French presidential election, 2022

[Hash]

Cleaned up the thread, without infractions. Keep it civil and clean.

[Hash]

LR online primary 'congress' first round results from today, before the runoff tomorrow/Saturday:

Éric Ciotti 25.59%
Valérie Pécresse 25%
Michel Barnier 23.93%
Xavier Bertrand 22.36%
Philippe Juvin 3.13%

t/o 80.80% (113,038 votes)

Pretty huge surprise: nobody saw Ciotti come close to winning. If you're wondering, all are basically hard-right and obsessively focused on immigration and security, except maybe Juvin. Ciotti is the most far-right: he'd support Zemmour 'his friend' over Macron, and had the most far-right platform of any candidate. Pécresse opportunistically also campaigned on pretty hard-right themes.

[Hash]

I had never even heard of Ciotti before. Who is he?

Deputy for the Alpes-Maritimes (for a Nice constituency) since 2007, boss of the powerful LR federation in the department since 2018, former president of the departmental council between 2008 and 2015. He began as a protégé of Christian Estrosi, the mayor of Nice since 2008, but since 2017-8, when Estrosi became more moderate and more macronista-inclined, the two men have become sworn enemies.

His far-right standing is somewhat more credible and genuine than the cynical, opportunistic hard-right posturing of Barnier, Bertrand and Pécresse. He's been known for his focus on security/authority, public order and immigration issues for most of the time he's been in parliament - in 2010, he was the author of a law which suspended or eliminated family benefits in case of frequent, unjustified school absenteeism of a student (repealed in 2013), and in 2011 of another law which created a military supervision for juvenile delinquents over 16.

He was a loyal Sarkozy and Fillon ally (and has made a point of saying that he was the only candidate who stood by Fillon in 2017 after the scandal). In 2017, he refused to endorse Macron in the runoff, and now says that the only difference between LR and RN is LR's capacity to govern. He claims to be loyal to the heritage of Pasqua and Séguin (i.e. souverainiste right), although he's a hardcore '80s neoliberal which Séguin wasn't.

[Hash]

What's the difference between workers and employees?

Employees is a broad category of, well, public and private sector employees - secretaries, administrative assistants, office workers, salespersons, service workers, cashiers, retail employees, support staff, caregivers, servers, personal care workers, estheticians, salaried hairdressers, housekeepers and so forth. Public sector employees under this definition are from the lowest categories of the public sector, those jobs which don't require a university degree. Police, firefighters and military are also considered employees under the Insee definition. Employees are overwhelmingly women (75%).

Manual workers (ouvriers) includes skilled and unskilled workers in industry, crafts/trades and agriculture. Again, it's a broad category which includes carpenters, fitters, electricians, plumbers, mechanics, repairmen, welders, bakers, butchers, cooks, drivers, truckers, taxi drivers, longshoremen, equipment operators, unskilled manual workers, sanitation workers and agricultural workers including fishermen and forestry workers. It is an overwhelmingly male-dominated category (79%).

Both categories are sometimes grouped together and considered lower socio-professional categories (CSP-): the levels of education and earnings of both groups are similar (the lowest levels of educational achievement, although employees are more likely to have a high school diploma, Bac or professional certificate, than workers, and workers are more likely to have no degree). In modern France, workers are found in rural areas, on the most remote outskirts of large urban areas, but are increasingly absent from most large cities, particularly the most white-collar and prosperous urban cores, like Paris (also, for historical reasons, there are fewer workers in the south/southeast); employees are found throughout the country, and they probably make up much of the 'proletariat' in the cities and inner suburbs.

