The Social Stratification of Fame Arnout van de Rijt Stony Brook University Charles Ward Stony Brook...

53
The Social Stratification of Fame Arnout van de Rijt Stony Brook University Charles Ward Stony Brook University Steven Skiena Stony Brook University Eran Shor McGill University

Transcript of The Social Stratification of Fame Arnout van de Rijt Stony Brook University Charles Ward Stony Brook...

The Social Stratification of Fame

Arnout van de RijtStony Brook University

Charles WardStony Brook University

Steven SkienaStony Brook University

Eran ShorMcGill University

Missing in Stratification Research

incomewealth

educationoccupational prestige

statushealth

x fameIntroduction Measuring Fame Data Sources Analysis Strategy Results

Definition of FameEmergent “Sociology of fame and celebrity” (Ferris 2007):

• “pure renown – literally the sum of all people who have heard of a person’s name.”(Currid-Halkett 2010:29,66)

• continuous, must include also intermediates: "local newscasters, minor league athletes, or local politicians" (Ferris 2010:393)

fame

Introduction Measuring Fame Data Sources Analysis Strategy Results

Claim to Fame: High Mobility

Sociology of fame & celebrity claims:

Fame exhibits high mobility

The hierarchy of fame exhibits continual change.

Makes reference to popular notion of fleeting fame.

Introduction Measuring Fame Data Sources Analysis Strategy Results

Celebrity Status

“Well-known for being well-known” (Boorstin 1961)

Paris Hilton Kim Kardashian

Musical Chicago

Short public attention span to celebrities

Introduction Measuring Fame Data Sources Analysis Strategy Results

Ordinary people swiftly rise to fame,only to be replaced in the next season

Reality TV

Introduction Measuring Fame Data Sources Analysis Strategy Results

Who’s news? Who’s not?

In / Out Lists

http://www.washingtonpost.com

Introduction Measuring Fame Data Sources Analysis Strategy Results

Scholars on High Mobility (1/3)

• "Fame bubbles can burst as quickly as they formed" (Cowen 2000:15)

• “types who command media attention one day and are forgotten the next" (Rojek 2001:20-1)

• “upward and downward mobility is a continuous characteristic” (Rojek 2001:21)

• "ephemeral nature of fame" (Marshall 2004:3)• …

Introduction Measuring Fame Data Sources Analysis Strategy Results

Scholars on High Mobility (2/3)• …• “it can be attached to and detached from

individuals relatively easily." (Marshall 2004:3)• “celebrity does not usually last very long…has a

flexible association with wealth” (Ferris 2007:373)

• “increased mobility” of fame (Ferris 2007:375)• “time span between the rise and evaporation of

celebrity is getting shorter.” (Currid 2010:219)• …

Introduction Measuring Fame Data Sources Analysis Strategy Results

Scholars on High Mobility (3/3)

• …• “status on speed. It confers honor in days, not

generations; it decays over time, rather than accumulating; and it demands a constant supply of new recruits, rather than erecting barriers to entry.” (Kurzman et al. 2007:347)

• “celebrity status is likely to be less stable than more traditional forms of status.” (Milner 2010:383)

Introduction Measuring Fame Data Sources Analysis Strategy Results

Consensus without Evidence

• Claim that fame exhibits high mobility– Scholarly consensus– Concordant with popular notion of fleeting fame

• But: no systematic evidence has been acquired to confirm this claim

Fame as a Stratification Variable

incomeeducationwealthstatus

political powerhealth

fameIntroduction Measuring Fame Data Sources Analysis Strategy Results

Mechanisms for Stratification

1. Fame is associated with other resources that exhibit low mobility

2. Cumulative advantage in careers:Merton 1968 (science)Lang & Lang 1988 (visual arts)Allen & Parsons 2006 (sports)Salganik et al. 2007 (music)

3. Reinforcement in news making & habitual journalism:Molotch & Lester 1974 (habitual news making)Bielby & Bielby 1994 (recycling of past stars)Oliver & Myers 1999 (routine journalism)Vasterman 2005 (self-reinforcing themes)

Introduction Measuring Fame Data Sources Analysis Strategy Results

Research Question

Fame: Fleeting or Stratified?

Introduction Measuring Fame Data Sources Analysis Strategy Results

Overview

Measuring Fame

Data Sources

Analysis Strategy

Results

Introduction Measuring Fame Data Sources Analysis Strategy Results

Measuring Fame as Media Coverage• Close correspondence between how widely known an

individual is and how often he or she is referenced in the media

Consistent with:• Agenda-setting theory (McCombs & Shaw 1972):

Mass media determine what people judge to be important

• Sociological definition of fame (Ferris 2007): Volume, not sentiment (infamy = fame); “any publicity is good publicity”

Introduction Measuring Fame Data Sources Analysis Strategy Results

Duration of a News Item

2. Consumption: It takes 35 hours for 1/2 of an article’s readers to click on the article (Barabási 2010:49–50)

1. Production: 2/3 of a news thread occurs in 24 hours (Leskovec et al. 2009:503)

Introduction Measuring Fame Data Sources Analysis Strategy Results

Public Attention to a Person

But: The persons involved in a news item may outlive the news item by reappearing in

a later news item.

