Better together?

Two of Poland’s opposition parties - the conservative Civic Platform (PO) and the liberal Nowoczesna (N)  - formed the Civic Coalition (Koalicja Obywatelska, KO) earlier in the year to take the fight more effectively to the governing populist-nationalist Law and Justice (PiS) party in the upcoming local elections. While the nature of their intended cooperation in next autumn’s general election remains uncertain, with much hinging on how they perform at local level, pollsters have begun to treat the coalition as a going concern at national level.

While few polls to date have asked respondents specifically about their support for KO rather than their support for PO and N separately, there are sufficient data to gain a sense of where such a coalition might stand. While PiS remain unassailable on over 40% of the vote, nearly a third (32%) of declared voters (those who would vote, and know for whom they would vote) would opt for KO. The two major forces are significantly ahead of their competitors, with only the radical-right Kukiz’15 movement polling in double figures (11%), while the social-democratic Democratic Left Alliance (SLD) just about stays above the electoral threshold on 7%, and the agrarian centrist Polish Peasant Party (PSL) adopts its customary mid-term position just below the threshold.

IMG_8343.PNG

There is currently little to go on by way of trends, given the sparsity of the data. What appears to be movement in the graph below is more likely to be smoothed-out noise. If conventional vote intention polls are any guide, no significant change should be expected; for all the volatility of Polish politics at present, support for parties is remarkably stable.

IMG_8345.PNG

Current levels of support, weighted to reflect support for the relevant parties in each of Poland’s 41 constituencies in 2015, suggest that were an election to be held tomorrow, the situation for the two main forces would not change markedly, with PiS falling just short of a majority (a majority they would presumably swiftly achieve by tempting switchers from the organisationally fragile Kukiz’15 movement), while KO would have slightly fewer than the combined share of seats for PO and N in 2015. The only parties whose fortunes would change significantly would be the SLD, which would re-enter the Sejm after a fallow parliamentary term, and PSL, which would finally drop out of parliament for the first time since 1989.

IMG_8346.PNG

Are the coalition members better off together? At most, we can say “Moderately so”.  KO certainly provides a lifeboat for N, which has dipped below the threshold in several recent polls. Running in a coalition helps PO get over the psychologically important 30% barrier, so long a bane of PiS during their sojourn in the wilderness of opposition, and a threshold PO has struggled consistently to surmount during the last three years. But these are cold comforts, or at best lukewarm ones. The absence of failure is hardly the most resounding measure of success.

Cucumber season

In Polish politics, the term 'cucumber season' (sezon ogórkowy) refers to the downturn in political activity during the summer months. Parliament goes into recess, politicians restrict their activity to Instagramming carefully curated holiday photos, and everyone disappears to their allotments to pick cucumbers. Cool, soothing, inoffensive cucumbers. Everyone likes cucumbers. A bit of sour cream and the ubiquitous dill is procured, and we have a mizeria - a classic Polish salad. Everyone likes mizeria.

It would be fair to say that the cucumber season is no longer entirely in vogue in Polish politics, which has seen a summer full of controversy and contestation, from protests over the unconstitutional purge of the Supreme Court to the abject cynicism of the "just-talking-to-Varsovians" pre-campaign campaign by Warsaw mayoral candidates Patryk Jaki and Rafał Trzaskowski. With so much at stake right now, nobody can relax for a minute. Instead, cucumber season has shifted to the realm of public opinion, at least where vote intention is concerned. After a certain degree of fluctuation in the first quarter of 2018, the last few months have seen remarkable stability in attitudes towards parties.

The governing Law and Justice (PiS) party has restored its double-digit lead, continuing to poll in excess of 40% when undecided voters (typically around 10% of declared voters overall) are discounted, while its main rival, Civic Platform (PO) has apparently plateaued on around 27%. The liberal party Modern (Nowoczesna) appears to have arrested the slide that briefly took it below the electoral threshold in May, but remains unable to pull itself above 5% for now. The social-democratic Democratic Left Alliance (SLD) has consolidated its mini-resurgence; while last year it often polled below the threshold, it now enjoys a stable 8% of support, which is approximately the same share commanded by the nationalist Kukiz'15 movement. The Polish Peasant Party (PSL) maintains its customary position just below the threshold, while minor parties Together (Razem) and Freedom (Wolność) both have around 2% support.

