Does Twitter-based campaigning win votes? New evidence published
Welcome to the latest in my occasional series highlighting interesting findings from academic research. Today, “Does Campaigning on Social Media Make a Difference? Evidence from candidate use of Twitter during the 2015 and 2017 UK Elections” by Jonathan Bright, Scott A Hale, Bharath Ganesh, Andrew Bulovsky, Helen Margetts and Phil Howard.
Their conclusion is a version of “yes, it does”:
We find support for the idea that Twitter based campaigning is associated with voting outcomes, a finding which is robust to a variety of different model specifications and a strong empirical test using a two wave panel design.
How big is the impact of Twitter which they find? Small but consistently positive across different ways of trying to measure it:
Our most conservative model suggests that tweets would need to be increased by 175% to generate a 1% increase in vote share.
If you are tempted to say something about correlation and causation at this point, hold on. Not surprisingly for such a strong cast of researchers, they’ve heard of this too and address the point head-on in their paper. Their statistical approaches are designed to separate out spurious from causal correlation.
However:
We address only one social media platform in this paper (Twitter). We do not know the extent to which the use of Twitter correlates with use of other types of social media (such as Facebook and Snapchat), hence we are unable to say to what extent it is Twitter itself which makes the difference, as compared to other platforms.
A crucial further caveat is that this study – in line with most studies trying to understand the impact of particular tactics or effort – tries to adjust for variations in campaign spend by using the locally declared expenditure. That is, it tries to separate out the impact of using Twitter from the impact of spending more money on a campaign, an important separation as it may simply be that campaigns which spend more money also do more on Twitter, but it is the money and not Twitter which produces votes.
Yet, the long and short local campaign limits are now only a small part of overall actual campaign expenditure given the massive scope to target quite legally national campaign expenditure in specific constituencies. My own estimates for the 2017 election, for example, based on sources from across parties is that for every £1 that got declared locally, around £8 were spent via the national limits. As a result of such problems, a few political scientists have stopped using local campaign expenditure totals to try to act as a proxy for overall campaign levels. A few, but far from all.
Here is the research paper “Does Campaigning on Social Media Make a Difference? Evidence from candidate use of Twitter during the 2015 and 2017 UK Elections” in full:
If you are a Liberal Democrat wanting to make the most of online campaigning, see my round-up of tools and services. You can read the other posts in the Evidence-based campaigning: what the academic research says series here.
UPDATE: Further evidence, this time from Greece, here.
I guess there could be a philosophical objection to any social media platform that facilitates micro targeting of voters and that objection has to do with the submerging or disguising of the issues or the debate. Arguably in a democratic election, or referendum, all discussion ought to be, at least potentially, public. All the demos should be able to discuss all the issues, and private communication which could include illegal or covert blandishments or bribes or the disclose of confidential or personal information should be avoided.
This goes double, or treble when one confronts a situation such as we have with Cambridge Analytica, admittedly not twitter, whereby dispositions and personality traits which give insights into potential voting preferences or exploitable personal “weaknesses” could be used, probably at the margins, to swing a close run result like the EU referendum.
What about a candidate talking to a voter on their doorstep Mike? That’s more private than, say, a candidate using a public profile on Twitter.
Yes, but you don’t reach many people in one-on-one conversation (despite the fantasies of enthusiastic candidates). Whereas social media can be used to target specific messages at groups of voters whilst excluding others. A risky approach, unless done through a deniable proxy, but probably worth the risk. I think that’s the sort of thing Mike is talking about, rather than generic announcements on a public Twitter profile.
Sorry if I wasn’t clear Malcolm: my point was that one to one conversations on the doorstep are just as private and able to be varied from person to person as targeted digital activities. You can target who you talk to on the doorstep and deliberate avoid certain people too. You don’t have to be online to do such selections amongst voters.
So we should be careful not to over-react to thinking that the newer of the two of those is somehow dreadful when the former, often seen as type of campaigning to be praised, shares many features of the latter. The ability to do each at scale is, of course, an important difference that does matter.
From the abstract I read:
“We find support for the idea that Twitter based campaigning is associated with voting outcomes…”
“associated with” – that sounds like a correlation. Dr Pack – since when did causation result from correlation?
In other words I question whether your statement in reply to the question “Does Twitter-based campaigning win votes?” i.e. “a version of “yes, it does”” is really justified.
Um, the research paper goes into a lot of statistical detail about confounding variables and the like to separate out coincidental correlation from causal correlation. So I’m not quite sure sure why you pose the question about correlation and causation given the paper addresses this issue head on and gives a detailed case about why there is more than coincidence going on?