Re: Anyone good at statistics...??

r norman wrote:
On Tue, 1 Sep 2009 21:03:18 +0100, "Manx" <Manx@xxxxxxxxxxxxxx> wrote:

Hi foks.

Just wondering about all those surveys that ask hundreds or thousands of people what they think of Darwinism, Creationism, etc: I'm told that when the figures are extrapolated to produce an estimate for the entire population, there's some kind of 'confidence level' that can be calculated from the size of the sample.
Is this right so far...?

So... what can a survey say if only *three people* were asked? What does the statistical analysis say that makes it clear to everyone that three is a lousy number of people to ask...? Hoe does the 'three-ness' of your survey show up in your final results?

Does my question make sense...?

Ta for all thoughts!

M.

This is a well worked out problem in sampling. See, for example,
Margin of Error and Confidence Levels Made Simple
http://www.isixsigma.com/library/content/c040607a.asp

which has a table of percent margin of error (at the 95% confidence
level) vs. sample size. A sample size of 50 has a margin of error of
14% and the error grows very rapidly for smaller sample sizes.

That table only goes down as far as a sample size of 50.

For sample sizes way below that (such as the OP's three), the usual assumption of normality no longer holds. For a sample size of only three, it's impossible to say what the distribution looks like, and so some non-parametric test would need to be used instead.


--
Steven L.
Email: sdlitvin@xxxxxxxxxxxxxxxxxxx
Remove the NOSPAM before replying to me.

.

Relevant Pages

  • Re: WHOS UP FOR MADITORY POLING
    ... margin of error unless you get exactly 100% participation ... A 10 question mini poll might be good for examining some ... cant we make the survey 20 questions long and INCLUDE ... one about refs? ...
    (rec.sport.disc)
  • Re: Anyone good at statistics...??
    ... people what they think of Darwinism, Creationism, etc: ... there's some kind of 'confidence level' that can be calculated ... Hoe does the 'three-ness' of your survey ... Margin of Error and Confidence Levels Made Simple ...
    (talk.origins)
  • margin of error of mean from 3-valued (0,100,200) mutiple-choice survey question
    ... I'm trying to figure out the right formula to compute margin of error ... for the mean value of a single multiple-choice survey question whose ... can be as small as 25 responses. ... My job is to produce web-based reporting tools for our survey data. ...
    (sci.stat.math)
  • Re: Gods Numbers
    ... I believe polls have a sampling ... uncertainty is often expressed as a margin of error. ... sampling error of 3% for the estimated percentage of the whole ... The AMERICAN RELIGIOUS IDENTIFICATION SURVEY of 2001 had a error ...
    (talk.origins)
  • Re: Only 43% will vote pap in Queenstown
    ... how come survey never provide margin of error? ... 1% confidence level? ... oppositions who cannot govern! ...
    (soc.culture.singapore)