Bright Patterns as an

Ethical Approach to
Counteract Dark
Patterns

A Closer Investigation of The Ethics of Persuasive

Design

Main Subject area: Informatics

Specialisation in: User Experience Design
Author(s): Hellen Truong & Axel Dalbard
JÖNKÖPING 2022-06-30
Certificate of Completion
This final thesis has been carried out at the School of Engineering at Jönköping
University within Informatics. The authors are responsible for the presented opinions,
conclusions and results.

Examiner: Bruce Ferwerda

Supervisor: Marko Tkalcic
Scope: 15 hp
Date: 2022-06-30

i
Attestation of Authorship
We hereby declare that this submission is our work, best upon research that we have
conducted.
To the best of our knowledge and belief, it contains no material published or written by
another person – except where explicitly defined in the Acknowledgements or listed in
the References and properly cited.
Nor does it contain any material of mine/ours that, to a substantial extent, has been
submitted for the award of any other degree or diploma of a university or other
institution of higher learning.
Hellen Truong
Axel Dalbard

Acknowledgements
We want to begin our thesis by expressing our gratitude towards the people who
contributed to this body of work, to the people that enabled this paper to flourish into
the complete version that you are currently reading.
First of all, we want to thank the people who participated in the experiment of this study.
We want to thank you for your time and patience, and we want to let you know that
your efforts are greatly appreciated.
Secondly, we want to thank the UX designers who participated in the interviews for
this study. We also want to thank you for your time and patience, and we want to let
you know that we appreciate your efforts immensely.
Last but certainly not least, we want to thank our supervisor Marko. Thank you, Marko,
for sharing your wisdom and knowledge to point us in the right direction during this
thesis. Your help has meant a lot to us, and we think that you have been a great
supervisor.

ii
Abstract
The purpose of this study is to explore the ethical dilemma in design that User
Experience (UX) designers encounter in their workplace, consumers’ perception of
ethics in bright patterns and dark patterns, and consumers’ decisions between bright
patterns and dark patterns. The former aims to understand the reason behind the
prevalence of dark patterns, while the latter aims to determine whether bright patterns
are a potential ethical approach that designers can adopt in the future. In this study, the
methods semi-structured interview and within-subjects experiment with follow-up
interview were conducted to gather empirical data. For both methods, a content analysis
was selected to analyze the empirical data, which resulted in findings that answered the
research questions of this study. The findings show that (1) authority to decide how
designers should address ethics in design is more distributed to those investing in a
product development project, (2) most ethical issues revolve around challenges of
working in an ethical manner rather than bad practices of incorporating ethics in design
work, (3) designers adopt one or more ethical approaches to stay ethical and /or address
ethical issues that arise in their workplace, (4) some specific bright patterns and dark
patterns have no influence on consumer decisions while other specific bright patterns
and dark patterns influence consumer decisions, and (5) consumers perceive bright
patterns to be more ethical than dark patterns in terms of freedom of choice and
transparency.

Keywords: Bright patterns, Dark patterns, Design ethics, Freedom of choice,

Persuasive design, Transparency.

iii
Table of Content
Certificate of Completion .............................................................. i
Attestation of Authorship ............................................................. ii
Acknowledgements........................................................................ ii
Abstract ......................................................................................... iii
Table of Content ........................................................................... iv
List of Figures ............................................................................... vii
List of Tables ................................................................................. xi
1 Introduction ........................................................................... 12
1.1 BACKGROUND ................................................................................................ 12

1.2 RELATED WORKS ........................................................................................... 13

1.2.1 Dark Patterns .......................................................................................... 13

1.2.2 Dark Patterns Raise Ethical Concerns ................................................... 15
1.2.3 Bright Patterns Address Ethical Issues .................................................. 16
1.2.4 Knowledge Gap ..................................................................................... 16
1.3 PURPOSE AND RESEARCH QUESTIONS ............................................................ 17

1.3.1 Expected Outcomes ............................................................................... 18

1.4 SCOPE AND DELIMITATIONS ........................................................................... 18

1.4.1 Replacing Dark Patterns With Bright Patterns ...................................... 18

1.4.2 Grey Zone .............................................................................................. 18
1.4.3 Perspective of Stakeholders ................................................................... 19
1.5 DISPOSITION ................................................................................................... 19

2 Method and Implementation ................................................ 20

2.1 METHOD CHOICES .......................................................................................... 20

2.1.1 Participant Recruitment: Screener Survey ............................................. 20

2.1.2 Semi-structured Interview ...................................................................... 21
2.1.3 Experiment: Within-Subjects Study Design .......................................... 21

iv
 2.1.4 Content Analysis .................................................................................... 22
2.2 ARTIFACT CREATION ..................................................................................... 24

2.2.1 Design Choices of the Artifact ............................................................... 26

2.3 DATA COLLECTION AND PROCEDURE............................................................. 38

2.3.1 Interview With UX Designers ............................................................... 38

2.3.2 Experiment With Follow-up Interview .................................................. 38
2.4 VALIDITY AND RELIABILITY .......................................................................... 39

2.4.1 Semi-structured Interview: Validity and Reliability .............................. 39

2.4.2 Experiment: Validity and Reliability ..................................................... 41

3 Theoretical Framework ........................................................ 43

3.1 TYPES OF DARK PATTERNS ............................................................................ 43

3.2 ETHICS ........................................................................................................... 47

3.2.1 Definition of Ethics ................................................................................ 47

3.2.2 Bright Patterns ....................................................................................... 47
3.3 THEORIES ....................................................................................................... 48

3.3.1 Dual Process Theory .............................................................................. 48

3.3.2 Cognitive Biases .................................................................................... 49

4 Results and Analysis............................................................... 51

4.1 RESULTS OF INTERVIEWS WITH UX DESIGNERS ............................................ 51

4.1.1 Ethical Issues ......................................................................................... 53

4.1.2 Countermeasure for Ethical Issues......................................................... 54
4.1.3 Ethical Approach ................................................................................... 55
4.1.4 Authority Level ...................................................................................... 55
4.2 RESULTS OF EXPERIMENT WITH FOLLOW-UP INTERVIEW .............................. 56

4.2.1 Consumer Decisions .............................................................................. 56

4.2.2 Perceived Freedom of Choice ................................................................ 62
4.2.3 Perceived Transparency ......................................................................... 65
4.2.4 Effect of Dark Patterns........................................................................... 69

v
 4.2.5 Effect of Bright Patterns ........................................................................ 70
4.2.6 Other Categories .................................................................................... 72
4.3 SUMMARY OF FINDINGS ................................................................................. 73

4.3.1 Findings of the Interviews with UX designers ...................................... 73

4.3.2 Findings of the Experiment With Follow-up Interview ......................... 73

5 Discussion ............................................................................... 75
5.1 RESULTS DISCUSSION..................................................................................... 75

5.1.1 Authority Distribution in the Workplace of Designers .......................... 75

5.1.2 Ethical Issues in the Workplace of Designers ........................................ 76
5.1.3 Ethical Approaches in the Workplace of Designers .............................. 77
5.1.4 Bright Patterns as an Ethical Approach ................................................. 78
5.1.5 Influence on Consumer Decisions ......................................................... 78
5.1.6 Persuade With Transparency ................................................................. 79
5.1.7 Freedom of Choice ................................................................................. 79
5.1.8 Consumers’ Distrust and Dislike of Dark Patterns ................................ 79
5.1.9 Consumers and the Dual Process Theory .............................................. 79
5.1.10 Dark Patterns Affecting Cognitive Biases ............................................. 80
5.2 METHOD DISCUSSION..................................................................................... 80

5.2.1 Ecological Validity Implications ........................................................... 80

5.2.2 Sample Size Implications ....................................................................... 81
5.2.3 Latent Content Analysis Implications .................................................... 82

6 Conclusions and Further Research ...................................... 84

6.1 CONCLUSIONS ................................................................................................ 84

6.1.1 Practical Implications............................................................................. 85

6.1.2 Scientific Implications ........................................................................... 85
6.2 FURTHER RESEARCH ...................................................................................... 86

7 References ............................................................................. 87
8 Appendices ............................................................................. 91

vi
 8.1 APPENDIX 1 .................................................................................................... 92

8.2 APPENDIX 2 .................................................................................................... 93

List of Figures
Figure 1: Women's home page for both prototypes. 25
Figure 2: Price Comparison Prevention Dark Pattern in Prototype A. The consumer can
choose between the two highlighted t-shirts, but the price is hard to compare. The t-
shirt in the bottom row is discounted with 5% off, but it does not show the initial price.
27
Figure 3: False Urgency Scarcity Dark Pattern in Prototype A. The consumer can
choose between the two highlighted options, but one of them has a false urgency
scarcity message on it to put pressure on the consumer. 28
Figure 4: Sneak into Basket Dark Pattern in Prototype A. When the consumer goes to
his/her shopping bag, an extra item has been added. The user can choose to remove the
item, but it is still annoying and causes friction. 29
Figure 5: Trick Questions Dark Pattern in Prototype A. When the consumer chooses
payment options, they are met with this newsletter pop-up. Two options are presented
that look very similar at first, enhancing confusion. 30
Figure 6: Roach Motel Dark Pattern in Prototype A, Part 1. When a consumer wants to
delete their account, they need to go through several links to do so. First, they must go
to FAQ. 31
Figure 7: Roach Motel Dark Pattern in Prototype A, Part 2. When the consumer has
navigated to the FAQ, they need to find the highlighted "Delete data" link to get one
step closer to terminating their account. 31
Figure 8: Misdirection Dark Pattern in Prototype A. When the consumer wants to delete
their account, the color and size of the "never mind" option has been highlighted to try
to sway the consumer to not delete their account. 32
Figure 9: Price Comparison Prevention Bright Pattern in Prototype B. The 3-pack is
now a bit cheaper and lists price per t-shirt. The other option has a clearer discount and
shows the initial price before the discount. 33
Figure 10: False Urgency Scarcity Bright Pattern in Prototype B. In this prototype, there
are no scarcity messages shown to the consumer. 34
Figure 11: Sneak into Basket Bright Pattern in Prototype B. Instead of sneaking in a
product to the consumer's shopping bag, consumers are met with this pop-up. This

vii
functions as a recommendation to the consumer, and they can choose to add the item or
decline. 35
Figure 12: Trick Questions Bright Pattern in Prototype B. In this newsletter pop-up,
consumers need to either check/uncheck one box to continue, or they can close it down
with the cross in the upper right corner. 36
Figure 13: Roach Motel Bright Pattern in Prototype B. In this prototype, the consumer
does not have to navigate through an unnecessary number of links to terminate their
account. 37
Figure 14: Misdirection Bright Pattern in Prototype B. In this protype, the consumer
does not get misdirected by the use of color and size. The big action button deletes the
account. 37
Figure 15: Note. An illustration of the “Trick questions” dark pattern. From Brignull,
H. (2017, September 19). Trick questions. Dark Patterns.
https://www.darkpatterns.org/types-of-dark-pattern/trick-questions 43
Figure 16: Note. An illustration of the “Sneak into Basket” dark pattern. From
Wintermeier, N. (2020b, June 22). Dark Patterns Examples in eCommerce: What they
are & why to avoid them. Crobox. https://blog.crobox.com/article/dark-patterns 44
Figure 17: Note. An illustration of the “Roach Motel” dark pattern. From Wintermeier,
N. (2020b, June 22). Dark Patterns Examples in eCommerce: What they are & why to
avoid them. Crobox. https://blog.crobox.com/article/dark-patterns 45
Figure 18: Note. An illustration of the “Price Comparison Prevention” dark pattern.
From Brignull, H. (2018, June 6). Price comparison prevention. Dark Patterns.
https://www.darkpatterns.org/types-of-dark-pattern/price-comparison-prevention 45
Figure 19: Note. An illustration of the “Misdirection” dark pattern. From Dunbar, J.
[@JessicaDunbar]. (2019, June 2). Guess I didn’t cancel my @ipsy membership last
month. I hope the $10 was worth it. I was planning to take a break for the summer, now
I’ll never subscribe to ipsy again. #Ux #design #unethical #ifellforit #ipsy cc/
@darkpatterns [Tweet]. Twitter.
https://twitter.com/JessicaDunbar/status/1135219427056738309?fbclid=IwAR23gPlQ
1LW2wW0e78AkEHeRcxUKk8Q8w5v-yh2mm8_mGD5aAW9N37fdMqM 46
Figure 20: Note. Fake Urgency Scarcity dark pattern. From Wintermeier, N. (2020b,
June 22). Dark Patterns Examples in eCommerce: What they are & why to avoid them.
Crobox. https://blog.crobox.com/article/dark-patterns 47
Figure 21: Note. A visual representation of the DPT. From Gorbatenko, E. (2020, March
8). Trading: fast and slow. Online Forex Broker.
https://fbs.eu/en/analytics/articles/trading-fast-and-slow-26041 49

viii
Figure 22: Consumer decisions for Task 1 where participants are instructed to select
three white t-shirts with the lowest price among other t-shirts and add them to their
shopping bag. The blue bars represent the number of participants that chose a) a 3-Pack
of t-shirts or b) three individual t-shirts in Prototype B (bright patterns). The orange
bars represent the number of participants that chose a) a 3-Pack of t-shirts or b) three
individual t-shirts in Prototype A (dark patterns). 57
Figure 23: Consumer decisions for Task 2 (Men’s) where participants are instructed to
select a pair of jeans in the colour medium blue and add it to their shopping bag. The
blue bars represent the number of participants that chose a) slim fit jeans or b) ripped
jeans in Prototype B (bright patterns). The orange bars represent the number of
participants that chose a) slim fit jeans or b) ripped jeans in Prototype A (dark patterns).
58
Figure 24: Consumer decisions for Task 2 (Women’s) where participants are instructed
to select a pair of jeans in the colour medium blue and add it to their shopping bag. The
blue bars represent the number of participants that chose a) straight jeans or b) mom
jeans in Prototype B (bright patterns). The orange bars represent the number of
participants that chose a) straight jeans or b) mom jeans in Prototype A (dark patterns).
58
Figure 25: Consumer decisions for Task 3 where participants are instructed to proceed
to checkout to pay for the items in their shopping bag. In Prototype A, a product has
been sneaked into their shopping bag. The orange bars represent the number of
participants that a) kept the sneaked-in product or b) removed the sneaked-in product
from their shopping bag. 59
Figure 26: Consumer decisions for Task 3 where participants are instructed to proceed
to checkout to pay for the items in their shopping bag. In Prototype B, a pop-up
recommends a clothing item to participants when they are in their shopping bag. The
blue bars represent the number of participants that a) added the recommended item to
their shopping bag or b) declined the recommended item. 60
Figure 27: Consumer decisions for Task 3 where participants are instructed to proceed
to checkout to pay for the items in their shopping bag. During the checkout process,
participants are met with a newsletter pop-up. The blue bars represent the number of
participants that a) subscribed to the newsletter or b) declined in Prototype B (bright
patterns). The orange bars represent the number of participants that a) subscribed to the
newsletter or b) declined in Prototype A (dark patterns). 60
Figure 28: Consumer decisions for Task 4 where participants are instructed to delete
their account. The blue bars represent the number of participants that a) easily found
the “Delete account” page or b) instantly clicked the “Delete” button in Prototype B
(bright patterns). The orange bars represent the number of participants that a) easily

ix
found the “Delete account” page, b) were stuck or struggled to find the “Delete account”
page, c) instantly clicked the “Delete” button, or d) hesitated or misclicked the “Delete”
button in Prototype A (dark patterns). 61
Figure 29: Participants’ preferred prototype between Prototype A and Prototype B. The
blue bar represents the number of participants that preferred Prototype B (bright
patterns), which was the preferred prototype for all participants. 62
Figure 30: Consumers’ perceived freedom of choice in Prototype A related to specific
dark patterns. The blue bar represents the number of participants reporting freedom
regarding Trick Questions. The orange bars represent the number of participants
reporting no freedom or limited freedom regarding a) Sneak into Basket and b) Trick
Questions. 63
Figure 31: Consumers’ perceived freedom of choice in Prototype B related to specific
bright patterns. The blue bars represent participants reporting freedom regarding a)
Recommended item and b) Newsletter. The orange bars represent the number of
participants reporting no freedom or limited freedom regarding a) Recommended item
and b) Newsletter. 63
Figure 32: Participants’ report of which prototype they found to provide them with the
highest freedom of choice between Prototype A and Prototype B. The blue bar
represents the number of participants that reported Prototype B (bright patterns) to
provide the highest freedom of choice, which was all participants. 65
Figure 33: Consumers' perceived transparency in Prototype A related to specific dark
patterns. The blue bars represent participants reporting high transparency regarding a)
Trick Questions, b) Price Comparison Prevention, and c) Roach Motel. The orange bars
represent the number of participants reporting low transparency regarding a) Sneak into
Basket, b) Trick Questions, c) Price Comparison Prevention, d) Roach Motel, and e)
Misdirection. 66
Figure 34: Consumers’ perceived transparency in Prototype B related to specific bright
patterns. The blue bars represent the number of participants reporting high transparency
regarding a) Recommended item, b) Newsletter, c) Price Comparison, d) “Roach
Motel”, and e) “Misdirection”. The orange bar represents the number of participants
reporting low transparency regarding Newsletter. 66
Figure 35: Participants’ report of which prototype they found to have the highest
transparency between Prototype A and Prototype B. The blue bar represents the number
of participants that reported Prototype B (bright patterns) to provide the highest
transparency, which was all participants. 69
Figure 36: The reported effects that certain dark patterns in Prototype A had on
participants. The blue bar represents the number of participants feeling a negative

x
emotion regarding Sneak into Basket. The orange bar represents the number of
participants feeling distracted regarding False Urgency Scarcity. The grey bar
represents the number of participants feeling misled regarding Misdirection. The yellow
bar represents the number of participants feeling confused regarding Trick Questions.
69
Figure 37: The reported effects that “Recommended item” in Prototype B had on
participants. In this case, participants only reported the effect that the bright pattern
Recommended item had on them. The blue bar represents the number of participants
feeling a negative emotion. The orange bar represents the number of participants feeling
a positive emotion. 71

List of Tables
Table 1: A coding scheme presenting the responses from the semi-structured interviews
coded into four categories and then re-coded into subcategories to quantify the
occurrences of similar responses within a subcategory. .............................................. 51

xi
1 Introduction
This chapter introduces the research area of this study by discussing previous work that
has been carried out in the field of research. The knowledge gap that this study intends
to fill is identified, thereby motivating this study’s purpose and research questions. This
chapter also includes the scope and delimitations of this study, as well as a disposition
of the remainder of this study.

