Users’ Willingness to Share Data on the Internet: Perceived Benefits

and Caveats

Martina Ziefle, Julian Halbey and Sylvia Kowalewski

Human-Computer Interaction Center, RWTH Aachen University, Campus-Boulevard 57, 52074, Aachen, Germany

Keywords: Internet, Social Network Sites, Perceived Privacy, Willingness to Share Data, Trust, Gender, Conjoint Study.

Abstract: One of the major challenges of the ongoing digitalization and the ubiquitous usage of pervasive computing in

all fields of our lives is to steer a sensible balance between benefits and drawbacks of using the Internet and

to implement an appropriate data handling when using digital media. The broad availability of data, in line

with the enormous velocity of information retrieval, is open to abuse and malpractice, with privacy threats as

the most serious barrier. The consumers and their attitudes and behaviors when using the Internet play an

important role in the discussion about privacy protection. The aim of the current study was to analyze Internet

usage behaviors and users’ willingness to share their data when using digital services and social network sites.

In a two step empirical approach, we first explore users’ perceptions of privacy in the context of Internet usage

and social network sites by means of a focus group approach. In a second step, a quantitative study was carried

out. Using a conjoint measurement approach, user scenarios were created from combinations of different

levels of anonymization extent, data type, and benefits from sharing the data. The respondents’ task was to

decide under which conditions they would be willing to share their data. 80 volunteers (50,6% women)

between 14 and 60 years of age participated in the conjoint study.

1 INTRODUCTION

Historically, never might there have been a bigger

technological challenge for democratic societies than

the sensible, responsible, and open-minded handling

of Big Data that goes hand in hand with the

tremendous chances and drawbacks of the Internet of

Things (Karabey, 2012, Katal et al., 2013). In the

IDC's sixth annual study of the digital universe

(Gantz and Reinsel, 2012), a comprehensive

overview of the Big Data challenge is outlined. Using

a longitudinal analysis approach (comprising data

collection between 2005 until 2020), the authors

provide a detailed report about the reach of digital

data, their quantity and the significant growth of data

over the respective time. Quoting from the executive

summary, analyses up to 2012 showed:

“From 2005 to 2020, the digital universe will grow

by a factor of 300, from 130 exabytes to 40,000

exabytes, or 40 trillion gigabytes (more than 5,200

gigabytes for every man, woman, and child in 2020).

From now until 2020, the digital universe will about

double every two years. […] Between 2012 and 2020,

emerging markets' share of the expanding digital

universe will grow from 36% to 62%. A majority of

the information in the digital universe, 68% in 2012,

is created and consumed by consumers -watching

digital TV, interacting with social media, sending

camera phone images and videos between devices

and around the Internet, and so on. Yet enterprises

have liability or response-bility for nearly 80% of the

information in the digital universe. They deal with

issues of copyright, privacy, and compliance with

regulations even when the data zipping through their

networks and server farms is created and consumed

by consumers. […] The amount of information

individuals create themselves — writing documents,

taking pictures, down-loading music, etc. — is far

less than the amount of information being created

about them in the digital universe” (ibid, pp. 1-2).

From this impressing and at the same time

alarming data report, three major points should be

noted:

(1) The ongoing digitalization in all fields of our

life enables huge benefits for consumers on different

levels – on a private and a professional as well as on

a societal level.

(2) The availability of data and information opens

up a significant knowledge gain in – for societies –

Zieﬂe, M., Halbey, J. and Kowalewski, S.

Users’ Willingness to Share Data on the Internet: Perceived Beneﬁts and Caveats.

DOI: 10.5220/0005897402550265

In Proceedings of the International Conference on Internet of Things and Big Data (IoTBD 2016), pages 255-265

ISBN: 978-989-758-183-0

255

very important fields: medical and health issues,

mobility and transport, production and business,

learning and education, to just mention a few.

(3) Naturally, the broad availability of data, the

measurability of information in important branches,

and the enormous velocity of information gathering

lends itself to abuse and malpractice. The damage is

– though not exactly calculable – huge and can

happen on different levels (privately, professionally,

societally). Among the most prominent drawbacks,

the danger of privacy loss by eaves-dropping as well

as shoulder reading, technologies come to the fore.

Consumer behaviors are tracked while using social

network sites and while surfing the Internet (Takabi

et al., 2010; Szongott et al., 2012).

For private consumers, but also for the

professional context, it is thus important to find a

sensible balance between the benefits and drawbacks

of using the Internet and establish an appropriate data

handling when using digital media. This balance can

only be reached by a broad understanding of Internet

usage behaviors and a joint approach from different

disciplines (Dartmann et al., under revision). In order

to achieve a far-reaching privacy preservation

strategy, different research and policy approaches

can be observed. While technical disciplines work

mostly on the development of privacy enhancing

technologies (k-anonymity or differential privacy

technologies, e.g., Sweeny, 2002; Dwork, 2006,

Dritsas et al., 2006), there is, naturally, a prominent

research input from the field of law and legal

regulations (Mayeda et al., 2016, Trestenjak, 2016),

as well as approaches to support privacy protection

behaviors in ecommerce and digital services from an

economic and market perspective (Phelps et al.,

2001; Matsusaki, 2016). Increasingly, normative

studies dealing with the establishment of a digital

etiquette and guidelines for safe and conscious

digital behavior are pursued. Already in 2004,

Ribble, Bailey and Ross (2004) claimed that across

all of society there is a need to establish a mindset of

technology appropriate behaviors in the context of

technology education, referred to as Digital

Citizenship.

