[OPE] Philosophical development of graduate students today

I worked for about three years professionally in survey questionnaire design
for a government statistics office, and typically the three main issues you
start off with before even going so far as devising valid measurement
concepts are:

(1) what is the purpose or goal of gathering information by means of the
chosen survey instrument,
(2) how will the data set that results be processed/what is done with it,
(3) what is the response burden likely to be.

Usually before running a pilot to test the questionnaire, those who
commission the survey are interviewed, to ascertain their views in detail
and consider various options.

The first question is important because the problem which obtaining
information via a survey is supposed to solve, may possibly be solved
without using any survey instrument, and, because the survey is limited in
scope, only that information which is exactly relevant should be collected,
and no more. Failing that, you get a waste of resources.

The second question is important because if you collect data in a certain
format, it directly affects your very ability to process that data, and if
you collect unusable data or data which you don't even use, then why collect
it in the first place? Lack of a clear answer to this issue results not just
in waste, but also creates problems for further analysis of the data.

The third - usually most decisive - question is important, because depending
on circumstances, it may be ethically unfair or practically unjustifiable to
impose a complex questionnaire on respondents, if doing so imposes an extra
burden on them. Moreover you have to be sure that respondents can actually
answer all the questions as put, since if they cannot answer them or some of
them easily, then your survey results will be of poor quality and the
inferences drawn from the obtained distributions will be dubious. In this
sense, poor quality questionnaire design will create not just low quality
data, but may also contain response error due to design faults and
question-wording effects, the magnitude of which cannot be detected in the
results unless they are specifically tested for, e.g. by split ballot
experiments varying the question routine (an additional complication and
expense). Normally you do not run a questionnare unless you are sure that
the respondent can give a definite answer to all questions asked, as proved
by the pilot.

Dr Paul Cockshott says: "We want our doctoral students to have a conceptual
lexicon drawn from multiple areas of science and philosophy." Then you can,
of course, require students to specify all the books they have or have not
read, but in fact students may find this an arrogant and pretentious
exercise, and insofar as they answer, provide answers which are unreliable,
rather than finding the survey fun to take part in.

So a research statistician - in contrast to a fumblebutt - is, when faced
with this project, likely to say first of all (if he is at all good) -

"How could you best go about promoting that doctoral students have a
conceptual lexicon drawn from multiple areas of science and philosophy, and
who is really responsible for this? Is a survey really necessary, and if it
is, how exactly is it conducive to furthering this aim?".

If you, as the philosopher king, cannot provide a well-reasoned answer to
this from your own throne, then it's best to think twice before embarking on
a survey at all.

Jurriaan

_______________________________________________
ope mailing list
ope@lists.csuchico.edu
https://lists.csuchico.edu/mailman/listinfo/ope
Received on Sat Aug 29 08:01:44 2009