1.1 Background
We currently live in a world that has become vastly digitized. Shopping, news,
entertainment and many other things ingrained into our daily lives are examples of
industries that have taken a shift towards the digital route. Websites and mobile
applications are constantly fighting for every second of our attention, and the designers
of these services have had to come up with ways to win our attention, one of them being
persuasive design.
Persuasive design takes the human perspective of understanding and optimizing the
user experience of a service (Wintermeier, 2020a). This usually requires an
understanding of consumer and behavioral psychology. Persuasive design is for the
most part ethical, but it can also turn unethical, which might lead to dark patterns.
Dark patterns are carefully designed user interfaces that result in manipulating the
consumer, also known as a user into carrying out something they do not want to do,
find difficult to do, or are not aware they are doing (Wintermeier, 2020b). Persuasive
design can easily turn into dark patterns as soon as it strips away the consumer’s ability
to choose by either manipulation or limitation of the consumer’s choices in a way that
is beneficial for the service (Wintermeier, 2020a). This can be attained by confusion,
deception, and manipulation that can take many different shapes in the design of a user
interface.
Businesses have practiced dark patterns for a long time, but consumers around the world
have recently become more and more aware of this unethical design practice online.
Consumers are aware that dark patterns are designs that manipulate them into making
decisions that benefit the business at the cost of themselves (Bongard-Blanchy et al.,
2021; Mathur et al., 2019; Narayanan et al., 2020). It is believed that public awareness
of dark patterns will shield consumers from being manipulated by dark patterns, but
this does not make any difference. Studies have found that consumers are unwilling to
oppose dark patterns despite being aware of them (Bongard-Blanchy et al., 2021; Maier
et al., 2020; Westin et al., 2019).
If consumers have no intention to shield themselves from dark patterns, then businesses
will not have any reason to stop practicing dark patterns to gain profits like money,
attention, and data (Narayanan et al., 2020). It might not be the right approach to let

12
consumers oppose dark patterns but rather the designers who should take the
responsibility to be more ethical in their design. Designers are perhaps pressured by
their stakeholders to practice dark patterns to meet specific metrics or goals (Kight &
Gram-Hansen, 2019). Kight and Gram-Hansen (2019) argue that research should focus
on the rights of designers, and not only focus on the rights of consumers. Other
academics stress the importance to address ethical issues in persuasive design and have
therefore recommended ethical approaches that designers can use to oppose dark
patterns (Kight & Gram-Hansen, 2019; Narayanan et al., 2020; McLean, 2020; Mildner
& Savino, 2020; Benner et al., 2021; Karagoel & Nathan-Roberts, 2021). One ethical
approach that has been researched is bright patterns, which is the counterpart of dark
patterns (Graßl et al., 2021).
Bright patterns have shown to be a promising ethical approach to address ethical issues,
but this subject has only been researched in the context of privacy cookie consent, thus
further investigation about this subject will be addressed in this study. This is to ensure
that bright patterns influence consumers to make decisions that benefit themselves, and
not only the company, in other contexts than privacy cookie consent. If designers are
not careful with how they use persuasive techniques to persuade consumers, then bright
patterns could turn into dark patterns (Wintermeier, 2020a). A study on this ethical
design might provide new, interesting knowledge about the context of the creation of
dark patterns and the ways to counteract dark patterns. Additionally, it might also help
with clarifying the blurred fine line that separates bright patterns from dark patterns.
This can be achieved by providing guidelines that clearly define when persuasive
techniques are considered as bright or dark.

1.2 Related Works

In this section, previous work in the field is discussed to highlight the knowledge gap
that this study intends to fill. Further knowledge concerning dark patterns and the ethics
of persuasive design is discussed in Chapter 3 – Theoretical Framework alongside
theories used in this study.

1.2.1 Dark Patterns

Dark patterns have been found to be prevalent from websites to mobile applications,
and more (Di Geronimo et al., 2020; Mathur et al., 2019). Di Geronimo et al. (2020)
showed that 1,787 dark patterns were found among 240 mobile applications, which
resulted in 95% of the mobile applications containing one or more dark patterns. They
also concluded that mobile applications have on average more than seven instances of
dark patterns. Mathur et al. (2019) performed a web crawl of eleven thousand shopping
websites, identifying 1,818 instances of dark patterns on 1,254 websites (11.1%) out of
the eleven thousand shopping websites. Additionally, they identified that popular
shopping websites, based on Alexa rankings, were more likely to contain some

13
instances of dark patterns. Another interesting finding is that they managed to identify
22 third-party entities that help shopping websites to create and incorporate dark
patterns on their websites.
Now, most consumers are too accustomed to dark patterns that they do not protect
themselves from them, according to M. Bhoot et al. (2020). They identified that
although most of their study participants claimed to have never been tricked by any
websites, the participants failed at identifying Brignull’s 12 types of dark patterns and
fell victim to at least one of them. This means that the participants had some false sense
of confidence that they were not affected by dark patterns. Although the participants of
the study were somewhat aware of dark patterns, they noted that they were accustomed
to them to the point that they accepted them as being a part of the internet experience.
This means that the participants would persistently use the website despite being aware
of its incorporation of dark patterns if participants’ motives to use the website were high
enough. Finally, participants claimed that they distrusted websites that employed dark
patterns and preferred not using them unless necessary.
Di Geronimo et al. (2020) carried out an online experiment consisting of 589
participants. The online experiment resulted in most participants (55%) not spotting the
malicious designs in the application containing dark patterns while some were unsure
(20%), and the remaining were successful in finding malicious designs in the
application (25%). In an additional control task, most participants (86%) successfully
identified that the application did not contain any dark patterns. This is interesting
considering that participants find it easier to notice the absence of dark patterns rather
than the prevalence of them. Participants that were not able to identify dark patterns or
were unsure reasoned their choices by arguing that dark patterns had become so
common, and a part of the normal interaction flow when using applications and
websites. The authors concluded that if a consumer is knowledgeable about dark
patterns, it will help him or her to detect them, despite their prevalence.
Some studies mention the need for raising awareness about dark patterns as a method
to combat them. Karagoel and Nathan-Roberts (2021) argue that raising awareness of
dark patterns in different platforms is one of many ways to prevent designers from
manipulating consumers with malicious designs. Di Geronimo et al. (2020) suggest
increasing consumers’ awareness of dark patterns as future research to combat dark
patterns. However, other studies such as Bongard-Blanchy et al. (2021) found that
raising consumers’ awareness of dark patterns is not sufficient to protect them from the
influence of dark patterns. Consequently, these studies show that shining light and
raising awareness of dark patterns is not sufficient to overcome them, since consumers
are dependent on certain websites or applications that incorporate dark patterns and will
therefore keep using them. Instead, ethical measures must be taken to counteract dark

14
patterns efficiently (Bongard-Blanchy et al., 2021; Mathur et al., 2019; Narayanan et
al., 2020).

1.2.2 Dark Patterns Raise Ethical Concerns

Many academic works on ethics in persuasive design suggest ethical approaches that
designers should consider in their design work to address ethical issues (Kight & Gram-
Hansen, 2019; Narayanan et al., 2020; McLean, 2020; Mildner & Savino, 2020; Benner
et al., 2021; Karagoel & Nathan-Roberts, 2021). Narayanan et al. (2020) argue that
consumers lose trust in the system over time when they realize they are being
manipulated into doing something that leads to undesired outcomes. Karagoel and
Nathan-Roberts (2021) argue that it is important to increase consumers’ awareness of
being persuaded through transparency in a similar way as other academic works
(Benner et al., 2021) have articulated. Karagoel and Nathan-Roberts (2021) claim that
it is considered unethical when consumers’ decisions are forced, and not deliberated, in
their oblivious state.
Some academics (Atkinson, 2006; Karppinen & Oinas-Kukkonen, 2013) have other
thoughts about the ethics of transparency than Karagoel and Nathan-Roberts (2021).
Atkinson (2006) argues that persuasion made with ethically good intention can be
considered ethical, even though consumers are not aware of the intention to persuade
them, while Karppinen and Oinas-Kukkonen (2013) argue that an information system
is not automatically considered ethical just because it is transparent. Benner et al. (2021)
share Atkinson’s statement that consumers' unawareness is not necessarily bad and then
goes on to explain that too much disclosure can lead to reverse outcomes like negative
attitudes towards products or services and a decrease in the effect of persuasion. As
previously mentioned in section 1.2.1 – Dark Patterns, some academic works on dark
patterns have already examined consumers’ awareness to only conclude that it will not
change their behavior to oppose dark patterns.
As aforementioned, persuasive design is considered ethical when consumers have the
right to freedom of choice. Benner et al. (2021) recommend the ethical approach called
nudging to respect consumers’ freedom of choice, and even for their autonomy.
Nudging is considered ethical but can also become unethical when used with dark
intentions like setting a fake countdown timer to nudge consumers to act with urgency
(Narayanan et al., 2020). Academics have addressed these unethical issues, or dark
patterns defined in 3.1 – Types of Dark Patterns, through approaches such as ethics or
ethical standards in the design process (Narayanan et al., 2020; McLean, 2020;
Karagoel & Nathan-Roberts, 2021). McLean (2020) recommends designers to mitigate
the negative effects of dark patterns through codes of professional conduct for
designers, such as The ACM Code of Ethics and Professional Conduct and The Code
of Professional Conduct developed by the User Experience (UX), to provide ethical
guidance.

15
Both codes of professional conduct advocate designers to be honest, but the example
about the fake countdown timer goes against this very conduct. Ethical issues come
with designers’ unethical design practice of dark patterns. Academics discuss anti-dark
patterns approaches that designers and consumers alike can consider to address ethical
issues, but these approaches have not been adopted in the real world. Some academic
works share the same view that no research has investigated how designers in the
workplace address ethics - whether designers assimilate the suggested approaches,
whether these approaches work in practice or not, and whether designers have the right
to free speech regarding ethics in design (Kight & Gram-Hansen, 2019; Benner et al.,
2021). Fansher et al. (2018) seem to have already examined the latter with the
conclusion that some designers “shame” other designers or companies that adopt
unethical design practices on social media. It can be argued that designers reach out to
social media to vent out their concerns about ethical issues because they have no right
to determine ethics in design. Still, it is not clear with only an assumption, thus there is
a need to investigate further into the ethical dilemma that designers encounter in their
workplace.

1.2.3 Bright Patterns Address Ethical Issues

Kight and Gram-Hansen (2019) discuss that it can be the management that determines
the standards of ethics rather than the designers themselves. Arguably, managers may
put pressure on designers to use any means necessary to increase revenue, preventing
designers to be ethical in their design process. Graßl et al. (2021) have contributed to
the insight that designers can address ethical issues with the ethical approach called
bright patterns. They found out that more than one-third (36.1%) of the participants
changed their decisions (or were nudged) to the privacy-friendly option (“do not agree”)
in privacy cookie consent when the option was preselected. Bright patterns could be the
future approach to address ethical issues, but more research in other context than
privacy cookie consent is required before academics investigate whether designers can
assimilate the approach and whether the approach works in practice. Other ethical
approaches mentioned earlier could also be investigated, but it would be advantageous
to further investigate bright patterns that have already been researched to be promising.

1.2.4 Knowledge Gap

The knowledge gap that this study intends to fill has been identified following the
survey of related works.
Designers are held responsible for creating dark patterns and not being sufficiently
ethically aware in their design work. Previous work has examined dark patterns from
the consumers’ perspective and suggested that the designers’ perspective should be
further investigated. This study aims to provide an insight into the ethical dilemmas that
designers encounter in their design work, which thus far has been heavily understudied.

16
Bright patterns have been considered as an ethical approach to counteract dark patterns
by academics. However, little research has been carried out to truly examine their
potential in counteracting dark patterns. This study aims to further investigate bright
patterns, and the potential that they have in counteracting dark patterns, by
incorporating them in multiple parts of the interface and pit them against more types of
dark patterns.

1.3 Purpose and Research Questions

The purpose of this study is twofold. Firstly, the purpose is to explore the ethical
dilemma in design that UX designers encounter in their workplace to understand the
reason behind the prevalence of dark patterns. Secondly, the purpose is to
explore consumers’ perception of ethics in bright patterns and dark patterns, and
consumers’ decisions between bright patterns and dark patterns to determine whether
bright patterns are a potential ethical approach that designers can adopt in the future.
To ensure that the purpose of this study is fulfilled, two groups of research questions
have been defined.
Academics share the opinion that research should focus on designers and not only
consumers. Consumers are aware of dark patterns but refuse to oppose them, while
designers are blamed for crafting dark patterns. If consumers will not oppose dark
patterns, then only designers can take the responsibility to make it happen. The problem
is that designers can be unethical, thus the three following research questions aim to
look closer into what it is that may influence designers’ work with ethics in design:
RQ1a: Who has the right to determine how designers should address ethics
in the designs at their workplace?

RQ1b: What are the ethical issues that designers encounter in their designs
at their workplace?

RQ1c: How do designers approach ethical issues that occur in their designs
at their workplace?

Bright patterns provide freedom of choice and transparency for consumers unlike dark
patterns, which are disingenuous and limit consumers’ choices. Hence, RQ2a aims to
investigate whether bright patterns’ influence on consumer decisions differ from the
influence that dark patterns employ, and if that is the case, how the consumer decisions
differ.
It is also important to note that ethics are very subjective. A designer might craft a bright
pattern that they deem is ethically good, but that does not necessarily mean that the
consumer agrees that it is ethically good. The consumer must feel that the bright pattern
is ethically good for the bright pattern to be ethically good. Hence, RQ2b aims to

17
investigate whether consumers’ perceptions of ethics differ in bright patterns and dark
patterns.
RQ2a: How do bright patterns (freedom of choice and transparency) in
comparison to dark patterns (lack of choice and non-transparency)
influence consumers’ decisions?

RQ2b: What is the difference between consumers’ perceptions of ethics

(freedom of choice and transparency) in bright patterns and dark patterns?

1.3.1 Expected Outcomes

This study aims to shed light on the ethics of persuasive design both from designers’
and consumers’ perspectives. Ideally, this study results in a deeper understanding of the
ethical dilemmas that designers encounter in their design work. Additionally, this study
results in the ethical approach of bright patterns being embraced by consumers,
becoming successful in counteracting dark patterns and possibly replacing them in the
future.

1.4 Scope and Delimitations

In this section, the scope and delimitations of this study are defined.

1.4.1 Replacing Dark Patterns With Bright Patterns

Some dark patterns are easier to turn into bright patterns than others. Other dark patterns
(for example Privacy Zuckering, Disguised Ads, and Friend Spam) are very difficult to
incorporate into an interactive prototype like the ones developed for this study’s
experiment. These dark patterns are prominent in fully developed websites or mobile
applications that run ads or have access to consumers’ personal data. Hence, the
prototypes used in the experiment do not include dark patterns that are difficult to
convert to bright patterns, nor do they include dark patterns that require running ads or
having access to consumers’ personal data.

1.4.2 Grey Zone

Some dark patterns are blatantly manipulative while other dark patterns are more subtle
e.g., misdirecting consumers by using colours and font sizes in such a way that
purposefully distract consumers from a certain thing. Sometimes, designs can appear to
be a dark pattern but is not, which is a grey zone that is not considered in this study. For
example, many free mobile applications contain a lot of ads that the consumer cannot
get rid of unless he or she pays a fee. Although the ads can irritate the consumer, these
are not dark patterns since they allow the mobile application to generate revenue. All
things considered, the consumer has a choice, either to endure the ads and use the app
for free or to pay a one-time fee to get rid of the ads.

18
1.4.3 Perspective of Stakeholders

As previously mentioned in section 1.1 - Background, designers might be pressured by

their stakeholders to incorporate dark patterns into their design to meet certain goals or
metrics. This study only investigates the replacement of dark patterns with bright
patterns from the consumers’ and designers’ perspectives and does not include the
perspective of stakeholders or business executives that might be inclined toward using
dark patterns in the pursuit of meeting certain metrics.

1.5 Disposition
A brief overview of the report structure is presented in this section. Following Chapter
1 - Introduction, the report has five other chapters as follows:
Chapter 2 - Method and Implementation, this chapter describes and discusses the
method choices and method implementation. The former involves the definition and
motivation of the chosen methods for data collection and data analysis, but also
evaluation of the validity and reliability of the methods. The latter involves a thorough
description of the implementation of the chosen methods. The chapter also includes a
brief walkthrough of the two prototypes created for the experiment of this study.
Chapter 3 - Theoretical Framework, this chapter presents state-of-the-art literatures,
theories, and concepts relevant to the research questions. These are used later for
understanding the results from the empirical investigation.
Chapter 4 – Results and Analysis, this chapter presents the findings from this study
including the findings from the interview with UX designers, as well as the findings
from the experiment with follow-up interview. The chapter is concluded by
summarizing the findings and answering this study’s research questions.
Chapter 5 – Discussion, this chapter provides answers to the first group of research
questions by discussing the findings of the semi-structured interview with UX
designers. Both the findings from the semi-structured interviews and experiment with
follow-up interview are discussed by referencing to previous work and theories.
Additionally, this chapter discusses the implications of this study’s method and
implementation choices.
Chapter 6 – Conclusions and Further Research, this chapter presents the conclusions
drawn from this study along with its practical and scientifical implications, and the
suggestions for potential further research.