[Hash]

For reference, the vote transfers in 2017 looked something thus (looking at Ifop and Ipsos election day surveys):

Ifop (their numbers seem to contradict each other a bit):
Mélenchon: FBM 52-54% | Panzergirl 12-14% | Abst./blank/invalid 32-36%
Hamon: FBM 73-79% | Panzergirl 3-4% | Abst./blank/invalid 17-24%
Fillon: FBM 51-55% | Panzergirl 21-23% | Abst./blank/invalid 21-26%

Ipsos:
Mélenchon: FBM 52% | Panzergirl 7% | Abst./blank/invalid 41%
Hamon: FBM 71% | Panzergirl 2% | Abst./blank/invalid 27%
Fillon: FBM 48% | Panzergirl 20% | Abst./blank/invalid 32%

Today, in Ifop and Ipsos' rolling:

Mélenchon: FBM 27-29% | Panzergirl 20-23% | Abst./etc. 48-53%
Jadot: FBM 49-57% | Panzergirl 7-18% | Abst./etc. 33-36%
Pécresse: FBM 45% | Panzergirl 27-28% | Abst./etc. 27-28%
Zemmour: FBM 15-13% | Panzergirl 74-79% | Abst./etc. 6-13%

So some noteworthy gains by Panzergirl with Mélenchon voters and even the moderate left, although she was losing by a landslide in 2017 and now she's only losing by a small margin. Remains to be seen how solid her transfers are with those people. Some somewhat smaller gains also from the 'moderate' right voters. But abstention will be even higher in 2022 than in 2017 (and 2017 abstention was already high), particularly on the left.

[Hash]

I guess we were due for one shock poll from a little-known pollster that will bring out bad takes from people who have no clue what they're talking about/know zilch about French politics.

AtlasIntel is a Brazilian pollster which seems to have been the 'most accurate' pollster in the 2020 US election as well as in Argentina in 2019. But they have no track record whatsoever in France, and their recent poll in Colombia wasn't terribly accurate on some results (though not terrible).

I'm not a specialist in analyzing technical details of polls, but it's strange how they have no crosstabs or weighting by socioprofessional category like all other pollsters, but have crosstabs by religion (including Muslim and Jewish), and that they have a weird '55-100' age category (rather than 65+).

Also, their first round numbers are actually not all that favourable to Panzergirl compared to other polls: she's at 21.3% (decimals!) vs 27.8% for Macron, and they have Pécresse quite a bit lower (5.3%) and Zemmour a fair bit higher (12%) (also they find Pécresse and Hidalgo to be more unpopular than Zemmour), which implies that Macron gets very bad transfers in the runoff - indeed, they find that Macron's runoff transfers from Mélenchon and especially Pécresse are much worse than in Ifop/Ipsos' regular tracking polls.

Also, the poll is filled with a bunch of questions (Ukraine-Russia, EU, immigration, Putin, sanctions, atomic bombs etc.), and some of the questions/answers seem to have been run through Babelfish a few dozen times: "êtes-vous en faveur ou contre l'invasion de l'Ukraine pour l'armée de la Fédération Russe ?" or whatever the hell this headache is:



Finally, just looking through the presentation, there are a few basic French grammar mistakes and weird phrasings (pas voté, 'ne votera pas') that makes me think at least some of their release was made with the help of Google Translate. Let's see if they even register this poll with the national polls commission (probably not, they don't speak French).

PS: can we avoid vile grand remplacement nonsense here?

[Hash]

Also, I forgot the obvious in my post above: this poll is an outlier: there has not been one other poll which shows Panzergirl leading in the runoff. All other polls basically have her at 46-48% now. On top of that, their Mélenchon/Macron runoff numbers are also an unusually favourable outlier for Mélenchon, who gets 46% compared to roughly 40-42% in all other polls.

[Hash]

I voted earlier today, after a very long line-up which took an hour and a half to get through. Cookies for anyone who can guess who I voted for.

[Hash]

Polling summary time:

Macron - almost all firms have him on 26%, but a handful push a little higher. None below 26%, none above 28%.

Le Pen - more diversity here: a high of 25% and a low of 21% with most hovering in between the two, generally slightly tilting towards the lower end.

