Becker 2020 Gold

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Abstract

It’s Not Just About Individuals’ disclosure of personal health information

the Product: How (PHI) can hold substantial benefits for both users and
providers, but users are often reluctant to disclose,
even if they gain benefits such as better
Persuasive personalization. While previous research has dealt
with message framing and information quality in a
Communication health-related context, these factors have been
observed separately. To our best knowledge, we are

Affects the among the first to have examined both factors


(attribute framing and argument strength) and their
interactions concerning PHI disclosure. Thus, we
Disclosure of conducted a web-based experiment with 529
participants to examine the impacts of two persuasive

Personal Health message techniques (attribute framing and argument


strength) on individuals’ PHI disclosure. We reveal that
individuals tend to disclose more PHI when they
Information experience persuasive messages with more positively
framed health wearable (HW) attributes or messages
with higher argument strength based on the reasons
Moritz Becker for the data collection. We enable researchers to
LMU Munich uncover the impacts of persuasive messages in highly
sensitive data environments and provide practitioners
with workable suggestions on how to affect individuals’
Christian Matt
PHI disclosure behaviors.
University of Bern
Keywords: Persuasive Messages; Personal Health
Thomas Hess Information (PHI); Data Disclosure; Attribute Framing;
LMU Munich Argument Strength; Health Wearables (HW).

Introduction
Health wearables (HWs) as a special form of health
information technology (health IT) automatically
gather individuals’ personal health information (PHI) to
provide users with medical feedback for managing
their health and wellbeing on a personal basis. PHI
refers to “personal data related to the physical or
mental health of a natural person, including the
provision of health care services, which reveal
information about his or her health status” (European
Commission of Justice and Consumers, 2016, p. 33).
Individuals perceive PHI as a very sensitive resource,
including serious privacy concerns about how their
PHI is used, disclosed, and protected (Milne, Pettinico,
Hajjat, & Markos, 2017). They are especially
Acknowledgments concerned about possible undesirable economic and
social consequences resulting from the misuse of such
This research project was funded by the German information (Bansal & Gefen, 2010). Thus, 55 percent
Federal Ministry of Education and Research (BMBF) of Europe’s population is willing to disclose their PHI
within the grant 16KIS0494. A previous version of this to physicians, but only 7 percent to HW providers
article has been presented at the Hawaii International (Lutter, Meinecke, Tropf, Böhm, & Esser, 2017).
Conference on System Sciences 2019 as “Becker, M.,
& Matt, C. (2019). The Impact of Persuasive According to privacy calculus theory, individuals are
Messages on the Disclosure of Personal Health willing to voluntarily disclose PHI if they expect that the
Information.” Furthermore, we gratefully acknowledge perceived value from data disclosure will outweigh the
the support of Isabella Burger. perceived costs (Plachkinova, Andrés, & Chatterjee,

The DATA BASE for Advances in Information Systems 37 Volume 51, Number 1, February 2020
2015; Wieneke, Lehrer, Zeder, & Jung, 2016). (logical, no, and illogical arguments) in a 3x3
Previous research suggests that it is not only the experimental setting, we reveal that individuals tend to
product or privacy policies’ characteristics that may disclose more PHI when they experience persuasive
affect individuals’ willingness to disclose PHI but also messages with more positively framed HW attributes
how the information is presented (Adjerid, Acquisti, & and messages with higher argument strength
Loewenstein, 2014; Samat & Acquisti, 2017). Thus, concerning data collection. We enhance the
HW providers could use specific communication theoretical understanding of individuals’ information
strategies, such as persuasive messages, as part of processing of persuasive messages concerning PHI
product presentation to influence individuals’ minds, disclosure and enable practitioners (HW providers and
attitudes, and practices (Stiff & Mongeau, 2003). In health policymakers) to learn how privacy
this sense, HW providers may be able to increase communication strategies can affect individuals’ PHI
individuals’ perceived benefits by adapting the disclosure. Further, we responded to the call for
production description and reframing the HW research by Angst and Agarwal (2009) to identify new
attributes (attribute framing). Framing refers to the predictors of persuasion in the context of health IT by
phenomenon of persuasion in which “simple and investigating individuals’ PHI disclosure.
unspectacular changes” in product presentation lead
to changes in choice” (Kühberger, 1998, p. 205). Conceptual Foundations
Research into healthcare has shown that message
content is more likely to positively influence behavior if Privacy Calculus and Personal Health Information
the message frame leads to positive thoughts and Health wearables studies have been conducted from
associations (Angst & Agarwal, 2009). Thus, attributes different perspectives to investigate the nature of
of HWs that are more positively framed (gain-framed individuals’ information disclosure behaviors. They
vs. neutral vs. loss-oriented) may increase individuals’ have adopted different theories, such as theory of
PHI disclosure. reasoned action, privacy calculus theory, and social
Apart from the product presentation, HW providers contract theory (Zhang et al., 2018). Among them,
may also think about adapting the presentation of their privacy calculus theory is widely adopted to investigate
privacy policies so as to lower the perceived risks. consumer privacy concerns because it integrates both
Privacy policy statements can be influenced via the risks and benefits of privacy concerns. The privacy
different arguments for data collection. It is known that calculus model assumes that individuals perform a
individuals produce more favorable responses to risk-benefit analysis before they can decide whether to
messages with strong than with weak arguments disclose their personal information (Dinev & Hart,
(Petty & Cacioppo, 1986). Following this, HW 2006). Following this, persuasive messages’ impacts
providers that directly present reasons for data on PHI disclosure depends on individuals’ cost-benefit
collection and thus use higher argument strength analyses. Thus, a calculus view of information privacy
levels (logical arguments vs. no arguments vs. illogical suggests that individuals measure the tradeoff
arguments) may be able to increase individuals’ PHI between the usage and the potential negative
disclosure. Both attribute framing and argument outcomes of sharing PHI. Individuals are willing to
strength have barely been evaluated (Chong, Ge, Li, voluntarily disclose PHI if they expect that the
& Proctor, 2018; Kühberger, 1998; Samat & Acquisti, perceived value of data disclosure will outweigh the
2017; Shen & Kollar, 2015). In the few existing cases, perceived risks (Wieneke et al., 2016). Privacy
they were mostly examined in combination with calculus assumes a primarily economic view and
personality traits such as self-esteem, perceived relates to other theories in the field of economics, such
prestige, and information sensitivity. Also, they have as utility maximization theory, expectancy theory of
mostly been investigated in non-health-related motivation, and expectancy value theory (Pavlou,
contexts, while the multidimensional effect of two 2011). Privacy calculus postulates that individuals’
persuasive message techniques on PHI disclosure behavioral intentions and follow-up actions are
has not yet been examined (Li, Wu, Gao, & Shi, 2016). positively affected by the potential gain of disclosure
We ask: and are negatively affected by their expected loss from
a potential privacy violation.
RQ: How do attribute framing and argument strength
influence individuals’ PHI disclosure? However, since not all information is considered
equally sensitive by individuals, the information type
To answer this research question, we conducted a they are asked to disclose affects their perceptions,
web-based experiment with 529 participants to processing, and behaviors (Matt et al., 2019; Milne et
analyze the influences of these two persuasive al., 2017; Yang & Wang, 2009). Individuals are more
messages techniques on individuals’ PHI disclosure. willing to provide certain information types about
By varying attribute framing (gain-oriented, neutral, themselves compared to other information they
and loss-oriented frames) and argument strength perceive as more sensitive. They evaluate information