19
2 Method and Implementation
In this chapter, the method and implementation choices of this study are defined,
including the validity and reliability of said methods. This chapter also includes a brief
walkthrough of the two prototypes that were used in the experiment of this study.

2.1 Method Choices

In this section, the method and implementation choices of this study are defined and
motivated.

2.1.1 Participant Recruitment: Screener Survey

Participants were recruited through two separate online surveys that were distributed
on social media such as Facebook and LinkedIn. Facebook is not only a useful tool to
reach out to a multitude of people, but also to target people like UX designers, who
belong to a certain community group. LinkedIn is also a useful tool since it contains a
lot of professional contacts and the ability for posts to travel through their networks.
However, distributing the two surveys on Facebook and LinkedIn did not help to recruit
many participants for both the semi-structured interviews with UX designers and the
experiment with a follow-up interview. Therefore, other measures had to be taken to
recruit participants. Personal and professional contacts were approached to be recruited
as initial participants, thereafter, the snowball sampling method was used to recruit
more participants. The snowball sampling method refers to acquiring an initial smaller
sample of participants, and those participants then reach out to their own contacts and
recommend them as valid participants (Parker et al., 2019). This method was
predominantly used for the interviews with UX designers, but also for the experiment.
One survey aimed to recruit 15-20 participants for the experiment, whereas the other
survey aimed to recruit 5-10 participants for the semi-structured interview. Ideally,
more participants should be recruited but the number of participants was limited due to
the timeline of this study. Both surveys aimed to recruit participants over 18 years old
for the similar reason that it would be time-consuming to recruit children who need
their legal guardians’ consent to be part of this study.
A survey is a fast and cheap method used to recruit many participants in a short
time. Some surveys possibly fall short when they contain too much information or
demand too much time and attention from the respondents. However, simple surveys
that do not demand too much time and attention from the respondents have a higher
success rate (Saunders et al., 2015). The surveys used to recruit participants in this study
were very simple, only asking if users were interested in partaking in the study and
asking for their contact information.

20
2.1.2 Semi-structured Interview

A semi-structured interview is a suitable method to explore the first three research

questions (RQ1a, RQ1b, and RQ1c). The method is suitable because it is a type of
interview that involves interviewers asking predetermined questions and unplanned
follow-up questions to gather in-depth qualitative data. The former guides interviewers
to focus on and not divert from the topic entirely, whereas the latter provides
interviewers the opportunity to probe interviewees to explain or build on their answers
(Saunders et al., 2015). These are the advantages of the semi-structured interview,
which the other types of interviews do not possess. Structured interviews have a
structure that hinders interviewers from asking questions for in-depth data, while
unstructured interviews have no structure and might therefore produce irrelevant data.
A semi-structured interview will not only produce rich data, but also provide time for
the interviewees to open up about sensitive issues. As some individuals may view ethics
as a sensitive topic, the semi-structured interview can be viewed as the most suitable
method for this study.
Semi-structured interviews can either be held online or offline (Saunders et al., 2015).
The former refers to interviews that take place online through communication software
anywhere and anytime, whereas the latter refers to face-to-face interviews where
interviewers meet participants in the real world. Online interviews make it easier for
interviewers to reach participants from a long distance, which is more difficult to do
with offline interviews. Considering the effort that participants must make to partake in
an offline interview, it increases the risk of participants cancelling the interview.
Therefore, online semi-structured interviews were carried out to explore the first three
research questions (RQ1a, RQ1b, and RQ1c).

2.1.3 Experiment: Within-Subjects Study Design

An experiment is a suitable method to explore the fourth research question

(RQ2a) whilst a structured interview is a method that is more suitable to explore the
fifth research question (RQ2b).
Since RQ2a seeks to explore bright patterns and dark patterns influence on consumers’
decisions, an experiment that is used to investigate whether one variable (independent)
causes a change in another variable (dependent) (Saunders et al., 2015) is deemed as a
suitable method. However, the independent variable is not bright patterns or dark
patterns but is an artifact with two values being the bright patterns and dark patterns
(see section 2.2 – Artifact Creation). The dependent variable is the consumer decisions.
Experiments are usually conducted in a laboratory setting in which participants who are
randomly assigned to the experimental group and control group will be tested. This is
the classical approach, but the within-subjects design is a more practical approach for

21
this study. The within-subjects design requires fewer participants because they are
exposed to all conditions (Saunders et al., 2015).
Since RQ2b seeks to explore the difference in consumer’s perception about ethics
in bright patterns and dark patterns, a structured interview that is great for data
comparisons is therefore a suitable method. A structured interview involves
interviewers to ask the same predetermined questions in the same order, while other
types of interviews, such as semi-structured and unstructured interview, are open to
unplanned questions (Saunders et al., 2015). It would be difficult for interviewers to
compare data between participants if the questions are different in each interview.
The experiment and structured interviews were carried out online on Zoom. Carrying
out the experiment and structured interview online was beneficial since it allowed
participants to partake in the experiment from the comfort of their own homes. This
ensured that participants did not need to make the effort to travel to a physical location
to participate in the experiment. Additionally, this opened the possibility to include
participants outside of the geographical location where the study was carried out.
The prerequisites needed for the experiment included that the participants could
communicate clearly with a microphone, show themselves by using a webcam, share
their screen while they perform the tasks of the experiment and that they have given
consent to be recorded during the process. When the participants had completed the
tasks of the experiment, they were allowed to stop sharing their screen and the
structured interview began. Every conversation with participants was recorded to
ensure that nothing got lost in translation, in contrast to taking notes, which ultimately
facilitated the analysis of the collected data.

2.1.4 Content Analysis

Content analysis is a qualitative method that is used to analyze the data collected from
both the semi-structured interviews with UX designers, as well as the data collected
from the experiment’s follow-up structured interviews. Content analysis involves a
procedure and outcome that consists of coding the data, or units of analysis, into
categories to present the number of occurrences in each category. The analyzed data is
then visualized using quantitative methods like tables and graphs (Saunders et al.,
2015).
Content analysis can adapt two forms, these are manifest content analysis and latent
content analysis.
Manifest Content Analysis
In manifest content analysis, words and phrases that are physically present in a dataset
are analyzed and coded into predetermined categories. The number of occurrences of
data is then presented in each category. Hence, manifest content analysis has

22
limitations, but these limitations enhance the consistency and transparency of the
analysis, which in turn increases reliability (Saunders et al., 2015).
Latent Content Analysis
Latent content analysis builds upon manifest content analysis in several ways. It goes
beyond coding words and phrases and coding them into predetermined categories in the
sense that it includes interpreting the meaning of the units of analysis. Latent content
analysis is not limited to words and phrases as units of analysis, it also includes
sentences and paragraphs. Latent content analysis is also not limited to predetermined
categories, it also includes establishing and adding more categories along the analysis
process. Since latent content analysis is more open and subjective than manifest content
analysis due to its lack of limitations, it is fair to state that it reduces the reliability of
the analysis. However, the consistency and transparency of the analysis can be
enhanced by clearly defining categories and their properties (Saunders et al., 2015).
Analysis Process
The analysis process of the experiment’s follow-up structured interview began by
adapting a manifest content approach. One of the experiment transcripts was analyzed
using the manifest content approach with the following predetermined categories:
perceived freedom of choice, perceived transparency, and automatic behavior. This
experiment transcript served as a pilot test in terms of analysis to determine if the
manifest content approach was appropriate. The result of this pilot test concluded that
the manifest content approach was in fact inappropriate, and the predetermined
categories were deemed insufficient, as they did not cover all relevant findings from the
transcript.
Therefore, the latent content approach was adopted instead, which included the
previous predetermined categories as well as new categories that were established while
analyzing. The process then consisted of each researcher analyzing the transcripts and
coding data separately. The researchers then compared their codes and discussed the
categories’ relevancy to the research questions, to establish a common ground and
finalize the analysis of a transcript. Finally, the analyzed data was compiled into an
Excel spreadsheet and quantified into several graphs.
After the content analysis of the experiment’s follow-up structured interviews were
done, the next step was to use the latent content approach to analyze the data of the
semi-structured interviews. The analysis process of the experiment’s follow-up
structured interviews was the same for the semi-structured interviews, excluding the
presentation of the analyzed data. It was performed twice to further analyze the
compiled data, and this led to identifying subcategories of the categories that were
identified in the first round of analysis. Then, the analyzed data was quantified and
presented in a table.

23
Motivation for Adopting Latent Content Analysis
There are multiple reasons why latent content analysis was selected for both the semi-
structured interviews with UX designers and the experiment’s follow-up structured
interviews. Latent content analysis allows for a richer data analysis that is not as limited
as manifest content analysis, in the sense that the meaning of the units of analysis is
interpreted, that it includes paragraphs and sentences, and that further categories can be
established along the analysis process.
Since semi-structured interviews allow for in-depth and unplanned questions that result
in rich qualitative data, it is in line with the benefits of adopting a latent content analysis
approach. Therefore, to analyze the collected data that answers the first group of
research questions, latent content analysis was deemed an appropriate approach.
On the other hand, structured interviews are often perceived as quantitative in nature.
However, some of the questions that were asked in the experiment’s follow-up
structured interviews were very broad which could result in rich and elaborate answers.
Therefore, latent content analysis was also deemed an appropriate approach to analyze
the collected data that answers the second group of research questions.
Coding Rules
To increase the validity and reliability of the analysis, and to ensure that the researchers
were consistent across their separate analyses, a couple of coding rules were
established. These are the following:

• Grammatical tenses are perceived as indifferent. This means that a certain word
or phrase in one grammatical tense is equal to the same phrase in another
grammatical tense. Example: “It was difficult to understand” is considered equal
to “It is difficult to understand”.
• Filler words such as “like”, “you know”, and “basically” are meaningless words
that mark a pause or hesitation in speech. Therefore, these words are treated as
such and are not carefully examined.

2.2 Artifact Creation

As previously mentioned in section 2.1.3 - Experiment: Within-Subjects Study Design,
the independent variable, which was an artifact with two values: bright patterns and
dark patterns, was investigated in the experiment. The artifact was built from scratch in
the digital prototyping tool Figma which was accessible through Jönköping University.
A prototype is a draft version of a product’s design with limited interactive features
(Figma, n.d.) and was therefore a good approach to investigate the dependent variable:

24
consumer decisions. Consumers’ or participants’ decisions were observable through
their interaction with the artifact or prototype (see section 2.3 - Data Collection and
Procedure for the experiment execution).
Figure 1 shows two prototypes that were built in Figma, one of them being a dark
patterns prototype (Prototype A), and the other being a bright patterns prototype
(Prototype B). Both prototypes were designed to resemble a clothing e-commerce
website that consumers can find online in the real world. These prototypes were inspired
from actual e-commerce websites such as Zalando and ASOS. A pilot testing was
conducted to discover whether the prototypes were flawed before the actual experiment
with participants took place. From the pilot testing, some flaws that were discovered as
few test participants (such as family and friends) interacted with the prototypes could
be addressed immediately. The prototypes had both a women’s page and a men’s page,
but the following figures only show examples of the women’s page.

Figure 1: Women's home page for both prototypes.

25
Prototype A was designed to include 6 dark patterns as follows: Price Comparison
Prevention, False Urgency Scarcity, Sneak into Basket, Trick Questions, Roach Motel,
and Misdirection (see section 3.1 - Types of Dark Patterns for definitions). To decide
which dark patterns were to be included in the prototypes used in the experiment,
several factors came into play. Firstly, thirteen different types of dark patterns such as
the twelve identified by Brignull (2019) as well as False Urgency Scarcity were
identified. As previously mentioned in section 1.4.1 - Replacing Dark Patterns With
Bright Patterns, several constraints had to be adhered to when choosing the dark
patterns that were going to be included in the prototypes. Firstly, four dark patterns that
require an application to run ads or have access to consumers’ personal data (e.g.
Disguised Ads and Privacy Zuckering) were excluded from the experiment since such
features were not possible to implement in the prototypes. Out of the nine remaining
dark patterns, three of them (Hidden Costs, Bait and Switch, and Confirmshaming)
were dismissed because they were more difficult to convert to bright patterns than the
other six, which are included in this prototype.
Prototype B was designed to include 6 bright patterns that we considered to be bright
patterns based on the definition in section 3.2.2 - Bright Patterns. Currently, types of
bright patterns have yet to be established like dark patterns. As aforementioned, a bright
pattern is a type of persuasive design that provides consumers with both freedom of
choice and transparency (Graßl et al., 2021; Wintermeier, 2020a).

2.2.1 Design Choices of the Artifact

Both prototypes were designed in a certain way to ensure that the variables this study
aims to investigate were covered in the experiment. Some features of the prototypes
were interactive to ensure that what we measured was relevant to the purpose of this
study. Once again, the purpose of this study is to determine whether bright patterns are
a potential ethical approach by comparing the influence that bright patterns and dark
patterns have on consumer decisions.
Prototype A
In Prototype A, the design choices were made in accordance with the definition of the
dark patterns implemented in the prototype. Price Comparison Prevention refers to
design choices that complicate the consumer’s ability to compare the prices of different
items. This aims, like all the other dark patterns, to trick consumers into making
decisions they did not intend to make (Wintermeier, 2020b). If a consumer intends to
buy the cheapest white t-shirt but chooses one that is not the cheapest, this shows that
the consumer has fallen victim to the dark pattern. As shown in Figure 2, the prototype
was partly designed to complicate the consumer’s ability to compare the prices between
the items she or he wants to buy. One item is presented with the original price and the
other item on discount is presented with a price that could be interpreted as the discount

26
or original price. The prototype was designed to only present the consumer with two
choices to measure the influence that Price Comparison Prevention has on the
consumer’s decision.

Figure 2: Price Comparison Prevention Dark Pattern in Prototype A. The consumer can
choose between the two highlighted t-shirts, but the price is hard to compare. The t-
shirt in the bottom row is discounted with 5% off, but it does not show the initial price.

27
False Urgency Scarcity refers to design choices that urge the consumer to buy a product
of interest to him or her. Figure 3 presents this dark pattern. One item presents a scarcity
message whereas the other item does not have such message. If a consumer feels a sense
of urgency when he or she views the item with the scarcity message and chooses to buy
it, this shows that the consumer has fallen victim to the dark pattern.

Figure 3: False Urgency Scarcity Dark Pattern in Prototype A. The consumer can
choose between the two highlighted options, but one of them has a false urgency
scarcity message on it to put pressure on the consumer.

28
Sneak into Basket refers to a situation when an item is added into the consumer’s
shopping bag without informing the consumer. As shown in Figure 4, an item is added
into a shopping bag, and this item can either be missed or seen by the consumer. If a
consumer does not realize the added item or realize it later in the checkout process, then
he or she has bought an item without the intention to do it. The consumer has fallen
victim to the dark pattern unless he or she notices the added item and removes it.

Figure 4: Sneak into Basket Dark Pattern in Prototype A. When the consumer goes to
his/her shopping bag, an extra item has been added. The user can choose to remove the
item, but it is still annoying and causes friction.

29
Trick Questions refers to information with language that confuses a consumer into
making an unintended decision. Figure 5 presents this dark pattern in the form of a
newsletter registration popup that includes two checkboxes with information aiming to
confuse the consumer into registering. A consumer has fallen victim to the dark pattern
when he or she does not have the intention to register the newsletter but ticks the
checkbox for registering the newsletter unintentionally.

Figure 5: Trick Questions Dark Pattern in Prototype A. When the consumer chooses
payment options, they are met with this newsletter pop-up. Two options are presented
that look very similar at first, enhancing confusion.

Roach Motel refers to design choices that make it easy for a consumer to easily get into
a situation but harder to get out of the same situation. As shown in Figure 6 and Figure
7, the link that leads to the desired situation, which is to find the page for deleting the
account, is hidden within other links and presented in a label that is uncommonly used
in real e-commerce websites. If a consumer clicks on the wrong links, it shows that he
or she has a hard time finding the page for deleting the account. The consumer has fallen
victim to the dark pattern unless he or she succeeds in clicking the right links.

30
Figure 6: Roach Motel Dark Pattern in Prototype A, Part 1. When a consumer wants to
delete their account, they need to go through several links to do so. First, they must go
to FAQ.

Figure 7: Roach Motel Dark Pattern in Prototype A, Part 2. When the consumer has
navigated to the FAQ, they need to find the highlighted "Delete data" link to get one
step closer to terminating their account.

31
Misdirection refers to design choices, such as colors and font sizes, aiming to distract a
consumer from making a certain decision. Figure 8 presents this dark pattern. On the
right-hand side, the button deletes the consumer’s account, and it is the reverse for the
other button on the left-hand side. The “Never mind” button is colored in bright red to
grab the consumer’s attention and distract him or her from deleting the account. If the
consumer intends to delete the account but clicks the button that does not delete the
account, then he or she has fallen victim to the dark pattern.

Figure 8: Misdirection Dark Pattern in Prototype A. When the consumer wants to delete
their account, the color and size of the "never mind" option has been highlighted to try
to sway the consumer to not delete their account.

Prototype B
The bright patterns implemented in Prototype B mimicked the dark patterns
implemented in Prototype A but were designed to provide higher freedom of choice
and transparency for consumers. In the following paragraphs, the way that the bright
patterns implemented in Prototype B differ from the dark patterns implemented in
Prototype A is explained.

32
The Price Comparison Prevention bright pattern was designed in such a way that the 3-
Pack option is a bit cheaper and easy to divide by three to attain the price per t-shirt. To
further facilitate the price comparison between the two options, the price per t-shirt of
the 3-Pack is listed. Additionally, the other option has a clearer discount and shows the
initial price of the t-shirt prior to the discount. These factors provide consumers with
high transparency and gives them the opportunity to make an informed decision for
their purchase. The bright pattern counterpart of Price Comparison Prevention is shown
in Figure 9.

Figure 9: Price Comparison Prevention Bright Pattern in Prototype B. The 3-pack is

now a bit cheaper and lists price per t-shirt. The other option has a clearer discount and
shows the initial price before the discount.

