In the blogosphere, words have it that there is a wind of change in Ijok — the Chinese are voting against the government. Screenshots is talking a lot about it as well as local media. I have to disagree on their analysis though, unless my calculation is wrong. I did some test statistics and I do not see how one could assert what they assert.
Let us see the pattern at Batang Berjuntai, a Chinese area.
According to Screenshots, the results are:
2004: BN 761; PKR 306
2007: BN 605; PKR 624
We shall do some computation for 2004 first. Assuming there were only two candidates in 2004 as well as ignoring spoilt votes, the standard error for it is roughly 1.01%. For 2007, the standard error is approximately 0.99%.
Therefore, the standard error for voting in Batang Berjuntai based on the two years is about 1.42%. And nope. To find the standard error of the voting pattern, we do not take the average.
Now, in 2004, the difference between the two party is 37.48% out of total votes garnered by both. In 2007, it is 20.57%. That means the difference between the two elections between the two parties is -16.91%.
I suspect the difference is meaningless. So, my null hypothesis is this: the difference is really 0.00%. The alternative hypothesis is that the difference is real, i.e. not zero. Doing a z-test, -16.91% minus 0.00% is -16.91%. Divide that by the standard error of 1.42%. and one will get -0.12 for z. This means my hypothesis is located only 0.12 standard error away from the mean, i.e. strong case for the null hypothesis which translates to no real change. Hence, my skepticism of the allegation of a swing.
In Pekan Ijok however, the alternative hypothesis seems to hold.
The bottom line is that the signal is mixed and inconclusive. The best way to make an authoritative conclusion is to know each and every person’s vote. Or at least, a sample with clear link to ethnicity. Without that comprehensive knowledge, it would be best to stand guard instead of making hasty conclusion. Patting oneself at the back does not achieve anything. Congratulating oneself for “truthiness” is laughable.
p/s — A more sensible explanation by Ong that goes beyond race and religion, in the comment section:
The incessant reference to Chinese voters swinging to opposition is certainly missing the point that the voters in so called chinese areas are really urban voters -eg Ijok toan has only 67% chinese. Similarly the malay and Indian areas are really small holders and estate voters respectively. Seen in this way the hike in agriculture prices are definitely playing a role to win voters to BN/Status Quo. At the same time corruption in government contracts, beraucratism etc are increasingly putting off the urban voters.
People, please go beyond race and religion. It is tiring listening to the same old tune over and over again. Are we so ingrained in communal politics that only race and religion could make sense of everything?
7 replies on “[1198] Of wind of change in Ijok? Stop making hasty conclusion please”
The incessant reference to Chinese voters swinging to opposition is certainly missing the point that the voters in so called chinese areas are really urban voters -eg Ijok toan has only 67% chinese. Similarly the malay and Indian areas are really small holders and estate voters respectively. Seen in this way the hike in agriculture prices are definitely playing a role to win voters to BN/Status Quo. At the same time corruption in government contracts, beraucratism etc are increasingly putting off the urban voters.
Dear Ideas: My system passed off your initial comment as spam. It is a false positive. I apologize.
Dear Peter: Believing is not fact. Any one of us could believe anything regardless of truth. Stephen Colbert calls it truthiness. Besides, your statement does not address the issue. The issue here is, is there an increase in Chinese vote for the opposition? As I have shown for Batang Berjuntai, I cannot reject the hypothesis that there is no increase for that particular area.
Whatever you say I still strongly believe that more than 60% of the Chinese voters voted against BN!
Wow, an example of where my statistics class actually did some good for me in the real world – I don’t feel like a complete dunce after reading your post. :p
I suspect that the main reason BN won in Ijok was that they poured in so much money and effort to winning the constituency (vote-buying and phantom voters aside). They’ll have a tougher time in the GE, especially in urban areas.
friggin hell it my earlier response disappeared into cyberspace…
anyway, it’s a good approach and jeff’s analysis is dodgy la…aiyah…
ok let’s look at this
According to Screenshots, the results are:
2004: BN 761; KeADILAN 306
2007: BN 605; KeADILan 624
so total votes
2004: 761 + 306 = 1,067
2007: 605 + 624 = 1,229
=>increase of turnout per 2004 of about 200 or 20% although we should check/compute against total voters to account for increase/decrease in total pool.
compare BN 2004 vs 2007 –>reduced majority.
but PKR votes more than doubled in 2007.
i suspect phantom votes were used to “top-up” to reduce the majority.
in any case, it is solid to say that PKR has doubled its number of votes, leading to the conclusion that there is a swing to PKR.
but good hypothesis test! that’s the way to do it.
tks for reminding me of this useful and more accurate method which i’ve abandoned after uni to employ more “malaysian” “guess-ology”.
test
good test. i like your blog cos you take an educated approach. jeff ooi’s analysis is dodgy la…aiyoh!
The results to quote jeff are
2004: BN 761; KeADILAN 306
2007: BN 605; KeADILan 624
————————–
but take a look at the total turnout.
in 2004: total = 761+306 = 1,067
in 2007: total = 605+624 = 1,229
ok i know should compare with total voters to compare turnout but in absolute numbers, it has increased about 20% in 2007.
note the reduced number of votes for BN.
and the almost double for PKR.
suspect that voters were brought in to “top-up” the results.
accounting for phantom voters at this ballot box of about 200-300, PKR would have won here and it would indicate a swing of Chinese votes towards BN.
not so easy huh analysing malaysian politics ;D
but def hypothesis testing should be employed :D
tks for reminding me of the scientific methods which i’ve ditched since uni :D