second survey
because of perceived immunity. The paper only performs a that was focused on cybersecurity and privacy perceptions.
qualitative analysis, and the number of participants is limited to We eliminated the surveyee from our analysis if the answers
14. The third category of work is on mobile users. Researchers to some questions between the two surveys did not match.
compared the threats for Android and iOS users [20], [1] Also, we made sure that every Amazon Mechanical Turk user
while Benenson et al. compared the user behaviors of Android participated in our surveys only once.
and iOS users concerning security and privacy issues [5]. In
another work, Falaki et al. studied application usage in terms IRB approval: Our methodology and survey questions have
of reducing the energy consumption of Android and Windows been submitted to our institution’s IRB board, and we obtained
phone users [8]. In addition, Benenson et al. studied the mental IRB approval. Our work was exempted under category #2 by
models of smartphone users towards IT security, and found that the Human Subject Research Protection department, as we did
the users with good security knowledge tend to use additional not collect any sensitive or personally identifiable information
technical protection means [6]. Finally, Chin et al. studied user from participants and all the participants were anonymous to
confidence in smartphone security and privacy, and looked us. Participants were shown a consent form to inform them
into the security and privacy perception on smartphones when about our research. We only accepted participants who were
compared to desktop computers [7]. Interestingly, while this 18 years old or older.
study also recruited MacOS and Windows users, it used
computer users as a baseline to discover smartphone-specific A. Model
issues but did not delve into the issues regarding MacOS and We crafted two different regression models before de-
Windows users. However, none of these studies worked on the signing our survey. In the first model, we investigated the
security and privacy perception of MacOS and Windows users. relationship between the primarily used OS and the per-
To the best of our knowledge, no prior work has attempted ception of which OS (i.e., MacOS or Windows) provides
to study the differences in cybersecurity and privacy percep- more cybersecurity and privacy. In the second model, we
tions of users with regard to the popular MacOS and Windows investigated the relationship between primarily used OS and
operating systems. A number of works have paid attention to some cybersecurity precautions such as AV software usage.
the merits of performing surveys on Amazon Mechanical Turk, Upon forging the initial regressions, we developed upon it
and how representative and generalizable the collected data is. and added independent variables that were likely to be related
Many recent studies, including [12] and [25], have leveraged to the dependent variable. By adding these new independent
Amazon Mechanical Turk for survey research. Redmiles et variables, we aimed to minimize the bias on the primarily used
al. highlighted its utility for capturing a U.S. representative OS variable. Hence, we concluded with two different multiple
demographic in security and privacy contexts [25]. regression models. Note that all regression models were crafted
before we conducted our surveys.
III. M ETHODOLOGY In the first model, the first independent variable is the
Proper sample selection is crucial in online surveys to primarily used OS. The key insight here is that the primarily
achieve valid results [3], [29]. We utilized the Amazon Me- used OS might be a strong driver for the cybersecurity and
chanical Turk platform, a crowd-sourcing marketplace [2]. privacy perception of that OS. That is, the frequent usage of
Recent research has determined that Amazon Mechanical Turk the OS might create an attachment to that OS (i.e., familiarity).
is representative of the U.S. population (between the ages 18 Or, the user might have chosen to use that OS because it has
and 49 with some degree of college education) with respect been mandated by work, or because of price reasons. We also
to privacy and security topics [25]. Hence, in our survey, we added a variable that we call the “desired” OS. This variable is
focused on this group. To make sure that our survey data was collected by asking users which operating system they would
of high quality, we selected HIT approval rates greater than or use if they were not constrained by price or their job – that
equal to 95%. Also, we recruited participants who are located is, we try to capture the “wish” OS of the user. The second
in the United States. For our main survey, we compensated the and third variables in our model are the perceived reputations
participants with $1 for well under 10 minutes. We conducted of Microsoft and Apple in cybersecurity and privacy. The
our study between December 2020 and March 2021. key insight here is that we believe the perceived reputation
represents hearsay information, advertisement power, historical
Online surveys have five main advantages: They are easy to events, and former experiences with the OS. Also, reputation
prepare, easy to answer, they help avoid desirability bias, they is one of the key factors in the perception of cybersecurity and
allow to make sure that surveyees are receiving the same set privacy [31]. The fourth variable, IT, is whether the surveyee
of questions in the same regard, and there is no time constraint works in an Information Technology (IT) or cybersecurity-
for users [3], [29]. At the same time, there is one important related job. By using this variable, we aim to measure if
disadvantage as well: The lack of an interviewer [3], [29] the surveyee is knowledgeable in cybersecurity. Clearly, the
who can register user reactions and clarify misunderstandings. cybersecurity knowledge a user has can affect her perception
Hence, we tried to create easy-to-follow text and directives. and understanding of cybersecurity. The fifth variable in our
Furthermore, we kept our survey as simple as possible, and model is the level of daily computer usage (i.e., experience
we conducted preliminary experiments to improve our survey with computers). We also hypothesize that demographic factors

3
might affect the security and privacy perception. Previous work Mechanical Turk and among some of our colleagues. The ini-
also studied the effect of demographics on security perception tial results of these trials showed that surveyees were in favor
and found a significant relationship [13]. Hence, finally, we of the MacOS both in terms of providing more cybersecurity
added age, education, and gender variables. To investigate and privacy. Note that Apple MacOS users had a stronger
the effect of eight independent variables on the dependent opinion about their OS when we compared it with Windows
variable that indicates which OS provides more cybersecurity users. After these trials, we decided to make three changes.
and privacy, we designed the following equation: First, we made operating system names explanatory to avoid
confusion such as using ‘Apple MacOS’ instead of ‘MacOS’.
Second, we explicitly defined the terms cybersecurity and
OS Comparison = β0 + β1 OS or Desired OS + privacy at the start of our survey.
β2 M S P erceived Rep. +
β3 Apple P erceived Rep. + (1) Survey Content: Our survey starts with background ques-
β4 IT + β5 Daily Computer U sage + tions about daily computer usage and computer-related job
β6 Age + β7 Education + β8 Gender + ϵ experience. Then, we try to get an idea of backup behavior,
physical security behavior, web privacy behavior, and usage
In the second model, our independent variable is the of software security programs. Hence, we asked about AV
primarily used OS. We believe the primarily used OS might be software usage, frequency of performed backups, frequency of
related to cybersecurity behavior. We used four different de- the private mode of the browser, and whether the participant
pendent variables that we dubbed as the cybersecurity behavior covers the webcam with a physical cover to gain a basic
variable. Our dependent variables are whether AV software is understanding of how secure and privacy-aware the participant
being used, if data is being backed up regularly, if the user’s is. We selected these four questions which are the subset of the
computer camera lens is covered, and how often the user uses recommendations for laptop security [26], [19]. Our aim was
the privacy mode of her browser. By asking for these cyberse- to investigate if there is a relationship between the primarily
curity precautions, we aim to determine behavioral differences used OS and these behaviors.
between the users of these two OSes. Since we have four
different dependent variables, we have four different equations Perception Queries: We also asked the participants questions
for each precaution. For each dependent variable, we added about their perception of the reputation of Apple and Microsoft
the IT and daily computer usage variables. Similar to the first with respect to cybersecurity and privacy, and which operating
equation, we believe, these could be related to cybersecurity system they mainly use. Furthermore, we directly asked the
behavior. Moreover, we controlled the demographic features participants which operating system they believe cares more
such as age, education, and gender as well. Hence, in order to about cybersecurity and privacy in general. We gave the par-
understand whether cybersecurity behavior changes are related ticipants a 7-point Likert scale that they could use to evaluate
to the primarily used OS, we derived the following equation: both operating systems. We chose a 7-point Likert scale to
give more freedom to evaluate the level of confidence while
Cybersecurity Behavior = β0 + β1 OS+ comparing the operating systems.
β2 IT + β3 Daily Computer U sage + (2)
β4 Age + β5 Education + β6 Gender + ϵ Quantitative and Qualitative Analysis: As we described in
Section III-A, we crafted detailed regression models. Then,
By crafting these regression models, we aim to understand we used R language [23] to conduct the quantitative analy-
the underlying factors that influence users’ perceptions and sis. Although our analysis was mostly based on quantitative
behaviors concerning cybersecurity and privacy across the two analysis, we also posed some qualitative questions to distill a
major OS platforms. better understanding of the participants’ thoughts and expres-
sions. For example, we asked why the participants believed
B. Survey that the specific operating system they chose offered more
cybersecurity or privacy, and collected textual feedback. We
Study Design: We conducted our study in two batches: a used thematic analysis [32], a process that involves reading
preliminary and a main survey. In the preliminary survey, our through a collection of data and searching for patterns in the
goal was to gather data to identify workers who fit our criteria, data to identify themes to process these qualitative questions.
as mentioned above. Specifically, we looked at the primarily We created a spreadsheet from the analysis of the open-
used OS, the age of the participant, and education information. ended questions and imported the spreadsheet into Nvivo, a
For the primary OS used, we accepted textual input in the event qualitative data analysis software [16], where we analyzed the
that a participant’s primary OS was not MacOS or Windows. data and created codes by using inductive coding, an approach
Upon collection of this data, we selected all MacOS and where codes are created from the data itself [18]. Then,
Windows users who were aged between 18 and 49, with at we proceeded to categorize the codes into themes, again by
least some college-level education. Then, we assigned these using an inductive approach, allowing us to derive meaningful
participants qualifications on Amazon Mechanical Turk so that patterns from the coded data.
they could participate in our main survey. We set a target
number of participants around 100 for both groups.
Finally, we collected demographic information at the end of
Preliminary Findings and Modifications: To keep our study the survey. The full list of the survey questions is in Appendix
simple and understandable, we conducted trials on Amazon A.

