Friday, April 29, 2011

When in doubt, turn to the data (Generational Shift Effect)

"Younger voters in their 20s are more likely to vote for opposition parties than for the PAP. Conversely, older voters in their 50s and above are more likely to vote for the PAP."

If you do not agree with the above statement, you may safely ignore the rest of this post. If you do agree with the statement, then read on.

Nomination Day has come and gone and the battle lines have been drawn. The chips are in the air now, and who knows where they will fall?

I asked myself the same question and decided a little data sleuthing was in order.

I am no pundit, and I lack mystical powers that allow me to read the sentiment on the ground. What I have however, is census data, and a belief that younger folks are more likely to want change, while older folks would prefer the status quo.

The logical question to ask then, is in which constituencies do younger folks comprise a larger percentage of the population now, while older folks comprise a smaller percentage of the population?

Demographic data by geographical region is available from Census 2000 and Census 2010. Note that geographical regions in Census 2010 are demarcated by the URA 2008 Masterplan available here. The most distinctive differences between the geographical regions demarcated by Census 2010 and Census 2000 are that Census 2010 includes 3 new regions: Singapore River, Mandai and Punggol. However, only Punggol among the three has a significant population today. Otherwise, I am assuming here that differences in geographical regions between the two census are small.

Those Singapore citizens who were aged 40 and above back in year 2000 are now aged 50 and above. In every single region, they comprise a smaller percentage of that region's population today due to the effects of mortality. In contrast, those Singapore citizens who were aged 10-19 in year 2000 are now aged 20 - 29 and of eligible voting age. Not all geographical regions have experienced an increase in the percentage of population who are of this young age group, due to marriage as well as moving out of the family home. However, in most geographical regions, they now do constitute a larger percentage of the population than in year 2000.

Why did I choose these age cut-off points? Because previous data analysis revealed that new PRs in the last 10 years are overwhelmingly in their 20s and 30s today. I can't separate out the new PRs from the citizens among the younger Singapore residents, but I can safely assume that census data relating to residents aged 50 and above today is unadulterated by the addition of new PRs .

For every geographical region, I calculated 4 numbers:

(1) Percentage of population older than 10 that was aged 10-19 in year 2000
(2) Percentage of voting population aged 20-29 in year 2010

(3) Percentage of population older than 10 that was aged 40 and above in year 2000
(4) Percentage of voting population aged 50 and above in year 2010

I define here the Generational Shift Effect (GSE)

GSE = [(2) - (1)] - [(4) - (3)]

or more simply

GSE = (2) - (1) - (4) + (3)

The GSE is an indicator of how much younger a population in a geographical region has become since the last census in 2000, and by extension, how much more likely a geographical region is likely to swing away from the PAP in this election as compared to the election in 2001.

Below is a table showing the regions with GSEs larger than 10%.

Larger image available here.

The next step is to match the geographical regions in the census data with the electoral map. The most detailed electoral map I could find is available here.

I lack the technical skills to superimpose the 2008 URA Masterplan with the electoral map, so all I did was to eye-ball the two maps and judge which constituency each geographical region with GSE > 10% fell in. These affected constituencies are indicated in the rightmost column in the table above.

Naturally, some geographical regions are small, so they are swamped by the rest of the constituency they fall into. However, a few geographical regions occupy enough of a constituency that I deem it possible that the GSE will manifest itself in these constituencies in this election. These constituencies have been bolded in the table above. Tanjong Pagar is a walkover constituency, so it has not been bolded.

The constituencies in which the GSE is most likely to manifest itself are, IMHO, Radin Mas, Pioneer, Moulmein-Kallang, Sembawang, Sengkang West and Punggol East.

These constituencies won't necessarily be lost by the PAP; it's just that I predict the fight will be closer than most people expect.

As I have previously mentioned, one of the effects of gerrymandering (and walkovers) is that the PAP has distorted the voting signals upon which political parties rely on to judge ground sentiment over the years. If the GSE is as significant as I expect it to be, it may come as a surprise to some people should it exert itself in this election.

On a separate note, interestingly, Tanjong Pagar is a constituency in which the GSE is very strong. However, it is a walkover constituency helmed by Lee Kuan Yew. When Lee Kuan Yew retires from his political career, I fully expect the GSE to exert itself in the following election. Many of his staunchest supporters are growing as old as he is and dying off. The numbers reflect this.


Pris said...

it is a pity i stumbled across this interesting post only after the GE. am curious to know what your data analysis of Aljunied GRC yielded, and if it hinted however subtly at the takeover.

mjuse said...

To Pris:

the short answer is that i didn't do any data analysis for aljunied at all. and i do not intend to at this moment.

i think political commentators have aljunied well-covered. there will be endless ink spilled in the coming days ahead as to why wp won, or why pap lost.

my post identified the GSE as a potential political force. but as i have pointed out, it is likely to exert a significant effect only in certain constituencies.

aljunied is not one of those constituencies. the wp's presence, the strength of their candidates, and widespread dissatisfaction with lim hwee hwa and cynthia phua were likely more important factors in the pap's ouster.