October 2022 Projections Update

Seat Map

Four polls in October, from Ireland Thinks, IPSOS/MRBI, B&A and RedC. The first doesn’t have detailed breakdowns so is not considered in the regional calculations, but the latter three do – even if the Irish Times annoyingly dripfed the IPSOS results instead of just publishing them all together outright.

In fact that IPSOS poll in general extremely annoying – not the poll itself, but the furore it caused in the Irish Times, who spent all day doing victory laps about how Fine Gael are back, baby, and the budget is the best thing since sliced tenants. This was, of course, nonsense, but par for the course for a media environment that continues – although it absolutely knows better – to choose to behave as if it has zero political or statistical literacy when it comes to polls.

Indeed, Pat Leahy was too busy writing multiple columns about how the government was on the comeback trail that he missed some insights that are showing bad signs for Fianna Fáil in particular – something supported by other polling companies this month.

I’ll go more into the problems with media coverage of polls and Fianna Fáil’s problems below. I’ll also be talking about why PBP are gaining seats now, some notes on recent issue polling, and a note on comparative methodology which will hopefully illuminate a bit why different people have different projections from the same data. There’s a lot of writing below, so if you would prefer to skip straight to the seat changes, click here.

Misreading polls and the “budget bounce”

Where to even begin with this. October’s IPSOS poll was, on face of it, a good one for the government, with the parties up a combined 5 points on the July poll (FG up 3, FF and Greens each up 1). The poll’s commissioners, the Irish Times, lost the plot and published by my count six or seven articles about how this vindicated the government and the budget and was a sign of their fortunes reversing. It was first light on the fifth day, at dawn they looked to the east, and there, sitting astride a white horse, was the long awaited “budget bounce”.

Except this is complete nonsense – leaving aside the fact that taking a single poll and drawing sweeping conclusions is always inaccurate, the data doesn’t really support that when it’s dug into. If we compare the regional figures to the rolling polling average (RPA) from immediately before the poll:

What we see here is where the poll detected changes in support from the RPA taken immediately prior to this poll. Any poll in isolation will usually show some degree of variance from averages, so that’s not particularly interesting. But you can do a lot with this – for example, I could create a narrative here that the Green Party is going to be wiped out for sure because their support is down 66% in Dublin since the GE – whereas in reality an average of polls show us that they are only down by about half this much, and have a shot at keeping a number of seats.

Similarly, the apparent bump for FG and FF is mostly in one region each; and while FF polling in Connacht-Ulster is all over the place in general so we should give a pass on this, the FG number in Munster is a huge outlier. That doesn’t mean this isn’t a good poll for FG; but it does mean we should be cautious about over-emphasising a single data point, and definitely shouldn’t be drawing huge conclusions about what drove it.

And if we consider the IPSOS polls from April and July, we probably find an actual answer. We have to take national figures here, as we don’t have regional figures for July:

In general this is all fairly standard fluctuations, but one thing jumps out in July – a 4 point drop for FG and a 4 point rise for Independents & Others. No other polling companies detected this movement in July. This then reverses precisely in October. The most likely thing here seems that the July poll was a particular outlier, and the big positives for FG this month are a reversion of that.

Again, this isn’t bad news for FG, but it does not support the conclusions drawn in the media. Indeed, if we compare to RedC, who also showed a 3 point rise for FG (from September, not July, which is an important distinction), they had that support being offset by a drop in support for their two government partners. Indeed, contrary to the Irish Times, RedC led off their analysis with the budget having “limited impact“.

While RedC’s approach is more sensible (and more statistically rigorous) than the Irish Times’ wild interpretations, the whole narrative about good or bad news for the government and the impact of the budget ultimately hinges on single point swings for Fianna Fáil and the Greens. This is silly.

Oh and to finally show how polling on the budget doesn’t really support any narrative, we have some issue polling from IPSOS. 65% of voters think “government action” in the budget will help the cost of living. That looks like vindication of the budget, right?

Except the same polling showed that 59% of people think “government action” will make inflation worse, and 48% think the entire cost of living crisis is out of the government’s hands. You can’t make a coherent case out of these numbers unless you cherry pick one. People are simultaneously optimistic, pessimistic and neutral depending on what question is asked; this is frequently the case with issue polling, and selecting one answer to support whatever you are trying to conclude is dishonest.