33
Since the False Urgency Scarcity dark pattern implies that the website is lying about its
popularity or limited stock to consumers, the bright pattern equivalent to this is to not
present any False Urgency Scarcity messages at all. In that sense, consumers are
provided with high transparency since they are neither being lied to nor urged to make
an impulsive decision. The bright pattern counterpart of False Urgency Scarcity is
shown in Figure 10.

Figure 10: False Urgency Scarcity Bright Pattern in Prototype B. In this prototype, there
are no scarcity messages shown to the consumer.

34
Regarding Sneak into Basket, consumers are forced to either buy an item they did not
choose or remove the item from their shopping bag. This violates both consumers’
freedom of choice and transparency since they did not choose the item in the first place,
and the website does not notify them of the extra item. To convert this dark pattern into
a bright pattern, consumers are met with an item recommendation popup instead.
Consumers can choose to either accept or decline the item, but the difference is that
they are being asked first, the item is not forced upon them. This provides consumers
with higher freedom of choice and transparency. The bright pattern counterpart of
Sneak into Basket is shown in Figure 11.

Figure 11: Sneak into Basket Bright Pattern in Prototype B. Instead of sneaking in a
product to the consumer's shopping bag, consumers are met with this pop-up. This
functions as a recommendation to the consumer, and they can choose to add the item or
decline.

35
The Trick Questions bright pattern is designed in such a way that consumers are only
presented with one checkbox rather than two. They can tick the checkbox to opt into
the newsletter, leave the checkbox empty or close the newsletter pop-up down with the
cross to opt-out of the newsletter. This provides consumers with higher freedom of
choice and transparency. The bright pattern counterpart of Trick Questions is shown in
Figure 12.

Figure 12: Trick Questions Bright Pattern in Prototype B. In this newsletter pop-up,
consumers need to either check/uncheck one box to continue, or they can close it down
with the cross in the upper right corner.

36
Roach Motel was implemented in Prototype A to make the account deletion
unnecessarily difficult for consumers, by embedding the path to the “Delete account”
page in several links. To convert Roach Motel to a bright pattern, the path to the “Delete
account” page is far easier to find, which makes the account deletion process
straightforward. This provides higher transparency for consumers. The bright pattern
counterpart of Roach Motel is shown in Figure 13.

Figure 13: Roach Motel Bright Pattern in Prototype B. In this prototype, the consumer
does not have to navigate through an unnecessary number of links to terminate their
account.

The bright pattern counterpart of Misdirection is designed in such a way that it does not
deceive or distract consumers by using colors and font sizes in a certain way. To delete
the account, the big bright red action button is the “Delete” button while the smaller
button next to it fulfills the purpose of cancelling the account deletion. The bright
pattern counterpart of Misdirection is shown in Figure 14.

Figure 14: Misdirection Bright Pattern in Prototype B. In this protype, the consumer
does not get misdirected by the use of color and size. The big action button deletes the
account.

37
2.3 Data Collection and Procedure
In this section, the data collection techniques of this study are described and specifically
applied to their respective parts of the study.

2.3.1 Interview With UX Designers

Interviews were conducted to gather the information that answer the three research
questions: RQ1a, RQ1b, and RQ1c. Four participants were recruited for the interviews,
consisting of three female participants and one male participant. As instructed in the
screener survey, the interview was held online via Zoom on the date and time agreed
between the participant and interviewers.
Before the interview, each participant was informed about the interview such as the
purpose and duration, and data management by us. Participants approved of being video
and audio recorded in the screener survey, but they were asked about the approval again
to confirm whether they had changed their mind or not.
During the interview, each participant was asked eight questions and sometimes a few
follow-up questions. The first three questions were warm-up and easy questions to
make each participant more comfortable during the interview, while the others were the
main questions. One interviewer delivered the three warm-up questions, but both
interviewers took turns to deliver the five main questions to allow one to rest during the
interview which could run for at least 40 minutes.
After the interview, each participant was thanked for participating in the interview.
Each participant was questioned about whether he or she had any other questions and
contacts that might be willing to participate in our interview. Some greetings were
exchanged between the participant and the interviewers before the interview ended.

2.3.2 Experiment With Follow-up Interview

15 participants were recruited for the experiment with follow-up interview, consisting
of 12 male participants and three female participants.
Participants were informed prior to the experiment that the purpose of the experiment
was to observe how they use the prototypes and to examine what their decision-making
process looks like. Although this was true, no information was disclosed regarding
bright and dark patterns to the participants to retain the element of surprise.
All participants were randomly assigned to start with one prototype and then switch to
the other prototype to ensure that there were no order effects. The prototypes were then
used by participants to carry out a set of smaller tasks where they would be confronted
with bright or dark patterns during each task. The prototypes represent an online
clothing store, and the tasks were the following:

38
 1. Find three plain white t-shirts with the lowest price among the t-shirt selection
and add them to your shopping bag.
2. Find a pair of jeans in the colour medium blue and add them to your shopping
bag.
3. Go through the checkout process to pay for the items in your shopping bag. Note
that you can make any choices you want if they are possible.
4. Terminate the account.

Once a participant had completed the tasks on one of the prototypes, they were
questioned about the decisions they made specifically related to the tasks. When the
tasks and follow-up questions were finished on one prototype, participants had to repeat
the same process on the other prototype. To conclude the experiment, participants were
then questioned about their perceived transparency and freedom of choice of the two
prototypes. The questions that were asked to participants during the follow-up
interviews are shown in Appendix 2. The experiment with follow-up interview lasted
approximately 20-30 minutes.
RQ2a aims to investigate whether bright patterns’ influence on consumer decisions
differs from the influence that dark patterns employ and if that is the case, how the
consumer decisions differ. By exposing consumers to bright and dark patterns and
examining their decisions in the experiment, RQ2a could be answered.
RQ2b aims to examine the difference between consumers’ perceptions of ethics
(freedom of choice and transparency) in bright patterns and dark patterns. RQ2b was
partly answered by the questions related to the tasks in the follow-up interview since
their perceived freedom of choice and transparency were identified implicitly.
However, during the concluding questions of the follow-up interview, participants were
explicitly asked which of the two prototypes they perceived as being the most
transparent and providing the most freedom of choice.

2.4 Validity and Reliability

In this section, the validity and reliability of the method choices of this study are
discussed.

2.4.1 Semi-structured Interview: Validity and Reliability

Saunders et al. (2015) introduce data quality issues associated with semi-structured
interviews: reliability, forms of bias, cultural differences, generalisability, and validity.
Since the semi-structured interviews did not involve participants with different cultural
orientations, the cultural differences were not taken into consideration in this study.
However, the other data quality issues were considered during the preparation and
conduct of the semi-structured interview to ensure data quality.

39
Reliability concerns whether other researchers obtain the same results and reach the
same conclusions through a repetition of the research method procedure that another
researcher carried out in a study (Saunders et al., 2015). According to Saunders et al.
(2015), semi-structured interviews are not intended to be repeatable or replicable since
they provide data at a specific time in a certain situation that reflects the reality which
may be subject to change. However, they state that explanations of how the data were
obtained and analyzed are sufficient to show that findings of a study are reliable. To
ensure a reliable study, this paper presents elaborated explanations to why certain
methods were chosen and how methods were used to conduct this study.
Biases are also related to the reliability of this study. Saunders et al. (2015) state types
of bias that researchers need to consider in interviews, these are interview bias,
interviewee or response bias, and participation bias. They explain that interviewer bias
occurs when interviewers’ tone or non-verbal behavior influences interviewees’
responses, interviewee or response bias occurs when interviewees are willing to
participate in an interview but unwilling to reveal information about a topic being
explored, and participation bias occurs when interviewees are less willing to keep
taking part in an interview that requires a large amount of time. Since researchers can
overcome these biases in the way interviews are prepared and conducted (Saunders et
al. 2015), an interview guide was constructed to minimize biases as much as possible.
As shown in Appendix 1, the interview guide ensured that all interviewees were told
about the time duration and asked questions that were not direct and misleading. It may
have helped to minimize the biases because the interviewees were willing to take part
in the interview to the end and reveal information to the questions being asked in the
interview. Efforts were also made to minimize the interviewer bias: questions were
delivered to interviewees in a neutral tone, nods were often made to signal that they
were heard, and recordings and transcriptions ensured that interviewees’ responses
were interpreted accurately and not lost in the memory limits of the human brain.
According to Saunders et al. (2015), a common argument is that semi-structured
interviews are often based on a small sample and are therefore not generalizable. The
argument says that a small sample provides findings that are not applicable to a broader
group of people or situations (Saunders et al., 2015). However, generalizability can be
achieved by allowing other researchers to design a similar study based on readings
about the full description of the research questions, design, context, findings and
resulting interpretations in a research paper (Saunders et al., 2015). Since the semi-
structured interview was based on a small sample, the findings of this study were not
generalizable, but this paper provides detailed descriptions about the study to ensure
that the findings can be generalizable in the future.
Semi-structured interviews may produce data of high validity in comparison to
generalizability (Saunders et al., 2015). Validity refers to the extent to which the

40
researcher can gather relevant data and accurately infer meanings that interviewees
intended to convey in their responses (Saunders et al., 2015). Researchers can achieve
a high level of validity by asking clarifying questions, probing meanings, and exploring
responses from a variety of angles (Saunders et al., 2015), which were done to ensure
valid data in this study. Biases may diminish the validity (Saunders et al., 2015), but
this is not an issue since actions have been taken to minimize the types of biases that
are common in semi-structured interviews.

2.4.2 Experiment: Validity and Reliability

Saunders et al. (2015) state that most experiments are carried out in a laboratory milieu
rather than in the field. This provides a higher level of control over several factors
included in the research process such as the sample selection and the limits of the
experiment, which improves the internal validity of the experiment. However, it makes
it more difficult to establish external validity since once again, experiments are carried
out in a laboratory milieu rather than in the field. In a perfect scenario, the dark and
bright patterns implemented in the two prototypes of this study would have been
implemented on a famous clothing e-commerce website just for testing purposes to
increase the “reality” of the experiment. Since that was not possible, a better solution
would have been to spend a lot of time creating interactive prototypes that are as
functional and as “real” as possible, possibly by programming the prototypes. However,
the two prototypes for this study have been developed with respect to the limited
timeframe of this study and are as functional and “real” as possible according to those
limitations.
Another major point that Saunders et al. (2015) make is that a standard experiment
usually contains hypotheses, a null hypothesis, and an alternative hypothesis. The null
hypothesis serves as a predictor of whether there will not be a meaningful difference or
relationship between the variables (independent and dependent variables). In contrast,
the alternative hypothesis serves as a predictor of whether there will be a meaningful
difference or relationship between the variables. Thereafter, both hypotheses are tested
statistically and they either get accepted or rejected. However, there are no hypotheses
brought forward in this study because this study adopts an exploratory approach. Little
is known about bright patterns in comparison to dark patterns, and therefore, this study
did not intend to validate or reject a certain claim, but rather shed light on an uncharted
area of research. Consequently, the research questions presented in section 1.3 -
Purpose and Research Questions were deemed sufficient to shed light on this obscure
area of research, not needing any hypotheses.
Budiu (2018) states the importance of randomization, more specifically related to
experiments that have a within-subjects experimental design. She insists that
randomization of tasks is imperative to ensure that there are no order effects that can
influence the experiment. This is something that has been taken into consideration in

41
this study as the participants were required to complete tasks on two similar prototypes.
However, the order in which participants were instructed to interact with the prototypes
was randomized for each participant, ensuring that there were no order effects, and
boosting validity.
The experiments were recorded throughout the entire experimental process, which
means that there was no need for notes to be taken during the experiment. This
minimized the risk of key information being missed or ignored and enhanced validity.
Additionally, the ability to replay the recorded experiments allowed for richer data
analysis where little to no information had the risk of getting lost in translation.
As aforementioned, reliability concerns whether other researchers can obtain the same
results and reach the same conclusions through a repetition of the research method
procedure that another researcher carried out in a study (Saunders et al, 2015).
Considering that the prototypes used in this experiment were very specific tools that
depicted the results of this study, if other researchers would use them to conduct a
similar study and ask participants the same follow-up questions, the results would most
likely turn out to be the same. However, other factors need to be considered such as the
sample size and the demographics of the sample. The larger the sample size is, the more
varied the results will be. Therefore, if other researchers would replicate this study with
more participants, the results might turn out differently. Additionally, if they would try
to replicate this study with participants that have a different demographical background
than the participants in our study, the results might turn out differently. All things
considered, these two factors might influence the reliability of this study, as do the
prototypes.
The first iteration of the experiment was conducted by both researchers together to
ensure that when the researchers later performed the iterations separately, they did it
the same way. This ensured that all iterations of the experiment were consistent, and
that the reliability of the experiment is increased. If the iterations of the experiment
lacked this consistency, it would be harder for others to replicate the experiment.
As previously mentioned, all experiments were recorded which allowed for richer data
analysis and enhanced validity. Since no data were left out and were as unbiased as
possible due to this, other researchers should be able to replicate this experiment, which
enhanced reliability.

42
3 Theoretical Framework
This chapter provides a deeper understanding of the ethics of persuasive design,
including dark patterns and bright patterns. Theories used in this study are also
presented in this chapter.

3.1 Types of Dark Patterns

As previously mentioned, a dark pattern is a type of persuasive design that sacrifices
consumers’ freedom of choice for the benefit of the service by confusing, deceiving, or
misleading them. In this section, a walkthrough of the six dark patterns that were used
in the experiment of this study is presented to further understand them.
Trick Questions
This type of dark pattern usually appears in online forms, often when registering with
a service. While filling in a form, a consumer might be presented with a question that
tricks them into giving an unintended answer, meaning that the form asks another thing
than the consumer initially thought. Altered order of checkboxes and confusing
language are other factors that are included in this dark pattern. Figure 15 shows an
example of this dark pattern with contradicting and confusing checkbox options.

Figure 15: Note. An illustration of the “Trick questions” dark pattern. From Brignull,
H. (2017, September 19). Trick questions. Dark Patterns.
https://www.darkpatterns.org/types-of-dark-pattern/trick-questions

43
Sneak into Basket
When a consumer wants to purchase something on an e-commerce website, they will
add one or multiple products to their shopping cart until they are satisfied and then
proceed to checkout. Occasionally, some websites might sneak in an additional product
into the consumer’s shopping cart, which consumers do not realize until they proceed
to checkout. Figure 16 shows an example of this dark pattern by sneaking in a Sports
Direct magazine to the consumer’s shopping cart. This creates friction that disrupts the
user experience since the consumer will have to remove the magazine from his or her
shopping cart.

Figure 16: Note. An illustration of the “Sneak into Basket” dark pattern. From
Wintermeier, N. (2020b, June 22). Dark Patterns Examples in eCommerce: What they
are & why to avoid them. Crobox. https://blog.crobox.com/article/dark-patterns
Roach Motel
Roach Motel is a dark pattern that a consumer might encounter when they easily get
into a situation on a website or an app but find it very hard to get out of that same
situation. This dark pattern is very common regarding online subscriptions or accounts.
It is very easy to create an account or subscribe, but in contrast, it is often very difficult
to cancel the subscription or terminate the account. A popular example of this dark
pattern is the task of closing an Amazon account, as showcased in Figure 17.

44
Figure 17: Note. An illustration of the “Roach Motel” dark pattern. From Wintermeier,
N. (2020b, June 22). Dark Patterns Examples in eCommerce: What they are & why to
avoid them. Crobox. https://blog.crobox.com/article/dark-patterns
Price Comparison Prevention
This dark pattern occurs when a retailer complicates the consumer’s process of
comparing the price of different items, rejecting consumers the ability to make an
informed and rational decision. Figure 18 shows an example of this dark pattern since
the first two products are essentially the same, but the website makes it hard for
consumers to compare the price of the two products.

Figure 18: Note. An illustration of the “Price Comparison Prevention” dark pattern.
From Brignull, H. (2018, June 6). Price comparison prevention. Dark Patterns.
https://www.darkpatterns.org/types-of-dark-pattern/price-comparison-prevention

45
Misdirection
Misdirection is a type of dark pattern that works by incorporating design elements that
purposefully grab the consumer’s attention to a certain part of the user interface to
distract him or her from another part of the user interface. This can be done using
different font sizes and colours. Figure 19 shows a consumer trying to cancel her
account/subscription. The “never mind” option is highlighted by using strong colour
and capital letters, while in contrast, the “continue cancellation” option is discrete.

Figure 19: Note. An illustration of the “Misdirection” dark pattern. From Dunbar, J.
[@JessicaDunbar]. (2019, June 2). Guess I didn’t cancel my @ipsy membership last
month. I hope the $10 was worth it. I was planning to take a break for the summer, now
I’ll never subscribe to ipsy again. #Ux #design #unethical #ifellforit #ipsy cc/
@darkpatterns [Tweet]. Twitter.
https://twitter.com/JessicaDunbar/status/1135219427056738309?fbclid=IwAR23gPlQ
1LW2wW0e78AkEHeRcxUKk8Q8w5v-yh2mm8_mGD5aAW9N37fdMqM
False Urgency Scarcity
False Urgency Scarcity refers to a service using nudges to notify the consumer that the
product they are interested in is in high demand. This exposes the cognitive biases that
humans are attracted to and respond to things that are scarce. This dark pattern can
present itself in the form of notifying that there is only x number of products left in
stock, x amount of people is looking at the product at the same time as the consumer,
and that it has been bought by x number of people the very same day. Countdown timers
and limited-time messages are other forms of pressure tactics commonly appearing with
this dark pattern (Mathur et al., 2019).
This is not particularly a dark pattern if the service has low stock or that a product is in
high demand, however, if the pressure tactics and warnings that are shown to the
consumer are untruthful, then it becomes a dark pattern. Booking.com is found guilty
of using such pressure tactics, as shown in Figure 20.

46
Figure 20: Note. Fake Urgency Scarcity dark pattern. From Wintermeier, N. (2020b,
June 22). Dark Patterns Examples in eCommerce: What they are & why to avoid them.
Crobox. https://blog.crobox.com/article/dark-patterns

3.2 Ethics
Academics have recognized ethics as an important concern in persuasive design for two
decades with regard to Berdichevsky and Neuenschwander and Fogg who published
the first academic work on ethics in persuasive technology (Kight & Gram-Hansen,
2019).

