|
Russian Federation112 Posts
On April 27 2012 08:08 chuky500 wrote:
The 3 point error was for regular surveys. The 20-18 mistake was based on official results that were already known in most places. That's why it's crazy to have such a big error. That's why I say they announced false results on purpose to make the results look more obvious.
The margin of error means the "amount of things you don't know". 3% of the 36 million counted votes represent 1.1 million votes. If you announce a result within a 3 point margin it means for each candidate you have a window of 2.2 million votes you don't know. 0.5 represents 180 000 votes. When you announce 0.5% with a 3 point precision it means you're confident enough to say there'll be exactly 180 000 votes within a 2.2 million range. And then what stops you from announcing 18 votes within that range ? The point is you don't know enough to tell what you think you know. If you don't know 2.2 millions don't tell 180 000, tell the safest figure, the one in between, tell 1.1 million with a 1.1 million precision.
But well the 3 point precision doesn't actually represent anything since the method used isn't scientific at all (quotas method). It's just a figure they claim to look serious. 2 weeks ago journalists said the raw figures of surveys on the radio for Le Pen, they said only 2% people actually said they would vote for her. This means on the 1000 people asked, that's only 20 people that confessed to vote for her. Out of these 20 people they managed to make highly scientific calculations and say she'd have about 16%. Believe the 3 point precision if you want, I don't.
Please don't flood on this thread for that. Create one called "art of statistics" if you enjoy it 
|
On April 27 2012 08:16 kwizach wrote:
I don't really know what to add to what I already said - your comments show that you don't seem to understand how surveys work and certainly don't understand what the margin of error is.
The way you confused % with points, and used the precision on surveys for actual results was blatant.
|
On April 27 2012 05:53 chuky500 wrote:
1st : surveys don't say 3% precision they say 3 point precision.
2nd : A difference between announcing a result of 20% when the actual result is 18% is a 10% mistake not 2%. If you don't believe me calculate it with the raw figures. And the precision wasn't even told here since it was intermediate results.
I know what you mean and I do not need to calculate it. Difference between 20% and 18% is 2 percentage points. Error is 0.1 (10%) but they are not talking about error when they say 2(or 3) percentage points. They are talking about margin for error and if the reality is 18% you are within 3% margin of error if you predicted 20%. Read my post again , you are mixing percentages as values and percentages as errors.
|
On April 27 2012 08:32 chuky500 wrote:Show nested quote +On April 27 2012 08:16 kwizach wrote:
I don't really know what to add to what I already said - your comments show that you don't seem to understand how surveys work and certainly don't understand what the margin of error is. The way you confused % with points, and used the precision on surveys for actual results was blatant.
I didn't confuse anything. I made it very simple for you to understand what your error was, and you still don't get it, as your comment on what they should have said for Cheminade clearly shows.
|
I need to go to bed so this is the last time I answer. Read my post again if you didn't understand : the 3 point margin is for surveys. Surveys is when they phone people and ask them questions about what they eat, how much they get paid and among other things who they vote for. This 3 point margin don't apply to the results they announce on TV at 8 o' clock because results are based on actual, definitive votes. Counting votes started at 6 o' clock almost everywhere and are finished by the time these results are announced. And for the first time this year, french people in foreign countries or in isles had already voted while they used to vote after french people from the metropolitan area.
So on one hand you have more definitive votes counted by the time results are announced, compared to previous years, and on the other hand, the results announced are more wrong than in past years. 20% compared to 18% wouldn't mean much for a regular survey but for counted votes it's something that deserves to be noted.
Now good night.
|
On April 27 2012 09:02 chuky500 wrote:
I need to go to bed so this is the last time I answer. Read my post again if you didn't understand : the 3 point margin is for surveys. Surveys is when they phone people and ask them questions about what they eat, how much they get paid and among other things who they vote for. This 3 point margin don't apply to the results they announce on TV at 8 o' clock because results are based on actual, definitive votes. Counting votes started at 6 o' clock almost everywhere and are finished by the time these results are announced. And for the first time this year, french people in foreign countries or in isles had already voted while they used to vote after french people from the metropolitan area.
