Overseas voters speak: LLL less than KPK in numbers

[steffychun]

http://www.todayonline.com/singapore/overseas-votes-punggol-east-election-counted

The Workers’ Party’s Lee Li Lian, who won the seat with 54.52 per cent of valid votes cast in Singapore, had seven overseas votes. With the seven votes added, her final vote share is 16,045 (54.50 per cent).

The People’s Action Party’s candidate, Dr Koh Poh Koon, had 19 overseas votes. These pushed his original share of 43.71 per cent up slightly to 43.73 per cent, or 12,875 votes.

Mr Kenneth Jeyaretnam of the Reform Party, who took 353 votes (1.2 per cent of the vote share) did not get any overseas votes. Neither did Mr Desmond Lim of the Singapore Democratic Alliance, who had 0.57 per cent of the vote share (168 votes).

Haha at least KJ and DL didnt get any overseas votes.

 

[wwabbit]

As with the local votes, they both lost to the 3rd best candidate.

The difference in PAP's result is statistically significant though, way outside the 95% confidence interval (52.2-88.4% for PAP, x = 19, N = 26). So you'll have to wonder what is it that foreign voters that make them prefer Dr Koh.

My guess is they didn't get to experience the warmth and sincerity of Ms Lee Li Lian.

 

[Dreamer1]

I was expecting PAP to be sacked by we the citizens of Singapore on 2016,now it looks it is impossible,PAP must go now,down with fucking PAP's stupid needs to have 8 million sardin fishes in this little island,their red dot,if they do not go,then we have to go,looks like PAP refuses to give an alternative.

人民行动党走(路)!打倒人民行动党!!!!!!!

PAP karung sekarang, turun dengan PAP!!!!!!!
PAP கீழே இப்போது வேலையிலிருந்து PAP!!!!!!

sack PAP now,down with PAP!!!!!!!

 

[cass888]

Minuscule sample size of 26.

So you'll have to wonder what is it that foreign voters that make them prefer Dr Koh.

 

[wwabbit]

Minuscule sample size of 26.

You obviously don't understand stats at all. I've even calculated the confidence interval for you to see, but then again you probably don't even understand what a confidence interval is.

 

[GoldenDragon]

No point winning an overseas battle but getting thrashed in the real war.

 

[Tualan]

Overseas voters mostly are CSs working for embassy or GLCs , this explains all.

 

[FredTan76]

Most overseas scholars are YP members

 

[Fool_Me_Hard]

That is as accurate as the Straits Times poll with 50 Punggol East Residents

 

[steffychun]

As with the local votes, they both lost to the 3rd best candidate.

The difference in PAP's result is statistically significant though, way outside the 95% confidence interval (52.2-88.4% for PAP, x = 19, N = 26). So you'll have to wonder what is it that foreign voters that make them prefer Dr Koh.

My guess is they didn't get to experience the warmth and sincerity of Ms Lee Li Lian.

Thanks for the econometrics.

 

[wwabbit]

That is as accurate as the Straits Times poll with 50 Punggol East Residents

It's not that simple. If you are not familiar with statistics, you should not brush away the results of polls where N is small. They still yield usable information, you just need to know how to separate the signal from the noise.

The results of the ST poll is actually consistent with the final results of the election. The main problem with the ST poll is not in the small N, but the large number of undecided voters. The most important rule with polling is that you cannot discard the undecided voters. Professional polling companies will always call back the undecided voters to check with them again. The odds of each candidate winning has to take the undecided voters into account. Often it is possible to predict how the undecided voters will vote, by extrapolating from previous polling records of how undecided voters vote, or from tracking polls.

While the ST poll did not make any prediction with any significant amount of certainty, one significant data we can get out of the ST poll is how undecided voters eventually voted. It is very clear that an overwhelming majority (mean = ~81%) of undecided voters from nomination day eventually chose to vote for WP.

Since the 43.7% local support for the PAP is outside of the confidence interval of the overseas support for PAP (52.2-88.4%), we can be very sure that overseas voters are more likely to vote for PAP. It is definitely not a statistical anomaly, this is a real signal.

 

[steffychun]

It's not that simple. If you are not familiar with statistics, you should not brush away the results of polls where N is small. They still yield usable information, you just need to know how to separate the signal from the noise.