In conclusion, what I find particularly frustrating about this is that there is absolutely no way Pat Leahy and the Irish Times don’t know this. They may be many things, but they aren’t stupid or innumerate, so in my book it’s hard to consider this as anything other than using whatever numbers are convenient to reach whatever conclusions they want. In an environment where both the media itself and polling are experiencing crises of trust, in Ireland and globally, this is a dangerous game to play.

Fianna Fáil slipping in Dublin

Last month I noted that Fianna Fáil didn’t seem affected by a property registration scandal, and while I don’t know what’s driving it, they certainly have suffered this month in Dublin. All three pollsters saw them drop compared to their prior poll – RedC by 3 points, B&A by 7, and IPSOS by 8 (albeit that we have to go back to April, as the IPSOS July poll did not provide breakdowns).

This isn’t manifesting as a disaster right now – the rolling average has them a shade below where they were in GE 2020, but it does wipe out all the seats that looked like likely pick-ups for them, as well as one incumbent. If this is not a blip (and while it’s unusual for a blip to map across all pollsters, it is not unprecedented), their other incumbents could start looking in trouble – though the only one that would be in line for immediate change is probably Dublin Bay North – they have more of a cushion in Fingal, West, South-West and Dún Laoghaire.

Still, it is notable – Fianna Fáil slowly chasing down Fine Gael in the polls, and even overtaking them in seat projection, has been the trend for a while. Whether this is the sign of a meaningful reversal in their fortunes or just a temporary thing remains to be seen, but is worth keeping an eye on.

Positives for PBP/Solidarity

I rarely write about PBP/Solidarity (henceforth “PBP” for convenience) on this site because their polling is rarely particularly interesting – it’s usually pretty steady and the number of seats they should win doesn’t fluctuate a tremendous amount. I’ve had them on three seats since December 2021 except for one month, and I’ve never had them go up or down by more than one seat month on month. Thus seeing them go up by two this month is, I think, worth discussing a little more.

We are going to focus on Dublin because (a) that is where the seat gains are and (b) outside of Dublin, their prospects of picking up seats is minimal, even if Mick Barry is able to hold on in Cork North Central.

PBP have been averaging around 4% in Dublin polls this year, with this month’s 4.9% representing a high this year. This might not look particularly significant, but it is. When we are dealing with relatively small vote bases, small swings can have a big impact – the increase of 0.8% since September represents a relative movement of 19%. These kind of fluctuations are naturally fragile and caution should be exercised, but three of their five best polls in Dublin this year have come in September and August. That’s an indication that an upward trend might be beginning.

And this is enough to put them up two seats – incumbent Geno Kenny in Dublin Mid-West, and a return for Ruth Coppinger in Dublin West (of course, this is based on her running again, something I have no clear information on, so have to assume). This does show the challenge in projecting smaller parties – small changes in polls can have significant impacts, but that’s also the case in real life, where going up or down by a very small amount can have a very big impact.

Also, here’s a fun fact that I’m not sure means anything but still: their previous highest on the RPA in Dublin was 4.5%, and that was in October 2021. There’s something about that month and revolutions…

Now, you might be wondering about why PBP are up when the swings this month show them down 1.6 points in Dublin on their GE2020 performance, and there’s a simple answer to this: PBP are underpolled in Dublin by about, well, about exactly that much – this is based on a historical analysis of all polls leading up to the 2020 election. This is the first month since I began tracking where the model has PBP matching their performance in the last GE here.

Most polling bias comes out in the wash between different pollsters, but PBP in Dublin being underpolled in Dublin is consistent. And for a party that got 6.5% of the vote in Dublin in 2020, 1.6% is very significant – compare that to another small party, the Green Party, who in Dublin were overpolled by 0.2%, and you get an understanding why this is something that, at this scale, needs to be adjusted for.

Interesting issue polling this month

Aside from the aforementioned polling on the budget there were a few interesting things coming out of specific issue polling in the IPSOS and RedC polls. I’ll cover the IPSOS one first as I have a real problem with how some of the questions were phrased – something that is not down to the polling company, but the people who commissioned the poll; so we can continue today’s theme of laying the blame for this squarely at the feet of the Irish Times.