The DATA BASE for Advances in Information Systems 38 Volume 51, Number 1, February 2020
about themselves and consider the information to Persuasive Communication in the Health Context
have different sensitivity levels, which affects
According to the definition of Stiff and Mongeau
individuals’ risk perceptions (Angst & Agarwal, 2009).
(2003), any message that seeks to reshape, force, or
Further, individuals are more willing to provide
change individuals’ responses can be referred to as
demographic and lifestyle information to marketers
persuasive communication. Through conformity
compared to health information or personal identifiers
effects such as compliance or identification, online
such as name, address, and social security number.
persuasion has become increasingly important owing
Requests for more sensitive information reduce trust
to the digital age and social media. However, there has
beliefs and intentions to disclose, increasing risk
been little research into online persuasion, which has
perceptions (Malhotra, Kim, & Agarwal, 2004; Roback
mainly been used for product advertising and health
& Wakefield, 2013).
communication messages (Stiff & Mongeau, 2003;
Individuals generally want control over their Zhang et al., 2018). Most studies have focused on
information and awareness of the information types information disclosure regarding e-commerce
collected about them. It is easier for them to trust transactions, use of online services and social network
organizations if they understand which information sites, and technology acceptance (e.g. Baruh, Secinti,
type is gathered, the way in which it is collected, and & Cemalcilar, 2017). Focusing on studies that
that the user can manage the way information is specifically analyzed decision-making and persuasive
treated in privacy policies (Becker, 2018). However, communication in the health context, we identified
owing to PHI’s very high sensitivity, even elaborate three main research streams (Table 1).
privacy policies did not fully mitigate individuals’
The first research stream analyzes sharing behaviors
privacy risk perceptions (Adjerid, Acquisti, &
concerning personal health record data with clinicians
Loewenstein, 2019; Samat & Acquisti, 2017). It is
and public health entities (e.g. Weitzman, Kelemen,
argued that when requested information is very
Kaci, & Mandl, 2012). The second stream examines
sensitive, the underlying context becomes more
individual factors that influence the adoption of health
important. Anderson and Agarwal (2011) compared
technologies, treatments, and disclosures. The third
the sensitivity and the disclosure of PHI and the
research stream measures the influence of perceived
different PHI types (general health, mental health, and
risks and benefits on behavioral intentions and framing
genetic information) and found that the requested PHI
concepts so as to enhance benefits concerning health
type had no significant impacts on the privacy
decisions (Bartels, Kelly, & Rothman, 2010; Edwards,
perceptions, the trust in the technology used for a
Elwyn, Covey, Matthews, & Pill, 2001).
transaction, and individuals’ willingness to disclose
PHI. Thus, all PHI types are perceived as sensitive for According these research streams, we developed our
individuals, and they do not carefully distinguish conceptual framework of how persuasive
between different PHI subtypes when deciding about communication affects PHI disclosure (Figure 1). We
their disclosure (Anderson & Agarwal, 2011; Milne et used the two established techniques—attribute
al., 2017). framing and argument strength—as persuasive
messages (Samat & Acquisti, 2017; Zhao, Strasser,
Cappella, Lerman, & Fishbein, 2011). This means that
message content can be presented in different frames
(attribute framing) and different levels of usefulness of
the information (argument strength).
Table 1. Exemplary Studies on Persuasive Communication and PHI Disclosure

Research
Focal object References Research topic
stream
e.g. Angst & Agarwal, 2009; Jena, Sharing personal health record data
Sharing
1 2015; Weitzman et al., 2012; Zhang with clinicians and public health
medical data
et al., 2018 entities
e.g. Alrige et al., 2014; Anderson &
Health-IT Agarwal, 2011; Krishnamurthy et al., Adapting technologies, treatments,
2
adoption 2001; Li et al., 2016; Schöning, Matt, and disclosures
& Hess, 2019
e.g. Bartels et al., 2010; Edwards et Framing effects on individuals’
3 Framing effects al., 2001; Elbert & Ots, 2018; Samat decisions (e.g. health behaviors,
& Acquisti, 2017 disclosure intentions)

