In its poll published right before the election for Moscow Mayor (Russian-language link), the country’s most-respected pollster Levada reported that 52% of Moscow voters had told it they planned to come out on election day and cast their ballot (this number was clearly hedged, however, since more than half of it was comprised by people who said it was only “very likely” that they’d vote, not sure).
Moscow has roughly 7.25 million eligible voters. Thus, Levada was indicating that up to 3.77 million Muscovites might be at the polls on election day. Of those, Levada reported that 18% would vote for Alexei Navalny, or in other words that Navalny would collect about 678,600 votes. By contrast, Levada reported, 58% would support incumbent Sergei Sobyanin, meaning the latter would rake in about 2,186,600 votes. Levada’s data overstated the number of votes Navalny would get on election day (and the same for Sobyanin).
On election day, nauseatingly huge numbers of Muscovites stayed at home, far more than Levada’s data imagined could occur. This reduced the number of votes collected by all candidates, but the incumbent was affected most dramatically.
Navalny actually got only 632,697 votes for mayor, 45,903 (or 7%) less than Levada had predicted. But Sobyanin suffered far more from the weak turnout, actually getting only 1,193,178 votes, roughly a stunning 992,822 (or 45%) less than Levada had predicted. Sobyanin ended up with just over 51% of the total and Navalny with just over 27% because Navalny brought a larger percentage of his supporters to the polls than Sobyanin did. It’s telling that Sobyanin could take this massive hit and still crush Navalny by such a wide margin.
Only 2,286,972 valid ballots were received on election day, roughly 1.5 million less than Levada’s data had shown there might be. In other words, only 31% of Moscow voters actually cast ballots on election day, vastly fewer than Levada had suggested might do so.
So Levada was almost exactly right about the number of votes Navalny would get on election day, but it vastly overestimated the number of votes that would be cast and it failed to recognize that the vast majority of voters who would stay at home would be Sobyanin voters. Levada said Navalny would get a fifth of half of the Moscow electorate to support him, but he ended up with less than third of less than a third of all Moscow voters supporting him. However, Levada never asked the large group of potential voters who said they were very likely to vote but not sure who they planned to vote for. Hence, it’s not possible to claim that Levada was “wrong” about turnout. They clearly said the turnout figure was not guarantied, and they did not attempt to project what would happen if voters were overestimated their probability of going to the polls.
The low turnout for Sobyanin was eminently predictable, however. Given that Levada had predicted that Navalny, Sobyanin’s closest competitor, would lose in a massive forty-point landslide of over 1.5 million votes, many Sobyanin supporters could quite foreseeably have decided there was no need to bother voting. Levada’s polling was actually quite helpful to Navalny, in other words, because it strongly encouraged Sobyanin to be overconfident and, like the famous hare who raced the tortoise, to slack off on his efforts to bring out the vote.
On the other hand, the outcome was still pretty shocking. Given the sensational fact that Navalny had just been convicted of serious criminal charges and sentenced to prison for five years, given that this was the first “Western-style” election in Russia’s history, and given that Navalny had repeatedly said that his own fate and that of the country were riding on it, one might have expected turnout to exceed Levada’s prediction rather than falling vastly short of it.
When the dust had settled, Sobyanin had defeated Navalny by “only” 24 points instead of 40 and by “only” half a million votes rather than a million and a half. His margin was half what it could have been in terms or percentages, a third what it could have been in terms of votes numbers. Had turnout been even lower, say about half what it actually was, Navalny might well have been elected Mayor. If it had been a quarter of what it actually was, Navalny might have won in a landslide. For that matter, had only Navalny’s supporters gone to the polls, Sobyanin wouldn’t have received any votes at all!
For some contrast, in the last mayoral election in New York City the incumbent Michael Bloomberg also won, but defeated his rival by only 50,000 votes and just five points. Sobyanin’s crushing margin of victory over Navalny was five times bigger than Bloomberg’s (ten times bigger if looking at number of votes in the margin rather than points), and Bloomberg was one of New York’s most well-respected mayors ever. Bloomberg got only 50.6% of the votes, a lower share than Sobyanin collected, and Sobyanin raked in more than twice as many votes as Bloomberg received.