UK General Election Day 17 Twitter Analysis

We’re going to mix things up a little starting from today. I’m going to drop the word chart, as I don’t think it really gives us much information, and I’m going to add two new charts, one Posts Per Hour by Party, and the other a map of the UK marked with geotagged tweets and the party that was the subject of the tweet.

But we’ll kick off with the hashtag chart:

image

As you can see, chatter about the BBC Debate dominated Twitter on Thursday April 16. Followed closely by the #SKYNEWSPOLL which showed that Nicola was perceived to have “won the debate”; although what it means to have “won the debate” is very subjective, so the poll doesn’t really mean anything; other than it is safe to say that, without exploring the policies in depth, most people agreed with what Nicola was saying.

image

Nicola Sturgeon, the SNP and SkyNews are all riding high in the chart, as the “CyberNats” enjoy their leader’s success, this popularity rubs off on the news agency who declares her the winner. For comparison, a polling agency “Survation” who didn’t think Nicola did quite as well, is down at number 473.

Natalie Bennett and Leanne Wood also make the top 10, marking a hat trick for the female party leaders.

Now, let’s look at the first of our new charts, and that’s a map of all the geotagged tweets and the subject of the tweet:

Thu16

The pins mark the location of the tweet and the colour of the pin signifies which party was the topic of the tweet.

A couple of quick caveats before we comment on this chart. Firstly, those people who geotag their tweets are a “self selecting sample”, meaning they decide to tag or not to tag, and so they will not be statistically significant, so things we say about this sample, only refer to this sample, and can’t be projected onto the population as a whole.

The second thing is that although this chart shows the subject party of the tweet, it makes no comment on whether or not the tweet is positive in negative in nature.

That said, it is clear from this sample, that Labour dominate the chart (probably due to the fact that the clashes between Sturgeon and Miliband were judged to be the most significant of the debate).

Now it’s a little tricky to analyse what that means. On the one hand, Labour are the dominant subject, but we can tell from the hashtag frequency that pro SNP tags dominate the chart, so it could be said that pro SNP tags and Labour as a dominant subject is bad for Labour as the balance of probability is that they are negative comments about the party.

On the other hand, we know from previous posts that the SNP have very good tag discipline around a few key hashtags, leading to their dominance of the chart; it is possible that, at least some of these tweets are pro Labour, but are either not tagged at all, or are tagged with multiple pro Labour tags.

What is clear is that a lot of people are talking about Labour, and even if the majority of the tweets are negative, if Labour can engage with these people and change their minds, then they have a good chance of winning this election.

Now onto the next of our new charts, the Posts Per Hour by Party Chart which again shows the subject party of the tweet, but this time codes it against the hour of the day that it was posted. Here we can see from mid afternoon, through the evening, to the debate, and on through the night as analysis, punditry and supporters waded in:

image

A couple of things to note with this chart too. Firstly, as in the last chart, this one shows the subject party of the tweet but is silent on the nature of the sentiment. The second thing to note is that I made the Y axis a logarithmic scale, so the data was more clearly displayed, but this means that for every vertical gradient the values go up by an order of magnitude.

As you can see from this chart, Labour are the dominant subject of tweets (winning in every hour), all through the afternoon and evening, sometimes beating the SNP by an order of magnitude.

Finally, we’ll look at the trigrams chart:

image

No prizes for guessing that the SNP dominate the chatter with the news that their leader did well in the debate. The second most talked about story was the claim that the Sun newspaper faked a story to attack Ed. Miliband. This is the claim that the Sun faxed over their front page, claiming Miliband lost the debate to Cameron, before the debate had even started.

Let’s take a look at what else was in the news and how the Twitteratti reacted to it:

During the debate UKIP’s Nigel Farage accused the audience – which was independently selected and broadly representative – of being “left-wing”. This happened after he was booed by the audience, “BOOED BBCDEBATE AUDIENCE” makes number 89 in the chart.

Leanne Wood, of Plaid Cymru, and the Green Party’s Natalie Bennett took on Mr Farage over his attitude to migrants. Migrants were mentioned 6 times, with this: “LEGITIMACY MIGRANTS SUPPORTING” at number 942.

Richard Desmond, whose publishing company owns the Daily and Sunday Express, gave £1m to UKIP, prompting “DONATES 1M UKIP” at 532.

Nick Clegg said a vote for the Lib Dems could prevent a right-wing coalition of the Conservatives, UKIP and the DUP – or “Blukip”, putting “KEEP BLUKIP OUT” in at 1,207.

The Scottish Tories launched their manifesto. This did not register in the top 2,000 trigrams.

Well that about wraps it up for this post, until next time, keep crunching those numbers! Smile

Advertisements
This entry was posted in Data Science, Social Media, Statistics, UKGeneralElection2015 and tagged , , , . Bookmark the permalink.

One Response to UK General Election Day 17 Twitter Analysis

  1. Pingback: UK General Election 2015 Day by Day Analysis | Gary Short

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s