4
 Perceived privacy Perceived privacy
comparison comparison

Perceived cybersecurity Perceived cybersecurity

comparison comparison

Microsoft's perceived Microsoft's perceived

privacy reputation privacy reputation
Variables

Variables
1 1
Microsoft's perceived 2 Microsoft's perceived 2
3 3
cybersecurity reputation 4 cybersecurity reputation 4
5 5
6 6
7 7
Browser private Browser private
mode usage mode usage

Apple's perceived Apple's perceived

privacy reputation privacy reputation

Apple's perceived Apple's perceived

cybersecurity reputation cybersecurity reputation

0.00 0.25 0.50 0.75 1.00 0.00 0.25 0.50 0.75 1.00
Percentages Percentages

Fig. 1. Only Windows users. Normalized responses to Likert scale survey Fig. 2. Only MacOS users. Normalized responses to Likert scale survey
questions. For rows 1 and 2, the scale ranged from Extremely Windows to questions. For rows 1 and 2, the scale ranged from Extremely Windows to
Extremely MacOS. For rows 3, 4, 6, and 7, the scale ranged from Poor Extremely MacOS. For rows 3, 4, 6, and 7, the scale ranged from Poor
Cybersecurity or Privacy to Excellent Cybersecurity or Privacy. For row 5, Cybersecurity or Privacy to Excellent Cybersecurity or Privacy. For row 5,
the scale ranged from Never to Always. Only related rows should be the scale ranged from Never to Always. Only related rows should be
compared. compared.

TABLE I. D ETAILED D EMOGRAPHIC S TATISTICS

C. Participant Data
Age MacOS Windows Total
Preliminary Survey: In total, we recruited 1,974 participants 18-24 12 7 19
in the preliminary survey. The distribution of primarily-used 25-29 30 27 57
operating system information for these participants is as fol- 30-39 44 47 91
40-49 17 24 41
lows: 70% of the participants were using Microsoft Windows,
Gender
16% were using Chrome OS, 13% were using Apple MacOS, Female 62 52 114
and 1% were using Linux. After collecting this information, we Male 39 51 90
selected the Amazon Mechanical Turk workers for our main Prefer not to say 2 2 4
survey as we described in Section III-B. Education
Associate degree 7 11 18
Main Survey: At the end of our main survey, we had Bachelor’s degree 52 53 105
Master’s degree 29 22 51
211 participants who met our selection criteria. However, we Ph.D. degree or higher 3 2 5
eliminated some of the surveyees since they had discrepancies Some college education 12 17 29
in their responses to the first and second surveys. By excluding Computer Usage
these data points, we collected survey data for 103 unique Less than 1 hour 2 0 2
From 1 up to 4 hours 15 19 34
MacOS and 105 unique Windows users. In Section IV, we From 4 up to 8 hours 55 31 86
statistically investigate the collected data. From 8 up to 12 hours 26 44 70
12 hours or more 5 11 16
Gender, Education, and Age: Our population is almost
balanced based on gender. That is, there are 114 female (55%)
and 90 male (43%) participants. Most of our surveyees had a
bachelor’s degree (50%). Also, most of the participants were
in between of 30 and 39. 91 of the participants were between OS users. As depicted, for Microsoft’s perceived privacy and
the ages of 30 and 39 (44%). Detailed demographics statistics cybersecurity reputation variables, Windows users tend to
presented in Table I. give higher scores compared to MacOS users. MacOS users
tend to believe that Windows provides neither perfect nor
Computer Usage: Table I also presents the results for
poor cybersecurity – with a slight tendency towards thinking
computer usage. The majority of our participants are heavy
more positively about this issue. The situation is reversed
computer users which aligns with our expectations since we
for Apple’s perceived privacy and cybersecurity reputation
conducted our survey on Amazon Mechanical Turk. Also,
variables, where MacOS user responses concentrated more on
35 surveyees (17%) reported that they work in an IT or
higher scores. Almost half of the MacOS users score Apple’s
cybersecurity-related job. Among those 35, 21 were Windows,
reputation as 6 or 7 on the Likert scale. Overall, MacOS users
and 14 were MacOS users.
tend to be neutral on the privacy reputation of Microsoft, while
Reputation Perceptions: Figures 1 and 2 present the nor- Windows users have a positive opinion about Apple’s privacy
malized results to reputation perception questions for each reputation.

5
Private Mode Usage: Figures 1 and 2 present the normalized OS as being extremely better than Windows in contrast to the
results to browser private mode usage question. MacOS users Windows users who were judging their own OS.
tend to accumulate around the middle, while Windows users’
scores to be either higher or lower. Also, 33% of the MacOS Figure 4 depicts the surveyee responses to whether Win-
users reported that they are not using AV software, 15% dows or MacOS cares more about user privacy. The majority
reported they do not back up their data at all, and 64% stated of the MacOS users (i.e., 65%) believed that MacOS doing a
that they do not cover the lens of their computer camera. better job in securing user privacy than Windows. In contrast,
Among Windows users, 12% reported that they are not using 28% of the Windows users did not believe that MacOS was
AV, 24% stated that they do not back up their data, and 45% doing a better job in securing user privacy when compared
stated that they are not covering the lens of their computer to Windows. Many Windows users were neutral, and tend to
camera. Perhaps unsurprisingly, Windows users tended to be believe that there is not a major difference in privacy between
more cautious when it came to malware protection, and attacks MacOS and Windows (i.e., 40%). 32% of the Windows users
that malware could launch such as recording video from the in our study had the belief that MacOS was doing a better job
camera of the computer. in privacy when compared to Windows – a significant number.
Figures 1 and 2 present normalized results for MacOS
32 and Windows with respect to what users perceive about their
29 29
cybersecurity and privacy. The results show that both MacOS
Number of users

and Windows users tend to believe that MacOS is provides

21 22 20
18 more than Windows when it comes to cybersecurity and
privacy. Almost all of the MacOS users perceive their OS at
12 11
8
least as good as Windows – most of them think MacOS is
4
better. Windows users also share this sentiment, although not
0 1 1 as strongly as MacOS users.
1 2 3 4 5 6 7 We also looked into what the desired OS was among our
Perfect Windows represented by 1, and perfect MacOS represented by 7
surveyees. The desired OS parameter in our model represents
Windows users MacOS users
the answer to which OS the surveyee would use if she was
not constrained by price or employment. 7 MacOS and 21
Windows users stated that they would switch their OS. The
Fig. 3. Results to the question of which OS offers more cybersecurity. results suggest that overall, MacOS users are very confident
about using their OS and do not wish to migrate to Windows.
42
Also, users who desire Windows as an OS seem to believe that
Windows is better than MacOS when it comes to privacy.
34
Number of Users

27
22
A. Cybersecurity Regression Analysis
17 18
15 In this section, we used R language [23] to analyze the
8 9 8 data and to apply the regression models we described in
6
Section III-A. We decided to use ordinal logistic regression
0 1 1
since our dependent variable is ordinal. Table II presents the
1 2 3 4 5 6 7 simplified cybersecurity regression results and we present the
Perfect Windows represented by 1, and perfect MacOS represented by 7 detailed regression results in Appendix B1.
Windows users MacOS users OS Perception: We started by regressing the primarily used
operating system as an independent variable with the per-
Fig. 4. Results to the question of which OS cares more about privacy. ceived cybersecurity comparison of MacOS and Windows as
a dependent variable where 7 represents MacOS offers more
cybersecurity than Windows and 1 represents the opposite. The
IV. R ESULTS AND D ISCUSSION primarily used OS parameter is a dummy variable where 1
represents being a Windows user, and 0 represents being a Ma-
In this section, we describe the statistical analysis of our cOS user. We determined that there is a statistically significant
results. Figure 3 shows the surveyee responses to whether relationship between these variables. We also calculated the
Windows or MacOS provides more cybersecurity. Perhaps odds ratio, which is 0.246. That is, the odds of a MacOS user
unsurprisingly and confirming the folk wisdom, the majority of having a higher security comparison score are approximately
the MacOS users (i.e., more than 80%) believed that MacOS 4.07 times the odds of a Windows user having a higher security
offers more cybersecurity than Windows. For Windows users, comparison score.
the majority of the users perceived MacOS to be better than
or equal to Windows (i.e., 77%). In fact, almost half (i.e., Influence of Company Reputation: In the second regression,
49%) of the users tend to believe that MacOS is better than we added MacOS’ and Windows’ perceived cybersecurity
Windows when it comes to cybersecurity. Our results suggest reputation variables. We determined that there is a statistically
that, as perhaps many people would expect, MacOS users seem significant relationship between the perceived reputations of
to have more sympathy and appreciation for their OS. MacOS the companies and the dependent variable. Also, when we
users were more enthusiastic when it came to ranking their added the reputation of the companies to the equation, the

6
 TABLE II. C YBERSECURITY REGRESSION TABLE . O NLY SIGNIFICANT LOG ODDS VALUES , ROUNDED TO THE THIRD DECIMAL , ARE REPORTED .
S IGNIFICANCE LEVELS ARE INDICATED WITH STARS (∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001). A LL ESTIMATES ARE BASED ON STANDARDIZED
REGRESSION MODELS . A DETAILED VERSION OF THIS TABLE IS AVAILABLE IN THE A PPENDIX .