Melenchon - between 16% and 18% with most settling on 17%.

Zemmour - French firms pretty much all have him on a functional 9%, but foreign firms who have had a go show higher figures.

Pecresse - French firms have her on either 9% or 8%, but those foreign firms who have had a go show lower figures.

Jadot - between 4% and 5%.

Roussel - between 2% and 3% with the consensus being a functional 3%.

Lassalle - between 2% and 3%.

Hidalgo - 2%.

Dupont-Aignan - between a functional 2% and 3%.

Poutou - 1%.

Arthaud - microscopic.

What I'm mostly getting here is that there is a lot of herding going on and that even moderate polling errors could upset a lot of apple carts if they were to occur.

An interesting metric measured by the two top pollsters, Ifop and Ipsos, is the 'vote certainty' - about a quarter or a bit less of those certain to vote were still not completely settled on their choice. Macron and Panzergirl have their vote well tied down, with 80-90% of their voters certain of their choice for them; Zemmour and Mélenchon also have pretty solid votes, with 75-80% of their voters certain of voting for them in the end. On the other hand, Pécresse and especially Jadot have a good chunk of their current voters still not completely certain (around 65-70% of Pécresse voters are certain, 45-55% of Jadot voters are certain).

The last big Ifop poll for Le Monde showed the following second choices of those who were uncertain of their choice:



Macron and Mélenchon would appear to be those who stand the best chance of benefiting from Jadot and Pécresse's wavering voters (although Panzergirl can hope that some Zemmour hesitators vote for her strategically). The same poll also showed the low-high ranges if half of the hesitating voters switched to their second choice:

Macron 25-28.5%
Panzergirl 21-24.5%
Mélenchon 15.5-19%
Zemmour 8-10%
Pécresse 7-9.5%
Jadot 4-6%

Jadot?

Congratulations!

[Hash]

If the runoff is Macron vs Melenchon who do you think you'll vote for?

I have no strong opinions about such a runoff because I strongly dislike both individuals. I'd feel very comfortable and happy voting for April Ludgate (my 2017 first round vote) in this scenario.

I imagine I'll need to hold my nose and immediately vomit in the likely Macron/Le Pen runoff, though.

[Hash]

I voted earlier today, after a very long line-up which took an hour and a half to get through. Cookies for anyone who can guess who I voted for.

I thought you were Canadian?

I have dual citizenship.

[Hash]

FWIW, this forum has an unwritten tradition/rule against sharing leaked exit polls/estimates before the official time - both as a courtesy to people who like the excitement and surprise of waiting for the official time, and because a lot of leaks are fakes/inaccurate. I'm not sure if I'll enforce it strongly this year, but I am putting that out there to let people know.

What we do have that isn't a leak is Ipsos' abstention estimate: 26.5%. Higher than in 2002, so not a record high, but would be the second lowest turnout. It is roughly at the upper end of the good pollsters' estimates, so to me this suggests that their final polls should be rather accurate barring some last-minute swings/surges.

[Hash]

Ifop's turnout estimate is a bit higher: 75%. This is, I think, about 2% higher than their last turnout estimate in the last rolling poll.

FWIW, most polls showed that turnout intentions were lowest among left-wingers.

[Hash]

What happened? Well, basically, as the second choice/hesitators numbers I shared yesterday, Pécresse hesitators went to Macron in the end and sank her further while there was strategic voters on the left breaking for Mélenchon (as expected) which shaved some points off from Jadot and Roussel.

Le Pen at 23% is still a good number for her - better than in 2017 - and the combined far-right is stronger than it's ever been.

