If you want to understand why the 2016 election won’t look anything like the the 2012 election (or any other election in the past two decades) but you only have time to look at one indicator, you should look at trends in KMT party identification. It’s easy to get lost in the little details (and I indulge in little digressions all the time), but I always try to remind myself to keep one eye firmly on the big picture.
-
Follow
Following
Already have a WordPress.com account? Log in now.
Frozen Garlic 
December 15, 2015 at 5:02 am |
Very good read again. Thank you Mr.Garlic. I put some of my thoughts down here. Well let’s first claim that all numbers are rough (very rough) estimate by myself and definitely not to peer reviewed journal standard. But hey this is just a comment in blog so please don’t grill me for that.
As a statistician we love more data points and good categorized variable, so I took a look at the pan-blue vs pan-green (Taiwan Indicators Research) graph more closely. I recognized a trend — Pan-Blue steadily lost their support from 2006 and vice versa for Pan-Green til now. (Arbitrary cut off point I admit but we’ll put that in context later). I performed an eye-ball regression (Warning: this can only be done by arm chair statistician and you should never do this at home) and I estimate that Pan-Blue lost 1.4% per year, Pan-Green on the other hand gain beta averagely 1% per year.
But however if I just conclude that Pan-Blue lost 1.4% of vote yearly to Pan-Green and neutral/no answer because they are performing really badly in past 10 years — then I think I’ll just get an “F” in politics 101 term paper. What I’m more interested in is this: Is this just a trend of coincidence because of recent events in past 2 years(sun flower etc) and made the trend “looks like” it’s a long term trend (heavy-headed the regression) but it’s in fact not. Or, is there an underlying fundamental issue which caused this plate shift (in a very slow pace though)?
Well, something definitely shift very slowly for our electorate cohort. New electorate (首投族）and obviously people who deceased. I checked the national census regarding birth/death rate. I concluded that in our study period (2006-2015) we roughly added in 0.3M and lost 0.15M electorates every year (ok this is very very rough estimate obviously there’s more to be accounted for but let’s move on at this time).
So the 0.3M new electorate not only replace the deceased cohort but also create an influx of net 0.15M yearly. And of course we need to time 0.7 for everything here to account for vote rate. And divided by a rough number of recent presidential total votes. You’ll get (drum rolls) — 1.6% shift per year! (well obviously this is just approximate number of course again)
However, if I make conclusion now that the new electorates are the cause of pan-blue demise in slow death form then I’m making some serious logic flaw here. For this conclusion to be considered seriously, if the numbers need to work, then we need to assume all new electorates (or almost of them) vote for pan-green and all deceased electorates voted for pan-blue — which is obviously not correct or even remotely close to fact.
However, some part it might be true, in yesterday’s presidential race survey PFP slaughtered pan-blue in age 20-39 group (3 to 1 roughly) and if you just considered our study group (20-29), it’s an astonishing close to 4 to 1. Well that has to account for something right? But for my argument to make (at least some) sense then they should vote like that consistently in a trend — which I have no data to prove or disapprove it. I think the original data definitely has age group variable and it would be very interesting to see in what percentage (or scale) that this new electorate cohort contribute to the long term pan-blue votes decay. Well obviously you need to have a delicate statistical model to adjust for that.
My gut feeling told me it contributed to certain degree to what extent then some serious study will have to be conducted to have a quantifiable answer. KMT notified this issue since 2 or 3 years ago and tried to respond positively (activities, social media marketing, blah, blah) and negatively (block it so age limit is still 21 not 18). I think the most important thing for KMT to study is if this group’s (and future new electorates) vote is sway-able or it’s like some people called it natural born pro-independence (天然獨）. In the latter they’re probably screwed unless they changed their name to TKMT. In the other case, they shall be able to fight back given time and better performance.
And you know what happened 20 years ago before we introduced this new study cohort? (Or, I should call it the new electorate decade cohort). End of martial law era (1987). Common wisdom told us that this young generation is more open to express their opinion (including political beliefs) and they’re not imprinted with the stigmatized green-camp supporter ink anyhow. Obviously this is just narrative since they have obviously no idea what martial law is after they sat in their history class as “historical context”
Well, I guess at least someone can frame it as a interesting research topic ” Study of Political Affiliation of post martial law generation on recent Taiwan elections”.
I’m surprised KMT didn’t make bigger efforts in terms of fighting for younger generation votes. Wang Bin-Chong? Please, give me a break.(well at least speaker Shu is somehow attractive — but hey! you can’t just appeal to half of the population). Or they are making efforts but not on the right track though. Think about it. For a mere 1% in 10 years it’s 10% and it could tip the balance in every district fundamentally.