“Digital citizenship can be defined as the norms of

behavior with regard to technology use. As a way of

understanding the complexity of digital citizenship

and the issues of technology use, abuse, and misuse,

we have identified nine general areas of behavior that

make up digital citizenship” (Ribble et al., 2004, p. 7).

In their work, the authors claim different usage

policy behaviors that should to be respected by

teachers and students when interacting with

technology (Ribble et al., 2004). Recent work

strengthens this view (Hornung and Schnabel, 2009;

Rouvroy and Poullet, 2009; Young and Quan-Haase,

2013). It is claimed that media competency as well

as knowledge about potential dangers and caveats in

the context of Internet usage and pervasive

computing have to be globally developed. The

responsibility, though, is not limited to policy, law,

and education entities, but it is shifted to end users

and consumers, referred to as the individual

responsibility of control of own personal data and

informational self-determination (Rouvroy and

Poullet, 2009; Tene and Polonetsky, 2012). Then,

individual consumers must take care of their self-

protection when using ecommerce services and

digital media, and – beyond technical, legal, and

policy efforts – consumers have the duty to be

responsible for their electronic actions and deeds.

Beyond this warranted claim for the

establishment of such digital ethics and the broad

public awareness that the responsibility for own

digital behaviors is a necessity (Marx, 1998), still,

from a social science point of view, there are some

concerns, if not doubts, that this mere (normative)

and rational claim is effective in the end (Lauffer and

Wolff, 1977; Kalwar, 2008). The concerns are based

on empirical findings which corroborate that users’

behaviors and their mental mindset or rationality

diverge widely.

Privacy Paradox. Users – though being seriously

concerned when asked about the fear of losing their

privacy in the digital universe – nevertheless do

mostly not protect themselves and are neither

discreet nor careful with their personal data in the

Internet (Lahlou 2008; Kowalewski et al., 2015;

Boyd and Hargittai, 2010). This phenomenon is

called privacy paradox (Awad et al., 2006; Norberg

et al., 2007).

Perceived Privacy. The factual risk of privacy loss

is not identical with the perceived risk of privacy loss

and the perception of control (Spiekermann, 2007).

Users might have not only a different understanding

of dangers and caveats and a different perception of

control but also a different appreciation of the

temporary benefits of using social network sites.

Thus, users might decide to take the risk of data

sharing as they perceive to be in control. Likewise,

users might decide to share data because the

temporary benefit is higher for them than the

potential risk.

Context. The perceived benefits, which might

motivate users to share their data on the Internet, and

the perceived risks, which might prevent users from

sharing their data, may also be impacted by the

IoTBD 2016 - International Conference on Internet of Things and Big Data

256

respective usage context (Nissenbaum 2011;

Kowalweski et al, 2015). Recent work showed that

the tradeoff between benefits and barriers are quite

different in a medical context (when it comes to

intimate data concerning personal health and illness,

Wilkowska and Ziefle, 2012) in comparison to the

sports context in which users might be even keen to

share data (van Heek et al., 2014). Also, the closer

digital data are related to personal homes, the more

observant and reluctant are consumers to allow

others to use and see their personal data (Ziefle et al.,

2011; van Heek et al., 2015; 2016).

Culture. Understanding digital behaviors and

privacy concerns requires also the understanding of

the respective situation and culture (Hargittai, 2007).

Krasnova and Veltri (2010) examined privacy

concerns of Facebook members in the USA and

Germany. The findings revealed culture-specific

differences: While German users expect more

negative consequences (identity loss, damage, and

privacy-related violations), American users – though

also aware of cyber crime and privacy violations –

still do report higher trust in the service provider and

higher levels of perceived control. Thus, the

perception of privacy varies between different

cultures.

User Diversity. User diversity is another important

factor with respect to Internet behaviors. Here,

several cognitive and affective factors might be

relevant: One is the users’ (low) level of knowledge

about factual risks and malpractice (Kowalewski et

al., 2015). A second factor relates to different

personalities and interacting styles (those who are

willing to take risks vs. those who are more fearful,

e.g., Karim et al., 2009). A third factor relates to the

competency in using technical devices and digital

media. The digital competency shapes the way users

interact with the Internet (Akther, 2014) and is

indirectly related to age effects (as persons with a

different upbringing with technology do have

another mental model of how technology works, e.g.,

Fogel and Nehmed, 2009; Freestone and Mitchell,

2004).

Gender. The usage of digital services and social

network sites is gender-sensitive (Kennedy et al.,

2003; Kim et al., 2007). The same applies for privacy

concerns when using digital services which are more

pronounced in women (Fogel and Nehmad, 200),

Wilkowska et al., 2010). On the one hand, there is

extensive research evidence that women – in contrast

to men – report a lower self-efficacy when using

digital devices which makes them much more careful

in the interaction with technology (Durndell and

Haag, 2002). In addition, frequent social network site

users show greater risk taking attitudes (Fogal and

Nehmad, 2009), and are usually male. In line with

this, women report higher privacy concerns and fears

of being victims of cyber crime (Halder and

Jaishankar, 2011). On the other hand, women

outperform males in terms of electronic

communication with peers on social network sites.

Women show a higher emotional involvement and

social engagement in digital communication (Sun et

al., 2016), accompanied by highly positive collective

self-esteem and motivation to befriend peers and stay

in contact with them (Barker, 2009; Thelwall et al.,

2008; 2010).

