The collapse of pre-election sociology: A view from Russia
This article was originally published in “Meanwhile in the Baltics…“, a collection of articles written by the graduates of 2016 Solidarity Academy – Baltic Sea Youth Dialogue, organised by the European Solidarity Centre in partnership with the Council of the Baltic Sea States.
February 13, 2017 - George Makarenko - Articles and Commentary
In 2016 classic political sociology died. Having celebrated its 80th jubilee in November, over the course of elections in the United States, it suffered from a tremendous stroke, from which it did not recover. At the beginning of December, when Italy held its constitutional referendum, it tried to wake up, get up off the sofa and stand up. But instead, it fell into an even deeper coma, from which it has not awoken.
The birth and death of political sociology (by which I understand the system of pre-electoral public opinion surveys) is strongly connected with presidential elections in the United States. At the beginning of the 20th century (and no later than the 1920) the main organisation responsible for predicting election results was the weekly journal Literary Digest. Before every presidential election, the publication would send questionnaires to millions of its subscribers. The recipients were expected to indicate who they were going to vote for and send the completed form back to the editorial office.
Each time, Literary Digest was able to correctly name the next US president. However, in 1936 the journal sent ten million queries and received 2.3 million responses and, based on the data gathered, Alfred Landon was meant to become the next president.
However, in the election on November 3rd 1936, the Democratic candidate, the then-president Franklin Delano Roosevelt, easily defeated Landon, winning in 46 out of 48 states. But the election had one more winner: a young sociologist, George Gallup. Before the voting began, he got in a public debate with the management of Literary Digest, claiming that the prognoses of the journal were inaccurate, since they were based on views expressed by its subscribers only – who in large part belonged to a higher social strata. Instead of two or three million, Gallup surveyed 50,000 people, having divided them into different groups based on their income level, place of origin, gender, age etc. In the end, his prognosis was the most accurate. Two years later Literary Digest closed down, unable to deal with a prolonged crisis.
The test of time
In nearly 80 years of public opinion research, based on the method developed by Gallup, the company has correctly named almost every winner of presidential races in the United States, with the exception of years 1948 and 1976. Another failure to predict the election result occurred in 2012, when Gallup gave Barrack Obama 48 per cent of votes and his opponent – Mitt Romney – 49 per cent. That year Obama retained the presidency.
The most recent prognosis by Gallup before the 2016 election asserted that support for Donald Trump was at 35 per cent, while for Hillary Clinton it was 40 per cent. But Gallup was not the only one predicting Clinton’s victory, as so did almost all other sociologists. Many social research agencies claimed that Clinton’s victory would be between 95 and 99 per cent certain. FiveThirtyEight website’s estimates were the most generous for Trump and claimed he had a 23 per cent chance to win.
After the election, in December 2016, Swiss Das Magazin published an investigation on how Trump’s victory became possible. “It cannot be claimed that sociologists lost the election, because their prognoses were mistaken. Quite the opposite: sociologists won, but only those ones who used the latest methods”, the publication concludes. Instead of analysing wide groups of population, the newer methods are based on studying Facebook activities of individual users. Subsequently each of the users was recommended a certain context-based advertisement, to mobilise Trump’s electorate or to compel Clinton’s followers to start questioning her actions.
Alexander Nix, the CEO of data and communications agency Cambridge Analytica, who used this system in Trump’s pre-election campaign, challenged Gallup’s classic sociology. Or rather the way Clinton’s team treated the results of the survey: having divided society into groups, the campaigners prepared certain solutions to women, with other appeals they tried to express support for African Americans etc. Trump’s campaign, on the contrary, focused on an individual approach to each voter or to small groups of people – for example the residents of one house.
Shortly before the US election, one more important failure of political sociology had taken place – United Kingdom’s decision to leave the European Union. Before the vote, the majority of polls suggested a three to ten per cent victory for the “Remain” supporters. However, in the end the “Leave” vote won by a four per cent margin.
On December 4th 2016, a constitutional referendum took place in Italy. Public opinion surveys showed that the opponents of the reforms of Prime Minister Matteo Renzi would win by around five per cent. No one predicted that the actual voting gap between the two positions would be close to 20 per cent.
Polling in Russia
From the point of view of survey methodology, the leading Russian sociological centres, no matter how pro-Kremlin they may be, can be compared to Clinton’s pre-election team rather than the internet-savvy Trump’s technologists. Social research surveys in Russia are based on the same old system developed by Gallup, slightly modified to better suit the Russian reality.
