Why are surveys and rankings about Singapore always spectacular?
March 18th, 2010 | 
Author: By Hurr Riyahi
Singapore is one place, probably the only place, I have seen whereby the survey or ranking information of practically anything is always spectacular.
There is hardly a time you can find Straits Times publishing any results of surveys or rankings which shows any imperfections in the Singapore system.
If one takes a look at the reports put out by government ministries, statutory boards, government linked agencies etc, you will find a same trend.
This then makes the average Singaporean believe that there are far fewer imperfections in the Singapore system than any other system in the world. But why is survey or ranking information in Singapore always so spectacular?
To illustrate I will use a recent Channel News Asia article which claim that a HDB survey found that elderly Singaporeans have sufficient income for daily expenses. http://www.channelnewsasia.com/stories/singaporelocalnews/view/1043152/1/.html
1) Response Rate
This is the first question to ask about any survey. In the case of the HDB survey, one should ask how many elderly Singaporeans were invited and how many eventually responded.
The minimum response rates that a survey should have vary but for a survey like this, you need to have sent several thousand invitations and have received at least 60%-70% response.
In many Singapore surveys the invitation numbers are low and more importantly the response rate is often less than 50%. Using the replies from less than half the number of people invited to participate in a survey, one can easily find spectacular findings which will then hold no value or meaning.
In some types of surveys, 40%-50% response rate is still acceptable but given Singapore’s small size and other factors, there should be very few surveys for which we can accept 40%-50% response rate.
In the HDB survey, 80% said they have sufficient income for daily needs, 97% are satisfied with their flats and more than 80% are proud of their homes. Such high proportions are likely due to low response rate.
If only 200 people responded out of 1000 invitations, its more likely that you get 160 (80%) people giving positive responses. If instead 600 people had responded, it will be less likely for you to get 480 (80%) positive replies.
This is assuming all the people who responded answered all the questions in both situations. In reality, that is not the case which then actually will push up or pull down the rates extremely when few people respond to a questions.
In fact the high 90% rates mentioned in HDB report suggest that extremely few people answered those questions. Imagine 200 people responded out of 1000 invitations. When only 30 answer a question, its easy to get a 27 Yes (90%). But if 600 people responded and only 100 answered a question, its unlikely to expect a 90 Yes (90%).
2) Pathetic sampling
Singapore surveys rarely meets the standards of sampling needs. Often the sampling is pathetic. In the HBD survey they claim 8 out 10 elderly Singaporeans have sufficient income for daily expenses.
In surveys like this, the people conducting the survey should know beforehand that the lower income are likely to be underrepresented in the survey. That is because they are working long hours that they dont have time for this, their literacy maybe low and they dont understand it, they maybe sick and are unable to spend their effort on it, they are lonely and care less for it etc etc.
Hence if you end up sampling the more affluent elderly Singaporeans you end up with such spectacular results. Combined with the earlier case of low response rate and poor sampling, one will get spectacular findings.
3) Selecting those who can make the results spectacular
Another way to get spectacular findings is through flawed selection criteria. For instance the findings claim elderly Singaporeans having sufficient income for daily expenses. But those selected for the survey is elderly in whose household they or their spouse is the head. This is silly. The survey must include all elderly Singaporeans.
4) Mixing up terms
The HDB survey findings is about income. Yet they cite savings and CPF savings as source of income. An income is amount of money one receives for labour, services, sale of goods or from investment. The interest from savings is income. Savings itself is not income. So by wrongly including these, the income is inflated. What the reality could really be is that majority of elderly Singaporeans are not having sufficient income and are relying on personal and CPF savings.
5) Omitting findings that are not sexy
One of the most hilarious things about Singapore survey findings is that there is hardly much findings reported. Having spent such a large amount of money doing a survey, it is a waste if the survey can only yield that single one line of finding or a few.
Actually what really happens is that the study does yield other findings but are not shared with the public as they show the imperfections in the system. Likewise we only see the spectacular findings of rankings. The other findings are omitted. But it is these findings which really are required to be shared with the public so that the cracks in the system can be rectified.
A young girl who continually choose only to keep looking at her glamour studio photos to judge her beauty even when she reaches retirement age, will never be able to do anything about her ugly wrinkles and will be instead be growing uglier by the day.