2 QUESTIONS ADDRESSED AND

LOGIC OF PROCEDURE

In this study, we take a social science perspective and

argue that an effective policy of privacy protection

behaviors cannot be effective and sustainable before

we understand the behaviors of Internet users by

means of empirical evidence. Therefore, a two-step

empirical procedure was undertaken. In a first step,

focus groups were carried out in which participants

discussed the perceived benefits and caveats of

sharing their data as well as their general experience

with the Internet and the usage of social network

sites. On the base of these argumentation lines, a

quantitative conjoint study was carried out in order

to study respondents’ privacy preferences and their

willingness to share their data.

Characteristically for the conjoint measurement

approach is that selected attributes (probability of

being identified (extent of anonymization), the data

type and benefits from sharing the data) are

combined to different usage scenarios for which

participants had to decide if they would be willing to

share their data under the respective conditions. This

procedure allows us to empirically describe which of

the factors is relevant to which extent and in what

way might other conditions modulate the willingness

to share their data.

As gender was revealed to be a decisive factor for

Internet behavior, a comparison between female and

male users was focused at.

The following questions guided this research.

(1) What are the most important factors for Internet

usage and willingness to share data?

(2) What is the worst case scenario under which

participants would not share their data at all and

which is the best case scenario?

Users’ Willingness to Share Data on the Internet: Perceived Beneﬁts and Caveats

257

(3) Are there gender differences in the preferences

and the decision scenarios?

3 METHODOLOGY

We followed a two step empirical procedure.

Focus Group Prior to the main study, focus groups

were carried out in which participants (N = 8, 20-50

years of age) discussed the most important issues with

respect to privacy and Internet usage. The

participants’ argumentation lines were analyzed in

detail (not reported here). The factors that have been

evaluated as most important (benefits and barriers) by

participants were taken as basis for the selection of

attributes used in the conjoint analysis.

Conjoint Analysis A choice-based conjoint analysis

approach was selected, as it mimics the complex

decision processes in real world scenarios in which

users have to evaluate more than one attribute that

influences the final decision (Luke and Tukey, 1964).

Contrary to traditional surveys, in which participants

answer single factors separately from each other,

conjoint analyses simulate real-world user decisions

in which users weigh potential benefits against

perceived barriers. In the context here, the tradeoff

between keeping one’s own privacy vs. sharing data

on the Internet was experimentally studied.

Methodologically, the given decision scenarios

and tradeoffs consist of multiple attributes and differ

from each other in the attribute levels. As a result, the

relative importance of attributes deliver information

about which attribute influences the respondents’

choice the most. Part-worth utilities reflect which

attribute level is valued the highest.

3.1 Questionnaire

First, the questionnaire structure is described as are

the instructions which were given to respondents,

followed by the selection and description of the

attributes used for the decision scenarios.

3.1.1 Structure

The questionnaire was arranged in five sections.

The first section addressed demographic characteristics

of the participants. Also, we asked if participants

were familiar with, respectively aware of, the

importance of privacy in the context of Internet

usage. Answers could be given on a four-point scale,

ranging from 1 = “the topic is quite novel to me” to

4 = “I am familiar with the topic.”

In the second part, the experience of Internet

usage was assessed, asking for the frequency of (1)

shopping in the Internet, (2) using social networks

sites, (3) searching for information for leisure

activities, and (4) asking for price comparisons on

the Internet. The frequency could be stated within the

following graduations: several times daily, daily,

several times a week, lesser than that, and never.

The third part introduced the topic of the study,

followed by a detailed description of the single

factors out of which the scenarios were formed. It

was important that the respondents understood the

reason for the study and the decisions that would be

presented later on. Therefore, we gave respondents

the following general introduction:

Internet users have a right to decide what is going

to happen with their data. Principally, the society

as a whole and every single individual can profit

from the data mass generated on the Internet. What

is important is an approach that satisfies the

interests of all parties concerned. Here, privacy

preserving technologies can step in as they

anonymize data and thereby detach them from

one’s person. However, this procedure also

reduces the usability of that data as one cannot, for

example, link a gender to a person anymore. Thus,

a complete anony-mization might not be

reasonable in every case.

The study aim is to find a solution that adheres to

the interests of the data owners (i.e., you as internet

user) and the ones utilizing said data.

In the fourth part of the questionnaire, the single

factors were introduced and explained in detail (see

section 3.1.2.).

The fifth and last part finally presented twelve

scenarios generated from combinations of the single

factors. Data was collected in an online survey

conducted in Germany in 2015. Completing the

questionnaire took about 20 minutes.

3.1.2 Selection of Attributes

The selection of attributes was based on focus group

study outcomes. Taken from the argumentation

patterns, we assume that the preferences concerning

data sharing on the Internet are influenced by

different characteristics that have the highest utility

for users.

For the conjoint study here, we selected the most

important attributes raised by participants:

• the data type

• the benefit of sharing the data

IoTBD 2016 - International Conference on Internet of Things and Big Data

258

• the extent to which anonymization is

guaranteed.

Type of Data

Shopping Preferences that are based

on previous online purchases.

Leisure-Time Interests: In this case,

you would share hobbies and other

interests – such as music, movies,

books, or events – by agreeing that

the contents of websites visited by

you are analyzed.