The DATA BASE for Advances in Information Systems 39 Volume 51, Number 1, February 2020
Figure 1. Theoretical Framework
Framing refers to the phenomenon whereby “simple leads to positive thoughts and associations. Thus,
and unspectacular changes” in product presentation message frames that lead to predominantly negative
lead to changes in choices (Kühberger, 1998, p. 205). thoughts may not anticipate substantial changes in
This research shows that the argumentation in individuals’ attitudes or behaviors.
privacy-relevant information interacts with the effect of
Argument strength: An argument’s message includes
changes in the objective privacy risk as a result of
information, which has two components: information
disclosure and individuals’ propensity to disclose
quality and information quantity. Information quality
personal information. It can be assumed that attribute
can be defined as “[…] the usefulness of the available
framing would increase individuals’ perceived benefits
attribute information in aiding a decision maker to
and that argument strength would lower the perceived
evaluate his/her true utility associated with an
risks (Samat & Acquisti, 2017). We will now describe
alternative,” while information quantity can be
both techniques in some detail.
described “[…] as the number of items or attributes
Attribute framing: Concentrating on attitude changes describing an alternative” (Keller & Staelin, 1987, p.
caused by persuasive communication, a well-known 200). When holding quantity fixed, an increase in
influence on individual choices, is message framing. quality leads to greater confidence in individuals’
Framing has two logically analogous ways to present decision-making. Likewise, when holding quality fixed,
identical information. It refers to either highlighting an increase in quantity negatively affects individuals’
benefits and conforming to the message advocacy or confidence (Krishnamurthy, Carter, & Blair, 2001). The
to accentuating costs when failing to comply (Shen & combination of the two is called argument strength; it
Kollar, 2015). Two message framing types commonly strongly influences arguments’ persuasiveness,
used in marketing and in health decision-making are affecting individuals’ attitudes and behaviors.
goal framing and attribute framing. Attribute frames Argument strength relates to the information’s
can be implemented by presenting either desirable vs. usefulness. Strong arguments are perceived as more
undesirable attributes or the presence vs. absence of convincing than weak or no arguments (Cialdini, Petty,
(un)desirable attributes (Block & Keller, 1995). Also & Cacioppo, 1981).
some information system research studies have
investigated framing effects on individuals’ behaviors Research Model and Hypothesis
(Becker & Matt, 2019; Samat & Acquisti, 2017). We Development
concentrate on attribute framing, since HWs’ features
can be better described in a positive or negative style HW providers can increase individuals’ perceived
than in a goal accomplishment setting. Attribute benefits by adapting the online production description
frames can be based on value, award a financial and reframing the HW attributes (attribute framing). On
advantage, or generally deal with gains and losses. the other hand, they can decrease the perceived risks
Possible framing elements could be urgency, by using different arguments for data collection
persistence, simplicity, or visuals and metaphors that (argument strength). We assume that using these two
must be assessed and arranged depending on fit and persuasive message techniques can influence
effectiveness (Bartels et al., 2010). According to Angst individuals’ risk-benefit-tradeoff perceptions and
and Agarwal (2009), message content is more likely to therefore affect their PHI disclosure (Figure 2).
positively influence behaviors if the message frame

Figure 2. Research Model

The DATA BASE for Advances in Information Systems 40 Volume 51, Number 1, February 2020
Attribute Framing strength scores. Nonetheless, a disputable view on
argument strength can be discovered in the
Positive frames have a superior effect on attitude
photocopier experiment of Langer, Blank, and
change when they promote a product or its benefits
Chanowitz (1978). In the experiment, participants are
(Krishnamurthy et al., 2001). Research shows that a
asked to imagine the following scenario: “You are in a
mere framing of choices (in terms of gains and losses)
long queue, waiting to make photocopies. You are next
significantly affects individuals’ decisions. Labeling
and then someone comes up to you and says, ‘May I
ground beef as 75 percent lean instead of 25 percent
use the machine, because I have to make copies?’
fat significantly affected participants’ perceptions of
How likely would you be to let them cut into the line?”
beef quality (Levin & Gaeth, 1988). Privacy research
Besides this illogical argument, which Langer et al.
has also shown that presenting information with a
(1978) called placebic information (i), the alternative
positive or negative framing can also change
scenarios were that the person either gave no
individuals’ perceptions and awareness in disclosure
arguments (ii), asking simply “May I use the
decisions (Acquisti, Adjerid, & Brandimarte, 2013;
photocopier?” or (iii) provided a logical argument: “May
Acquisti, Brandimarte, & Loewenstein, 2015; Gluck et
I use the photocopier, because I’m in a rush?” Since
al., 2016). Framing ease individuals’ privacy concerns.
the respective success rates were: (i) 93 percent, (ii)
Framing notices with more positive, misleading, or
60 percent, and (iii) 94 percent, Langer et al. (1978)
misdirecting statements can direct individuals’
concluded that individuals exhibit mindless behavior
attention away from the implications of privacy aspects
and pay no attention to the substantive elements that
to product aspects, lowering their privacy awareness
are relevant for successful resolution. Langer et al.
(Gluck et al., 2016). In other words, positively framed
(1978) referred to this as mindless behavior.
product descriptions may decrease negative thoughts
Individuals act mindlessly and reenact prior scripts,
about possible risk concerning PHI disclosure.
sequences of events in a familiar situation using
Message content is more likely to positively affect
heuristic cues, or shortcuts as long as an argument is
behaviors if the message frame causes positive
offered (Salovaara, Lyytinen, & Penttinen, 2019). This
thoughts and associations (Angst & Agarwal, 2009).
illustrates the impacts of persuasive messages by
Message frames that lead to mainly negative thoughts
assessing the effect of information quality on
may not anticipate considerable changes in individuals’
behaviors. Despite the fact that, for logical arguments,
behaviors. When attribute framing is applied, the
individuals perceive higher argument strength levels
product depicts the object of the frame and arguments
than for illogical or no arguments, the study uncovered
for the product usage are the object’s attributes, which
another interesting aspect. It revealed that inquiries
impact on the decision (Krishnamurthy et al., 2001).
led to a positive outcome significantly more often if
We hypothesize the following:
some reasoning was included in the inquiry, while the
Hypothesis 1 (H1): Individuals who experience argument’s quality was unimportant. Regarding the
persuasive messages with more positively framed use of PHI, HW providers tend to convey the data
HW attributes tend to disclose more PHI than handling in their privacy statements and thus do not
those who experience neutral or loss-oriented directly offer information about data use to individuals.
frames. However, directly presenting reasons for data
collection may influence individuals’ behaviors. Thus,
Argument Strength providing stronger arguments for data collection to
individuals may lower the perceived risks, which would
Argument strength is directed at individuals’ rational positively influence individuals’ willingness to disclose
judgment rather than the effect of reinforcing or more PHI. We hypothesize the following:
improving their beliefs (Bhattacherjee & Sanford,
2006). They engage in thorough cognitive activity, Hypothesis 2 (H2): Individuals who experience
assessing the presented information and thinking persuasive messages with higher argument
about it. It is known that individuals produce more strength levels tend to disclose more PHI.
favorable responses to messages with strong than
with weak arguments (Petty & Cacioppo, 1986). Methodology
Individuals who are influenced by strong arguments
are likely to hold a strong, accessible attitude to the Data Collection
information. Thus, when a message can be carefully We performed an expert panel and a pilot test to
scrutinized, individuals will likely have more thoughts validate the messages and instruments prior to the
or arguments (Lukin, Anand, Walker, & Whittaker, final data collection. Based on this, we conducted
2017). Following this, logical arguments provide minor changes to the instructions and to certain item
higher levels of argument strength than explanations wordings. We conducted the final web-based
with no argument. In this understanding, explanations experiment in February 2018 using Qualtrics. We
with illogical arguments receive the lowest argument included time stamps to record the time a participant

