Charles,
I believe that I have seen 0%, 25%, 75%, 100% (and 50% for dont
know).
Given the well known tendency of people to overstate reactions, if I
were forecasting I would use 0%, 5%, 50%, 75% (with 5% for dont
know!)
Peter
, Priority: normal
Date sent: Fri, 15 Nov 2002 18:54:57 +0100
Send reply to: Charles Raux <[log in to unmask]>
From: Charles Raux <[log in to unmask]>
Organization: LET
Subject: Stated Preferences responses weighting
To: [log in to unmask]
A colleague of mine is looking for a clarifying answer regarding SP
data coding: During a stated preferences survey, different
possibilities of responses are offered to respondents with regard to
the characteristics of a new mode (i.e. a new light rail). Hereafter
some possibilities offered : "I will : [1] Certainly choose my current
mode [2] Probably choose my current mode [3] Probably choose the
new
mode (as described in the survey) [4] Certainly choose the new mode
(idem) [5] I don't know"
Question :
Are there (from state-of-practice or state-of-the-art) weights usually
applied to these responses, from [1] to [5] (e.g. if x% of people
respond "I will certainly choose the new mode" what is the weight to
use in the model for "Yes" : 100% of these responses, 90%, 80% ?).
Which arguments can be used to choose one kind of weighting rather
than another one.
Thank you for your help.
---------------
Charles RAUX,
Laboratoire d'Economie des Transports
CNRS-Université Lumière Lyon 2-ENTPE
email : [log in to unmask]
http://www.ish-lyon.cnrs.fr/let
|