Thursday 12 March 2009

Does 'Not Unhappy' = Happy?

With all the internet chatter among Singaporeans on how train commuters have to constantly enjoy the whiffs emanating from fellow passengers' armpits on a regular basis - so much so that one can recognise and identify a fellow commuter by smell (are you nodding your head? Eww!) - I was surprised to read the headlines '90% of Singaporeans happy with level of graciousness' (carried in TODAY as well) and '1 out of 10 Singaporeans unhappy with current level of graciousness' in today's local news.

source: IMETRO

If the reports are to be believed only a mere 1 in 10 Singaporeans are unhappy with the standard of graciousness and CNA goes as far as making it sound as if the other 9 declared that they were happy. But in actual fact, the survey conducted by don't-know-who which was commissioned by the Singapore Kindness Movement, polled only 502 Singaporeans (nationality by whose definition?) and only half of them actually said they were happy. So, apart from the 50-odd folks who plainly said they were unhappy, there were about 200 who DID NOT SAY whether they were happy or unhappy.

So it is misleading to suggest '90% are happy', isn't it? Once again, we get a lesson in how to have fun with statistics by using a sample size that is 0.0125% of the population and making incredible assumptions to make an erronous sweeping statement.

The reports also indicate that those polled ranked Singapore fifth in a list but stops short of letting us know exactly how many were in the list of choices. For a 500-respondent survey, I wouldn't be surprised if the list only had 5 options to begin with - i.e. Singapore is placed last! Let's see if more information comes along - perhaps the Singapore Kindness Movement will publish the exact/full survey findings on its website (Note: link may be broken/down).

3 comments:

Anonymous said...

“People support CPF cuts because there are no protest outside parliament.” - PM Lee

Anonymous said...

btw, if you had really studied statistics in uni you would know that 502 is a good sample size representative of the population if they were randomly selected. it is a t distribution with 501 degrees of freedom, meaning to say it emulates a normal distribution.

fyi the gallup(spelling?) poll in usa measures only a few thousand people out of the entire us population to predict who wins presidency with amazing accuracy repeatedly except once in the 1930s.

Ganga said...


I did not study statistics at University and have never claimed to have done so.

I guess you've said it all with 'if they were randomly selected' - which is certainly the question that begs to be answered.

In any case, my grouse is not if the sample size is representative of the population but if the results/findings are interpreted with impartiality and presented without prejudice.

Even if the whole population was polled and for a certain question 70% said poor, 20% good and 10% bad, it is still morally wrong to say 'ONLY 10% said bad' and imply that 90% said 'good' when that was not the true circumstances.

That was what happened here and, in fact, often the case when such data is presented in the mainstream media. Is it really so difficult to tell it as it is?