Only crazy people vote

Priorities. (click to see original tweet)

Everyone in the US knows that the more removed an election is from a presidential election, with emergency special elections inhabiting the limit point away, the more turnout will be depressed. Furthermore, everyone in the US knows, since the Christian Coalition rode this pony into power, that the lower turnout is, the fewer votes you need to win. So if you excite your base and get them to vote in a number not representative of the population, with just a small number of votes, you can look very important.

This fact of democracy is among the themes emerging out of Sunday’s elections, suggesting that, nationwide, the Front National’s performance (about 12%) was the result of a base more motivated to vote than the average French population, so that when you have global abstention at greater than 50%, the FN’s performance can look out of whack. But does it stand up? Libération began to address this question in their article “FN et abstention : le cocktail que dit non,” which provided two maps showing region-by-region rates of both votes for far right parties (usually the FN) and for abstention. I took crummy photos of the maps and reproduce them here.

Abstention rate. (click to enlarge)

A glance at the abstention rate suggests, and this is helped by how Libé chose to color it, that France seems split in two, with the western half more inclined to vote than the eastern half. Yet Franche-Comté, the sole region in the east with a below 50% abstention rate bucks this trend. Similarly, it’s unclear whether the 51.2% rate in Brittany is that much different from, say, the 53.7% rate in Bourgone. Furthermore, given the national rate of 53.64%, having two colors above and three below the rate further confuses things. It’s useful to have this information, but it needs a bit more context.

Regarding the right-wing vote, Libé was content to lump it in correspondence with the rate of abstention. “A l’est d”une ligne Marseille-Rouen, soit on boude les urnes, soit on vote Le Pen,” they write. Yet in this image, the coloration is even more severe. Not counting Corsica (which is, interestingly, off the pace of both abstention and extreme right voting), there is only one color category for the entire swath of France that voted below the national average. Furthermore, the coloring suggests a stronghold in PACA, where Le Pen père was at the top of the list, and in Nord-Pas-de-Calais, where Le Pen fille was the top of the list. Yet Marine Le Pen’s performance was bested in Alsace, and Picardie brought in a huge result–a result so surprising, that one of my coworkers gasped upon looking at the map.

Extreme Right vote. (click to enlarge)

Libé, or, at least, the article’s author, Eric Aeschimann, commits a grave sin in pushing home his point, however. He provides two examples of the correspondence between the two variables, explaining how in Vitrolles, near Marseille, the abstention rate hit 62%, yet the FN collected 21.5% of the vote. The numbers were similar in Lens: 60.5% and 21.4%. Yet this is (all together now) cherry picking.

The two questions, then, that this post hopes to answer are: is there a demonstrable correlation between the abstention rate and the vote total received by the far right? Might there be a correlation with the other parties, as well? That is, could Libération be picking on the FN out of general, national shame over the 12% result?

Time to make a dataset! I put the abstention rates for each region into a table along with the performances as given in today’s Libé. This conflated the FN and other “far Right” parties, thereby raising their values a bit, but not, I hope, by too much. I then added the results for the PS, UMP, and Europe-Écologie, which Libé reprinted in their own maps. The far Left fall beneath the scope of even this leftwing newspaper, so I relied on online results to tabulate a combined total for the Front de Gauche and the Nouveau Parti Anticapitaliste.

The vote totals, among the 22 regions, distributed into boxplots, show the variance between vote totals in the various regions. Some element of explanation is necessary for some of the outliers. First, the outlier for the PS, its 7% showing in Languedoc-Rossillon, is the result of the explusion from the PS of Georges Frêche, who, still very popular in his region, fronted his own leftist list, which carried the plurality in the region (34%), leaving the official PS list in the dust. The green coalition did not field a list in Corsica (which is itself an outlier in terms of abstention rate), which explains their 0 there. Nevertheless, the squat box for E-É shows that they enoyed very little variance in the nation as a whole. The two outliers are Rhône-Alpes (17.8%) and Île-de-France (16.6%). Just behind them, interesting, is Alsace at 15.6%. Alsace, of course, was one of the only two regions carried by the UMP in 2004, suggesting the possibility of a combined green-pink front to knock over UMP this upcoming Sunday. Finally, the far-left outlier is from Auvergne, where the separate FG and NPA lists combined to net 18% of the vote. Lest one consider this an NPA stronghold, 14 points of that total come from the FG list.

But now that the reader is familiar with my color scheme, I can add the second chart, which plots the performances of the five groups per region against the abstention rate in the region. To show where this argument is going to go, I’ve also included color-coded regression lines for each party that show the correlation between the group/party’s performance and the region’s rate of abstention. The regression lines help one guess other, theoretical values.

So here are the numbers to go with the chart. For the far-Right parties, the adjusted R^2 was .44. R gave a statistically significant level of correlation (p < .01). For the UMP, the adjusted R^2 was .05 with a p of .17. For the PS, -.04, with a p of .72. For the ecologists: .35 and p < .01. And, finally, the for the far Leftists: an adjusted R^2 of .05 and a p-value of .15.

What do these numbers tell us? First, they give credit to Libé‘s suspicion. There is a positive, strong correlation between abstention and the amount of votes the far Right received at the ballot box. The dark blue line racing up the chart suggests that the worse the turnout, the better the result for the FN. For the far Left, PS, and UMP, the correlation is not strong at all, yet the model does suggest that apologists for the far Left are not entirely without reason in complaining that they were particularly harmed by the bad turnout, especially considering their astoundingly negative regression line. Of course, it seems entirely unlikely that if the turnout rose to 60%, the FG/NPA would get 40% of the vote, but that’s the problem with R^2 that close to zero: awful predictive value.

But Libé is not entirely off the hook here, since note that Europe-Écologie also showed a statistically significant relationship between low turnout and high vote totals. This suggests that Green voters are as passionate as far right voters about showing up at the polls, as they do better the fewer people vote. On the other hand, when I remove the big 0 in Corsica from the E-É total, the statistical significance collapses. In fact, only the far right maintains its level of strong correlation with Corsica taken out of the picture, though also further shrinking the sample size, which is its own risk.

Of course, these models are rather simplistic and make outrageous assumptions, like that the distribution of Greens, xenophobes, and commies is more or less evenly distributed in France. Furthermore, it would be helpful to compare these totals with other totals with other rates of abstention over time, and so on and so on. But it’s better when the model confirms your suspicion (“only crazy people vote”) than when it doesn’t.

Tags: Europe-Écologie, France, Front de Gauche, Front National, GIS, Libération, Nouveau Parti Anticapitaliste, PS, statistics, UMP