Using Facebook Data to Predict 2016 US Presidential Election - - PowerPoint PPT Presentation

▶
using facebook data to predict 2016 us presidential
SMART_READER_LITE
LIVE PREVIEW

Using Facebook Data to Predict 2016 US Presidential Election - - PowerPoint PPT Presentation

Feb 25, 2024 127 likes •825 views

Using Facebook Data to Predict 2016 US Presidential Election Keng-Chi Chang Chun-Fang Chiang Ming-Jen Lin Department of Economics National Taiwan University 2018-05-29 Prepared for Innovations in Political Methodology and China Study

slide-1
SLIDE 1

Using Facebook Data to Predict 2016 US Presidential Election

Keng-Chi Chang Chun-Fang Chiang Ming-Jen Lin Department of Economics National Taiwan University 2018-05-29

Prepared for Innovations in Political Methodology and China Study International Conference

0 / 32

slide-2
SLIDE 2

In This Paper

  • Previous social media ideology measures

▸ Are mostly for elites, and uses “following” of a fan page

But people also consume news and process info. through posts

  • We use 19B “likes” on posts of 2K US fan pages to scale ideology

Also account for media, interest groups, parties, etc, and users Pages share similar ideology should share “likes” from similar users Adds time, post content, and region (guessed states) dimensions

  • We predict 2016 US presidential election using this measure

Derive state level FB support rates based on spatial model Compare with actual vote shares and state polls

  • We nd under minimal assumptions, Facebook support rates:

Predicts election quite well and shares similar trends with polls Overestimates winner’s vote share, but may enhance prediction

1 / 32

slide-3
SLIDE 3

In This Paper

  • Previous social media ideology measures

▸ Are mostly for elites, and uses “following” of a fan page ▸ But people also consume news and process info. through posts

  • We use 19B “likes” on posts of 2K US fan pages to scale ideology

Also account for media, interest groups, parties, etc, and users Pages share similar ideology should share “likes” from similar users Adds time, post content, and region (guessed states) dimensions

  • We predict 2016 US presidential election using this measure

Derive state level FB support rates based on spatial model Compare with actual vote shares and state polls

  • We nd under minimal assumptions, Facebook support rates:

Predicts election quite well and shares similar trends with polls Overestimates winner’s vote share, but may enhance prediction

1 / 32

slide-4
SLIDE 4

In This Paper

  • Previous social media ideology measures

▸ Are mostly for elites, and uses “following” of a fan page ▸ But people also consume news and process info. through posts

  • We use 19B “likes” on posts of 2K US fan pages to scale ideology

▸ Also account for media, interest groups, parties, etc, and users ▸ Pages share similar ideology should share “likes” from similar users ▸ Adds time, post content, and region (guessed states) dimensions

  • We predict 2016 US presidential election using this measure

Derive state level FB support rates based on spatial model Compare with actual vote shares and state polls

  • We nd under minimal assumptions, Facebook support rates:

Predicts election quite well and shares similar trends with polls Overestimates winner’s vote share, but may enhance prediction

1 / 32

slide-5
SLIDE 5

In This Paper

  • Previous social media ideology measures

▸ Are mostly for elites, and uses “following” of a fan page ▸ But people also consume news and process info. through posts

  • We use 19B “likes” on posts of 2K US fan pages to scale ideology

▸ Also account for media, interest groups, parties, etc, and users ▸ Pages share similar ideology should share “likes” from similar users ▸ Adds time, post content, and region (guessed states) dimensions

  • We predict 2016 US presidential election using this measure

▸ Derive state level FB support rates based on spatial model ▸ Compare with actual vote shares and state polls

  • We nd under minimal assumptions, Facebook support rates:

Predicts election quite well and shares similar trends with polls Overestimates winner’s vote share, but may enhance prediction

1 / 32

slide-6
SLIDE 6

In This Paper

  • Previous social media ideology measures

▸ Are mostly for elites, and uses “following” of a fan page ▸ But people also consume news and process info. through posts

  • We use 19B “likes” on posts of 2K US fan pages to scale ideology

▸ Also account for media, interest groups, parties, etc, and users ▸ Pages share similar ideology should share “likes” from similar users ▸ Adds time, post content, and region (guessed states) dimensions

  • We predict 2016 US presidential election using this measure

▸ Derive state level FB support rates based on spatial model ▸ Compare with actual vote shares and state polls

  • We nd under minimal assumptions, Facebook support rates:

▸ Predicts election quite well and shares similar trends with polls ▸ Overestimates winner’s vote share, but may enhance prediction 1 / 32

slide-7
SLIDE 7

Facebook Data

  • Facebook provides fan page data through Graph API
  • Specify fan page ideological universe

1475 fan pages of national politicians Members and candidates of Senate, House, and Governors Top 1000 pages related to 2016 presidential election In Aug 2016, nd all pages mentioned “Trump” and “Clinton” Weight by likes, comments, shares, nd top 1000 pages Includes all major news outlets, interest groups, parties, etc NYT, Fox News, NRA, RNC, Occupy Wall St, Tea Party, 9GAG, ...

