Sarkozy seems more dangerous than I thought | 79 comments (79 topical, editorial, 0 hidden)

So the polls don't have a "don't know/decline to answer" category?

Talk about unprofessional.

"It's the statue, man, The Statue."

by Carrie (migeru at eurotrib dot com) on Tue Apr 17th, 2007 at 08:01:53 AM EST
[ Parent ]

They do report the number of people who didn't name a candidate, but to put it into context, one has to do some minor number crunching.
For example, in the latest Ipsos, Sarko gets 28.5% of the votes, and 14% of those planning on voting didn't name a candidate. Which means Sarko actually received 24.5%. And the number of people who didn't name a candidate varies quite a bit from pollster to pollster.
In other words, don't be surprised if the second round is between Besancenot and Voynet.

"The basis of optimism is sheer terror" - Oscar Wilde

by NordicStorm (m<-at->sturmbaum.net) on Tue Apr 17th, 2007 at 08:18:26 AM EST
[ Parent ]

They do (4.00 / 4)

but they post the results brought back to 100% of those that express a view point.

Not sure if they update the relevant error interval.

Le Monde's graphs with polls (for sub.) do give the info on "no opinion":

See for CSA: it's 21% (the white shaded area). It varies from 7% to 24% according to the poll outfits.

(click for larger)

In the long run, we're all dead. John Maynard Keynes

by Jerome a Paris (etg@eurotrib.com) on Tue Apr 17th, 2007 at 08:29:04 AM EST
[ Parent ]

Re: They do (none / 1)

Sheesh, their definition of "Abstention" is "people certain to vote but not expressing a preference".

"It's the statue, man, The Statue."

by Carrie (migeru at eurotrib dot com) on Tue Apr 17th, 2007 at 08:32:55 AM EST
[ Parent ]

Re: They do (none / 1)

"certain to vote"? From an opinion poll? How do they know?

by Colman (colman at eurotrib.com) on Tue Apr 17th, 2007 at 08:36:26 AM EST
[ Parent ]

Re: They do (none / 1)

Question 1: Do you intend to vote? Certainly/Maybe/Certainly not
Question 2 (if "Certainly" above): Who would you vote for?

"It's the statue, man, The Statue."

by Carrie (migeru at eurotrib dot com) on Tue Apr 17th, 2007 at 08:37:48 AM EST
[ Parent ]

Let's introduce some push polling (4.00 / 6)

Question: if you knew that Sarkozy being elected president would be the end of the republic and indeed the entire continent, would you be more or less inclined to vote for him?
If you answered more inclined: do you work as a journalist at the Financial Times?

"The basis of optimism is sheer terror" - Oscar Wilde

by NordicStorm (m<-at->sturmbaum.net) on Tue Apr 17th, 2007 at 08:45:41 AM EST
[ Parent ]

Re: They do (4.00 / 2)

Why don't they ask "Are you a lazy git with no sense of civic duty"? If ever there was a question unlikely to get a sensible result ...

by Colman (colman at eurotrib.com) on Tue Apr 17th, 2007 at 08:48:33 AM EST
[ Parent ]

Re: They do (none / 1)

If the question is potentially embarrassing, you just tell the polled person to toss a coin (this being over the phone, the pollster cannot see the outcome), and to tell you the following:

heads: yes
tails: the true answer

If both answers are potentially embarrassing, you toss two coins and say:

two heads: yes
two tails: no
one head and one tails: the true answer

"It's the statue, man, The Statue."

by Carrie (migeru at eurotrib dot com) on Tue Apr 17th, 2007 at 11:46:19 AM EST
[ Parent ]

Re: They do (none / 1)

Huh?

by Colman (colman at eurotrib.com) on Tue Apr 17th, 2007 at 11:50:42 AM EST
[ Parent ]

Re: They do (none / 1)

It's a standard mechanism to remove bias. The tradeoff is that the margin of error is increased.

