Post on 29-Nov-2014
description
Privacy in Mobile Personalized SystemsThe Effect of Disclosure Justifications
Bart P. KnijnenburgDepartment of Informatics, UC Irvine
Samsung R&D Research
Alfred KobsaDepartment of Informatics, UC Irvine
Gokay SaldamliSamsung R&D Research
INFORMATION AND COMPUTER SCIENCES
Mobile apps need personal data
Mobile applications often use personalization
This requires personal information
- Demographical data (e.g. age, hobbies, income)
- Contextual data (e.g. app usage, calendar, location)
INFORMATION AND COMPUTER SCIENCES
Let users control their disclosure
Problem: Many people are not comfortable disclosing diverse personal information
FTC, CPBoR: let users decide
Privacy calculus: trade off between benefits and risks
INFORMATION AND COMPUTER SCIENCES
Help users decide what to disclose
Problem: This trade-off is difficult!
Lack of knowledge about positive and negative consequences
CPBoR: informed choicePrevious research: justifications
INFORMATION AND COMPUTER SCIENCES
Justification types
Explain the reason why the information is requestedMay prove the legitimacy of the disclosure request
Highlight the benefits of disclosurePrivacy calculus: tip the scales in favor of the benefits
Appeal to the social norm Eschew privacy calculus by conforming to the majority
INFORMATION AND COMPUTER SCIENCES
Our starting point
Previous work: Justifications seem to work
- They increase disclosure
- They increase user satisfaction - not always tested
Our goal: Find out which one works best
INFORMATION AND COMPUTER SCIENCES
Experiment
INFORMATION AND COMPUTER SCIENCES
Experiment
INFORMATION AND COMPUTER SCIENCES
Gender, etc.
Manipulations
Context data first Demographical data first
Location, etc.
Location, etc.
Gender, etc.
INFORMATION AND COMPUTER SCIENCES
Manipulations
5 justification typesNoneUseful for youNumber of othersUseful for othersExplanation
INFORMATION AND COMPUTER SCIENCES
Which one is best?
Which increases disclosure the most?
Which increases satisfaction the most?
INFORMATION AND COMPUTER SCIENCES
0%#
10%#
20%#
30%#
40%#
50%#
60%#
70%#
80%#
90%#
100%#Context#first# Demographics#first# Context#first# Demograpics#first#
Disclosure*behavior**
Demographics*disclosure * *Context*disclosure*
Results
INFORMATION AND COMPUTER SCIENCES
*"
1"
**"*"
***"*"
*"
0%"
10%"
20%"
30%"
40%"
50%"
60%"
70%"
80%"
90%"
100%"Context"first" Demographics"first" Context"first" Demograpics"first"
Disclosure*behavior**
Demographics*disclosure * *Context*disclosure*
none" useful"for"you" #"of"others" useful"for"others" explanaDon"
Results
INFORMATION AND COMPUTER SCIENCES
***"***"
**"
#1,00"
#0,75"
#0,50"
#0,25"
0,00"
0,25"
0,50"
0,75"
1,00"
Perceived(value(of(disclosure(help(
Results
Perceived value of disclosure help:
3 items, e.g. “The system helped me to make a tradeoff between privacy and usefulness”
Higher for all except “number of others”
*"
1"
**"*"
***"*"
*"
0%"
10%"
20%"
30%"
40%"
50%"
60%"
70%"
80%"
90%"
100%"Context"first" Demographics"first" Context"first" Demograpics"first"
Disclosure*behavior**
Demographics*disclosure * *Context*disclosure*
none" useful"for"you" #"of"others" useful"for"others" explanaDon"
INFORMATION AND COMPUTER SCIENCES
*"
#1,00"
#0,75"
#0,50"
#0,25"
0,00"
0,25"
0,50"
0,75"
1,00"
Perceived(privacy(threat(
Results
Perceived privacy threat:3 items, e.g. “The system has too much information about me”
Higher for “useful for others”
*"
1"
**"*"
***"*"
*"
0%"
10%"
20%"
30%"
40%"
50%"
60%"
70%"
80%"
90%"
100%"Context"first" Demographics"first" Context"first" Demograpics"first"
Disclosure*behavior**
Demographics*disclosure * *Context*disclosure*
none" useful"for"you" #"of"others" useful"for"others" explanaDon"
INFORMATION AND COMPUTER SCIENCES
**"1"
$1,00"
$0,75"
$0,50"
$0,25"
0,00"
0,25"
0,50"
0,75"
1,00"
Trust&in&the&&company&
Results
Trust in the company:4 items, e.g. “I believe this company is honest when it comes to using the information I provide”
Generally lower, especially for “useful for others”
*"
1"
**"*"
***"*"
*"
0%"
10%"
20%"
30%"
40%"
50%"
60%"
70%"
80%"
90%"
100%"Context"first" Demographics"first" Context"first" Demograpics"first"
Disclosure*behavior**
Demographics*disclosure * *Context*disclosure*
none" useful"for"you" #"of"others" useful"for"others" explanaDon"
INFORMATION AND COMPUTER SCIENCES
**" **"***"
1"
$1,00"
$0,75"
$0,50"
$0,25"
0,00"
0,25"
0,50"
0,75"
1,00"
Sa#sfac#on)with))the)system)
Results
Satisfaction with the system:6 items, e.g. “Overall, I’m satisfied with the system”
Lower for any justification!
*"
1"
**"*"
***"*"
*"
0%"
10%"
20%"
30%"
40%"
50%"
60%"
70%"
80%"
90%"
100%"Context"first" Demographics"first" Context"first" Demograpics"first"
Disclosure*behavior**
Demographics*disclosure * *Context*disclosure*
none" useful"for"you" #"of"others" useful"for"others" explanaDon"
INFORMATION AND COMPUTER SCIENCES
Conclusion
Justifications did not have the expected effectsNo increase in disclosureNo decrease in perceived threat, no increase in trust Satisfaction is lower
...but participants liked the disclosure help!
INFORMATION AND COMPUTER SCIENCES
Reflection
Why did this happen?
Possible reason 1: Justifications are seen as persuasionBut participants liked the disclosure help
Possible reason 2: Low percentages cause disappointmentDisclosure only starts to increase at around 90% for the “number of others” justification
Possible reason 3: Justifications carry an implicit warningThey signal that the disclosure decision is not trivial
INFORMATION AND COMPUTER SCIENCES
Discussion
None of our justification messages seemed to work very wellIs there a “golden justification”?
Different justifications may work for different types of usersHas anyone tried “tailored” disclosure help?
We provided objective information for privacy decisionsShould we do this even if it reduces users’ satisfaction?
Thank youbart.k@uci.edu :: www.usabart.nl :: @usabart
INFORMATION AND COMPUTER SCIENCES
Discussion
None of our justification messages seemed to work very wellIs there a “golden justification”?
Different justifications may work for different types of usersHas anyone tried “tailored” disclosure help?
We provided objective information for privacy decisionsShould we do this even if it reduces users’ satisfaction?