  • Collect all 24M posts in 2015 and 2016 on these pages
  • And user’s 19B reactions (mostly likes) to these posts

2 / 32

slide-8
SLIDE 8

Facebook Data

  • Facebook provides fan page data through Graph API
  • Specify fan page ideological universe

▸ 1475 fan pages of national politicians

↝ Members and candidates of Senate, House, and Governors Top 1000 pages related to 2016 presidential election In Aug 2016, nd all pages mentioned “Trump” and “Clinton” Weight by likes, comments, shares, nd top 1000 pages Includes all major news outlets, interest groups, parties, etc NYT, Fox News, NRA, RNC, Occupy Wall St, Tea Party, 9GAG, ...

  • Collect all 24M posts in 2015 and 2016 on these pages
  • And user’s 19B reactions (mostly likes) to these posts

2 / 32

slide-9
SLIDE 9

Facebook Data

  • Facebook provides fan page data through Graph API
  • Specify fan page ideological universe

▸ 1475 fan pages of national politicians

↝ Members and candidates of Senate, House, and Governors

▸ Top 1000 pages related to 2016 presidential election

↝ In Aug 2016, nd all pages mentioned “Trump” and “Clinton” ↝ Weight by likes, comments, shares, nd top 1000 pages Includes all major news outlets, interest groups, parties, etc NYT, Fox News, NRA, RNC, Occupy Wall St, Tea Party, 9GAG, ...

  • Collect all 24M posts in 2015 and 2016 on these pages
  • And user’s 19B reactions (mostly likes) to these posts

2 / 32

slide-10
SLIDE 10

Facebook Data

  • Facebook provides fan page data through Graph API
  • Specify fan page ideological universe

▸ 1475 fan pages of national politicians

↝ Members and candidates of Senate, House, and Governors

▸ Top 1000 pages related to 2016 presidential election

↝ In Aug 2016, nd all pages mentioned “Trump” and “Clinton” ↝ Weight by likes, comments, shares, nd top 1000 pages ↝ Includes all major news outlets, interest groups, parties, etc ↝ NYT, Fox News, NRA, RNC, Occupy Wall St, Tea Party, 9GAG, ...

  • Collect all 24M posts in 2015 and 2016 on these pages
  • And user’s 19B reactions (mostly likes) to these posts

2 / 32

slide-11
SLIDE 11

Facebook Data

  • Facebook provides fan page data through Graph API
  • Specify fan page ideological universe

▸ 1475 fan pages of national politicians

↝ Members and candidates of Senate, House, and Governors

▸ Top 1000 pages related to 2016 presidential election

↝ In Aug 2016, nd all pages mentioned “Trump” and “Clinton” ↝ Weight by likes, comments, shares, nd top 1000 pages ↝ Includes all major news outlets, interest groups, parties, etc ↝ NYT, Fox News, NRA, RNC, Occupy Wall St, Tea Party, 9GAG, ...

  • Collect all 24M posts in 2015 and 2016 on these pages
  • And user’s 19B reactions (mostly likes) to these posts

2 / 32

slide-12
SLIDE 12

Facebook Data

  • Facebook provides fan page data through Graph API
  • Specify fan page ideological universe

▸ 1475 fan pages of national politicians

↝ Members and candidates of Senate, House, and Governors

▸ Top 1000 pages related to 2016 presidential election

↝ In Aug 2016, nd all pages mentioned “Trump” and “Clinton” ↝ Weight by likes, comments, shares, nd top 1000 pages ↝ Includes all major news outlets, interest groups, parties, etc ↝ NYT, Fox News, NRA, RNC, Occupy Wall St, Tea Party, 9GAG, ...

  • Collect all 24M posts in 2015 and 2016 on these pages
  • And user’s 19B reactions (mostly likes) to these posts

2 / 32

slide-13
SLIDE 13

Data Summary

Time Period 2015-01-01 to 2016-11-30 Total Reactions 19,085,783,534 US Political User Likes 16,180,488,916 Total Users 366,840,068 US Political Users 29,412,610 Total Posts 24,788,093 Total Pages 2132 Politicians 1225 News Outlets 560 Political Groups 211 Other Public Figures 93 Others 43

3 / 32

slide-14
SLIDE 14

Estimation: Shared Users Matrix

  • Measure ideology of pages, then measure those of users

↝ Similar to Bond and Messing (2015, APSR)

  • First build the page by page afliation matrix

Number of shared users (based on likes) between pages

Trump FoxNews TeaParty Clinton CNN NYTimes Trump 2243216 1078513 128225 32731 120963 25842 FoxNews 1078513 2449174 148016 87084 186850 63401 TeaParty 128225 148016 242089 1528 10738 2162 Clinton 32731 87084 1528 1768980 351210 367021 CNN 120963 186850 10738 351210 1201156 216163 NYTimes 25842 63401 2162 367021 216163 986613

4 / 32

slide-15
SLIDE 15

Estimation: Shared Users Matrix

  • Measure ideology of pages, then measure those of users

↝ Similar to Bond and Messing (2015, APSR)

