Slightly better than chimpanzee...

There was a popular joke that made the rounds in the late 1990's about the different levels of efficiency of the election commissions in various Asian countries. At the time, many countries in the region were experiencing a surge of interest in elections and politics. The joke goes something like this:

A journalist wants to know which country in Asia is fastest in announcing the results of their national elections. First, he went to Indonesia and met with the Election Commission (EC) there. The Indonesian EC said, "we are a big country and our system isn't really good, so we can only announce the official results three months after the election." The journalist next went to India and asked the EC there the same question. The Indian EC said, "we have a huge population but our system is good, we can announce the results three days after the election". Next, to Singapore, and the journalist was told, "we are very efficient, we announce the results in three hours". The journalist finally went to Malaysia and told the Malaysian EC how efficient Singapore was, to which the Malaysian EC chief retorted, "we can do even better, we can announce the results three months before the election!"   

This was of course during the time when elections in Malaysia were more predictable and if someone did really want to fix the election, he could have done so quite easily. Civil society was not strong, election monitoring bodies almost non-existent and opposition parties most often did not have enough counting and polling agents. Elections can still be fixed today, of course, but to do so without anyone noticing it would be extremely difficult.  

Fixed or not, to predict (correctly) election results is not easy. We often rely on experts and pundits for their views but how valid really are their views? Well, according to Professor Philip Tetlock, in his book Expert Political Judgment: How Good Is It? How Can We Know? (2005), an 'expert' is only slightly more accurate than a dart-throwing chimpanzee! A rather sad statement, isn't it? But the evidence is overwhelming.

Here, in Malaysia, we only need to go back to the 2008 General Election to find examples of how inaccurate many expert predictions were. If you browse through all mainstream newspapers the editions just prior to the election, you will find one of the most talked about issues was 'who from Gerakan will replace Koh Tsu Koon as Penang Chief Minister?'; a question totally oblivious of even the possibility of Gerakan losing in Penang. And among pro-Barisan Nasional experts, a view often drummed-up was the 'death' of Parti Keadilan Rakyat after the election and the end of PAS' rule in Kelantan.

What happened then was really quite amusing. I can still recall watching at around 7 pm on election day how excited an RTM presenter was about wanting to know the latest tally in Permatang Pauh. She was so excited because she was expecting a result that signals the end of PKR. PKR, of course not only retained Permatang Pauh but won 30 other parliamentary seats in 2008.      

Local political experts and politicians are very much more cautious this time around. Pro-government experts have lately become more philosophical. Those with statistics and numbers are gaining more prominence.

A few days ago, Professor Redzuan Othman from the University of Malaya presented 'selected findings' from his survey on the election. He reported that 43 percent of his respondents preferred Anwar Ibrahim as Prime Minister, and 37 percent preferred the incumbent Najib Abdul Razak. That is indeed good news for the opposition. Pakatan Rakyat supporters are ecstatic and many are now predicting a resounding victory.

I am not a member of any political party but for this election, I have openly expressed my wish to see PR win. I honestly believe a PR win would be good for the country, and good for Umno and BN too;) But a PR victory, with one week to go, is far from certain. Seven days is a long time in politics. Many things can still happen, many 'videos' can still be produced, and many mistakes can still be made. If everything falls into place, PR may not even need Sabah and Sarawak to win federal. But make just one mistake and PR may end up with only Penang and Kelantan to govern.        

Professor Redzuan's findings would be more meaningful if this was a direct presidential election. But that's not the case here. We have a really unfair system where even if a coalition garners more than 50 percent of the popular votes, it may only get 25 percent seats in parliament. The magic number is 112. Whoever gets at least 112 parliamentary seats will form the next federal government. And at this point, nobody really knows who will that be.

The only way any pundit or expert can validly predict the outcome is to do what Nate Silver did for the US Presidential Election last year. He correctly predicted the outcome in all 50 states. More remarkably, in 48 of those, the actual votes received by each candidate fell within his margin of error.

Nate Silver did not collect any data on his own. He uses data from various polling and media agencies, averages them and run a series of regression analyses to cater for various adjustments. The method isn't really that complicated. But you need a diverse range of present and historical data to get it right, and we just don't have that here in Malaysia.

That, however should not stop us from making our own predictions. Unlike 2008, I haven't been able to observe how things are on the ground but I still would like to offer my own humble slightly-better-than-chimpanzee's prediction. Later...