Misreading the Newspoll Results

The post by Dennis Shanahan this morning in The Australian was so pitiful in its attempt to derive a story in the Newspoll figures that wasn't there (or was there but was three months too late in the telling) prompted me to download the figures that are freely available from the Newspoll website and recreate them as a graph below (the figures are copyright Newspoll):

What is clear from this is that 

1. It's still a long time until the next election.  The x-axis of the graph is deliberately out until August 2013 to show how far we have to go - but could even be longer. 
2. The Two-Party-Preferred trends over the past six months is obvious and has not, as Shanahan writes, "been wiped out".
3. That neither Gillard nor Abbott are particularly popular.
4. That the number of uncommitted voters has been rising steadily.  Estimates of two-party preferred are thus more and more unreliable.

It was interesting that the Newspoll figures were not linked to his article - perhaps people would have realised that Labor are actually, for the three months in question, up in 4 out of 5 states and that Gillard, who comes in for quite a bit of flak in the article, is now preferred PM in 4 out of 5 states, including, significantly, WA.  Conversely, the Mumble piece the same day in the same newspaper did link the Newspoll stats and discussed them quite intelligently.

One has to wonder why the article was written at all.  It is a slow news day?  Is it politically self-fulfilling - say it long enough and loud enough and it happens?  Or is it a real fear that the NSW and Victorian results are trending back to Labor (as Mumble discusses) and that Queensland might do the same over the next year?  If Queensland was to emulate NSW then the TPP trend can only continue.

For an even more intelligent look at the data, have a look at Pollbludger too.