A Turkish electoral official shows a “Yes” vote as ballots are counted in the referendum on expanding the powers of the president on April 16, 2017 in Istanbul. (Ozan Kose/Agence-France Presse via Getty Images)

Turkey’s recent referendum, which ended its parliamentary system by transferring all executive powers to the president, has been a source of contention within its borders, as well as outside them. President Recep Tayyip Erdogan’s Justice and Development Party, or AKP, along with his government ministers and MPs, led active “Yes” campaigns in an array of European countries. Those rallies and public meetings targeted not only active AKP supporters, who are the overwhelming majority of Turkish immigrant populations in Europe, but all Turks eligible to vote.

Governments in Europe expressed displeasure with what they viewed as the one-sidedness of the rallies and the tensions they provoked, including violent clashes between supporters and opponents outside the Turkish embassy in Brussels. The Austrian chancellor called for an European Union-wide ban on such political campaigning, while Germany and the Netherlands enforced their own obstructions and bans. The rhetoric that ensued was caustic and the tensions palpable. Erdogan called these actions “not different from the Nazi practices of the past.” German ministers in turn argued that his comments were absurd and German Chancellor Angela Merkel issued a stern rebuke.

Though these rallies and their ban undoubtedly strained relations between Europe and Turkey, did they actually have an effect on the outcome of the referendum?

Why Turkish citizens abroad were important for this referendum

About 2.5 million Turkish migrants or dual nationals living in Europe are eligible to cast a vote in their local Turkish consulate for elections in the homeland. More than 1.1 million participated in the referendum, a critical number given the narrow victory of the “Yes” camp by roughly 1.3 million votes.

Given the influence of this bloc, Mustafa Yeneroğlu, the AKP politician responsible for grass-roots mobilization abroad, highlighted the vitality of the campaign activities in Europe for “increasing turnout rates” among the immigrant communities. Mevlüt Çavuşoğlu, the Minister of Foreign Affairs, upon cancellation of his rally in Hamburg, said the Germans were “blocking ‘yes’ voters and supporting ‘no’ voters.”

How we measured the impact of rallies

German and Dutch interference in the campaign process was not systematically targeted toward Erdogan’s strongholds in those countries, as empirically confirmed by a comparison of local AKP vote shares with and without bans and cancellations. Rather, they were driven by domestic calculations such as the Dutch elections and recent political disputes between Germany and Turkey. We used this to measure the effect of bans, comparing electoral results from areas that had rallies or rally cancellations with those that had neither a rally planned nor one banned.

Turkish citizens abroad participate in the elections at the Turkish consulates closest to their residential area. Each consulate can thus be considered an electoral precinct. We looked at referendum results by precinct (35 total) for all 11 European countries where the number of registered Turkish voters exceeds 5,000, excluding non-E.U. countries and the U.K. We compiled data on rallies from the official Web pages and Facebook pages of AKP and assigned them to these precincts based on proximity. We divided the electoral precincts in areas where at least one rally took place, areas where at least one prospective rally was canceled and areas where no AKP rallies where planned or banned.

We find that despite the rhetoric around the rallies and their ban, they had no effect on voter turnout or on the outcome of the referendum. The results of our statistical analysis, where we examine how the presence or cancellation of a rally affected the “Yes” vote in an electoral precinct compared to the baseline level (vote shares and turnout rates in the 2015 elections), also suggest a null effect. Rallies or bans did not have differential effects in places where support for AKP was higher either.

Here’s what did matter in this referendum

Instead, we find that deep identity cleavages were the clear determinants of voting patterns. Specifically, areas where the Kurdish party (HDP) and/or the main opposition party (CHP) had higher vote shares in the previous elections were significantly more likely to vote “No” in the referendum, as compared to areas with high AKP support, which carried the “Yes” vote. A one percentage point increase in the opposition party or Kurdish vote share led to a 1.20 and 0.73 percentage point respective decrease in the “Yes” vote.

Overall, our results confirm that the deep-seated cleavages between Turks and Kurds, as well as divisions within Turks — be they Kemalists, leftists or Erdogan supporters — ultimately determined the outcome of the referendum. The European rallies and the ban against them in Germany and the Netherlands, as well as the caustic rhetoric that ensued, had no significant effect on the “Yes” vote.

They did, however, further strain the already tense relations between Turkey and Europe at a critical time for regional stability and increased the nationalist discourse and polarization in Turkey. This suggests that in cases where divisions run deep, foreign meddling of this nature may deteriorate relations without effecting the desired change. Turkey hosts about 3 million Syrian refugees and controls the spigot on migrant flows to Europe. Erdogan has so far made no efforts to repair these relationships, despite expectations that he may go on a “charm offensive” after a victory in the referendum. While the rallies and bans did not influence the election as some had hoped — or feared — their true political cost is yet to be fully seen.

Tugba Bozcaga is a PhD candidate in political science at MIT.

Fotini Christia is an associate professor in political science at MIT.