[Hash]

Three basic scattered impressions based on looking at maps:

- Huge numbers for Mélenchon in the cities and banlieues: he clearly coalesced a large swathe of the left behind him, strategically, and while it wasn't enough to matter for the runoff, it made for some impressive headline numbers in a lot of areas, particularly the greater Paris region (49% in the 93!) but also victories in towns like Strasbourg and Nantes which, even from 2017, you wouldn't expect to be Mélenchon-favourable. There's definitely a class element to it but his electorate is kind of cuts across class in urban cores, perhaps more than in 2017 (when you had a lot of left-liberals for Macron and Hamon's more bobo-centric electorate). On the other hand, his results in some 'white-working class' (sorry for using this US term) areas and rural areas are not particularly impressive, lower than 2017 in places.

- Macron is the candidate of the old mainstream right. This was not the case in 2017, but it already kind of was in 2019, and got clearer this year. He swept the very affluent suburbs and resort towns which had voted Fillon in 2017 - like, for example, Neuilly-sur-Seine or western Paris. Of course, Macron's electorate is still quite different from the old mainstream right's electorate: he is missing a lot of the traditionally conservative rural areas, particularly in the south and east, and has retained regions like Brittany, which are more left-leaning from a pro-European, Christian democratic centrist tradition, and is stronger in the cities than the right was in 2007/2012.

- I expected the phenomenon of Zemmour doing well in right-wing affluent areas but I didn't expect it'd be that hilarious: he is consistently outperforming Panzergirl, sometimes by a not insignificant margin, in nearly all traditional wealthy right-wing areas -- 17.5% in Paris 16ème, 18.8% in Neuilly-sur-Seine, 12.9% in Saint-Cloud, 22.4% in Saint-Tropez (behind Panzergirl but not by much). A lot of those votes came from people who voted for Fillon in 2017 because Panzergirl never did that well in those places: I feel like Zemmour got the stereotypical racist reactionaries who think Panzergirl likes the yucky poor people a bit too much.

[Hash]

As I've been doing since 2017, here are quick graphs based on Ipsos' analysis of the electorates. As usual, interpret this as you would any poll crosstabs, particularly with the smallest candidates.

Gender:


Age:


CSP:


Monthly income:


Self-perceived social class:


Diploma:


Employment status:


Satisfaction with life:


Religion/religiosity:


2017 vote:


2019 EP vote:


I hope to make similar graphs based on Ifop's numbers once they come out, as well as my usual maps at the constituency level.

[Hash]

Quick and dirty map of headline results at the departmental level:

[Hash]

On the topic of the first arrondissement and prisoners' votes, BFM TV calculated them:

Mélenchon 45.78%
Le Pen 20.28%
Macron 18.63%
Zemmour 3.82%
Pécresse 2.61%
Lassalle 2.03%
Poutou 1.69%
Jadot 1.62%
Roussel 1.06%
Hidalgo 1.06%
Arthaud 0.87%
NDA 0.56%

Unsure about turnout or how many votes were cast, the article cites 13,672 detainees who chose to vote by mail before the election. You could calculate them by adding up the actual 1st arrdt. precincts and figuring the difference with the official results for the arrdt. reported by the Ministry.

https://www.bfmtv.com/politique/elections/presidentielle/resultats-presidentielle-pour-leur-1er-vote-par-correspondance-les-detenus-plebiscitent-melenchon_AN-202204110613.html

[Hash]

Expat vote results are out by consular constituency and they're hilarious. Zemmour won in Israel, Russia (minus Yekaterinburg constituency) - no comment - Thailand and the Almaty constituency. He also won 31% in the Bahamas and 25.4% in Miami. I will just say one thing about this: LOL.



There's no embassy in North Korea, the embassy in Yemen is closed but the consular constituency still exists and reported 0 votes, the embassy in Syria is closed but 2 votes were cast (one each for Zemmour and Pécresse), 1 vote was cast in Afghanistan and it was for Macron and the election couldn't be held in Shanghai because ZeroCovid disenfranchises people.

Some constituencies are non-contiguous (Malawi is with RSA, Nunavut is with Montreal, Jamaica is with Panama, PR is with Miami etc.)

