28 January 2013

Modelling the by-elections: What was the real vote swing against the PAP?

Previously on this blog, I conducted a post mortem of the 2011 general election because Alex Au had made several claims about groundwork and campaigning being essential in determining voting outcome.

I disproved Au's claims through available electoral data and statistical modelling and explained the GE2011 results in several GRC contests to show that the Workers Party had in fact underperformed relative to other opposition parties in contests in wards with highly unpopular ministers, that Aljunied was simply won because of the 6.6% average vote swing against the PAP, and that the average voter in 2011 was simply voting for or against the ruling party.

Although Alex Au has not quite acknowledged our corrections of his thought, his analysis of the Punggol East by-elections attempt to explain the Workers Party win over the People's Action Party, the Reform Party, and the Singapore Democratic Alliance by analysing the swing vote.

His conclusion? There was a 10.83% swing against the PAP since 2011.

A swing to the left, a swing to the right

We argue that Au mistakes the eventual result of the Punggol East by-election to comprise solely of a swing against the PAP. Again assuming the Punggol East voter is simplistically voting for or against the PAP like in 2011, we in addition propose a more sophisticated model of the Punggol by-election vote as an aggregate of two swings.

Aggregate swing = sex scandal + vote against PAP

Given the public uproar in the intervening months against the various sex and sex and corruption scandals, we posit that a sex scandal results in political costs for the incumbent. But how many votes does a sex scandal cost?

We look towards the Hougang by-election, where WP's Yaw Shin Leong resigned from his party and vacated his seat in parliament following a protracted sex scandal.

As it turned out, the WP won Hougang with a marginal change in vote share.

GE May 2011 Hougang SMC: PAP 35.2%%
May 2012 Hougang SMC: PAP 37.92%

For the moment, let's all be naive and simplistic. In an opposition ward of more than 20 years, highly insular and loyal Hougang voters in a by-election one year after a general election are unlikely to change their voting preferences and attitudes towards both the incumbent party and the ruling party. If so, the cost of a sex scandal leading to a by-election is merely 2.7%. This conclusion may be surprising but we note the relative absence of public outrage against the sex scandal in the run-up to the by-election.

Note however the public outrage against the prime minister's initial refusal to call for a by-election, the constitutional challenge that followed, and the positive reaction to Kenneth Jeyaratnam's call for donations online to defray the costs of the case. These factors in addition to the breakdowns in infrastructure since 2011 and the outbreak of the underaged sex scandal would have caused a small swing against the PAP, which would then offset the real cost of a sex which would be at say about 5%, causing an aggregate shift of 2.7%.

Two swings in the same direction

Let's now look at the Punggol East by-election.
Aggregate swing = cost of sex scandal for incumbent + 'national' vote swing against ruling party

GE May 2011 Punggol East SMC: PAP 54.54%
Jan 2013 Punggol East SMC: PAP 43.71%

Here, the incumbent is the ruling party so the final swing of 10.83% is an aggregate of two swings working in the same direction. Taking the projected real cost of a sex scandal as 5%, the actual vote swing against the PAP in Punggol East since 2011 is about 6%. This is our naive estimation.

Now if we take into account local issues that caused unhappiness in the ward - such as unfinished construction projects, inadequate and shoddy infrastructure works that make Sengkang feel more like a slum than Potong Pasir, and the high property prices in the ward - the actual vote swing against the PAP would be far less than 6%.

In other words, we're positing that in Punggol East,
aggregate swing = cost of sex scandal for incumbent + cost of local issues + 'national' vote swing against ruling party

And yes, we're saying that if you did the math before the by-election, you would have put good money to bet on a PAP loss.

Now, we're not saying Alex Au is politically illiterate. He's actually more aware and informed than the average Singaporean but it seems that again, the task has fallen to the priestess of Ise to correct Au's political mis-diagnosis!

1 comment:

Anonymous said...

Brilliant analysis.

Now we realise that he knows how to pay himself the highest but could not make analysis like this to have called and misjudged the by-election.

And by the way, the court eunuchs dare not tell the emperor that he has no clothes - the loss is purely a by-election effect.