Ordinal Logistic Regression:

Perceived cybersecurity comparison of MacOS and Windows
(1) (2) (3) (4) (5) (6) (7) (8)
OS (Windows = 1) −1.403∗∗∗ −0.963∗∗∗ −0.989∗∗∗ −0.941∗∗∗ −0.937∗∗∗ −0.897∗∗∗ −0.971∗∗∗
Apple’s cybersecurity 0.948∗∗∗ 0.962∗∗∗ 0.986∗∗∗ 1.017∗∗∗ 1.028∗∗∗ 1.048∗∗∗ 0.983∗∗∗
reputation
Microsoft’s cybersecurity −0.666∗∗∗ −0.701∗∗∗ −0.726∗∗∗ −0.746∗∗∗ −0.746∗∗∗ −0.773∗∗∗ −0.798∗∗∗
reputation
IT 0.816∗∗ 0.747∗∗ 0.794∗∗ 0.695∗ 0.723∗ 0.677∗
Computer usage ... ... ... ... ...
Age ... ... ... ...
Education ... ... ...
Gender ... ...
Desired OS −1.340∗∗∗
Observations 208 207 205 205 205 205 205 205
Note: ∗ p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01

log odds value of the primarily used OS decreased from - B. Privacy Regression Analysis
1.403 to -0.963 while remaining statistically significant. This
decrease demonstrates that by controlling the reputations, we In this section, we discuss the regression results on the
captured some of the bias on the primarily used OS. The odds OS privacy questions. Table III presents the privacy regression
ratio value for the reputation of MacOS is 2.579, which means results.
that the odds of having a higher dependent variable value (i.e., OS Perception: In the first regression, we used the perceived
towards MacOS offers excellent security) are approximately privacy comparison of the Windows and MacOS variables as
2.579 times greater for each unit increase in Apple’s perceived the dependent variable. The first independent variable was the
cybersecurity reputation. For Microsoft, the odds ratio value primarily used OS. There is a negative relationship between
is 0.513, which represents a negative relationship between the these two variables. The log odds value of the independent
dependent and independent variables. variable is -1.609 (odds ratio is 0.200), and there is a sta-
tistically significant relationship. That is, if we switch from
a Windows user to a MacOS user, the odds of having a
Influence of IT Background: In the third regression, we
higher value for the privacy comparison (towards MacOS) are
added the IT variable. Interestingly, we found a statistically
approximately 5 times of a Windows user. Hence, a MacOS
significant relationship between the dependent and IT variable.
user, relative to a Windows user, finds that MacOS cares more
The log odds of the IT variable was 0.816 and the odds
about privacy than Windows.
ratio was 2.262 (i.e., the odds of having a higher dependent
variable value are approximately 2.262 times larger, when we Influence of Company Reputation: In the second regression,
switched from a non-IT worker to an IT worker.) Note that we added Apple’s and Microsoft’s perceived privacy reputation
apart from small coefficient changes, there was no change in to the equation. We found that Apple’s perceived privacy
the significance of the previous variables. In the sequential reputation has a positive coefficient, and Microsoft’s has a
regressions, we added the daily usage of the computers, the negative coefficient. The odds ratio for Apple’s and Microsoft’s
age of the participants, the education level of the participants, perceived privacy reputation are 3.113 and 0.452. Similar
and their gender. However, we could not find any statistically to cybersecurity analysis, we found a positive relationship
significant relationship in any of these categories. between the dependent variable and Apple’s perceived pri-
vacy reputation, and a negative relationship with Microsoft’s
perceived privacy reputation. Note that both of the variables
In the complete model, the log odds for primarily used OS,
have a statistically significant relationship with the dependent
Apple’s cybersecurity reputation, Microsoft’s cybersecurity
variable.
reputation, and IT variables are -0.971, 1.048, -0.773, and
0.723, respectively, all of which are statistically significant. When we added Apple’s and Microsoft’s reputation to the
equation, similarly to our analysis of cybersecurity perception,
the log odds value of the primarily-used OS decreased from
Influence of Desired OS: In a final step, we added the desired -1.609 to -0.951 while keeping its statistical significance. This
OS variable to understand if a surveyee’s preference for an decrease demonstrates that by controlling the reputations for
OS makes a difference in their judgment of the cybersecurity privacy, we again captured some of the bias on the primarily
capabilities of that OS. We did indeed find a statistically sig- used OS.
nificant relationship between the desired OS and the dependent
variable. The log odds of the desired OS variable is -1.340 and Influence of IT Background: In the third regression, we
the odds ratio is 0.262. This demonstrates that individuals who added the IT variable. We found a positive relationship and
are using their desired OS in practice tend to perceive their OS statistical significance between the IT and the dependent
as being more secure than users on a platform that is not their variable. Furthermore, we controlled the computer usage, age,
favorite. education level, and gender variables. However, once again,

7
TABLE III. P RIVACY REGRESSION TABLE . O NLY SIGNIFICANT LOG ODDS VALUES , ROUNDED TO THE THIRD DECIMAL , ARE REPORTED . S IGNIFICANCE
LEVELS ARE INDICATED WITH STARS ( ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001). A LL ESTIMATES ARE BASED ON STANDARDIZED REGRESSION MODELS . A
DETAILED VERSION OF THIS TABLE IS AVAILABLE IN THE A PPENDIX .

Ordinal Logistic Regression:

Perceived privacy comparison of MacOS and Windows
(1) (2) (3) (4) (5) (6) (7) (8)
OS (Windows = 1) −1.609∗∗∗ −0.951∗∗∗ −0.969∗∗∗ −0.931∗∗∗ −1.006∗∗∗ −1.044∗∗∗ −1.099∗∗∗
Apple’s privacy 1.136∗∗∗ 1.165∗∗∗ 1.175∗∗∗ 1.172∗∗∗ 1.161∗∗∗ 1.240∗∗∗ 1.200∗∗∗
reputation
Microsoft’s privacy −0.793∗∗∗ −0.837∗∗∗ −0.844∗∗∗ −0.855∗∗∗ −0.835∗∗∗ −0.917∗∗∗ −0.945∗∗∗
reputation
IT 0.869∗∗ 0.774∗∗ 0.850∗∗ 0.852∗∗ 0.760∗ 0.718∗
Computer usage ... ... ... ... ...
Age ... ... ... ...
Education ... ... ...
Gender ... ...
Desired OS −1.569∗∗∗
Observations 208 207 205 205 205 205 205 205
Note: ∗ p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01

we could not find any significant relationship between these we introduced the IT, daily computer usage, and demographic
variables and the dependent variable. In the equation where we variables to our regression. The coefficient was 1.314 and the
added all the variables in our model, we can summarize the significance is preserved.
log odds variables of primarily-used OS, Apple’s reputation,
Backup Behavior: In the second model, we investigated
Microsoft’s reputation, and IT as -1.099, 1.240, -0.917, and
the performing backup behavior. The coefficient of the first
0.760. These four variables have a statistical significance with
equation is -0.715, and we found a statistically significant
the dependent variable.
relationship between primarily used OS and backup behavior.
Influence of Desired OS: As a final step, we switched the In the complete model, the coefficient of the primarily used
primarily used OS variable with the desired OS variable. We OS for performing backups became -1.157, and there was
found that the log odds of the desired OS is -1.569, and again a statistically significant relationship. That is, when we
statistically significant. That is, individuals who are using switched from a MacOS user to a Windows user, the log odds
their desired OS are predicted to perceive their OS as caring of backup behavior decreases by -1.157. Interestingly, we also
more about privacy than other users. Finally, Apple’s and found a statistical significant relationship between working in
Microsoft’s perceived privacy reputation variables and IT are IT job and backup behavior. When we switched from a non-
still significant, and their coefficients are 1.200, -0.945, and IT worker to an IT worker, the log odds of backup behavior
0.718. increases by 1.450. Another statistical significant relationship
was in computer usage. When we switched from a computer
C. User Behavior Regression Analysis user who uses its computer less than 1 hour to 1 to 4 hours or
8 to 12 hours, the log odds of backup behavior increases by
In this section, we investigated the relationship between the 3.778 and 3.970.
primarily used OS and the four cybersecurity and privacy be-
haviors (i.e., backing up files, covering the lens of the computer Covering of the Lens of the Computer Camera Behavior:
camera, using AV software, and using the browser in private In the third model, we checked the covering of the lens of the
mode). Table IV presents the behavioral regression results. computer camera behavior. We found that the coefficient of the
For this analysis, AV software usage, performing backups, and first model is 0.751 with a statistically significant relationship.
covering the lens of the computer camera are binary variables In the full model, the coefficient of the primarily used OS
where 0 represents a ‘no’ and 1 represents a ‘yes’. The usage of was 0.717, and its significance persisted. That is, when we
browsers in private mode variables is on a Likert scale (1-7). To switched from a MacOS user to a Windows user, the log
investigate this relationship, we formed a multiple regression odds of covering the lens of the computer camera behavior
model that we previously described in Section III-A. As AV increases by 0.717. Again, the relationship between being an
software usage, performing backups, and covering the lens of IT worker and covering the camera was statistically significant
the computer camera are binary variables, we use binomial and positive.
logistic regression for those dependent variables. For the usage Browsing in Private Mode: In the fourth model, we regressed
of browsers in private mode variable, we use ordinal logistic the usage of the private mode with the primarily used OS.
regression. The coefficient of the initial model is 0.108, and the full
model is -0.078. However, we could not find any significant
AV Usage Behavior: In the first model, we analyzed AV
relationship for the usage of the private mode variable. We only
usage. In this regression, we found that there is a statistically
found a statistically significant relationship between being an
significant relationship between primarily used OS and AV
IT employee and private mode usage.
usage. The coefficient of the primarily used OS is 1.293. This
positive correlation between the two variables means that when Overall, this analysis showed that there are some distinct
we switched from a MacOS user to a Windows user, the log cybersecurity behavioral differences between MacOS and Win-
odds of the usage of the AV increases by 1.293. Afterward, dows users. Specifically, we observed significant variations in