Other articles by Hurr Riyahi:
>> PAP’s model: Starting ahead and finishing last
>> What is the PAP’s long term economic vision?
>> Security threat posed by foreigners versus Muslims
>> How often is there tolerance in Singapore or Malaysia
Posted in Opinion
[Apologies to the statisticians. I am trying to use lay language here.]
While I generally agree with most of the comments in this piece, I would like to highlight an error. Specifically this piece makes an argument for a large sample size.
For most inferential statistical methods, getting a large sample size doesn’t do much for you. If you look at the F, t, chi-square, etc. distributions that underlie most statistical tests, the numbers don’t change spectacularly after an N of about 30-50. For most survey type studies IIRC [I'm not looking the figures up, so I could be wrong] with Yes/No questions, an N of about 1300 gets you to a confidence interval of +/- 2 percent. For policy studies where you’re blasting an entire population, it isn’t hard to get 1300 people. Most news surveys around the world, for example, regularly get that sample size.
In fact getting a large sample size is often BAD, because you’ll find statistically meaningful trends (statistical significance), which are not PRACTICALLY meaningful. This is because as sample size increases, a statistical test can detect smaller effects. For example, we might find that [fictitious example] people who work indoors earn $1 more on average than those who work outdoors. We found a difference, but its not one we would really CARE about.
The key issue one has to address is the representativeness of the sample one draws from. The article highlights this issue, but seems to confuse the issue of large samples with representative samples. I could sample everyone in Ang Mo Kio, except one person. While this would be a large sample, it wouldn’t be representative of people in Singapore. It would be better if I could get 50 people drawn from the pool of everyone in Singapore such that there is no difference between these 50 and the rest of the Singapore population.
I can think of a number of reasons for having a large sample size, which aren’t treated well here.
(1) Multiple hypothesis testing- The survey asks many questions. One has to have a larger sample size here, because we have to adjust for the likelihood of spurious findings if we ask too many questions from our data. Traditionally, we do this by being conservative and only detecting the really big effects. In statisticians’ language, we opt for a small statistical significance threshold.
(2) Some statistical techniques such as structural equation modeling require large samples. Even here, the recommendation is to STOP at about N=250.
(3) A test for NO material difference (statistical power). Here, we’re trying to demonstrate that something we postulate has an effect actually has no effect.
Like or Dislike:
0 
1
Cecil
What you have written seems to violate the Law of Large numbers. For your basic statiscs, I don’t believe that there is any statistical paper or theory which postulates that the way to address the issue of spurious findings is to take a smaller sample size.
Like or Dislike:
1 
0
During my younger days, my lecturer once said
Statistics only show one side of the story.
Why take these surveys and rankings with a pinch of salt? Don’t even digest it.
Like or Dislike:
1 
0
From a sampling point, any sampling scheme has the potential to be biased depending on the way the initial seed is selected. Thus while the mathematical model behind the sampling is sound, manipulative individuals can still attempt to skew the results to their likings.
Having said that, I am not entire sure about the sampling approach taken here. Taking a survey of elderly is a very difficult process and my guess is the drop out rate is extremely high. The other problem of a survey if voluntary in nature is the tendency to skew towards people who have very strong opinions. Thus the survey will skew to the good or bad but hardly representative of the neutral grey area.
My comments for Cecil:
Multiple hypothesis testing: I doubt that it is a major concern if proper multiple comparison tests are used. Robust techniques such as Scheffe’s test and Tukey-Kramer test can be adopted. However, if Duncan multiple or SNK test is used, I will be seriously worried as these tests can be quite liberal.
Structural Equation Modeling: Good stuffs but bigger data is still more preferred in this case. Remember that if this is micro array data, we might have some problems.
Mediation effect (point of indifference): Well similar to hypothesis testing. Still have problem of testing accuracy.
Three points that I am interested in:
Are the questions double loaded?
How many questions are being asked?
Are there filter survey questions?
Regards,
M
Like or Dislike:
0 
0
On-line survey is the best . In order to avoid duplicate entries, all must key in their NRIC no . This can be inplemented very easily but it will also mean the results will be very very bad against the ruling party. SO the point is the surveys and interviews are merely for window dressing the papayas only .