Location Data: In this case, you

would permanently share your

location when using your

smartphone, cellphone, or computer.

Lifestyle Habits: You would actively

use apps meant to, e.g., record your

level of fitness or driving habits. The

apps would chronicle and analyze

data that is important to you but also

to health insurance funds or car

insurance companies.

Figure 1: “Attribute: type of data.”.

Anonymization

The same query (90-year-old man

from a certain location) suddenly

generates four equal results and

individual persons are no longer

clearly distinguishable. The 90-year-

old is anonymous in a group of four,

or, in other words, the probability of

identifying him lies at 25%.

There is also the possibility to adapt

characteristics like age or gender in

such a way that there are at least 10

persons with the same characteristics.

This puts the probability of

identification at 10%.

In a group of two persons, the

probability of clear identification is

50%.

Without anonymization, the

likelihood of identification is 100%.

Figure 2: Attribute “Extent of anonymization”.

The first attribute that has been varied is the type of

data. Respondents were instructed as follows:

You can decide on the type of data you are willing to

share. This is information you can distribute in

general on the Internet. We differentiate four types of

data (that, to date, are already recorded without the

users’ knowledge:

Types of Benefits

Commendations: You periodically

receive commendations via email

or social networks. For example,

(1) the cheapest offer of a product

you have searched for or looked at

on the internet several times in the

recent past, (2) an insurance that is

tailored to your lifestyle habits and

offers better conditions than your

current insurance package, or (3)

events or leisure time activities in

your proximity.

Discounts: You periodically

receive coupons via email. For

example, for (1) online-shops you

frequent, (2) leisure-time activities

in your proximity (cinema, public

swimming pool, etc.), or (3)

insurance offers that are tailored to

your personal lifestyle habits.

Global benefits: Your shared data

aid the public at large and not only

singular big corporations. For

example, researchers could use

your data for research purposes that

benefit society or medicine, cultural

institutions or clubs could advertize

in a more targeted manner, or

software developers could utilize

the user data to develop free

software.

Figure 3: Attribute “Benefits”.

The second factor was the extent of

anonymization. The following instruction was given

to participants in order to clarify what was meant by

the single anonymization levels:

You can decide on the probability that your data

can be linked to your person. Hypothetically,

general data about people is collected. One could

then assume this data is anonymous if a person’s

name is not recorded as well. How-ever, there are

further characteristics such as age or zip code that

enable an identification without access to one’s

Users’ Willingness to Share Data on the Internet: Perceived Beneﬁts and Caveats

259

name. Anonymization technologies prevent this by

slightly modifying the data. Imagine, e.g., a village

that has two 86-year-olds, one 87-year-old, and a

92-year-old. Instead of recording the precise age,

only an age-group, e.g., aged 85 to 95, would be

stored.

The third factor was the benefit from sharing data.

It was of specific importance that benefits do reflect

different persepctives (from local to global), as these

dimensions were mentioned in the discussions during

the focus groups.

The attribute “benefits” was introduced simply by

the question: As an Internet user, how do I benefit

from this?

3.1.3 Exemplary Decision Scenario

A combination of all corresponding levels would

have led to 48 (4x4x3) possible combinations. As

those decisions are quite taxing on the participants,

we reduced the number of choice tasks to 12 random

tasks. A test of design efficiency confirmed that the

reduced test design was comparable to the

hypothetical orthogonal design.

In each of the choice-based-conjoint decision task

four sets of scenario configurations were presented.

No restrictions were put on the level combinations,

because all chosen attribute levels were combinable.

Overall, participants had to evaluate twelve choice

tasks, each consisting of three different combinations

of the attributes types of data, extent of

anonymization, and the types of benefits. Also, a

“none” option was available in case that none of the

scenarios seemed appropriate to respondents.

Participants were instructed to select the senario they

preferred the most. There was no possibility to skip

tasks. In order to improve comprehensibility, attribute

levels were presented by pictrogramms in addition to

written information.

An exemplary decision scenario is given in Figure 4.

Figure 4: Example decision scenario.

3.2 Sample

Overall, the data of 80 participants was analyzed. Age

ranged between 14 years (youngest participant) and

60 years (oldest participant), with a mean age of 32.9.

The sample was quite gender-balanced, with 50.6 %

women.

Asked about prior knowledge about privacy and

Internet usage, the sample reported to be quite aware

of the topic (M=3.1/out of four points max).

However, a significant gender difference was found:

male respondents (M=3.5) reported to be more aware

of the topic than females (M = 2.9).

With respect to the frequency of using the

Internet, participants reported to use social network

sites on a daily basis (M = (2.2) and seek information

about leisure time activities several times a week (M

= 3.2), whereas online shopping and price

comparisons via Internet are accomplished less often

(M = 3.4). Outcomes are depicted in Figure 5.

Figure 5: Relative importance of attributes.

4 RESULTS

The data analysis (estimation of part-worth utilities)

was done with the Sawtooth Software (SSI Web, HB,

SMRT). In order to identify the main impact factors

on users’ preferences to share their data on the

Internet, we calculated the relative importance of

each attribute. Then, part-worth utilities were

analyzed (on the basis of Hierarchical Bayes) as to

understand which of the three factors is the most

relevant attribute across all decisions made and

relative to all other attributes. When interpreting part-

worth-utilities, it should be noted that these are data

that are scaled to an arbitrary additive constant within

each attribute. Thus, it is impossible to compare

utility values between attributes. Comparisons of

differences between attribute levels are possible if

using zero-centered differentials (part-worth utilities

that are scaled to sum to zero within each attribute).