So on one hand you have more definitive votes counted by the time results are announced, compared to previous years, and on the other hand, the results announced are more wrong than in past years. 20% compared to 18% wouldn't mean much for a regular survey but for counted votes it's something that deserves to be noted.
Now good night.
First of all, you mentioned surveys in your original post, and got their margin of error completely wrong:
On April 25 2012 02:01 chuky500 wrote:
Why all year long were surveys announced with a 0.5 point precision while even the actual results are announced with a 2 point precision !
Second, the results published at 8 o'clock are not official results. They are projections based on a few tens of thousands or hundreds of thousands votes - not even an million (last time I checked). If you base your projections on the results obtained in pollling stations that are together not sufficiently representative of the entire population, your projections are likely to differ from the actual results. This is what happened in this case, because the votes for Sarkozy and Le Pen were a bit differently distributed in the polling stations that were not used for the projections than in the ones that were used, while for the other candidates the distribution was roughly the same.
Finally, regarding Cheminade, sorry but this is clearly you not understanding what a margin of error is.
|
Lets make things straight if some people have doubts about what a margin of error is (basicly):
p is the real percentage you want to know, p is unknown unless you survey the whole population.
p_hat is an estimate of p calculated from a survey
e is the margin of error calculated with statistical formula based on p_hat and some hypothesis/approximation, the bigger the sample the smaller e is.
p_hat +/- e creates a confidence interval, it means there is a high probability (usually 95%) that the real p is between p_hat - e and p_hat + e. At least if p_hat is unbiased.
|
I thought it as a thread about politics, not statistics....
|
Projections aren't based on a few tens of thousands. Like I said I count votes in my town of 150 000 inhabitants intra muros. You start counting at 6 and it's usually done within 45 minutes. 30 minutes later you know the results for the whole region, 800 000 votes counted in mine. Every region counts at the same time. Everyone phones everyone, everyone knows what's counted and even though some small towns are slow or have to recount, the results are pretty accurate. There's no way institutes doesn't know these results. Add that to the fact that for the first time DOM-TOM and foreign countries were already counted because they voted one day before. That's 900 000 votes in DOM-TOM, 400 000 in foreign countries. The only figures that institute don't know by the time they announce their projections is the votes in cities like Paris, Lyon, Marseille but since they're survey institutes they already know how these cities voted in past elections and know how to predict how they'll vote. That's the only part of the prediction that has incertitude, everything else is known for a fact.
Regarding the Cheminade survey example you're saying his result is within a window of 0-3.5 This window is the same as saying his score is 1.75 within a 1.75 margin, still the same window of 0-3.5 To sum up this like saying a margin of 3 is as accurate as a margin of 1.75, almost twice as innacurate !
|
In other news, Sarkozy started to put some more extremism in his speeches. Polls say he's getting more and more behind (at least 54% for Hollande).
Next tuesday (may 1st), french people will follow some traditions: protests from left wingers, some kind of celebration from the far right, and Marine Le Pen (leader of the far right) will announce who she gives support to for the secound round (presumably no one).
Next wednesday evening there will be a debate between Hollande and Sarkozy on TV. The next morning we may very well know for sure who will be elected.
And sunday we vote!
|
On April 27 2012 17:50 chuky500 wrote:
Projections aren't based on a few tens of thousands. Like I said I count votes in my town of 150 000 inhabitants intra muros. You start counting at 6 and it's usually done within 45 minutes. 30 minutes later you know the results for the whole region, 800 000 votes counted in mine. Every region counts at the same time. Everyone phones everyone, everyone knows what's counted and even though some small towns are slow or have to recount, the results are pretty accurate. There's no way institutes doesn't know these results. Add that to the fact that for the first time DOM-TOM and foreign countries were already counted because they voted one day before. That's 900 000 votes in DOM-TOM, 400 000 in foreign countries. The only figures that institute don't know by the time they announce their projections is the votes in cities like Paris, Lyon, Marseille but since they're survey institutes they already know how these cities voted in past elections and know how to predict how they'll vote. That's the only part of the prediction that has incertitude, everything else is known for a fact.