To study fame we must take the person name as unit of analysis.

Introduction Measuring Fame Data Sources Analysis Strategy Results

Operationalization of FameNumber of appearances of a name in newspaper records

Introduction Measuring Fame Data Sources Analysis Strategy Results

7x Marc Meyers

1

7

6

5

2 3 4

Data Sources

Lydia news analysis system:(details in: Bautin et al. 2010)

• Dailies corpus: For 2,500 newspapers:all person names online 2004-

• Archival corpus: For 13 newspapers:all person names in scans 1977-

Introduction Measuring Fame Data Sources Analysis Strategy Results

Data Processing

• Lydia extracts person names from text through NLP algorithms study ‘typical’ person name in the news (all mentions of a name)

• Classification of sections: entertainment, business, sports

• NLP: gender, ethnicity, sentiment, geography

Introduction Measuring Fame Data Sources Analysis Strategy Results

Analysis Challenge: Common Names

1

2

3

Introduction Measuring Fame Data Sources Analysis Strategy Results

Analysis Challenge: Common Names

Common names may refer to multiple people (e.g. ‘Michael Jackson’)

Strategy: Determine commonality via U.S. Census data.Uncommon := expected frequency in U.S. population < 1, assuming independence of first & last names 71% uncommon

Outcome: Analysis of subsample of uncommon names shows robustness of key findings

Introduction Measuring Fame Data Sources Analysis Strategy Results

NYT Fame Across Two Decades

Introduction Measuring Fame Data Sources Analysis Strategy Results

Deck Stacked in Favor ofHigh Mobility Hypothesis

• Event-based news coverage(book release, tournament, movie premiere) inflates mobility

• Newspaper format changes inflates mobility

Introduction Measuring Fame Data Sources Analysis Strategy Results

Mobility Analysis

Classic method: Mobility table

Typical unit of time: Years(e.g. Aaberge et al. 2002)

Cross-tabulate fame in one year by fame in the subsequent year

Introduction Measuring Fame Data Sources Analysis Strategy Results

Mobility Table

Fame in current year

Fame in

past year

# sentences 0 1-10 11-100 101-1000 1001+ total

1-10 67 28 5 0 0 100%

11-100 14 29 53 5 0 100%

101-1000 0 3 24 68 4 100%

1001+ 0 0 0 30 70 100%

Introduction Measuring Fame Data Sources Analysis Strategy Results

Mobility Table

Fame in current year

Fame in

past year

# sentences 0 1-10 11-100 101-1000 1001+ total

1-10 67 28 5 0 0 100%

11-100 14 29 53 5 0 100%

101-1000 0 3 24 68 4 100%

1001+ 0 0 0 30 70 100%

Introduction Measuring Fame Data Sources Analysis Strategy Results

Mobility Table

Fame in current year

Fame in

past year

# sentences 0 1-10 11-100 101-1000 1001+ total

1-10 67 28 5 0 0 100%

11-100 14 29 53 5 0 100%

101-1000 0 3 24 68 4 100%

1001+ 0 0 0 30 70 100%

Introduction Measuring Fame Data Sources Analysis Strategy Results

Mobility in Fame

Introduction Measuring Fame Data Sources Analysis Strategy Results

Mobility in Fame

Sudden loss of fame

Introduction Measuring Fame Data Sources Analysis Strategy Results

Results So Far

•Results so far:-Mobility in stratification of fame appears low

• How robust is this finding across different categories of individuals?

• Stratification of ‘celebrities’ may exhibit greater mobility than that of institutional fame (e.g. politicians)

Introduction Measuring Fame Data Sources Analysis Strategy Results

4 Ways to Identify Celebrities

1. Only names that appear in ‘tabloids’

2. Names that appear for 50+% in entertainment sections of newspapers

3. For all names, only count appearances in entertainment sections

4. Only movie actors (IMDb)Introduction Measuring Fame Data Sources Analysis Strategy Results

‘Tabloids’ in Database

6 scandal-, crime-, gossip-, fashion- or celebrity-oriented journals:

• Sun (UK)• USA Weekend• Hollywood Reporter • New York Post• New York Daily News• Women‘s Wear Daily

