timothy d bowman dissertation defense
TRANSCRIPT
Investigating the Use of Affordances and Framing
Techniques by Scholars to Manage Personal and Professional
Impressions on TwitterTimothy D. Bowman, Ph.D. Candidate, Indiana
UniversityResearch Professional, Université de Montréal
MO
TIVA
TIO
NTH
EORI
ESIM
PLIC
ATIO
NS
MET
HO
DS
&
RESU
LTS
MOTIVATION
CONTROLLING IMPRESSIONS
CONTROLLING IMPRESSIONS
CENSO
RED
SUSP
ENDED
RESCIN
DE
D
• Twitter use is increasing (Brenner & Smith, 2013)
• Around 10% of scholars on Twitter with variation by field(Ponte & Simon, 2011; Rowlands et al., 2011)
• 20% of scientific articles shared on Twitter (Haustein et al., 2014; Holmberg and Thelwall, 2014).
Brenner & Smith, 2013
• Few people examining impression management of scholars in Twitter
• Studies of scholars tend to focus on social media in the classroom or on scholarly output (altmetrics)
• Populations of scholars on Twitter tend to be limited
LITERATURE GAP
THEORIES & QUESTIONS
The Presentation of Self in Everyday Life (Goffman, 1959)
PRESENTATION OF SELF
BACKSTAGEINFORMAL
TALKRELAXED
ROLE
BARRIER
GIVE OFFGIVE
DRAMATIC INTERACTION
FRONT STAGE
SIGNS
PROPSBARRIER
IMPRESSIONMANAGEMENTExpressing certain information in order to impress certain ideas upon an audience during social interaction
Frame Analysis: An Essay on the Organization of Experience (Goffman, 1974)
FRAME ANALYSIS
FRAME RIM
PRIMARY FRAME
KEYING
FABRICATION
SIGNS/SYMBOLS
BRACKETS
ROLES
EXPERIENCE
ENGROSSABLES
The Theory of Affordances. (Gibson, 1977)
AFFORDANCE
CONTEXT
SOCIAL RULES
EXPERIENCE
PERCEPTION
1. In what ways do scholars utilize affordances to manage impressions on Twitter?
2. In what ways do scholars frame interactions to manage impressions on Twitter?
3. What are the differences in the use of framing strategies and affordances by scholars for managing the presentation of their professional and personal selves on Twitter?
RESEARCH QUESTIONS
METHODS & RESULTS
Phase One Online survey of full, associate, & assistant professors
Phase Two Tweet categorization in Amazon’s Mechanical Turk (AMT)
Phase Three Follow-up survey and tweet categorization withmost active professors on Twitter
RESEARCH METHODS
• 62 Association of American Universities (AAU) Member Schools (2 Canadian)
• 16,862 Full, Associate, and Assistant Professors
• 8 Departments – Anthropology, Biology, Chemistry, Computer Science, English, Philosophy, Physics, and Sociology
PHASE ONE: SAMPLING
• 19 questions
• Five sections
• Matrix questions
• Likert-scale questions
• 8.5% response rate
PHASE ONE: INSTRUMENT
χ2 (7, n=1,910) = 0.182, p = .0005, Cramér’s V = 0.182
PHASE ONE: TWITTER USE
Computer ScienceEnglish
SociologyAnthropology
BiologyPhilosophy
PhysicsChemistry
50.0%37.5%36.9%
29.0%27.5%27.1%
24.3%20.7%
68% 32%
NO YES
(n=613)(n=1,297)
(n=224)
(n=299)
(n=271)
(n=169)
(n=367)
(n=144)
(n=267)
(n=169)
(N=1,910)
(N=1,910)
Non-White
White
45%
41%
PHASE ONE: TWITTER USE
Male
Female
38%
49%(n=615)
(n=1,200)
6 Years or Less
7 to 9 Years
10 Years or More
39%
41%
25%(n=1,262)
(n=196)
(n=363) 35 and Under
36 to 45
46 to 60
61 and Over
44%
36%
28%
11%(n=271)
(n=841)
(n=517)
(n=194)
(n=229)
(n=1,580)
χ2 (2, n=1,821) = 0.217, p = .0005, Cramér’s V = 0.217) χ2 (3, n=1,823) = 0.125, p = .0005, Cramér’s V = 0.125.
