The Three Divides of French Electoral Geography
       |           

Welcome, Guest. Please login or register.
Did you miss your activation email?
May 09, 2021, 09:03:57 AM

  Talk Elections
  Other Elections - Analysis and Discussion
  International Elections (Moderators: Gustaf, أندراوس, Hash)
  The Three Divides of French Electoral Geography
« previous next »
Pages: [1]
Author Topic: The Three Divides of French Electoral Geography  (Read 2963 times)
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« on: August 28, 2016, 04:40:05 AM »

Well folks, this is officially my 40,000th post. It kinda scares me, but I figured that, seeing how the past 5,000 or or 10,000 posts have been almost entirely worthless, I might try to start making up for that. I thought a way to do so could be to share a few of the findings from the research project I've been working on this summer. As the thread suggests, it's about French electoral geography, so I'm hoping at least a few people here will find it interesting. Smiley

One thing I've been doing for this project is using Principal Components Analysis on election results by (pre-2015) canton in metropolitan France. For those who aren't familiar, PCA is a statistical technique that reduces the dimensionality of a dataset - meaning that it takes a dataset with n variables and allows you to summarize it in i<n components. The way it does that is that it finds the linear function that best explains the aggregate variance of the dataset (the first component), then takes the variance that remains unexplained and finds another linear function (the second), and so forth until all the variance is explained.

The whole project is a lot more messy, but I found amazingly clear results for the 2012 presidential election, so I figured I'd share those. What I ended up finding there is that 94% of the variance across cantons could be explained by just three components (that's a lot even for a PCA). This means that, with just these three variables, you know almost everything of how each canton voted. In addition, these components have very clear substantive interpretations (in that their correlations to various candidates makes a lot of sense). So I mapped the value each canton takes in all three components (not manually though, I used a geocoding website) and here's what it looks like.

The first dimension, which, alone, explains about half of all the variance, can be seen as representing the left-right divide. All the left-wing candidates load negatively on this component, while right-wing ones load positively. When put on a map, it provides a very familiar picture of France's political divide. It's basically the map that you see in a runoff between a right-wing and a socialist candidate. The left is strong in the Pyrenees, the Massif Central, most of Brittany, most of the Paris area except the West, and scattered old industrial strongholds in the North. The right, meanwhile, dominates the East from Champagne to Lyon, the Mediterranean shore (including Corsica), and most of the suburban/rural Northwest besides Brittany.



The second component, meanwhile, captures the divide between "populist" or "protest" candidates and those of the political "establishment". Basically, candidates that you see as traditionally being against the "system" load positively there, while more mainstream ones load negatively. This divide explains an additional 30% of the variance. Here, you see very different patterns (which is bound to happen because, by design, the components of a PCA are completely uncorrelated with one another). This is the traditional East-West divide that you often see in analyses of the FN vote: populism dominates everywhere North and East of the Paris area as well as in almost all of Languedoc and Provence, and is also strong in the Garronne valley, North-central France, and Corsica. The "establishment vote", meanwhile, is concentrated in the big cities and their immediate suburbs (Paris, Lyon, Toulouse, Nantes, Bordeaux - but not Marseille) in the "Grand Ouest" (the area stretching from lower Normandy to Poitou), the Massif central, the Southwestern corner, and the Alps-Jura region.



The third dimension, which explains an additional 15% or so, is not as easy to interpret substantively. Basically, it pits abstention against almost all other candidates (to various extents), so it can be seen as a divide over turnout. Cantons that load negatively are less likely to turn out irrespective of their position on the left-right and populist-establishment axes, but beyond that it's hard to say what characterizes them. On the map, however, it appears clearly that abstaining areas are clustered in very specific regions: almost all of Île-de France, thin lines along the Northeastern border and Côte d'Azur, and Corsica. Most of the rest (but especially Brittany and the Southwest) turns out above average.



