Do 68% of Russians really support the invasion of Ukraine? | No. 186 – March 6, 2022
How can you expect to reliably survey a police state that doesn't have freedom of expression or an independent media?
On March 4th a Russian blogger named Anatoly Karlin shared the results of a new survey of Russian citizens conducted by “VCIOM,” a pollster that I think calls itself the “Russian Public Opinion Research Center” (and used to be called the “All-Russian” or “All-Union” “Center for the Study of Public Opinion”). I say “I think” because I do not speak Russian and Google Translate only claims to. VCIOM is also sometimes abbreviated VTsIOM, but these appear to be the same pollster.
According to the poll, 68% of Russians approve of the “special military operation” (their words) in Ukraine, while 22% oppose it. Karlin added commentary, saying: “As I said, there is no meaningful distinction between Putin and the Russian people on this question.”
Wow, that is quite the claim! There is no meaningful distinction between Vladimir Putin and the Russian people on support for the invasion of Ukraine? I don’t know much about Russians, but I do know about polls and mass opinion. One flag that the polls might not be measuring opinion correctly is that about ~10,000 Russians have been arrested for protesting Putin’s war so far, meaning tens or hundreds of thousands have been protesting actively under threat of state-sanctioned violence and prosecution.
That suggests to me that there may just in fact be some difference between the public and Putin. I mean, taken literally, Putin is 100% in favor of the “special military operation” and the public per this poll is only 66%. But okay maybe we are supposed to take this seriously and not literally….
But Karlin’s comment does bring up an interesting set of questions. How good is polling under an autocratic regime? Can we trust polls in wartime in said regime? What could cause them to go wrong? There are a few things to consider. None of them reflect all that well on the state of Russian public opinion “polling” today.
Let’s first talk about methodological issues.
It is hard to measure opinions in autocratic regimes
With this poll in particular, two things immediately jump out at me. The first is measurement error. Reading through the results of the poll indeed shows that the surveyors didn’t call the war a “war,” but used the term “special military operation” instead. Apparently this is due to direct guidance by the government that pollsters cannot call the war a “war” or an “invasion” or the like. This is also presumably why TV news is not calling the war a “war.”
Because the way you ask a question changes how people respond to it — something pollsters call “measurement error” — the selective language here could obviously bias responses to the poll. I imagine Russians are likelier to support an “operation” than a “war,” for example. One hint that this might be the case is that another poll that called the conflict a “war operation” picked up just bare majority support for Russian action in Ukraine. And it still wasn’t calling a spade a spade.
The second big error you would make in putting 100% of your faith in this poll is that you’re assuming the people responding to the survey are interchangeable with those who aren’t. In most contexts, this is not so big a deal — but you can easily see why things would be different in a police state during an ongoing war. In the American (and, to some extent, English) context, for example, we are familiar with the way so-called “differential nonresponse bias” can cause polls to under- or over-estimate support for one political party over others. But social scientists have also found that nonresponse among supports of the political increases in less democratic countries. This causes polls to overestimate support for incumbent government.
One paper by Timothy Johnson, Geon Lee, and Young Ik Cho called "Examining the association between cultural environments and survey nonresponse” finds higher rates of survey nonresponse in areas of the US with stricter social cultures and rigid collectivism. While that is obviously not a 100% accurate analog for Russia today — perhaps not even a strong one — the findings are pointing us in a direction that is consistent with our theory.
A related problem is so-called “item nonresponse,” where someone will agree to take a survey but refuse to answer certain questions. One study of polling microdata collected across European countries by Livia Popp finds that item nonresponse to particular questions tends to be highest in “flawed democracies,” per the ratings from Varieties of Democracy (V-DEM) project at the University of Gothenburg. The survey data Popp compiled shows that respondents in less democratic countries are much likelier to say they have no opinion on questions about faith in the government and public parties, whether elections are conducted fairly, whether TV news stations tend to favor the governing party and at what point in the left-right political spectrum the respondent would place themselves. In other words, people were least likely to give their real opinions on questions that could get them in the most trouble with the incumbent government, and that was particularly true in countries with more oppressive governments.
The third major problem is there is some really good evidence that people will lie to pollsters if they feel the government might persecute them for their ”real” opinions. One study of survey responses from 228,000 Africans living across 37 countries tested for systematic bias in more oppressive regimes. The author found that when people believe that the government, rather than an independent research institute, has commissioned the survey, then there is a “significant bias” among responses to questions regarding trust in and approval of the government — as well as perceptions of corruption among government leaders.