3.2.1 Definition of Ethics

Ethics is a term that is subjective. A designer and consumer can have different views
on what is ethical, making it important to define a concurrent view of ethics in this
study. It is considered unethical when designers exploit knowledge of human
psychology and behavior to neglect consumer benefits (Karagoel & Nathan-Roberts,
2021), coerce or deceive consumers, target vulnerable groups like children and those
with intellectual disabilities, and obstruct consumers’ right to freedom of choice (Kight
& Gram-Hansen, 2019). However, it is considered ethical when designers preserve
consumers’ freedom of choice (Wintermeier, 2020a).

3.2.2 Bright Patterns

Bright patterns are the counterpart of dark patterns (Graßl et al., 2021). As previously
mentioned in section 1.1 - Background, dark patterns manipulate through persuasion
techniques consumer decisions to desired outcomes, like increased revenue, that benefit
the company. Bright patterns persuade rather than manipulate consumers to make
decisions that benefit themselves, and not only the company. Graßl et al. (2021) found
out that two out of three design nudges (default and obstruction), which may persuade
or manipulate depending on how it is used, influence consumers’ privacy choices.
Consumers were nudged to either select the privacy-unfriendly option (“agree”) or the
privacy-friendly option (“do not agree”) and had the possibility (in default and aesthetic
design nudges) to select another option (“manage options”). Nudges towards the

47
privacy-friendly option were called bright patterns, whereas nudges towards the
privacy-unfriendly option were called dark patterns.
Graßl et al. (2021) assume that nudges that direct consumers towards the privacy-
unfriendly option would only benefit the company. It is debatable whether the privacy-
unfriendly option would not be in the interest of consumers. Arguably, consumers
benefit from giving companies consent to manage their private data. Giving consent
enables companies to save consumers’ private data such as street addresses and card
information to reduce their time and effort to checkout in e-commerce for future
purchases. Bright patterns are not only beneficial for consumers, but also for
companies.

3.3 Theories
In this section, the theories used in this study are presented.

3.3.1 Dual Process Theory

The American Psychological Association (APA) Dictionary of Psychology defines the

Dual Process Theory (DPT) as ”a decision-making theory stating that judgment and
reasoning involve two separate processes: intuitive decision making and rational
decision making”. Kahneman (2011) provides a further interpretation of the DPT and
has managed to popularize it in his highly acclaimed book “Thinking, Fast and Slow”.
In the book, Kahneman discusses the psychology behind heuristics, more specifically
how humans use Systems 1 and 2 in their decision-making process, as shown in Figure
21. System 1 represents the intuitive decision-making process that occurs in the brain,
it operates quickly and automatically on a subconscious level. System 2 on the other
hand, represents the rational decision-making process that occurs in the brain, it
operates on a much slower, controlled, and very conscious level. For example, System
1 is activated in the form of flinching or ducking when an object is thrown at us, while
System 2 fires up when we try to solve a mathematical equation. System 1 operates on
cognitive biases to facilitate quick decisions, and persuasive design works by making
use of these cognitive biases (Wintermeier, 2020a). However, persuasive design can
easily turn into dark patterns when it exploits these cognitive biases and uses them to
the consumer’s disadvantage (Wintermeier, 2020b). The DPT and its interpretation by
Kahneman aid in understanding persuasive design and distinguishing bright patterns
from dark patterns, which proves useful when developing the prototypes that are used
during the experimentation part of this study.

48
Figure 21: Note. A visual representation of the DPT. From Gorbatenko, E. (2020, March
8). Trading: fast and slow. Online Forex Broker.
https://fbs.eu/en/analytics/articles/trading-fast-and-slow-26041

3.3.2 Cognitive Biases

Dark patterns exploit consumers’ cognitive biases, which are mental shortcuts that
distort perception, interpretation, and judgment, to influence them into making intuitive
or irrational decisions (Kim et al., 2021). Kim et al. (2021) identified dark patterns
affecting consumers’ cognitive biases which are described in the following four
sections.
False Discounts Claims and Anchoring Bias
False discount, a dark pattern tactic for displaying inaccurate pricing information about
products or services, is employed to mislead consumers into making incorrect
assumptions about the pricing of products or services. It influences a common cognitive
bias referred to as price anchoring that can influence consumers’ purchase decisions.
Price anchoring refers to consumers using initial information to make subsequent
decisions. Subconsciously, consumers tend to anchor the original pricing information,
which might be a number that is intentionally set high, in mind when it is displayed
next to the discounted price. Consumers are misled into assuming that they can get a
product or service with a high discount and therefore make the decision to buy the
product or service.
Hidden Cost and Sunk Cost Fallacy
As previously mentioned in 3.1 - Types of Dark Patterns, hidden cost is a dark pattern
tactic employed to sneakily charge high costs to consumers. Hidden costs are only
revealed to consumers during the purchase completion, such as taxes and other fees.

49
Consumers may disregard the additional costs and go ahead with the purchase: they
have fallen into the “sunk cost fallacy” trap. Sunk cost fallacy is the cognitive bias that
occurs when consumers have invested resources such as money, emotion, effort, or time
in making the purchase to the extent that they cannot discard the purchase.
Activity Notification and Bandwagon Effect
Activity notification refers to a dark pattern tactic that notifies consumers about online
activities of other consumers to get their attention to products or services and accelerate
their purchase decisions. For example, consumers could be informed about popular
products or services among others or products or services that others have bought.
Activity notification influences the cognitive bias called bandwagon effect which
influences consumers to buy products or services that others are purchasing.
High-Demand and Low-Supply Messages and Scarcity Bias
High-demand and low-supply messages are dark pattern tactics employed to rush
consumers into buying products or services. Consumers feel the need to act fast to
complete a purchase, otherwise, they will miss out on valuable products or services.
Scarcity bias is the cognitive bias that influences consumers into believing products or
services to be valuable. Four types of messages that influence scarcity bias are
countdown timer message, limited-time message, low-stock message, and high-demand
message. Countdown timer message indicates that a deal or discount will expire soon,
but it remains even when the time runs out. Limited-time message indicates that a deal
or sale will close soon without specifying a deadline. The low-stock message and high-
demand message are described in section 3.1 - Types of Dark Patterns.

50
 4 Results and Analysis
In this chapter, the findings from the semi-structured interviews with UX designers are
presented. Additionally, the findings from the experiment and the follow-up structured
interviews are presented. To conclude the chapter, a summary is presented to answer
the research questions of this study.

4.1 Results of Interviews With UX Designers

The findings of the semi-structured interviews are visualized in Table 1 presenting a
coding scheme of quantified interview responses grouped into categories
and subcategories. A coding scheme was created to extract relevant interview responses
that answer the first group of research questions:
RQ1a: Who has the right to determine how designers should address ethics
in the designs at their workplace?

RQ1b: What are the ethical issues that designers encounter in their designs
at their workplace?

RQ1c: How do designers approach ethical issues that occur in their designs
at their workplace?

Some interview responses were not grouped into any of the four categories and their
subcategories shown in Table 1. Although these are not presented in Table 1, they are
introduced in the following four sections for the identified categories: ethical approach,
ethical issues, countermeasure for ethical issues, and authority level. At first, there were
six categories that were identified, but only four categories were relevant to answer the
research questions. Ethical approach answers RQ1c, authority level answers RQ1a, and
ethical issues and countermeasure for ethical issues answer RQ1b.
Table 1: A coding scheme presenting the responses from the semi-structured interviews
coded into four categories and then re-coded into subcategories to quantify the
occurrences of similar responses within a subcategory.

Category Subcategory Responses Occurrence

Ethical approach User-centric process “... getting user feedback...to really 3

have a user-centric process and make
sure you include people.”
“I think you have to know who you’re
designing for...”
“... user testing to see if you guide
someone in the wrong direction.”

51
 Good design intention “... try to do things good from the 2
beginning.””
“... use my powers for good to move
those kinds of barriers...”

Guidelines/ “… we have several guidelines...we 3

have like a checklist...”
References
“... GDPR helps us quite a bit...”
“......we have the accessibility
(WCAG)”

Persuasion “... I try to work as much as I can with 2

visuals, like a storytelling."
“I can find ways to point out and make
it hurt for those people who build it...”

Reflection “I think we need to beware as designers 2

that whatever we do has an impact on a
person... I would try to persuade and
ask questions...”
“I always think about consequences...I
always think of the whole picture...I
always raise the questions...”

Ethical issues Design tradeoffs “... trade-off of how inclusive you want 2
to be versus other things.”
“...you can design friction... makes
everyone else super annoyed.”

Excessive data collection “... it might not be necessary for the 2

client to collect all information...”
“... everybody wants more data more
data...”

Ethical grey zone “... there’s always a thin line when it’s 2
considered non-ethical.”
“… it’s about getting carried away and
not really realizing that what we’re
doing is not an ethical choice.”

Tunnel vision “... when we get rid of the stakeholders 3

that basically holding us back from

52
 products that will appeal to more
clients."
“... they want to sell more or have
higher scores or things like that, but not
seeing like the whole picture."
““Oh this number went up”, it’s like
“Yeah but did we make the product
better? Did we make things for users
better?””

Countermeasure Job termination “I felt really bad, so I ended my work 3

for ethical issues there.”
“I actually decided to leave the
company...”
“... I would do that and then look for
another job if they didn’t listen to me."

Question “... focus on “why do we need to bring 2

in this feature when we could focus
perhaps on a different segment that
would not need this feature?””
“...you can always try to tell them why
this is unethical or this is wrong.”

Authority level Group authority “I’m not alone but I think I have a 2
voice...”
“So it’s not entirely in my control, my
power...”

4.1.1 Ethical Issues

Ethical issues is a category for grouping interview responses indicating ethical issues
within design that interviewees encounter(ed) in their workplace. The interview
responses were further grouped into four subcategories: design tradeoffs, excessive data
collection, ethical grey zone, and tunnel vision.
Design tradeoffs refer to a situation that involves giving up or diminishing something
in return for making a certain design choice related to ethics. One participant encounters
a tradeoff when making design choices regarding accessibility while another participant
finds that designing friction diminishes the user experience.

53
Excessive data collection refers to a situation where companies or clients collect a large
amount of personal data and only utilize some of the collected data. Two participants
contemplated whether it was necessary for their company or client to collect the amount
of personal data that was collected then.
Ethical grey zone refers to a situation where designers happen to make unethical design
choices, even if they did not have the intention to do it. One participant finds it difficult
to determine whether something is ethical or unethical, while another participant gets
carried away with making design choices that are feasible and not realizing that these
are unethical choices.
Tunnel vision refers to parties involved in a product development, impeding a product
from delivering good user experience. One participant expressed that clients demand
features that are less user-focused while another participant stated that product owners
have the goal to sell more products rather than deliver products of good experience.
Other responses that were not categorized include ethical issues that are not mutual
between the interviewees. One participant faces two ethical issues, and these are that
bias makes it difficult to gather user feedback, and rules and laws lead to ethical
challenges. According to a participant, it becomes difficult to design an ethical product
where the motivation for using the product is to earn money, which may lead to users
becoming addicted. As stated by a participant, the participant can consider a design
choice to be unethical but someone else may have the opposite opinion.

4.1.2 Countermeasure for Ethical Issues

Countermeasure for ethical issues is a category for grouping interview responses

indicating designer’s reactions to and handling of ethical issues that arise in their
workplace. The interview responses were further grouped into two subcategories: job
termination and question.
Job termination refers to a situation where designers quit their job due to being involved
or nearly involved in developing unethical products. Upon launching unethical
products, two participants felt guilty and quit their job. One participant was not
involved in developing unethical products but reported that working on unethical
products would result in quitting the job.
Question refers to a situation where designers find a design choice to be unethical and
raise the issue. One participant raises issues by asking questions, while another
participant does it by explaining why design choices are unethical and wrong.
Other responses that were not categorized include countermeasures for ethical issues
that are not mutual between the interviewees. One participant tries to guide clients into
making ethical choices, while another participant avoids working for companies that
contribute to a negative impact on the public welfare.

54
4.1.3 Ethical Approach

Ethical approach is a category for grouping interview responses indicating the actions
that designers take to approach design in an ethical way. The interview responses were
further grouped into five subcategories: user-centric process, good design intention,
guidelines/references, persuasion, and reflection.
User-centric process refers to a user-focused mindset, where users are always in focus
when designing products. One participant claims that it is important to know the target
group while two other participants emphasize the importance of getting user feedback.
Good design intention refers to designers having a good intention when designing a
product. One participant strives to always design with good intention, while another
participant focuses on removing accessibility barriers.
Guidelines/references refers to guidelines in various forms such as GDPR, WCAG, and
company checklists. One participant has a checklist of things to consider when
designing a product, while another participant utilizes GDPR. A third participant
utilizes WCAG to design with accessibility in mind.
Persuasion refer to various approaches that designers take to persuade stakeholders into
making ethical choices. One participant persuades by using storytelling while another
participant utilizes empathy.
Reflection refers to critical thinking about the impact of design choices. One participant
reflects on the impact design choices have on people, while another participant thinks
about people and other aspects.
Other responses that were not categorized include ethical approaches that are not mutual
between the interviewees. As reported by a participant, an ethical approach is to not
work for certain companies that are believed to be unethical. One ethical approach is to
have the goal of keeping loyal and happy customers, according to a participant. A
participant stated that seeking other people’s opinions on design choices is helpful in
making ethical products, compared to another participant that learns from others and
stays updated with the latest knowledge about ethical design.

4.1.4 Authority Level

Authority level is a category that refers to interview responses indicating the individual,
group or organization making the final decision to addressing ethics. One subcategory
that was identified is group authority.
Group authority refers to whom the decisional power is distributed to. Ethical decision
is not distributed to one individual but enhanced in a group of individuals, according to
two participants.

55
Other responses that were not categorized include authority levels that are not mutual
between interviewees. As reported by one participant, if a product is requested by a
certain market, some salesperson might have a higher authority level and delegate
requirements to designers that influence the design and features of a product. Another
participant reported that authority can be given to another party such as the company’s
legal department before making a final decision regarding ethics. One participant
reported that she felt that designers had very much influence on what they decide, whilst
another participant reported that she felt that developers and architects are very
dominant and difficult to convince when it comes to making decisions.

4.2 Results of Experiment With Follow-up Interview

The findings from the experiment with follow-up structured interview aim to answer
the second group of research questions. To offer some recollection, the second group
of research questions is presented below.
RQ2a: How do bright patterns (freedom of choice and transparency) in
comparison to dark patterns (lack of choice and non-transparency)
influence consumers’ decisions?

RQ2b: What is the difference between consumers’ perceptions of ethics

(freedom of choice and transparency) in bright patterns and dark patterns?
As aforementioned, the experiment itself aims to answer specifically RQ2a whilst the
follow-up structured interviews aim to answer RQ2b. The consumer decisions related
to the tasks of the experiment have been quantified in several graphs and are visualized
in 4.2.1 - Consumer decisions.
The consumer’s perceptions of ethics in bright and dark patterns have been analyzed
from the follow-up structured interviews and coded into several categories. Code
examples within the different categories have been quantified into several graphs and
are visualized from 4.2.2 - Perceived freedom of choice to 4.2.6 - Other categories,
where each section represents a category that has been identified during the analysis
process.

4.2.1 Consumer Decisions

This section presents the findings of the experiment in relation to consumer decisions.
To see a detailed description of how the two prototypes differ in relation to the tasks,
see section 2.2 - Artifact Creation.
Task 1
In Task 1, participants were instructed to select three white t-shirts with the lowest price
among other t-shirts and add them to their shopping bag. Figure 22 below shows the
difference in consumer decisions between bright and dark patterns for Task 1.

56
Figure 22: Consumer decisions for Task 1 where participants are instructed to select
three white t-shirts with the lowest price among other t-shirts and add them to their
shopping bag. The blue bars represent the number of participants that chose a) a 3-
Pack of t-shirts or b) three individual t-shirts in Prototype B (bright patterns). The
orange bars represent the number of participants that chose a) a 3-Pack of t-shirts or b)
three individual t-shirts in Prototype A (dark patterns).

The 3-Pack of t-shirts was the cheapest option in both prototypes, and thereby the
“correct” option to choose in this task. In the bright patterns prototype (Prototype B),
the 3-Pack listed the price per t-shirt, which made the price comparison simple for the
two available options. In the dark patterns prototype (Prototype A), the 3-Pack did not
list the price per t-shirt, and the two available options were closer in terms of price
which made the price comparison a bit more complicated. Participants reported that the
reasons why they chose the three individual t-shirts instead of the 3-Pack was that they
mistook the 3-Pack for an individual t-shirt or that they simply did not see the 3-Pack.
Task 2
In Task 2, participants were instructed to select a pair of jeans in the colour medium
blue and add it to their shopping bag. Figure 23 and Figure 24 below show the difference
in consumer decisions between bright and dark patterns for Task 2.

57
Figure 23: Consumer decisions for Task 2 (Men’s) where participants are instructed to
select a pair of jeans in the colour medium blue and add it to their shopping bag. The
blue bars represent the number of participants that chose a) slim fit jeans or b) ripped
jeans in Prototype B (bright patterns). The orange bars represent the number of
participants that chose a) slim fit jeans or b) ripped jeans in Prototype A (dark
patterns).

Figure 24: Consumer decisions for Task 2 (Women’s) where participants are
instructed to select a pair of jeans in the colour medium blue and add it to their
shopping bag. The blue bars represent the number of participants that chose a)
straight jeans or b) mom jeans in Prototype B (bright patterns). The orange bars
represent the number of participants that chose a) straight jeans or b) mom jeans in
Prototype A (dark patterns).