8
 TABLE IV. B EHAVIORAL REGRESSION TABLE . O NLY SIGNIFICANT LOG ODDS VALUES , ROUNDED TO THE THIRD DECIMAL , ARE REPORTED .
S IGNIFICANCE LEVELS ARE INDICATED WITH STARS (∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001). A LL ESTIMATES ARE BASED ON STANDARDIZED
REGRESSION MODELS . A DETAILED VERSION OF THIS TABLE IS AVAILABLE IN THE A PPENDIX .

Logistic (Binomial) Regression: Ordinal Logistic Regression:

AV Backup Cover Private
(1) (2) (3) (4) (5) (6) (7) (8)
OS (Windows = 1) 1.293∗∗∗ 1.314∗∗∗ −0.715∗ −1.157∗∗∗ 0.751∗∗∗ 0.717∗∗ 0.108 −0.078
IT 0.628 1.450∗∗ 0.770∗ 1.062∗∗∗
Computer usage ... ... ... ...
Age ... ... ... ...
Education ... ... ... ...
Gender ... ... ... ...
Constant 0.664∗∗∗ 15.171 1.827∗∗∗ 0.379 −0.579∗∗∗ 1.099
Observations 208 206 206 204 208 206 208 206
Log Likelihood −105.330 −99.515 −99.415 −86.973 −139.650 −130.500 NA NA
Akaike Inf. Crit. 214.661 231.031 202.831 205.945 283.300 293.000 NA NA
Note: ∗ p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01

TABLE V. PARTICIPANT COMMENTS ON WHY THEY CHOSE ONE OS OVER ANOTHER FOR CYBERSECURITY (W: W INDOWS USERS , M: M AC OS USERS )

User Opinion W M
MacOS is malware proof 8 27
The positive reputation of MacOS 12 15
The positive reputation of Windows 5 -
The “bad” reputation of Windows 1 -
Better technology on MacOS 4 14
Better technology on Windows 4 -
The enclosed ecosystem and better monitoring of MacOS (i.e., difficult to break) 5 2
Relatively small user base of MacOS provides better cybersecurity 6 8
Wide usage of Windows provides cybersecurity advantages (e.g., more people audit the system) 3 -
The financial power of Apple provides better cybersecurity 3 -
The financial power of Microsoft provides better cybersecurity 1 -
MacOS is a UNIX-based system, hence, making it automatically more secure - 5
Total 52 71

the use of AV software, the practice of performing backups, areas to acquire a feeling for how the surveyees perceive the
and the habit of covering the lens of the computer camera. cybersecurity of an OS.
These behavioral differences might stem from a belief held
1) Cybersecurity: We identified ten themes corresponding
by MacOS users that their platform is less susceptible to
to users’ operating system perception on cybersecurity, by
malware. Additionally, the behaviors might be influenced by
looking at the responses on why the participants chose one OS
the perception that MacOS offers superior cybersecurity and
over another for cybersecurity comparison. In summary, those
ensures greater privacy, as discussed in previous sections.
ten themes are: a) Same Security Outlook, b) Cybersecurity
Another interesting result was that, all else being equal, IT
Investment, c) Ecosystem Control, d) Low Vulnerability and
workers are more likely to perform backups, cover the lens of
Virus Incidence, e) Mac-specific Security, f) Reputation and
the computer camera, use the private mode of the browser. We
User Safety, g) Strong Security Measures, h) Achieve Cy-
hypothesize that these differences originate from the result of
bersecurity Excellence, i) Ease of Use and Familiarity, and
better awareness of the security and privacy issues.
j) User-Centric Approach.
It’s worth noting that during the analysis of the cyberse- Table V also presents the categorized responses, which
curity, privacy, and behavior regressions, we added interaction are malware resistance, reputation, better technology, a closed
variables into our models. However, these interaction variables ecosystem, user base size, wide usage, financial backing, and
did not demonstrate any significant relationships. OS’s foundation (i.e., UNIX-based). Interestingly, 27 MacOS
and 8 Windows users commented that MacOS is malware-
proof or less vulnerable to malware. For instance, a participant
D. Qualitative Analysis
said:
In this section, we present our analysis of cybersecurity
and privacy-related open-ended questions. We used NVivo P118: ”I think MacOS is less vulnerable as it is not
qualitative analysis software [16], and created a codebook with widely used as windows so hackers don’t target it”.
categories, themes, and codes from our qualitative data, as we Another participant stated:
previously described in Section III-B. The complete codebook
can be found in the Appendix. We also categorized the par- P119: ”I feel like MacOS is less susceptible to
ticipant responses to why they think that Windows or MacOS viruses and I have never had a virus on my mac
might offer more cybersecurity or privacy into different subject compared to my old windows computer”.

9
 Additionally, reputation was also one of the common topics V. M AIN F INDINGS AND ACTIONABLE R ESULTS
– 15 MacOS and 12 Windows users mentioned the positive rep-
In this section, we present the main findings and actionable
utation of MacOS whereas only 5 Windows users mentioned
results.
the positive reputation of Windows. Also, a higher number of
MacOS users believed that MacOS has better technology. For Confirming the folk wisdom: Many computer scientists and
example, a MacOS participant mentioned: users are aware of the folk wisdom that MacOS is superior
to Windows in terms of security (i.e., often referred to as the
P144: ”Their reputation shows that they have inter- “religious OS wars”). In this work, we studied the perception
nalized cyber security and make it a priority in their of users with respect to the security and privacy of these
products”. operating systems. The main finding of our work is that many
It is also worth noting that although 38 Windows users MacOS users appear to be convinced that their operating
were positive for MacOS cybersecurity, none of the MacOS system is better than Windows with respect to cybersecurity
users were positive for Windows. This observation aligns with and privacy. Notably, this sentiment also seems to be shared
our findings from the quantitative analysis section, where we by many Windows users as well. This finding, perhaps, is to
highlighted that the majority of the MacOS users believe that be expected, but we were able to verify it empirically and
MacOS is at least as good as Windows in cybersecurity. were able to confirm the folk wisdom that MacOS is indeed
perceived to be more “secure” by many users.
2) Privacy: We identified eight themes corresponding to
user operating system perception on privacy, by analyzing the Differences in cybersecurity habits: Our work also found that
responses on why the participants chose one OS over the there is a statistically significant difference between cyberse-
other for privacy. In summary, those eight themes include curity habits such as covering the camera lens of the computer,
a) Same Privacy Outlook, b) Customer Trust and Satisfac- performing backups regularly, and using AV software between
tion, c) Security and Protection, d) User Data Protection and MacOS and Windows users. In fact, MacOS users seem to
Enhancement Efforts, e) Customer-Centric Focus, f) Privacy- be taking some of these actions less seriously. The reasoning
Centric Business, g) Product Improvement and Quality, and h) can be identified by our qualitative analysis that showed that
Reputation and Media. MacOS users are confident that their system is malware-
proof, and has better technology and prestige. IT workers
Table VI also presents the categorization of user opinions. also seem to take cybersecurity habits more seriously. We
Just as with cybersecurity, these opinions are organized into found a statistically significant relationship between being an
reputation, privacy-centric technology, and the foundational IT worker and backup, covering the lens of the camera, and
type of OS (i.e., UNIX-based). Additionally, we identified using the private mode behaviors. We can hypothesize that IT
new themes, including personal sentiments and experiences, workers are more aware of the possible security and privacy
company-consumer relationships, the potential for data sales, issues.
and the extent to which privacy is wielded as a marketing
strategy. Reputation and primary OS matter: Our analysis showed
that the primarily used OS and the reputation of the vendor
Reputation was again one of the common subjects – 23
are the main determinants of the perception of more cyberse-
MacOS and 8 Windows users mentioned the positive privacy
curity and privacy. The difference between the perception of
reputation of MacOS. Interestingly, 13 MacOS and 4 Windows
MacOS and Windows users is, as we determined, statistically
users thought that Apple cares about the privacy of their users
significant. In this manner, OS developer companies, especially
and does not sell data. For instance, a participant expressed:
Microsoft, could benefit from trust-building campaigns that
P207: ”They want to keep updating their systems so highlight security enhancements to reposition their brand.
that they can keep protecting their customers”. Caution for organizations: Our empirical results suggest
Another participant stated: that the mental models of users do not match reality in this
space. We have shown that many MacOS users have the
P140: ”Apple wants to offer a high quality product perception that their OS will protect them against attacks and
so they are consistent in making sure that consumers’ that the risk of being compromised is less when compared to
products are safe to use through free updates of their Windows users. This perception may lead to a more relaxed
OS”. assumption about necessary security risks among MacOS
In contrast, only one Windows user noted that Microsoft users. We also showed that IT workers favor MacOS more
cares about the privacy of its users. We also observed that 9 than non-IT workers in the perceived cybersecurity and privacy
Windows users reported that MacOS has privacy-supporting comparison between the two operating systems. IT workers
technology and that it is automatically better than Windows play an important role in the security and privacy posture
because it is a UNIX-based system. For example, a participant of organizations. If this favoritism towards MacOS relaxes
said: some assumptions about risks among IT workers, it could have
tremendous effects on organizations. As a result, organizations
P205: ”Given that MacOS is Unix based it has better that have users on MacOS need to be especially prudent about
security measures built in”. the behaviors of their users and their inclination to be more
relaxed about necessary cybersecurity precautions.
Similar to the cybersecurity analysis, none of the MacOS
users mentioned positively for Windows. Also, fewer Windows Security training and OS improvements: We believe there
users mentioned positively for MacOS in the privacy compar- are two main determinants of improving the security and
ison, compared to the responses for cybersecurity. privacy of OS users. First, security training needs to focus