NEVER BELIEVE IN SURVEYS AND INTERVIEWS IN SG . ITS CRAP
Like or Dislike:
0 
0
We are No 1 here & No 1 there. But when come to press freedom, we are…? What did you say………….
Did the MSM report this..
Like or Dislike:
0 
0
Although i felt that this article is holds little water and the argument is kinda weak, i still believe that PAP are more then capable and totally likely to had manipulate the surveys and statistics.
Example of manipulation: Phrasing of questions that will inevitably put PAP in a positive light.
Omitting the questions that have negative responses.
And the killer blow: Regarding a surveyee with 1 positive answer out of 100 questions as positive
Like or Dislike:
0 
0
Let him tell us WHY!!! And how this is so AND why to IS his pappy and pappies so perfectly perfect too!!! Since his papa went from PM down to SM and then up to MM to Mentor… The WHOLE Parliament of pappyists which comprises the PM,DPMs, SMs,Ms & MPs of 82 in numbers!!!
Like or Dislike:
0 
0
4) Mixing up terms
“they cite savings and CPF savings as source of income”
I’m drawing ‘income’ from my bank every end of the month….so bank is the source of my income. /laugh @ hdb
Like or Dislike:
0 
0
I wish to put it this way.
Yesterday i conduct a survey asking just two questions.
1) “Are you happy with your current jobs?”
2) “Will you switch your job in 6 months time?”
The feedback is largely negative with 100% of the respondents saying “No” to the first question and 66% say “Yes” to teh second question.
————————–
Now the real facts: My target audience consist of only 3 persons.
“What you said is rubbish”, you might say. And I do agree. The point I wish to say is, there are 3 kinds of lies. Lies, damn lies, and statistics. All statistical figures can/might/will be manipulated to serve an agenda.
Whatever the result says, please take it with a pinch of salt and analyse it with your common sense.
Like or Dislike:
0 
0
Welcome to Singapore and “Hear only the good news” and as for the rest, blanko lah, erased lah, covered up la and worst kenna twisted left, right, centre . . . what else?
Like or Dislike:
0 
0
PAP way of showing why they deserve big salaries and boss us around. (((
Gd analysis of weakness of statistical methodology.
Tr readers should send this article to ST and Today every time they report a ranking. And to the organisation that released study.
Hopefully, they will then become more discriminating.
Like or Dislike:
0 
0
Hear only the good stuff … that’s part of the Singapore Way
Like or Dislike:
0 
0
Hi Aurvandil,
What I think Cecil is trying to put across is that statistical sample does not necessarily needs to extremely big to be representative. What is needed is proper sampling schemes and proper selection with proper questions asked.
Regards,
M
Like or Dislike:
0 
0
Hi Penknife,
What statistics show is suggestive, what it hide is crucial!
I do believe that proper statistics done is beneficial to the general population. Poor statistics has the potential to create disasters for the general population.
Look at Florence Nightingale’s brilliant use of statistics to save lives compared to Lehman Brother’s terrible use of statistics to handle risk.
Statistics can be misused. But trying to hide behind false statistics is asking for trouble when the number does not tally.
Regards,
M
Like or Dislike:
0 
0
Singapore always No 1 lah! Top in everything, including losing money. Even in press freedom we moved from 154 to 133, the fastest rate of growth. So, yes, Singapore is always No 1.
For surveys to reflect the truth, or something close to the truth, honesty is key. You need to display honesty by including a range of questions. For instance, if you ask people whether they want PAP to rule for 10 years or 20 years, all answers would be in PAP’s favor. However, if you include a question on whether people want PAP in power, the findings would be different.
Like or Dislike:
1 
0
M
Didn’t Cecil write this:
In fact getting a large sample size is often BAD, because you’ll find statistically meaningful trends (statistical significance), which are not PRACTICALLY meaningful. This is because as sample size increases, a statistical test can detect smaller effects.
Like or Dislike:
0 
1
Today’s surveys are like yester-years referendums.
For example the national referendum in 1962 for merger with Malaysia.But actually there was no option to vote against the merger among the three options presented to the people by PAP.