58
For the men’s jeans, the ripped jeans had a false urgency scarcity message on them in
Prototype A, and the same goes for the mom jeans regarding the women’s jeans. Most
participants reported that they chose jeans based on their personal preference and taste.
Task 3
In Task 3, participants were instructed to proceed to checkout and pay for the items in
their shopping bag. In Prototype A, participants were met with a sneaked-in product in
their shopping bag, as well as a newsletter pop-up. In Prototype B, participants were
met with an item recommendation pop-up, as well as a newsletter pop-up. Figure 25 and
Figure 26 show the difference in consumer decisions between bright and dark patterns
for the extra item included in Task 3, whilst Figure 27 shows the difference in consumer
decisions between bright and dark patterns for the newsletter included in Task 3.

Figure 25: Consumer decisions for Task 3 where participants are instructed to proceed
to checkout to pay for the items in their shopping bag. In Prototype A, a product has
been sneaked into their shopping bag. The orange bars represent the number of
participants that a) kept the sneaked-in product or b) removed the sneaked-in product
from their shopping bag.

59
Figure 26: Consumer decisions for Task 3 where participants are instructed to proceed
to checkout to pay for the items in their shopping bag. In Prototype B, a pop-up
recommends a clothing item to participants when they are in their shopping bag. The
blue bars represent the number of participants that a) added the recommended item to
their shopping bag or b) declined the recommended item.

Figure 27: Consumer decisions for Task 3 where participants are instructed to proceed
to checkout to pay for the items in their shopping bag. During the checkout process,
participants are met with a newsletter pop-up. The blue bars represent the number of
participants that a) subscribed to the newsletter or b) declined in Prototype B (bright
patterns). The orange bars represent the number of participants that a) subscribed to
the newsletter or b) declined in Prototype A (dark patterns).

60
In Prototype A, most participants removed the sneaked-in product from their shopping
bag. The reasons why some participants kept the sneaked-in product in their shopping
bag include that they either did not notice the sneaked-in product, or that they thought
it was a mistake from the website. In Prototype B, most participants declined the
recommended item. Those participants who added the recommended item to their
shopping bag reported that it was a pleasant recommendation.
Almost all participants declined the newsletter in both prototypes due to their reported
preconceived disinterest or even disliking of newsletters, however one participant was
intrigued enough to subscribe to the newsletter in both prototypes due to the possibility
of receiving discounts. The vast majority of those that declined the newsletter in
Prototype B closed it down with the cross instead of choosing the decline option.
Task 4
In Task 4, participants were instructed to delete their account. In order to do that, they
must first find their way to the “Delete account” page, and once they have found the
page, they can simply delete their account. In Prototype A, the way to the “Delete
account” page was far more complicated than the one in Prototype B. Prototype B has
a big red “Delete button” in the “Delete account” page whilst Prototype A has the
“Never mind” option as the big red button and a smaller, more hidden “Delete” button
next to it. Figure 28 shows the difference in consumer decisions between bright and
dark patterns for Task 4.

Figure 28: Consumer decisions for Task 4 where participants are instructed to delete
their account. The blue bars represent the number of participants that a) easily found
the “Delete account” page or b) instantly clicked the “Delete” button in Prototype B
(bright patterns). The orange bars represent the number of participants that a) easily
found the “Delete account” page, b) were stuck or struggled to find the “Delete

61
account” page, c) instantly clicked the “Delete” button, or d) hesitated or misclicked
the “Delete” button in Prototype A (dark patterns).

In Prototype A, some participants found their way to the “Delete account” page with
ease whilst some struggled or even got stuck and needed hints from the researchers to
continue. In contrast, all participants found their way to the “Delete account” page with
ease in Prototype B. Some participants instantly clicked on the discrete “Delete” button
in Prototype A whilst some did it with some hesitation or even misclicked on the “Never
mind” option. In Prototype B however, all participants instantly clicked on the big red
“Delete” button without hesitation.
Preferred Prototype
When all tasks were completed, participants were asked which prototype they preferred.
Figure 29 below shows the unanimous decision about which prototype participants
reported as their preferred prototype between Prototype A and Prototype B.

Figure 29: Participants’ preferred prototype between Prototype A and Prototype B.

The blue bar represents the number of participants that preferred Prototype B (bright
patterns), which was the preferred prototype for all participants.

4.2.2 Perceived Freedom of Choice

Perceived freedom of choice is a category that encapsulates to what extent participants

feel that they a) have the freedom to make their own decisions, b) are limited in the
ability to make their own decisions, and c) have no freedom to make their own
decisions.
Figure 30 and Figure 31 below show the difference in consumers’ perceived freedom
of choice between dark patterns and bright patterns.

62
Figure 30: Consumers’ perceived freedom of choice in Prototype A related to specific
dark patterns. The blue bar represents the number of participants reporting freedom
regarding Trick Questions. The orange bars represent the number of participants
reporting no freedom or limited freedom regarding a) Sneak into Basket and b) Trick
Questions.

Figure 31: Consumers’ perceived freedom of choice in Prototype B related to specific

bright patterns. The blue bars represent participants reporting freedom regarding a)
Recommended item and b) Newsletter. The orange bars represent the number of
participants reporting no freedom or limited freedom regarding a) Recommended item
and b) Newsletter.

First and foremost, not all dark and bright patterns that have been implemented in both
prototypes were reflected upon in terms of freedom of choice by participants. The
reason for this is that not all dark and bright patterns are related to freedom of choice,

63
some are related to transparency instead. Hence, the dark and bright patterns depicted
in Figure 30 and Figure 31 are those that participants reported a strong opinion about
regarding freedom of choice. Secondly, not all participants reported an opinion
regarding the freedom of choice of a specific bright or dark pattern. This means that
these participants did not adopt a standpoint towards a specific bright or dark pattern,
but rather towards the whole prototype(s) and experience.
The two dark patterns that participants reported a strong opinion about regarding
freedom of choice are Sneak into Basket and Trick Questions, with their respective
bright pattern counterparts.
Extra Item
In Prototype A, most participants reported having no freedom or limited freedom of
choice when faced with a sneaked-in product in their shopping bag in Task 3.
In Prototype B, most participants reported having the freedom to make their own
decisions when faced with the recommended item pop-up. However, some participants
reported not having freedom or having limited freedom of choice regarding the
recommended item, which they motivated by stating that the item recommendation felt
pushy.
Newsletter
In Prototype A, some participants reported having the freedom to make their own
decisions when faced with the poorly worded newsletter, whilst others reported having
no freedom or limited freedom to make their own decisions.
In Prototype B, most participants reported having the freedom to make their own
decisions when faced with the better worded newsletter.
Prototype With Highest Freedom of Choice
When all tasks were completed, participants were asked which prototype they found to
give them the most freedom of choice. Figure 32 below shows the unanimous decision
about which prototype participants reported as providing the highest freedom of choice
between Prototype A and Prototype B.

64
Figure 32: Participants’ report of which prototype they found to provide them with the
highest freedom of choice between Prototype A and Prototype B. The blue bar
represents the number of participants that reported Prototype B (bright patterns) to
provide the highest freedom of choice, which was all participants.

4.2.3 Perceived Transparency

Perceived transparency is a category that encapsulates to what extent participants feel

that a) the information presented in the prototype is clear, comprehensive, and/or
complete, b) the information presented in the prototype is unclear, incomprehensible,
incomplete, and/or confusing.
Figure 33 and Figure 34 below show the difference in consumers’ perceived
transparency between dark patterns and bright patterns.

65
Figure 33: Consumers' perceived transparency in Prototype A related to specific dark
patterns. The blue bars represent participants reporting high transparency regarding a)
Trick Questions, b) Price Comparison Prevention, and c) Roach Motel. The orange
bars represent the number of participants reporting low transparency regarding a)
Sneak into Basket, b) Trick Questions, c) Price Comparison Prevention, d) Roach
Motel, and e) Misdirection.

Figure 34: Consumers’ perceived transparency in Prototype B related to specific

bright patterns. The blue bars represent the number of participants reporting high
transparency regarding a) Recommended item, b) Newsletter, c) Price Comparison, d)
“Roach Motel”, and e) “Misdirection”. The orange bar represents the number of
participants reporting low transparency regarding Newsletter.

66
Just like perceived freedom of choice, not all dark and bright patterns that have been
implemented in both prototypes are reflected upon in terms of transparency by
participants. The reason for this is that not all dark and bright patterns are strongly
related to transparency, some are related to freedom of choice instead. Hence, the dark
and bright patterns depicted in Figure 33 and Figure 34 are those that participants
reported a strong opinion about transparency. Once more, not all participants reported
an opinion regarding the transparency of a specific bright or dark pattern. This means
that these participants did not adopt a standpoint towards a specific bright or dark
pattern, but rather towards the whole prototype(s) and experience.
Extra Item
In Prototype A, two participants reported that the sneaked-in product in Task 3 reflected
low transparency from the website. These participants explicitly shared the opinion that
the website was unclear to them in that specific situation.
In Prototype B, two participants reported that the item recommendation in Task 3
reflected high transparency from the website. These participants expressed that the
website was clear to them in that specific situation.
Newsletter
For the newsletter in Task 3, participants were almost equally divided when it came to
the level of transparency provided by Prototype A. Those that reported the newsletter
provided high transparency motivated their reasoning by stating that it was clear and
intuitive enough, whilst those that reported the newsletter provided low transparency
motivated their reasoning by stating that the two options of the newsletter were too
similar in their wording and thereby unclear.
In Prototype B, most participants reported that the newsletter reflected high
transparency from the website, whilst a few reported low transparency. Those that
reported the newsletter provided high transparency motivated their reasoning by stating
that it was intuitive enough to know what they had to do to proceed. Those that reported
the newsletter provided low transparency suggested that only the cross in the upper
right corner was intuitive, whilst the checkbox and “Continue” button were unclear.
Price Comparison of T-Shirts
In Prototype A, participants were equally divided when it came to the level of
transparency of the price comparison in Task 1. Those that reported the price
comparison provided high transparency stated that it was simple to calculate which t-
shirt was the cheapest option. Those that reported the price comparison provided low
transparency stated that it was unclear since they had to do the math themselves and
that the website did not indicate the price per item.

67
In Prototype B, most participants reported that the price comparison reflected high
transparency from the website. They expressed that having the price per item listed for
the 3-Pack made the price comparison very easy and enhanced clarity.
Finding the “Delete account” Page
Regarding finding the “Delete account” page in Task 4, most participants reported that
the website reflected low transparency in Prototype A, whilst a few reported high
transparency. Those that reported low transparency in Prototype A motivated their
reasoning by stating that the way to the “Delete account” page was unnecessarily
complicated, based on the links being unclear and misplaced. Those that reported high
transparency in Prototype A stated that the path was similar to other websites that they
have previously used.
In Prototype B, most participants reported that the website provided high transparency
regarding the path to the “Delete account” page. They stated that the links were clear,
and that the information was in the right place, which made it very easy to find the page.
However, some participants felt that it was almost too easy.
Deleting the Account
In Prototype A, two participants reported that when they had found their way to the
“Delete account” page, they felt that the actual deletion of the account reflected low
transparency. They stated that they were more used to the “Delete” button being the big
red one, so when the “Never mind” button was the big red one instead, those participants
expressed it would be easy to make the wrong choice.
In Prototype B, most participants reported that the actual deletion of the account
reflected high transparency. They motivated their reasoning by stating that they were
used to the “Delete” button being the big red one, therefore they felt it was intuitive.
Prototype With Highest Transparency
Again, when all tasks were completed, participants were asked which prototype they
found to have the highest transparency. Figure 35 below shows the unanimous decision
about which prototype participants reported as having the highest transparency between
Prototype A and Prototype B.

68
Figure 35: Participants’ report of which prototype they found to have the highest
transparency between Prototype A and Prototype B. The blue bar represents the
number of participants that reported Prototype B (bright patterns) to provide the
highest transparency, which was all participants.

4.2.4 Effect of Dark Patterns

Effect of dark patterns is a category that refers to which extent a participant was affected
by a dark pattern and their emotion tied to it.
Figure 36 below shows the reported effects that certain dark patterns in Prototype A
had on participants.

Figure 36: The reported effects that certain dark patterns in Prototype A had on
participants. The blue bar represents the number of participants feeling a negative

69
emotion regarding Sneak into Basket. The orange bar represents the number of
participants feeling distracted regarding False Urgency Scarcity. The grey bar
represents the number of participants feeling misled regarding Misdirection. The
yellow bar represents the number of participants feeling confused regarding Trick
Questions.

Not all dark patterns were explicitly mentioned by participants, and not all participants
explicitly reported an effect that a certain dark pattern had on them. The dark patterns
depicted in Figure 36 represents the dark patterns that participants explicitly reported
some effect or emotion about.
Sneak into Basket
The dark pattern that had the highest reported effect on participants was Sneak into
Basket when there was a sneaked-in product in participants’ shopping bag during Task
3. Almost all participants reported feeling a negative emotion when faced with this dark
pattern. These negative emotions include feeling angry, frustrated, upset, annoyed, that
it did not feel good, and that it almost felt illegal.
False Urgency Scarcity
Most participants reported feeling distracted to some extent when faced with the False
Urgency Scarcity dark pattern during Task 2. To elaborate, participants reported that
the False Urgency Scarcity messages were attention-grabbing and that they were the
first things they noticed on the “Jeans” page. In addition to feeling distracted, some
participants reported feeling stressed, curious, and that the message was a bit pushy.
Misdirection
Approximately half of the participants reported feeling misled when faced with the
Misdirection dark pattern during Task 4. Participants reported feeling misled in the
sense that they mistook the big red “Never mind” button for the “Delete” button.
Trick Questions
Two participants reported feeling confused when faced with the Trick Questions dark
pattern during Task 3. These participants reported feeling confused in the sense that the
wording of the options to either accept or decline the newsletter was very similar whilst
simultaneously confusing.

4.2.5 Effect of Bright Patterns

Effect of bright pattern is a category that refers to which extent a participant was
affected by a bright pattern and their emotion tied to it.

70
Figure 37 below shows the reported effects that the bright pattern “Recommended item”
in Prototype B had on participants.

Figure 37: The reported effects that “Recommended item” in Prototype B had on
participants. In this case, participants only reported the effect that the bright pattern
Recommended item had on them. The blue bar represents the number of participants
feeling a negative emotion. The orange bar represents the number of participants
feeling a positive emotion.

Again, not all bright patterns were explicitly mentioned by participants, and not all
participants explicitly reported an effect that a certain bright pattern had on them. The
bright pattern depicted in Figure 37 represents the bright pattern that participants
explicitly reported some effect or emotion about.
Recommended Item
Most participants reported that the item recommendation pop-up during Task 3 affected
them. Five participants reported feeling a negative emotion whilst four participants
reported feeling a positive emotion. Those that reported feeling a negative emotion
stated that the item recommendation was a bit pushy and bothersome and that they felt
annoyed by it. Those that reported feeling a positive emotion stated that the item
recommendation was nice and that it was a good thing for an e-commerce website to
do.

71
4.2.6 Other Categories

Other categories were identified, although they are not as strongly connected to the
second group of research questions compared to the previous categories, but they still
explain participants’ motivations and thought processes.
Automatic Behavior
Automatic behavior is a category that encapsulates when a participant rapidly
performed an action based on instinct and/or habit, on a subconscious level. This
category is closely related to the DPT mentioned in 3.3.1 - Dual Process Theory.
Seven participants that closed or declined the newsletter pop-up and declined the
recommended item pop-up motivated their decision by stating that it is based on
instinct, it is a reflex for them to do that because that is what they always do.
When selecting a pair of jeans, five participants stated that they selected the first pair
that they saw, acting on instinct without giving it too much thought.
Suggestions for Persuasion
Suggestions for persuasion is a category that identifies participants’ suggestions for
how the prototype(s) could have successfully persuaded them.
Some participants that declined the recommended item in Protype B provided some
suggestions for how the website could have successfully persuaded them to add or
consider adding the recommended item to their shopping bag. Three participants
suggested that the recommended item could have been on sale. Three participants
suggested that the recommended item could have been something that they felt they
needed, such as underwear or socks. Five participants suggested that the recommended
item could have been more relevant to their style, i.e., that it was a similar style to the
other items in their shopping bag. Lastly, one participant suggested that other types of
items could have been recommended such as shoes and jackets rather than another t-
shirt.
One participant provided a suggestion for how the website could have successfully
persuaded them to subscribe to the newsletter. This participant suggested that if the
website offered a discount code for subscribing to the newsletter, he would have done
it.
Prior Experience
Prior experience is a category that refers to heuristics that participants have formed
based on their previous experience with other websites. Although this category is
similar to automatic behavior, it differs in the sense that it does not encapsulate actions.
This category was identified during the analysis process; however, the authors feel that
this category is not relevant to the research questions. The codes that have been

72
established in this category have already been answered in other categories, therefore,
the category Prior experience was dismissed.

4.3 Summary of Findings

In this section, a summary of the findings that answer the research questions is
presented.

4.3.1 Findings of the Interviews with UX designers

Authority level involves gathering a group to enhance the decisional power in

preventing unethical design choices to be implemented. This answers the research
question RQ1a.
Ethical issues involve giving up or diminish something in return for getting something
else, seeing companies and clients gathering personal data that are not utilized by them,
defining and realizing when to draw the line between ethical and unethical design, and
meeting internal and/or external stakeholders with a tunnel-vision. These are ethical
issues that designers encounter in their designs at their workplace, which answer the
research question RQ1b.
Ethical approach involves putting users in the center, having the intention to do good,
following guidelines or references, persuading internal and external stakeholders, and
reflecting on the impact of design choices in various contexts.
Countermeasure for ethical issues involve quitting the job due to having a bad
conscience for launching an unethical product or reluctance to work for companies that
produce these kinds of products. It also involves questioning issues that are believed to
be unethical.
Both ethical approach and countermeasure for ethical issues answer the research
question RQ1c.

