18 September 2015

Modelling the 2015 general election: numbers, outcomes... and theory?

Singaporeans voted on 11 September 2015 and returned the People's Action Party to power with a supermajority and a 9.9% swing in its favour. This was despite 5 years of poor performance by the PAP, where policy and governance failures erupted in the public eye. We at Illusio conduct a postmortem to uncover, from the numbers, if what went wrong for the opposition can be best explained from the voting model we have championed.

Note: Our numbers are compiled from the Electoral Department's final tally. These are not the same numbers from polling day; overseas vote have been added to all the districts. We are using the final tally numbers for 2015, as the 2011 numbers also include overseas votes. And since the People's Action Party is the dominant party and our model is called the dominant party model, we are interested in votes for the PAP and against the PAP.

The spreadsheet is here for your consideration.

And this graph should again remind you about Dunleavy's dominant party model.

And to recap, here are the key numbers
Average 2015 PAP vote: 69.99%
Average 2011 PAP vote: 60.12%
Average PAP swing: 9.87%

Workers Party numbers, outcome, and theory

Workers Party (all wards contested)
Average PAP vote: 60.3%
Average PAP swing: 7.56%

On the surface, the Workers Party appears to deserve its billing as Singapore's dominant opposition party, with a halo of effectiveness when compared to other opposition parties.

Workers Party (all wards contested, minus Hougang and Aljunied)
Average PAP vote: 64.56%
Average PAP swing: 11.49%

The PAP vote was lower but its swing in the wards where the Workers Party was not the incumbent was larger than the average national swing. And outside its strongholds, WP performed just like any opposition party (note the typical PAP swing), just with a bonus. Taken together, the numbers don't lie: this is proof that the WP markets itself as a dominant opposition party to the electorate, and largely succeeds. Yet its Manifest Destiny strategy was repudiated, with voters in East Coast and Marine Parade GRCs saying loud and clear:

To add insult to injury, the razor thin margin in Aljunied sends a signal that to rational voters, the WP A-team is almost equal to the generic team the PAP fielded. While Kenneth Jeyaretnam boasted fatuously that he had sent a team to Ang Mo Kio that would give Lee Hsien Loong nightmares, it is Lee who has sent a team to Aljunied that will give Low Thia Khiang nightmares for the next 5 years!

We have previously identified WP as a political actor that is not adopting any rational strategy that Dunleavy says viable opposition parties end up taking in a dominant party system. The results demonstrate WP's Manifest Destiny strategy is not viable in Singapore's dominant party system, proving our initial concerns right.

But how do the numbers square with our model?

Election strategy aside, WP does have a coherent policy stand that we can then map onto Dunleavy's model. As many people have described, the WP is "PAP-lite". We have a better name for the WP: the 10% discount party. The PAP aims for a target population of 6.9 million, the WP replies by saying it wants only 5.8 million.

WP cannot say that there has been a policy failure, that KPIs and assumptions should be reexamined, and population and economy models rebuilt, much less how it will rebuilt them. It will however consistently tell you maybe we should aim for 10% less of what the PAP wants, and that's a result they can live with.

I'd look like that too if you say you believe a 10% discount off what PAP wants is a result you can live with, and what you think I can live with.

In Dunleavy's model, such a policy position is described as a "convergent strategy". Yet Dunleavy reminds us: "Getting closer to the policy position of the higher efficacy dominant party is also dangerous for any rival, because if P1 shifts its policy stance towards them then all or part of their support may quickly desert the opposition party." Which incidentally is what has happened in this GE, with Lee reminding voters that his administration is "on it", offering minor tweaks to CPF, housing, transport, immigration... Rightly or wrongly, Lee convinced voters the PAP has moderated its stance. And of course, the rational voter would choose the PAP as the most effective party to implement a 10% discount on its own policies.

Singapore Democratic Party

Average PAP vote: 68.71%
Average PAP swing: 5.47%

Exactly as predicted by the model, the PAP achieved the smallest national swing against the SDP, far smaller than the national average swing. The model explains that the SDP should be more resistant to national swings than other opposition parties because it has adopted the "clear water strategy", putting a clear and coherent ideological distance from the dominant party. The model also predicts correctly that as a niche party, the SDP will never have the best performance at the polls.

National Solidarity Party

Average PAP vote: 72.51%
Average PAP swing: 13.48%

"The collapse of the NSP" is the narrative that everyone on the PAP or opposition side believes in. From a party scandal to multiple last minute defections to a gaffe by a former MP and an incoherent rally speech that went viral, it was the butt of jokes. If voters at the 2015 election did not view it as a viable party, then its numbers become the benchmark for which we must measure the success or failure of the rest of the opposition.