Full details here: https://www.diplomatie.gouv.fr/IMG/pdf/elections_pr_1er_tour_2022_-_pourcentages__cle0f9d7c.pdf

[Hash]

Some while ago I shared graphs of Ipsos' analysis data of the first rounds - now, with some delay, here is Ifop's data, both equivalent and some additional details. My warnings about small subsamples, MOE etc. obviously stands.

Gender


Age

Unlike Ipsos, Ifop doesn't find a strong-ish far-right vote for Panzergirl and Zemmour with 18-24 young voters (Zemmour is in fact weakest here, and Panzergirl is at 18% rather than 26%), and Mélenchon's dominance is even bigger. Broad patterns in other age groups quite similar, though.

CSP

Basically Ifop finds Mélenchon to be stronger with manual workers (27 vs 23 in Ipsos), but weaker with the cadres and intermediate-grades.

Employment status


Diploma/education


Income


Religion

Unlike Ipsos, this data comes from a separate poll for La Croix. Now, for obvious reasons, I have very strong doubts about the 'regularly practicing Catholics' subsample here - I'd trust Ipsos' numbers far more! On the other hand, the Muslim numbers here are very interesting: Mélenchon is absolutely dominant, with nearly 70% (!) - not very surprising, given that all other major candidates engaged in various shades of, uh, you know. In 2017, the Muslim vote (per Ifop) was: Mélenchon 37%, Macron 24%, Hamon 17%, Fillon 10%, Le Pen 5%.

Union membership

This also comes from a separate poll.

Partisan self-identification

(lol at the PS numbers)

2017 presidential vote


2017 presidential vote - runoff


2012 presidential vote

Very interesting stats! For Macron, the numbers compared to Ifop's same poll in 2017 show how his electorate moved right: in 2017, 45.5% of his electorate had voted Flanby in 2012 and only 17% had voted for Sarkozy -- put otherwise, only 17% of Sarkozy's 2012 voters went to Macron in 2017, while 48% of Flanby's 2012 vote had gone to Macron in 2017. Macron still does retain a sizable chunk of voters who were 'left-leaning' pre-2017 (left leaning enough to vote Flanby in 2012), so I guess you could say the realignment he caused was not ephemeral. But still, his 2022 base is much more right-wing than in 2017...

Ideology

There was no similar question from them in 2017 on this but there was a question from Ipsos on similar wording in 2017 (unfortunately, the 'neither left nor right' response was scrapped here by Ifop, which is a shame). Macron still wins over a broad centrist electorate straddling left and right, but more right-leaning now: 39% of 'centre-leftists' voted Macron as did 55% of 'centre-rightists', and 23% of 'right-wingers'. In 2017, Macron won (per Ipsos): 23% of 'left-wingers', 47% of 'rather left-wing', 60% of centrists and 23% of 'rather right-wing'. By makeup, Macron's 2022 electorate is 44% centre-right, 27% centre-left, 21% right-wing and 7% left-wing.

[Hash]

Interesting polling data from Ifop about vote choice by media (print, radio, TV) consumption habits:

Nightly news:


More proof that 24h news channels are a cancer on society and rots brains, and that I should watch Arte even more. TF1's right-wing lean is now split between Macron and the far-right.

Radio:


Yeah, no surprise, I only listen to Radio France.

Newspapers and magazines:



No surprise, far-right rag Valeurs Actuelles had 49% support for the far-right. As for the rest, well, it's clear confirmation that Mélenchon coalesced the left very well.

Various TV shows/channels:


*pretends to be shocked* as regular CNEWS viewers lean to the far-right.