  • First build the page by page afliation matrix A

↝ Number of shared users (based on likes) between pages

Trump FoxNews TeaParty Clinton CNN NYTimes Trump 2243216 1078513 128225 32731 120963 25842 FoxNews 1078513 2449174 148016 87084 186850 63401 TeaParty 128225 148016 242089 1528 10738 2162 Clinton 32731 87084 1528 1768980 351210 367021 CNN 120963 186850 10738 351210 1201156 216163 NYTimes 25842 63401 2162 367021 216163 986613

4 / 32

slide-16
SLIDE 16

Estimation: Transform to Ratios

  • Transform A to matrix of ratios G, where gij = ai j/aii

↝ 0.44 = Pr(Trump ∩ FoxNews) Pr(FoxNews) = Pr(Trump∣FoxNews)

  • Can interpret columns as features and rows as observations

↝ Col 1 is how each row similar to “Trump” feature

Trump FoxNews TeaParty Clinton CNN NYTimes Trump 1.00 0.48 0.06 0.01 0.05 0.01 FoxNews 0.44 1.00 0.06 0.04 0.08 0.03 TeaParty 0.53 0.61 1.00 0.01 0.04 0.01 Clinton 0.02 0.05 0.00 1.00 0.20 0.21 CNN 0.10 0.16 0.01 0.29 1.00 0.18 NYTimes 0.03 0.06 0.00 0.37 0.22 1.00

5 / 32

slide-17
SLIDE 17

Estimation: Dimension Reduction

  • Compute the principal components of G after standardizing
  • PC1 is the dimension explains the largest variation

Unsupervised Guess and verify PC1 is related to “ideology”

  • User ideology

mean ideology of pages user liked

  • Guess user’s state residence by their likes on national politicians

Like more politicians from NY More likely from NY

6 / 32

slide-18
SLIDE 18

Estimation: Dimension Reduction

  • Compute the principal components of G after standardizing
  • PC1 is the dimension explains the largest variation

↝ Unsupervised ⇒ Guess and verify PC1 is related to “ideology”

  • User ideology

mean ideology of pages user liked

  • Guess user’s state residence by their likes on national politicians

Like more politicians from NY More likely from NY

6 / 32

slide-19
SLIDE 19

Estimation: Dimension Reduction

  • Compute the principal components of G after standardizing
  • PC1 is the dimension explains the largest variation

↝ Unsupervised ⇒ Guess and verify PC1 is related to “ideology”

  • User ideology = mean ideology of pages user liked
  • Guess user’s state residence by their likes on national politicians

↝ Like more politicians from NY ⇒ More likely from NY

6 / 32

slide-20
SLIDE 20

5 10 15 20 0.00 0.02 0.04 0.06

Scree Plot for Principal Component Analysis

k-th Principal Component Proportion of Variance Explained

7 / 32

slide-21
SLIDE 21

Trump Clinton NYTimes Fox News

0.00 0.25 0.50 0.75

  • 2
  • 1

1 2

Estimated Facebook Ideology Score Density

Public Figure Political Groups News Outlets

8 / 32

slide-22
SLIDE 22

The New York Times Politics and Washington The New York Times Opinion Section USA TODAY The New York Times The Wall Street Journal Washington Post Chicago Tribune Boston Herald The Christian Post The Washington Times 0.0 0.3 0.6 0.9

  • 1.5
  • 1.0
  • 0.5

0.0 0.5 1.0

PC1 (First Principal Component) Density

PC1 Density of Newspaper Pages

9 / 32

slide-23
SLIDE 23

CNN Breitbart MSNBC PBS Fox News Opinion The Rachel Maddow Fan Page. Fox News NRA News The Federalist Papers ABC News 0.0 0.5 1.0 1.5 2.0

  • 2
  • 1

1 2

PC1 (First Principal Component) Density

type_sub

radio tv website

PC1 Density of TV, Radio, Website Pages

10 / 32

slide-24
SLIDE 24

Validation for Congressional Politicians

ρ = 0.92 ρR = 0.50 ρD = 0.22

Schumer McConnell McCain Pelosi Sanders Ryan Rubio Warren Cruz Booker

  • 1.0
  • 0.5

0.0 0.5 1.0

  • 2
  • 1

1 2

Estimated Facebook Page Ideology Score, 2015-01 to 2016-11 DW-Nominate Score of 114th Congress

Democratic Party Independent Republican Party

Using politician and top 1000 page matrix

11 / 32

slide-25
SLIDE 25

Validation for Media

New Yorker New Republic NYTimes WashPost MSNBC BuzzFeed Politico CNN WSJ ABC News USA Today The Hill WSJ Opinion AWM Fox News National Review The Blaze Breitbart 0.5 1

