Indyref – the changing nature of the opinion polls


Share Button

A word cloud created using the 1707 Act of Union.

That graphic above is a word cloud. The source document is the highly important 1707 Act of Union. It’s part of what led to the formation of the United Kingdom, as we know it today. Think of the things that would be different (or not even exist) without it.

On September 18th, there is going to be a vote which could end up being just as important as the source of what you see above. There will be a referendum in Scotland about independence. At present, they have a devolved Assembly with a certain set of powers. However, the Scottish National Party (leading the ‘Yes’ vote) want more. They want Scotland to be a completely separate entity (well apart from the currency, which they want to keep). They want to be a separate part of the European Union and not be restricted by old mathematics, such as the Barnett formula.

Given that this is such a hugely important vote, I think it’s worth looking at the current state of public opinion, as well as how it compares to historical figures. I will also explain some reasons for the increases and decreases.

I have an opinion on independence – just like many other people. However, I want this post to be as objective as possible. I haven’t taken part in any of the opinion polls that form part of what you see below either. It’s also worth mentioning that only British and EU citizens living in Scotland can vote anyway. I may have some Scottish elements to my family tree, but I am born, bred and live in England.

For this post, I have used data from this page, but also added more recent polls using articles found on the site.

Some polls also had a ‘Won’t vote’ option. However, I did not include that as not every poll has it. In addition, I have to point out that one or two of the polls seems to have figures that add up to more than 100%.

Headline figures

Data source – UK Polling Report. Created using Plotly.

You can find a larger version of this here and can also have a look at the full dataset.

One common way of looking at poll data is considering the ‘headline figures’. The polls and surveys that are commissioned have multiple questions and different wordings (depending on the organisation and related events). Whilst it’s always good to look at the details, the headline figures still give you a good idea about how people feel.

From this chart, you can see that there seems to be a gradual decline in the number of people who are unsure about which option to go for. The Yes vote seems to be gradually increasing, whilst ‘No’ has been going down (despite some peaks and troughs along the way). This could indicate a very close result on September 18th.

This recent article shows just how much the gap has narrowed recently.

The many different pollsters and those who commission them

Data source – UK Polling Report. Created using Plotly.

This chart uses the same dataset as the last one. You can find a larger (and interactive) version of this chart here.

All major opinion poll organisations are members of the British Polling Council and use a certain set of procedures. Whilst sample sizes don’t go into the millions, they are widely considered representative. However, the methodologies can still vary and the polls are done at different times. It’s worth taking a look at the differences of the headline figures for each of the pollsters.

What you see above is a box plot. In this case, I have taken all the ‘No’ figures and grouped them by pollster. You can see that most of them are bunched together in groups, despite the peaks and troughs in the earlier chronological chart. However, there are some ‘outlying’ figures – those out of the bunches. It’s most noticeable with TNS-BMRB and YouGov.

With the former, the highest figures were in the first three of their polls in the UK Polling Report list – the most recent being in August 2013. Since then, most of their ‘No’ percentages have been in the low 40’s, which reflects increasing support for independence – especially as the ‘Don’t Know’ figure is roughly the same. From the 28/02/13 poll to the one done on 07/08/14, the ‘No’ percentage has gone down by 7% and ‘Yes’ has only gone down by 1%.

As for the latter, the ‘Yes’ vote has gone up dramatically (from 28% to 47%), whilst ‘No’ has decreased by 10% (to 45%). YouGov are regular pollsters, so their results are better able to change based on current events. The ‘Don’t Know’ percentage gets lower and lower.

There seems to be more organisations (that do regular polling) with lower ‘No’ percentages than those with high ones.

Averages

Data source – UK Polling Report. Created using Plotly.

The interactive version is here and the dataset is here.

As I have stated previously, there has been numerous polls conducted/commissioned. You don’t always see massive increases or decreases (e.g. above +/- 10%) from poll to poll, despite the differences in methodology and regularity. Sometimes, the results fall into what you call the ‘margin of error’. That means if (in this case) the percentages have a difference of 3% or less, you could argue that they are level.

Over time, the averages become worth looking at because they are based on a substantial amount of data.

As you can see from the chart, the No vote has a noticeable lead. However, in all three cases it’s not as big as the difference between ‘Yes’ and ‘Don’t Know’.

You would think that this means ‘No’ should be victorious. However, it does conflict with the recent polling data, that shows a narrower gap. The above doesn’t take into account changing publicity. The older polls also don’t factor in the recent television debates, which have had some impact on what people think.

Finally…

As I mentioned earlier, it could be a very close result. However, there will be many voters who have not taken part in these polls. Polls aren’t always completely accurate (remember ‘Cleggmania’?) In addition, it’s hard to accurately gauge the levels of apathy. However, I would think the turnout should be high as the outcome will have a dramatic effect on the outcome of more than one nation. It’s just a shame that not everyone in the United Kingdom has the option to participate in the vote.

When the results of the referendum are announced, I will provide an analysis as soon as possible. In addition, I will be producing graphics and other things (e.g. something about the anatomy of an opinion poll) leading up to the vote. Links will be posted on Twitter, my Facebook page and the Google+ community. I hope you find it all informative and useful. If you have any questions, I will attempt to answer them.

I am also interested in your opinions about the upcoming vote. Should it be yes or no? Can you give your reasons?

So, what do you think?

Share Button