10
 TABLE VI. PARTICIPANT COMMENTS ON WHY THEY CHOSE ONE OS OVER ANOTHER FOR PRIVACY (W: W INDOWS USERS , M: M AC OS USERS )

User Opinion W M
The positive reputation of MacOS 8 23
The positive reputation of Windows 9 -
Privacy-supporting technology on MacOS 6 -
MacOS is a UNIX-based system, hence, making it automatically more secure 3 -
Good feelings and experience with MacOS 1 6
Privacy is a marketing strategy for MacOS 3 4
Apple cares more about their clients 4 5
Microsoft cares about their clients 1 -
MacOS does not share or sell user info with third parties - 8
More people use Windows so it is more likely to sell/leak our data - 2
Total 35 48

more on the role of users in maintaining their security posture. capture the majority of market share and user perceptions
Individuals should become aware of emerging trends and learn associated with these leading platforms.
what cannot be offered or guaranteed by operating systems.
Second, regardless of any misconceptions or flawed mental Survey questions and actionable results: To make sure that
models users may have, operating system vendors should our survey is easily understandable for average users and
incorporate as many built-in cutting-edge features as possible to ensure participant engagement, we kept the number of
to compensate the errors in the user mental models. For questions limited. However, this choice has the disadvantage
example, operating system vendors should add as much built- of having limited depth and consequently, a limited range of
in features as they can into operating systems (e.g., enhanced actionable insights we could derive. While we aimed for a
built-in AV programs, password managers, and a physical balance, there is always a trade-off between the length of
button for the microphone). Although some of the companies the survey and the participant’s attention span. While inherent
have already implemented or currently implementing some of biases such as brand reputation and personal OS preferences
the suggested precautions, they are not generally implemented play a significant role in shaping cybersecurity perceptions,
(e.g., many companies do not insert a physical cover for the the underlying beliefs are multifaceted. Self-report and recall
computer camera.) A recent example of a computer camera biases were also inevitable in our case since the participants
vulnerability [35] underscores that, although rare in advanced self-reported from their past experiences. Further studies might
operating systems, vulnerabilities are always possible. Also, consider exploring other factors, such as personal experiences
since the reputation of the companies is one of the main with malware or cyberattacks, exposure to advertising, or
determinants, the extent of the advertisements about cyber- peer opinions, to gain a more comprehensive understanding
security and privacy capabilities could be narrowed down of what drives these perceptions. We used having an IT
(e.g., each advertisement could contain a piece of advice to or cybersecurity-related job as an indicator for cybersecurity
improve cybersecurity awareness). Known techniques in other knowledge to shorten our survey. To improve the reliability of
domains such as behavioral economics and psychology could the results, future studies might use a better scale to measure
be helpful in this domain as well. For instance, the concept cybersecurity knowledge.
of nudging [30] can be used more systematically by operating
system vendors to modify a user’s perceptions when it is not
very realistic. Future work-1: An interesting future research direction would
be conducting interviews about personal experiences with
malware and cyber attacks. Subsequent studies can also focus
VI. L IMITATIONS AND F UTURE W ORK
on the effect of educating users, try to gain more insights into
Crowdsourcing platform: We performed experiments on why individuals do or do not take some security precautions,
Amazon Mechanical Turk. One potential limitation is the and how we can make these users feel more secure while
inherent bias that might come with this choice. Participants they are using their OS, by asking more detailed questions.
on this platform are likely to be more tech-savvy and familiar For example, researchers can inform the surveyees about the
with the intricacies of operating systems. One way to overcome detailed capabilities of each operating system, and then, ask
this issue is to directly engage with the participants. The them to compare the operating systems.
additional ability to interview participants face-to-face might
have been useful, and we might have had the opportunity to Future work-2: Given the ubiquity of mobile devices, explor-
ask follow-up questions. However, according to [25], our work ing user perceptions about mobile operating system security
on Amazon Mechanical Turk should be representative of the could offer invaluable insights. Comparing these findings with
U.S. population on the 18-49 age group who possessed at least desktop OS perceptions might also highlight unique challenges
some college-level education, with some skew towards more or beliefs specific to mobile platforms.
technology-inclined individuals. This skew, in our case, is not
a problem as we were interested in recruiting participants who
were computer users. Our purpose is to provide a secure, private, and stress-free
environment for users while helping them take some security-
The choice of operating systems: By focusing predominantly privacy precautions. This work was one of the steps toward
on two operating systems, MacOS and Windows, our study this goal. By understanding these, we will gain more insights
might inherently favor or overlook nuances present in other into the perception and, hence, will develop usable and secure
systems. Our intent, however, was strategic, as we aimed to solutions.