  • 2
  • 1

1 2

Estimated Facebook Ideology Score Share of Republican-Affiliated User

Magazine Newspaper Radio TV Website

12 / 32

slide-26
SLIDE 26

User Ideology Density by States

0.0 0.5 1.0 1.5 2.0

  • 1
  • 0.5

0.5 1

Massachusetts

0.00 0.25 0.50 0.75 1.00

  • 1
  • 0.5

0.5 1

Washington

0.0 0.2 0.4 0.6

  • 1
  • 0.5

0.5 1

Michigan

0.0 0.2 0.4 0.6

  • 1
  • 0.5

0.5 1

Pennsylvania

0.0 0.3 0.6 0.9 1.2

  • 1
  • 0.5

0.5 1

Texas

0.00 0.25 0.50 0.75 1.00

  • 1
  • 0.5

0.5 1

Wyoming 13 / 32

slide-27
SLIDE 27

User Ideology Density by States

10 20 30

  • 1
  • 0.5

0.5 1

Massachusetts

0.0 0.5 1.0 1.5 2.0 2.5

  • 1
  • 0.5

0.5 1

Washington

1 2

  • 1
  • 0.5

0.5 1

Michigan

0.0 0.5 1.0 1.5

  • 1
  • 0.5

0.5 1

Pennsylvania

2 4 6 8

  • 1
  • 0.5

0.5 1

Texas

2 4 6

  • 1
  • 0.5

0.5 1

Wyoming

Politician-Only Method (Bond and Messing 2015)

14 / 32

slide-28
SLIDE 28

Media Ideology Dynamics

ThinkProgress.com MSNBC Washington Post New York Times Bloomberg CNN WSJ ABC News Fox News Breitbart Federalist Papers NRA News CNSNews.com

  • 1

1 2 06-01 2015 08-01 2015 10-01 2015 12-01 2015 02-01 2016 04-01 2016 06-01 2016 08-01 2016 10-01 2016 12-01 2016 02-01 2017 04-01 2017

Estimated Facebook Ideology Score 15 / 32

slide-29
SLIDE 29

Politician Ideology Dynamics

Warren Clinton Sanders Johnson Trump Ryan Rubio Cruz Warren Sanders Clinton Johnson Trump Ryan Rubio Cruz

  • 1.0
  • 0.5

0.0 0.5 1.0 05-01 2015 07-01 2015 09-01 2015 11-01 2015 01-01 2016 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016 01-01 2017

Estimated Facebook Ideology Score

16 / 32

slide-30
SLIDE 30

FB Support Rates, Polls, and Vote Shares

  • Apply the Hotelling-Downs spatial model for voting: Voters

support candidates closer to their own ideological location

  • In each state, we compare:

FB support rate: Share of user’s ideology closer to Trump or Clinton May not be a precise estimator for vote shares Since turnouts may not be the same across states But adding assumptions may look like tting the data Polls: State polling averages calculated by FiveThirtyEight Actual vote shares in 2016 election

17 / 32

slide-31
SLIDE 31

FB Support Rates, Polls, and Vote Shares

  • Apply the Hotelling-Downs spatial model for voting: Voters

support candidates closer to their own ideological location

  • In each state, we compare:

▸ FB support rate: Share of user’s ideology closer to Trump or Clinton

May not be a precise estimator for vote shares Since turnouts may not be the same across states But adding assumptions may look like tting the data Polls: State polling averages calculated by FiveThirtyEight Actual vote shares in 2016 election

17 / 32

slide-32
SLIDE 32

FB Support Rates, Polls, and Vote Shares

  • Apply the Hotelling-Downs spatial model for voting: Voters

support candidates closer to their own ideological location

  • In each state, we compare:

▸ FB support rate: Share of user’s ideology closer to Trump or Clinton

↝ May not be a precise estimator for vote shares ↝ Since turnouts may not be the same across states ↝ But adding assumptions may look like tting the data Polls: State polling averages calculated by FiveThirtyEight Actual vote shares in 2016 election

17 / 32

slide-33
SLIDE 33

FB Support Rates, Polls, and Vote Shares

  • Apply the Hotelling-Downs spatial model for voting: Voters

support candidates closer to their own ideological location

  • In each state, we compare:

▸ FB support rate: Share of user’s ideology closer to Trump or Clinton

↝ May not be a precise estimator for vote shares ↝ Since turnouts may not be the same across states ↝ But adding assumptions may look like tting the data

▸ Polls: State polling averages calculated by FiveThirtyEight

Actual vote shares in 2016 election

17 / 32

slide-34
SLIDE 34

FB Support Rates, Polls, and Vote Shares

  • Apply the Hotelling-Downs spatial model for voting: Voters

support candidates closer to their own ideological location

  • In each state, we compare:

▸ FB support rate: Share of user’s ideology closer to Trump or Clinton

↝ May not be a precise estimator for vote shares ↝ Since turnouts may not be the same across states ↝ But adding assumptions may look like tting the data

▸ Polls: State polling averages calculated by FiveThirtyEight ▸ Actual vote shares in 2016 election 17 / 32

slide-35
SLIDE 35

Predicting Vote Shares and Outcomes

AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY

ρ = 0.73 95% CI [0.56, 0.84] 30 40 50 60 25 50 75

Share of Facebook User Closer to Clinton (10-01 to 11-07) 2016 Clinton Vote Share

Rep wins 2016 & 2012 Swings from Obama to Trump Dem wins 2016 & 2012

18 / 32

slide-36
SLIDE 36

Compare with Major Forecasters

Battleground States E.V.† Winner FB 538 NYT PEC* Florida 29 Trump Pennsylvania 20 Trump Wisconsin 10 Trump Michigan 16 Trump Ohio 18 Trump Iowa 6 Trump Montana 3 Trump Alaska 3 Clinton Maine 2 Clinton Trump’s Electoral Vote 306 292 235 216 215

† Electoral Votes. * Princeton Election Consortium.