[Hash]

Here's a shaded map of the leading candidate by constituency (shaded by margin/majority)



The map which I was most eager to complete was Zemmour's map which is hilarious



His best constituencies were Paris-4 (16.8%) which includes extremely wealthy parts of the 16e and 17e, Paris-14 (16.4%) which has the rest of the 16e, Alpes-Maritimes-8 (16.6%) which includes Cannes and Alpes-Maritimes-6 (14.5%) which includes Cagnes-sur-Mer and Saint-Laurent-du-Var. Most of his other top results (over 10%) were in the Alpes-Maritimes, Var, Bouches-du-Rhône and Vaucluse. In general, it's a very retro 1980s FN map - similar to the FN's first map, in the 1984 EP elections (with some notable exceptions, like the Seine-Saint-Denis, 'too diverse' now to vote for any far-right candidate), which had patterns which didn't hold for that long (particularly the strong support in affluent areas).

I will make similar maps at the constituency level for all relevant candidates as well as various other 'thematic' maps of interest.

[Hash]

I've finished my candidate maps by constituency (not doing the two Trots because that's statistical noise)

Macron:


I'll soon do a map comparing his 2017 vote to his 2022 vote to reveal the expected patterns, but it's already obvious that he has a more traditional centre-right map than in 2017 - quite obvious in the western Parisian bassin and the inner west (40.2% in Cholet's constituency in Maine-et-Loire, which was solidly right-wing in the increasingly distant past, 41.9% in northeastern Vendée, Philippe de Villiers' old stomping grounds in the bocage, 37.7% in the Vitré constituency in Ille-et-Vilaine which was the most right-wing seat in Brittany not too long ago). His best constituency with 46.7% in Paris-14th (south of the 16e), followed by 46.6% in Hauts-de-Seine-9th (Boulogne-Billancourt) and 45.9% in Paris-4th (north of 16e and parts of 17e). His centre-right patterns are also quite obvious in the Lyonnais, Haute-Savoie, Gironde, Massif Central (though with poorer results in Haute-Loire, Lozère, Corrèze and Creuse). His map is also basically a map of which parts of France are doing well or are more optimistic, or 'France Triple A' to use a term used in a recent Ifop analysis.

Panzergirl:


With stiff competition from Zemmour along the Mediterranean in the southeast, Panzergirl's map is even more northeast/east focused. The vast majority of her best results are concentrated in this northeast quadrant in the Pas-de-Calais, Aisne, Somme, Nord, Marne, Haute-Marne etc.. Amusingly, her own seat (Pas-de-Calais-11th), where she won 45.2%, isn't her best constituency - that 'honour' goes to Pas-de-Calais-10th  (Bruay-la-Buissière, which has a RN mayor, councillors and deputy) with 47.8% followed by 46.2% in Pas-de-Calais-12th, held by RN deputy Bruno Bilde. Her weakness in nearly every major metropolitan centre of importance, even those in the northeast quadrant where she is strongest, is obvious. I'll have more data to share later on the Le Pen vote in relation to major metropolitan centres.

Mélenchon:


As we've discussed before, Mélenchon's 2022 map is strikingly a very urban-focused map - not much in common with the 'radical left' map (or old Communist map) that he still kind of had the influences of in 2012 and 2017. Nearly all but a handful of his 120 or so best constituencies are in urban areas (either cities or inner suburbs), the only exceptions being the Ariège and the Drôme (3rd constituency specifically). In the details, I find it quite telling that in the Somme, his best constituency (with 24.6%) is the 2nd constituency rather than Ruffin's 1st constituency (22.1%) - simplifying things a lot, the former is the wealthier/middle-class urban seat of Amiens (south and northeast parts) while the 1st constituency is a gerrymandered seat drawn by Marleix in the last redistricting as a leftist vote sink. Mélenchon's support was so strong (and concentrated) in certain urban areas that he won over 50% of the vote (!) - something which no other candidate did in metro France - in no less than 12 constituencies, all in Seine-Saint-Denis (his top 5 seats are all there), Bouches-du-Rhône (including 'his' seat, a leftist vote sink which still works), Hauts-de-Seine, Val-d'Oise and Paris. Those who know about the historic political 'reputations' of certain cities will find it quite striking (bizarre in a sense?) that a 'radical left' candidate does so well in Lyon, Strasbourg, Caen, Rouen, Bordeaux etc. On the other hand, Mélenchon's results keep getting worse and worse in old left-wing working-class/industrial basins - he did face PCF competition this time but even that shouldn't be overstated too much.