Singapore People's Party

Average PAP vote: 72.92%
Average PAP swing: 14.36%

If the NSP numbers indicate that it is not a viable party, then the SPP has within 4 years become a non-viable party in the eyes of the electorate.

Readers will note that we at Illusio never considered the SPP a major opposition party in our rally season analysis. In an election where the PAP declared town council management and town council master plans to be the most important issues, the SPP dutifully trotted out candidates who would promise their best to be good local governors, releasing their own town council master plans. Yes, Ravi Philemon turned on his attack mode to level 11 on the PAP. Yet it was painfully obvious he was the sole SPP candidate to do so.

The SPP in 2015, in an election that was sweet for the opposition (considering PAP's policy and governance failures), chose to fight an election strategy from 1984, i.e. fielding hardworking backbench MP candidates with no personal ambition beyond representing their constituency and their causes. That is, fielding more 1984 era Chiam See Tong cloneshardworking, sincere, and apolitical compared to the rest of the field.

The SPP results are worth mulling over in detail. According to Chiam's own political modelling, he had presented a slate that would succeed in pushing at least 1 or 2 candidates into parliament. Voters though, rejected his modelling as though they no longer lived in 1984.

An iconic scene from the film Good Bye, Lenin! where a young East German decides to hide the fall of the Berlin Wall from his mother, who was in a coma when it happened. Mother was not fooled. Neither were the Singapore voters.

Assuming there was a real fear of a freak election result, the SPP would have been rewarded for presenting a slate that was least likely to challenge the PAP directly. Assuming there was a real LKY factor, SPP's slate would've reminded voters of the type of opposition they voted to parliament in the good old days of LKY. In the reality of 11 September 2015, voters rejected, with very small variation, the entire SPP slate.

Lina Chiam's commendable showing as NCMP (an indication of a strong policy team behind her) put at least 5% between herself and her compatriots, but was insufficient for a win at Potong Pasir. Dunleavy's model predicts their strategy would be non-viable; the results prove it doubly by also showing that rational voters would prefer a generic candidate like Sitoh from the dominant party over the NCMP with a credible policy team behind her.

Singapore Democratic Alliance
(contested the same GRC in 2011 and 2015)
PAP vote: 72.89%
PAP swing: 8.1%

A failure, if we go by definition that the NSP was a failure this election.

Reform Party
Average PAP vote: 78.61%
Average PAP swing: 10.39%

A failure, given it has the lowest average vote of this election, far lower than the NSP. It's not because they challenged Lee on his home ground; they polled equally badly in West Coast GRC and cleanly split the vote in Radin Mas  with Han Hui Hui, who might as well have been a RP candidate clone . This is a party whose leader came up with the single good idea of the entire election (COEs for foreign workers!), refused to flesh it out and make it credible and convincing enough to capture the imagination of voters nationwide, and instead concentrated on a manifesto with 99 other barely fleshed out proposals. And delivered that manifesto late, 1 week into the campaign and 3 days before the polls.

The Reform Party had one job. One job!

Singaporeans First Party (new party)
Average PAP vote: 78.51%

As we pointed out: In Dunleavy's model, Mr Tan Jee Say positioned himself and his team as a former policymakers and insiders. Of all the parties, SFP took the deeply convergent approach. We at Illusio had predicted from his general strategy and hardhitting and detailed rally speeches a commendable showing. The opposite has happened: SFP polled lower than the NSP clown show.

We cannot say the data has disproved the model. While positioning himself to be a policy insider and running on competence, Tan made 3 mistakes which may have fatally undermined his positioning:

He failed to get ESM Goh and the anchor ministers in both wards to endorse and recognise his capability and competence as a former policymaker and PMO secretary. Tan needed to remind voters of his competence and credentials.

He fielded his A-team in the wrong ward, going up against Chan Chun Sing in Tanjong Pagar instead of Tharman Shanmugaratnam in Jurong. What's the oppositions best and only policymaker doing? If Singapore was in fact run incompetently as Tan and the opposition claimed, then part of the blame would be on Tharman's head (remember the relation between labour, economy, and government?), and it was Tan's job to prove it.

Early gaffes and missteps were played up and reverberated over the electorate, undermining the image of SFP as the party of competent insiders. We refer to the party botching the Tamil translation of the SingFirst motto twice, the failure to turn off spellcheck in its manifesto, and so on.

Unfortunately, Tan Jee Say faced a media that was willing to pounce on every mistake he made and magnify any mistake into a buffoonish blunder

At this point, we can only say SFP failed because it failed to execute the model's prescribed strategy correctly, and not because the model was wrong.

