There has been some ruminating about the inaccuracy of the national pre-election polls with the internet pollsters coming off worst. However what about locally?
Polling in Northern Ireland has a somewhat dodgy reputation and when the Belfast Telegraph produced its pre-election Inform Communications poll there was much howling from UCUNF. This was unsurprising as it correctly predicted that they would flop. David Cameron even criticised it during his flying visit:
"No, not at all because the poll I saw in the Northern Irish paper this morning, this is a poll not carried out by a proper polling organisation and indeed it is by a lobbying company."
TV outlets declined to cover the poll offering the excuse that the sample was too low. However, while that may have been arguably true of individual constituencies it was not true of Northern Ireland as a whole. The total Northern Ireland sample of 3200 was substantial and larger than the sample for many national polls. Its results did closely reflect the actual result with most well within the standard margin of error:
DUP 26% (P) 25% (A) -1%
SF 25% (P) 25.5 (A) +0.5%
SDLP 17% (P) 16.5% - 0.5%
UCUNF 13% (P) 15.2% +2.2%
Others 19% (P) 17.8% -1.2%
Key (P) Poll (A) Actual
As regards the individual constituencies its predictions were good. There was greater variation between the predicting in vote % and result, however, it called the winner in 17 of the 18 constituencies correctly (under-reported Alliance in East Belfast). It correctly predicted the top two in 15 of the 18 constiuencies and the top four in 12 of the 18 constituencies.
PS spotted Mark McGregor had done something similar on slugger while I was writing this but decided to post it anyway.