19 / 32

slide-37
SLIDE 37

Compare with Major Forecasters

Battleground States E.V.† Winner FB 538 NYT PEC* Florida 29 Trump ○ × × × Pennsylvania 20 Trump ○ × × × Wisconsin 10 Trump ○ × × × Michigan 16 Trump Ohio 18 Trump Iowa 6 Trump Montana 3 Trump Alaska 3 Clinton Maine 2 Clinton Trump’s Electoral Vote 306 292 235 216 215

† Electoral Votes. * Princeton Election Consortium.

19 / 32

slide-38
SLIDE 38

Compare with Major Forecasters

Battleground States E.V.† Winner FB 538 NYT PEC* Florida 29 Trump ○ × × × Pennsylvania 20 Trump ○ × × × Wisconsin 10 Trump ○ × × × Michigan 16 Trump × × × × Ohio 18 Trump Iowa 6 Trump Montana 3 Trump Alaska 3 Clinton Maine 2 Clinton Trump’s Electoral Vote 306 292 235 216 215

† Electoral Votes. * Princeton Election Consortium.

19 / 32

slide-39
SLIDE 39

Compare with Major Forecasters

Battleground States E.V.† Winner FB 538 NYT PEC* Florida 29 Trump ○ × × × Pennsylvania 20 Trump ○ × × × Wisconsin 10 Trump ○ × × × Michigan 16 Trump × × × × Ohio 18 Trump ○ ○ ○ ○ Iowa 6 Trump ○ ○ ○ ○ Montana 3 Trump Alaska 3 Clinton Maine 2 Clinton Trump’s Electoral Vote 306 292 235 216 215

† Electoral Votes. * Princeton Election Consortium.

19 / 32

slide-40
SLIDE 40

Compare with Major Forecasters

Battleground States E.V.† Winner FB 538 NYT PEC* Florida 29 Trump ○ × × × Pennsylvania 20 Trump ○ × × × Wisconsin 10 Trump ○ × × × Michigan 16 Trump × × × × Ohio 18 Trump ○ ○ ○ ○ Iowa 6 Trump ○ ○ ○ ○ Montana 3 Trump × ○ ○ ○ Alaska 3 Clinton × ○ ○ ○ Maine 2 Clinton × ○ ○ ○ Trump’s Electoral Vote 306 292 235 216 215

† Electoral Votes. * Princeton Election Consortium.

19 / 32

slide-41
SLIDE 41

Compare with Major Forecasters

Battleground States E.V.† Winner FB 538 NYT PEC* Florida 29 Trump ○ × × × Pennsylvania 20 Trump ○ × × × Wisconsin 10 Trump ○ × × × Michigan 16 Trump × × × × Ohio 18 Trump ○ ○ ○ ○ Iowa 6 Trump ○ ○ ○ ○ Montana 3 Trump × ○ ○ ○ Alaska 3 Clinton × ○ ○ ○ Maine 2 Clinton × ○ ○ ○ Trump’s Electoral Vote 306 292 235 216 215

† Electoral Votes. * Princeton Election Consortium.

19 / 32

slide-42
SLIDE 42

Predicting Electoral Votes

Facebook FiveThirtyEight Even Actual

150 200 250 270 300 350 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Predicted Electoral Votes for Trump

20 / 32

slide-43
SLIDE 43

Trump: FB (Dotted), Polls, and Vote Shares

40 45 50 55 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Pennsylvania

45 50 55 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Ohio

40 45 50 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Nevada

40 50 60 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Maine

40.0 42.5 45.0 47.5 50.0 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

New Hampshire

45 50 55 60 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Florida

40 44 48 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Michigan

40 50 60 70 80 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Wisconsin

40 50 60 70 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Iowa 21 / 32

slide-44
SLIDE 44

Clinton: FB (Dotted), Polls, and Vote Shares

25 30 35 40 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Alabama

20 25 30 35 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

West Virginia

20 25 30 35 40 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Kentucky

25 30 35 40 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Tennessee

20 25 30 35 40 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Kansas

20 25 30 35 40 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Arkansas

30 33 36 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Idaho

33 36 39 42 45 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

Missouri

30 40 50 60 03-01 2016 05-01 2016 07-01 2016 09-01 2016 11-01 2016

North Dakota 22 / 32

slide-45
SLIDE 45

Polls Overestimates Clinton in Red and Swing States

AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY

  • 5

5

  • 20

20 40

Clinton: Facebook Support - Actual Vote Share Clinton: 538 Vote Share - Actual Vote Share

Dem wins 2016 & 2012 Rep wins 2016 & 2012 Swings from Obama to Trump

23 / 32

slide-46
SLIDE 46

FB Overestimates Trump in Red and Swing States

AL AK AZ AR CA CO CT DE FL GA HI ID IL IN IA KS KY LA ME MD MA MI MN MS MO MT NE NV NH NJ NM NY NC ND OH OK OR PA RI SC SD TN TX UT VT VA WA WV WI WY

