|Date: ||Tue, 28 Jun 2005 09:16:19 -1000|
|Reply-To: ||Bob Schacht <firstname.lastname@example.org>|
|Sender: ||"SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>|
|From: ||Bob Schacht <email@example.com>|
|Subject: ||Re: International differences in rating questions|
|Content-Type: ||text/plain; charset="us-ascii"; format=flowed|
At 07:49 AM 6/28/2005, David Lindsay wrote:
>On 24 Jun 2005 at 16:52, A P wrote:
> > I'm in the process of analyzing data from an international customer
> > loyalty/satisfaction study. In the study we ask the respondent to rate
> > their satisfaction (on a scale of 1-10) with a wide variety of different
> > experiences.
> > Part of this analysis will include comparing the results between
> > countries. One of the questions I anticipate getting is how does
> > culture play into the differences we see.
>My experience is that problems can occur in the
>translation of scales e.g. translating "slightly agree"
>into say Italian. In other words, nuances in word
>meanings have different interpretations when translated
>back into english.
I think David brings up an excellent point. Although A P professes an
interest in "cultural" differences, she has yet to identify any variables
in this study that have "cultural" content. Previously, the only cultural
variable evident was that perhaps country of residence was tracked. This of
course serves as a rather crude proxy for "culture," but at least its more
than nothing. David implies a second variable of relevance: What language
was the satisfaction questionnaire administered in? How many
language-specific versions were there? Were people from, say, Switzerland,
able to choose between the German, French and Italian versions of the
questionnaire? Better yet, did all respondents in all countries have a
choice of which questionnaire to use?
> > Does anyone have any experience/information on how to handle cultural
> > differences when analyzing data from rating questions?...
Unless A P is imagining that a priori cultural differences have been
demonstrated and have already been established, the ability to handle
"cultural differences" requires questions in the survey instrument with
cultural content in order to establish whether such differences exist. The
"a priori" strategy assumes that studies such as Benedict's "The
Chrysanthemum and the Sword" have somehow established cultural patterns
that can be inferred backwards, but this involves a logical fallacy, the
fallacy of affirming the consequent, IIRC, as well as naive anthropology
(read post-WWII reviews of studies of national character). For example, to
expand upon David's suggestion, suppose this cultural stereotype was that
Italians choose "slightly agree" in a certain context (I am not suggesting
David would actually do this). But if in that certain context someone
chooses "slightly agree," does that imply that he or she is Italian? I
don't think so. In fact, I suspect that all cultural a priori's should be
viewed with a great deal of skepticism.
However, if A P's instrument tracks both the respondent's country, and the
language in which the questionnaire was administered, these then constitute
two crude indicators of culture, allowing the data to be grouped
accordingly, from which one can test the questionnaire responses using
Contingency table analysis (e.g. chi-square) or ANOVA, so that one can
actually determine whether differences exist with regard to certain questions.
Robert M. Schacht, Ph.D. <firstname.lastname@example.org>
Pacific Basin Rehabilitation Research & Training Center
1268 Young Street, Suite #204
Research Center, University of Hawaii
Honolulu, HI 96814