11
 VII. C ONCLUSION [13] J. E. Klobas, T. McGill, and X. Wang, “How per-
ceived security risk affects intention to use smart home
In this work, we conducted surveys with a large number devices: A reasoned action explanation,” Computers &
of MacOS and Windows users on Amazon Mechanical Turk. Security, vol. 87, p. 101571, 2019. [Online]. Available:
[Link]
We aimed to understand what the differences in perception
are among MacOS and Windows users with respect to the [14] M. Lindorfer, B. Miller, M. Neugschwandtner, and C. Platzer, “Take
a bite-finding the worm in the apple,” in 2013 9th International Con-
cybersecurity and privacy of these operating systems. Our ference on Information, Communications & Signal Processing. IEEE,
results confirm the folk wisdom and show that many Windows 2013, pp. 1–5.
and MacOS users indeed have the perception that MacOS is a [15] S. Lipner and M. Howard, “Inside the windows security push: A twenty-
more secure and private operating system when compared to year retrospective,” IEEE Security & Privacy, 2023.
Windows. At the same time, our results suggest that Windows [16] Lumivero, “NVivo - Lumivero,” (2023),
is overall considered by users not to be an insecure operating [Link]
system and that the reputation of the operating system has [17] Malwarebytes, “State of Malware Report,” Feb. 2020. [Online]. Avail-
indeed improved over the last decade through the investments able: [Link] State-of-
[Link]
made by Microsoft in its cybersecurity. We hope that under-
[18] P. Medelyan, A., “Coding Qualitative Data: How To
standing the perception of users in this regard will pave the way Code Qualitative Research (2023) — Thematic.” (2023),
for determining new trends, predicting potential new attack [Link]
surfaces, and proposing usable solutions. [19] Microsoft, “Keep your computer secure at home,” 2023,
[Link]
secure-at-home-c348f24f-a4f0-de5d-9e4a-e0fc156ab221.
ACKNOWLEDGMENT [20] I. Mohamed and D. Patel, “Android vs ios security: A comparative
study,” in 2015 12th International Conference on Information Technol-
This work was partially supported by the U.S. National Sci- ogy - New Generations, 2015, pp. 725–730.
ence Foundation (Awards: NSF-EAGER-2219920 and NSF- [21] Moonlock, “Mac Security Survey 2023,” 2023,
1663051), Cyber Florida, and Microsoft. The views expressed [Link] Security Survey [Link].
by the authors in this paper are their own, not those of the [22] A. J. O’Donnell, “When Malware Attacks (Anything but Windows),”
funding entities. IEEE Security Privacy, vol. 6, no. 3, pp. 68–70, 2008.
[23] R Core Team, R: A Language and Environment for Statistical
Computing, R Foundation for Statistical Computing, Vienna, Austria,
R EFERENCES 2023. [Online]. Available: [Link]
[24] C. Ravi and R. Manoharan, “Malware detection using windows api
[1] F. Al-Qershi, M. Al-Qurishi, S. Md Mizanur Rahman, and A. Al-Amri, sequence and machine learning,” International Journal of Computer
“Android vs. ios: The security battle,” in 2014 World Congress on Applications, vol. 43, no. 17, pp. 12–16, 2012.
Computer Applications and Information Systems (WCCAIS), 2014, pp.
1–8. [25] E. M. Redmiles, S. Kross, and M. L. Mazurek, “How Well Do My
Results Generalize? Comparing Security and Privacy Survey Results
[2] Amazon, “Amazon mechanical turk,” Oct. 2023. [Online]. Available: from MTurk, Web, and Telephone Samples,” in 2019 IEEE Symposium
[Link] on Security and Privacy (SP), 2019, pp. 1326–1343.
[3] H. L. Ball, “Conducting Online Surveys,” Journal of human lactation, [26] S. Sham, “6 steps to practice strong laptop security,” 2020,
vol. 35, no. 3, pp. 413–417, 2019. [Link]
[4] G. Belding, “Windows OS Security Brief History,” Oct. 2019. [On- security/.
line]. Available: [Link] [27] Statista, “Global market share held by operating systems
os-security-brief-history/ for desktop PCs, from January 2013 to July 2023,” (2023),
[5] Z. Benenson, F. Gassmann, and L. Reinfelder, “Android and ios users’ [Link]
differences concerning security and privacy,” in CHI’13 Extended windows-7/.
Abstracts on Human Factors in Computing Systems, 2013, pp. 817– [28] S. G. Stats, “Desktop operating system market share worldwide,”
822. Oct. 2023. [Online]. Available: [Link]
[6] Z. Benenson, O. Kroll-Peters, and M. Krupp, “Attitudes to it security share/desktop/worldwide
when using a smartphone,” in 2012 Federated Conference on Computer [29] V. M. Sue and L. A. Ritter, Conducting online surveys. Sage, 2012.
Science and Information Systems (FedCSIS), 2012, pp. 1179–1183. [30] R. Thaler and C. Sunstein, Nudge: Improving Decisions
[7] E. Chin, A. P. Felt, V. Sekar, and D. Wagner, “Measuring user about Health, Wealth, and Happiness, ser. A Caravan
confidence in smartphone security and privacy,” in Proceedings of the book. Yale University Press, 2008. [Online]. Available:
eighth symposium on usable privacy and security, 2012. [Link]
[8] H. Falaki, R. Mahajan, S. Kandula, D. Lymberopoulos, R. Govindan, [31] C. W. Turner, M. Zavod, and W. Yurcik, “Factors that affect the
and D. Estrin, “Diversity in smartphone usage,” in Proceedings of perception of security and privacy of e-commerce web sites,” in Fourth
the 8th international conference on Mobile systems, applications, and International Conference on Electronic Commerce Research, Dallas TX.
services, 2010, pp. 179–194. Citeseer, 2001, pp. 628–636.
[9] S. Flinn and J. Lumsden, “User perceptions of privacy and security on [32] F. Villegas, “Thematic Analysis: What it is and How to Do It. Ques-
the web,” in PST. Citeseer, 2005. tionPro.” (2023), [Link]
[10] Intel, “PC vs. Mac: The Big Debate,” Mar. 2021. [33] P. Vinod, R. Jaipur, V. Laxmi, and M. Gaur, “Survey on malware
[Online]. Available: [Link] detection methods,” in Proceedings of the 3rd Hackers’ Workshop on
tips-and-tricks/[Link] computer and internet security (IITKHACK’09), 2009, pp. 74–79.
[11] Y. Jin, J. Lim, I. Yun, and T. Kim, “Compromising the macos kernel [34] R. Wash, “Folk models of home computer security,” in Proceedings of
through safari by chaining six vulnerabilities,” in Black Hat 2020. the Sixth Symposium on Usable Privacy and Security, 2010, pp. 1–16.
Black Hat, 2020. [35] William Gallagher, “Apple pays record $100,500
[12] P. G. Kelley, “Conducting usable privacy & security studies with to student who found Mac webcam hack,” 2022,
Amazon’s Mechanical Turk,” in Symposium on Usable Privacy and [Link]
Security (SOUPS)(Redmond, WA. Citeseer, 2010. student-who-found-mac-webcam-hack.

12
[36] M. Williams, “Are Macs more secure than PCs? Not ◦ 1 - Poor Privacy
always. Here’s why,” Jan. 2021. [Online]. Available: ◦ 7 - Excellent Privacy
[Link]
• Q7: Which operating system are you primarily using?
[37] Y. Ye, T. Li, D. Adjeroh, and S. S. Iyengar, “A survey on malware ◦ Apple MacOS
detection using data mining techniques,” ACM Computing Surveys ◦ Microsoft Windows
(CSUR), vol. 50, no. 3, pp. 1–40, 2017.
[38] T. Yin, Z. Gao, Z. Xiao, Z. Ma, M. Zheng, and C. Zhang, “{KextFuzz}:
• Q8: Which operating system would you use if you
Fuzzing {macOS} kernel {EXTensions} on apple silicon via exploiting
were not constrained by price or your job?
◦ Apple MacOS
mitigations,” in 32nd USENIX Security Symposium (USENIX Security
◦ Microsoft Windows
23), 2023, pp. 5039–5054.
• Q9: Which operating system do you think offers more
cybersecurity? (1 to 7 Likert scale)
A PPENDIX ◦ 1 - Extremely Windows
◦ 7 - Extremely MacOS
A. Demographic Survey Questions
• Q10: Which operating system do you think cares more
• Q1: Which operating system are you primarily using? about privacy? (1 to 7 Likert scale)
◦ Chrome OS
◦ 1 - Extremely Windows
◦ Fedora
◦ 7 - Extremely MacOS
◦ Microsoft Windows
◦ Solaris • Q11: Why do you think this operating system offers
◦ Apple MacOS more cybersecurity?
◦ Free BSD
◦ Debian ◦ short answer text
◦ Other • Q12: Why do you think this operating system cares
• Q2: Which option best describes your age group? more about privacy?
◦ Under 18 ◦ short answer text
◦ 18 - 24
◦ 25 - 29 1) Behavior Questions:
◦ 30 - 39
◦ 40 - 49 • Q1: Are you using an antivirus program?
◦ 50 - 59
◦ Yes
◦ 60 - 69
◦ No
◦ 70 or older

• Q3: What is the highest level of education you have • Q2: Do you back up your data?
completed? ◦ Yes
◦ No
◦ Some High School
◦ High School Diploma • Q3: Do you cover your webcam with a physical cover?
◦ Some College ◦ Yes
◦ Associate Degree ◦ No
◦ Bachelor’s Degree
◦ Master’s Degree • Q4: In a typical week, how often do you use the
◦ Ph.D. degree or Higher private mode of your browser? (1 to 7 Likert scale)
• Q4: Worker ID ◦ 1 - Never
◦ 7 - Always
◦ short answer text

B. Main Survey Questions

• Q1: How much time do you spend on your computer
each day?
◦ Less than 1 hour
◦ From 1 hour up to 4 hours
◦ From 4 hours up to 8 hours
◦ From 8 hours up to 12 hours
◦ 12 hours or more

• Q2: Are you working in an Information Technology

(IT) or Cybersecurity-related job?
◦ Yes
◦ No

• Q3: How do you feel about Apple’s reputation in

cybersecurity? (1 to 7 Likert scale)
◦ 1 - Poor Cybersecurity
◦ 7 - Excellent Cybersecurity

• Q4: How do you feel about Apple’s reputation in

privacy? (1 to 7 Likert scale)
◦ 1 - Poor Privacy
◦ 7 - Excellent Privacy

• Q5: How do you feel about Microsoft’s reputation in

cybersecurity? (1 to 7 Likert scale)
◦ 1 - Poor Cybersecurity
◦ 7 - Excellent Cybersecurity

• Q6: How do you feel about Microsoft’s reputation in

privacy? (1 to 7 Likert scale)

13
 TABLE VII. D ETAILED CYBERSECURITY REGRESSION TABLE

Ordinal Logistic Regression:

Perceived cybersecurity comparison of MacOS and Windows
(1) (2) (3) (4) (5) (6) (7) (8)
OS (Windows = 1) −1.403∗∗∗ −0.963∗∗∗ −0.989∗∗∗ −0.941∗∗∗ −0.937∗∗∗ −0.897∗∗∗ −0.971∗∗∗
(0.262) (0.271) (0.275) (0.287) (0.291) (0.293) (0.298)

Apple’s cybersecurity 0.948∗∗∗ 0.962∗∗∗ 0.986∗∗∗ 1.017∗∗∗ 1.028∗∗∗ 1.048∗∗∗ 0.983∗∗∗

reputation (0.122) (0.124) (0.126) (0.129) (0.130) (0.131) (0.132)