  • 5
  • 40
  • 20

20

Trump: Facebook Support - Actual Vote Share Trump: 538 Vote Share - Actual Vote Share

Dem wins 2016 & 2012 Rep wins 2016 & 2012 Swings from Obama to Trump

24 / 32

slide-47
SLIDE 47

Discussions

  • Strengths of Facebook based prediction:

▸ Revealed preference instead of self-report ▸ Low cost and almost in real time ▸ Trace individuals repeatedly over time ▸ Overestimation for winners can help to make predictions

  • Weaknesses, compared to polls or surveys:

Not representative Can reweight if more social-demographic information is known Hard to link with ofine behaviors

  • Ex. “Strong supporter” vs. “Likely voter”
  • Can complement each other if more research try to link the two

25 / 32

slide-48
SLIDE 48

Discussions

  • Strengths of Facebook based prediction:

▸ Revealed preference instead of self-report ▸ Low cost and almost in real time ▸ Trace individuals repeatedly over time ▸ Overestimation for winners can help to make predictions

  • Weaknesses, compared to polls or surveys:

▸ Not representative

↝ Can reweight if more social-demographic information is known

▸ Hard to link with ofine behaviors

↝ Ex. “Strong supporter” vs. “Likely voter”

  • Can complement each other if more research try to link the two

25 / 32

slide-49
SLIDE 49

Discussions

  • Strengths of Facebook based prediction:

▸ Revealed preference instead of self-report ▸ Low cost and almost in real time ▸ Trace individuals repeatedly over time ▸ Overestimation for winners can help to make predictions

  • Weaknesses, compared to polls or surveys:

▸ Not representative

↝ Can reweight if more social-demographic information is known

▸ Hard to link with ofine behaviors

↝ Ex. “Strong supporter” vs. “Likely voter”

  • Can complement each other if more research try to link the two

25 / 32

slide-50
SLIDE 50

Working on: Effect of Fake News

  • Joint with Chun-Fang Chiang, Brian Knight, and Ming-Jen Lin
  • Would consuming fake news change people’s ideology or

information consumption?

  • If so, what kind of fake stories have larger effect, and why?
  • Fake news pool on Facebook:

Top 40 fake stories, 536 posts, 130 pages Posts link to fake domains, 139,074 posts, 177 pages

26 / 32

slide-51
SLIDE 51

Working on: Effect of Fake News

  • Joint with Chun-Fang Chiang, Brian Knight, and Ming-Jen Lin
  • Would consuming fake news change people’s ideology or

information consumption?

  • If so, what kind of fake stories have larger effect, and why?
  • Fake news pool on Facebook:

Top 40 fake stories, 536 posts, 130 pages Posts link to fake domains, 139,074 posts, 177 pages

26 / 32

slide-52
SLIDE 52

Working on: Effect of Fake News

  • Joint with Chun-Fang Chiang, Brian Knight, and Ming-Jen Lin
  • Would consuming fake news change people’s ideology or

information consumption?

  • If so, what kind of fake stories have larger effect, and why?
  • Fake news pool on Facebook:

▸ Top 40 fake stories, 536 posts, 130 pages ▸ Posts link to fake domains, 139,074 posts, 177 pages 26 / 32

slide-53
SLIDE 53

1 2 3

  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 2

Ideology (2016 Jan to Apr) Density

Users Do Not Like Fake Post Users Like Fake Post

Individual Ideology

27 / 32

slide-54
SLIDE 54

2 4 6 8

  • 0.5
  • 0.25

0.25 0.5

Ideology Difference (2016 Jul to Nov - 2016 Jan to Apr) Density

Users Do Not Like Fake Post Users Like Fake Post

Individual Ideology Differnece

28 / 32

slide-55
SLIDE 55

Strategies for Identication

  • Challenges:

▸ People “like” fake post may be very different ▸ Pages posting fake posts may attract very different users ▸ Some stories may be “too fake” for people to believe, even backre

  • For each fake post, we:

Find nonfake pages very similar to fake page through different matching methods as control Find potential followers of these pages, instead of “likes” Compare the ideology of these fake and nonfake followers before and after fake page unexpectedly started posting fake story

1 1 1 1

29 / 32

slide-56
SLIDE 56

Strategies for Identication

  • Challenges:

▸ People “like” fake post may be very different ▸ Pages posting fake posts may attract very different users ▸ Some stories may be “too fake” for people to believe, even backre

  • For each fake post, we:

▸ Find nonfake pages very similar to fake page through different

matching methods as control Find potential followers of these pages, instead of “likes” Compare the ideology of these fake and nonfake followers before and after fake page unexpectedly started posting fake story

1 1 1 1

29 / 32

slide-57
SLIDE 57

Strategies for Identication

  • Challenges:

▸ People “like” fake post may be very different ▸ Pages posting fake posts may attract very different users ▸ Some stories may be “too fake” for people to believe, even backre

  • For each fake post, we:

▸ Find nonfake pages very similar to fake page through different

matching methods as control

▸ Find potential followers of these pages, instead of “likes”

Compare the ideology of these fake and nonfake followers before and after fake page unexpectedly started posting fake story