4.3.2 Findings of the Experiment With Follow-up Interview

The paragraphs related to consumer decisions answer RQ2a while the paragraphs
related to perceived freedom of choice and transparency answer RQ2b.
Consumer Decisions
The findings of the experiment indicate that there was no significant difference between
bright patterns and dark patterns in terms of how they influenced consumer decisions,
except for the bright and dark patterns in Task 4.
The Roach Motel dark pattern heavily complicated the completion of Task 4 for half of
the participants in the sense that they struggled to find the “Delete account” page or
even got stuck and failed to find it. On the other hand, the bright pattern counterpart of

73
Roach Motel facilitated the completion of Task 4 for all participants since they all found
the “Delete account” page with ease.
The Misdirection dark pattern complicated the completion of Task 4 for a third of
participants in the sense that they hesitated to click the “Delete” button or clicked on
the big red “Never mind” button instead. However, the bright pattern counterpart of
Misdirection facilitated the completion of Task 4 for all participants since they all
instantly clicked on the “Delete” button.
Consequently, RQ2a can be answered. Based on these findings, it can be concluded that
there was no significant difference between most bright and dark patterns in terms of
how they influenced consumer decisions. However, there was a significant difference
in Task 4 where the dark patterns Roach Motel and Misdirection influenced the
consumers to make the wrong decisions to complete the task, while their bright pattern
counterparts influenced the consumers to make the right decisions to complete the task.
Perceived Freedom of Choice and Transparency
Regarding the dark patterns Sneak into Basket and Trick Questions participants faced
in Task 3, the findings indicate that most participants felt limited freedom or no freedom
of choice while some participants felt freedom of choice. On the other hand, almost all
participants felt freedom of choice when facing the bright patterns in Task 3.
Additionally, the findings indicate that almost all participants reported that the dark
pattern Roach Motel reflected low transparency in Task 4. Participants were equally
split regarding the level of transparency that was reflected when facing the dark pattern
Price Comparison Prevention in Task 1, and Trick Questions in Task 3. Moreover, few
participants reported that the dark patterns Sneak into Basket and Misdirection reflected
low transparency. On the other hand, most participants reported that the bright pattern
counterparts of Trick Questions, Price Comparison Prevention, Roach Motel, and
Misdirection reflected high transparency. Lastly, few participants reported that the
bright pattern counterpart of Sneak into Basket (Recommended Item) reflected high
transparency, while a few participants reported that the bright pattern counterpart of
Trick Questions reflected low transparency.
Overall, all participants concluded that they perceived Prototype B and its bright
patterns to provide both the highest freedom of choice and the highest transparency,
which answers RQ2b.

74
5 Discussion
This chapter provides answers to the first group of research questions presented in
section 1.3 – Purpose and Research Questions by discussing the findings of the semi-
structured interview with UX designers. Both the findings of the semi-structured
interviews and experiment with follow-up interview are discussed by referencing to
previous work presented in section 1.2 – Related Works and theories presented in
Chapter 3 – Theoretical Framework.
Additionally, this chapter discusses the implications of this study’s method and
implementation choices.

5.1 Results Discussion

This section provides discussions of the findings that answer the research questions
which in turn fulfill the purpose of this study.
As a reminder, the purpose of the study is twofold. Firstly, the purpose is to explore the
ethical dilemma in design that UX designers encounter in their workplace to understand
the reason behind the prevalence of dark patterns. Secondly, the purpose is to explore
consumers’ perception of ethics in bright patterns and dark patterns, and consumers’
decisions between bright patterns and dark patterns to determine whether bright patterns
are a potential ethical approach that designers can adopt in the future.

5.1.1 Authority Distribution in the Workplace of Designers

Kight and Gram-Hansen (2019) discuss that it can be the management that determines
the standards of ethics rather than the designers themselves. According to the results
presented in section 4.1 – Results of Interviews With UX Designers, the right to
determine how designers should address ethics in design at their workplace is not
delegated to one individual person. It showed that half of the participants have the right
to voice opinions, but they are not always heard by others involved in a project together
with them. Instead, they are required to make arduous efforts in talking with and
gathering others to enhance their authority in getting approval to follow through with
their decisions.
According to the results in section 4.1 – Results of Interviews With UX Designers, it
seems that designers must consider many things when designing an ethical product.
Often, designers are the intermediator between internal and external stakeholders. If
designers work for a company, they may talk to developers about whether design
choices are feasible or not. It may limit ethical design choices that can be implemented,
which can make it more difficult for designers to design ethical products. As stated in
the results, one participant had to make an effort in working with web developers to
find a middle ground for a design choice that is ethical and feasible. Moreover,
designers must think about the goals of the stakeholders, which can also make it

75
difficult to implement certain ethical design choices. It might explain why designers do
not have full authority to decide how ethics are addressed in design and must make an
effort in achieving temporary decisional power. The more stakeholders, the less
authority designers hold.

5.1.2 Ethical Issues in the Workplace of Designers

As far as we know, there is no research about whether designers encounter ethical issues
in design at their workplace. It is therefore not possible to discuss the findings that
answer the research question RQ1b by referencing to previous studies presented in
section 1.2 – Related Works.
According to the findings, designers encounter the four following ethical issues: design
tradeoffs, excessive data collection, ethical grey zone, and tunnel vision. The latter is
the most significant ethical issue among the participants, and the remainder of the
ethical issues are encountered equally by them. Most participants often find themselves
in a complicated situation where they sometimes must face shortsighted internal and/or
external stakeholders (tunnel vision). They struggle with making ethical design choices
when the stakeholders only focus on their goals and do not look beyond these goals.
Half of the participants have been in a situation where they must give up or diminish
something in return for getting something else (design tradeoffs), encounter companies
and clients that gather personal data that are not utilized by them (excessive data
collection), are oblivious to whether they are ethical or unethical as there is a thin line
when a design is considered ethical or unethical (ethical grey zone).
Often, profitable companies like e-commerce have the goal to increase revenue,
compared to non-profitable companies. There are also external stakeholders that invest
in profitable companies to achieve the same goal as the internal stakeholders. As stated
by one participant, stakeholders only look at the numbers and do not look at the impact
that a product has on consumers. Most designers work for profitable companies, and
this might explain why designers face shortsighted internal and external stakeholders.
It might also explain why tunnel vision is the most significant ethical issue for
designers.
Moreover, profitable and non-profitable companies may have more than one target user
for whom they design products for. It means that they must design products that can be
used by multiple target groups, which can lead to design tradeoffs. As explained by one
participant, design friction is implemented to hinder consumers with problematic
behavior to become addicted to gaming products, but it also makes other users
annoyed.
Designers’ encounters with companies or clients gathering excessive personal data
from users may have to do with another ethical issue, namely the tunnel vision. If
companies have the goal to get numbers up, they need to know more about their target

76
users to be able to design products that appeals to these users. Gathering a large amount
of personal data from users can help companies achieve their goals, but they might be
shortsighted and do not consider what and how much data are needed.
As stated in the results, designers can make unethical design choices without the
intention to do bad. It is an ethical issue that can be explained by the fact that ethics is
a term that is subjective: a designer can have views on ethics that are different for
another. Ethics can be said to have a blurry definition, where there is a thin line between
when a design is considered ethical and unethical. This might explain why designers
can sometimes be unethical.

5.1.3 Ethical Approaches in the Workplace of Designers

According to Kight and Gram-Hansen (2019), and Benner et al. (2021), there is no
research that has investigated how designers address ethics in their workplace. RQ1c
aims to answer this. The findings revealed five ethical approaches that designers adopt
to stay ethical and/or deal with ethical issues at their workplace. Most participants stay
ethical by putting users in the center and/or following guidelines/references that may
include GDPR (General Data Protection Regulation), WCAG (Web Content
Accessibility Guidelines), and a company checklist. Moreover, half of the participants
stay ethical with one or more of the following approaches: good intention, persuasion,
and reflection. Half of the participants try to have the intention to do good, utilize
persuasion to persuade internal and external stakeholders to adopt ethical design
choices, and/or reflection to reflect on the impact of their design choices in various
contexts. Most participants approach ethical issues by quitting their jobs at companies
that produce unethical products, while half of the participants question issues that they
believe are unethical.
If the designers have an education degree, they have most likely learned to put users in
the center and consider laws regarding collections of personal data (GDPR) and
accessibility guidelines (WCAG). It might explain why most designers use these
approaches to stay ethical or design ethical products. Moreover, designers
may have learned about ethics or similar through their education, and this might be why
they try to design products with the intention to do good. It can also be that the
“intention to do good” is derived from the designer’s morals and beliefs. This can lead
designers to persuade stakeholders to adopt ethical design choices and reflect on the
consequences of their design choices. Although the following statement is presented as
numbers in the results, it implies that a participant that tries to have good intentions
would also try their best to persuade stakeholders and/or reflect on design choices. Good
intention can also explain why designers raise questions about ethical issues
Authority might explain why some designers approach ethical issues by quitting their
job. One explanation can be that designers feel neglected in the sense that their opinions

77
are not worth considering. If they are not given any chances to have a voice, they cannot
do anything other than quit the job. Although designers could try to make the company
pay attention to them, but this would require arduous effort. It would be easier for
designers to quit their job and find new ones. As presented in the results, one participant
would probably look for another job when the company decides to develop a product
that he or she claimed to be unethical.

5.1.4 Bright Patterns as an Ethical Approach

Graßl et al. (2021) proposed that designers can address ethical issues by adopting bright
patterns as an ethical approach. They investigated bright patterns in the context of
privacy cookie consent and found that more than one-third of the participants were
nudged to the privacy-friendly option in privacy cookie consent when the option was
preselected. Hence, they concluded that bright patterns are a promising ethical approach
that designers can adopt in the future to counteract dark patterns. Our study intended to
explore bright patterns beyond privacy cookie consent and compare multiple bright
patterns to their dark pattern counterparts. In our study, participants were positively
influenced by some bright patterns to make the right decisions to complete the tasks.
This is consistent with the findings of Graßl et al. (2021), supporting that bright patterns
are a promising ethical approach that designers can adopt in the future to counteract
dark patterns.

5.1.5 Influence on Consumer Decisions

As indicated by the findings of our experiment, there was no significant difference

between dark and bright patterns in terms of how they influenced consumer decisions.
This resonates with the findings of the study carried out by M. Bhoot et al. (2020),
where it was shown that the participants had become accustomed to dark patterns to the
point that they accepted them as being part of the regular internet experience. It also
resonates with the findings of the study carried out by Di Geronimo et al. (2020) where
participants argued that dark patterns had become so common, and a part of the normal
interaction flow when using applications and websites. These studies’ findings as well
as our study’s findings indicate that people have accepted that dark patterns are part of
the internet experience, hence they have learned how to work around dark patterns to
get the outcome that they desire. We believe this might account for the lack of
significant difference between dark and bright patterns in terms of how they influenced
the consumer decisions in our experiment, although it should be verified in future work.
Consequently, dark and bright patterns might not differ in terms of consumer decisions,
but as the findings of our experiment indicate, they differ significantly in terms of
consumers’ perceptions of ethics, as described in the following three paragraphs.

78
5.1.6 Persuade With Transparency

Previous works have argued that it is important to increase consumers’ awareness of

being persuaded through transparency (Benner et al., 2021; Karagoel & Nathan-
Roberts, 2021). The bright patterns that were implemented in Prototype B of our study
were aimed to be as transparent as possible regarding persuasion. Since all participants
reported that Prototype B (bright patterns) was the prototype with the highest
transparency, the findings of our study are consistent with the suggestions of previous
works regarding transparent persuasion.

5.1.7 Freedom of Choice

Karagoel and Nathan-Roberts (2021) claimed that it is considered unethical when

consumers’ decisions are forced, and not deliberated, in their oblivious state. In our
study, most participants reported feeling no freedom or limited freedom of choice
regarding certain dark patterns, whilst most participants reported feeling freedom of
choice regarding the certain dark patterns’ bright pattern counterparts. Overall, all
participants reported that Prototype B (bright patterns) was the prototype with the
highest freedom of choice. This shows that participants’ reported perception of ethics
in bright and dark patterns are consistent with the claims brought forward by Karagoel
and Nathan-Roberts (2021).

5.1.8 Consumers’ Distrust and Dislike of Dark Patterns

In the study conducted by M.Bhoot et al. (2020), participants claimed that they
distrusted websites that employed dark patterns and preferred not using them unless
necessary. In our study, all participants reported Prototype B (bright patterns) to be their
preferred prototype compared to Prototype A (dark patterns). These findings are
consistent with the findings from M.Bhoot et al. (2020).

5.1.9 Consumers and the Dual Process Theory

Kahneman (2011) proposed with the DPT that humans use System 1 and System 2 in
their decision-making processes. System 1 represents the intuitive decision-making
process while System 2 represents the rational decision-making process. Wintermeier
(2021a) stated that System 1 operates on cognitive biases to facilitate quick decisions,
and persuasive design works by making use of these cognitive biases. Additionally,
Wintermeier (2020b) stated that persuasive design easily can turn into dark patterns
when it exploits these cognitive biases and uses them to the consumer’s disadvantage.
The DPT was reflected in the findings since many participants reported making
decisions based on instinct when closing the newsletter or selecting jeans. This supports
the theory that persuasive design works by making use of System 1.

79
5.1.10 Dark Patterns Affecting Cognitive Biases

Kim et al. (2021) identified dark patterns affecting consumers’ cognitive biases, two of
them being the activity notification and bandwagon effect, as well as the high-demand
and low-supply messages and scarcity bias. Activity notification influences the
cognitive bias called bandwagon effect which influences the consumers to buy products
that others are purchasing. High-demand and low-supply messages influence
consumers to buy products from a place of urgency. Scarcity bias is a cognitive bias
that influences consumers into believing that a product is valuable. These previously
mentioned dark patterns affecting consumers’ cognitive biases are reflected in Task 2
of the experiment in Prototype A (dark patterns) in the false urgency scarcity messages.
Although a minor number of participants were affected by the false urgency scarcity
messages in Prototype A, it still supports the findings from Kim et al. (2021).

5.2 Method Discussion

In this section, the implications of this study’s method and implementation choices are
discussed.

5.2.1 Ecological Validity Implications

As aforementioned in section 2.4.2 – Experiment: Validity and Reliability, experiments

can provide a high level of control over several factors included in the research process
which improves the internal validity of the experiment. However, external validity is
hard to establish since experiments are carried out in a laboratory milieu rather than in
the field. This lack of external validity is referred to as low ecological validity.
Ecological validity is defined as “a measure of how test performance predicts behaviors
in real-world settings” (Gouvier, 2019).
If the experiment of this study was conducted on a real e-commerce clothing website
with real users spending their real, hard-earned money, the findings would have
indicated high ecological validity. This is because the findings would have reflected a
real-world setting, with users mindlessly browsing and ending up buying something.
However, it was not possible for us to conduct this experiment on a real e-commerce
clothing website. Instead, we had to be content with conducting the experiment on a
simulated e-commerce clothing website. This implies that the participants were aware
that this was an experiment, their own money was not at stake, there was a learning
curve on how to use the simulated website(s), and they had specific instructions to
follow. All these factors influence the findings, and thereby, low ecological validity is
indicated by the findings.
Consequently, within the scope of the study, we still feel that the prototypes efficiently
allowed the experiment and follow-up structured interview to provide some insightful
results.

80
5.2.2 Sample Size Implications

For the semi-structured interviews with UX designers, we were initially aiming to

recruit 5-10 participants, but we ended up with four participants. For the experiment
with follow-up structured interviews, we were initially aiming to recruit 15-20
participants, and we ended up with 15 participants. As previously mentioned in section
2.1.1 – Participant Recruitment: Screener Survey, we would have wished to recruit
more participants, but we have faced some obstacles along the way that has prohibited
us from doing so.
The type of participants that we aimed to recruit for the interview with UX designers
have very specific demographical requirements. These potential participants have a
specific occupation and are required to have at least one year of professional experience
working as a UX designer and being employed by a company. These potential
participants are very busy working full-time jobs. Some of them might not have time to
participate or might just simply not be interested in participating in a 40-minute
interview. When recruiting these participants, we began by sharing our screener survey
in Facebook community groups for UX designers where it failed to garner attention
from potential participants. We also shared the survey on LinkedIn in the hopes that it
would garner the attention of UX designers through our professional networks, but
ultimately, this was also an unsuccessful attempt.
The type of participants that we aimed to recruit for the experiment do not have any
specific demographic requirements other than being over 18 years of age, which made
it slightly easier to recruit this type of participants compared to the UX designers.
However, some of these potential participants are also busy working full-time jobs or
studying at a full-time rate. They also might not have the time or interest to participate
in a 30-minute experiment to carry out a handful of tasks and answer a couple of
questions. When recruiting these participants, we also shared this survey on LinkedIn
in the hopes that it would garner the attention of professional contacts such as
colleagues or fellow students, however, this was also an unsuccessful attempt.
As a result of this issue in terms of recruiting participants for both parts of this study,
we had to reassess and change our recruitment strategy. We began reaching out to
personal and professional contacts and recruited them as initial participants. Thereafter,
the snowball sampling method was adopted to recruit more participants predominantly
for the interviews with UX designers, but also for the experiment. Although the
snowball sampling method is, in theory, a great way to recruit more participants, it only
gets you so far. There is no guarantee that participants will know other potential
participants and recommend them. There is also no guarantee that in the case they do
know other potential participants, that they will reach out to these potential
participants.

81
Considering these obstacles that we have faced in the recruitment stages of this study
combined with the limited timeframe of this study, we have done our best to recruit as
many participants as possible. The small sample size of this study might lower its
validity and reliability. However, we feel that this study still provides some insightful
findings that should not be dismissed for the following reasons.
As previously mentioned in 2.4.2 – Experiment: Validity and Reliability, this study has
adopted an exploratory approach. There is a lack of research concerning bright patterns
and their potential in counteracting dark patterns in the scientific literature, and to the
best of our knowledge, there is no research concerning the ethical issues and ethical
approaches that designers face on a daily basis. The goal of this study was not to
statistically validate or prove a certain claim, but rather to shed light on an otherwise
obscure area of research.
Despite a small sample size, this study has managed to explore the topic of bright
patterns beyond the context of privacy cookie consent and shown that they can be a
promising ethical approach for designers to adopt.
Furthermore, this study has manged to explore some of the ethical issues that designers
encounter in their designs at their workplace, the ethical approaches that designers
adopt to face those ethical issues, and the level of authority that designers have in
addressing these ethical issues. As previously mentioned in section 2.4.1 – Semi-
structured Interview: Validity and Reliability, the findings from the semi-structured
interviews are not generalizable considering the small sample size. However, this paper
provides detailed descriptions about this aspect of the study to ensure that the findings
can be generalizable in the future.