Microsoft’s cybersecurity −0.666∗∗∗ −0.701∗∗∗ −0.726∗∗∗ −0.746∗∗∗ −0.746∗∗∗ −0.773∗∗∗ −0.798∗∗∗

reputation (0.114) (0.117) (0.119) (0.120) (0.121) (0.123) (0.124)

IT 0.816∗∗ 0.747∗∗ 0.794∗∗ 0.695∗ 0.723∗ 0.677∗

(0.369) (0.375) (0.374) (0.381) (0.385) (0.388)

From 1 hour −1.483 −1.272 −1.244 −0.455 −0.757

up to 4 hours (1.233) (1.228) (1.238) (1.427) (1.389)

From 4 hours −1.252 −1.042 −0.997 −0.309 −0.661

up to 8 hours (1.192) (1.187) (1.202) (1.376) (1.345)

From 8 hours −1.300 −1.141 −1.134 −0.469 −0.825

up to 12 hours (1.206) (1.198) (1.209) (1.376) (1.345)

12 hours −1.009 −0.837 −0.709 −0.037 −0.377

or more (1.273) (1.266) (1.275) (1.450) (1.410)

Age 25-29 0.004 −0.020 −0.068 −0.315

(0.492) (0.504) (0.508) (0.514)

Age 30-39 −0.248 −0.374 −0.485 −0.505

(0.472) (0.484) (0.490) (0.498)

Age 40-49 0.311 0.151 0.134 −0.041

(0.525) (0.540) (0.542) (0.541)

Associate −0.223 −0.229 −0.374

degree (0.558) (0.560) (0.565)

Bachelor’s −0.137 −0.128 −0.079

degree (0.401) (0.403) (0.409)

Master’s 0.422 0.469 0.412

degree (0.460) (0.463) (0.466)

Ph.D. −0.217 −0.009 0.144

degree or higher (0.907) (0.918) (0.933)

Female −1.286 −1.023

(1.099) (1.102)

Male −0.859 −0.511

(1.092) (1.097)

Desired OS −1.340∗∗∗
(0.307)

Observations 208 207 205 205 205 205 205 205

∗ ∗∗ ∗∗∗
Note: p<0.1; p<0.05; p<0.01

14
 TABLE VIII. D ETAILED PRIVACY REGRESSION TABLE

Ordinal Logistic Regression:

Perceived privacy comparison of MacOS and Windows
(1) (2) (3) (4) (5) (6) (7) (8)
OS (Windows = 1) −1.609∗∗∗ −0.951∗∗∗ −0.969∗∗∗ −0.931∗∗∗ −1.006∗∗∗ −1.044∗∗∗ −1.099∗∗∗
(0.271) (0.283) (0.287) (0.294) (0.300) (0.308) (0.310)

Apple’s privacy 1.136∗∗∗ 1.165∗∗∗ 1.175∗∗∗ 1.172∗∗∗ 1.161∗∗∗ 1.240∗∗∗ 1.200∗∗∗

reputation (0.127) (0.129) (0.131) (0.132) (0.133) (0.138) (0.139)

Microsoft’s privacy −0.793∗∗∗ −0.837∗∗∗ −0.844∗∗∗ −0.855∗∗∗ −0.835∗∗∗ −0.917∗∗∗ −0.945∗∗∗

reputation (0.117) (0.120) (0.121) (0.121) (0.122) (0.127) (0.127)

IT 0.869∗∗ 0.774∗∗ 0.850∗∗ 0.852∗∗ 0.760∗ 0.718∗

(0.378) (0.385) (0.392) (0.394) (0.395) (0.395)

From 1 hour −0.332 −0.139 −0.100 −0.350 −0.939

up to 4 hours (1.418) (1.464) (1.497) (1.853) (1.990)

From 4 hours 0.220 0.388 0.471 0.123 −0.406

up to 8 hours (1.382) (1.426) (1.472) (1.829) (1.982)

From 8 hours 0.192 0.364 0.478 0.069 −0.529

up to 12 hours (1.392) (1.435) (1.478) (1.835) (1.983)

12 hours 0.204 0.325 0.517 0.096 −0.390

or more (1.453) (1.493) (1.531) (1.884) (2.024)

Age 25-29 0.252 0.302 0.228 0.022

(0.532) (0.545) (0.550) (0.552)

Age 30-39 0.322 0.267 0.094 0.121

(0.504) (0.514) (0.522) (0.526)

Age 40-49 0.909∗ 0.814 0.708 0.626

(0.552) (0.573) (0.575) (0.571)

Associate −0.477 −0.487 −0.766

degree (0.572) (0.592) (0.593)

Bachelor’s −0.664 −0.634 −0.598

degree (0.417) (0.418) (0.420)

Master’s −0.181 −0.087 −0.101

degree (0.463) (0.466) (0.464)

Ph.D. 0.001 0.364 0.476

degree (1.013) (1.034) (1.055)

Female 0.464 0.877

(1.139) (1.270)

Male 1.370 1.842

(1.138) (1.275)

Desired OS −1.569∗∗∗
(0.322)

Observations 208 207 205 205 205 205 205 205

∗ ∗∗ ∗∗∗
Note: p<0.1; p<0.05; p<0.01

15
 TABLE IX. D ETAILED BEHAVIORAL REGRESSION TABLE

Logistic (Binomial) Regression: Ordinal Logistic Regression:

AV Backup Cover Private
(1) (2) (3) (4) (5) (6) (7) (8)
OS (Windows = 1) 1.293∗∗∗ 1.314∗∗∗ −0.715∗ −1.157∗∗∗ 0.751∗∗∗ 0.717∗∗ 0.108 −0.078
(0.362) (0.402) (0.366) (0.441) (0.284) (0.318) (0.244) (0.267)

IT 0.628 1.450∗∗ 0.770∗ 1.062∗∗∗

(0.563) (0.701) (0.414) (0.360)

From 1 hour −15.484 3.778∗∗ 0.773 1.155

up to 4 hours (942.637) (1.927) (1.688) (1.726)

From 4 hours −15.803 2.558 −0.027 0.953

up to 8 hours (942.637) (1.819) (1.656) (1.700)

From 8 hours −15.449 3.970∗∗ 0.044 1.411

up to 12 hours (942.637) (1.871) (1.657) (1.706)

12 hours −16.323 2.621 −0.381 0.878

or more (942.637) (1.904) (1.730) (1.744)

Age 25-29 0.784 −0.699 −0.143 0.539

(0.625) (0.868) (0.570) (0.459)

Age 30-39 0.732 −0.661 0.001 0.388

(0.595) (0.847) (0.551) (0.446)

Age 40-49 0.822 −0.455 0.601 0.498

(0.688) (0.908) (0.610) (0.499)

Associate −1.149 −0.229 −1.025 0.116

degree (0.829) (0.770) (0.669) (0.554)

Bachelor’s −0.775 −0.060 −0.211 0.184

degree (0.642) (0.573) (0.460) (0.391)

Master’s −0.623 0.292 −0.787 0.320

degree (0.712) (0.684) (0.526) (0.443)

Ph.D. −1.542 15.413 −0.054 1.258

degree or higher (1.188) (1,014.996) (1.058) (0.997)

Female 1.198 −0.988 −1.634 −0.292

(1.288) (1.643) (1.416) (1.024)

Male 0.975 −1.117 −1.614 −0.526

(1.291) (1.627) (1.411) (1.026)

Constant 0.664∗∗∗ 15.171 1.827∗∗∗ 0.379 −0.579∗∗∗ 1.099

(0.208) (942.636) (0.288) (1.973) (0.205) (1.732)

Observations 208 206 206 204 208 206 208 206

Log Likelihood −105.330 −99.515 −99.415 −86.973 −139.650 −130.500 NA NA
Akaike Inf. Crit. 214.661 231.031 202.831 205.945 283.300 293.000 NA NA
Note: ∗ p<0.1; ∗∗
p<0.05; ∗∗∗
p<0.01