Thus PAP ‘won’ with a majority(73%) who were ‘in favor of merger’.
Barisan Socialis were strongly against the idea of referendum,they urge their supporters to submit blank votes in protest of the “rigged” referendum. And 26% of the votes were left blank as a result. This move had been anticipated by the ruling PAP government, as seen by the insertion of a clause that stated that all blank or defaced votes would be counted as a vote for merger with Malaysia.
Ironically, separation is exactly what would happen three years later.
Like or Dislike:
0 
0
Here’s one you won’t see in the local mainstream media:
Per capita govt expenditure on health in int’l dollars
http://www.nationmaster.com/graph/hea_per_cap_gov_exp_on_hea_in_int_dol-capita-government-expenditure-international-dollars
Like or Dislike:
0 
0
And so… Florence Nightingale SAVED MANY LIVES that would otherwise have been either lost of lived in UTTER misery to the very end!!!
LKY’s ways are the piggies Animal Farm ways!!!
Like or Dislike:
0 
0
Take the results of all surveys regarding Singapore with a large dose of salt!!
If those surveys which are not favourable or low ranking, you will never see them in the MSM. Only those surveys or reports which are No.1s will make headlines.
Who will have the time and resources to verify the surveys or reports are true or taken without any prejudice??
The very fact that surveys done on Singapore are always
’spectacular’ is indeed spectacular by itself!!
Like or Dislike:
0 
0
Aiyoh! If you can use “statistics” to lie, you can also use “surveys” to lie as well! Just ensure those “surveyors” whom perform those surveys have an interest in whatsoever!
So simple, even a baby knows how to cry to garner attentions, a little kid knows how to behave liken to a spoil-child to garner for the adults to provide what he long-for or desire-for!
Like or Dislike:
0 
0
I do not know whether pappies’ integration programme for new citizens includes “Hear only the good stuff. The Singapore Way. Chapter 1″. Wayang lecturers at their best.
Like or Dislike:
0 
0
@Aurvandil: I will stand by my statement, and I think statisticians will agree that statistical significance is not the same as practical significance.
I think most (if not all) statisticians will also agree that sample size is not as important a criterion as representative sampling. What you really want with sample size is:
(1) A sample size large enough that the statistical test you run can run on the data- basically that the sample distribution “approaches” the theoretical distribution the test relies on.
(2) A sample size large enough so you can detect the effect size you want. Look up Cohen’s classic work on statistical power on this (Statistical Power Analysis For The Behavioral Sciences). If you’re online, just look up the terms “statistical power,” “effect size,” “practical significance.”
If we have a large, representative sample, we can do more than a small, representative sample. However, there is a cost associated with gathering data. The question a researcher has to ask is, “How much of a sample do I really need given my cost?” The answer is that we often need a surprisingly small sample to get conclusions we can work with.
Also, with a large sample, we are exposed to a different problem. We have to be careful not to conclude too much, because our sample is so “powerful” that we can see very minute things. If we think of these minute things as important, we can end up enacting policy that “overcorrects” or solves a problem that doesn’t need solving.
Regarding the law of large numbers, the incremental benefit you get from the N+1th trial is always less than the benefit you get from the Nth trial. If you look at the statistical tables at the back of any statistics textbook, you’ll see that after N of about 30 or 50, the number is so close to the theoretical curve, that for most situations, you don’t need to worry anymore.
@M: All multiple hypotheses tests work by shrinking alpha (the threshold for significance). They might divide or get an nth root of the number of tests. At the end of the day, they have a smaller alpha.
Also, re: SEM. SEM is a wierd statistical technique in that the tester wants the null hypothesis supported. If you have too large an N, a SEM analysis will not get you what you want, because you’ll FIND a difference. Hair et al. (Multivariate Data Analysis With Readings)- a book many social scientists cite because it dares to give threshold numbers – recommends the N=250 upper limit.
Like or Dislike:
0 
1
Statistics could be manupulated and massaged to appear rosy,
obviously self-interest is at stake.