This matters a lot in this particular case because the VTsIOM poll is aligned with the Russian government! Karlin calls it a “state-run” polling organization, which actually seems a bit overstated based on other sources, but the firm is definitely not independent. According to one account, when Vladimir Putin got upset at his approval ratings in VTsIOM’s polls, they changed the way they measured it and incidentally issued him better numbers.
All of these factors make polling really hard in Russia right now. Given that the government has now shut down independent media and is actively imprisoning protestors and dissenters, I think we are right to be skeptical of the data — especially the polls being embraced by the Kremlin.
But beyond measurement error and nonresponse, we also need to talk about how people actually form their opinions in autocratic regimes.
Expressing “true” opinions in police states is also hard
Another big problem for measuring the “truth” about peoples’ feelings is that humans are very, very social creatures. We tend to adopt the beliefs of our friends, families and coworkers. That is in part because they socialize us when we haven’t yet formed opinions, but also because we face pressures to be more “agreeable” to people we like or want to impress.
Research by the social scientists Sam Greene and Graeme Robertson have found that contemporary Russians in the modern autocratic state exhibit particularly high levels of “agreeableness.” They say that this is “not because they're conformist, per se, but because there are powerful psychological incentives to get along.” There are some direct political consequences of this; The incentives toward consensus, Greene and Robertson, claim, have led to increased perceived “support” for Putin’s social conservatism and his annexation of Crimea in 2014.
Over time agreeableness can even turn into reinforcement. The adoption of an attitude for the sake of appealing to one’s in-group gets stronger every time the attitude and one’s identity is tested. This further distorts the landscape of what we might consider “unadulterated opinion” in autocratic regimes. Discovering someone’s true preferences, after you have stripped away all their agreeableness and reinforcement, becomes increasingly elusive as time passes. At some point, the truth becomes the alternative and the social becomes one’s truth.
On Twitter last week, Greene summarized the state of public opinion polls in Russia, saying they “are not measuring what people think about the war. Rather, they're measuring (1) how many people are willing to talk openly about the war, and (2) what those people think other people think about the war.
Public opinion is conditional on freedom and a fact-based information ecosystem
Finally, let’s talk about the free exchange of ideas and information. Last week, I wrote to you about an old quote from George Gallup about US public opinion to World War II. “To be intelligent,” Gallup wrote, “public opinion must be based upon facts—the facts which only a free press can give the people.
In Russia today, the press is anything but free. Journalists are not allowed to call the war in Ukraine a “war,” as I’ve noted. But it’s not just that. The government has also embarked on a particularly intense campaign of video propaganda on their state TV channels and on social media. They have preferred to show film of Russian military units advancing in separatist-held areas instead of military strikes on civilians, for example. They have also claimed the conflict was instigated by the West in an intentional assault on Russian security. And on and on.
The biasing of information Russians are receiving from their government obviously alters the formation of public opinion in Russia. Similar to how Americans were brainwashed into supporting interment of citizens with Japanese heritage during World War II, so too are Russian’s being manipulated to follow Vladimir Putin down his path to all-out war.
. . .
So, then, let’s now restate the original question. First, we asked: can polls in authoritarian regimes reliably measure true public opinion? Given the above data and social science literature, I think the answer is a resounding “no.”
But now let’s ask: Do Russians support the war in Ukraine? Yes, or no?
It’s a trick question! The best answer is probably that we just don’t know!
I know that sounds like a cop-out, but the real argument I’m making here is for you to not put faith in most of the polls you see coming out of Russia. I certainly won’t. (And that’s coming from me and I love the polls.) Plus, I think we’ve learned a lot about how polls work along the way.
Posts for subscribers
If you liked this post, please share it — and consider a paid subscription to read additional posts on voter behavior, public opinion polling, and democracy.
Subscribers received one extra posts over the last week:
That’s it for this week. Thanks very much for reading. If you have any feedback, you can reach me at this address (or just respond directly to this email if you’re reading in your inbox). And if you’ve read this far please consider a paid subscription to support the blog.
Another credibility problem is the fact that any two Slavs will have three opinions; any one Slav will have two opinions. I speak from years of observing 4 Slavic grandparents and their fellow bridge players.