16
Category Sub-Category Theme Code Description Ref.
Neither offers more security I don't think any operating system offers more cyber security. 4
Both OS Same security Outlook
Roughly equal I think they're roughly equal 27
Money to spend on cyber security It has more money to spend on it 2
More advanced and proactive I feel that they are more advanced and prepared to be proactive instead of reactive. 3
Cybersecurity Investment Not impacted by viruses or other threats Apple has a history of not being impacted by viruses or other threats. They are known in the industry as being one of the most secure 7
systems.
Put more money and work into their security They have put more money and work into their security system. 2
Better designed Better designed 2
Enclosed ecosystem a more enclosed ecosystem 1
Feels secure with simplistic layout It just feels more secure with a simplistic layout. 1
Inability to customize or download third party The inability to customize or download third party apps 1
apps
Ecosystem Control
Less enticing to hackers Mac is less enticing to hackers. 1
Less modifiable which provides more security It's less accessible, less modifiable which means it's more secure. 1
Limits what can be installed I think MacOS does a good job about limiting what can be installed on their devices 1
Not open-sourced or require anti-virus software MacOS is not open-sourced and generally does not require anti-virus software. 1
Safer due to apps being vetted Heard something about it being safer since apps are vetted 1
Harder creating viruses harder to make viruses for 3
Less issues with viruses I have had way less issues with viruses since switching to a MAC 6
Less prevalent Less prevalent, so less of a target. 3
Low Vulnerability and Virus Incidence Less prone to viruses Apple systems are not prone to viruses like Windows based operating systems are. 15
Less report of viruses People report fewer viruses 1
Less vulnerable due to not being widely used I think MacOS is less vulnerable as it is not widely used as windows so hackers don’t target it 3
Have not experience slowness I have never experienced slowness on my Mac 2
Have not heard issue with Mac system I hardly ever hear of any issues with this system 1
Mac-specific Security MacOS rarely having virus or malware It is rare to hear about a MacOS device having a virus or malware on it. 4
Not much hacking due to lower market share MacOS is more secure just because people aren't trying to hack it as much because it has lower market share 3
Good reputation of being secure I feel that MacOS has a good reputation as being very secure. 18
Mac OS Greater transparency There are just less viruses written for OSX. Also, it's built on a Linux kernel, there's greater transparency as to what's going on under the 1
hood. You can always issue a ”top” command and ”kill -9” any rouge processes.
Known for privacy and innovation Apple is known for their privacy and innovation in addition to their focus on the customer. 2
Reputation and User Safety Offers superior security Apple products in general offer superior security. 2
Protect user's information I have heard explicitly that Apple protects users' info, while Microsoft profits off of it 1
Reputation shows internalized cyber security Their reputation shows that they have internalized cyber security and make it a priority in their products 1
Their reputation Just the reputation 1
User safety To keep the users safe 2
Attack resistant and less exploitable I feel as if Apple is more attack-resistant and less exploitable than Windows. 5
Better security measures built in Given that MacOS is Unix based it has better security measures built in. 1
Browser add-ons Free add-ons for the browser 1
Built in anti-virus software It has built in anti-virus software 1
Good security features by providing fingerprint In MacOS speed, security features are good compared to windows. It provides fingerprint authentication; it blocks the lot of unwanted 2
authentication and blockage of unwanted activities activities.
OS Good security measures like encryption Apple is known for their security measures, including their encryption, so it feels more safe 2
Cybersecurity
High priorities on security I think the company places higher priorities on security 1
Perception Strong Security Measures
Higher security level They simply have a higher level of security which is very hard to breach 5
Offers fingerprint authentication and drive encryp- It offers fingerprint authentication and drive encryption with the T2 Security Chip, and will generally be safer running a non-Windows 1
tion operating system
Provides Mitigation Strategies Provide Mitigation Strategy Guidance 1
Secures personal information for securing personal information and preventing attack 2
Superior in cyber security I have read about how superior it is in cyber security 1
TouchID security They secure everything with touchID which can't be replicated 1
Virus proof I've heard MacOS is almost virus proof. 2
Vocal about privacy Apple is very vocal about privacy when other companies aren't. 2
Been around longer with more cybersecurity im- I think they have been around longer and have had more work with cyber security 3
provements
Built-in antivirus that scan and remove malware Windows has a great built in antivirus that scans and removes malware automatically. 1
Built-in safeguards and firewalls Windows has more built in safeguards and firewalls as compared to Mac. 1
Heavily invested on security and privacy Microsoft are a big company that is heavily invested on security and privacy 1
More defence and better encryption More line of defence and better encryption codes 1
Achieve Cybersecurity Excellence
More software developed There's just more software developed for Windows, so there are more options for security. 2
Not many reports of attacks less reports of attacks. 1
Prevent attacks I think they both offer it to prevent attacks. 2
Protects identities, device and information Protects identities, device and information. 1
Removes malware automatically Windows has a great built in antivirus that scans and removes malware automatically. 1
Tested and developed necessary tools for robust I believe that because this operating system has been around longer and is more widely used in business and government, that they have 1
cybersecurity tested and developed the necessary tools to provide more robust cyber security.
Windows OS Don't have issue I have always used Windows and have never had an issue. 2
Familiarity with system I am more familiar with it. 3
Feel safer to use Safer to use 1
Harder to install programs on MacOS Because it's harder to install programs on MacOS 1
Ease of Use and Familiarity
Less concerned gathering and selling user data Windows isn't as concerned with gathering and selling personal data to advertisers 1
No keylogger it doesn't have a keylogger 1
Widely used OS Because I think that it is the operating system that is used more worldwide. 1
Broader user base Broader user base 3
Built-in programs for user protection It has built in programs so even the most stupid and non-techsavvy consumers get some protection. 3
Cares about customers Microsoft cares more about its customers. Apple just wants money. 1
User-Centric Approach
More advanced Because they are more advanced compared to others. 1
User trust It is the one I have always used and trust. 3

17
Category Sub-Category Theme Code Description Ref.
Both OS care due to money I think they both care because they both care about money 1
Equally comparable with privacy I feel that both systems are comparable when it comes to privacy. 25
Both OS Same Privacy Outlook
Neither are better than the other in privacy I don’t think either brand is really better than the other in the privacy department. They both have their faults. 14
No good privacy policies They both do not have good privacy policies. 1
Better reputation To maintain a better reputation 6
Consumer trust It wants people to trust them. 2
Create products for consumers that are safe Apple wants to offer a high-quality product so they are consistent in making sure that consumers’ products are safe to use through free 1
updates of their OS.
Customer centric product They have always had a customer-centric product and are known for their superior privacy. 1
Does not care about its customers Apple doesn’t care about its customers 1
For the public image Public image more than anything else. 1
Goes the extra mile to satisfy their customers Apple goes the extra mile for their customers 1
Customer Trust
and Satisfaction Maintain good security with updates They make a focused effort to maintaining great security with updates 1
Marketing purposes Marketing purposes 1
More consumers and wants to keep good image I think because it has more consumers and wants to keep its good name it tries to take care of them 1
Provide better products for their customers It wants to provide the best privacy for their customers for them to purchase their wide variety of apple products 1
Reputation for being invasive Apple has a reputation for being incredibly invasive. 1
To emphasize they prioritize privacy Tends to emphasize that they prioritize privacy 1
Value their clients They value their clients more as compared to others. 1
Continues to create more privacy updates they keep creating more privacy updates 1
Crucial for modern day It is a crucial aspect in modern day 1
Devices are used widely They have to care more because their devices are used more widely 1
Does not allow anyone to decrypt devices easily Apple as a whole seems to care more than others. See law enforcement anger at their refusal to decrypt devices of the accused. 2
Keep user information safe I think that their operating system has more security features to keep your information safe. 2
Security and Protection
Mac OS Keeps updating system to protect customers They want to keep updating their systems so that they can keep protecting their customers 1
Protect consumers They want to protect their consumers. 1
To prevent hacking or surveillance it has improved its data security by preventing hardware that leads to hacking or surveillance 1
Allow users to browser privately I feel like I can use private browsing modes and shut off my information from being shared. For example, private browsing. 1
Allow users to encrypt files I feel it is a bit more secure and my files are encrypted better, ensuring more privacy 1
Cares about user privacy I believe that Apple as a company strongly cares about its users privacy 5
OS Caring
About Privacy Corrected privacy flaws Apple has acknowledged and corrected their privacy flaws in the past 1
Perception Creating a secure place Because they choose to put money and effort into creating a secure place. 4
User Data Protection and Enhancement Efforts
Protect privacy to cover up their actions I think Apples reputation for wanting to protect privacy has more to do with covering up their own actions and not consumer privacy 2
Recognize the need for privacy and embrace it I feel that they have recognized the need for privacy and embraced it. 1
Transparency in privacy policies You are always shown privacy policies and asked if you want to share your data 1
Upfront controls after installation Upfront controls immediately after OS installation 1
Cares about their customers Because I believe it has more customers, and would like to do the best possible to keep them satisfied. 3
Customer-Centric Focus Keep customer satisfied Because I believe it has more customers, and would like to do the best possible to keep them satisfied. 1
Users trust Because they want users to trust them and use their products 2
Bad for business if privacy is not integrated its bad business for them if they don't 1
Privacy controls Upfront controls immediately after OS installation 2
Privacy is part of innovation Windows cares more about privacy because it comes from a smarter company. Apple has only made slight tweaks to an iPhone while 1
Microsoft continues to innovate. Privacy is part of that innovation.
Protect personal life Want to make sure you protect as much of personal life 2
Privacy-Centric Business Protecting user More geared toward protecting the user 1
Provide user flexibility over settings It offers more user flexibility, and control over security settings. 1
Provides opt out option to protect user privacy They give you the option to opt out of this and that, so you can think that they take your privacy safety. 1
Use privacy as a selling point for their product They use it as a selling point for their product. 1
Windows OS Value transparency and honesty on keeping user I feel that Windows cares more because they are more open and honest about their operating system and how they keep users safe. 2
safe
Improve products They want their products to be top notch 1
Product Improvement and Quality
To keep bad stuff out They work harder to keep the bad stuff out. 1
Keep collecting user's data To keep collecting your data 1
Less worried about media They are less worried about being in the media and integrating with social media 1
Reputation and Media Not concerned with data gathering They are not as concerned with data gathering so their privacy policy is more geared toward protecting the user 1
Protects reputation Because it protects their reputation 3

18