Like or Dislike:
1 
0
Mr Mah is one of the worst abusers of statistics. His basic mode of operation when questioned about HDB flats is
1) Flash you a number on how much to pay for your HDB flat
2) Refuse to provide relevant data/methodology of the cost of building,way of determining price etc
3) Use strange and unorthdox statistics to convince you that HDB flats are affordable
4) Scold you for refusing to buy his unpopular flats
Goh Meng Seng has written in to HDB to try and determine how they got those strange numbers quoted in the blog posting. HDB has refused to release the questioniare (national secret?!). From their reply, it was however possible to determine that the most likely reason was because of the scale. The following is from HDB
Residents’ satisfaction was measured on a 4-point scale, ranging from “Very Satisfied”, “Satisfied”, “Dissatisfied” to “Very Dissatisfied”. The proportion of households who were satisfied consists of those who indicated either they were “Very Satisfied” or “Satisfied”.
Goh Meng Seng made this observation about the scale:
The 4 point scale is a non-standard method to generate a satisfaction score. The more standard scale is to use a 5 point scale or a 10 point scale. It is unknown why HDB chose to use such an unorthodox scale. A known research problem with such a scale is that it tends to give inflated results. This is because respondents who are indifferent (neither satisfied nor dissatisfied) are not given a choice. When forced the tendency of such indifferent respondents is to give a 3 rating, inflating the results.
For those interested to learn more, you can check out the following links
http://singaporealternatives.blogspot.com/2010/02/call-for-clarity-from-mah-bow-tan-and.html
http://singaporealternatives.blogspot.com/2010/02/no-clarity-yet-but-more-questions-on.html
Like or Dislike:
0 
0
Hi Aurvandil,
This statement from Cecil is not necessarily wrong. Given any data of sufficiently large sample, any event is more likely to occur. Depending on the type of statistical tests, they almost always have sample size in the calculation formula.
What happens in some scenario is that the sample size can affect the formula causing the false positive effect to appear unnaturally. This scenario can occur in some situation where bootstrapping is not done properly. Given in this particular context, his statement is not necessarily wrong.
The other problem is the repeated measure problem. We are not sure whether the respondents are unique individuals or they have been surveyed multiple times. This can be another major issue.
Bigger data is preferred only if they are good data. Good data referring to accurate data and not data that are randomly filled in without concern for accuracy. Big data which contain too much noise (errors) can also result in erroneous situation. If there are white noise, it can be handled. Otherwise the analysis is almost useless given that artificial noise are added to the mix.
Law of big number refers to the convergence to the truth given that the data is true. If the data is false, it will converge to the false result.
By the way in the Journal of American Statistical Association, there is a law/conjecture/theory called the Law of Truly Large Number by Diaconis, Mosteller(JASA, 1989) in the paper on the method of studying conincidences. I think this is what is referred to by Cecil. A rare and unusual paper about the effects of too much data. I do not think this is the mainstream statistical thoughts but still it has a place in statistics.
Regards,
M
Like or Dislike:
0 
1
Minister Raymond Lim also tried to do “magic” with statistics. At the recent budget debate, he presented the findings of the latest survey on satsifaction with public transport. Like Mr Mah, all percentages in the high 90s.
If you look at the survey more closely, you find that a 10 point scale was used. Anyone who gave a 6 to 10 score was classified as being “satisfied”. This therefore deploys the same trick of grouping indifferent respondents into the satisfied classification.
Even with this “kelong”, he was unable to hide the fact that many Singaporeaans are unhappy with the crowdness in our public transport. Almost 50% of the respondents have a 1 to 5 score. He therefore deplloyed 5) from the blog posting and omitted mentioning it altogether, burying it as a table in the appendex.
I made a comment on this in Goh Meng Seng’s blog and you can read about it here.
http://singaporealternatives.blogspot.com/2010/03/transpot-minister-vs-hdb-minister.html
Like or Dislike:
0 
0
M
If what you write is true, then we should quickly stop the Population Census 2010. They are wasting a huge amount of money ($12 mil) to sample 200,000 plus households.
Like or Dislike:
1 
0
Another point to add on selective surveying:
All they have to do is interview all the RC, CC and PA members living in the area (who count as average Singaporeans too) and there you go, spectacular results!
Like or Dislike:
1 
0
Hi Cecil,
All multiple comparison tests operate on the same basis by shrinking Alpha in relation to the number of hypothesis tested as well as the population size. I have to admit that I never liked this approach and have resigned from a job because the boss insist on using this approach which is not appropriate in analyzing data.
