Scottish Opinion/MoS Scottish Westminster voting intention

Ok, so I'm a bit late with this post, but never mind. Most people with any interest will already have seen results of the MoS poll, but here they are again. There doesn't seem to be any online version of the poll or the newspaper article, but the results are posted on the SNP's site among others. The poll samples the opinions of 650 people between the 15th and 20th May when the expenses scandal was getting into full swing. There is a great lack of detail in the results and the sample size is smaller than most full polls, but these results should be more representative than the sub-samples I've reported on below. The results are as follows:

Tory.....11%
Labour...27%
LibDems..11%
SNP......43%
Others....8%

So firstly, a massive boost for the SNP with an unprecedented lead in Westminster voting intention. Secondly, the big two London parties are down heavily on what we've seen from other polls. As well as the SNP, the 'others' appear to be the main beneficiary. The Lib Dems continue their poor polling of recent times. If the results of this poll was replicated in the election, the SNP would go well beyond the tipping point that would see them win most existing Labour seats.

So to put this in the context of previous polls the graph below shows the results of all the large sample Scotland-only polls carried out this year. The two YouGov polls in January and March are remarkable in their similarity. However in, by April, YouGov are showing the SNP making gains on Labour. The most recent two polls are TNS/System 3 and this one from Scottish Opinion, so they are not necessarily directly comparable, but the trend is still interesting. The TNS/System 3 poll shows broadly the same results as the YouGov poll from the same period. The latest poll is one of three things - it is either an outlier, captures the temporary disenchantment with Labour and the Tories or it indicates the beginning of a sea change in Scottish politics. It will take a few more polls before we can determine whether this change will be sustained.

Populus/ITN Westminster poll

Another poll on Westminster voting intentions. This time it's from Populus for ITN. The main story being reported is the majority in favour of an immediate UK election and a reduction in likelihood to vote. UK politics and the UK government are clearly in a state at the moment, but it still doesn't seem to be having that much impact on voting intention. This poll shows a the big three UK parties at pretty much the same level we've saw before the expenses scandal, so maybe turnout will be the biggest loser and in a few cases this will allow the 'smaller' parties in. The situation is obviously a bit different in Scotland with the SNP now staking a claim to have usurped Labour as the biggest party.

So what does this poll have to say about the position in Scotland? Well, it shows the SNP well ahead on 40% with Labour slipping back to 27%. Populus use much smaller sample sizes to cover Scotland than YouGov so we expect to see much more variability (this poll questioned just 88 Scots). That said, if we look back at the sub-samples from all the Populus polls since the turn of the year, while Labour were well ahead in the January poll, since then, the SNP are showing a consistent lead. The Tories also seem clear in third place with the Lib Dems struggling around the 10% level.



So make of this what you will, but the lead for the SNP over Labour does appear to be sustained. It also backs up what we're seeing in other polls. The combined sampes for these polls show the following:

Tories.....15%
Labour.....32%
Lib Dems...13%
SNP........37%
Others.....3%

YouGov/DT poll

So today saw the Daily Telegraph publish the results of a YouGov UK poll of Westminster voting intention.

Headline figures from a sample of 241:

Tories....20
Labour....28
Lib Dems..16
SNP.......30
Others.....6

YouGov tend to have relatively large sample sizes, but the Scottish sample is still low numbers and won't be weighted to the Scottish population. So is there any value at all in looking at these figures? Well, to begin with, we can look at how these figures compare with previous samples. The graph below shows the last Scottish sub-samples from the last seven YouGov polls.



So you can see the variability that appears in these small samples. This does seem to show the SNP moving ahead of Labour.

These samples can also be combined to produce a larger sample size. Albeit this will retain the poor weighting. Anyway, the most recent seven polls combine to give a sample of 1408 over the period of 13 March 09 to 16 May 09. This gives the following results:

Tories....21
Labour....33
Lib Dems..13
SNP.......28
Others.....5

So Labour still ahead here - probably as expected. Taking the combination of these sub samples means we're looking at a much longer period, so more recent factors (expenses) will be less visible. In future I'll also show a rolling average keeping the sample size at >1000. This should show results with a respectable sample size, but show a trend over time, evening out the sharp changes of the individual sub-samples. It will lag behind the real situation though.

Anyway, this post is really about getting the blog up and running and it should give an indication of the type of data I'll be trying to post and hopefully it will be of interest to some. Obviously there are huge issues involved with sample sizes and methodology and I, or someone else, will try to tackle them here as well. I'll also try a bit more of the interpretation of the results as well.

Introduction

There are not a lot of large sample political polling done in Scotland but there are regular UK-wide polls that allow a sub-sample of the Scottish population to be viewed. These data are reported widely around the blogosphere, so I thought it would be useful to archive the results for comparison and also to look at how these sub-samples can be combined to give a more statistically significant result. Not being an expert on political polling methodology, any constructive feedback on how to best use this data would be very welcome.

I know some will say that it is a waste of time to even bother looking at these polls, but people do, so we may as well try and make best use of them. If I keep this running up to the election, we should be able to see whether the polls proved to be any help in predicting the result.