Social influence & Political mobilization

Research Letter Daniyar Mukhanov & Sijo Emmanuel

Introduction

Can political behaviour spread through an online social network? 61 million Facebook users during 2010 US congressional elections

Effect of social transmission

Most methods are ineffective (1% to 10%)However, 0.01% makes difference

Main Results

- A randomized theory

- People older than 18

- On November 2, 2010. US congressional selection

Three groups of users

- Social message 60,055,176 (99%)

- Informational 611,096 (1%)

- No message 611,044 (1%)

Message

Cannot compare the treatment groups with the control groups

No option to click ‘I Voted’ button or polling place link

Social message vs Informational message

Users who received the social message were

2.08% more likely click on ‘I Voted’ button

0.26% more likely to click polling-place information link

Information seeking and political self-expression doesn’t guarantee that a particular user will actually vote

Experiment and direct effects

Experiment Results: Analysis of direct effects

Measured the effect of experimental treatment on validated voting Examination of public voting records

Users who received social message were 0.39% more likely to vote than user who did not receive any message

Similarly 0.39% difference in voting between those who received social and informational message

Seeing faces of friends significantly contributed to overall effect of the message on real-world voting

These results show that online political mobilization can have a direct effect on political self-expression.

Experiment Results: Analysis of direct effects

Sample had an average of 149 Facebook friends

Many of these relationships constitute ‘weak-ties’

Mobilisation can spread online more effectively through ‘strong ties’

Close friends stronger behavioural effect on each other than

acquaintances

Counted the number of interactions between each pair of friends

Categorized them by decile, ranking from lowest - highest %

interactions

According to a validation study, friends in the highest decile are most likely to be close friends in real life

Interaction and Close real-world relationship

Experiment Results: Analysis of indirect effects

Observe per-friend treatment increases as tie-strength increasesObserved treatment effects fall outside the null distribution for

expressed votes

Significantly different from chance outcomes As interaction increases, the observed per-friend effect on friends

treatment on a user's expressed voting also increases Horizontal grey bars represent null distribution derived from

simulations of identical networks Incidence and topology of the behaviour and treatment are the same Assignments of the treatment are randomly assigned

Expressed voting

Experiment Results: Analysis of indirect effects

Observe treatment effect is near zero for weak

ties Treatment effect spikes upwards and falls outside

the null distribution for the two top deciles Strong ties are crucial for the spread of real-world

voting behaviour Treatment effect for polling-place search

gradually increases Several of the effects falling outside the 95%

confidence interval of null distribution

Validated votingPolling-place search

Experiment Results: Analysis of indirect effects

Experiment ResultsThe sample had 60,491,898 (98%) users with at least 1 close friend

Average user having about 10 close friends

Results suggest users were about 0.011% more likely to engage in an act of political self-expression

By clicking ‘I Voted’ button than they would have been had their friend seen no message

Each close friend who received a social message was on average 0.099% more likely to express

voting Ordinary facebook friends may affect online expressive behaviour but do not affect private

or real-world political behaviours In contrast, close friends have influenced all three

In many cases it was not possible to change target’s behaviourUsers may have already voted by absentee ballot before election day They may have logged into Facebook too late to vote or influence other users voting behaviour

All effects measured are intent-to-treat effects rather than treatment-on-treated effects

Does close friends matter?

Social transmission

- Friends: 886 000 expressed votes

- Close friends: +559 000 expressed votes

- Close friends of close friends +1 000 000 expressed votes

Close friends

- 282 000 validated votes

- 74 000 polling place researches

Some important notes

US midterm elections turnout increases from 36.3% (2002) to 37.2% (2006) to 37.8% (2010)

Facebook social message increased turnout by 60 000 direct and 280 000 indirect votes

Results

- Online political mobilization works.

- Social mobilization is significantly effective than informational mobilization alone.

- Close friends have about four times more influence.

- Online social networks influence offline behaviors.

Conclusion?

Ethics