Jumping to the 'losers', we begin with Pécresse:


This map does a good job at showing that her support is fairly evenly concentrated at low levels throughout the country (many in a range of 3.3%-5%). The map doesn't do that great of a job at showing that Pécresse's vote did spike in a few places over 8%, even 10% - Corrèze, Cantal's 2nd constituency and traditional right-wing strongholds in Yvelines, Paris and Hauts-de-Seine (her best result was 14.2% in Paris-14th... but that's obviously a pathetic result for the right there, given that Fillon still got 57% there in 2017). With the exception of the usual old wealthy right-wing spots, she's quite weak in urban areas but a bit stronger in some, though by no means all, more rural regions - a reflection of her old retiree electorate.

Jadot:


Not an unusual map for the Greens, with their usual strongholds and dead zones, although obviously at lower levels of support than in 2019. Jadot broke 10% in three constituencies, two in Paris and one in Nantes.

Lassalle:


As in 2017, Lassalle has a favourite son effect radiating out of his constituency, where he won 20.8% and finished in second, with other results over 10% in Pyrénées-Atlantiques, Hautes-Pyrénées, Landes, Haute-Garonne, Lozère and Corsica. His vote is both heavily concentrated in rural areas and in the southeast, and there is a strong correlation with the Occitan language, except along the Mediterranean and in Provence.

Roussel:


With Roussel, we find a more traditional Communist map - with the very notable absence of most old PCF strongholds in urban areas, which is quite obvious in Seine-Saint-Denis but also in Marseille and suburban Lyon. He won by far his best result in his constituency, Nord-20th (12.4%), an old Communist stronghold in the Nord's industrial and coal mining basin, and his four other top results were also in the Nord-Pas-de-Calais mining basin. He also did 'well' in old rural communist heartlands like the Allier, Limousin, Dordogne as well as smaller old industrial regions where the PCF did well (Gard's mining area, parts of the Somme and Seine-Maritime etc.). On the other hand, it's quite striking how poorly he did in urban areas, particularly in the Paris region: Roussel's support was concentrated (as Al's nice maps at the commune level in the Petite Couronne show very well) in the last remaining Red Belt suburbs which have PCF mayors. Obviously Mélenchon took a lot of the left-wing vote, and the PCF federation in the 93 was the only one which was in favour of the party supporting Mélenchon and there was likely an 'ethnic factor' in these places too: Mélenchon won the immigrant vote in a landslide, and Roussel didn't do well there (neither did Jadot or Hidalgo).

NDA:


A pretty rural and eastern map for Dupont-Aignan (at fairly low levels everywhere: even in his own seat, Essonne-8th, he got just 6.4%) with strong results, outside of his favourite son vote, in Alsace, Moselle and Haute-Savoie.

Hidalgo:


Behold Hidalgo's pathetic map! She didn't even break 5% anywhere, coming closest with 4.6% in Landes-3rd, an old PS stronghold. Some slightly less pathetic results in the old PS strongholds in the southwest and Limousin, as well as in Brittany. Compared to results elsewhere, she didn't do too poorly in Paris (still a very bad result for the place she's mayor of) but did extremely poorly basically everywhere else in IDF and the broader Parisian basin. Obviously Mélenchon cannibalized a huge chunk of the leftist vote in urban areas, but perhaps Hidalgo as the city-centre mayor was a particularly poor fit, even for the left, outside of Paris but within proximity of it? Also, congratulations to the dying PS for doing just as 'well' (poorly) in rural Alsace than in Marseille and the NPDC mining basin. Good job guys.