People's Power Party (new party)
Average vote: 76.9%

We thank Mr Goh Meng Seng for gamely setting up a new party on the (not literal) eve of the elections, carrying out a minimal number of rallies, and making zero appearances in the party political broadcasts offered by the state-owned television stations. In political science parlance, Goh has kindly provided for this election a "generic opposition party" with a slate of "generic opposition candidates".

In 2015, a generic opposition party would've polled 76.9%. By definition therefore, a generic opposition party would've been a failure this election. Let us repeat: in 2015, a generic opposition party would be a failure. The 2015 general election was a bloodbath for everyone except the SDP and WP.

The PAP won because it shifted its policy stance in the last few days towards the WP's 10% discount position. Only Dunleavy's model predicts this will result in a complete desertion of the would-be WP swing voters in East Coast and Marine Parade, who correctly and rationally concluded that all things equal, the PAP, not the WP, would be the more effective party to enact a 10% discount on its own policies.

The WP, by positioning itself as the dominant party of the opposition and adopting a convergent strategy towards the dominant party, effectively condemned the rest of the opposition to failure and a bloodbath.

WP misplayed its cards. Keeping the town council issue unresolved and devolving it into a political ping pong match did not endear itself as a plucky underdog to the electorate. The more the town council issue stayed alive during the rally, the more the electorate turned against the WP and worse, against all other opposition parties. "If WP is the best party and can't manage a town council properly or can't manage attacks on its town council management, why should any other opposition party?"

The myth of WP as the dominant opposition party needs to go. Part of it is pure propaganda. Part of it is stage-managed working of crowds at rallies. It is the wrong strategy to win elections. But it is the right and effective strategy for destroying and degrading their "competition" in the opposition camp.

It is not a sustainable strategy. It is a deluded and arrogant strategy that lacks a sense of proportion: whoever heard of a party of 6 seats in a parliament crowning itself the dominant party of the opposition? And then proceeding over the next 4 years to undercut every other opposition party because hey, it has a Manifest Destiny that will make it the government in 50 years but it doesn't really need any allies to win any elections now? And it would rather go into coalition as a junior partner with the PAP than with any other party?

There are lessons from this election for the PAP. Everyone's full of advice, saying the obvious about humility and arrogance, delivering what they promised, etc. We at Illusio too have advice for the WP: it's the same thing about humility and arrogance, delivering what you promise in parliament better, etc. Or you could be the next SPP in GE2020.

For the opposition at large: GE2015 was where you failed to unseat a dominant ruling party, given the best circumstances ever. For your homework leading up to GE2020, you may choose to hone your skills in the meantime by learning how to unseat a dominant opposition party, as a purely intellectual exercise involving voter modelling, party positioning, and campaign competency. Your survival next election, as well as resilience against WP sabotage (intentional or otherwise) is crucial.


Anonymous said...

Fascinating piece, thanks for sharing. Loved the application of Dunleavy's dominant party model to Singapore; learnt something new. I agree entirely on WP and SDP; that was my instinctive view as well; nice to see it being corroborated with some academic rigour.

I wonder how you'd square this (reasonable) strategic analysis with the notion that Singaporean voters essentially seem to have made up their minds a month before the actual elections (that is to say, the support simply hadn't moved during the campaigning phase): https://www.facebook.com/quadsg/photos/a.513712962117785.1073741830.505111996311215/513712968784451/?type=3

Anonymous said...

Sorry, I should have been more specific. The team also found the following(http://www.quad.sg/#findings):
1) That "Opposition supporters might be likely to vote for other opposition parties if their preferred party was absent"
2) That "21% of national respondents indicated that they would vote for who they thought was the best candidate locally regardless of party."

Essentially, how does the model take in to account sentiment on the ground on such seeming intangibles as personal appeal and the fact that no Opposition party competes across the country.

akikonomu said...

Let's look at QUAD's claims, findings, and theories.

They quote the Minimal Effects Model to justify only carrying out their survey once before the campaign. This model was developed to describe PRESIDENTIAL ELECTIONS in the USA. A literature review will show that the minimal effects model has been
1. challenged over the decades,
2. shows less relevance in parliamentary elections and plebiscites, and
3. states that any survey done A YEAR before the US presidential election is sufficiently predictive of the results amd not one month before

What QUAD has not done is conduct a series of surveys a year before the election and on various days before and during the campaigning, and shown that of all the survey results, the earliest match the actual voting results closest.

What QUAD has not done is to furnish a series of preliminary surveys for various GEs in Singapore, and then proven those preliminary surveys closely match all the acting voting results.