With the exceptions of SLD, which would enter parliament, and Nowoczesna, which would lose a substantial percentage of the seats it gained in 2015, a hypothetical election this weekend would not have much of an impact on the distribution of power in the Polish parliament. PiS would still command a slight majority.

I write this overlooking my in-laws' plot of land outside Warsaw, where chilli peppers grow next to the wilting remainder of this season's cucumbers. Chilli-pepper season would perhaps be a better analogy for Polish politics right now, but it looks like voters have got used to the heat.

State of play

After a period of relative quiet, a flurry of polls towards the end of May suggests that levels of support for parties have stabilised over the last month after some clear shifts since the start of the year.

According to the most recent calculations of the pooled polls model, Law and Justice (PiS) remains clearly in the lead on 38%, which is roughly the same amount of support it obtained in the 2015 elections. Civic Platform lags behind on 26%. All other parties have less than 10% support, with the Democratic Left Alliance (SLD) and the Kukiz'15 social movement on 9%, while the Polish Peasant Party (PSL) and the Modern party (Nowoczesna) are on 5%.

Looking at trends since the start of the year, there are several noteworthy developments. During March, support for PiS fell by a significant amount, but the party has arrested that decline since the beginning of April. During the same period, support for PO has steadily risen, although the rate of growth has slowed in the last couple of months. Several recent polls put PO on nearly 30%, although this is partly offset by the regular monthly CBOS polls, in which PO always does significantly worse than in other polls.

After stabilising its position in March, Nowoczesna has slipped again since the start of April, in several cases polling well below the 5% threshold. There is no certainty that the party would return to parliament if an election were held tomorrow. On the other hand SLD, which at the start of the year was not polling significantly above the threshold, has established a stable base of support sufficient to gain it seats in parliament. Support for Kukiz'15 and PSL has remained largely stable since January.

On current polling, PiS would not have a single-party majority and would have to rely on Kukiz'15, which would make for a numerically comfortable but ideologically volatile coalition, or on PSL, which would make for an ideologically more stable but numerically wafer-thin majority. It should be kept in mind that it is by no means certain that PSL and Nowoczesna would make it into parliament: if both were to drop definitively below the threshold, PiS would probably gain the seats it needs for a single-party majority (perhaps with a few defections from Kukiz'15), while PO would be the biggest beneficiary from the absence of Nowoczesna.

Seats_weighted.png

Modern downhill technique

While everyone else sat in the sunshine during Poland’s extended May holiday, the liberal Nowoczesna (Modern) party went skiing.

After a week that saw the departure of three prominent legislators, including founder and former leader Ryszard Petru, the release today of an IBRIS poll conducted over the weekend will not have steadied nerves. While all single polls are subject to the usual caveats, and 2.9% is at the lower end of the margin of error on earlier Nowoczesna polls, the direction of travel is consistent with recent tendencies.

IMG_1603.PNG

The pooled polls model puts Nowoczesna right on the electoral threshold, with a probability of 0.54 that it is over the threshold (0 meaning not over; 1 meaning definitely over). 

IMG_1604.PNG

If we assume that Nowoczesna is over the threshold, and that constituency-level support for the party is much the same as it was at the last election, then it would currently have 7 seats in the Sejm, 21 fewer than it gained in 2015.

IMG_1606.PNG

The map of support for Nowoczesna shows that it would have two seats in the Warsaw metropolitan constituency, but otherwise would be limited to single seats.

The consolation is that one of these seats borders the Nowy Sącz constituency, which incorporates the Tatra mountains. Handy for some skiing.

IMG_1605.PNG

Caveat suffragator - the limits of the PO&N coalition

In recent weeks, pollsters have started asking voters whether they would opt for a coalition of Civic Platform (PO) and Nowoczesna* (N), rather than asking which of these two parties they would vote for. I am sceptical of two things: one, the value of polls that ask about fictional coalitions; two, the likelihood of these two parties maintaining a common front for long enough to fight the 2019 election on a joint list.

