home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math.stat
- Path: sparky!uunet!cs.utexas.edu!torn!news.ccs.queensu.ca!mast.queensu.ca!dmurdoch
- From: dmurdoch@mast.queensu.ca (Duncan Murdoch)
- Subject: Re: unweighted mean of means
- Message-ID: <dmurdoch.115.715529297@mast.queensu.ca>
- Lines: 23
- Sender: news@knot.ccs.queensu.ca (Netnews control)
- Organization: Queen's University
- References: <8edG3vu00WBN868HYr@andrew.cmu.edu> <winzar.73.0@newsman>
- Date: Thu, 3 Sep 1992 14:08:17 GMT
-
- In article <winzar.73.0@newsman> winzar@newsman (Hume Winzar) writes:
- >Of course this does not get you arround the systematic bias created by the
- >propensity to answer TRUE more often.
-
- That bias results from a mis-designed questionaire. If the correct choice
- is randomly chosen (with equal probabilities) and the question worded to
- fit, then any system of picking answers that doesn't depend on knowledge
- about the statement will have equal chances of picking the right and wrong
- answers. You should also try to make knowledge of the subject matter
- important, e.g. don't do this:
-
- As everyone knows, 2+2 = 5. True False
-
- 2+2=4. True False
-
- As everyone knows, unbiased estimates of the variance are essential.
- True False
-
- Most sponsored surveys are just about this blatant in wording their
- questions to elicit the "correct" answer.
-
- Duncan Murdoch
- dmurdoch@mast.queensu.ca
-