χ 2 (2, n=1,824) = 0.066, p = .018, Cramer’s V = 0.18
PHASE ONE: SOCIAL MEDIA USE
ScilinkEpernicus
BioMedExperts.comMySpace
ScribdSlideShare
TumblrOther
BloggerPinterest
MendeleyInstagramWordPressWikipedia
Academia.eduResearchGate
YouTubeTwitter
Google+LinkedIn
0.4%0.7%0.9%2.0%2.7%2.9%
4.8%5.2%5.7%7.2%7.3%7.4%
15.4%16.7%
21.5%26.2%27.8%
32.0%50.2%
57.8%69.9%(N=1,63
9)
Professor (n=253)
Associate (n=177)
Assistant (n=178)
42%
29%
29%
PHASE ONE: TWITTER USERS
English
Computer Science
Biology
Sociology
Physics
Anthropology
Philosophy
Chemistry
18.3%
18.3%
16.5%
16.3%
10.6%
8.0%
6.0%
6.0%
(n=112)
(n=112)
(n=101)
(n=100)
(n=65)
(n=49)
(n=39)
(n=35)
PHASE ONE: TWITTER USERS
< 1 year
1 to 2 years
2 to 3 years
3 to 4 years
4 to 5 years
5 to 6 years
> 6 years
22%
28%
21%
15%
6%
3%
5%
Male
Female
62%
37%
29%
42%
29%
PHASE ONE: ACCOUNT TYPE
BOTH PROFESSIONALPERSONAL
PHASE ONE: ACCOUNT TYPE
Philosophy (n=33)Biology (n=93)
English (n=102)Sociology (n=88)Chemistry (n=31)
Computer Science (n=102)Anthropology (n=45)
Physics (n=59)
55%22%19%25%
35%28%
42%44%
21%49%60%
41%39%
37%33%
25%
24%29%
22%34%
26%34%
24%31%
PersonalBothProfessional
PHASE ONE: AFFORDANCE USE
Add PhotoAdd Location
Address Message AtMention Someone
Use HashtagsEmbed URLs
Delete a tweetFavorite a tweetReply to a tweetRetweet a tweet
19%5%5%
29%26%
34%34%26%
34%27%
80%94%93%
68%68%
58%58%
61%53%48%
3%6%8%8%13%14%
25%
Mostly or Always Sometimes Rare or Never
PHASE ONE: AFFORDANCE USE
Profile PictureBio information
Apps Allowed Access to TwitterPrivacy SettingsHeader Picture
Twitter Sends EmailLanguage Specified
Twitter Connected to FacebookCountry
Time ZoneTheme
Geo TaggingTwitter Sends Text MessagesTwitter Personalizes Interface
Widget(s) CreatedPhone Number Specified
Sleep Settings
60.3%52.1%
43.3%40.8%
38.0%24.9%24.6%23.5%22.9%
19.8%14.4%
8.2%6.8%5.1%4.8%
2.5%1.7%
87.3% 87.3%
16.2%
PHASE ONE: FINDINGS
• Age, academic age, department, and gender associated with having Twitter account
• The majority of professors indicated using their Twitter account for both personal and professional communications
• There were differences in perceived affordance use
Phase One Online survey of full, associate, & assistant professors
Phase Two Tweet categorization in Amazon’s Mechanical Turk (AMT)
Phase Three Follow-up survey and tweet categorization withmost active professors on Twitter
RESEARCH METHODS
• 445 Twitter accounts
• 289,934 tweets collected
• 75,000 tweets in AMT
PHASE TWO: DATA COLLECTION
PHASE TWO: DATA COLLECTION
Group Name Average Tweets
per Day (TPD)Total Scholars
in GroupTotal Tweets
CollectedPercentage of
Total TweetsTweets Used
in AMTTEN
(intense) 8 to 24 9 29,064 10.02% 7,518
NINE 5 to 8 8 25,863 8.92% 6,690EIGHT 4 to 5 6 19,321 6.66% 4,998SEVEN 3 to 4 10 24,532 8.46% 6,346
SIX 2.5 to 3 10 25,508 8.80% 6,598FIVE 2 to 2.5 10 22,195 7.66% 5,741FOUR 1.5 to 2 13 23,018 7.94% 5,954THREE 1 to 1.5 29 43,831 15.12% 11,338TWO 0.5 to 1 33 30,463 10.51% 7,880ONE
(infrequent) < 0.5 317 46,139 15.91% 11,935
445 289,934 100.00% *75,000
*Confidence Interval 0.4 at 99% Confidence Level
0.0
5.0
10.0
15.0
20.0
25.0
• 12,056 Human Intelligence Tasks (HIT)
• 7 tweets per HIT
• 1 control question per HIT
• 3 Turkers per HIT
PHASE TWO: INSTRUMENT
Anthro Bio Che
mComp
Sci Eng Philo Phys Soc Average
HASHTAGS 4.4% 5.5% 5.2% 5.2% 4.9% 4.6% 6.4% 7.4% 5.5%
URLs 0.7% 1.2% 0.3% 1.1% 0.5% 1.7% 0.8% 1.1% 0.9%
MENTIONS 11.6% 16.3% 12.9% 9.2% 13.4% 10.6% 13% 20% 13.4%
RETWEETS 241 273 137 244 291 171 124 205 211
PHASE TWO: AFFORDANCES
Philosophy
English
Anthropology
Sociology
Biology
Computer Science
Physics
Chemistry
1.96
1.41
1.18
1.06
0.73
0.67
0.53
0.52Female Male
0.801.02
PHASE TWO: TWEET ANALYSIS
Mean tweets per day
PHASE TWO: AMT RESULTS
PERSONAL
PROFESSIONAL
UNKNOWN
NON-ENGLISH TOTAL