There is a caveat to the caveat; these two parties have agreed to fight November's local elections on the same ticket. Yet in turn, there is also a caveat to that caveat: over the last few days, Nowoczesna have seen three prominent members abandon ship, including party founder and unceremoniously defenestrated leader Ryszard Petru. This leaves Nowoczesna a qualitatively different beast to the one it was when most of these polls were conducted.

Nevertheless, since it is likely that pollsters will continue to ask voters about their intentions toward this coalition - hopefully not at the cost of continuing to poll them on their intentions toward the component parties - I have modelled these pooled polls separately, and will return to them from time to time.

As the most recent figures show, the gap between the two major players is significantly closer if Civic Platform and Nowoczesna join forces, with the PO&N coalition on around 34% of the vote, and Law and Justice (PiS) on around 39%. For the most part, this closing of the gap can be attributed to the accumulation of the two parties' independent vote shares, but there is a 3-4% 'coalition bonus' that accrues to the joint endeavour. Nevertheless, PiS almost certainly remains ahead of PO&N.

The story among the remaining parties remains relatively unchanged. The Democratic Left Alliance (SLD) just edges into third position with around 9% of the vote, with the Kukiz'15 social movement on around 8%. Perennial rotating coalition candidate the Polish Peasant Party (PSL) just dips below the threshold, while none of the remaining parties come close to breaching it. 

There is little to look at so far when it comes to trends, as there have only been six relevant polls. Nothing yet suggests that the prospect of a coalition has stimulated any movement in the direction of increased support for this initiative, but it is too early to rule out the possibility that it might.

If the current figures are translated into seats, each party's national scores weighted by its constituency-level support in 2015, then a PiS-Kukiz'15 coalition offers the only viable route to a governing majority. PiS would lose 23 seats compared with their 2015 figure, while PO&N would gain 14 compared with the combined totals for the constituent parties at the last election.

Looking at the absolute differences in seat shares for PiS (light-dark blue) and PO&N (light orange), two things stand out: PiS remains dominant in its eastern and south-eastern strongholds, and PO&N prevails - albeit less convincingly - in the west and major urban centres. This is unsurprising: there was a strong correlation between the size of Civic Platform's and Nowoczesna's constituency-level vote shares in 2015. 

This suggests that the breathless response of the liberal media to this last set of polls needs to be tempered with a degree of caution. Yes, a PO&N coalition gives Poland's liberal-conservative opposition a greater overall share of the vote; and yes, there seems to be a bonus for cooperation. However, as things stand, it is likely that a significant proportion of this support will pile up in constituencies where the two parties do better than average already. In the absence of any marked decline in support for PiS, it will struggle to make significant inroads into less favourable terrain. It is not enough for the opposition to succeed; PiS must fail.

* I use Nowoczesna's Polish name rather than the awkwardly adjectival English translation "Modern".

How to kill zombies.

The 'zombie map' of misleadingly colour-coded regional differences has made another unwelcome return on social media. In a previous post, I attempted to illustrate the regional strengths of parties in a more useful way, and the following map builds on that. Instead of plotting each party's seats in each constituency against the district magnitude, in the plot below I use the party's highest constituency seat share as the upper limit (see the colour bar in the legend for specific party totals).

Plotting the data like this allows us to visualise the extent to which - assuming 2015 constituency levels of support - national-level polling would result in concentrations of seats in particular areas or, conversely, a more geographically dispersed allocation of seats.

For example, we can see that Kukiz'15's support is broad and shallow: it currently polls at around 10%, gaining 44 seats which are spread across all but one constituency. By comparison, while SLD is currently polling around 8%, its reach in certain areas is more limited, particularly in the south-east.

This map also highlights the particular geographical concentration of support for PiS, while making it clear that the party also draws significant support from constituencies in the west. Although that map of PiS versus PO votes looks striking and does bear out some legacies of the past, the plots above give a more useful visualisation of actual party strengths. Poland is not as divided as the zombies would have you believe.

Maps and legends.

Poland's political geography has often been the topic of excited - if at times rather overheated - comment, much of which tends to focus on the correspondence between the lines along which Poland was partitioned and voting patterns in the present day. The legacies of partition - and related legacies of economic and institutional development and cultural consolidation (or de-consolidation) - have clearly played a role in influencing electoral choices, although the definitive article that parses the various strands of correlation and causation remains to be written. 