The DATA BASE for Advances in Information Systems 41 Volume 51, Number 1, February 2020
spent on every page. First, we excluded unrealistic keeping them in an intuitive and natural environment.
responses from the analyses. Second, to reduce After the product presentation which resembled an
variance of individuals’ attitudes concerning former official product website, the participants received the
experiences, we excluded de facto and former users questionnaire and had to answer attention check
of this specific HW (Fitbit Charge 2). Thus, we questions. The participants received three test
collected 605 responses from non-users of Fitbit questions relating to the product (Fitbit Charge 2), the
Charge 2. After screening the responses, based on the features (heartbeat and activity tracking), and the
timestamps and deleting the incomplete responses, provider (Fitbit). Participants who failed to answer one
we ended up with 529 participants. The ages ranged of these attention check questions were rejected from
from 18 to 78, with a mean of 29.1 years (SD = 10.28); continuing the experiment. All other participants were
54.6 percent were male and 45.4 percent were female. then randomly split into nine treatment groups (Table
Two-thirds of the participants (68.1 percent) had at 3).
least a two-year college degree or higher; 65.6 percent
The persuasive message treatments consisted of
were employees or were self-employed, and 31.4
loss-oriented, neutral, or gain-oriented framed HW
percent were students (Table 2).
attribute descriptions, and the privacy policy
statements included either logical, illogical, or no
Experimental Design
arguments for the data collection. We adapted the
Since HWs are very diverse in terms of functioning and treatments from Raj, Charles, and Alisha (2006) and
appearance, we needed to create a similar product modified them to fit the context (see Table A.1 in the
and provider vision by concentrating on a specific Appendix). After two manipulation checks, we
subgroup of the examined technology. We used the measured individuals’ PHI disclosure adapting the
Fitbit Charge 2 from the provider Fitbit, one of the most scale of MacKenzie and Spreng (1992) to our context.
frequently sold HWs (Lee, Kim, & Welk, 2014). We Thus, the participants had to specify on a seven-point
used the original Fitbit Charge 2 website and replaced semantic scale (1 = unlikely/likely, 2 = not
its HW attributes and arguments for data collection probable/probable, 3 = impossible/possible, and 4 =
with our treatments. unwilling/willing) the extent to which they would reveal
the PHI to the provider Fitbit. We then recorded
Thus, we ensured that participants received sufficient
demographics such as gender, age, employment, and
knowledge about the provider and HW product, while
education.

Table 2. Sample Description


Demographics Frequency Percentage
Gender
Male 289 54.6%
Female 240 45.4%
Age
18 to 22 149 28.2%
23 to 27 142 26.8%
28 to 31 81 15.3%
32 to 35 75 14.2%
36 to 78 82 15.5%
Education level
None 16 3.0%
High/Secondary school 136 25.7%
Bachelor 170 32.2%
Master’s/Diploma/Magister 182 34.4%
Ph.D. 25 4.7%
Employment status
Employed 271 51.2%
Self-employed 76 14.4%
Student 166 31.4%
Unemployed 16 3.0%

The DATA BASE for Advances in Information Systems 42 Volume 51, Number 1, February 2020
Table 3. Treatment Overview

Attribute framing

Loss-oriented framing Neutral framing Gain-oriented framing


(LOF) (NF) (GOF)

Logical arguments Group 1 Group 4 Group 7


(LAG) (n = 66) (n = 57) (n= 68)
Argument
strength

Illogical arguments Group 2 Group 5 Group 8


(IAG) (n = 46) (n = 75) (n = 60)
No arguments Group 3 Group 6 Group 9
(NAG) (n = 57) (n = 54) (n = 46)

Manipulation Checks and Group Assignment conducted a Kolmogorov-Smirnov test on the


collected samples.
To verify that our framing manipulation was successful
(i.e. showed a notable difference in the perception of The p-value (0.42) assumed a normal distribution for
the framing level), we used the manipulation check the dependent variable PHI disclosure. We also used
questions of Petty, Cacioppo, and Schumann (1983). the Levenes test and obtained an F-value of 1.951 and
The manipulation checks for attribute framing, using a p-value of 0.51, failing to reject the equal variance
ANOVA (F = 31,221, p < 0.001; MAverage = 2.74) and dispersion hypothesis.
the Bonferroni test, showed that the three descriptions
of the HW attributes were perceived differently, Results
depending on the framing levels. Gain-oriented
framing (MGOF = 3.59; p < 0.001) revealed significantly First, we used a two-way ANOVA to identify the main
higher positive HW attitudes than neutral framing (MNF effects and interaction effects. Table 4 shows the
= 2.60; p < 0.001) and loss-oriented framing (MLOF = analysis results. From the F-statistic, we found that
2.02; p < 0.001). Thus, the effectiveness of the attribute framing’s main effect reached a significant
implementation of the three framing levels was level (F = 11.833, p < 0.001). Concerning the framed
supported. HW attributes (H1), individuals distinguished between
all three framing levels.
We used the nine manipulation check questions of
Zhao et al. (2011) for argument strength. The ANOVA Second, to closely examine H1 and H2, we conducted
(F = 43,951, p < 0.001; MAverage = 3.58) and the the mean comparisons among multiple groups using
Bonferroni test showed that the argument strengths the Bonferroni test, which is considered suitable for
were perceived differently for the three factor levels. such analyses (Kirk, 2012). As Table 5 shows, and as
hypothesized in H1, a more positive framing level
Logical arguments (MLAG = 4.38; p < 0.001) revealed a increased individuals’ PHI disclosure. The contrasts
significantly higher perceived strength than the for loss-oriented vs. neutral (p < 0.05), neutral vs. gain-
descriptions with illogical (MIAG = 3,43; p < 0.001) and oriented (p < 0.01), and gain-oriented vs. loss-oriented
no arguments (MNAG = 2.79; p < 0.01). Thus, the (p < 0.001) were significant. Loss-oriented framing
effectiveness of the implementation of the three (MSLOF = 3.27, SDLOF = 0.14) had a smaller mean
argument levels was supported. value than neutral framing (MSNF = 3.79, SDNF = 0.13)
The different groups can be considered independent, and gain-oriented framing (MSGOF = 4.26, SDGOF =
since the subjects are randomly assigned. We 0.14). Thus, H1 was supported.

Table 4. Results of the Two-way ANOVA

Treatment (hypothesis) F Sign.


Attribute framing (H1) 11.833 < 0.001

Argument strength (H2) 39.287 < 0.001

Attribute framing x argument strength 1.039 0.386

The DATA BASE for Advances in Information Systems 43 Volume 51, Number 1, February 2020
Table 5. Bonferroni Group Comparisons (H1 and H2)

Group A Group B Mean difference (A - B) Sign.


