voter mobilization effects of poll reports during the 2012 ... · pdf filevoter mobilization...

Post on 11-Mar-2018

216 Views

Category:

Documents

2 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Voter Mobilization Effects of Poll Reports During the 2012 Presidential

Campaign

D AV I D L . V A N N E T T E & S E A N J . W E S T W O O D S TA N F O R D U N I V E R S I T Y

D E PA R T M E N T O F C O M M U N I C AT I O N

68th Annual AAPOR Conference – May 17, 2013

Agenda INTRODUCTION   Poll effects   Polls in the 2012 election CURRENT PROJECT   Design   Data RESULTS CONCLUSIONS LIMITATIONS

2

Poll effects WHAT ARE POLL EFFECTS?

›  Bandwagons ›  Underdogs

3

Poll effects POSITIVE EFFECTS OF POLLS? NEGATIVE EFFECTS OF POLLS?

4

What can we hope for? POSITIVE EFFECTS OF POLL REPORTS?   Impersonal influence (Mutz,1998)

›  Thoughtful reflection?

5

What can we hope for? POSITIVE EFFECTS OF POLL REPORTS?   Impersonal influence (Mutz,1998)

›  Thoughtful reflection?   Stimulate information-seeking?

6

What can we hope for? POSITIVE EFFECTS OF POLL REPORTS?   Impersonal influence (Mutz,1998)

›  Thoughtful reflection?   Stimulate information-seeking?   Increase political engagement?

7

What can we hope for? POSITIVE EFFECTS OF POLL REPORTS?   Impersonal influence (Mutz,1998)

›  Thoughtful reflection?   Stimulate information-seeking?   Increase political engagement?   Increase turnout?

8

What should we fear? NEGATIVE EFFECTS OF POLL REPORTS?   West coast voters and early Election Day returns

(Fuchs, 1966) ›  Turnout ›  Preferences

9

Poll effects

Negative effects of poll reports?   West coast voters and early Election Day returns

(Fuchs, 1966) ›  Turnout ›  Preferences

  Voter decision-making (Bartels, 1988) ›  Candidate viability

10

Poll effects Negative effects of poll reports?   West coast voters and early Election Day returns

(Fuchs, 1966) ›  Turnout ›  Preferences

  Voter decision-making (Bartels, 1988) ›  Candidate viability

  Voter information (Iyengar et al, 2004) ›  The substantive information cost of the horse race

11

Other findings Polls may shift:   Attitude strength and candidate preference (Ceci &

Kain, 1982)

12

Other findings Polls may shift:   Attitude strength and candidate preference (Ceci &

Kain, 1982)   Candidate preference but not vote intention

(Ansolabehere & Iyengar, 1994)

13

Other findings Polls may shift:   Attitude strength and candidate preference (Ceci &

Kain, 1982)   Candidate preference but not vote intention

(Ansolabehere & Iyengar, 1994)   Assessment of candidate traits (Hardy & Jamieson,

2004)

14

Other findings Polls may shift:   Attitude strength and candidate preference (Ceci &

Kain, 1982)   Candidate preference but not vote intention

(Ansolabehere & Iyengar, 1994)   Assessment of candidate traits (Hardy & Jamieson,

2004)   Vote choice (Marsh, 1985)

15

A silver lining?

  Bandwagon and underdog effects are generally very small (Hardmeier, 2008)

16

A silver lining?

  Bandwagon and underdog effects are generally very small (Hardmeier, 2008)

  *but even small treatment effects can have real-world implications (Rosenthal & Rubin, 1982) ›  Elections are often won or lost by very small margins

17

Polls in the 2012 Election   900% increase in trial-heat polls from 1984-2000

(Traugott, 2005)

18

Polls in the 2012 Election   900% increase in trial-heat polls from 1984-2000

(Traugott, 2005)   18,150 polls aggregated by Huffington Post leading up

to Election Day 2012 ›  “New polls every day”