1 1 1 1

29 / 32

slide-58
SLIDE 58

Strategies for Identication

  • Challenges:

▸ People “like” fake post may be very different ▸ Pages posting fake posts may attract very different users ▸ Some stories may be “too fake” for people to believe, even backre

  • For each fake post, we:

▸ Find nonfake pages very similar to fake page through different

matching methods as control

▸ Find potential followers of these pages, instead of “likes” ▸ Compare the ideology of these fake and nonfake followers before

and after fake page unexpectedly started posting fake story

1 1 1 1

29 / 32

slide-59
SLIDE 59

Strategies for Identication

  • Challenges:

▸ People “like” fake post may be very different ▸ Pages posting fake posts may attract very different users ▸ Some stories may be “too fake” for people to believe, even backre

  • For each fake post, we:

▸ Find nonfake pages very similar to fake page through different

matching methods as control

▸ Find potential followers of these pages, instead of “likes” ▸ Compare the ideology of these fake and nonfake followers before

and after fake page unexpectedly started posting fake story

Ideologyit = α 1(Afuert) + γ 1(FollowFakei) + β 1(FollowFakei)1(Afuert) + εit

29 / 32

slide-60
SLIDE 60

Users Follow Fake Page (Treatment) Users Not Follow Fake Page (Control) Parallel Line

0.65 0.70 0.75 0.80 0.85

  • 2

2 Week Aer Post Mean Follower Ideology

"BREAKING: Official Set to Testify Against Hillary Found Dead" by Western Journalism

30 / 32

slide-61
SLIDE 61

Story Level Ideology Change for Following Pages Sharing Pro−Trump Fake News Week +1 to −1, DiD Estimates with Individual Fixed Effects and 99.9% CI

5−NN Matching

  • 10−NN Matching

5−Nearest PS Matching 10−Nearest PS Matching Pentagon furious Clinton nuclear response time Clinton financial connection to Saudi Arabia Wikileaks: Clinton sold weapons to ISIS Pope Francis endorses Trump Trump sends own plane to transport marines Obama refuses to leave office if Trump elected Clinton HIV secret revealed Clinton goes to Texas Muslim fundraiser Associate to testify against Clinton dead Stanford University: Dem election fraud Trump protester: I was paid to protest Official to testifiy against Clinton dead Uncounted Sanders ballots on Clinton server Clinton ISIS email leaked ISIS leader calls voters support Clinton Clinton disqualified holding Federal office Clinton tells nuclear launch response time Bill Clinton 2000 sex partners, Hillary lesbian Billy Graham STUNNING statement on Trump Putin: Emails reveal Clinton threatens Sanders Graham: Christians must support Trump Clinton email reopens, Comey asks immunity Clinton to be indicted, prayers answered

−0.05 −0.025 0.025 0.05 0.075

  • 31 / 32
slide-62
SLIDE 62

Story Level Ideology Change for Following Pages Sharing Pro−Trump Fake News Week +1 to −1, DiD Estimates with Individual Fixed Effects and 99.9% CI

5−NN Matching

  • 10−NN Matching

5−Nearest PS Matching 10−Nearest PS Matching Pentagon furious Clinton nuclear response time Clinton financial connection to Saudi Arabia Wikileaks: Clinton sold weapons to ISIS Pope Francis endorses Trump Trump sends own plane to transport marines Obama refuses to leave office if Trump elected Clinton HIV secret revealed Clinton goes to Texas Muslim fundraiser Associate to testify against Clinton dead Stanford University: Dem election fraud Trump protester: I was paid to protest Official to testifiy against Clinton dead Uncounted Sanders ballots on Clinton server Clinton ISIS email leaked ISIS leader calls voters support Clinton Clinton disqualified holding Federal office Clinton tells nuclear launch response time Bill Clinton 2000 sex partners, Hillary lesbian Billy Graham STUNNING statement on Trump Putin: Emails reveal Clinton threatens Sanders Graham: Christians must support Trump Clinton email reopens, Comey asks immunity Clinton to be indicted, prayers answered

−0.05 −0.025 0.025 0.05 0.075

  • 31 / 32
slide-63
SLIDE 63

Story Level Ideology Change for Following Pages Sharing Pro−Trump Fake News Week +1 to −1, DiD Estimates with Individual Fixed Effects and 99.9% CI

5−NN Matching

  • 10−NN Matching

5−Nearest PS Matching 10−Nearest PS Matching Pentagon furious Clinton nuclear response time Clinton financial connection to Saudi Arabia Wikileaks: Clinton sold weapons to ISIS Pope Francis endorses Trump Trump sends own plane to transport marines Obama refuses to leave office if Trump elected Clinton HIV secret revealed Clinton goes to Texas Muslim fundraiser Associate to testify against Clinton dead Stanford University: Dem election fraud Trump protester: I was paid to protest Official to testifiy against Clinton dead Uncounted Sanders ballots on Clinton server Clinton ISIS email leaked ISIS leader calls voters support Clinton Clinton disqualified holding Federal office Clinton tells nuclear launch response time Bill Clinton 2000 sex partners, Hillary lesbian Billy Graham STUNNING statement on Trump Putin: Emails reveal Clinton threatens Sanders Graham: Christians must support Trump Clinton email reopens, Comey asks immunity Clinton to be indicted, prayers answered

