David Routh and Nick Cox have kindly provided excellent references to
previous work. I am glad that I had not seen these before summarizing my
results, as they now offer independent confirmation.
Nick Cox pointed to Mosteller F & Youtz C, "Quantifying Probabilistic
Expressions" in Statistical Science Vol 5 No 1 Feb 1990. It is worth
quoting from their appendix to complement my comments:
"Under the most carefully controlled conditions, respondents do as they
please, as experimenters and survey scientists well know. Although we
asked the respondents [science writers in this survey] to give estimates
in whole numbers, some gave answers such as 0.001, <50, 10-20, which
required editing." M&Y used the same procedure as myself, substituting the
midpoint for a range and rounding odd values.
"In work with scales in questionnaires, it is a familiar finding that
respondents occasionally get turned round and give the complement of the
number they intend. ... When such errors occurred on the more extreme
expressions, we changed the estimates to their complements"
This was one explanation I thought about for the responses 100% and 98%
for "never" and 5% for "almost always", but my respondents were contacted
and I left them as outliers.
On Thu, 28 Nov 2002, David A. Routh wrote:
> Thanks for the interesting extracts from your report on quantifiers.
>
> There is an extensive literature on quantifiers, ranging through
> linguistics, psychology, cognitive science, logic, etc.
>
> Some of the work has concerned the meaning and use of frequency and
> probability. I am too busy right now to pass on the most
> relevant references, but a good overview is by two psychologists in
> Glasgow University, viz., Moxey, L.M. & Sanford, A.J. (1993).
> Communicating Quantities: A Psychological Perspective. Hove: Lawrence
> Erlbaum Associates.
It is one of the frustrations of working in data analysis that questions
of methodology are scattered throughout the literature on all subjects and
may be difficult to identify. One of the criticisms of the RAE exercises
(one of many!) has been that academics making a substantial contribution
to research by assisting with data analysis were not given credit because
the work was assessed only against its contribution to the topic studied.
R. Allan Reese Email: [log in to unmask]
Associate Manager GRI Direct voice: +44 1482 466845
Graduate School Voice messages: +44 1482 466844
Hull University, Hull HU6 7RX, UK. Fax: +44 1482 466436
====================================================================
Under the most carefully controlled conditions, respondents do as they
please, as experimenters and survey scientists [, census enumerators and
administrators] well know. - adapted from Mosteller & Youtz, 1990.
|