[Hash]

Macron 2022 vs. Macron 2017:



Gains, some very strong, in traditionally right-wing areas (the notable exception being parts of the southern Massif Central particularly Lozère and Haute-Loire), and losses or only modest (below average) gains in left-wing areas, something which is of course quite clear in Paris and the Petite Couronne but also in the southwest. His biggest gains were Paris-14th (+19.4%), Paris-4th (17.2%), Hauts-de-Seine-6th [Neuilly] (+15.1%), Vendée-4th (14.1%) and Hauts-de-Seine-9th [Boulogne-Billancourt] (+13.9%). His biggest losses were nearly all in eastern Paris or Seine-Saint-Denis, notably in Paris-6th (-6.9%) and Paris-17th (-6.8%). There is a positive correlation between Macron's gains and Fillon's 2017 vote, at the constituency level (RSQ 0.39).

Mélenchon 2022 vs. Mélenchon 2017:



This map really shows how Mélenchon's gains this year were concentrated in urban areas, beginning with IDF. His biggest gains nearly all came from Seine-Saint-Denis, Paris, Val-d'Oise, Hauts-de-Seine and Val-de-Marne and nearly all of them came in constituencies where Macron did worse than in 2017 -- his biggest gains were in Val-d'Oise-8th (+18.8%), 93-2nd (+18.5%), 93-6th (+17.7%), 93-4th (+17.5%) etc.; his biggest losses were in Pas-de-Calais-10th (-6.7%), Nord-16th (-5.5%), Bouches-du-Rhône-13th (-5.1%) and Roussel's Nord-20th (-5.1%). His losses do have a lot to do with Roussel, but while there is a positive correlation between Roussel's vote and the change in the Mélenchon vote it is a very weak one (RSQ 0.15).

However, if you compare Mélenchon + Roussel in 2022 to Mélenchon alone in 2017, the map is much 'greener'



Mélenchon alone did worse than five years ago in 201 metropolitan constituencies, but Mélenchon and Roussel's combined vote was higher than Mélenchon's 2017 vote in all but 21 metropolitan constituencies. However, regardless of how you look at it, the general outline is the same: Mélenchon (alone or + Roussel's votes) got very big positive swings towards him in urban/metropolitan areas, particularly in IDF, while his vote in many non-urban areas, including rural areas and some smaller urban or old industrial areas where the PCF was formerly strong (Meurthe-et-Moselle, Gard), was far more unimpressive (even with Roussel's vote added, the positive swings were largely modest and below natl average).

After the runoff, I'll hopefully have more first round analysis/maps in addition to some runoff analysis/maps. I hope it's offering a somewhat interesting take on the results.

[Hash]

So is this mostly the north of France outside of Paris where the votes have been counted?

Mostly rural France nationwide.

On that point, two more southern departments are joining Lozère:

Alpes-de-Haute-Provence: 51.45% Le Pen, 48.55% Macron. 67% voted for a candidate, and did not abstain or vote blanc.
Cantal: 56.1% Macron, 43.9% Le Pen. 69.43% voted for a candidate.

We're seeing Macron around 13-14 points below his 2017 result throughout the mostly rural and Southern departments that have fully reported so far. So either the projections are significantly off (unlikely) or, more likely, he's lost more ground than average in those places compared to more urban areas. In other words, the cleavage between metropolitan and peripheral France has gotten even deeper.

I guess I should be happy to have another data point for my postdoctoral research, but I'd like nothing more than to see it proven wrong.

So far.in Ile de France it seems to be more like 5-6%.

The IdF results aren't representative yet: even within those departments, it's the relatively less urbanized areas that report first. So if anything Macron might end up even higher there. But we'll see.

Panzergirl underperforming her 'potential' in IDF and urban areas is not new: I made this map in 2017 based on a rough calculation (based on poll data) of vote transfers:



I look forward to making a similar map later for this election, let's see how it looks - but I broadly agree with your conclusion that the runoff results reinforced the periphery vs core divide even more than in 2017.