5.2.3 Latent Content Analysis Implications

Latent content analysis was chosen as the appropriate method to analyze the data
collected from both the interviews with UX designers and the experiment with follow-
up interview. In retrospect, we feel that latent content analysis was advantageous in
extracting all the relevant data required to answer the research questions of this study,
allowing us to present the findings in a structured manner.
The broad, open-ended, and sometimes unplanned questions asked during the semi-
structured interviews with UX designers prompted interviewees to deliver rich answers.
Some of the questions asked during the experiment’s structured follow-up interviews
were narrow and related to specific tasks, but some questions were broad which also
prompted participants to deliver rich answers. These rich answers could be thoroughly
analyzed using latent content analysis since it is not strictly limited to words and
phrases, but also includes paragraphs, unlike manifest content analysis.

82
The only disadvantage with using latent content analysis is that it was very time-
consuming. The recorded interviews and experiments were transcribed in order to
perform a content analysis. The analysis process consisted of analyzing the transcripts
and coding the data into categories, discussing the categories’ relevance to the research
questions to finalize the codes and categories, and quantifying the data. These different
steps of the analysis process resulted in the analysis process being very time-
consuming. A possible workaround would have been to use software to automize the
analysis process, instead of us having to manually perform the content analysis.
However, since we adopted a latent content analysis approach that includes interpreting
the meaning of the units of analysis, software might not have been as advantageous in
extracting relevant data that answers the research questions of this study.

83
6 Conclusions and Further Research
This chapter presents the conclusions drawn from this study along with its practical and
scientific implications, and the suggestions for potential further research.

6.1 Conclusions
On the web, dark patterns have been used to manipulate people into making decisions
they never intended to do. Designers are held responsible for creating dark patterns, but
little is known about the reasons behind this. However, this study has addressed this
knowledge gap.
Designers cannot be blamed for creating dark patterns. Most designers struggle with
working in an ethical manner for various reasons. The authority to decide how designers
should address ethics in design is more distributed to stakeholders. It makes it more
difficult for designers to adopt ethical design choices, but they have found strategies to
work around this problem. Namely, designers try to find other designers with similar
ideas and then gather together to confront the stakeholders. It helps designers in
persuading stakeholders to go in the desired direction, though it requires great efforts.
Sometimes, designers cannot work around it, and this could lead designers into quitting
their jobs. Other strategies that designers adopt is to draw upon references to make it
easier to persuade the stakeholders. These references include GDPR and WCAG.
Other reasons that make designers struggle are that stakeholders focus on only getting
numbers up and can consider something to be ethical that someone else might not.
The ethical dilemmas that designers encounter revolve more around challenges of
working in an ethical manner than bad practices of incorporating ethics in design work.
Designers struggling to be ethical does not necessarily mean that they are unethical and
create dark patterns. However, if designers create dark patterns, a possible explanation
can be that they struggle with stakeholders. It can later end with them quitting the job.
Organizations must regulate the way ethics are incorporated in design to work towards
a digital world without dark patterns. However, the most efficient way would be to
involve the government to establish laws that prevent organizations from creating dark
patterns. Dark patterns are an issue that cannot be addressed on the individual level, but
rather on the organizational or governmental level.
Little research has been carried out to examine the potential that bright patterns have in
counteracting dark patterns. Therefore, the aim of this study was to further investigate
bright patterns and the potential that they have in counteracting dark patterns. This was
achieved by conducting an experiment to examine the difference in the influence that
bright patterns and dark patterns have on consumer decisions. In the experiment,
participants were instructed to fulfill a series of minor tasks on two simulated e-
commerce clothing websites. In Prototype A, participants encountered several dark

84
patterns, and in Prototype B, they encountered several bright patterns. When all tasks
were completed, participants were asked a couple of predetermined questions to
understand the reasoning behind their decisions, as well as their perceptions of ethics
(freedom of choice and transparency) of the two simulated websites.
The findings of the experiment indicate that there was no significant difference between
most bright and dark patterns in terms of how they influenced consumer decisions.
However, there was a significant difference in Task 4 where the dark patterns Roach
Motel and Misdirection influenced the consumers to make the wrong decisions to
complete the task, while their bright pattern counterparts influenced the consumers to
make the right decisions to complete the task. Regarding consumers’ perceptions of
ethics (freedom of choice and transparency) in bright and dark patterns, all participants
reported Prototype B (bright patterns) to provide the highest freedom of choice and
transparency. These findings indicate that bright patterns are embraced by the
experiment’s participants, and that bright patterns in fact have the potential to
counteract dark patterns. However, further research needs to be carried out to further
investigate the potential of bright patterns.

6.1.1 Practical Implications

For companies to not lose designer employees, they must consider incorporating ethics
in their work procedure. If designer employees find their company to be unethical, they
may decide to quit their job. This means that companies must start a recruitment process
to hire new employees, which would cost them resources.
Companies that are unethical would not only lose employees, but also the profits that
come from their users. A website with dark patterns may provide short-term profits to
the company, but when users find a better website, they could leave the
company’s website anytime. It means that companies that have websites with dark
patterns will not be profitable for them in the long term.

6.1.2 Scientific Implications

The ethics of persuasive design from the consumers’ perspectives, but predominantly
from the designers’ perspectives is an area of research that has been heavily
understudied prior to this study.
Previous studies have merely mentioned the importance of investigating the designers’
perspectives, but this study has taken it a step further by actually investigating the
designers’ perspectives on the ethics of persuasive design. This study has also
investigated the potential that bright patterns have in counteracting dark patterns
beyond privacy cookie consent, applying it to a larger scale.
Although the findings of this study have contributed to filling the identified knowledge
gap, future studies can continue to fill and expand on it to gain more insights into the

85
ethical issues and approaches that designers adopt, as well as the potential that bright
patterns have in counteracting dark patterns.

6.2 Further Research

Further research can be carried out to expand on the findings of this study. The
prototypes used in the experiment of this study did not incorporate all types of dark
patterns that have been identified in previous works. Future studies can investigate the
difference in consumer decisions and consumers’ perceptions of ethics concerning
these dark patterns and their bright pattern counterparts.
Additionally, this study’s experiment included prototypes that were designed as desktop
applications. Future studies can conduct a similar experiment with prototypes designed
as mobile applications to investigate whether the type of platform and device produce
different results.
One of the limitations of this study included the experiment having low ecological
validity since the prototypes used in the experiment are simulated websites where
participants do not spend real money and are instructed to fulfill a series of tasks. It
would be interesting for future studies to run a similar experiment in a real-world setting
where a real website is used by real users spending real money, and for users to be able
to navigate the website more freely. This might result in the consumers’ perceptions of
ethics in bright and dark patterns to be more intense and authentic.
Another limitation concerned the small sample size of this study, for both the interview
with UX designers as well as the experiment. It would be interesting for future studies
to run a similar study with a larger sample size, to see if the findings would differentiate
from the findings of this study.

86
7 References

Atkinson, B. M. C. (2006). Captology: A Critical Review. Persuasive Technology,

3962, 171–182. https://doi.org/10.1007/11755494_25

Benner, D., Schöbel, S., & Janson, A. (2021, July 20). It is only for your own good, or
is it? Ethical Considerations for Designing Ethically Conscious Persuasive
Information Systems. Americas Conference on Information System (AMCIS).
Americas Conference on Information System (AMCIS), Montreal, Kanada.
https://www.alexandria.unisg.ch/publications/263525

Bongard-Blanchy, K., Rossi, A., Rivas, S., Sur Alzette, E., Luxembourg, S., Doublet,
Luxembourg, V., Koenig, Luxembourg, G., Lenzini, Doublet, S., & Koenig, V.
(2021). “I am Definitely Manipulated, Even When I am Aware of it. It’s
Ridiculous!” -Dark Patterns from the End-User Perspective.
https://doi.org/10.1145/3461778.3462086

Brignull, H. (2019). Types of dark patterns. Dark Patterns.

https://www.darkpatterns.org/types-of-dark-pattern

Budiu, R. (2018, May 13). Between-Subjects vs. Within-Subjects Study Design. Nielsen
Norman Group. https://www.nngroup.com/articles/between-within-subjects/

Di Geronimo, L., Braz, L., Fregnan, E., Palomba, F., & Bacchelli, A. (2020). UI Dark
Patterns and Where to Find Them. Proceedings of the 2020 CHI Conference on
Human Factors in Computing Systems.
https://doi.org/10.1145/3313831.3376600

87
Fansher, M., Chivukula, S. S., & Gray, C. M. (2018). #darkpatterns: UX Practitioner
Conversations About Ethical Design. Extended Abstracts of the 2018 CHI
Conference on Human Factors in Computing Systems, 1–6.
https://doi.org/10.1145/3170427.3188553

Figma. (n.d.). Guide to prototyping in Figma. Retrieved March 31, 2022, from
https://help.figma.com/hc/en-us/articles/360040314193-Guide-to-prototyping-
in-Figma

Gouvier, W. D. (2019). ecological validity. In Encyclopædia Britannica.

https://www.britannica.com/science/ecological-validity

Graßl, P. a. J., Schraffenberger, H. K., Zuiderveen Borgesius, F. J., & Buijzen, M. A.

(2021). Dark and bright patterns in cookie consent requests. Journal of Digital
Social Research, 3(1), 1–38. https://repository.ubn.ru.nl/handle/2066/231659

Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus And Giroux.

Karagoel, I., & Nathan-Roberts, D. (2021). Dark Patterns: Social Media, Gaming, and
E-Commerce. Proceedings of the Human Factors and Ergonomics Society
Annual Meeting, 65(1), 752–756. https://doi.org/10.1177/1071181321651317

Karppinen, P., & Oinas-Kukkonen, H. (2013). Three Approaches to Ethical

Considerations in the Design of Behavior Change Support Systems. Persuasive
Technology, 7822, 87–98. Springer Berlin Heidelberg.
https://doi.org/10.1007/978-3-642-37157-8_12

Kight, R., & Gram-Hansen, S. B. (2019). Do Ethics Matter in Persuasive Technology?

Persuasive Technology: Development of Persuasive and Behavior Change
Support Systems, 143–155. https://doi.org/10.1007/978-3-030-17287-9_12

88
Kim, W.G., Pillai, S.G., Haldorai, K., & Ahmad, W. (2020). Dark patterns used by
online travel agency websites. Annals of Tourism Research, 88(2021), 1-6.
https://doi.org/10.1016/j.annals.2020.103055

M. Bhoot, A., A. Shinde, M., & P. Mishra, W. (2020). Towards the Identification of
Dark Patterns: An Analysis Based on End-User Reactions. IndiaHCI ’20:
Proceedings of the 11th Indian Conference on Human-Computer Interaction,
24–33. https://doi.org/10.1145/3429290.3429293

Maier, M., & Harr, R. (2020). Dark Design Patterns: An End-User Perspective. Human
Technology, 16(2), 170–199. https://doi.org/10.17011/ht/urn.202008245641

Mathur, A., Acar, G., Friedman, M. J., Lucherini, E., Mayer, J., Chetty, M., &
Narayanan, A. (2019). Dark Patterns at Scale: Findings from a Crawl of 11K
Shopping Websites. Proceedings of the ACM on Human-Computer Interaction,
3(CSCW), 1–32. https://doi.org/10.1145/3359183

McLean, A. (2020). Dark Patterns. Canadian Journal of Nursing Informatics, 15(2).

https://www.proquest.com/docview/2561886665/C450E71EFC0540D8PQ/1

Mildner, T., & Savino, G.-L. (2021). Ethical User Interfaces: Exploring the Effects of
Dark Patterns on Facebook. Extended Abstracts of the 2021 CHI Conference
on Human Factors in Computing Systems, 464.
https://doi.org/10.1145/3411763.3451659

Narayanan, A., Mathur, A., Chetty, M., & Kshirsagar, M. (2020). Dark Patterns: Past,
Present, and Future. Communications of the ACM, 63(9), 42–47.
https://doi.org/10.1145/3397884

89
Parker, C., Scott, S., & Geddes, A. (2019). Snowball Sampling. SAGE Research
Methods Foundations. https://eprints.glos.ac.uk/6781/

Saunders, M., Lewis, P., & Thornhill, A. (2015). Research Methods for Business
Students (7th ed.). Pearson.

Westin, F., & Chiasson, S. (2019). Opt out of privacy or “go home.” Proceedings of the
New Security Paradigms Workshop. https://doi.org/10.1145/3368860.3368865

Wintermeier, N. (2020a, April 7). Persuasive eCommerce Design: Convert Browsers to

Buyers. Crobox. https://blog.crobox.com/article/persuasive-design-ecommerce

Wintermeier, N. (2020b, June 22). Dark Patterns Examples in eCommerce: What they
are & why to avoid them. Crobox. https://blog.crobox.com/article/dark-patterns

90
8 Appendices

Appendix 1: The questions asked in the semi-structured interview with UX designers.

Appendix 2: The questions asked during the follow-up interview of the experiment.

91
8.1 Appendix 1
Warm-up Questions

• How long have you been working as a UX designer?

• Do you work in a small or large company?
• Do you work in a team or do you work alone?
o Which different roles are included in your team? (developers?
designers?)
o Does your team have multiple UX designers or are you the only one?

Main Questions

• What do you consider to be your most ethical qualities as a person?

• What do you consider to be ethical and unethical design?
o Could you give examples of ethical and unethical design?
• Have you ever witnessed unethical design arise in your workplace?
o If yes: What happened? What did you do (confront or comply)? What
was the outcome? How many times have you been asked to do this?
o If no: Suppose you are in this situation, what would you do? Why?
• Have you ever had a company that asked you to perform an unethical design?
o If yes: What happened? What did you do (confront or comply)? What
was the outcome? How many times have you been asked to do this?
o If no: Suppose you are in this situation, what would you do? Why?
o Have you ever been asked to create an unethical design feature (Meaning
that the whole program is not unethical, but a certain part of the design
/ a feature is)?
• What do you do to stay ethical in your workplace (in regard to design)?
o Why?

92
8.2 Appendix 2
Interview questions

• Task 1 (both prototypes)

o If 3-Pack: Why did you choose the 3-Pack of t-shirts instead of choosing
3 individual t-shirts (the other option)?
o If 3 t-shirts: Why did you choose the 3 individual t-shirts instead of
choosing 3-Pack (the other option)?
o Did you find it easy or difficult to compare the prices?
▪ Why?
• Task 2 (both prototypes)
o Why did you choose this jeans over the other jeans?
o If NOT urgency scarcity message: Did you notice the jeans with the
red message in the image?
▪ Yes: What were your thoughts/feelings?
o If urgency scarcity message: What did you feel when you saw the red
message on the image?
• Task 3
o Bright Patterns (Checkout)
▪ Why did you choose one option over the other?
▪ If NO THANKS: What would make you add the item in your
bag?
▪ What were your thoughts when you saw the popup recommend
an item to you?
▪ Did you feel that you feel that you had the freedom to make your
own decisions?
• Yes: What made you feel that way?
• No: What made you feel that way? What would be
needed to make you feel like you have the freedom to
make your own decisions?
o Bright Patterns (Newletter)
▪ Why did you choose the option you chose?
▪ Was the information in the newletter clear or ambiguous?
• Clear: What made the information clear?

93
 • Ambiguous: What made the information ambiguous?
How would you make the information clearer?
▪ Did you feel that you feel that you had the freedom to make your
own decisions?
• Yes: What made you feel that way?
• No: What made you feel that way? What would be
needed to make you feel like you have the freedom to
make your own decisions?
o Dark Patterns (Checkout)
▪ If noticed sneaked-in product: How did you feel about having
an extra clothing item in your bag that you did not add yourself?
• If removed: Why did you remove the extra item from
your bag?
• If not removed: Why did you keep the extra item in your
bag?
▪ If unnoticed sneaked-in product: Did you notice the extra item
in your bag?
• No: What are your thoughts/feelings now that you found
out that there is an extra item in your bag?
o Dark Patterns (Newsletter)
▪ Why did you choose the option you chose?
▪ Was the newsletter sign-up clear? Did you understand what you
needed to do to get the outcome you desired?
▪ Did you feel that you feel that you had the freedom to make your
own decisions?
• Yes: What made you feel that way?
• No: What made you feel that way? What would be
needed to make you feel like you have the freedom to
make your own decisions?
• Task 4
o Bright Patterns
▪ Did you feel that it was easy or difficult to find your way to the
delete account page?

94
 • Easy: What made it easy?
• Difficult: What made it difficult?
• How did it differ from your expectations?
▪ Did you feel that it was easy or difficult to delete your account?
• Easy: What made it easy?
• Difficult: What made it difficult?
o Dark Patterns
▪ Did you feel that it was easy or difficult to find your way to the
delete account page?
• Easy: What made it easy?
• Difficult: What made it difficult?
• How did it differ from your expectations?
▪ If clicked/almost clicked NEVER MIND button: Did you feel
deceived by the buttons?
• Why did you feel that way?
• Yes: What changes would you make to not feel
deceived?

Concluding Questions

• Which prototype did you prefer?

o Why?
• Which prototype did you find to give the most freedom to make the decisions you
want?
o If prototype B (Bright Patterns): Why? How is it different from prototype A
(DP)?
o If prototype A (Dark Patterns): Why? How is it different from prototype B
(BP)?
• Which prototype did you find to hold the clearest information?
o If prototype B (Bright Patterns): Why? How is it different from prototype A
(DP)?
o If prototype A (Dark Patterns): Why? How is it different from prototype B
(BP)?

95