Sunday, 6 November 2016

My transcript of interesting interview on BBC Radio 5 Live re social media analytics predicting Trump win

Here's my transcript of a 4 November BBC Radio 5 interview with Jean Pierre Kloppers, CEO of BrandsEye about their analysis of Twitter responses to the US election. His company apparently correctly predicted the Brexit vote and are now saying that Trump will win based on similar trends. It’s an interesting alternative perspective that moves beyond traditional polling. 

However, the jury is out on whether social media analytics tells us much of any importance within the context of an election.

I've highlighted the best bits below

——

My transcript from http://www.bbc.co.uk/programmes/b080x5sv 3h15m

Jean Pierre Kloppers, CEO of BrandsEye: “We’ve seen in the last week a remarkable shift in online sentiment towards Trump in all of the battle ground states. Two days ago, Hillary was still ahead in New Hampshire. Yesterday we saw even in New Hampshire on social media pushing past the 50% mark people talking positively and advocating Donald Trump.”

Interviewer: So just explain to us how these social media polls work?

JPK: OK so it’s not a poll per se. We look at all conversation from the US on social media. And the challenge with social media conversation as you’re well aware, is twofold. One is you get a view normally from the people you’re already connected with…you get the social media echo chamber effect and it’s hard to see through that. The challenges in seeing through it is accurately determining sentiment in social media is a nightmare, because people speak so sarcastically, with local nuance and use vernacular - it’s hard to understand that. That’s the challenge of it. 

So what we do is we take a representative sample of all of that conversation and can look state by state and week by week to accurately determine what are people actually saying. So if there are 60,000 people in Florida talking about the candidates, how do they feel about these candidates. And so it becomes, inadvertently, a poll when you look at it from that perspective. But it’s an unsolicited poll - people just sharing their own opinions of their own volition. So you get something that the polls often miss which is the energy and the volume and emotion that comes with the sentiment being expressed on social media.

And what happened last week with the whole [Clinton] email saga is it gave a lot of people licence to get back onto social media to support Trump. And we hadn’t seen that in weeks before that. Especially post the 2005 audio that was shared where he talked about touching women inappropriately. I think a lot of Trump supporters post-that were a bit ‘ok we can’t post our opinions’. And certainly on social media this last week, that has changed. 

In states like Pennsylvania over 90% of people speaking from Pennsylvania are supporting Trump on social media. 

Interviewer: So your poll shows that Trump’s ahead. You’ve been correct before on something else that was pretty big?!..

JPK: It was pretty big, you could say that. Yeah, we saw the same trend in Brexit the week before the referendum. What happened there was we didn’t know what to do with this data, because we saw. three days before. 58% of people from the UK on Twitter promoting the Leave camp. And we thought this was bizarre because it was not what all the other polls were saying. It was not what the media was saying. It seemed like the Remain camp had it in the bag. But that’s not how people in the UK, certainly in the outlying regions were feeling. And if that’s anything to go by this time, we’re seeing the same trend, just far more exaggerated in the US. 

Interviewer: So you were the only polling company to predict Brexit and now you’re saying that Trump’s ahead?

JPK: Yeah correct. And...social media, it’s not a poll, because you can’t have 90% of Pennsylvania voting Trump, you know that’s never going to happen. What we have seen is that it gives an indication of which way the surprise is going to go. And I think what we’re seeing in the US is, you know, the Nate Silvers (of http://fivethirtyeight.com) of the world are putting Hillary’s chances at 65% to 70% of winning the election. And what that can do is cause people to not come out and vote - certainly on the Democrat side. On the Republican side I think that what it’s doing is getting the people who wouldn’t ordinarily have voted, social media is now giving those people the sniff that ‘hey, maybe we can win if we get out and vote’. I think it’s certainly mobilising people who wouldn’t have entered the conversation before to get out there and both get involved in the conversation online and I also think it’s going to translate into more people than we expected getting out to vote for Trump. 

And the big question is, in that silent majority of people who aren’t speaking on social media, are they just going to stay home, or will the help of Obama, Bernie Sanders, the other kind of big names on the Democrat side who are out there campaigning for Hillary, will they be able to move those people to get out on Tuesday to go and vote? 

No comments:

Post a Comment

Note: only a member of this blog may post a comment.