Some of the folk wisdoms about Poland's electoral geography continue to outlast their empirical falsification, but we can say two things for sure:

  • PiS and PO have tended to be more successful in different areas.
  • At the 2015 election, PiS made significant inroads into former PO strongholds (which turned out not to be so strongly held).

Assuming as I do - in the absence of better information - that the geographical distribution of votes in the 2015 election still remains a relevant variable for understanding how national polling results translate into seat shares, current regional strengths look something like the following.

PiS's stronghold is in the east and south. In the Rzeszów constituency, PiS would, on current polling, have 12 of the 15 seats available (note that the shading of the plots corresponds not to the individual district magnitudes, but to the gradient from 0 to 20, which is the largest district magnitude, in the Warsaw metropolitan constituency). By contrast, in Gdynia, which has a district magnitude of 14, PiS would have only 5 seats.

Support for PO has tended to be more evident in the west and north, and in major cities. The map roughly captures this, although the differences are not as stark as in previous years, and in any case this starkness was often a product of the colour-coding strategies of newspapers' graphic designers, who have an understandably limited tolerance for shades of grey.

 Instead of the clash-of-the-primary-colours maps we tend to see in newspaper articles and memes, a better way to get a handle on the current state of play in Poland's electoral geography is to look at the absolute differences between projected seat shares for PiS and PO. The map below is based on a colour coding that runs from orange (PO have all the available seats) through white (PO and PiS have an even number of seats) to dark blue (PiS have all the available seats). The numbers correspond to the extent of the advantage enjoyed by the party which is ahead in that constituency.

As the map shows, PiS's dominance is evident in the fact that it clearly outdoes PO in its own strongholds, while restricting PO to only a couple of constituencies in which it outperforms PiS (only in Poznań would PO's advantage over PiS exceed a single legislator). While there is clearly a geographical gradation to the electoral politics of Poland's Big Two parties, western Poland is far from terra incognita to PiS.

Maps for the minor parties are less revealing, partly because these parties do not necessarily have strong electoral profiles, but also because these maps show where parties gain seats, and not where they accumulate votes (which might give more of a clue about where these parties have the potential to be stronger, but that's a future blogpost). 

The decline in support for Nowoczesna over the last couple of years is evident in the rather washed-out character of its electoral map, although in any case support for the party was not especially regionally concentrated in 2015. While Nowoczesna could still count on two seats in Warsaw and one in the surrounding commuter-belt 'doughnut', for the most part it would be picking up scraps after PO in western regions.

Support for Kukiz'15 is broadly dispersed, and this is reflected in the fact that it would - on a current 9% vote share - capture at least one seat in all but three constituencies, with Wałbrzych, Nowy Sącz and Poznań resistant to the party's charms, and Kielce and Opole - party leader Paweł Kukiz's home turf - more open to them, resulting in two seats apiece.

The heartland of PSL in central and eastern Poland is still in evidence, although on current polling it could hope for no more than a single seat in any given constituency.

In recent weeks, SLD has made something of a comeback in the polls. While its geographic appeal has been harder to pin down than that of PiS or PO, from the performance of the ill-fated United Left alliance in the 2015 elections we can conclude that it would do badly in the south-east of Poland (where it has generally tended to struggle), but pick up a handy couple of seats in Warsaw thanks to its enduring appeal among the substantial contingent of urban leftists who remain unconvinced by the parvenu Razem's long-term viability as an electoral option.

Some movement, but no crisis.

After a pretty fallow first quarter of 2018, there has been a lot of interest in polls over the last couple of weeks, sparked by a shock Kantar poll that had PiS down by 12 percentage points. Several subsequent polls have shown PiS losing ground, and PO and SLD gaining slightly. However, CBOS's monthly poll gives PiS 46% of the vote, which is very similar to the figure it has averaged for much of the last six months.