−0.05 −0.025 0.025 0.05 0.075

  • 31 / 32
slide-64
SLIDE 64

Story Level Ideology Change for Following Pages Sharing Pro−Trump Fake News Week +1 to −1, DiD Estimates with Individual Fixed Effects and 99.9% CI

5−NN Matching

  • 10−NN Matching

5−Nearest PS Matching 10−Nearest PS Matching Pentagon furious Clinton nuclear response time Clinton financial connection to Saudi Arabia Wikileaks: Clinton sold weapons to ISIS Pope Francis endorses Trump Trump sends own plane to transport marines Obama refuses to leave office if Trump elected Clinton HIV secret revealed Clinton goes to Texas Muslim fundraiser Associate to testify against Clinton dead Stanford University: Dem election fraud Trump protester: I was paid to protest Official to testifiy against Clinton dead Uncounted Sanders ballots on Clinton server Clinton ISIS email leaked ISIS leader calls voters support Clinton Clinton disqualified holding Federal office Clinton tells nuclear launch response time Bill Clinton 2000 sex partners, Hillary lesbian Billy Graham STUNNING statement on Trump Putin: Emails reveal Clinton threatens Sanders Graham: Christians must support Trump Clinton email reopens, Comey asks immunity Clinton to be indicted, prayers answered

−0.05 −0.025 0.025 0.05 0.075

  • 31 / 32
slide-65
SLIDE 65

Story Level Ideology Change for Following Pages Sharing Pro−Clinton Fake News Week +1 to −1, DiD Estimates with Individual Fixed Effects and 99.9% CI

5−NN Matching

  • 10−NN Matching

5−Nearest PS Matching 10−Nearest PS Matching Palin to become Trump VP Rupaul: Trump touched me inappropriately Trump critical condition choking own bullshit Ireland accepts Trump refugees Rage Against the Machine anti Trump album Trump: Giving Canada independence mistake Trump U offers Palin honorary climate degree Mexico will close border if Trump elected Trump: I will overtern shocking gay marriage Trump picks Stacey Dash as VP Pence: Michelle Obama most vulgar FLOTUS Palin endorses Cruz Sauron endorses Trump Trump sues Chicago after forced to cancel rally

−0.075 −0.05 −0.025 0.025

  • 32 / 32
slide-66
SLIDE 66

Story Level Ideology Change for Following Pages Sharing Pro−Clinton Fake News Week +1 to −1, DiD Estimates with Individual Fixed Effects and 99.9% CI

5−NN Matching

  • 10−NN Matching

5−Nearest PS Matching 10−Nearest PS Matching Palin to become Trump VP Rupaul: Trump touched me inappropriately Trump critical condition choking own bullshit Ireland accepts Trump refugees Rage Against the Machine anti Trump album Trump: Giving Canada independence mistake Trump U offers Palin honorary climate degree Mexico will close border if Trump elected Trump: I will overtern shocking gay marriage Trump picks Stacey Dash as VP Pence: Michelle Obama most vulgar FLOTUS Palin endorses Cruz Sauron endorses Trump Trump sues Chicago after forced to cancel rally

−0.075 −0.05 −0.025 0.025

  • 32 / 32
slide-67
SLIDE 67

Story Level Ideology Change for Following Pages Sharing Pro−Clinton Fake News Week +1 to −1, DiD Estimates with Individual Fixed Effects and 99.9% CI

5−NN Matching

  • 10−NN Matching

5−Nearest PS Matching 10−Nearest PS Matching Palin to become Trump VP Rupaul: Trump touched me inappropriately Trump critical condition choking own bullshit Ireland accepts Trump refugees Rage Against the Machine anti Trump album Trump: Giving Canada independence mistake Trump U offers Palin honorary climate degree Mexico will close border if Trump elected Trump: I will overtern shocking gay marriage Trump picks Stacey Dash as VP Pence: Michelle Obama most vulgar FLOTUS Palin endorses Cruz Sauron endorses Trump Trump sues Chicago after forced to cancel rally

−0.075 −0.05 −0.025 0.025

  • 32 / 32
slide-68
SLIDE 68

Story Level Ideology Change for Following Pages Sharing Pro−Clinton Fake News Week +1 to −1, DiD Estimates with Individual Fixed Effects and 99.9% CI

5−NN Matching

  • 10−NN Matching

5−Nearest PS Matching 10−Nearest PS Matching Palin to become Trump VP Rupaul: Trump touched me inappropriately Trump critical condition choking own bullshit Ireland accepts Trump refugees Rage Against the Machine anti Trump album Trump: Giving Canada independence mistake Trump U offers Palin honorary climate degree Mexico will close border if Trump elected Trump: I will overtern shocking gay marriage Trump picks Stacey Dash as VP Pence: Michelle Obama most vulgar FLOTUS Palin endorses Cruz Sauron endorses Trump Trump sues Chicago after forced to cancel rally

−0.075 −0.05 −0.025 0.025

  • 32 / 32