Regarding the Cheminade survey example you're saying his result is within a window of 0-3.5 This window is the same as saying his score is 1.75 within a 1.75 margin, still the same window of 0-3.5 To sum up this like saying a margin of 3 is as accurate as a margin of 1.75, almost twice as innacurate !
Those two are not the same, there is probability distribution on the interval making those two statements different.
|
On April 27 2012 17:50 chuky500 wrote:
Projections aren't based on a few tens of thousands. Like I said I count votes in my town of 150 000 inhabitants intra muros. You start counting at 6 and it's usually done within 45 minutes. 30 minutes later you know the results for the whole region, 800 000 votes counted in mine. Every region counts at the same time. Everyone phones everyone, everyone knows what's counted and even though some small towns are slow or have to recount, the results are pretty accurate. There's no way institutes doesn't know these results. Add that to the fact that for the first time DOM-TOM and foreign countries were already counted because they voted one day before. That's 900 000 votes in DOM-TOM, 400 000 in foreign countries. The only figures that institute don't know by the time they announce their projections is the votes in cities like Paris, Lyon, Marseille but since they're survey institutes they already know how these cities voted in past elections and know how to predict how they'll vote. That's the only part of the prediction that has incertitude, everything else is known for a fact.
Read my post again - the number of votes already counted is irrelevant to the argument I was making, namely that they are projections based on the results at some polling stations and not all. If the already-counted polling stations that are used to make the projections are not sufficiently representative, the projections will likely differ from the official results. This is what happened. Do you have an issue with this explanation or can we close this discussion?
On April 27 2012 17:50 chuky500 wrote:
Regarding the Cheminade survey example you're saying his result is within a window of 0-3.5 This window is the same as saying his score is 1.75 within a 1.75 margin, still the same window of 0-3.5 To sum up this like saying a margin of 3 is as accurate as a margin of 1.75, almost twice as innacurate !
Do you really not understand the difference between a result of 0.5% with a margin of error of three percentage points and a result of 1.75% with a margin of error of 1.75 percentage point? They're two different estimations...
|
On April 27 2012 17:39 Kerm wrote:
I thought it as a thread about politics, not statistics....
Reading my political science lesson: statistics and geography are the main tool to administrate the state.
Just kidding
|
Lol, sorry if this has already been brought up.
Francois Hollande's campaign video (with rap music) -_-
|
On April 23 2012 20:22 DirtyCash wrote:Show nested quote +On April 23 2012 17:45 Agathon wrote:Funny to see that people who vote for the racist party live in places where...there's almost no french with foreign roots (like in my birth place) ^^ In the same time, in Paris, a multicultural city, Le Pen is barely at 5% We've some work to do to stop this ignorance in our countryside, it's such a shame. How sheltered can you be exactly? the place that votes the most for the "racist party" as you call it is historically the south-east region, and if you think "there's almost no french with foreign roots" there you are the ignorant one. Keep ignoring immigration issues, and you will see the FN get stronger each election.
I live in Paris, and there is 6% Front National vote. You go into the tube in Paris and on whole lines there are not a single white person. Fun fact, nobody is bothered with that. They are mostly loyal and honest citizen, and we know it by experience. So people don't vote Le Pen.
Now. My parents have a house in the Meuse. One of the most rural, poor, economically unattractive region in France. I have never seen a Muslim, a black, and arab or an asian person there. In our village, Le Pen is at 35%.
Le Pen vote is vote of fear, of lies, of manipulation, of reject. People in this little village are dead scared of these people they know only through the voice of manipulative far right politicians and propaganda TV news. They have barely seen one of them in their whole fucking life. They fantasize about those barbarian people who want to install Chariah in France. That's ridiculous.
Le Pen is very strong in the North (low immigration high poverty), Alsace and Lorraine (low immigration high poverty), Vendée (low immigration very conservative) and the South East, that's true. It is super strong in countryside and rural areas (no immigration) and very weak in big cities (high immigration). So your observation is biaised.
Let's make it clear.
There is no "immigration issue". There are issues with poverty and its correlated criminality. A recent independant report showed that immigration was actually very beneficial to France economically. There are thousand of illegal immigrants who live in France since 10, 15, 20 years, work in Paris Brasseries, pay their social cotisations and don't have ANY rights.