2.2

3.2

3.4 3.4

-0

soc ial

networks

informing

about

leisure

activities

online

shopping

comparing

price s

Frequency

using

internet

for...

IoTBD 2016 - International Conference on Internet of Things and Big Data

260

4.1 Relative Importance

The relative importance scores of attributes on

preferences to share data in the Internet are depicted

in Figure 6.. As can be seen, the most important

attribute is the probability of being identified

(48.9%), followed by the type of data (30.8%). The

attribute which was evaluated as least important on

users’ preferences was the benefit of sharing the data

reaching a score of 20.3%.

Figure 6: Relative importance of attributes.

The relative importance scores show that users

evaluate the probability of being identified as the

most important attribute whenever they think about

sharing their data. Also, the type of data is relevant

for them. Interestingly, the benefits they could gain

from sharing the data is evaluated as least important,

compared to the other attributes under study. At this

point, it may be noteworthy that the ranking of

importance and even the single scores are quite

homogeneous across participants, neither impacted

by gender nor by the self-reported familiarity with the

topic (“how familiar are you with the topic privacy in

the Internet”).

4.2 Part-worth Utilities

The average zero-centered diff part-worth utilities for

all attribute levels are shown in Figure 7. From this

depiction it becomes obvious that the attribute

“probability of being identified” yielded the highest

range between part-worth utilities. When looking at

absolute utility values, the level “probability of 10%

of being identifiable” reached the highest utility score

(76.7), followed by the 25% probability, which was

still acceptable (reaching a positive value of 23.3). In

contrast, the 50% probability was not acceptable for

respondents (yielding a negative value of -36.2). The

attribute level which received the lowest utility value

was a 100% probability of being identifiable (-63.8).

Figure 7: Part-worth utilities (zero-centered diffs) for all

attributes and levels in the choice-based conjoint study.

When focusing on the type of benefits, also a

mixed picture of acceptance was found. If data

sharing leads to a global benefit for the public at large

(e.g., research efforts, free software development,

medical health or cultural issues), respondents would

be willing to share their data (positive utility score of

27.4).

Negative utility scores were revealed for general

discounts that are periodically offered by email

(-7.5). The lowest utility scores (-19.9) were received

by recommendations that specifically address

personal habits (insurance, lifestyle tailored offers,

leisure time activities in respondents’ proximity).

Figure 8: Part-worth utilities (zero-centered diffs) for all

attributes and levels in the choice-based conjoint study from

a gender perspective.

Looking at the type of data, “shopping habits” and

“leisure time activities” might be shared from the

respondents’ point of view (shopping habit: 29.9;

20%

31%

49%

0% 10% 20% 30% 40% 50% 60%

benefit

sharing

data

type

data

probability

bei ng

identified

Relativ e

importance

Users’ Willingness to Share Data on the Internet: Perceived Beneﬁts and Caveats

261

leisure time data: 23.3). However, sharing of

“location data” and lifestyle habits” receive negative

scores (lifestyle habits: -23.8; location data: -29.5).

Finally, we analyzed gender differences with

respect to part-worth utilities (Figure 8). Single

values can be taken from Table 1.

Table 1: Part-worth utilities in both gender groups.

While the preferences with respect to the main

attribute “probability of being identifiable” is the very

same for women and men, there still were gender

differences regarding the evaluation of the benefits of

sharing data and the type of data users would be

willing to share.

With respect to the benefits, personal

recommenda-tions are more negatively evaluated by

men (-25.3) than women (-17.4). Women evaluate

discounts as more negative (-11.4) compared to men

(-0.3), who are neutral in this regard. When looking

at the type of data, men refuse to share data about life

style habits (-29.1) more strongly than women (-

20.9). Women, in contrast, are much less willing to

share location data (-33.9) compared to men (-21.3).

Sharing data about leisure time activities is more

attractive to women (30.9), while men agree to share

data with regard to leisure activities to lesser extent

(9.2). Another difference between both gender groups

regards shopping data, which male respondents

would be much more willing to share (41.3) than

women (23.8).

5 DISCUSSION AND FUTURE

RESEARCH

This study revealed insights into users’ preferences

for sharing their data on the Internet. In order to

understand behaviors and the tradeoffs between

perceived benefits (that motivate users to share data)

and perceived barriers (that motivate users not to

share the data), we simulated usage scenarios in

which different attribute combinations were

experimentally varied. As attributes we explored

three major factors, which were discussed as the most

important factors in previously conducted focus

group. The first factor was the data type (shopping

preferences, leisure time interests, location data, and

lifestyle habits), referring to the fact that users might

have different privacy protection needs depending on

the respective context. The second factor referred to

the type of benefit which is to be expected from

sharing data on the Internet. There were three

different benefits, ranging from recommendations

(information about interesting events, offers) over

discounts (specifically tailored offers and bargains) to

global benefits (data sharing helps the public at large,

not only single company interests). Finally, the third

factor dealt with the probability of being identified,

which was also experimentally varied.

When looking at the single factors, the findings

show that the anonymization extent is, in fact, the

most relevant attribute for respondents, hinting at a

high awareness of the risk of eavesdropping. The type

of data was ranked as second most important

criterion. Users’ willingness to share data depends on

the respective usage context, conforming

Nissenbaum’s theory of contextual integrity

according to which users control their privacy needs

depending on the respective context (Nissenbaum,

2014). Shopping habits and leisure time activities

might be shared from the respondents’ point of view,

in contrast to lifestyle habits and location data which

are perceived as too personal to be shared.

