Profiling Facebook Users’ Privacy Behaviors

Pamela Wisniewski, Bart P. Knijnenburg, & Heather Richter Lipford

Workshop on Privacy Personas and Segmentation (PPS)

Introduction

} Privacy: An interpersonal boundary process by which a person or group regulates social interactions with others

}  SNS Privacy Research: Often frames privacy as SNS users’ decision to withhold or disclose personal information

}  SNS Privacy Behaviors: The subset of privacy features and/or settings that Facebook users leverage in order to manage interpersonal privacy boundaries }  Includes but is not limited to personal disclosure decisions

Facebook Privacy Options

News Feed

Chat

Timeline

Privacy Settings

Facebook Privacy Options Untag

Unshare

Block

Background

}  Setting profile as “Friends Only” (Stutzman et al. 2010)

}  Use of “advanced” privacy settings (Ellison et al. 2011) }  Selective sharing and customizing

privacy defaults }  Associated with higher levels of

social capital

}  Use of “Friend Lists” or “Circles” (Kairam et al. 2012; Watson et al. 2012)

Privacy Features

Privacy Settings

Disclosure Decisions

Previous Work }  21 Semi-Structured SNS user interviews

}  Feature-Oriented Domain Analysis

Disclosure

•  Self • Confidant

Relational

• Connection • Context

Network

• Discovery •  Intersection

Territorial

•  Inward-facing

• Outward-facing

Interactional

• Disabling • Blocking

SNS Interpersonal Boundary Types

Privacy Behaviors – Settings/Features

Methodology }  Web-based survey

}  Recruited participants 18-years-old or older with an active Facebook account

}  Asked to log into their accounts to report privacy behaviors

“To do this: You would have had to click on the drop down arrow at the top, right corner of a post on your News Feed as shown below.”

How often have you done the following to modify posts on your News Feed?” Ø  Hid a story Ø  Reported Story or Spam Ø  Changed friend subscription settings Ø  Unsubscribed from a friend Ø  Unsubscribed from status updates from a friend.

(1 = Never, 7 = Always)

Data Analysis Approach }  Adapted from Knijnenburg et al. 2013

}  Confirmed the multi-dimensionality of privacy behaviors }  Confirmatory Factor Analysis (CFA)

}  Classified users based on privacy behaviors }  Mixture Factor Analysis (MFA)

Results }  308 Participants

}  119 males, 189 females

}  Average Age 35.74, (sd = 12 years; 18 – 75 years-old)

}  31% identified as college students

}  91.6% reported having a Facebook account > 2 years

}  19.2% reported having a Facebook account > 6 years

Privacy Behavior CFA Factor Code Item Loading Altering News Feed (NWF) AVE: 0.777

NFH Hid a story 0.845 NFS Changed friend subscription 0.872 NFN Unsubscribed to a friend 0.908 NFP Unsubscribed to status updates 0.900

Timeline/Wall Moderation (WAL) AVE: 0.638

CWD Deleted content from Timeline/Wall 0.783 CWS Reported/marked content as spam 0.796 CWH Hid a story 0.817

Reputation Management (REP) AVE: 0.671

UNT Untagged a photo or post 0.800

TAK Requested friends to take down posts or photos 0.838

Limiting Access Control (LIM) AVE: 0.734

TAG Tag visibility privacy setting 0.683 SEE Wall/Timeline post visibility privacy setting 1.012 DEF Default privacy level Removed

Blocking people (BLP) AVE: 0.838

BLU Blocked a user 0.892 RES Added a user to restricted list 0.938

Blocking apps/events (BLA) AVE: 0.621

BLE Blocked an event invite 0.746 BLA Blocked an app invite  0.828

Restricting Chat (CHA) AVE: 0.777

SCF Gone “offline” on Facebook chat 1.013 SCH Default chat visibility 0.744

Selective Sharing (SEL) AVE: 0.829

POS Posted a status to a custom friend list 0.867 PIC Posting a photo to a custom friend list 0.952

Friend Management (FRM) AVE: 0.910

LIN Categorized new friends into friend lists 0.915 LIO Categorized existing friends into friend lists 0.991

Withholding Contact Info. (CON) AVE: 0.780

CIB Withheld/restricted cell phone number 0.742 CIP Withheld/restricted other phone number 0.946 CIM Withheld/restricted IM screen name 0.880 CID Withheld/restricted street address 0.949