Group comparisons of attribute framing on PHI disclosure (H1)
Loss-oriented mean: 3.27 Neutral mean: 3.79
-0.52 < 0.05
SD: 0.14 SD: 0.13
Neutral mean: 3.79 Gain-oriented mean: 4.26
-0.47 < 0.01
SD: 0.13 SD: 0.14
Gain-oriented mean: 4.26 Loss-oriented mean: 3.27
0.99 < 0.001
SD: 0.14 SD: 0.14
Group comparisons of argument strength on PHI disclosure (H2)
Illogical arguments mean: 3.92 No arguments mean: 2.88
1.04 < 0.001
SD: 0.14 SD: 0.14
No argument mean: 2.88 Logical arguments mean: 4.58
-1.70 < 0.001
SD: 0.14 SD: 0.13
Logical arguments mean: 4.58 Illogical arguments mean: 3.92
0.66 < 0.01
SD: 0.13 SD: 0.14

The main effect on PHI disclosure that arose from supported. Table 5 shows that persuasive messages
different arguments was significant (F = 39.287, p < containing arguments for data collection (even illogical
0.001) (see Table 4). Concerning the arguments for (MSIAG = 3.92; SDIAG 0.14)) elicited higher PHI
data collection (H2), individuals distinguished between disclosure than persuasive messages without
all three levels. The contrasts for illogical vs. no arguments (MSNAG 2,79; SDNAG 0.14). The interaction
arguments (p < 0.001), no arguments vs. logical effect between attribute framing and argument
arguments (p < 0.001), and logical vs. illogical strength was insignificant (F = 1.039, p = 0.386).
arguments (p < 0.01) were significant (see Table 5). Figure 3 illustrates the results, which we will now
discuss in some detail.
Persuasive messages with higher argument strength
tended to lead to higher PHI disclosure. H2 was

Figure 3. Impacts of Persuasive Messages on PHI Disclosure

The DATA BASE for Advances in Information Systems 44 Volume 51, Number 1, February 2020
Discussion Argument Strength
Motivated by the question whether persuasive Our results revealed that individuals who experienced
messages can affect individuals’ willingness to persuasive messages with higher argument strength
disclose PHI, we conducted a web-based experiment tended to disclose more PHI. Further, persuasive
with 529 participants to examine the influences of the messages containing logical arguments for data
persuasive message techniques, attribute framing, collection elicited higher PHI disclosure than
and argument strength on individuals’ PHI disclosure. persuasive messages with illogical or no arguments.
We tested three effects on PHI disclosure: (H1) the In sum, strong arguments are perceived as more
effect of framing HW attributes, (H2) the influence of convincing than weak or no arguments (Cialdini et al.,
argument strength concerning data collection, and the 1981). Yet, interestingly (also illustrated in Figure 3),
interplay between the two factors. By varying attribute persuasive messages that contained illogical
framing (loss-oriented, neutral, gain-oriented) and arguments for data collection received higher PHI
argument strength (logical, illogical, no arguments) in disclosure than persuasive messages without
a 3x3 factorial design, we found support for H1 and arguments. However, this is in line with other studies
H2. of cognitive processing in IS research (e.g. Curry,
Marshall, & Kawalek, 2014) and psychological
Attribute Framing research studies (e.g. Langer et al., 1978), where it is
called placebic information. In terms of theory of
Our results showed that individuals who experience controlled mindless behavior (Langer, 1992), it can be
persuasive messages with more positively framed HW argued that implementing illogical information
attributes tend to disclose more PHI. Valence-based concerning data collection is more effective than giving
associative processing is probably a valid explanation no information, since individuals don’t read privacy
of how attribute framing affected the content of policies carefully. A recent study showed that 25
individuals’ PHI disclosure behaviors, and thus why percent never read or immediately agreed to privacy
gain-oriented framing led to more positive evaluations policy statements (DIVSI, 2015). According to Langer
of the HW attributes than negative HW attribute (1992), arguments can either be processed in a
framing. Prior research usually investigated evaluation controlled mindless way or in an automatic mindful
effects of labeling a key attribute in positive vs. way. Mindlessness occurs when an individual pays no
negative terms without questioning differences in attention to new, relevant information (Salovaara et al.,
susceptibility to framing effects (e.g. Elbert & Ots, 2019). In our case, if the provided reason appears to
2018). By including a neutral frame, we were able to be irrelevant, the arguments for data collection are not
examine the magnitude of framing effects and found examined and evaluated by individuals. Thus, we can
that participants who received positive framing were state that the context dependency is ignored and that
more affected by attribute framing than those who the argument concerning PHI disclosure is processed
received negative framing. This indicates that when mindlessly. This means that individuals will
individuals experienced gain-framed HW attributes, it automatically and mindlessly process the arguments
probably outweighed the perceived risks of data and will ignore the context dependency.
collection; thus, they were more inclined to take risks
and disclose more PHI. This finding is also consistent Attribute Framing x Argument Strength
with the position that more effortful and less heuristic
processing may reduce susceptibility to cognitive Finally, as our two-way ANOVA analysis showed no
biases (Kuvaas & Selart, 2004). Since our scenario significant effect, we found no interaction between
contained more complex and abstract information than attribute framing and argument strengths. This result
is usually the case in attribute framing research, it can stands in contrast to studies in other contexts (e.g.
be assumed that in our context, individuals used more Bartels et al., 2010). We can state that, in the case of
effortful processing. They may have been better able PHI disclosure, adapting only one construct implies
to counterbalance the framing information with other that perceived benefits are enhanced or perceived
and more relevant information than those using less risks are reduced, but there is no interaction effect
effortful processing. Thus, for complex decisions such between the two factors that influence individuals’ risk-
as outweighing the PHI disclosure concerning the benefit tradeoff perceptions (Wieneke et al., 2016).
perceived risks and benefits, a facilitating effect of However, concerning the effects of attribute framing
more detailed processing should be more likely than and argument strength alone, it would be interesting to
an inhibiting or biasing effect (Petty & Cacioppo, evaluate which information processes are triggered by
1986). these two factors. According to the elaboration
likelihood model (ELM), it can be assumed that the
framed HW attributes will be processed via the central
route (more conscious and thoughtful), since the