Introduction Measuring Fame Data Sources Analysis Strategy Results

4 Ways to Identify Celebrities

1. Only names that appear in ‘tabloids’

2. Names that appear for 50+% in entertainment sections of newspapers

3. For all names, only count appearances in entertainment sections

4. Only movie actors (IMDb)Introduction Measuring Fame Data Sources Analysis Strategy Results

Top 10 ‘Entertainers’ in SampleNames with 50+% mentions in newspaper entertainment sections:

• Jamie Foxx -- musician / actor / comedian talk radio host

• Bill Murray -- actor / comedian• Natalie Portman -- actor• Tommy Lee Jones -- actor / film director• Naomi Watts -- actor• Howard Hughes -- film producer / director / entrepreneur / aviator /

engineer• Phil Spector -- record producer / song writer• John Malkovich -- actor / producer / director / designer• Adrien Brody -- actor / film producer• Steve Buscemi -- actor / film directorIntroduction Measuring Fame Data Sources Analysis Strategy Results

4 Ways to Identify Celebrities

1. Only names that appear in ‘tabloids’

2. Names that appear for 50+% in entertainment sections of newspapers

3. For all names, only count appearances in entertainment sections

4. Only movie actors (IMDb)Introduction Measuring Fame Data Sources Analysis Strategy Results

4 Ways to Identify Celebrities

1. Only names that appear in ‘tabloids’

2. Names that appear for 50+% in entertainment sections of newspapers

3. For all names, only count appearances in entertainment sections

4. Only movie actors (IMDb)Introduction Measuring Fame Data Sources Analysis Strategy Results

Mobility among Celebrities

Introduction Measuring Fame Data Sources Analysis Strategy Results

Results So Far

•Results so far:–Mobility in stratification of fame appears low–Mobility low even among ‘celebrities’

•How robust is this finding across media?

• Blogs may exhibit greater mobility than newspapers, given open democratic access

Introduction Measuring Fame Data Sources Analysis Strategy Results

Mobility in Blogs vs. Newspapers

Introduction Measuring Fame Data Sources Analysis Strategy Results

Correlations: Blogs vs. Newspapers

Introduction Measuring Fame Data Sources Analysis Strategy Results

Results So Far•Results so far:

–Mobility in stratification of fame appears low–Mobility low even among ‘celebrities’–Mobility lower on blogs than in newspapers

• We have measured fame as annual coverage

• How stable is fame at different scales?

Introduction Measuring Fame Data Sources Analysis Strategy Results

Trade-off in Window Size

• We have measured fame as annual coverage

• Trade-off in size of window:– Too narrow even famous names fluctuate– Too wide very brief fame is not detected

• Explore narrower windows: quarters, months

Introduction Measuring Fame Data Sources Analysis Strategy Results

Stability of Fame at Different Scales

Introduction Measuring Fame Data Sources Analysis Strategy Results

Results So Far•Results so far:

–Mobility in stratification of fame appears low–Mobility low even among ‘celebrities’–Mobility lower on blogs than in newspapers–Mobility low also for shorter time windows

•We have found that fame tends to persist from month to month, quarter to quarter, and year to year

•How long does fame last?

Introduction Measuring Fame Data Sources Analysis Strategy Results

Duration of Fame

• Employ “Archival” database of select newspapers for which coverage goes back to 1977 (scans).

• Study life course of a name. Define “new names” as those that never occurred during first 5 years of our data (1977-1981). Track coverage of new names during years since birth until we hit the present (right-censored).

Introduction Measuring Fame Data Sources Analysis Strategy Results

Calculating Life Course of a Name

• Select all new names from Archival corpus that did not occur in first 5 years (1977-1981)

• Bucket names by annual coverage volume• For each year since birth calculate average

coverage of all uncensored names in bucket• Normalize such that lifetime volume sums to 1• Graph normalized coverage by age

Introduction Measuring Fame Data Sources Analysis Strategy Results

Life Course of a Name

Introduction Measuring Fame Data Sources Analysis Strategy Results

Evaluation of Hypothesis

•Hypothesis of high mobility is rejected

–Mobility in stratification of fame appears low

–Mobility low even among ‘celebrities’

–Mobility low on both blogs and newspapers

–Mobility low for shorter time windows

–Fame, even at low levels ( dozen articles p.y.), is steadyIntroduction Measuring Fame Data Sources Analysis Strategy Results

Conclusion

•Scholarly consensus on high mobility in fame finds no support in news analysis

•Fame appears no less stable in social media than in newspapers and no less so among entertainers than among politicians

•Stratification of fame is more rigid than generally believed; celebrity culture and ‘celetoids’ appear of limited impact

•Just like other stratification variables, fame remains relatively stable throughout the life course.

Thank You