Withholding Basic Info. (BAS) AVE: 0.700

BAD Withheld/restricted “Interested In” 0.750 BAE Withheld/restricted religion 0.878 BAO Withheld/restricted political views 0.876

Concealing Network FRL Hid Friend list from profile Removed Denying Connection HID Hidden a friend request Removed

UNF Unfriended (frequency) Removed

χ2(295) = 432.59, p < .001; CFI = .987, TLI = .983; RMSEA = .039, 90% CI: [.031, .047]

Privacy Behavior User Classes }  Comparing across MFA model results

BIC Entropy LL N p-value 1 class 21998 -10534.652 162 2 classes 20829 0.915 -9916.195 174 < .001 3 classes 20479 0.915 -9706.503 186 0.1032 4 classes 20324 0.880 -9594.600 198 0.7248 5 classes 20183 0.905 -9489.752 210 0.1774 6 classes 20104 0.922 -9415.822 222 0.4441 7 classes 20163 0.904 -9411.090 234 0.7039

Class Distributions

Privacy Behavior User Classes }  Interactive Web Charts:

http://www.usabart.nl/chart/

Key Implications }  SNS users employ a subset of privacy features

}  Exhibiting a multidimensional pattern from which emerged unique privacy management strategies

}  Dimensionality was often tied to physical groupings within the interface design

}  Privacy strategies extend beyond disclosure decisions }  Self-Censors vs. Selective Sharers

}  Privacy behavior propensity provides valuable insights }  Low propensity highlights opportunities for privacy redesign and/or user

education (i.e. Friend List Management vs. Selective Sharing)

}  Profiling users offers new opportunities }  Privacy personalization, notifications, advice, and recommendations }  Better understanding of antecedents and outcomes associated with various

profiles (i.e. Feature Awareness, Privacy Desires, Social Benefits, etc.)

Feature Awareness vs. Privacy Behavior

15

C1  Privacy  Maximizers  

C2  Selec4ve  Sharers  

C3  Privacy  Balancers  

C4  Time  Savers  

C5  Self-­‐Censors  

C6  Privacy  Minimalists  

C1  Experts  13  (5.6)   6  (3)   28  (20.4)   4  (9.4)   1  (6.1)   5  (12.6)  

C2  Near  Experts   11  (6.9)   8  (3.7)   31  (25.4)   7  (11.8)   4  (7.6)   10  (15.7)  C3  

1  (4)   0  (2.1)   17  (14.6)   9  (6.8)   4  (4.4)   10  (9.1)  C4  

1  (3.8)   0  (2)   1  (13.9)   4  (6.5)   12  (4.2)   21  (8.6)  C5  Near  Novices   1  (5.8)   0  (3.1)   11  (21.4)   22  (9.9)   7  (6.4)   19  (13.2)  C6  Novices  

3  (3.9)   2  (2.1)   22  (14.3)   5  (6.6)   5  (4.3)   3  (8.8)  

Class-to-class Membership

Thank you! }  Questions?

}  Author Contact Information }  Pamela Wisniewski – pam@pamspam.com }  Bart Knijnenburg – bart.k@uci.edu }  Heather Richter Lipford – heather.lipford@uncc.edu

Boundary Type SNS Interface Controls Facebook MySpace Hi5 LinkedIn Ning

Disclosure Boundaries – Managing personal information

Confidant-Disclosures

 

 

Access Level Settings X X X X

Delete Posts or Comments X X X X X

UnTagging X X X

Moderation X X X X

Relationship Boundaries – Managing one’s interpersonal interactions

Connection Access Level - Friend Request X X X X

Deny Friend Request X X X X X

Unfriend/Remove Connection X X X X X

Context Group Labeling X X X X

Group Management X

Network Boundaries – Managing interactions between one’s connections

Discovery Access Level – Friend List X

Access Level – Profile X X X

Hide Connections X X X

Intersection See Relationship Context

Territorial Boundaries – Managing one’s virtual spaces

Inward-Facing Filters X X X X

Preference Settings X X X

Hiding X X X

Outward-Facing See Confidant-Disclosure

Interactional Boundaries – Managing access to self

Disabling Search (Finding You) X

Posts/Commenting X X X

Tagging X

Friend Requests X

Chat X X X

Blocking Blocking X X X X