2020 Sri Lankan Parliamentary Election
Did the Greens stay Home?
What the data says
Did the Greens stay Home?
The 2019 Presidential Election had 13.2M valid votes. That number dropped to 11.6M in the 2020 General election.
About 1.6M fewer valid ballots.
Some claimed that this was due to many UNP/SJB supporters staying home without voting. Others argued that they did vote, but many spoiled their ballots. Resulting in the SLPPs near-2/3 landslide.
True? False? What does the data say?
Read on to find out.
Valid Votes vs Green Support
Did the Greens Stay Home?
If so, we would expect a more significant reduction in the number of valid votes in UNP/SJB strongholds compared to SLPP ones.
Let us investigate this claim first.
For each polling division (and postal/displaced votes), we plot the % change in valid votes from 2019 to 2020, against the % of votes the NDF got in 2019. We plot the 2019 NDF wins in green, and the 2019 SLPP wins in red.
E.g. The valid votes in the Chilaw polling division dropped from 93,885 in 2019, to 75,415 in 2020. Or a % change in valid votes of -19.7%. In 2019 the NDF received 40.0% of the vote, to the SLPP’s 54.6%. Hence, a red plot.
We see the opposite of what we would expect if the “Green Stayed Home”. The most considerable reductions in valid votes are in polling divisions that the SLPP won in 2019. Our trend line (noisy, but significant) indicates an increase in valid votes, as the NDF vote in 2019 increased.
Of course, this only indicates that there were (relatively) more valid votes in NDF polling divisions. And fewer in SLPP polling divisions. It says nothing about specific voters. It could be that SLPP supporters in NDF divisions voted more. And UNP/SJP supporters in SLPP divisions voted less.
But nor can we claim that the Greens stayed home. More likely the opposite.
Rejected Votes vs Green Supporters
The other claim was the Greens did go and vote (our data does agree with this). But they spoiled their votes.
If this were the case, we would see more rejected votes in NDF polling divisions. Let us investigate this second possibility.
For each polling division (and postal/displaced votes), we plot the % rejected votes in 2020, against the % of votes the NDF got in 2019. As before, we plot the 2019 NDF wins in green, and the 2019 SLPP wins in red.
Unlike, before we do see a positive trend. The % rejected votes in 2020 increase with NDF support in 2019. Unfortunately, this trend is not statistically significant, as there is too much variation in the % of rejected votes. Particularly among smaller NDF divisions.
This graph is similar to one comparing NDF support in 2019 to rejected votes in 2015.
Hence, there is no evidence to support the ballot pooping theory either.
Conclusions and Take-Aways
On the one hand, caveat emptor. With 24-carota aurum knobs on.
The methods described in this article are not the best ways of answering our questions. A more in-depth and more detailed study, ideally, some combination of exit polling and voter polling in general, would give more significant results.
On the other hand, our results do indicate that some “popular intuitions” (e.g. “The Greens stayed home” or “The Greens spoiled ballots”) might be flawed. Or even contradict the data.
Hence, if you take away one thing from this article:
Don’t blindly accept your intuitions. Or those of your friends and neighbours. Or what you might read on WhatsApp. The truth might prove you wrong.
…
You can find the data used in this analysis at https://github.com/nuuuwan/public_data/blob/master/gen_elec_sl_2020/gen_elec_sl_2020.analysis_unp_home.csv. Originally extracted from http://elections.gov.lk.
