Tracking media sentiment about the UK general election

The UK has recently elected a new government to Westminster.

For the next few weeks we're using our media monitoring tools to track sentiment about the UK general election.

It was a torrid first week for the Conservatives: a damp campaign launch, disunity within the party and a set of polarising policy ideas that indicated a “protect the base first” campaign strategy.

Below is scatterplot of media sentiment across all the stories picked up about the Conservative Party by our media monitoring software. Stories are plotted chronologically from left to right.

The software picked up 937 stories that mentioned the Conservatives. Note the “bulge” in story bubbles from 22 May onwards when the general election was announced. Light blue shaded bubbles in the chart represent stories with a positive sentiment, grey bubbles are stories with negative sentiment.

The Conservative Party

The majority of stories written about the Conservatives in week 1 scored just higher than neutral (0.00) according to the algorithm. Overall Conservative Party news stories had an average polarity of +0.015 (red line) and a median polarity of +0.020.

A chart showing the sentiment of news stories that mentioned the Conservative Party

In the chart stories that are longer in length are depicted with a larger bubble, smaller bubbles are stories of shorter length. It’s interesting to note how shorter stories often “float” higher on the chart, correlating with positive sentiment scores. It’s possible this shows that shorter stories tend to be more positive in sentiment. By contrast many of the larger story bubbles sink around the neutral mark (and sometimes lower) perhaps suggesting that lengthier stories are more balanced or negative.

Story bubbles that appear to smudge together in a horizontal line represent repeat stories, such as syndicated news. They often share the same content and therefore the same sentiment score. They are worth including here because, while the content is the same, each story can be considered to have reached a different set of eyes. In campaigning, message discipline matters!

The Labour Party

Our media monitoring software picked up slightly more stories about the Labour Party than the Conservatives, 1,083 stories in total.

A chart showing the sentiment of news stories that mentioned the Labour Party

Again, note the “bulge” and “pulse” in sentiment scatter bubbles from 22 May. The bulge highlights a leap in the frequency and therefore media interest in stories about the Labour Party and the pulse shows something about the news-cycle, literally the news cycle’s heartbeat.

Labour recorded higher positive sentiment scores for a number of stories, scoring almost as high as +0.4.

Overall the majority of stories about the Labour Party were positive with median and mean sentiment scores inseparable at 0.0329 and 0.0327 respectively and significantly more positive than the sentiment attached to stories about the Conservative Party. Again shorter stories, (smaller bubbles) seemed to float higher / be more positive.

However, what’s most notable is how the sentiment around Labour stories tightened and clustered toward neutrality and higher as the week went on. At the start of the period (19-22 May) the stories were scattered across the +/- sentiment polarity with little by way of a pattern. However, after 22 May the picture was transformed, with a tightened concentration of stories between 0.0 (neutral) and +0.2. There was a trend away from negative sentiment stories towards the end of the period.

My impression is that Labour Party had a better opening week than the Conservatives.

Rishi Sunak Vs Keir Starmer

So how do the two party leaders fair in the sentiment stakes after a week of campaigning? Below is a salience scatterplot of stories that mentioned Keir Starmer, the Labour leader. A monumental 2,965 stories were detected about him by my software in the first week of campaigning. His overall polarity score outperformed both his own party and the Conservatives with a mean polarity of 0.050 (rounded up) and his median polarity score at 0.046 (rounded up).

A chart showing the sentiment of news stories that mentioned the Labour Party leader, Keir Starmer.

Again, it’s interesting to note the “bulge” and “pulse” in the plot and a firming upwards in terms of the overall scatter of sentiment scores. It will be interesting to see how this changes in the coming weeks of the campaign.What about the Prime Minister? The consensus among media commentators was that Rishi Sunak’s election launch was a farce. It seemed like a bad week for Sunak. What did the sentiment numbers say?

Well first off I’ll note that the software detected a great deal more coverage of Sunak than Starmer. There were 5,172 stories recorded that mentioned the Prime Minister in the first week of campaigning. But not all news is good news, it seems.

Although Sunak’s name dominated in terms of coverage, his sentiment polarity scores of +0.030 (mean polarity) and 0.029 (median) are poor by comparison to Starmer. Still better than the Conservatives (+0.015 / +0.020) but worse than the Labour Party (0.0329/0.0327).

That said, the scatterplot really starts to tighten up towards neutral for Sunak after 26 May, towards the end of week 1. It will be interesting to see if this trend continues for the Prime Minister after a very shaky start.

What about policy?

It’s early days and at the time this blog was written the campaign was still policy light.

One policy that had been announced was the Conservative’s idea of compulsory national service for 18-year-olds. Around 792 stories were detected about this by the software, with a glut stories with a fairly positive sentiment published before midday on the 26 May, but noticeably starting to peter out by the 28 May.

Overall the policy had a mean positive sentiment score of +0.070 and a median score of +0.058, in media terms, stories about national service had a more positive sentiment than coverage of Sunak, Starmer or their respective parties.

In conclusion

It seems the sentiment scoring was broadly in line with the narrative I’ve been hearing: that it was a bad week for Sunak. But it was an even worse week for the Conservative Party according to this analysis. That said, the policy of national service was popular by comparison.

I’ll be continuing this media sentiment tracking throughout the election period. Tune in next week when there will hopefully be more in terms of policy to track.

In the meantime, feel free to get in touch about this or any other aspect of the media monitoring tools!