Belfast Telegraph online Border poll - useful to gauge interest, but it's no forecast
You will have seen the online poll run by the Belfast Telegraph about demand for a border poll, and how people would perhaps vote if such a poll took place.
This produced a large response but as detailed here, they can't really be taken as representative in terms of what perhaps the result would be if a full actual Northern Ireland border referendum took place.
To be fair to the Belfast Telegraph it never claimed the poll to be representative, or taken as a prediction.
The poll is currently running at 75% Yes to a border poll, and 70% Yes to a United Ireland, and I don't need to tell you that common sense, and history, tells us that this is obviously not representative of Northern Ireland.
Polling companies like ours are 'easy' with these 'self-selecting' polls as they're known - i.e. just allowing anyone who wants to, to take part, as most people see them for what they are i.e. a bit of fun.
They can, however, give a very rough approximation as to current general feeling - in this case, it shows the EU referendum result has increased interest in the border poll debate, as a response of 50,000 is pretty high.
As opposed to these 'self-selecting' polls, representative polls like LucidTalk's (and others) select a representative sample of Northern Ireland opinion, i.e. basically we choose and target who we poll.
The current Belfast Telegraph poll is the other way around i.e. they invite the public (i.e. any of the Northern Ireland public) to take part.
This is where the very large error comes in, an error so large that you can't even calculate it, as you don't know who took part in the poll - in terms of age-group, gender, area, socio-economic group, and most importantly community background etc.
For example in terms of the last LucidTalk pre-EU referendum poll we received over 1,500 responses from our already Northern Ireland representative Opinion Panel, and then honed this down to a 'very' representative 1,075 Northern Ireland sample, totally matching the Northern Ireland demographic measurements as detailed.
In terms of our selected sample polls - when we know our e.g. 1,075 sample is representative of NI, we expand out from that (extrapolate as the mathematics gurus call it) to end up with a representative view of NI opinion.
Because we expand from a relatively small sample there is a calculable mathematical error of +/-3% either way, and this is published in all our reports.
However, it's the accuracy, representativeness, and quality of the initial sample, which counts - i.e. our 1,075 sample. This approach allowed us to predict the NI EU Referendum result to within 1% a week before the actual referendum (June 23).
In terms of the current Belfast Telegraph self-selecting online poll these tend to attract 'politically active-interested' higher earners, younger age-groups, urban dwellers, etc... - i.e. you end up with an unrepresentative sample of opinion, and this is regardless of the numbers taking part (although for 'marketing' purposes 50,000 does sound impressive).
LucidTalk (& other polling companies) select our samples, and then work hard to get them to be representative of Northern Ireland - e.g. in terms of our online Opinion Panel we make sure that there is a representative number of lower income earners, a representative number from say loyalist and republican working class areas, a representative sample of over 65s, and a representative sample of people from the west of Northern Ireland taking part etc.. etc..
Basically we match - in terms of our 1,075 sample - the Northern Ireland demographics as detailed above.
In terms of these latter demographics I would doubt that few from these groups are taking part in the current Belfast Telegraph online poll.
However sizable numbers from all these groups took part in the EU referendum, and would undoubtedly take part in a border poll referendum. Hence you can see where the large error comes in with these self-selecting polls?
In a sense the Belfast Telegraph 'border poll' complements the more accurate research carried out by the polling companies.
The Bel Tel poll did show high interest in the issue, and that probably the Brexit decision has perhaps caused some former pro NI in the UK (as long as it's in the EU) to perhaps think about a United Ireland that's in the EU (if the UK are leaving the EU).
However unless we have accurate polling we can't estimate the numbers related to this 'change to support a United Ireland based on Brexit' - it may be a few hundred, a few thousand, or tens of thousands.
With our next Northern Ireland-wide poll coming up in September we'll hopefully be able to put a few estimates on these numbers.
Bill White is Managing Director of Belfast polling and market research company LucidTalk. You can follow LucidTalk on Twitter: @LucidTalk.
Belfast Telegraph Digital