Archive for June, 2013

Was it Gerrymandering?

June 12, 2013

Accusations of gerrymandering have been hurled at the EC and the Government in the recent elections. On the face of it, the evidence is compelling. Most of the largest seats were won by the opposition, while most of the smallest seats were won by BN.

I’m not going to dwell too much on the smallest seats, because as we all know most of these seats are in Sabah and Sarawak. The agreement with these states during the formation of Malaysia stipulated at the time that 25% of seats in Parliament were reserved for those 2 states. Seats in Sabah and Sarawak have fewer voters simply due to the fact their combined population is a lot less than 25% of the country as a whole.

So, let’s look instead at the largest seats that recently were mostly won by the opposition. Remember, the accusation is that they were purposely left to be so large to minimize the number of opposition seats in Parliament.

The last redelineation exercise was done in 2003. The chart below summarizes the 1986 – 2004 election results for 15 of the seats that had the most voters in the recent GE 13.

Image

Blue indicates the seat was won by BN and red indicates it was won by the opposition. The ones left blank indicates that the seat did not exist at the time.

What is immediately apparent is that there is a heck of a lot more blue than red. In fact, in the 2004 GE, the first one after the last redelineation, every single one of these seats was won by BN. Furthermore, BN had consistently won most of these seats in the 4 GEs prior to that.

So my question to the conspiracy theorists is simple: How is it gerrymandering when all the seats that are now the largest were, at the time of the last redelineation, solid BN seats?

In Reply to hooleah

June 10, 2013

I recently commented on an article in The Malaysian Insider pointing out that there never was an urban Malay swing for PR in GE 13. You can find the article and my comment (if it hasn’t been removed) here.

In response, another reader using the name hooleah asked me to provide references and called my numbers fictitious. So I wrote a long reply detailing my reasoning and posted it.

For whatever reason, TMI declined to publish it. So I’m posting it verbatim here. I’ll let you decide if there’s anything wrong with it that warrants it to be censored.

For a start, you can go here: http://dahalmi.wordpress.com/2013/05/09/pilihan-raya-umum-ke-13-tahun-2013-analisis-undi-dan-kerusi-mengikut-kaum/
The numbers are a bit different from mine because he combines Selangor with KL and Penang mainland with the island, so there are some non-urban seats mixed in. But the conclusion is the same – even in the urbanised states most of the Malays supported BN.

The author is Malay so you may accuse him of being “UMNO funded” and not believe him. In which case you can just do the math yourself. You can use linear regression and it is not difficult. Here is a site that can do the calculation for you: http://www.alcula.com/calculators/statistics/linear-regression/. Just fill in the data – in this case the independent (x) and dependent (y) variables are:
x – percentage of Chinese registered voters in each seat
y – percentage of votes won by PR in each seat

Below is the data for the 11 KL parliamentary seats starting from P114 Kepong. You can look this up yourself if you don’t believe me.
x,y
90.31,81.65
40.77,57.76
39.76,54.42
51.77,64.61
33.23,48.04
23.7,48.58
74.64,72.70
27.04,50.79
89.69,85.58
84.14,81.22
39.47,56.39

Just copy and paste the numbers into the form and click “SUBMIT DATA”. It’ll give you a regression line equation and a chart that shows how well it matches the data. You can then enter a value for x = 100 to estimate the percentage of the Chinese who voted PR, and x = 0 for the percentage of non-Chinese who voted PR. If you click on “correlation coefficient” it’ll tell you how strong the relationship is between the 2 variables. In this case, it is 0.988 which is very strong and indicates this is a very good model for explaining the results for KL. Do the same for Penang island and you’ll get the same number I quoted.

Telling Lies

June 4, 2013

Since the May 5th General Election, there has been a lot of crap being dished out and called “analysis”. However, with this piece titled Metos Kejayaan UMNO, newly elected MP for Raub, Mohd Ariff Sabri, has taken the art of bullshitting to a whole new level. It is filled with spin, incredible leaps of logic and outright lies. In fact, almost every figure he passes off as fact is a fabrication. Let’s go through some of his gems:

Semua orang tahu UMNO bertanding dalam kerusi yang jumlah pengundi dalam kawasan nya jauh lebih kecil berbanding jumlah pengundi dalam kawasan parti lawan. Jika PAS bertanding dalam kawasan dimana jumlah pengundi secara purata berjumlah 77,000, UMNO bertanding dalam kawasan dimana jumlah pengundi secara purata adalah 36,000.
This is a lie, pure and simple. If you count the number of people who voted in seats contested by UMNO, the average per seat is approximately 51,000. The average for PAS is only 49,000

Demikian juga dengan PKR. PKR bertanding dalam kawasan dimana jumlah pengundi nya secara purata ialah 75,000.
Another lie. The average number of people who voted in areas contested by PKR is also approximately 49,000

Dalam tahun 2008, umno telah mendapat 2 juta undi orang melayu. Ketika itu jumlah pengundi melayu ialah 5.7juta. Lebih ramai orang melayu sejak 2008 sudah pun menolak umno. Erti nya, ketika itu umno hanya memperolehi 35% dari undi orang melayu.
I don’t know whether he’s lying or just being stupid. Did these 5.7 million Malays all live in seats in which UMNO contested? UMNO got 0 votes in, for example Pandan, simply because Pandan was contested by MCA. Does that mean there were no UMNO supporters there? Surely, for an accurate comparison, the total cannot include seats that UMNO did not contest.

Berapa jumlah pengundi Melayu kali ini? Jika 70% dari 11.2 juta pengundi dalam PRU13 ialah orang Melayu, maka jumlah pengundi Melayu ialah 7.8 juta.
Yet another lie. Instead of plucking a random number out of his ass, why couldn’t he simply have referred to the number given by the EC? The percentage of Malays out of the 13.3 million voters registered for GE13 was only 53%. Even if all of them went out to vote, that would only be 7 million. There is no reason to think that Malays were more likely to go out to vote than other races. If the turnout among Malays were the same as everyone else (85%), then the number would fall further to 6 million.

Sekali lagi buatlah perhitungan. Jumlah pengundi Melayu ialah 7.8 juta. UMNO hanya dapat 2.7 juta pengundi Melayu. Ertinya, 5 juta orang Melayu tidak menyokong UMNO dalam PRU13.
Again, only about 6 million Malays voted. UMNO only contested 119 out of 222 seats. So, it was simply impossible for many Malays to vote for UMNO even if they wanted to. MCA, MIC and Gerakan got 899,420, 291,814 and 153,081 votes respectively. How many of these more than 1.3 million voters were Malays who would have voted for UMNO if UMNO had placed a candidate instead?

Proper analysis has been done by others using sound statistical methods instead of this voodoo that Ariff Sabri employs. I suggest you take a look at this post by Sdr DAH Ikhwan who used regression analysis of data from individual seats to estimate how the different races voted. What he found was that support for BN among the Malays in Peninsular Malaysia was about 63%.

Edit 10am, 6/6/2013:

I’ve updated my figures for the average number of voters per seat for each of the parties as the original figures didn’t include independents and spoilt votes. They’ve changed slightly with UMNO up to 51k from 50k, PAS to 49k from 47k and PKR to 49k also from 47k. Details of how these numbers are derived are as follows:

UMNO PKR PAS
Number of seats contested 119 97 73
Number of registered voters in seats 7058181 5591580 4167109
Number of votes cast in seats 6063772 4720281 3589138
Average registered voters per seat 59312 57645 57084
Average votes cast per seat 50956 48663 49166