The DATA BASE for Advances in Information Systems 45 Volume 51, Number 1, February 2020
information context is complex for the individuals. In Implications for Practice
contrast, privacy policy statements and the presented
From a practical perspective, we have shown
arguments for data collection will probably be
persuasive messages’ impacts on individuals’ PHI
processed via the peripheral (less conscious and
disclosure. We revealed that HW providers could
thoughtful) route, since our results and other research
gather more PHI by just using effective combinations
(DIVSI, 2015) indicated that privacy policies are often
of attribute framing and argument strength (e.g. gain-
ignored or not read with care.
framed HW attributes and logical arguments for data
collection) in their product descriptions. Increasing
Implications and Limitations data disclosure via persuasive messaging
Implications for Theory substantially increases a company’s market share and
sales without affecting costs (Bhaskar & Vo, 2012). By
From a theoretical perspective, we responded to the implicitly decreasing the perceived risks, a broader
call by Angst and Agarwal (2009) to identify new customer base could be established, especially
predictors of persuasion in the context on health IT. among individuals above the age of 40, who have
We have added new insights about the drivers and higher concerns about technology and smart devices
issues of PHI disclosure. While previous research has (Becker, Matt, Widjaja, & Hess, 2017). As these
dealt with message framing and information quality in aspects help HW providers to overcome individuals’
a health-related context, these factors have been privacy concerns on disclosing PHI without actually
observed separately (Block & Keller, 1995; Shen & improving the product quality, we strongly recommend
Kollar, 2015). To our best knowledge, we are among judicious judgment in its application. Further, our
the first to have examined both factors (attribute results may help health-related organizations with
framing and argument strength) and their interactions their data collection efforts without changing or
concerning PHI disclosure. We have added value to adjusting the product and service offering, while
health IT research, since it can be assumed that the keeping user gratification and loyalty high. Our results
explanation lies in the insecurity regarding the advocate how to carefully frame information regarding
technology and HW providers’ non-transparent data HW benefits and privacy policies, because a
processing. For this reason, researchers should use persuasive message strongly affects individuals’
information processing models, especially ELM, to attitudes and usage intentions. This is interesting,
clarify the influence of route distinction for persuasive since political regulation conditions, such as the EU
messages in such highly sensitive data environments. General Data Protection Regulation (GDPR), are
When we looked at the results of our experiment, we being implemented in order to increase transparency
found PHI disclosure tendencies to be consistent with about data storage, disclosure, and usage (European
the norms of reciprocity. Thus, overlearned social Commission of Justice and Consumers, 2016; Vo &
scripts can be activated not only in a context in which Bhaskar, 2012). While such reforms are forcing
they do not make sense, but also in a context in which providers to increase their openness, they should plan
the trigger for the script makes explicit the non-human their communication strategy wisely so as to reduce
source of the information. As Langer (1992) noted, predictors of negative attitudes and behaviors.
mindlessness is distinct from mere overlearning Otherwise, published critical information about data
because the former may result from a single exposure processing leads to more negative attitudes toward
to a stimulus, as opposed to repeated exposures. In technology usage and lower usage intentions.
these cases, information is accepted uncritically,
without attention to other aspects of the situation. In Limitations and Future Research
ELM research, this is called credibility. When placebic
information is stated in neuroscience jargon (e.g. by Owing to PHI’s high sensitivity, we assumed that PHI
relating it to the activation of brain areas), circular disclosure has high personal relevance, implying that
explanations are mistakenly taken for good ones. participant elaboration was presumably high.
Scientific literacy seems insufficient to prevent this However, presenting rigid response options may also
bias type for longer and scientifically stated create mind blockages, locking respondents’ attention
explanations: while neuroscience experts correctly to preconceived answers. However, we clearly pointed
identify the information about brain activation as out our study’s purpose and guaranteed full anonymity
irrelevant and non-explanatory, individuals barely so as to mitigate such effects. Further, we used a web-
acquainted with neuroscience are as affected as naive based experiment, since it allows for high control while
ones (Weisberg, Keil, Goodstein, Rawson, & Gray, providing participants with appealing decision
2008). environment in which rich information on HWs’
purpose and capabilities could be delivered. However,
previous studies have criticized these aspects as
limitations of the web-based experiment methodology

The DATA BASE for Advances in Information Systems 46 Volume 51, Number 1, February 2020
(Hergueux & Jacquemet, 2015). Although web-based Anderson, C. L., & Agarwal, R. (2011). The digitization
experiments may have weaker experimental controls of healthcare: Boundary risks, emotion, and
compared to laboratory-based experiments and they consumer willingness to disclose personal health
have greater associated difficulties to ensure reliability information. Information Systems Research,
and internal validity, we used a web-based experiment 22(3), 469-490.
so as to provide a more realistic and natural setting for Angst, C. M., & Agarwal, R. (2009). Adoption of
the participants’ decision processes. As part of this, we electronic health records in the presence of
concentrated on a specific HW device to reduce privacy concerns: The elaboration likelihood
variance in individuals’ PHI disclosure concerning model and individual persuasion. MIS Quarterly,
different product visions and experiences. However, 33(2), 339-370.
future research should evaluate active HW users, as Bansal, G., & Gefen, D. (2010). The impact of personal
well as other devices, and may distinguish between dispositions on information sensitivity, privacy
individuals who use a fitness tracker and the concern and trust in disclosing health information
corresponding mobile application and individuals who online. Decision Support Systems, 49(2), 138-
refuse to use the mobile app but who own a fitness 150.
tracker. Also, other constructs that influence the Bartels, R. D., Kelly, K. M., & Rothman, A. J. (2010).
perceived benefit-risk tradeoffs should be tested and Moving beyond the function of the health
added to the model so as to obtain additional insights. behaviour: The effect of message frame on
This is relevant, since research into online persuasion behavioural decision-making. Psychology &
has revealed that personal factors (e.g. the need for Health, 25(7), 821-838.
cognition, self-esteem, or general privacy concerns) Baruh, L., Secinti, E., & Cemalcilar, Z. (2017). Online
and situational factors (e.g. trust or transparency) privacy concerns and privacy management: A
affect individuals’ privacy concerns and disclosure meta-analytical review. Journal of
decisions (e.g. Malhotra et al., 2004; Plachkinova et Communication, 67(1), 26-53.
al., 2015; Samat & Acquisti, 2017). In line with this, it Becker, M. (2018). Understanding users’ health
would also be important to elaborate whether results information privacy concerns for health wearables.
differ subject to individuals’ health status or subject to Proceedings of the 51st Hawaii International
their expectations after engaging with a product and a Conference on System Sciences (HICSS),
brand. Hawaii, USA.
Becker, M., & Matt, C. (2019). The impact of
References persuasive messages on the disclosure of
personal health information. Proceedings of the
Acquisti, A., Adjerid, I., & Brandimarte, L. (2013). Gone 52nd Hawaii International Conference on System
in 15 seconds: The limits of privacy transparency Sciences (HICSS), Hawaii, USA.
and control. IEEE Security & Privacy, 11(4), 72- Becker, M., Matt, C., Widjaja, T., & Hess, T. (2017).
74. Understanding privacy risk perceptions of
Acquisti, A., Brandimarte, L., & Loewenstein, G. consumer health wearables–an empirical
(2015). Privacy and human behavior in the age of taxonomy. Proceedings of the 38th International
information. Science, 347(6221), 509-514. Conference on Information Systems (ICIS), Seoul,
Adjerid, I., Acquisti, A., & Loewenstein, G. (2014). South Korea.
Framing and the malleability of privacy choices. Bhaskar, R., & Vo, A. (2012). Health care reform
Proceedings of the 13th Workshop on the requires it solutions to influence consumer
Economics of Information Security (WEIS), perception at a health care payer. Journal of
Pennsylvania, USA. Cases on Information Technology 14(2), 18-26.
Adjerid, I., Acquisti, A., & Loewenstein, G. (2019). Bhattacherjee, A., & Sanford, C. (2006). Influence
Choice architecture, framing, and cascaded processes for information technology acceptance:
privacy choices. Management Science, 65(5), An elaboration likelihood model. MIS Quarterly,
1949-2443. 30(4), 805-825.
Alrige, M., Alsudais, A., Plachkinova, M., Chatterjee, Block, L. G., & Keller, P. A. (1995). When to
S., Edwards, A., Rodriguez, J., & Weinstein, A. accentuate the negative: The effects of perceived
(2014). Ehr adoption in healthcare practices: efficacy and message framing on intentions to
Lessons from two case studies. Proceedings of perform a health-related behavior. Journal of
the 20th Americas Conference on Information Marketing Research, 32(2), 192-203.
Systems (AMCIS), Savannah, USA. Chong, I., Ge, H., Li, N., & Proctor, R. W. (2018).
Influence of privacy priming and security framing
on mobile app selection. Computers & Security,
78, 143-154.