All in all, I'm not that knowledgeable in French electoral geography, so I'm sure others will provide more detailed interpretations of what each axis means. Still, I hope the maps themselves were a worthwhile contribution. Smiley
Logged
DavidB.
Atlas Icon
*****
Posts: 11,923
Denmark


Political Matrix
E: 0.97, S: 5.30


Show only this user's posts in this thread
« Reply #1 on: August 28, 2016, 05:43:03 AM »

Very fascinating, thanks for sharing.
Logged
Filuwaúrdjan
Realpolitik
Atlas Institution
*****
Posts: 63,702
United Kingdom


Show only this user's posts in this thread
« Reply #2 on: August 28, 2016, 07:25:03 AM »

Map B has more in common with the traditional Catholic/Anti-Clerical divide than Map A does, which is interesting.
Logged
DC Al Fine
Atlas Icon
*****
Posts: 13,688
Canada


Show only this user's posts in this thread
« Reply #3 on: August 28, 2016, 07:32:00 AM »

What parties were included in the "populist" part of Map B? I assume FN and FG (or its constituent parties) were in there. Anyone else?
Logged
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« Reply #4 on: August 28, 2016, 08:18:56 AM »

Map B has more in common with the traditional Catholic/Anti-Clerical divide than Map A does, which is interesting.

Definitely. In this regard the old Catholic West has become the area where it remains hardest for the new populist right to make inroads (and as such it might be the area that allows Juppé to defeat Sarkozy in the primary... God willing). Of course it's never been an area where the hard-left could hope to make inroads, so politics there gravitates more toward the "center" (and of course it helps that Bayrou is originally from the party that used to embody the Christian-Democratic tradition in France when such thing still meant something).


What parties were included in the "populist" part of Map B? I assume FN and FG (or its constituent parties) were in there. Anyone else?

Le Pen is the candidate that loads most strongly on the "populist" side, yeah. Mélenchon from the FG also contributes to it, as does abstention. Arthaud and Poutou (Trotskyists), Dupont-Aignan (right-sovereignist), and invalid votes also load in this direction, but only marginally. I think it makes sense to understand it as an axis that has a lot of Le Pen and a bit of Mélenchon and nonvoters on one side, and Hollande, Sarkozy and Bayrou on the other.
Logged
Grand Wizard Lizard of the Klan
kataak
Jr. Member
***
Posts: 1,890
Vatican City State


Political Matrix
E: -4.52, S: 5.39

Show only this user's posts in this thread
« Reply #5 on: August 28, 2016, 04:33:22 PM »

Great post, very interesting.
Logged
Zanas
Zanas46
Sr. Member
****
Posts: 2,933
France


WWW Show only this user's posts in this thread
« Reply #6 on: September 01, 2016, 08:16:52 AM »

Very interesting work, thanks.

First observations :
- Map B seems to me like a standard recent FN vote map. Does it differ from such a standard FN vote map somewhere ?
- Map C is more interesting to me. Corsica has its own explanation, for being... you know, Corsica. Other than that two main zones can be spotted : Île-de-France and the borders. Île-de-France abstains the most because of a) immigrants who don't want or don't know how or why to vote, and b) urban way of life with many young people and alternative types. The borders, on the other hand, may be abstaining more than the country as a whole because a number of French people there just work on the other side and their lives are more affected by what's going on on the other side, so they lose interest in French politics. Which would explain why zones such as Genève, Nice, Béarn, Forbach or Maubeuge, widely ranging on the socio-economic spectrum, all tend to abstain more than the country.

I'm not explaining the spot of abstention around North-western Cantal.

I think some parts of map B also correlate with Protestantism, but not all. Causse, Vendée, Deux-Sèvres, Béarn, coastal Calvados, south of Lyon come to mind. Of course, other Protestant regions like the lower Rhone valley and Gard were lost in translation more recently. Bretagne and Paris have other explanations.
Logged
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« Reply #7 on: September 01, 2016, 09:38:06 AM »

- Map B seems to me like a standard recent FN vote map. Does it differ from such a standard FN vote map somewhere ?

That's a very good question. Le Pen's results account for about two thirds of that axis' variance, so yeah, there's bound to be considerable overlap. To see if there is any notable difference, I mapped the FN vote using the same scale (ie, a scale based on standard deviation).



Based on this, it looks like, compared to Le Pen, overall populism is weaker in the Grand Ouest, in central France, and in the Right-dominated Northeast, while being stronger in the Southwest, the eastern Paris suburbs (see all the purple in the 93 in Map B), and in the working-class North. This makes sense since Sarkozy's and Bayrou's vote loads on the pro-establishment side while Mélenchon's and abstention loads on the populist side. Still, you're right that Le Pen is the dominant influence here.


