Friday, March 16, 2012

Opinion Polls - measuring shadows cast upon the walls of a cave?

By: Choo Sing Chye

Leaping from a 59% approval rate to a recent pudgy 69% for our Prime Minister Najib Tun Razak is by no mean a small feat, considering the fact that his administration is mired in controversy after controversy.

Unmistakably our Budget is unbalance, corruption remains unabated, our economy seems to be capitulating to the natural gravitational pull, the talk of bankruptcy is in the air and against this backdrop, we got a 69% approval rating?

Hey, what the heck is going on? Are we talking Machiavellian Economics or French Politics here?

Many do not know that the French are a very idealistic lot, particularly when it comes to politics and money. They literally support both sides of the ideological poles – Left and Right. They will tell you that their Hearts are on the LEFT and their Purses/Wallets, on the RIGHT. (1)   So anything goes in French politics.

But anyway, any survey or polling done (do not care whoever does it) always gives me bad vibes. I mean just imagine by trying to ascertain the state of public opinion by questioning a few i.e. 1,022 respondents against the  huge population of 28 million Malaysians? (2)

How can we contemplate the existence of a vast discrepancy in reality and the poll measured? This problem of measuring opinion existed from the time when the notion of democracy began to flourish in Greece.

Apparently, the answer to this vast difference in reality and the poll measured is best described by the ancient Greek philosophers which they likened it to Plato’s myth of measuring shadows cast upon the walls of the cave. (3)

So for an independent opinion research firm (I mean Independent not profit oriented organisations), the main focus of its questions must be based on morality, wisdom, capability and honesty, not money alone. Therefore, it is imperative to frame questions so that the sampled population is made to focused on these values.

Unfortunately, questions asked can be made in favour of anyone if the researchers are not careful. For an example, I could easily get above 70% approval rate, if I ask this question: “Do you think the Prime Minister’s idea of giving RM500 to the poor is good ?” Well, it is difficult to say “no” to this question of giving money to the poor.

Therefore looking at the 69% approval rate for Najib and the reasons given i.e. the budget and the RM 500 relieve for the poor somehow does not reflect the core values that a Prime Minister should have to govern Malaysia effectively.

In fact, the 69% approval rate does not tell us anything about how good the PM is or how bad he is? It just tell us that the sampled population approved the RM 500 “Bantuan Rakyat 1Malaysia”, and perhaps the budget – this certainly does not correlate with the core values of a good PM.

In short, if a mere RM500 (BR1M) could do wonders to an approval rating, then we must be prepared for a long haul if we want see a better Malaysia in the future!!


I've just visited Malaysian Digest's portal and wish to share their poll with you which says:

Should Anwar Ibrahim be investigated over his alleged ties with Jewish organizations?  Yes or No

Is anything wrong with this poll by Malaysian Digest?

Now, look at the word 'investigated' give an impression that a crime had been committed so if I  choose 'No" is like Anwar is above the law and seems arrogant.


1) Peter Harris, Foundation of Political Science, Prentice
     Hall Simon & Schuster (Asia) Pte. Ltd. (1997)

2) For technicality aspects of survey/experiment,please read:
     Experimental Methodology* by: Larry B. Christensen

3) Peter Harris, Foundation of Political Science, Prentice
      Hall Simon & Schuster (Asia) Pte. Ltd. (1997)

*Example of the formula   (Experimental Methodology)for randomly selecting subjects from a population, and it’s usefulness in determining how many subjects should be included in the study for achieving a given level of accuracy.

n = x2NP(l - P) + d2(N1) + x2P(l - P)


n = the required sample size

X2 = the table value of chi-square for 1 degree of freedom at the desired confidence  
level (3.89)

N = the population size

P = the population proportion that it is desired to estimate (assumed to be .50 since
this would provide the maximum sample size)

d = the degree of accuracy expressed as a proportion (.05)

No comments: