This Research
Note compares the political opinion polls results of two polling organisations:
Newspoll and Roy Morgan
Research [Morgan]
since June 2001. Another major pollster, ACNielsen, is not included
because its political polling is not currently as regular as the two
examined.1 The Note will be updated
regularly to incorporate the latest results. It is published in conjunction
with another Research Note examining the issues that voters rank as
important and the party that voters prefer to handle these issues.
The Australian
reports Newspoll's figures and both Morgan
and Newspoll attract considerable attention across the media. Poll
results usually are reported in text and/or tables, but this Note
will present them as graphs, allowing an easier comparison of the
figures. As most polls generally are reported in isolation rather
than together, this regular comparative publication will be a useful
addition to the columns already devoted to polls.2
Future editions of this publication will touch
briefly on some of the major events that may affect poll results.
In this issue, however, we offer only a preliminary comparison of
poll statistics from June 2001, graphing first the results of each
poll for the primary vote, then the results of the polls by party
and finally the results for two-party preferred.
Note that the pollsters use different methods to
calculate the two-party preferred result. Newspoll measures only primary
votes and notionally distributes according to preference flows at
the last election; Morgan
measures full preferences.
The problem with notional distribution is that it cannot account for
significant shifts in voter preferences since the last election; that
is, voters may have altered their preferences since the last election
in response to major events or changes within parties (such as leadership).
1.
For further information on the poll
results, contact Greg Baker, of the Statistics Group. The figures will be posted
on the Intranet.
2.
This is not to argue that polls are
never compared. Reuters, for example, has
in the past examined trends across the polls.
