Archive for February 2016
Seats not finalised, but maybe its time to look at how the different companies did in their final polls of the campaign.
As I pointed out in 2011, MRBI had been the closest in 2007- people still rubbished them, and they went on to be the most accurate in 2011 also. And yet, still the mockers.
This time, they introduced another adjustment which I was dubious of at first, and I didn’t project their first poll on the blog as a result. Subsequently I met their MD for a coffee and he very kindly explained the reasoning and impact of that adjustment (along with other aspects of their vision of polling), and I was impressed enough with his reasoning to go back to projecting for their polls.
Just as well! This year they were looking to be the most accurate in both the pre-election poll and Exit poll, and managed it again with aplomb.
Figures below on the last polls and the exit polls from RedC, B&A, MillwardBrown, MRBI, the DIT poll and the IPR projections, along with the 2 exit polls.
Of the pre-election polls, MRBI were the closest, being a cumulative 11.36% out over over the 9 party/Oth totals. My estimates were second at 13.68%, RedC 3rd with 14.6%, Millward Brown 4th with 17.36%, B&A 5th with 20.6%, and DIT last (just!) with 21.04%.
In the battle of the Exits (which had much smaller margins of error due to bigger and-presumably – more accurate samples of actual voters), MRBI were also closer, being an impressive 8.04% out, compared to B&A’s decent 10.38%.
Impossible to meaningfully compare these with the 2011 figures as the smaller parties weren’t measured in those.
Anyways, the figures are below. Read ’em and weep. Etc etc.
As readers will be aware, I’m a Labour voter (usually) and the Labour result is one that’s not filled me with hope. There will be much argued about where they went right and wrong, both in Govt and the campaign, but we’ll leave that for elsewhere.
As I write, Labour have 5 seats and are in the hunt (with varying degrees of hopefulness) for 4 further seats, and it appears (slightly) odds on they won’t make 7 in the new Dáil (unless they come to some arrangement with the SocDems or e.g. Zappone).
Worth noting that this day last week, when projecting their performance in the Sindo/Millward Brown (poll=6%, seats=5) I stated the following;
“The couple of points LP are lower compared to RedC sees them miss out on a rake of seats (indeed falling short of the 7 required to be a Dail party), suggesting there is a very big tipping point for them between 6-8%… if they are about 7% on the day (quite possible on current polling) I suspect they will be involved in a lot of dogfights. If they fall as low as this on polling day, this could be the last time the party contests an election (at least in its current form), as consideration of a re-alignment of the centre-left would surely arise if both they and the SocDems were below that threshold. But we’ll see.”
This is pretty much what happened, although whether you draw the same conclusions is up to you.
On the day, they have received 6.6%, and while they may yet reach 7 seats, the issue of re-alignment of the centre left will probably remain. Interestingly, this was below the 2 exit polls, which gave 7.1% and 7.8% – those margins may not seems much, but the seat projections for them were 7 and 11, so 5-9 (tending to 6) is around what you’d expect for 6.6%.
Looking at the projections I made over this series (since I returned in December) that were under 8%, the seats projections were as follows (remember interactions with other parties also influence);
So 5-7 seats at 6% (averaging 6.3), 9 seats at 7%, and 7-11 seats in the space between 7 and 8%. Quite where 6.6% is going to land is still uncertain but currently at 5-9 (looking like 6-7), it looks like it will end at the upper end of the 6% range.
That’s all for this post, back off to look at the Dublin Bay figures – I’m had Labour at about 35% chance (AOR) and 20% (KH) last night, but want to look at them again now I’m sober…
A lot of partial tallies out there this year, making it harder to project. By best clal at this stage seems to show SF doing slightly better than I would have taken from the exit polls. Based on what we’ve heard (and projected into the gaps) I’m expecting something similar to the following now;
How d’ya like them apples…?
B&A exit poll shows a slightly bigger drop for FG/LP and bigger vote for the harder left. Most noticeable is the tipping point for LP, which I’d noted at sims around 7%, comes into play again, the 7.1% here gives 7 seats, compared to 11 seats on 7.8% (although interplay of other parties also contributes to this). This suggests their reaching the ‘speaking rights’ threshold is far from certain. Anyway, projections…
These figures, no chance of FG/LP core of Govt. Numbers there for something witgh a FF/SF/LP/SD core, but I don’t see it. FG/FF in some shape or form again.
FWIW, Prof Mike marsh did an official projection for RTE and he has been similiar enough in the past, and is again on this occasion (difference in brackets);
FG 46 (-4), LAB 9 (+2), FF 37 (-1), SF 27 (same), SD 7 (+2), PBP 6 (same), GP 4 (+1), IA 4, R 3(+1), INDs 13, OTHERS 2.