Quote
You must be logged in to read this quote.

This explanation makes a lot of sense, thanks. Smiley It's interesting that this urban-young lifestyle that leads to abstention seems to be also present on the Western side of Île-de-France (since immigrants definitely aren't what's driving it there).


Quote
You must be logged in to read this quote.

That's very interesting as well. I had very little idea of where the last remnants of Protestantism were (I'd never have guessed Vendée, for example!). I guess Protestants and "real" Catholics are now united in preferring "respectable" parties. Tongue
Logged
parochial boy
parochial_boy
YaBB God
*****
Posts: 3,890


Political Matrix
E: -8.38, S: -6.78

Show only this user's posts in this thread
« Reply #8 on: September 02, 2016, 01:29:16 PM »

Very interesting work, thanks.

First observations :
- Map B seems to me like a standard recent FN vote map. Does it differ from such a standard FN vote map somewhere ?
- Map C is more interesting to me. Corsica has its own explanation, for being... you know, Corsica. Other than that two main zones can be spotted : Île-de-France and the borders. Île-de-France abstains the most because of a) immigrants who don't want or don't know how or why to vote, and b) urban way of life with many young people and alternative types. The borders, on the other hand, may be abstaining more than the country as a whole because a number of French people there just work on the other side and their lives are more affected by what's going on on the other side, so they lose interest in French politics. Which would explain why zones such as Genève, Nice, Béarn, Forbach or Maubeuge, widely ranging on the socio-economic spectrum, all tend to abstain more than the country.


For what it's worth, the Genevois and Forbach both have very large immigrant populations as well; the darkest shade of brown on the Haute Savoie border with Geneva is Annemasse, which is very diverse.

On a sort of tangent, is there any reason that so many of the old rural centre-right strongholds have moved to the left in recent decades. I believe the secularisation of Catholic areas has been behind areas like Brittany becoming staunchly on the left, and places like Cantal trending left as well, but there must be some other driving factor, as you wouldn't normally expect rural areas to go left?
Logged
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« Reply #9 on: September 02, 2016, 02:24:32 PM »
« Edited: September 02, 2016, 02:26:13 PM by I did not see L.A. »

On a sort of tangent, is there any reason that so many of the old rural centre-right strongholds have moved to the left in recent decades. I believe the secularisation of Catholic areas has been behind areas like Brittany becoming staunchly on the left, and places like Cantal trending left as well, but there must be some other driving factor, as you wouldn't normally expect rural areas to go left?

In addition to the trend you mentioned, the leftward trend in the Massif Central area is also a product of personal vote. Chirac, who embodied the right for almost two decades, was from Corrèze and always was extremely popular there. Hollande isn't from Corrèze, but he's made it his electoral fiefdom since 1988 and as such was also quite popular there. As such, both were great fits for Corrèze and the neighboring area (which roughly include Lot, Cantal, Aveyron and Lozère).

As for Brittany, it had always been more amenable to the left than the rest of the Catholic West, and the trend you describe (in addition to the fact that the French right has moved further and further away from the traditional Christian-Democratic brand of leftism) has been enough to move it entirely to the left. You should ask Hashemite for a more detailed analysis, since he's the ultimate expert on French (and especially Breton) political geography.
Logged
Filuwaúrdjan
Realpolitik
Atlas Institution
*****
Posts: 63,702
United Kingdom


Show only this user's posts in this thread
« Reply #10 on: September 02, 2016, 06:31:04 PM »

It's not so much secularisation as the general weakening of that issue across the board. Note that elsewhere in the country a lot of anticlerical bastions that voted solidly Left for over a century are now strongholds of the Right. Le Var rouge sounds like a joke now.
Logged
DC Al Fine
Atlas Icon
*****
Posts: 13,688
Canada


Show only this user's posts in this thread
« Reply #11 on: September 03, 2016, 05:56:13 AM »

I guess that makes sense given that weekly mass attendance is <5% of the population. Or has it been like that for a long time and the divide was more about whether you 'identify' with the church or not?
Logged
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« Reply #12 on: September 03, 2016, 06:02:56 AM »

It's not so much secularisation as the general weakening of that issue across the board. Note that elsewhere in the country a lot of anticlerical bastions that voted solidly Left for over a century are now strongholds of the Right. Le Var rouge sounds like a joke now.