The results of the ST poll is actually consistent with the final results of the election. The main problem with the ST poll is not in the small N, but the large number of undecided voters. The most important rule with polling is that you cannot discard the undecided voters. Professional polling companies will always call back the undecided voters to check with them again. The odds of each candidate winning has to take the undecided voters into account. Often it is possible to predict how the undecided voters will vote, by extrapolating from previous polling records of how undecided voters vote, or from tracking polls.

While the ST poll did not make any prediction with any significant amount of certainty, one significant data we can get out of the ST poll is how undecided voters eventually voted. It is very clear that an overwhelming majority (mean = ~81%) of undecided voters from nomination day eventually chose to vote for WP.

Since the 43.7% local support for the PAP is outside of the confidence interval of the overseas support for PAP (52.2-88.4%), we can be very sure that overseas voters are more likely to vote for PAP. It is definitely not a statistical anomaly, this is a real signal.

You are really an econometric person

 

[MightyMouse]

It's not that simple. If you are not familiar with statistics, you should not brush away the results of polls where N is small. They still yield usable information, you just need to know how to separate the signal from the noise.

The results of the ST poll is actually consistent with the final results of the election. The main problem with the ST poll is not in the small N, but the large number of undecided voters. The most important rule with polling is that you cannot discard the undecided voters. Professional polling companies will always call back the undecided voters to check with them again. The odds of each candidate winning has to take the undecided voters into account. Often it is possible to predict how the undecided voters will vote, by extrapolating from previous polling records of how undecided voters vote, or from tracking polls.

While the ST poll did not make any prediction with any significant amount of certainty, one significant data we can get out of the ST poll is how undecided voters eventually voted. It is very clear that an overwhelming majority (mean = ~81%) of undecided voters from nomination day eventually chose to vote for WP.

Since the 43.7% local support for the PAP is outside of the confidence interval of the overseas support for PAP (52.2-88.4%), we can be very sure that overseas voters are more likely to vote for PAP. It is definitely not a statistical anomaly, this is a real signal.

Your assumption is that the sample population shares the same characteristics as the rest of the voters in Singapore. Given that the voters need to travel vast distances to vote, the likelihood is that the personnel in the voting locations such as the Embassy/IE Singapore staff would be those who would be voting. This would alter the voting profile dramatically given the propensity of civil servants to toe the government line. Furthermore, the paucity of actually ballots polled means that identification of opposition voter becomes easier. Leading to the safer option of voting in support of the government's candidate.

 

[steffychun]

Your assumption is that the sample population shares the same characteristics as the rest of the voters in Singapore. Given that the voters need to travel vast distances to vote, the likelihood is that the personnel in the voting locations such as the Embassy/IE Singapore staff would be those who would be voting. This would alter the voting profile dramatically given the propensity of civil servants to toe the government line. Furthermore, the paucity of actually ballots polled means that identification of opposition voter becomes easier. Leading to the safer option of voting in support of the government's candidate.

I dont think it's the embassy personnel voting. Not all are from PE. But yes distances matter. Singapore Embassies and HCs are few and widely scattered.

 

[myfoot123]

Overseas voters are those newly minited citizens who rent out their HDB in Singapore and retire in their homeland. They prefer Singapore to remain status quo.

 

[Thick Face Black Heart]

As with the local votes, they both lost to the 3rd best candidate.

The difference in PAP's result is statistically significant though, way outside the 95% confidence interval (52.2-88.4% for PAP, x = 19, N = 26). So you'll have to wonder what is it that foreign voters that make them prefer Dr Koh.

My guess is they didn't get to experience the warmth and sincerity of Ms Lee Li Lian.

95% confidence interval is 52-88%? I thought should be 56-90%. I use normal distribution.

 

[genetics]

95% confidence interval is 52-88%? I thought should be 56-90%. I use normal distribution.

Small sample cannot use normal distribution, think should use student-t distribution.

 

[Thick Face Black Heart]

Small sample cannot use normal distribution, think should use student-t distribution.

Very little difference in result.

 

[steffychun]

Care to enlighten the layman for those who dont know confidence intervals, statistically significant, x, n ?

Thanks

 

[wwabbit]

Should be using binomial distribution for this