"It's the statue, man, The Statue."

by Carrie (migeru at eurotrib dot com) on Tue Apr 17th, 2007 at 11:58:25 AM EST
[ Parent ]

Re: They do (none / 1)

Reference. I can't see straight off how that works, but then I'm doing two other things at the same time.

by Colman (colman at eurotrib.com) on Tue Apr 17th, 2007 at 12:01:00 PM EST
[ Parent ]

Re: They do (4.00 / 2)

Ah, make that "Reference please, if you have it?". Failed my politeness saving throw.

by Colman (colman at eurotrib.com) on Tue Apr 17th, 2007 at 12:02:07 PM EST
[ Parent ]

Re: They do (none / 1)

That was an exam question in an exam I took last year. I didn't know the formulas so I derived them on the spot. The idea is pretty simple.

Check out question 5(ii) here (PDF, and solutions). The question of reliability hinges on the difference between accuracy and precision.

"It's the statue, man, The Statue."

by Carrie (migeru at eurotrib dot com) on Tue Apr 17th, 2007 at 12:10:22 PM EST
[ Parent ]

Re: They do (4.00 / 3)

Not precisely; the "echantillon representatif" method used by French polling institutes does not allow for the calculation of a mathematical "margin of error."

Your point of course is entirely true for the more widely used "random sample" method, but in the case of French polls, its more accurate to say merely that the problem of screening to remove bias seeks to increase the accuracy of the result at the expense of reducing the precision of the poll.

But in fact we can't use "margin of error" to assess that precision, which is why I am convinced that things remain wide open for Sunday's result.

by desmoulins (gsb6@lycos.com) on Tue Apr 17th, 2007 at 12:03:14 PM EST
[ Parent ]

Re: They do (none / 1)

Excellent comment.

Not that "margin of error" means anything if there are more than two possible (yes/no) answers.

"It's the statue, man, The Statue."

by Carrie (migeru at eurotrib dot com) on Tue Apr 17th, 2007 at 12:11:52 PM EST
[ Parent ]

Re: Which is why I think (4.00 / 5)

They do have that category, but it doesn't correspond to the 40% who say (in other polls) they don't yet know who they're voting for. The different institutes give from 5% to 25% of "no choice" replies.

What they don't give is the number of outright refusals they get (possibly one out of two contacts); or the number of people without a land line who are never polled (around 17% of the electorate).

So they "reconstitute" the electorate and recalculate the percentages. The raw numbers, apparently, give Royal ahead. By the time they've finished cooking, Sarkozy is ahead. Le Pen is multiplied by three (and, imo, is still lower than the real level of intentions).

I'm not saying (or at least I'm not sure that) the pollsters deliberately manipulate in Sarko's favour. I am saying the polls mean a whole lot less than they are cracked up to mean. And that the pollsters play the gurus in the unrolling of the media drama - which is, by and large, framed in Sarko's favour. Public commentary on the election is largely dominated by this game.

by afew (afew(a in a circle)eurotrib_dot_com) on Tue Apr 17th, 2007 at 08:52:02 AM EST
[ Parent ]

Sarkozy seems more dangerous than I thought | 79 comments (79 topical, 0 editorial, 0 hidden)

Top Diaries

Speaking right: What an Taoiseach should say in Washington

Ambushing an Taoiseach

The problems with formal education

Net Zero Energy - March 2025

We do have the cards

Manipulating Irishness

Thomas Paine's Age of Reason

An Enchanted Isle

Recent Diaries

Keith Kellogg Is Out

Nordic States Join Baltics Again WWII Redux

Goodbye NATO/USA, Welcome Europe's E5

Zelensky and the Terror State Ukraine

EU Commissioner VDL In Full Panic Mode

The problems with formal education

The Battle for Kiev/Kyiv

VDL At the Crossroads of Europe - Part 1

The speech Micheál Martin should make in the White House:...

Either At the Table, Or Else On the Menu

A Happy AQ Warrior On Syrian Victory

Trump More Destructive Than Genocide Joe?

Terror HTS FF In Latakia, Syria

Speaking right: What an Taoiseach should say in Washington

Net Zero Energy - March 2025

Ukraine Stop the War

Dutch Labour Morphed Into A Huron War Party

Last Stand of Amazon Warrior Baerbock

Ambushing an Taoiseach

Slow Death of NATO

Occasional Series