The lowest overall importance received the type

of benefits. However, there were still differences

between the respective types of benefits. Users’

willingness to share data increases when the data

sharing results in a global benefit for the public at

large (e.g., research efforts, free software develop-

ment, medical health or cultural issues). General

discounts that are periodically offered by email were

negatively evaluated. The same applies for

specifically tailored recommendations (insurance,

lifestyle tailored offers), because they rely on quite

intimate data (personal habits). Thus, the more

specific and personal the benefit is that is offered in

return of sharing the data, the more negative is the

respondents’ attitude and the lower is the willingness

to share data.

In conclusion, on this data basis, a “best case” and

a “worst case” scenario of users’ willingness to share

data can be derived. As taken from the highest utility

ratings for each attribute, the most accepted scenario

Men Women

benefit

sharing

data

global

benefit 25.4 28.5

discounts -0.3 -11.4

personal

recommen dation -25.3 -17.4

type

data

about

lifestyl e

habits -29.1 -20.9

location data -21.3 -33.9

data

about

leisure

interests 9.2 30.9

data

about

shopping

habits 41.3 23.8

probability

being

identified

10% 78.6 75.8

25% 21.2 24.3

50% -34.4 -37.1

100% -65.3 -62.9

IoTBD 2016 - International Conference on Internet of Things and Big Data

262

for sharing their shopping or leisure activities data is

a probability of 10% of being identifiable and a global

benefit of data usage for the public at large (in

contrast to the exclusive benefits for companies). The

worst scenario is a 100% identification probability in

combination with location data or even the

identification of lifestyle habits that are used for an

individually tailored personal recommendation.

Another research focus was directed to gender

differences and the question if women and men have

a similar perspective on the willingness to share their

data on the Internet. The findings of the study showed

both gender-sensitive and gender-insensitive

findings. Men and women do have the same attribute

importance ranking (anonymization is most

important, followed by type of data, and the type of

benefit as least important criterion). They also apply

the same decision criterion for or against sharing the

data. Thus, the above mentioned worst case and best

case scenarios are valid for both gender groups.

However, gender differences showed with respect

to the evaluation of the benefits of sharing data and

the type of data users would be willing to share. In

regard to the type of benefits, women evaluate

personal recommendations as more positive and

discounts as more negative than men. Sharing

location data is a much stronger no-go for women

than for men. Also, women are much more reluctant

to share data regarding their shopping habits, in

contrast to men who are more open in this regard. On

the other hand, when it comes to the question of

sharing leisure time activities on the Internet, women

would share their data much easier than men would.

Even though our empirical research approach

provided valuable insights into conditions of users’

willingness to share their data, still, the outcomes here

provide only a first glimpse into a highly complex

phenomenon. Future studies will have to continue in

this line of research, considering methodological

limitations and broadening the research focus

(Dartmann et al., under revision).

A first point in this context regards the sample size

of this study. The findings should be replicated in

larger and more representative samples.

In this context, integrating a larger portion of

participants of higher and younger age could be

insightful in order to learn if the perceived benefit of

sharing data might be age-sensitive. Also, we only

used three attributes for the conjoint study. Even

though the attributes were empirically identified as

most important for Internet users, of course there are

further factors that should be integrated in future

studies to receive a full picture.

Furthermore, the findings do reflect only

empirical insights from one country - Germany.

Naturally, social values, societal patterns, policy

structures, and economic status of countries do

impact Internet behaviors, and thus also risk

behaviors which impact the willingness to share the

data in the Internet. Therefore, respondents of other

countries and cultures should be integrated in order to

get a more international picture.

In addition, other usage contexts need to be

examined. As such, data sharing in the medical

context (with more sensible and person-related data)

is an emerging field (Wilkowska & Ziefle, 2012) that

needs a closer look as does the mobility and transport

context in which car2x communication in automated

driving (Schmidt et al., 2016) raises other privacy

concerns that yet need to be explored. Overall it will

be interesting to explore such critical situations in

which data sharing on the one and data protection on

the other hand make sense at the same time – however

from the perspective of different interest groups.

ACKNOWLEDGEMENTS

This research has been funded by the Excellence

Initiative of German state and federal governments

(Project ‘NEPTUN, no. OPSF316).

The authors express their special thanks to

Chantal Lydinia and Juliana Brell for excellent

research support as well as two anonymous reviewers

for helpful comments on an earlier version of this

manuscript.

REFERENCES

Akhter, S. H., 2014. Privacy concern and online

transactions: the impact of internet self-efficacy and

internet involvement. Journal of Consumer Marketing,

31, pp. 118–125.

Awad, N. F. and Krishnan, M. S., 2006. The personalization

privacy paradox: an empirical evaluation of

information transparency and the willingness to be

profiled online for personalization. MIS quarterly,

30(1), pp. 13-28.

Barker, V. (2009). Older adolescents' motivations for social

network site use: The influence of gender, group

identity, and collective self-esteem. CyberPsychology

& Behavior, 12(2), 209-213.

Boyd, D. and Hargittai, E., 2010. Facebook privacy

settings: Who cares? First Monday, 15 (8).