The DATA BASE for Advances in Information Systems 47 Volume 51, Number 1, February 2020
Cialdini, R. B., Petty, R. E., & Cacioppo, J. T. (1981). Krishnamurthy, P., Carter, P., & Blair, E. (2001).
Attitude and attitude change. Annual Review of Attribute framing and goal framing effects in health
Psychology, 32(1), 357-404. decisions. Organizational Behavior and Human
Curry, M., Marshall, B., & Kawalek, P. (2014). It artifact Decision Processes, 85(2), 382-399.
bias: How exogenous predilections influence Kühberger, A. (1998). The influence of framing on
organizational information system paradigms. risky decisions: A meta-analysis. Organizational
International Journal of Information Management, Behavior and Human Decision Processes, 75(1),
34(4), 427-436. 23-55.
Dinev, T., & Hart, P. (2006). An extended privacy Kuvaas, B., & Selart, M. (2004). Effects of attribute
calculus model for e-commerce transactions. framing on cognitive processing and evaluation.
Information Systems Research, 17(1), 61-80. Organizational Behavior and Human Decision
DIVSI. (2015, 06.02.2018). Wie gehen Sie persönlich Processes, 95(2), 198-207.
üblicherweise mit den AGB bzw. Langer, E. J. (1992). Matters of mind: Mindfulness /
Datenschutzbestimmungen von Angeboten im mindlessness in perspective. Consciousness and
Internet um? Retrieved from Cognition, 1(3), 289-305.
https://de.statista.com/statistik/daten/studie Langer, E. J., Blank, A., & Chanowitz, B. (1978). The
/477294/umfrage/umfrage-zum-umgang-mit- mindlessness of ostensibly thoughtful action: The
datenschutzbestimmungen-und-agb-im-internet role of “placebic” information in interpersonal
Edwards, A., Elwyn, G., Covey, J., Matthews, E., & interaction. Journal of Personality and Social
Pill, R. (2001). Presenting risk information a Psychology, 36(6), 635-642.
review of the effects of framing and other Lee, J.-M., Kim, Y., & Welk, G. J. (2014). Validity of
manipulations on patient outcomes. Journal of consumer-based physical activity monitors.
Health Communication, 6(1), 61-82. Medicine & Science in Sports & Exercise, 46(9),
Elbert, S. P., & Ots, P. (2018). Reading or listening to 1840-1848.
a gain-or loss-framed health message: Effects of Levin, I. P., & Gaeth, G. J. (1988). How consumers are
message framing and communication mode in the affected by the framing of attribute information
context of fruit and vegetable intake. Journal of before and after consuming the product. Journal
Health Communication, 23(6), 573-580. of Consumer Research, 15(3), 374-378.
European Commission of Justice and Consumers. Li, H., Wu, J., Gao, Y., & Shi, Y. (2016). Examining
(2016). Regulation (EU) 2016/679 of the individuals’ adoption of healthcare wearable
European Parliament and of the Council of 27 April devices: An empirical study from privacy calculus
2016 on the protection of natural persons with perspective. International Journal of Medical
regard to the processing of personal data and on Informatics, 88(1), 8-17.
the free movement of such data, and repealing Lukin, S. M., Anand, P., Walker, M., & Whittaker, S.
Directive 95/46/EC (General Data Protection (2017). Argument strength is in the eye of the
Regulation) oj 2016 l 119/1. beholder: Audience effects in persuasion. arXiv
Gluck, J., Schaub, F., Friedman, A., Habib, H., Sadeh, preprint arXiv:1308.1164, 1-12.
N., Cranor, L. F., & Agarwal, Y. (2016). How short Lutter, T., Meinecke, C., Tropf, T., Böhm, K., & Esser,
is too short? Implications of length and framing on R. (2017). Zukunft der Consumer Technology –
the effectiveness of privacy notices. 12th 2017. Berlin: Bitkom e.V.
Symposium on Usable Privacy and Security MacKenzie, S. B., & Spreng, R. A. (1992). How does
(SOUPS). motivation moderate the impact of central and
Hergueux, J., & Jacquemet, N. (2015). Social peripheral processing on brand attitudes and
preferences in the online laboratory: A intentions? Journal of Consumer Research, 18(4),
randomized experiment. Experimental 519-529.
Economics, 18(2), 251-283. Malhotra, N. K., Kim, S. S., & Agarwal, J. (2004).
Jena, R. (2015). Sharing personal health information: Internet users’ information privacy concerns
Personalization versus privacy. Paper presented (iuipc): The construct, the scale, and a causal
at the Proceedings of the 21st Americas model. Information Systems Research, 15(4), 336
Conference on Information Systems (AMCIS), - 355.
Puerto Rico. Matt, C.; Becker, M.; Kolbeck, A.; Hess, T. (2019).
Keller, K. L., & Staelin, R. (1987). Effects of quality and Continuously healthy, continuously used? – A
quantity of information on decision effectiveness. thematic analysis of user perceptions on
Journal of Consumer Research, 14(2), 200-213. consumer health wearables. Pacific Asia Journal
Kirk, R. E. (2012). Experimental design: Procedures of the Association for Information Systems, 11(1),
for the behavioral sciences (4th ed.). Thousand Article 5.
Oaks, CA: Sage Publications Inc.