For example, VCIOM – Russian Public Opinion Research Centre (the oldest and largest sociological centre, completely owned by the state) is using the “Express” survey. Every week professional sociologists interview around 1,600 Russian citizens in their homes in half of the country’s regions. The regions are chosen on the basis of their representativeness; first, by population size and second, by their similarity to the average social markers (such as economy, urbanisation, share of resource sector and even political loyalties or instability).
During the process of conducting the survey, respondents are divided into groups by gender, age, education level, income and other social categories. The study is conducted using the so-called omnibus poll: the sociologists ask up to 20 different questions in one round (about politics, economy, society), from which they later compile a summary report on a given subject.
One of the most recent Express surveys, released on December 11th 2016, shows that the level of trust for Russia’s president had risen over the past week by 0.6 percentage points and reached 86 per cent. It is worth noting how the question was posed: “Do you, in general, approve of the actions of the president?” Clearly, the apolitical majority of the Russian population can “in general” support the actions of the president and the parliament, but “in particular” every citizen may have their own list of reservations about the authorities or questions to which he or she needs answers.
For a comparison, apart from VCIOM there are two more leading sociological institutes in Russia: FOM (Public Opinion Foundation – private, but actively cooperating with the authorities) and Levada Centre (private and not so open to dialogue with the state, for which it was included in the list of foreign agents at the end of September 2016).
According to FOM, on December 11th 2016 the popularity of the current president Vladimir Putin was on the level of 66 per cent (the question was: “Who would you vote for if the election was taking place on Sunday?”). While according to Levada Centre, Putin is supported by 84 per cent of the population (just like with VCIOM, the question was: “Do you, in general, approve of the actions of the president?”).
As we can see, the difference between “approval” and the “intention to vote” comes to around 20 per cent – and this is according to a rather loyal sociological centre. In fact, coming back to the question of predicting the election outcome, survey data in Russia always shows a lower number of votes for a leading candidate than they actually receive.
For example, in 2012, FOM and VCIOM predicted that Putin would receive around 59 per cent of the vote. The final result, however, was 63.6 per cent for the president. Before the parliamentary election in 2016, the last survey by VCIOM showed 41.1 per cent support for the ruling United Russia, while the final result gave the party 54.2 per cent. Even if we believe the reports saying there were mass falsifications, the results of the elections are far from the sociological prognoses.
The methods of Trump’s advisers can be used as a weapon on social media: if not to influence the election results, at least to affect social moods. However, the amount of Russian users on Facebook is limited – there are only around six million unique visitors a day, according to data from April 2016. V Kontakte social network, a Russian equivalent of Facebook, is much more popular and is visited by more than 70 million users a day (not only Russians, but also citizens of the rest of the former Soviet Union). However, the algorithm of the social network differs from Facebook, and V Kontakte, as a Russian company, has to adhere to certain rules. For example, the site’s personal user data are accessible not only to the company’s owners, but also the governmental agency “Roskomnadzor”.
The hidden stratum
The most interesting sociological surveys in Russia, which attract the most media interest, are thematic polls, as opposed to the “omnibus” ones. Indeed, it is one thing to ask about the relationship towards the head of state and another to share one’s own view on a given divisive issue. The results of such surveys continuously prove to be sensational and do not support the conventional stereotype of the “84 per cent” (the percentage of Russians who, according to data presented by the media, support the activities of the Kremlin since the beginning of the conflict in Ukraine).
For example, a recent survey by Levada Centre suggests that in 2016, 33 per cent of Russians put personal well-being above saving the current state system (in 2015, 27 per cent of respondents gave this answer and in 2014 – 21 per cent).
Moreover, additional research conducted by Levada Centre found that in November 2016, around 71 per cent of Russians supported rapprochement with the West. In March 2015 after the conclusion of the Minsk agreement on the situation in Ukraine, the same was true for 50 per cent of respondents.
In the end, it seems that Russians are rather “lucky” as a society, as they have not been divided into two camps (as Americans on Trump and the British on Brexit). A person who is ready to vote for Putin, can at the same time support normalisation of relations with the West – and the other way around.
Russians in general are not a politicised people. According to FOM’s data, 75 per cent do not plan to take part in any demonstrations (against the authorities or in their support) and 55 per cent will verbally support neither the authorities, nor the opposition. Therefore, public opinion surveys in Russia might get a second chance: at least when they are conducted on a narrow topic. It will not help in accurately predicting election results, but it may help to shape a better picture of where the country is heading. Perhaps.
Translated by Agnieszka Pikulicka-Wilczewska
George Makarenko is a Moscow-based Russian historian and journalist. He currently works as a staff writer at RBC Russian daily newspaper and news agency.