[PEN-L:11980] Re: Swing (renamed: Surveys)

[knowware@xxxxxxxx (Tom Walker)]

Michael Perelman wrote,

>>A question, such as the percentage favoring socialism, will be highly
>>sensitive to the phrasing -- probably moreso at the time.

And Wojtek Sokolowski replied,
> . . .
>In more general terms, surveys are rather useless, if not misleading, in
>testing opinions on value-charged items, since people would give what they
>consider the "politically correct" answer, rather than what they actually
>think.  In case of opinions, the validity (=consistency with the phenomenon
>intended to be measured) of such survey items are hard, if at all possible,
>to verify, but such verification can be made in items measuring observable
>behaviour.

Another problem with a result like 25% favouring socialism is that it
doesn't tell us anything else about what the survey respondents thought.
People also BS in ordinary conversation, but we often have enough other
information to discount their implausible claims of church attendance or
support for socialism. There's a funky little region of social science
methodology called "Q-methodology" that offers a trenchant critique of
survey research methodology and presents an alternative to the traditional
approach.

Here's how it works: first the researcher gathers a collection of statements
on the issue at stake from as many different viewpoints as can be tracked
down. Then a sample of say 40-60 of the statements is selected in a way to
retain the maximum diversity of opinion. The statements are shown to
interview subjects who are asked to rank them, e.g. from "most agree" to
"most disagree". The subjects' rankings are then evaluated statistically to
identify common factors in the ways people have ranked the statements.
Instead of simply adding up all the people who say they support socialism, a
q survey would report *how strongly* people say they support socialism (say
on a scale of +1 to +5) and it would report it in relation to how strongly
they held other opinions, such as say, "support for capitalism."

Obviously, if someone showed strong support for both socialism and
capitalism, the researcher might conclude that they must have a very unique
understanding of what the two terms mean. But it is unlikely that such an
idiosyncratic view would be expressed often enough to show up as a common
factor. It is more likely the common factors would reveal contexts for
"support for socialism" that nevertheless suggest a wide range of
definitions of socialism (i.e. some people would take socialism to mean a
workers' state, others would be thinking of reformist welfare policies, etc.).

A q methodology survey can be done with as few as 30 to 40 participants and
still produce meaningful results. If anyone wants to follow up on q
methodology, I'm currently doing a survey on attitudes toward shorter work
time. In a separate message, I'm sending the sample statements for that
survey (and if anyone wants to participate in the survey, contact me off the
list for instructions).

One thing that is already clear from just collecting and sifting through the
published statements is that the debate is asymetrical. That is to say,
opponents and advocates of shorter work time are talking about two different
things. It will be interesting to see how opinions shape up away from the
limelight of official "spokespeople".

Regards,

Tom Walker
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
knoW Ware Communications
Vancouver, B.C., CANADA
knowware@xxxxxxxx
(604) 688-8296
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
The TimeWork Web: HTTP://WWW.VCN.BC.CA/TIMEWORK/