However, in the context of the survey, such liberal use of multiple comparison tests might be objectionable and thus meriting an additional check on their approach.
As for SEM, I never really believe that 250 is enough even though the textbook suggest that. The rule of thumb for normality assumption is 30 observations. But, in practical work, that is rarely sufficient for most cases. Most of the real life application cases for SEM I have seen (a number of projects) use an upward of 500 to 1000 depending on the number of links.
I am just highlighting some aspects of statistics in the hope that people will not be blatantly abusing it.
Regards,
M
Regards,
M
Like or Dislike:
0 
0
thanks to statistics experts for clarifying though i prefer less jargon.
in computer terms,rubbish in,rubbish out. if survey questions and sample are not properly structured, of course the survey results will be slanted.
what’s the purpose of this survey? to convince us or to convince PAP govt themselves that they are doing a good job.
if the former, i am not worried as we have stats experts to debunk their claims but if the latter, then they are just deluding themselves, thus not correcting their mistakes.
Like or Dislike:
0 
0
Statistics say what the statisticians want it to say.
The question would be who pays the statisticians.
This isnt a Singaporean problem, its the same everywhere.
It is funny how a man in his pigeon hole, paying more in interest than principal on his mortgage, watching his savings and earnings diminish from accelerating inflation can knowingly and willingly delude himself that his government is doing a good job.
Stockholm syndrome indeed.
Like or Dislike:
0 
0
Let’s have simple hypothetical survey in Singapore for statistical purpose.
Question: Do you believe in God?
It doesn’t matter if the sampling size is small or large since the assumption is that the demographic is representative.Only about 14 to 15% of Singaporeans have no religion which makes the vast majority of 85% who believes in some sort of god.
So,the vast majority would vote that they do.But is that a reality that there IS indeed a God or scientifically proven to be.?..
But on the other hand what such statistics does is it tells those sitting-on-the-wall that the probability of the non existence of god is slim statically .
Now,let’s rephrase the same question.
Question:-How does your God look like?
I am sure majority of Singaporeans cannot answer that .And since majority cannot answer ,can it be than concluded statically that since no body really knows how his or her god looks like and therefore there is no God?…which logically makes sense.
Sorry to all believers.I am merely pointing to the fact that statistics are often used as evidence to support a claim by circular-logic,straw argument,false analogies over-generalization,misuse or abuse of statistics in argumentation,illogical reasoning,conclusion not necessarily that follows from the premise and etc etc….jumping to unjustified conclusion.
Like or Dislike:
0 
0
I wrote on Thu, 18th Mar 2010 2:53 pm ,
Typo—Please read the word “statically” AS statistically .
Sorry.
Like or Dislike:
0 
0
@True Singaporean
Majority are not always right.
When Galileo point out Earth revolve round the sun against the bible teaching that Earth is the centre of Universe, he was threaten with heresy by the church.
Sorry if Truth really hurts.
Like or Dislike:
0 
0
Well,the ironies is that instead of finding the truths,USERS of the statistical surveys sometimes knowingly collect ‘SAMPLES’ that ultimately WOULD represent their intended MOTIVES.
If you collect data using just a sampling at ORCHARD ROAD or SHENTON WAY,you would likely get some kind of results that would not neccesarily be representative of all singaporeans.
Alos,the way the data are being interpreted may be skewed towards certain biases.
For instance,govt agencies claimed ther eis increase in employment but we know alot of us are still out of jobs!
Like or Dislike:
0 
0
@MeeSiamWhereGotHumStupid on Thu, 18th Mar 2010 5:46 pm ,
Perhaps you are referring to the Tyranny of the majority .It is used in systems of democracy where majority rule. Is a criticism of the scenario in which decisions made by a majority under that system would place that majority’s interests so far above a dissenting individual’s interest that the individual would be actively oppressed, just like the oppression by tyrants and despots.
But ,Tyranny of the Minority also can exist or co-exist.Where a minority can exercise tyranny too.Like where elected officials follows the wishes of his/her boss only, even though most voters may disagree.
