The CEC has released top-level turnout numbers for the recent elections. They have not yet put the full file online, so we cannot see the sub-district breakdowns. Still, a lot of people have been wondering if turnout drove the unexpected election results. There are some interesting numbers here.
Remember, some places always have lower turnout than others. It is harder for people who live in Taipei to return to Hualien to vote than it is to get to Hsinchu. Also, it matters whether people generally expected the race to be close or not. In this table, I’m listing the turnout for each city and county this year. I also put the 2012 presidential election turnout to indicate whether turnout is naturally lower in a particular place. Finally, I give my subjective opinion of how close the race was expected to be.
| 2014 | 2012 | Diff | ||
| Taipei | 70.5 | 76.8 | -6.3 | A little close |
| New Taipei | 61.7 | 75.9 | -14.2 | Not close |
| Taoyuan | 62.7 | 74.6 | -11.9 | Not close |
| Taichung | 71.9 | 75.8 | -3.9 | Close |
| Tainan | 65.9 | 74.2 | -8.3 | Not close |
| Kaohsiung | 66.4 | 75.9 | -9.5 | Not close |
| Hsinchu County | 68.8 | 76.1 | -7.3 | Not close |
| Miaoli | 72.8 | 74.6 | -1.8 | Not close |
| Changhua | 72.9 | 73.5 | -0.6 | Very close |
| Nantou | 73.1 | 71.1 | 2.0 | A little close |
| Yunlin | 74.1 | 68.9 | 5.2 | Close |
| Chiayi County | 74.2 | 72.5 | 1.7 | Not close |
| Pingtung | 73.5 | 72.7 | 0.8 | Not close |
| Yilan | 70.5 | 72.5 | -2.0 | Not close |
| Hualien | 61.8 | 64.6 | -2.8 | Not close |
| Taitung | 67.8 | 61.8 | 6.0 | A little close |
| Penghu | 66.3 | 59.0 | 7.3 | Very close |
| Keelung City | 63.9 | 72.1 | -8.2 | A little close |
| Hsinchu City | 62.9 | 75.7 | -12.8 | not close |
| Chiayi City | 71.0 | 73.5 | -2.5 | A little close |
| Kinmen | 45.2 | 46.7 | -1.5 | ? |
| Lienchiang | 67.1 | 65.8 | 1.3 | ? |
This looks very interesting. If the surprises were a results of lots of blue voters staying at home, turnout should be markedly depressed in those surprising areas. The most surprising places were New Taipei, Taoyuan, and Hsinchu City, and those three all had double digit drops in turnout. The other places with high drops (Kaohsiung, Tainan, Keelung, Hsinchu County, and Taipei) were also places where the KMT did particularly poorly. This is very strong evidence in support of the hypothesis that blue voters stayed at home.
However, there is another group of districts where the DPP also did better than expected that did not follow this pattern. Changhua and Chiayi City saw very small drops from 2012, and Nantou, Yunlin, and Penghu actually went up.
The two groups of districts are quite different, and they might be experiencing different phenomena. The larger, more urbanized, mostly northern places might have seen blue voters stay at home, while the more rural, southern places might have seen intense mobilization of potential green voters or widespread conversion of former blue voters to the green side.
Remember, this is all speculation. Many of the big drops also occurred in places that were not expected to be competitive, and that might be the critical factor. If it was, then presumably equal numbers of blue and green supporters stayed at home and turnout did not decisively affect the results. The numbers look important, but without more evidence we shouldn’t jump to firm conclusions.
December 3, 2014 at 10:40 pm |
How does turnout compare to the last round of local elections in ’09/’10? I would think that that comparison may be more revealing given turnout is generally higher for Presidential elections.
December 3, 2014 at 11:15 pm |
The overall turnout of the presidential election is irrelevant to my point. The purpose is to see which places were most unlike a “normal” turnout. If the turnout in the presidential election had been lower, New Taipei and Yunlin would still have been the most extreme places. Also, the turnout in 2009/10 depended in part on how close and how intense those races were. New Taipei, if you recall, was extremely intense last time so the dropoff might be particularly dramatic this time, and that might be misleading.
December 3, 2014 at 11:19 pm
Also, turnout in the 2009 elections was much lower than that in 2010, so that would also mess up the comparison.
December 5, 2014 at 8:20 am
Then I suggest re-normalizing the 2012 figures to the overall 2014 turnout.
December 5, 2014 at 11:27 am
Add 6.8 to each number. Sorry, I’m so used to the idea that adding a constant is trivial that I often forget that it can make data presentation clearer.