As Sinn Féin find themselves dealing with an awful rant from one of their councillors about refugees, as a small group of people use platform manipulation to make a racist hashtag trend on Twitter in Ireland, IPSOS asked two questions about refugees. Both questions came back with negative outcomes, but both questions were phrased appallingly. The framing for refugee numbers was about “too many” coming to Ireland (61% agree) and accepting refugees from Ukraine “no matter how many arrive” (56% disagree).

This is irresponsible and commits to a framing where an endless amount of people are arriving here, and I have to seriously question what the Irish Times’ motivations were in having leading questions like this. I would also question if they understand how this contributes to an atmosphere of an increasingly emboldened far-right looking to exploit refugees to stoke up xenophobia, but then I remember that time they gave an uncritical platform to an internet Nazi and refused to acknowledge they did anything wrong. So they understand damn well.

RedC have a few things that are worth mentioning. One big insight is on whether or not people think Ireland can avoid a recession – slim majorities of FF and FG voters (52% and 53% respectively) believe Ireland can, but only 20% of SF voters feel the same. I know that’s hardly shocking news, but it is useful to be able to put a number to the vibe that SF are benefitting from people’s justified concerns about the economic future as well as their current conditions.

Also, a small one but a nice one – increased pedestrianisation of cities and towns has over 60% support in every demographic. A reminder that it really isn’t a majority of people who seek to interrupt these schemes, and hopefully something that Dublin City Council could bear in mind next time they are considering caving to three angry car park owners.

Oh, and one final thing from IPSOS, that I believe is much more important than the budget narrative when it comes to positives for the government – only 37% of people want a general election now, with 57% saying the government should serve a full term. While the number of people who want an election is significant, this more importantly shows that the calls from some parts of the opposition for an immediate election have a way to go before they represent an imminent threat to the government.

Why does the same data produce different projections?

So, here’s something interesting. Ireland Elects/Toghann Éire, who usually publish a seat range, this month have also published a precise projection of most likely seats. This broadly aligns with what I have projected, but there are some differences. I thought it would be interesting to talk about what might be driving those – bearing in mind I don’t know their exact methodology, this is a little bit speculative, but it might give some insight into why there’s no perfect way to project things.

I will preface this by saying Ireland Elects are great and this isn’t a criticism of what they are doing or of their outcomes – there’s validity in different methods, especially with something as uncertain as polling. This is about how different conclusions with equal potential validity can be reached, not an argument that what they are or are not doing ^{(also they have Labour on zero seats which is always extremely funny)}

We can see some of the differences by comparing their map to mine – I’ve laid these out in a chart below:

This looks pretty close, with only 8 variances, but some of these numbers are due to projections cancelling each other out; there are 17 seats where the two projections have different outcomes. These are below:

Let’s look at what I think is driving these differences, again bearing in mind that I am making assumptions about IE/TÉ’s methodology – which I am very much open to correction on if people have more information than I do.

Firstly, I think we have slightly different base polling averages. This isn’t massively significant – most differences are within one point, but may have an impact on smaller parties and may be a partial contributor to some of the differences on Labour and PBP. You can see my averages here and theirs, on their much nicer than mine website, here.

Secondly, and this is a big one: I think from looking at these differences that IE/TÉ use national figures rather than regional ones. There are two things that flag this strongly to me – Sinn Féin and Labour. On SF, IE/TÉ have them ahead of where I do in Dublin, while the opposite is true for Munster. Munster is where SF have the biggest swing since 2020 (+87.7%) and Dublin the smallest (+5.6%). Thus it logically follows that using regional figures will favour them in Munster, while using national ones will favour them in Dublin.

Similarly, Labour are much stronger than their national average in Dublin (+11.9% swing vs -9.1%), so that explains the difference on two of their seats, with Wexford very much being due to the challenges with Independents (see below).

I think there are good arguments for both to be valid. I use regional figures to get as close to the constituency level as possible and get a more nuanced view of the variance in party support by area. However, this does come at the increased risk of a greater MoE on all of these numbers.