The DATA BASE for Advances in Information Systems 48 Volume 51, Number 1, February 2020
Milne, G. R., Pettinico, G., Hajjat, F. M., & Markos, E. Weisberg, D. S., Keil, F. C., Goodstein, J., Rawson,
(2017). Information sensitivity typology: Mapping E., & Gray, J. R. (2008). The seductive allure of
the degree and type of risk consumers perceive in neuroscience explanations. Journal of Cognitive
personal data sharing. Journal of Consumer Neuroscience, 20(3), 470-477.
Affairs, 51(1), 133-161. Weitzman, E., Kelemen, S., Kaci, L., & Mandl, K.
Pavlou, P. A. (2011). State of the information privacy (2012). Willingness to share personal health
literature: Where are we now and where should we record data for care improvement and public
go? MIS Quarterly, 35(4), 977-988. health: A survey of experienced personal health
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration record users. BMC Medical Informatics and
likelihood model of persuasion. In B. Leonard Decision Making, 12(39), 1-10.
(Ed.), Advances in Experimental Social Wieneke, A., Lehrer, C., Zeder, R., & Jung, R. (2016).
Psychology (Vol. 19, pp. 123-205). New York, NY: Privacy-related decision-making in the context of
Academic Press. wearable use. Proceedings of the 20th Pacific
Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Asia Conference on Information Systems
Central and peripheral routes to advertising (PACIS), Chiayi, Taiwan.
effectiveness: The moderating role of Yang, S., & Wang, K. (2009). The influence of
involvement. Journal of Consumer Research, information sensitivity compensation on privacy
10(2), 135-146. concern and behavioral intention. ACM SIGMIS
Plachkinova, M., Andrés, S., & Chatterjee, S. (2015). Database: The DATABASE for Advances in
A taxonomy of mhealth apps--security and privacy Information Systems, 40(1), 38-51.
concerns. Proceedings of the 48th Hawaii Zhang, X., Liu, S., Chen, X., Wang, L., Gao, B., & Zhu,
International Conference onSystem Sciences Q. (2018). Health information privacy concerns,
(HICSS). antecedents, and information disclosure intention
Raj, A., Charles, S., & Alisha, A. (2006). Using framing in online health communities. Information &
and credibility to incorporate exercise and fitness Management, 55(4), 482-493.
in individuals’ lifestyle. Journal of Consumer Zhao, X., Strasser, A., Cappella, J. N., Lerman, C., &
Marketing, 23(4), 199-207. Fishbein, M. (2011). A measure of perceived
Roback, D., & Wakefield, R. L. (2013). Privacy risk argument strength: Reliability and validity.
versus socialness in the decision to use mobile Communication Methods and Measures, 5(1), 48-
location-based applications. ACM SIGMIS 75.
Database: The DATABASE for Advances in
Information Systems, 44(2), 19-38. About the Authors
Salovaara, A., Lyytinen, K., & Penttinen, E. (2019).
High reliability in digital organizing: Mindlessness, Moritz Becker holds a Ph.D. in Management from
the frame problem, and digital operations. MIS LMU Munich, Germany. His Doctoral thesis addressed
Quarterly, 43(2), 555-578. users’ perceptions, processing, and disclosure of
Samat, S., & Acquisti, A. (2017). Format vs. Content: healthcare data. Before joining the Ph.D. program, he
The impact of risk and presentation on disclosure received a master’s degree from the University of
decisions. Proceedings of the 13th Symposium on Mannheim. His work has appeared in the Pacific Asia
Usable Privacy and Security (SOUPS), Santa Journal of the Association for Information Systems and
Clara, USA. in the proceedings of the International Conference on
Schöning, C., Matt, C., & Hess, T. (2019). Information Systems and the European Conference
Personalised nudging for more data disclosure? on Information Systems, among others.
On the adaption of data usage policies format to Christian Matt is an Assistant Professor of
cognitive styles. Proceedings of the 52nd Hawaii Information Systems at the University of Bern,
International Conference on System Sciences Switzerland. He holds a Ph.D. in Management from
(HICSS), Maui, USA. LMU Munich and was a visiting scholar at the National
Shen, L., & Kollar, L. M. M. (2015). Testing moderators University of Singapore and the Wharton School of the
of message framing effect. Communication University of Pennsylvania. His current research
Research, 42(5), 626-648. focuses on digital transformation and value creation,
Stiff, J. B., & Mongeau, P. A. (2003). Persuasive as well as on digital customer interfaces and
Communication. New York, USA: Guilford Press. experiences. His work has appeared in Journal of
Vo, A., & Bhaskar, R. (2012). Health care reform Management Information Systems, Electronic Markets,
requires rethinking on the IT strategy. Journal of MIS Quarterly Executive, Business and Information
Cases on Information Technology 14(2), 65-72. Systems Engineering, and Internet Research, among
others.

The DATA BASE for Advances in Information Systems 49 Volume 51, Number 1, February 2020
Thomas Hess is a Professor of Management digital management systems, digital transformation
Information Systems at LMU Munich, Germany, where processes and the management of IT, Internet, and
he also serves as director of the Institute for media companies. His work has appeared in journals
Information Systems and New Media and as such as Journal of Management Information Systems,
coordinator of the Center for Internet Research and Communications of the ACM, Decision Support
Media Integration. He holds a Ph.D. in management Systems, Information Systems Journal, International
from the University of St. Gallen, Switzerland. His Journal of Electronic Commerce, and others.
research focuses on digital business models and

Appendix A
Table A.1. Screenshots of the Experimental Treatments (Attribute Framing / Argument Strength)

3x3 Experimental treatments (persuasive messages)

Attribute framing Argument strength

Logical Arguments (LAG)


Gain- oriented (GOF)

No Arguments (NOG)
Neutral-oriented (NF)

Illogical Arguments (IAG)


Loss-oriented (LOF)

The DATA BASE for Advances in Information Systems 50 Volume 51, Number 1, February 2020

You might also like