A word about CBOS is in order. It is no secret that CBOS polls tend to show higher levels of support for PiS than the average poll does. Typically, this is somewhere between 2pp and 4pp. It is also no secret that CBOS is a foundation financed from the state budget, and supervised by the office of the prime minister. As a result, every time a CBOS poll is published, it attracts a torrent of abuse on social media from opposition supporters who allege pro-government bias. I have no strong opinions either way on this issue, but think it is much more likely that CBOS's face-to-face methods create a "bonus" for the governing party thanks to the phenomenon of social desirability. In any case, the poll-pooling model is intended to mitigate these kinds of biases.

The results of the most recent model, run after the CBOS results were released, suggest that the recent fall in support for PiS has bottomed out. However, this may simply be an effect of the latest poll, so I will reserve judgement on that for now. What we can see from the latest poll is that PiS (40%) remains well ahead of its nearest competitor, PO (27%). Indeed, we have to go back at least a year to find a point at which PiS was not at least 10pp ahead. No other party gains more than 10% support, with Kukiz'15 (9%) and SLD (8%) definitely exceeding the 5% threshold for seats in parliament. Nowoczesna (6%) would probably make it into parliament on current polling, while PSL (5%) is as likely to lose out as it is to gain seats. No other parties have a chance of making it into parliament.

One recent trend is clear: support for PiS has fallen since the end of February, although it still remains around 40%. PO and SLD have seen slight increases in support during March. However, Nowoczesna continues to perform badly; while in December it had around 10% support, at present it has half this. PSL continues to bump along the electoral threshold, while all other parties are below it.

If we assume that constituency-level support for parties is comparable to what it was in the 2015 elections, then PiS comes close to the seats needed for a single-party majority. In this situation, it would likely have little trouble gaining the five seats it would need through transfers from other formations, particularly Kukiz'15 and PSL, and potentially also PO and SLD. 

It thus seems premature to speak of a "polling crisis" for PiS. As the map below shows, PiS (shade of blue) continues to outperform PO (shades of orange) in the majority of constituencies (the numbers in each constituency correspond to the advantage enjoyed by PiS (in the blue constituencies) or PO (in the orange constituencies). In several areas in the west of Poland where PO was formerly dominant, PiS remains ahead in votes and seats. The party's abrupt decision to return controversial ministerial bonuses and submit a bill to slash legislators' salaries by 20% seems a rather hasty response to a few relatively unfavourable polls.

Pooling the Poles

What is this blog for?

The purpose of this blog is simple: it pools polls of Poles.

That is, it uses a statistical model pioneered by Simon Jackman (2005) "that tracks changes in voter support over time by pooling the polls, and corrects for variation across polling organisations" to provide "a less biased and more precise estimate of vote intentions than is possible from any one poll alone". I have adapted this model using code kindly provided by Kai Arzheimer.

This model allows me to take all the information contained in the various polls published by CBOS, Kantar, Millward Brown, IBRIS, Pollster, Estymator (and, on occasion, other pollsters) to estimate current levels of support for all relevant parties, trends in support over time, and the likely seat shares of parties both nationally and in each constituency.

A caveat is required: if all the polls are consistently wrong, then the estimated totals will also be wrong. For this reason, poll-pooling models are often better at capturing the shifts of trends than they are at providing snapshot estimates of party strengths at any given point in time. However, at worst, a pooled model will do no worse than individual polls. Generally, it will do better. While Polish polling is volatile - as is the Polish electorate - a model very similar to the one I currently use managed to predict fairly accurately the results of the 2015 election.

Why is this blog needed?

Although reporting of polls in the Polish press has improved significantly over the last few years, there is still a tendency to pay too much attention to individual, eye-catching polls (or clusters of same) without paying attention to longer-term trends. As a result, too much credence is given to noise, and not enough to signal. Since I started pooling the Polish polls in 2014, I have tried to persuade people to pay more attention to trends. At the same time, it is clear that there is also an interest in where parties stand at any one given time, and how that translates into likely shares of seats in the legislature. By pooling the polls and then extracting estimates of present party strengths, I am able to provide a more balanced picture of the current situation than reports on individual polls can. Previously, I published these estimates on Twitter and in an RPubs mini-blog, but have come to believe it would be better to set up a dedicated blog to present these estimates and, where necessary, provide some comment on them.

Why are you doing this?

Because nobody else is.