Read on Twitter
There are plenty of interesting results in today's @LucidTalk poll for the Sunday Times. However, given the importance of the subject, we really must take great care when drawing any inferences from survey data. (1/7)
The headlines are misleading. Readers are told that a majority in Northern Ireland support holding a referendum within the next five years. It could well be a majority. But it might also be a minority, since the estimate of 50.7% has a margin of error of +/- 2.3 points. (2/7)
We’re also told that “a majority” think that Northern Ireland would leave the UK within 10 years. But then we’re told that it’s 52% of respondents who say this, after removing the 8% saying they don’t know. It is misleading to characterise this as a “majority.” (3/7)
When it comes to voting intentions, we’re then told that “a significant majority” of Alliance, Green Party, People Before Profit voters would support a united Ireland in a referendum. *But*, in fact, it’s only 38% (quite clearly not >50%). (4/7)
We’re then told that “a slight majority” of voters aged 18-44 would vote for Irish unity. *But* it’s only 47% who say they would. Not only is this clearly not >50%, but it’s also statistically indistinguishable from the level of support for remaining in the UK (46%). (5/7)
As I said, there are very interesting, highly relevant results in this poll. It is fair to say that they give unionists plenty to worry about. But the strength of the conclusions in the reporting do not match the data. (6/7)
All of this might seem pedantic, but since polling data on this particular issue will understandably generate plenty of interest and attention, journalists, editors and commentators have a responsibility to report any results accurately. (7/7)