Thirdly, Independents. This is the biggest driver – in nine of the 17 constituencies where there’s disagreement, it is over Independent candidate. And it’s not clean either – with four going one way and five the other. I think this mostly comes down to how hard it is to get a clean gauge on how Independents are doing, something I have discussed repeatedly on here.

If you read my bit about the PBP above, you’ll have seen that there are certain polling biases I adjust for. And INDs in Connacht-Ulster is one of them. In 2020 their polling average revealed a systemic underpoll of 3.3%. I attempt account for this; I doubt IE/TÉ do. This is probably why I have 4 IND candidates successful in the region that they don’t. I believe adjusting for an underpoll that is outside the MoE is the right thing to do, but I also understand the risk of polluting data with historical trends that may no longer hold true, so again, both approaches are valid here even if they produce different outcomes.

That said, I’m not surprised to see a number of highly marginal ones (eg Laois-Offaly, Galway East, Sligo-Leitrim) split differently. Independents could win all these seats. They could lose all of them. There’s a good chance the data is wrong about Danny Healy-Rae. It’s very difficult to project and I’m not surprised to see a difference here.

A final, also minor point: I’m not sure if IE/TÉ factor in transfer projections. I don’t see this as being key in any of the areas where we have different results, but it may be a contributory factor in combination with the above. Either way, again, this is fine – if you don’t include transfers you are missing a big chunk of data, but if you do include them you are dealing with extremely volatile data than doesn’t necessarily align with broader voting trends. There is no perfect solution here – either including or excluding them is inevitably problematic.

In conclusion, as you can see there’s multiple ways to operate with the same data where uncertainty exists, and I think there’s validity to most of these approaches. Most importantly though this should serve as a reminder of what this is – projections, and nothing more. The data isn’t perfect, the methodologies aren’t perfect and we are all presenting what is likely from those imperfect bases, not what is definitely going to happen or a hot opinion.

That’s really important to remember that when navigating the world of political commentary and analysis, which often seeks to cast things as absolutes and certainties, and is incentivised to seek out the big controversial narratives.

Seat Changes

Five changes again this month, all in Dublin, and four of them can be attributed at least in part to Fianna Fáil’s recent polling struggles in the capital.

Clicking on the linked name for each constituency should jump to the relevant section of this page. Changes on this page indicate changes from September 2022’s projections; changes on constituency pages indicate changes from current composition.

Note: The projections reflect, and always have reflected, most likely outcomes. So if a final seat is more likely for candidate X over Y, the model will show X winning the seat.

This does not mean the scenario where Y wins doesn’t exist, or even is necessarily unlikely (there’s a lot of marginal calls!). It also does not mean that every single “most likely” scenario will come true; statistically that in and of itself is probably not going to happen. This is true from from a simple probability point of view, even if we ignore deficiencies in underlying data. A projected result merely means that the model thinks X winning is the most likely outcome.

• Dublin Bay South (FG +1 FF -1)
• Dublin Central (SD +1, FF -1)
• Dublin Mid-West (PBP +1, FF -1)
• Dublin Rathdown (GP +1, FF -1)
• Dublin West (PBP +1, SF -1)

A reversion of last month’s changes here, as the model favours Fine Gael to take a second seat here at the expense of Fianna Fáil. This isn’t because of an increase in support for FG – their Dublin support is static, but because of the problems for FF in Dublin in recent polling; they have dropped from 15.5% to 13.8% on the RPA in the region since September.

Still, I expect this to continue to be volatile for a while for reasons I have discussed numerous times before – there are two seats here that should be viewed as very uncertain, and any combination of Fine Gael, Fianna Fáil, the Greens and Labour is very plausible.

I also came across some interesting scuttlebutt – there’s rumours of Fine Gael leader Leo Varadkar relocating to this constituency for the next election. I don’t believe it until I see something concrete, but if he were to, he’d comfortably take a seat. Whether that would help a second FG candidate by dragging them over the line, or hurt them by gobbling up first preferences that may not transfer, who knows?

Another symptom of Fianna Fáil’s bad month of polling in Dublin shows this potential pick-up disappearing for now, though there is also the impact of having more non-B&A polls, who are very harsh on the Social Democrats; their polling is up marginally this month versus September. Either way, Gary Gannon should now be favoured to hold his seat, but this is far from certain; the Greens aren’t far behind here still and all the caveats I mentioned last month still apply.