I don't think Var (and the rest of Côte d'Azur)'s rightward trend should mainly be ascribed to a change in issue salience. What changed first and foremost is the département itself - it went from a largely rural area of small landowners to a sprawly tourist-dominated abomination. And let's not forget about the pieds-noirs.
Logged
Zinneke
JosepBroz
Sr. Member
****
Posts: 2,133
Belgium


Show only this user's posts in this thread
« Reply #13 on: September 03, 2016, 06:20:58 AM »

It's not so much secularisation as the general weakening of that issue across the board. Note that elsewhere in the country a lot of anticlerical bastions that voted solidly Left for over a century are now strongholds of the Right. Le Var rouge sounds like a joke now.

I don't think Var (and the rest of Côte d'Azur)'s rightward trend should mainly be ascribed to a change in issue salience. What changed first and foremost is the département itself - it went from a largely rural area of small landowners to a sprawly tourist-dominated abomination. And let's not forget about the pieds-noirs.

Pieds-noirs, like Mélenchon?
Logged
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« Reply #14 on: September 03, 2016, 08:28:04 AM »

I guess that makes sense given that weekly mass attendance is <5% of the population. Or has it been like that for a long time and the divide was more about whether you 'identify' with the church or not?

Apparently, it's been mostly stable since the late 1980s (and even in the late 1960s, it was only 20%).

But yeah I think that some regions have maintained something of a "Catholic identity" even as religious practice faded. At least, it's clear that something about these regions still drives a preference for parties of the Christian-Democratic tradition and a strong distaste for the FN.
Logged
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« Reply #15 on: September 03, 2016, 09:41:35 AM »

It's not so much secularisation as the general weakening of that issue across the board. Note that elsewhere in the country a lot of anticlerical bastions that voted solidly Left for over a century are now strongholds of the Right. Le Var rouge sounds like a joke now.

I don't think Var (and the rest of Côte d'Azur)'s rightward trend should mainly be ascribed to a change in issue salience. What changed first and foremost is the département itself - it went from a largely rural area of small landowners to a sprawly tourist-dominated abomination. And let's not forget about the pieds-noirs.

Pieds-noirs, like Mélenchon?

I'm pretty sure Mélenchon is the exception, politically speaking. So was Camus, for that matter. Tongue
Logged
parochial boy
parochial_boy
YaBB God
*****
Posts: 3,890


Political Matrix
E: -8.38, S: -6.78

Show only this user's posts in this thread
« Reply #16 on: September 03, 2016, 12:23:56 PM »


I don't think Var (and the rest of Côte d'Azur)'s rightward trend should mainly be ascribed to a change in issue salience. What changed first and foremost is the département itself - it went from a largely rural area of small landowners to a sprawly tourist-dominated abomination. And let's not forget about the pieds-noirs.

I think so. Outside of the Mediterranean basin, a lot of the old anti-clerical left wing strongholds have stayed that way, notably the Pyrenees and the South West. Aude has trended right, but Ariege and the Hautes Pyrenees are still two of the most reliably left wing departments in the country.
Logged
DC Al Fine
Atlas Icon
*****
Posts: 13,688
Canada


Show only this user's posts in this thread
« Reply #17 on: September 03, 2016, 02:24:32 PM »

It's not so much secularisation as the general weakening of that issue across the board. Note that elsewhere in the country a lot of anticlerical bastions that voted solidly Left for over a century are now strongholds of the Right. Le Var rouge sounds like a joke now.

I don't think Var (and the rest of Côte d'Azur)'s rightward trend should mainly be ascribed to a change in issue salience. What changed first and foremost is the département itself - it went from a largely rural area of small landowners to a sprawly tourist-dominated abomination. And let's not forget about the pieds-noirs.

Pieds-noirs, like Mélenchon?

I'm pretty sure Mélenchon is the exception, politically speaking. So was Camus, for that matter. Tongue

Melenchon and Camus were Pieds Noirs?! Learn something new every day.
Logged
1,066,892 Likud voters can't be wrong!
Nathan
Atlas Star
*****
Posts: 28,109


Show only this user's posts in this thread
« Reply #18 on: September 03, 2016, 02:28:21 PM »

It's not so much secularisation as the general weakening of that issue across the board. Note that elsewhere in the country a lot of anticlerical bastions that voted solidly Left for over a century are now strongholds of the Right. Le Var rouge sounds like a joke now.

