Turnout was lower than many expected. The previous low for turnout in a presidential election was 74.4% (in 2012), but this time turnout dropped to 66.3%. Some drop was expected, since the presidential race was not close and because the election was scheduled so near to the Lunar New Year holiday and university students’ final exams. However, many people have also speculated that a disproportionate number of demoralized blue voters would stay at home. We’ll never have a definitive answer to whether this was the case because we don’t have exit polls. The post-election academic surveys will provide some evidence, but those won’t be release for several months.
In the meantime, we can look for crude patterns in the district-level turnout data. If the hypothesis is correct, we should probably see larger drops in turnout in blue districts than in green districts. I’m going to use the number of valid votes to calculate turnout. (The actual figure includes both valid and invalid votes. These numbers are now available on the CEC website, but I’m going to save time and just use valid votes. Arguably, valid votes are a better indicator, since many disgruntled blue voters may have cast invalid votes.) Overall, turnout in the 73 districts dropped from 74.0% to 65.2%, a drop of 9.2%. However, some dropped more. For example, Hsinchu County dropped 12.4% and New Taipei 9 dropped 12.0%. Some dropped less, such as Chiayi County 1 (6.0%) and Pingtung 2 (6.8%). Not coincidentally, Hsinchu County and New Taipei 9 are very blue, while Chiayi County 1 and Pingtung 2 are very green. Those examples are not misleading, though the pattern is not usually quite so stark. The correlation between the drop in turnout and Tsai’s 2012 vote share in the 73 districts is 0.55, a very strong relationship. However, much of that is driven by Kinmen and Lienchiang, where lower turnout was probably driven by a surge in household registrations by people who don’t actually live there. If we only look at the other 71 districts, the correlation is 0.26, which is still quite impressive for such a crude test. Thus, there is some evidence for the hypothesis that blue voters disproportionately stayed home.
However, let’s not overdo it. There are some people who seem to think that this was the main cause of the KMT’s defeat. There is simply no way that a drop of less than 10% in turnout can make up for a 25% gap in the presidential vote or even a gap of about 13% in the average legislative district race. Remember, many of those people who didn’t vote would have voted for the green side, so you can’t simply add 9.2% to the KMT’s total. There might have been a pattern, but it certainly did not drive the overall result. The KMT’s disaster cannot simply be blamed on poor turnout.
Frozen Garlic 
January 18, 2016 at 9:07 am |
Hi Nathan,
Thanks for the analysis! Putting the question of whether blue turnout decreased this year aside, can we say that OVERALL 2016 turnout (66%) was suppressed compared to previous presidential elections due to the unusual circumstances this year where Tsai was expected to win by around 20% points? In other democracies, non-competitive elections suppress turnout.
Taiwan could be an exception to this phenomena but I doubt it. The best way to test this would probably be to look at local elections in presidential off years, and to see if–all else being held equal–there is a negative correlation between the winner’s margin of victory and the turnout rate in the district. If so, we might interpret this year’s 66% turnout as relatively high, given the timing of the the Chou Tze-yu video release perhaps actually mobilizing a few more middle voters than would otherwise have voted…
January 18, 2016 at 9:28 am |
I’d assumed Tsai surpassing Ma’s 2012 total would discourage blues from assuming they can easily get a majority back, especially when this same wave happened in 2014, but the more comforting short-term reasons Turnout, Tzuyu, and Soong are what I’ve heard from them so far. My own father-in-law was one of the blues who declined to vote; he was a huge Hung fan turned off by the way the party treated her and its other errors. Tzuyu’s effect was just not big enough to be explanatory. As for Soong’s 13%, based on Tsai-Chu head-to-head poll data and polls on Tsai’s and Chu’s favorability among Soong voters, I think around half of his supporters would have gone to Tsai instead of Chu. You can’t just add all those votes to the blue camp if they run a Chinese nationalist like Hau Lung-bin next time.
January 18, 2016 at 11:32 am |
While it’s true that Tsai’s votes total was higher than Ma, it isn’t significantly higher (if memory serves Tsai’s eventual lead over Ma’s 2012 came to something like around 10k?) to support this thesis.
No doubt it’s true that this was an electoral disaster for the KMT, but much of the damage was also self-inflicted and I wonder if in the next cycle better candidates & policies were introduced, how much it would matter.
Frankly, I’m still of the opinion that had the KMT simply “Hung” on (pun intended), they probably would not have lost as much in the legislature. And we probably would have seen much higher vote splitting.
January 22, 2016 at 6:21 am |
The electorate changes with every election. Some people who voted in one year, don’t turn out for the next. Some voters pass away. People who were too young to vote last time, are now able to vote this time, etc.
What fascinates me are the people who change their votes. There’s obviously a segment of people out there who voted for Ma in 2012, and then voted for Tsai in 2016. Do you have any insights or information on the voters like this?
January 22, 2016 at 9:49 am |
Certainly I don’t have the research to back this up but the electorate changes that you speak of, they’re really just nibbling at the edges, it wouldn’t swing the needle too much. As you say, the more interesting and what tipped Tsai over the edge, are the people who changed their votes from 2012 to 2016.
January 22, 2016 at 7:39 pm |
[…] Frozen Garlic, a blogger in Taiwan who has been doing a lot of number crunching during the past week, probed district-level turnout data in “blue” and “green” districts this week, and found some clues there that would support this guess. […]