Casaló, L. V., Flavián, C. and Guinalíu, M., 2007. The role

of security, privacy, usability and reputation in the

Users’ Willingness to Share Data on the Internet: Perceived Beneﬁts and Caveats

263

development of online banking. Online Information

Review, 31, pp. 583–603.

Dartmann, G.; Demir, M.; Laux, H.; Lücken, V.; Bajcinca,

N.; Kurt, G.; Ziefle, M. & Ascheid, G. (under revision).

Privacy in Cyber-Physical Systems. In: S. Jeschke, C.

Brecher, H. Song, & D. B. Rawat (Eds.). Cyber-

Physical Systems - Foundations, Principles and

Applications. Elsevier Series, Intelligent Data-Centric

Systems".

Debatin, B., Lovejoy, J.P., Horn, A.-K., and Hughes, B.N.,

2009. Facebook and Online Privacy: Attitudes,

Behaviours, and Unintended Consequences. Journal of

Computer-Mediated Communication, 15(1), pp. 83–

108.

Dritsas, S., Gritzalis, D. and Lambrinoudakis, C., 2006.

Protecting privacy and anonymity in pervasive

computing: trends and perspectives. Telematics and

informatics, 23(3), pp. 196-210.

Durndell, A. and Haag, Z., 2002. Computer self efficacy,

computer anxiety, attitudes towards the Internet and

reported experience with the Internet, by gender, in an

East European sample. Computers in human behavior,

18(5), pp. 521-535.

Dwork, C., 2006. Differential privacy. Differential privacy.

In Bugliesi, M. et al. (eds.), Automata, Languages and

Programming, 4052, Lecture Notes in Computer

Science, Springer Berlin Heidelberg, pp. 1-12.

Fogel, J. and Nehmad, E., 2009. Internet social network

communities: Risk taking, trust, and privacy concerns.

Computers in Human Behavior, 25, pp. 153–160.

Freestone, O. and Mitchell, V., 2004. Generation Y

attitudes towards e-ethics and internet-related

misbehaviours. Journal of Business Ethics, 54(2), pp.

121-128.

Gantz, J. and Reinsel, D., 2012. The digital universe in

2020: Big data, bigger digital shadows, and biggest

growth in the far east. IDC iView: IDC Analyze the

Future, pp. 1-16.

Halder, D., & Jaishankar, K., 2011. Cyber gender

harassment and secondary victimization: A

comparative analysis of the United States, the UK, and

India. Victims & Offenders, 6(4), 386-398.

Hargittai, E., 2007. Whose space? Differences among users

and nonusers of social network sites. Journal of

ComputerMediated Communication, 13(1), 276-297.

Hornung, G. and Schnabel, C., 2009. Data protection in

Germany I: The population census decision and the

right to informational self-determination. Computer

Law & Security Review, 25(1), pp. 84-88.

Kalwar, S. K., 2008. Human behavior on the internet.

Potentials, IEEE, 27(5), pp. 31-33.

Karim, N. S. A., Zamzuri, N. H. A. and Nor, Y.M., 2009.

Exploring the relationship between Internet ethics in

university students and the big five model of

personality. Computers & Education, 53(1), pp. 86-93.

Karabey, B., 2012. Big Data and Privacy Issues. In:

Kurbanoğlu, S. et al. (eds.) E-Science and Information

Management

, Springer Berlin Heidelberg, p. 3.

Katal, A., Wazid, M. and Goudar, R. H., 2013. Big data:

Issues, challenges, tools and Good practices. In: Sixth

International Conference on Contemporary Computing

(IC3), IEEE, Noida, India, pp. 404-409.

Kennedy, T., Wellman, B. and Klement, K., 2003.

Gendering the digital divide. It & Society, 1(5), pp. 72-

96.

Kim, D. Y., Lehto, X. Y. and Morrison, A. M., 2007.

Gender differences in online travel information search:

Implications for marketing communications on the

internet. Tourism management, 28(2), pp. 423-433.

Kowalewski, S., Ziefle, M., Ziegeldorf, H., & Wehrle, K.

(2015). Like us on Facebook!–Analyzing User

Preferences Regarding Privacy Settings in Germany.

Procedia Manufacturing, 3, 815-822.

Krasnova, H. and Veltri, N. F., 2010. Privacy Calculus on

Social Networking Sites: Explorative Evidence from

Germany and USA. In: 43rd Hawaii International

Conference on System Sciences (HICSS), Hawaii, USA,

pp. 1–10.

Lahlou, S., 2008. Identity, social status, privacy and face-

keeping in digital society. Social science information,

47(3), pp. 299-330.

Laufer, R. S. and Wolfe, M., 1977. Privacy as a concept and

a social issue: A multidimensional developmental

theory. Journal of social Issues, 33(3), pp. 22-42.

Luce, R. D. & Tukey, J.W., 1964. Simultaneous conjoint

measurement: A new type of fundamental

measurement,” Journal of Mathematical Psychology, 1,

1-27.

Marx, G. T., 1998. Ethics for the new surveillance. The

Information Society, 14(3), pp. 171–185.

Matsusaki, H., 2016. The Data Needs and the Scientific

Methodologies of Marketing Studies: An Analysis

From Ecological Perspectives. In: Proceedings of the

1979 Academy of Marketing Science Annual

Conference, Springer International Publishing, pp. 374-

374.

Mayeda, G., 2016. Privacy in the Age of the Internet:

Lawful Access Provisions and Access to ISP and OSP

Subscriber Information. Alberta Law Review, 53(3), pp.

