I recently (finally) had the chance to take a look at the Turkish presidential elections data from August 2014, with the aim of comparing some of the measures of election irregularities I documented from the earlier local elections in March of the same year. The analysis from those elections were essentially a snapshot of potential irregularities with little to say about how this evolved over time.

What I focused on last year was the rather peculiar degree with which the share of invalid ballots appeared to be positive correlated, at the ballot-box level within voting stations, with AKP vote shares (or rather the vote share difference between the AKP and the largest opposition party). Now, with the presidential elections data, I can compare how this correlation differs between that election and the local elections. As for the interpretation of the correlations, here’s what I wrote in April last year:

“An obvious question mark in the analysis is to what extent any correlation represents systematic mistakes, not systematic fraud. Suppose the AKP has a higher support among the illiterate who are more likely to make mistakes when voting. In this case, we would not be surprised if there was a correlation between invalid ballots and AKP support. An explanation would be that those more likely vote for the AKP are also those more likely to make voting mistakes and have their ballots declared invalid. Given the large difference across districts in the large cities in Ankara and Istanbul, one can easily imagine this as a plausible explanation for the simple unconditional correlations.

It is here that the fixed effects used in the previous analysis becomes crucial, i.e.  including fixed effects (FEs) to regressions of vote shares on invalid ballots control for all factors that vary across the FEs. Adding FEs for districts (Ilce) means we’re only looking at variation across ballot boxes within districts, whereas adding FEs for voting station means only looking at variation across ballot boxes within voting stations.

When doing this, although voters going to the same station to vote may still differ along several characteristics, it is much more difficult to argue that this systematically affect their likelihood of making mistakes in voting. The strength of the FEs is thus not that they control for everything, but that they reduce these differences to the point where it is less likely that the remaining differences represent an competing explanation for the correlation.

Furthermore, to the extent that this represent fraud, one would expect the relationship between AKP’s vote share and invalid ballots to be stronger in races with significant competition and less likely in races where the AKP was safe. (For the obvious reason that there is little return to engaging in fraud in races where you’re expected to win without fraud.)”

Below I show such correlations between the AKP’s vote share and the invalid share of ballots, with red representing the August presidential elections and blue representing the local elections. These are essentially the same kind of graphs as the first set of ones I posted here, the main difference being that I now from the start subtract the voting station means from the ballot-box level data. Another difference is that instead of plotting the raw voting data (which, with large amounts of data, makes graphs overly crowded and visually less attractive) I’m plotting equal-sized binned means – essentially a scatterplot of grouped data, with groups made up of equally large number of observations (see here for an easy way to implement such grouped scatterplots in Stata).

I’ve chosen eight different provinces to look at, each among the two largest ones in four groups. The first is Ankara and Istanbul, the two largest, where CHP is the main opposition party; the second is Adana and Mersin, the two largest provinces where MHP is the main opposition party; the third is Gaziantep and Konya, the largest among central Anatolian AKP “safe seats”; and the last is the largest among the provinces where the Kurdish parties tend to be prominent.

What is noticeable is that the dispersion of invalid ballots (in red) is much smaller in the presidential elections (in blue) than in the local elections. (For example, the blue circles on the x-axis run from below -.04 to above .04, whereas the red circles run mostly from -.02 to .02.) This represents a rather consistent difference over time; local elections almost always tend to have larger invalid shares of ballots than in national elections. So, even though in several cities the red line tends to be steeper than the blue line, this doesn’t necessarily mean the presidential elections in cities like Ankara, Istanbul, or even Adana were necessarily more irregular than the local elections.  In fact, if one accounts for the difference in the distribution of the variables, looking at standardized coefficients, these are mostly rather similar.

What is rather striking, however, is the bottom-most graphs for Diyarbakir and Sanliurfa. In especially Diyarbakir, the correlation between invalid share of ballots and the AKP vote share is near-zero in the local elections, but the correlation in the presidential elections was – both compared to its own past local election and other provinces’ elections – observably steeper.

If we group the data by whether the province represents one in which Kurdish parties are quite active, this becomes even clearer. Here I show the correlations both within-voting-station (controlling for alan fixed effects) as well as within-district (controlling for the ilçe fixed effects). There are several tens of thousands of voting stations in Turkey, but only a bit less than one thousand districts, so the latter is likely to be less robust, and is mostly to allow smaller localities in Turkey’s south-east to be included as many of these tend to only have a few ballot boxes per voting station.

Again, for the less Kurdish areas, there is little difference in the correlation between the invalid vote share and the AKP vote share, but for the more Kurdish areas, the regression lines are almost orthogonal.

Clearly, something happened in the Kurdish areas of Turkey between the local elections and the presidential elections. But what?

Ever since the Kurdish political movement branched out to form the HDP and tried to target constituencies outside its traditional core, the degree of political competition in the East has arguably increased. Yet it is unclear at this point how such an increased competition would necessarily increase the correlation between invalid ballots and the AKP’s vote share within voting stations. The alternative is that the above graphs reflect something more sinister:

• In the case of Diyarbakir, for example, AKP stood next to no chance of winning the local elections, and any moderate amounts of vote fraud would not have any real effect on the overall election outcome. Thus, we could be observing no systematic correlation in the local March elections because it wouldn’t make much sense for the AKP to commit any vote fraud then (perhaps because the amount of fraud required would become too costly).
• In the presidential elections of August 2014, every vote counts up to the national vote total, and so even in Diyarbakir, the probability of winning the election increases linearly with every additional vote won. In this case, committing a small or moderate amount of vote fraud could be consistent with the now much larger correlation.

But whereas the documented relationship is consistent with this hypothesis, so far it remains just that.

Over the last year or so a disillusionment over issues ranging from negotiations with the PKK, Kobane and ISIS, as well as the government’s investment in state repression, appears to have turned the Eastern provinces into AKP’s soft political underbelly. If AKP loses out in the upcoming elections and the HDP manages to get above ten percent of the national votes – the minimum required to receive any representation in parliament – this would lead to a discontinuous drop in the amount of parliamentary seats the AKP wins (see here and here). And without a super-majority to implement a new constitution, or even an absolute majority to stay in power, the AKP and Erdoğan may find it increasingly difficult to stay in control.

Therefore, just as the electoral stakes increase, so may the demand for tampering with votes (on both sides).

The real political battlefield for Turkey’s future is likely to play out in the voting stations of Diyarbakir, Van, Mardin, and other cities in the southeastern region, perhaps more so than in any other region of Turkey. It is therefore crucial that this region also receives an adequate amount of election monitoring.