Oct 21, 2019
On Non-Voters
2019 Sri Lankan Presidential Election
Turnout is the proportion of the voters who actually voted, to the voters who were registered (hence, eligible) to vote. The turnout in the 2015 presidential election was 81.52%, and the highest in presidential election history. Of the 15,044,490 voters registered to vote 12,264,377 turned-up and voted on election day.
The lowest turnout was in the 1988 presidential election, wherein the midst of two civil conflicts (one with the LTTE, and one with the JVP), turnout plunged to 55.32%. The 2005 presidential election was also controversial, turnout-wise, with the LTTE ordering voters in the north to boycott the poll.
I define a “non-voter” as a voter who was registered to vote (hence, eligible), but didn’t or couldn’t (as many in 1988 and 2005) vote. If turnout is even across the country, “non-voting” does not affect the final result, as all parties will get the same proportion of additional votes if the non-voters voted. However, this is seldom the case. Usually, there are significant differences in turn-out across the country.
2015
As an aide to analyzing turnout, and non-voting, I first define “deltas” and “non-voter votes.”
Deltas
To win a presidential election, a candidate must win more than 50% of the total valid votes. I define a delta as the difference between the number of votes a candidate wins in a polling division or an electoral district, and 50% of the total valid votes in that polling division or electoral district. Hence, if the candidates total delta is greater than zero, the candidate wins the election.
This is a map of the deltas the NDF and UPFA got across the country in 2015.
For example, in the Colombo Electoral District, the total number of valid votes was 1,296,360. Half (50%) of that is 648,180. The NDF candidate won 725,073. Hence, the NDF delta is 76,893 (~77K).
More intuitively, deltas represent the “lead” a party had in an electoral district. And if a party’s “leads” added-up across the nation is positive, they win the election.
Non-Voter Votes
By “non-voter votes” I mean the votes the voters who would not have voted would have cast.
For example, in the 2015 presidential election, the Colombo-North polling division had 89,771 registered voters. Of these voters, 69,795 or 77.75% voted. 19,976 or 22.25% did not vote. These are the “non-voters” of Colombo-North.
Now, the NDF won 51,537 votes in Colombo-North, while the UPFA won 16,423. If we assume that the non-voters would have voted in the same proportion as the voters, the NDF and the UPFA would have got an additional 14,750 and 4,700 votes respectively.
Important: The assumption that “non-voters would have voted in the same proportion as the voters” is not always valid. For example, often non-voters might be apathetic to most parties and might have voted for a minor party, or even rejected their votes. Also, non-voters in a polling division may be biased towards a single party.
Assuming we accept the assumption as reasonable, we can aggregate polling division non-voter votes at the electoral district level. For example, adding Colombo-North and other Colombo electoral district polling divisions gives a total 171,037 and 124,361 non-voter votes for the NDF and the UPFA respectively.
We can plot the Vote Delta map, for just the non-voter votes. This represents the “additional lead’ a party would have got if the non-voters voted.
Most of the non-voters were in NDF leaning electoral districts, and hence non-voters voting would not have changed the outcome of the election.
2010
In 2010, the UPFA candidate won with 57.88% of the vote, to the NDF candidate’s 40.15%.
We have the same conclusion as 2015 for 2010. Except for the other party.
Non-voters votes would have emphasised the UPFA victory margin, even though there was a significant number of non-voters in NDF leaning electoral districts like Jaffna.
2005
The 2005 presidential election was the closest in Sri Lankan history. The UPFA won 50.29% of the vote, just 180,786 votes more than the UNP’s 48.43%.
Assuming our “non-voter votes” assumption holds, the UNP would have won 143K extra votes in Jaffna alone, possibly securing the election.
Compared to 2015, the UNP delta is significantly lower in many electoral districts, including Colombo. They might have won the election in spite of the LTTE boycott if they did better in some of these districts. See On Floating Votes for more thoughts on this.
1988
The 1988 Presidential Election was held amid two bloody conflicts: One with the LTTE in the north and east. The other with the JVP in many other parts of the country. As a result of President Jayawardena’s Indo-Sri Lanka Accord, Indian peace-keeping forces had also entered the country. Unsurprisingly, turnout was only 55.32%, down from 81.06% in 1982. Voting was not held in LTTE controlled parts of the country.
While voter turnout was very low in LTTE controlled areas, this is not completely visible in the delta statistic because the vote was split between the UNP, SLFP and SLPP in these areas. More turnout would not have significantly added to the UNP or SLFP delta. On the other hand, in Sinhala majority areas where the JVP insurrection was serious, the low voter turnout significantly affected the deltas.
Conclusions for 2019
On the one hand, turnout and non-voting will play a significant part in the November 16th election, especially if the results are close. On the other hand, it is difficult to say what candidates and parties should be doing about this, beyond encouraging as many supporters to vote. In other words, “Campaign 101” stuff.