I don't think Var (and the rest of Côte d'Azur)'s rightward trend should mainly be ascribed to a change in issue salience. What changed first and foremost is the département itself - it went from a largely rural area of small landowners to a sprawly tourist-dominated abomination. And let's not forget about the pieds-noirs.

Pieds-noirs, like Mélenchon?

I'm pretty sure Mélenchon is the exception, politically speaking. So was Camus, for that matter. Tongue

Melenchon and Camus were Pieds Noirs?! Learn something new every day.

Derrida as well, believe it or not.
Logged
Zinneke
JosepBroz
Sr. Member
****
Posts: 2,133
Belgium


Show only this user's posts in this thread
« Reply #19 on: September 04, 2016, 07:07:07 AM »
« Edited: September 04, 2016, 07:09:30 AM by JosepBroz »

It's not so much secularisation as the general weakening of that issue across the board. Note that elsewhere in the country a lot of anticlerical bastions that voted solidly Left for over a century are now strongholds of the Right. Le Var rouge sounds like a joke now.

I don't think Var (and the rest of Côte d'Azur)'s rightward trend should mainly be ascribed to a change in issue salience. What changed first and foremost is the département itself - it went from a largely rural area of small landowners to a sprawly tourist-dominated abomination. And let's not forget about the pieds-noirs.

Pieds-noirs, like Mélenchon?

I'm pretty sure Mélenchon is the exception, politically speaking. So was Camus, for that matter. Tongue

You're right, a quick google search shows the pieds-noir polled 18% in favour JMLP in 2007. Compared to his 10% nationally.
Logged
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« Reply #20 on: October 07, 2016, 01:08:46 PM »

So, I've been thinking of a new pet project - which I obviously won't have time for until who knows when, but still. I'd like to make a map using a multidimensional color scale similar to what Homely did here, that would map the left-right and populist-estabishment axis simultaneously. I suspect it could make for a really pretty map.

Before I start though, I was wondering which map key would look best on a map. The possibilities are endless, but those are the 4 I've come up with:



Which one do you like best?
Logged
ObserverIE
Jr. Member
***
Posts: 1,796
Ireland, Republic of


Political Matrix
E: -3.87, S: -1.04

Show only this user's posts in this thread
« Reply #21 on: October 07, 2016, 05:46:31 PM »

So, I've been thinking of a new pet project - which I obviously won't have time for until who knows when, but still. I'd like to make a map using a multidimensional color scale similar to what Homely did here, that would map the left-right and populist-estabishment axis simultaneously. I suspect it could make for a really pretty map.

Before I start though, I was wondering which map key would look best on a map. The possibilities are endless, but those are the 4 I've come up with:



Which one do you like best?

Personally, the third one, but I'd go with yellow for left establishment and a deeper green for centre establishment to make them more distinct - you have quite a few colour pairs that are hard to tell apart (for my eyes, at any rate).
Logged
God-Empress Stacey I of House Abrams
Antonio V
Atlas Institution
*****
Posts: 53,446
United States


Political Matrix
E: -7.87, S: -3.83

P P P

Show only this user's posts in this thread
« Reply #22 on: October 07, 2016, 06:37:06 PM »

you have quite a few colour pairs that are hard to tell apart (for my eyes, at any rate).

Yeah, I realize that's a real problem. It always happens when I try to create a continuous color scale (ie, one that associates every value a specific shade, rather than lumping them into broad categories). In order to do that, I need a mathematical function. But whether you use the RBG or the HSL systems, the mathematical difference isn't always proportional to the difference perceived by the human eye.

Sadly, those four scales are the least worse I have in this regard.
Logged
MASHED POTATOES. VOTE!
Kalwejt
Atlas Institution
*****
Posts: 57,474


Show only this user's posts in this thread
« Reply #23 on: October 08, 2016, 02:13:21 PM »

Fourth I guess.
Logged
Pages: [1]  
« previous next »
Jump to:  


Login with username, password and session length

Terms of Service - DMCA Agent and Policy - Privacy Policy and Cookies

Powered by SMF 1.1.21 | SMF © 2015, Simple Machines

Page created in 0.193 seconds with 12 queries.