19

Current Research STUDY DESIGN

20

Current Research STUDY DESIGN   1500 participants on Amazon Mechanical Turk

›  Participants were paid $1.00   Pre/post-election surveys   Three email communications between Oct 22-Nov 5   Nine-digit ZIP

›  Local poll results   Human subjects/IRB approval was received

21

Current Research EXPERIMENTAL CONDITIONS – RANDOMLY ASSIGNED 1.  Obama leading 2.  Romney leading 3.  Obama and Romney tied 4.  No poll information   Consistent within participants across 3 emails   Leading margins were 73%, 75%, and 77%

22

Current Research – Experimental stimuli SUBJECT LINE: {ROMNEY}{LEADING BY}{73%} IN LATEST LOCAL POLL MESSAGE BODY: ACCORDING TO A RECENTLY RELEASED POLL OF LOCAL VOTERS, IN {STATE DISTRICT}, {ROMNEY IS}{LEADING BY} {73%}. THOSE VOTERS WHO REPORTED BEING UNDECIDED WERE ASKED TO REPORT WHO THEY WERE MOSTLY LIKELY SUPPORT IN THE ELECTION. THERE HAS BEEN NO CHANGE IN THE POLLING NUMBERS IN THE PAST WEEK WHEN ACCOUNTING FOR THE POLL’S MARGIN OF ERROR. THE TELEPHONE POLL OF {527} LOCAL RESIDENTS WAS CONDUCTED BY A NONPARTISAN POLLING ORGANIZATION AND PAID FOR BY A GROUP OF LOCAL TELEVISION AND PRINT NEWS MEDIA OUTLETS

23

24

25

26

Interest in Politics

(Intercept) 3.338*** (0.119)

Obama lead 0.231 (0.165)

Romney lead 0.328 (0.171)

Obama/Romney tie 0.345* (0.173)

Republican 0.639*** (0.158)

Democrat 0.750*** (0.206)

Obama lead x Republican -0.235 (0.216)

Romney lead x Republican -0.541* (0.219)

Obama/Romney tie x Republican -0.575* (0.224)

Obama lead x Democrat -0.496 (0.289)

Romney Lead x Democrat -0.742** (0.282)

Obama/Romney tie x Democrat -0.658* (0.287)

Log-Likelihood -1166.133 AIC 2358.267 N = 839 Nagelkerk R-squared 0.055

27

Interest in Politics

Fairness of the Election

(Intercept) 3.338*** (0.119)

3.706*** (0.116)

Obama lead 0.231 (0.165)

0.072 (0.162)

Romney lead 0.328 (0.171)

0.072 (0.168)

Obama/Romney tie 0.345* (0.173)

0.244 (0.170)

Republican 0.639*** (0.158)

-0.559** (0.201)

Democrat 0.750*** (0.206)

0.946*** (0.154)

Obama lead x Republican -0.235 (0.216)

-0.160 (0.284)

Romney lead x Republican -0.541* (0.219)

0.293 (0.277)

Obama/Romney tie x Republican -0.575* (0.224)

-0.191 (0.281)

Obama lead x Democrat -0.496 (0.289)

-0.153 (0.211)

Romney Lead x Democrat -0.742** (0.282)

-0.144 (0.215)

Obama/Romney tie x Democrat -0.658* (0.287)

-0.206 (0.219)

Log-Likelihood -1166.133 -1149.436 AIC 2358.267 2324.872 N = 839 839 Nagelkerk R-squared 0.055 0.375

28

Limitations   Self-reports of turnout

›  Should really validate   Not a representative sample

›  Could use more Republicans (151/839) ›  Sample is young, educated, etc

29

Conclusion POLLS SEEM TO INFLUENCE TURNOUT THE EFFECTS AREN’T PERFECTLY UNIFORM POLLS ALSO SEEM TO INFLUENCE OTHER POLITICALLY RELEVANT VARIABLES

30

Thank you!

vannette@stanford.edu www.davidvannette.com

31

Appendix ODDS

(INTERCEPT) OBAMA UP 2.626906 1.328669

ROMNEY UP TIE

2.439689 1.970804

DEM REP 2.814671 1.910819

top related