This is a small project to see whether it is possible to make predictions about the outcomes of Lithuanian parliamentary elections.

I have only worked on a tighter question this time:

Given the results of the first round of parliamentary elections, is it possible to predict the results of the second round and the overall composition of the Seimas?

Here I present my rough working out of a possible approach.

This was a rough presentation of my results which I rushed to publish shortly after polls closed for the second round of the 2020 elections.

The answer to the question appears to be, yes, it is possible to make probabilistic predictions about second round results.

This document gives a slightly more coherent explanation both of the prediction algorithm and some analysis of its results.

In late 2020 I tried working on what could be done to try to estimate first round results and multi-member constituency results, given polling results and historical data. This is hard for a few reasons: polls are few and there are nearly no regional polls; parties change significantly from one election to the next; and constituencies have changed significantly, particularly in the between the past three elections.

In October 2024 a colleague reminded me that Lithuania was having Parliamentary elections (I now follow Lithuanian politics from a distance, out of interest, rather than professionally) and I challenged myself to try to run my model again.

It took several hours of work, but in the early hours of October 27, 2024, I was able to publish my prediction of the outcome of Lithuania’s 2024 Parliamentary Elections.

Less than 24 hours later, we have initial results and I’m putting together a very quick commentary of how well my model did. (Spoiler: not great, probably worse than in 2020.)