In the context of Singapore ,pseudo -statistics are flaunted as the wishes of the majority .By which the minority(PAP) exercise tyranny over the majority of voters.
Like or Dislike:
0 
0
lies, damned lies and statistics!
Brilliant and critical article.
Like or Dislike:
0 
0
the latest surprising survey has gotto be the one on public transport! what a joke!!
Like or Dislike:
0 
0
There are two issues here
1. Old people are inaccessible, so only the fit, able and eager to participate will provide the statistical results. The SAMPLE IS LIKELY TO BE BIAS. Worse still, those that were omitted of participation for any reason/s or declined to participate are the true representative sample. If the participation rate is 30% and the positive response is 8 out of 10, the actual positive correspond to 24% AS COMPARED TO 70% who declined participation and who in fact hold negative response. The result is negative response is 76% BUT THE SURVEY RESULTS WILL BE PUBLISHED AS 80% OF THE SUBJECTS AGREED THEY HAVE SUFFICIENT INCOME. IT IS FALSE CONCLUSION!!!!
2. Words like sufficiency of income has different meaning to sufficiency. Most probably they have no income and as long as the children put food on the table. It is “sufficient” but when it comes to medical, they won’t see a doctor because of lack of affordability but not interpreted by the retirees as “income”. In other words, the wording of the question “suficiency” may be constructed to get the EXPECTED POSITIVE OUTCOME. The survey worker might even “interpreted” the word “sufficiency” (which mean different things to different people) to the respondent to elicit the “correct” preferred survey results.
With old people survey, it will be low response rate and the biggest fault is that the sample declined or unable of participation makes the survey results so obtained unreliable of interpretation.
I any case, who wants to admit their family is a failure?
Like or Dislike:
0 
0
Also income is not evenly distribution BUT I THINK IT IS BADLY SKEWED WITH VERY FEW HIGH INCOME EARNER IN REALITY but if the survey result suggest it is “normal” that everyone has sufficient income, I am deeply suspicious. THERE IS NOT HEAVEN DETERMINED RULE THAT INCOME IS EVENLY DISTRIBUTED AMONG THE POPULATION.
Like or Dislike:
0 
0
half of Singapore are strange in that they will accept any findings of any survey that the Straits Times or Government publishes but question any negative finding of any survey that anyone external to this group conducts.
singaporeans will never accept that their country is imperfect like any other country. they will rather crash the system than repair whats broken.
Like or Dislike:
0 
1
The silent majority is not the stupid majority.
Like or Dislike:
0 
0
Another big mix up of terms:
‘Singaporeans’, ‘locals’ do not translate to ‘Singapore Citizens’
In PAP’s definition,
‘locals’ = ‘Singaporeans’ = ‘Singapore Permanent Residents’ + ‘Singapore Citizens’
But legally and in reality,
‘Singapore Permanent Residents’ = ‘FOREIGNERS’
Like or Dislike:
0 
0
Nomenclature on Fri, 19th Mar 2010 6:27 pm,
Aiyoh, always the same nonsense.
Hello, PR is how many percent of “locals”? 15%? Of 15%, how many live here more than 5 years? 10-12%?
5 years ago you ever complain about PR? No, right? So you only complain the new 3-5%, right? Be honest leh. You also target the old PRs? How many of them married to Singaporeans? How many bring whole family stay here?
3-5% can distort your statistics by a lot ah? Really? Distort until totally out, as in Singaporean is totally the loser in every survey and the 3-5% can cover all up?
Wah, very convincing argument siah. Faintz.
Like or Dislike:
0 
0
I never believe in statistics given out by government, GLCs..
I urge you all to do the same.
You will know for yourself whether life lately has been tough or not. No need to listen to PAPaya’s bullshit!
Like or Dislike:
1 
0
NOW I OsO KNOW… WHY hairy sent his son to become a Mathematical Statistician!!!…
To Debunk by way of Statistics all of his hairy apa’s fumbling and Goober di Goo stumblings… and then his own when he became PM too!!!… WOW… SURE runs in the FAMelee!!!
Like or Dislike:
0 
0
ummm, seems to me that Singapore is the opposite of Germany– where rankings are always rock bottom!!! LOL
Like or Dislike:
0 
1