The Irish Times/MRBI poll came out (almost) on time, and suggests the swing to FF recorded over the course of the campaign continued. FG are lower than in most polls, LP about average or slightly above, SF dipping in the direction the polls had indicated. Of the minor parties, the margin of error makes it less revealing, but on these figures a good poll for the Greens, and more at the disappointing end for AAA and the SocDems. Possibly.
OK, projection here, but (get one’s excuses in early), I have half a bottle of wine in me, and so if these are on the button I get *bonus* points.
NB, I am assuming a fair few non-transferable votes in the later stage of the counts, a few seats (notably LP) benefit from this, and if this is over-estimated LP and to some extent FG may suffer.
On these figures its hard to see past a minority FG govt with FF support (until they pull the plug), but only after weeks of talks. Even FG+LP+GP+SDs+Ren=72, and even that ragtag would need another 7 votes from the 19 in OTH (which seems impossible to me given the likely composition).
A FF-led Govt would be as implausible (if not, at the same time, impossible), if one ignores their ‘red-lines’. FF+SF+LP+GP+SD=81 but even if they agreed to talk to SF (who would have to agree to go in with them) FF would have to persuade LP (below 15 and licking their wounds) and Greens and SocDems to hitch their wagon to a caravan that could come unstuck very quickly. FF+SF+LP as a ‘core’ would come to 74, and one could at a stretch imagine 5+ Indos coming on board, but that, again would be a precarious arrangement, and not one many would relish explaining in November when it fell apart. Maybe.
So anyways, only a poll, but the Exit poll in 2011 was very close, and the final MRBI conducted a week earlier was slightly closer, so one imagines this is very close to being on the money. B&A will however have another Exit poll tomorrow on RTE at 7am, and may have polled later into the day, which might make it more reliable.
In the meantime, I’ll try to haul my ass up early tomorrow to run the sim on the B&A poll, and I’ll be while stuck at Dotski Jnr’s GAA match for some of the morning, hopefully the wonders of smartphone tech and my trusty election-day spreadsheet will allow me make decent extrapolations from the early tallies. In 2011, after 15 partial tallies, it predicted the final result with a total deviation of 2.6% across the 6 totals, compared to 4.5% across the Exit poll. Harder this time, given the new constituencies and parties – a disadvantage which won’t apply if we have a second election this year, but hopefully accurate enough for any candidates tailoring their speeches at the count centre.
Ok…last time out, predictions from most quarters (including the other blog whose method was annoying me…) were so far out it wasn’t too hard to out-predict them.
This time out I note in the last week or so projections from all sources, and predictions on Twitter are converging on what I would consider sensible (at least if you exlcude the usual headbangers). Presumably that means they have improved their methods, or I have disimproved mine.
My projections therefore look less radical, and more boring as a result.
But here they are;
FG 30% 55 seats
FF 20% 39
SF 15% 22
LP 7.5% 11
AAA 3.3% 4
SD 2.9% 5
GP 2.5% 1
REN 2.5% 2
OTH 16.3% 19
Rumours of an MRBI/ITimes exit poll tonight, and RTE will have a B&A exit poll tomorrow at 7. While MRBI have a better record, its worth noting that to get the poll out early (10:30 has been mentioned) they may have to have a very small sample of later voters which may skew the figures.
I’ll have some projection based on what comes out.
Also, as in 2011, I have a formula which I used to take the early tallies and project final result (in FPVs). Basically it takes my constituency projections, looks at the average deviation of most recent tally from that, and projects the same deviation to everywhere theres no tally.
Last time out I poated what it had after 15 partial tallies, & cumulatively it was 2.6% out for party totals, compare to 4.6% in the exit poll. This time new constituencies and parties may make that base less accurate, so it will eb intersting if it works as well this time.
Anyways, 10 o clock!
Well, that rumoured Millward Brown poll never surfaced, so was either a hoax or privately commissioned. So the last poll we have is one of very uncertain quality, being surveyed by DIT Market Research students (under what supervision I have no idea) for journal.ie. Method seems similar to the RedC phone approach.
But given I was geed up to do one last projection, I thought I’d run it anyway! 😉
Overall, the poll isn’t very far out from the professional companies, but appears to have higher rating for FG and lower for AAAPBP – other than that no great deviation. FG boost however small in vote terms, combined with a good LP showing (unlikely both would occur at the same time, given they chase many similar voters) would be enough however to put FG/LP within touching distance of Govt – indeed on these figures they could get a majority if the SocDems were interested. However, would a new party take a risk like that so soon into their existence, particularly for a Govt that could fall soon enough, given the size of its majority. LP would also have to be persuaded, being below 15, however, if it was almost that and they had an ally like the SocDems joining them, I suspect that could persuade them.
Anways, a school night so off to bed. Will post my projection before polls close tomorrow, although I notice online predictions on Twitter are broadly more sensible (and consistent) than they were in 2011 (or in certain site’s projections a few months ago) and so you are unlikely to find anything surprising in them.
Enjoy the rest of #GE16 ..