If you are not convinced please read this:
http://www.20minutes.fr/economie/722873-immigres-rapportent-plus-coutent-economie-francaise
here an interview with economist Alexandre Delaigue
http://www.liberation.fr/politiques/01012371431-l-immigration-est-elle-une-charge-ou-une-chance-pour-l-economie
and here from the Huffington Post, saying that the Bundesbank says that Germany will need 200 000 more immigrannts to save their economy. That's the god damn Bundesbank, not Jean Luc Mélenchon
http://www.huffingtonpost.fr/karim-amellal/immigration-france-allemagne_b_1463617.html?ref=economie
And more than everything, there is no problem at all with Islam in France, except for a couple of idiots fundamentalists, here and there who must be around 1 for 10 000 in the Muslim communities. And if we want to talk about them, let's also talk about the fuckers from institut Civitas and other Christian nuts who harass people and doctors in front of abortion centers or people going to watch "blasphematory" theatre plays. Is there "a problem with Christianity" in France? Of course not.
But French are stupid, and if you repeat day after day that the problem of France is hallal food, veil, and the skin colour of people, they start believing it. Of course it's easier to blame the black dudes than to look at the state in which our education system is, the state in which suburbs are, the skyhigh enmployement rate in poor areas, the way police treats people who happen not to be white, etc etc etc etc etc...
The fact that Le Pen lies, day after day after day after day and manipulates statistics to scare people should be a sufficient reason to suspect that there is something behind "immigration issue".
Lepénisation des esprits, they said. Thanks for Sarko and his crew of criminals to have helped a lot to make France a more closed minded, racist and conservative country. And yes, the FN is a very very racist party, in case you doubt about it.
|
Why do you even try to tell this to someone anymore.
Facts say: It's poverty/income diffrence.....
Right wing sais: It's people with diffrent skin colour!
Morons believe the right wing..
|
On May 02 2012 18:55 Velr wrote:
Why do you even try to tell this to someone anymore.
Facts say: It's poverty/income diffrence.....
Right wing sais: It's people with diffrent skin colour!
Morons believe the right wing..
Unfortunately planet earth is full of morons 
|
On May 02 2012 18:34 Biff The Understudy wrote:
Lepénisation des esprits, they said. Thanks for Sarko and his crew of criminals to have helped a lot to make France a more closed minded, racist and conservative country. And yes, the FN is a very very racist party, in case you doubt about it.
2007 : Le Pen scores 10%, people say "Sarkozy has taken an ultra right turn and has stolen all the voters from the FN because he talks like Le Pen does"
2012 : Le Pen scores 18%, people say "Sarkozy has taken an ultra rigth turn and is responsable for the rise of the FN
2002 : Sarkozy wasn't in the political scene and FN scores 16%
Ultra Right wing ideas and their popularity are not caused by Sarkozy, stop blaming him for everything.
|
On May 02 2012 18:58 Geiko wrote:Show nested quote +On May 02 2012 18:34 Biff The Understudy wrote:
Lepénisation des esprits, they said. Thanks for Sarko and his crew of criminals to have helped a lot to make France a more closed minded, racist and conservative country. And yes, the FN is a very very racist party, in case you doubt about it. 2007 : Le Pen scores 10%, people say "Sarkozy has taken an ultra right turn and has stolen all the voters from the FN because he talks like Le Pen does" 2012 : Le Pen scores 18%, people say "Sarkozy has taken an ultra rigth turn and is responsable for the rise of the FN 2002 : Sarkozy wasn't in the political scene and FN scores 16% Ultra Right wing ideas and their popularity are not caused by Sarkozy, stop blaming him for everything.
No Sarlozy got elected by adopting the discourse of Le Pen and making it legitimate. It worked once. Now we pay the consequences: Le Pen is higher than ever, and the bridge between right and far right are blown up.
|
I don't really think that the average FDG/NPA voter is really smarter than those of the FN tbh.
They have even more ressentiment lol.
Austerity will have a bitter taste anyway and it is always funnier when it is done by socialists =)
|
|
|
|
|
|
EPT
