Election Predictions

By Julian Dierkes

No, please do not worry, I am not actually going to make any predictions. But here is why and what that means…

Election Law

The Election Law prohibits polling during the campaign. The wording has also been interpreted to forbid any kind of speculation about the outcome of the election.

I am not sure whether that has ever been tested by an academic in court (or whether an academic has been tested by this in court), but there seems to be general agreement that this paragraph prevents speculation about outcomes.

I have been at events in past elections where hosts have told me explicitly to stay away from speculation about results, as they were concerned that this would make them liable as hosts. Obviously, this prohibition should and would apply to foreigners in Mongolia. Again, my legal expertise does not stretch far enough (actually, it barely exists at all), to know whether speculation on this blog or on Twitter, i.e. media hosted abroad, would be subject to this prohibition and how I might be prosecuted, but I do not have any interest in antagonizing anyone just for the purpose of sharing my guesses.

The reason for this prohibition on polls is generally that such polls would influence voters. This is not uncommon elsewhere, especially in a place like Canada with multiple time zones and concerns that if a party is showing especially well in the Maritimes (Atlantic provinces of Canada, ahead of the West Coast by up to 4.5h), voters in the Prairies or in British Columbia might choose not to vote, thinking that the election has already been decided. Or, if a poll says, politician A is winning a seat, their supporters might not bother to vote or supporters of an opposing candidate might show up in greater numbers.

I am not sure that there is evidence for such effects and the opposite effect is imaginable, i.e. if the election appears to be trending against my own views, would that not be all the more reason to go vote?


There are polls in Mongolia, of course. Methodologies continue to develop (see for example the fascinating Worldbank pilot project to survey rural population through random geographic cluster sampling). However, these polls are still hampered by the lack of a general social survey that would allow comparison of a sampled population to the broader population on representativeness and there are some very real practical hurdles to sampling, especially in urban ger districts, and in rural areas generally.

I have always taken Sant Maral Foundation‘s Politbarometer, for example, to be a useful indicator of longterm trends, especially on questions that are asked repeatedly to generate longitudinal trends, but have put little stake in these poll results when it comes to party or candidate vote shares, in part because there appears to be such a strong regional bias in these.

There is always much chatter about the polling conducted by the political parties, but I cannot see how they would overcome some of the practical and methodological hurdles to produce much more reliable results.

Exit Polling

One of the real challenges in understanding Mongolian elections is the absence of exit polling on polling day. I see this as a challenge that breeds mistrust in that there are no expectations of results that have any factual basis before the actual results come out. My conclusion from this post further down will be that I certainly do not feel like I can predict much about the outcome, it is a surprise every time. That makes it very easy for many voices, including loosing candidates who cry foul and allege voter fraud to assert their claims. Other than the absence of evidence for fraud, there is no counter-evidence. Exit polls (done well) are not infallible (as we know from many elections around the world), but they lay the basis for understanding voter decisions and thus enable more expectations in subsequent elections that enable explanations of why a result might have swung in a certain direction.

The Electoral System

I have discussed the unusual features of the multi-member majoritarian system with UBC Political Science colleague, Max Cameron, elsewhere. Since it is not used very much in national-level elections elsewhere, we do not really have firm expectations about its impact.

The two biggest perspectives are that a) MMM reinforces the advantage that incumbents have by privileging prominence, and perception of personality and qualifications, or b) that MMM will lead to many voters making surprising choices with their 2nd and/or 3rd votes. Even if we assume that staunch party supporters (assume that there are more of them among older and more rural voters, mostly but not entirely skewing toward the MPP) will vote party slates in their electoral district, that still leaves a potentially large number of voters who might vote for one or two choices nominated by their preferred party, but will give a second or third voice to an entirely different alternative. It is the latter calculation that must have spurred some of the many independents and small-party candidates to run in this election.

What this suggests is that some of the electorate will vote by party preference, while some voters will based their vote entirely on the candidate not his/her party allegiance.

To the extent that polling or previous history (voters seemingly preferring a back-and-forth between the two dominant parties) suggests anything, these suggestions are thus hampered by the unknown impact of voting in multi-member districts, especially in some of the Ulaanbaatar election districts where so many candidates are competing for seats and thus absolutely small numbers of votes might win a plurality.

My Myopic View of Mongolian Politics

The final element that keeps me from making predictions is my ignorance when it comes to many candidates which is exacerbating the many uncertainties associated with the electoral system even further.

Just for an exercise (and maybe to substantiate any small private bets I might make), I tried to go through the electoral districts and pick certain, likely and possible winners. I could have even gone further and associated probabilities with these gradations and thus come up with an estimate of likely seats. But it soon became clear to me that I really only know a small subgroup of national politicians well enough to have even a faint sense of their electoral chances, and almost no regional/local politicians. Some nationally prominent candidate might thus easily be trumped by a local hero and I would know anything about that as that local hero will have never showed up on my radar.

So, I quickly gave up on this exercise. While I (think I) understand a lot of the dynamics of national politics and while I do know a lot of the politicians active on the national stage (though primarily cabinet members and most visible politicians otherwise, there are some members of parliament that I would neither recognize nor associate any kind of legislative activity with), none of that adds up to enough information to base any predictions on.

Of course, I do have a gut feeling…

About Julian Dierkes

Julian Dierkes is a sociologist by training (PhD Princeton Univ) and a Mongolist by choice and passion since around 2005. He teaches in the Master of Public Policy and Global Affairs at the University of British Columbia in Vancouver, Canada. He toots @jdierkes@sciences.social and tweets @jdierkes
This entry was posted in Elections, Ikh Khural 2020, Party Politics, Politics, Public Opinion and tagged . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *