As someone with a political interest and a bit of a statistical one I found
the attached emails quite interesting.

Now that Scotland has proportional representation for the election of its
MSP's to the Scottish Parliament, accuracy of polling results for 'other'
parties becomes more important. A poll suggesting a 6% national vote for a
party might lead members of the public/media to perceive voting for this
party as a wasted vote (although because proportionality is made on a
regional basis 6% could actually elect 3-4 MSPs). Whereas only a 9% polling
result could very much be interpreted by the public as a basis for
guaranteeing election of 8-9 MSP's. Quite a difference for a small change!

In light of the below and if Mori are open to changing aspects of their
polling, what should we be suggesting they should be doing - now that it's
not just the accuracy of the big parties that count? Especially considering
that polling accuracy really needs to be achieved individually for each of
the 8 Scottish regions as the national figure gives little indication of
whether your vote counts in your region? Has much statistical work been done
in this area that we can draw on? 

Many thanks,

Mark Westcombe
Dept of Management Science
University of Strathclyde

-----Original Message-----
From: James Mackenzie
Sent: 25 October 1999 20:16
To: Scottish Green Party
Subject: [SGP] Mori explain why their Euro Poll in Scotland was so wrong
about us



I asked why their poll said we'd get 1% and the others (inc. the SSP) would
get 3%, when we got 5.7% and the SSP got 4%. The answer may be of interest
to the list.

yours
James

-----Original Message-----
[forwarded message from Mori begins]

Dear James

There is a problem, yes.  In polls before elections we are, of course,
asking people how they inticipate they will behave.  This introduces
problems due to differential turnout - which can be significant if the
"supporters" of one party fail to turnout.  What is also certainly true is
that, in surveys which ask party support questions without a prompt list
(our surveys were conducted over the telephone, without reading out the
names of the parties that would be contesting the election) the "smaller"
parties can be overlooked by respondents.  Once in the voting booth, with
the full lists available, some people will be drawn in a way that they did
not indicate they would be, prior to voting.  We have seen this happen
before.  In a sense, the rise in support that the Liberal Democrats usually
experience in General Elections is down to this rise in recognition during
the campaings.

Since 1989, we have made a point of indentifying the Green party in our
breakdown of party supporters, and smaller parties are grouped under
"Other".  We may amend our lists to include Scottish Socialist in future.

I hope this helps.  A poll conducted a week before the election is certainly
not predictive, but I think the work we did did provide a useful insight
into the public mood.

Let me know if you have any other questions, pleased to help if I can.

Best wishes


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%