2015-27.

Nissenbaum, H., 2011. A Contextual Approach to Privacy

Online. Daedalus, 140, pp. 32-48.

Norberg, P. A., Horne, D. R. and Horne, D. A., 2007. The

Privacy Paradox: Personal Information Disclosure

Intentions versus Behaviors. Journal of Consumer

Affairs, 41, pp. 100–126.

Phelps, J. E., D’Souza, G. & Nowak, G. J., 2001.

Antecedents and consequences of consumer privacy

concerns: An empirical investigation. Journal of

Interactive Marketing., 15, pp. 2–17.

Ribble, M. S., Bailey, G. D. and Ross, T.W., 2004. Digital

Citizenship:Addressing Appropriate Technology

Behavior. Learning & Leading with Technology, 32(1),

p. 6.

Rouvroy, A. & Poullet, Y., 2009. The right to informational

self-determination and the value of self-development:

Reassessing the importance of privacy for democracy.

In Gutwirth, S. et al. (eds.) Reinventing data

protection? Springer Netherlands, pp. 45-76.

IoTBD 2016 - International Conference on Internet of Things and Big Data

264

Schmidt, T.; Philipsen, R. & Ziefle, M. 2016. Share to

protect - Quantitative Study on Privacy Issues in V2X-

Technology. Full paper at the 18th International

Conference on Human-Computer Interaction, 2016, in

press.

Spiekermann, S., 2005. Perceived control: Scales for

privacy in ubiquitous computing. In 10th International

conference on user modeling. Available at SSRN:

http://ssrn.com/abstract=761109 or

http://dx.doi.org/10. 2139/ssrn.761109

Sun, Y., Wang, N., Shen, X. L., & Zhang, J. X., 2015.

Location information disclosure in location-based

social network services: Privacy calculus, benefit

structure, and gender differences. Computers in Human

Behavior, 52, pp. 278-292.

Sweeney, L., 2002. K-anonymity: A model for protecting

privacy. International Journal of Uncertainty,

Fuzziness and Knowledge-Based Systems, 10(05), pp.

557-570.

Szongott, C., Henne, B. and von Voigt, G., 2012. Big data

privacy issues in public social media. In: 6th IEEE

International Conference on Digital Ecosystems

Technologies (DEST), IEEE, Campione, Italy, 18-20

June 2012, pp. 1-6.

Takabi, H., Joshi, J. B. and Ahn, G. J., 2010. Security and

privacy challenges in cloud computing environments.

IEEE Security & Privacy, 6, pp. 24-31.

Tene, O. and Polonetsky, J., 2012. Big data for all: Privacy

and user control in the age of analytics. Nw. J. Tech. &

Intell. Prop., 11(5), pp. 239-273.

Thelwall, M., Wilkinson, D., & Uppal, S. (2010). Data

mining emotion in social network communication:

Gender differences in MySpace. Journal of the

American Society for Information Science and

Technology, 61(1), 190-199.

Thelwall, M. (2008). Social networks, gender, and

friending: An analysis of MySpace member profiles.

Journal of the American Society for Information

Science and Technology, 59(8), 1321-1330.

Trstenjak, V., 2016. General report: The influence of

human rights and basic rights in private law. In:

Trstenjak and Weingerl (eds.) The Influence of Human

Rights and Basic Rights in Private Law, Springer

International Publishing, pp. 3-61.

Van Heek, J., Schaar, A. K., Trevisan, B., Bosowski, P. &

Ziefle, M., 2014. User requirements for wearable smart

textiles. Does the usage context matter (medical vs.

sport)? In: S. Boll and F. H. Köhler (Eds.). User-

Centered Design. 4th International Workshop on User-

Centered Design of Pervasive Healthcare Applications

(pp. 205-209). Institute for Computer Science, Social-

Informatics and Telecommunications Engineering

(ICST).

Van Heek, J., Arning, K., and Ziefle, M., 2015. Safety and

privacy perceptions in public spaces: An empirical

study on user requirements for city mobility. In

Giaffreda, R., Caganova, D., Li, Y., Riggio, R., and

Voisard, A. (Eds.). Internet of Things 2014, LNICST

151, Springer Berlin Heidelberg.

Van Heek, J.; Arning, K. & Ziefle, M. (2016). How Fear of

Crime affects Needs for Privacy & Safety. Acceptance

of Surveillance Technologies in Smart Cities. Full

paper at the 5th International Conference on Smart

Cities and Green ICT Systems (Smartgreens 2016), in

press.

Wilkowska, W. and Ziefle, M., 2012. Privacy and data

security in E-health: Requirements from the user’s

perspective. Health Informatics J., 18, 191–201.

Wilkowska, W., Gaul, S., & Ziefle, M. (2010). A Small but

Significant Difference – The Role of Gender on

Acceptance of Medical Assistive Technologies (pp. 82-

100). Springer Berlin Heidelberg.

Young, A. L. and Quan-Haase, A., 2013. Privacy Protection

Strategies on Facebook. Information, Communication

& Society, 16, pp. 479–500.

Ziefle, M., Himmel, S., & Wilkowska, W. 2011. When

your living space knows what you do: Acceptance of

medical home monitoring by different technologies (pp.

607-624). Springer Berlin Heidelberg.

Users’ Willingness to Share Data on the Internet: Perceived Beneﬁts and Caveats

265