Full Agreement (3/3) 27,264 6,810 129 766 34,969
(47%)Partial Agreement (2/3)
19,403 15,692 1,993 262 37,355 (49%)
No Agreement 2,674 (4%)96% agreement for 2 out of 3 Turkers on all tweets
Hashtags URLs User Mentions Retweets
17% 15%
67%
17%28%
69%
56%
36%
Personal Professional
χ 2 (1, n=34,074) = 0.187, p = 0.0005, Cramer’s V = 0.187
χ 2 (1, n=34,074) = -0.089, p = 0.0005, Cramer’s V = 0.089 χ 2 (1, n=34,074) = 0.491, p = 0.0005, Cramer’s
V = 0.491
χ 2 (1, n=34,074) = 0.112, p = 0.0005, Cramer’s V = 0.112
PHASE TWO: AMT RESULTS
PHASE TWO: FINDINGS
• URLs, retweets, and hashtags occurred more often in professional tweets
• User mention was the only affordance to occur more in personal tweets
• There were also differences across personal and professional tweets by academic title, age, gender, department and Twitter activity
Phase One Online survey of full, associate, & assistant professors
Phase Two Tweet categorization in Amazon’s Mechanical Turk (AMT)
Phase Three Follow-up survey and tweet categorization withmost active professors on Twitter
RESEARCH METHODS
• 95 Most Active Scholars on Twitter
• 5 Tweets to Categorize – 2 Personal, 3 Professional
PHASE THREE: DATA COLLECTION
• 6 questions
• 5 tweets: 2 Personal, 3 Professional
PHASE THREE: INSTRUMENT
PHASE THREE: AFFORDANCE USE
Mentions URLs
Retweets Hashtags
Directed messages Punctuation, caps, etc.
MediaEmoticons
OtherNot used in this way
54.2%42.4%
44.1%42.4%
54.2%32.2%
55.9%30.5%
5.1%10.2%
84.7%84.7%
79.7%78.0%
61.0%61.0%
54.2%13.6%
6.8%3.4%
Professional Personal
PHASE THREE: AFFORDANCE USE
Description
Profile Image
Theme
Header (banner) Image
Location
Colors
Other
Not used in this way
25%
21%
6%
13%
8%
6%
2%
19%
77%
60%
32%
30%
23%
17%
11%
11%
Professional Personal
PHASE THREE: TWEET CATEGORIZATION
PERSONAL PROFESSIONAL TOTAL AGREEMENT
TURKERS (3/3) 102 153 255SCHOLAR 44 125 169
43% Agreement
82% Agreement 69% Agreement
OBSERVED TURKERS EXPECTED TURKERS
Personal
Professional
Personal
Professional
Personal 44 28 Personal 29 43Professional
58 125 Professional
73 110
Cohen’s Kappa = 0.26
Turker and Scholar Agreement
Turker and Scholar Coding: Cohen’s Kappa
PROF
ESSO
R
PROF
ESSO
R
TURKER: PersonalPROFESSOR: Professional
TURKER: ProfessionalPROFESSOR: Personal
INCORRECTLY CATEGORIZED TWEETS
CORRECTLY CATEGORIZED TWEET
TURKER: ProfessionalPROFESSOR: Professional
#citsci What motivates you to take part in @the_zooniverse or citizen science projects. Help us find out!. http://t.co/2tth5RVpmN
TURKER: PersonalPROFESSOR: Personal @k_garten @bashartak @sig_chi we should just
form CHI University and all hang out together all the time. #chi2013
PHASE THREE: FINDINGS
• It is difficult for audience members to distinguish between personal and professional tweets
• The framing of tweets are associated with specific affordances
• Professional tweets are perceived as containing more hashtags, URLs and are retweets
• Personal tweets are perceived as containing more user mentions
IMPLICATIONS & CONCLUSION
PERSONAL
PROFESSIONAL
SOCO-TECHNICAL FRAMEWORK
•Sample Population•Survey Response Rate/Design•Low Cohen’s Kappa
LIMITATIONS
• Distinguishing between personal and professional tweets is difficult• Framing of tweets by audience members depends on many
factors, one of which may be affordances• There are differences in tweeting behavior by various demographic
characteristics• One of the largest studies on U.S. professors on Twitter to date• Unique use of Amazon’s Mechanical Turk• Provides foundation for altmetrics research• Social media use is a hot topic
SUMMARY
1.Analyze events in online social media
2.Utilize theories and methods from multiple disciplines
3.Use quantitative and qualitative methods
FUTURE DIRECTION
Thank You
Timothy D. Bowman, Ph.D. Candidate, Indiana UniversityResearch Professional, Université de Montréal