This is a combination of Fianna Fáil difficulties and that slight tick up for PBP that gives Geno Kenny an edge here. This is still very tight and Kenny is probably the most transfer-friendly (or transfer-dependent depending on how you wish to spin it) person I have winning a seat at the moment. He can definitely do it – he got elected in 2020 off the back of Paul Gogarty’s transfers even after he himself seemed to believe it was over, so he should never be counted out.

Still, as I have said before, transfer projections should always be treated with a healthy degree of caution, and a lot of what will happen will depend on Sinn Féin’s strategy and performance – the model currently assumes they will run three candidates but only win two seats. If they only run two, or run three and get three, the calculus will change substantially – both would likely significantly reduce the amount of transfer available to hum.

Fianna Fáil struggles again in this one, and despite having a not-great month of polling themselves, the Greens move back ahead of them here. Other than that not much has changed since last time, this is still a constituency that loves doing strange things at election time

As there isn’t much else to note here, let me tell another story about how mad this place is. In 2014, Sorcha Nic Cormaic became the first ever Sinn Féin councillor elected in the Dundrum LEA. At the next election in 2019, they voted her out of her council seat so hard that she finished bottom of the poll, behind PBP, Aontú, Eirígí and a bloke who was running solely on the basis that he was Shane Ross’ mate. In GE 2020, less than a year after this, she got party’s best ever result in the Rathdown/former Dublin South constituency. This place just does stuff sometimes.

Here’s one we haven’t seen for more than a year – PBP/Solidarity favoured to win a seat in Dublin West, assuming that Ruth Coppinger is the candidate. I do not know if this will be the case or not, but the model assumes unless demonstrated otherwise that candidates will be the same as the prior election. Regardless, this is a good look for PBP/S and a result of a month where they have ticked up. While the Greens have ticked down and look a bit more distant from competing for the seat, that’s not what’s important here; it’s that PBP/S may now be able to overtake a second SF candidate.

This is very marginal and transfers will be important so I would not be surprised if this continues to fluctuate, but this would be a huge pick-up for them if they could pull it off at SF’s expense – especially given the narrative that SF are going to squash the rest of left – which polling in Dublin doesn’t necessarily support at this stage.

As mentioned in discuss Dublin Bay South this month, if there is truth to the rumours about Varadkar leaving here, I’m not sure what that would do to the FG vote. Nothing good I suspect but at this point, there is no point speculating on the basis of rumours.

Other notes

• Dublin Bay North: After a while under the radar, this is showing as very close again with Fianna Fáil struggling. Them, Labour and a theoretical second SF candidate are an almost negligible distance apart, and the Social Democrats are very close behind.
• Galway East: Unless polling trends change for Independents next month (which they might well!) we may see Fianna Fáil be favoured to hold their seat here ahead of Seán Canney (IND).
• Laois-Offaly: The model still shows Carol Nolan (IND) as very close to likely retaining her seat, probably overtaking a second candidate from Fianna Fáil. Then again this constituency might well not exist by the next election – a prime candidate to be redrawn – so who knows.
• Limerick County: Nothing really of note here, just a fun observation that right now the model shows this being settled on the first count. I don’t think that will necessarily happen in reality – I expect O’Donoghue (IND) to be a bit competitive, or one of the big three to mess up their candidate strategy – but it is interesting if not particularly significant.
• Sligo-Leitrim: Similar to Galway East, we may soon see Fine Gael be favoured to hold their seat here ahead of Marian Harkin (IND). But then again Marc MacSharry (IND, ex-FF) is back doing Marc MacSharry things again, so this entire constituency is a bit of a question mark.
• Wexford: Labour’s polling in Leinster has dropped off recently, so I suspect the model will show this changing soon unless they pick back up. They’ve been cycling up and down in the region for a while now though, so this may continue to fluctuate.

Thanks so much for reading! This website is done entirely in my spare time and run without ads, so if you want to donate, please do so via Patreon or Ko-Fi – any support is greatly appreciated and 100% of any donations received are invested into the costs of running and improving the site.