The weekly Essential Research result, released yesterday a day later than usual, pooped Labor’s party a little in failing to replicate the shift to them detected by other pollsters over the past fortnight. Labor, the Coalition and the Greens are all steady on the primary vote, at 32%, 49% and 10% respectively, although rounding has ticked the two-party result a point back in Labor’s favour to 56-44. The poll also finds 22% expecting the economy to get better over the next year against 45% worse, essentially unchanged on May; 29% and 37% ditto for personal finances; 47% concerned and 37% not concerned about job security, likewise little changed; and 30% thinking they will be better off and 32% worse off under a government led by Tony Abbott. There are also complex questions on the manufacturing sector, and to my mind rather loaded question on public sector job cuts.
We also have an entertaining release from Roy Morgan which replicates Newspoll’s quarterly exercise of breaking down accumulated federal poll results (in this case its face-to-face polling) by state, gender and age. The only substantial differences from Newspoll are that separate results are provided for each month, rather than just a single aggregate for all three, and figures are provided for Tasmania. I gather the monthly sample size would range from around 1100 in the case of New South Wales to barely 100 for Tasmania. Combining the three at least produces a reasonable number for Western Australia and South Australia, while the monthly samples for the larger states are large enough to be useful in their own right.
Tossing aside Morgan’s peculiar respondent-allocated two-party preferred results and using the preference flows of the 2010 election, we see Labor’s two-party result up from 46% in July to 47.5% in August in New South Wales (rounding as Morgan does to half a per cent); from 49.5% to 50% in Victoria; and, interestingly, from 38.5% to 44% in Queensland. Despite the small sample of about 300, the combined three-month result for Tasmania is worth a mention, as I believe it’s the first published result of federal voting intention in Tasmania since the election. It roughly bears out the reported Labor internal polling from Mark Riley of Channel Seven in showing an averaged swing over the period of 14% – enough to cost Labor all four of its seats if uniform, although the margins in Franklin and Lyons would be within the 6% margin of error.
Hopefully Morgan will make a habit of this, as it will at least allow us to see if Morgan’s apparent skew to Labor is more pronounced among particular cohorts, to the extent that this can be accurately measured through comparison with Newspoll. This will require many, many more observations than we have at present, but to get the ball rolling I have looked at the differences between Newspoll’s April-June quarterly result and the June and July figures from Morgan, figuring they should be close enough to comparable due to the poll trend during this period. The only cohorts which buck the trend of Morgan being roughly 2% better for Labor are Victoria, where Morgan produced a quirky 55-45 to the Coalition result in July; Western Australia, where Labor rated 6% higher in Newspoll, which can be written off for the time being given Morgan’s small sample; and among those aged over 50, where Morgan had Labor 4.5% higher.