KEY POINTS:
Helen Clark is popular among female voters as preferred Prime Minister in the latest Herald Digi-Poll survey, while Labour is virtually neck-and-neck with National in the race for the female party vote.
Winning the female vote is something National has openly targeted, and it has gone slightly ahead of its main rival in the poll - registering 44.4 per cent of female support, just ahead of Labour on 43.3 per cent.
However, the battle between the two leaders sees a clear-cut victory for Helen Clark, with the Labour leader registering 56.3 per cent support as preferred Prime Minister among females, and National's John Key 33.9 per cent.
The poll suggests the Prime Minister remains very much Labour's strength, as her party battles to peg back the large gap in various political polls.
The poll also shows Labour's attacks on Mr Key have had little impact. When the final week of the survey's calls was broken out, the party vote remained much the same as the overall survey result.
Geographically, the race for Auckland's vote appears to be being won by National, which registers 50.3 per cent support in the region, compared with Labour's 39 per cent.
Helen Clark's popularity as preferred Prime Minister in Auckland is only slightly higher than Mr Key's, a far smaller gap than the overall rating for the whole country - further emphasising National's popularity in Auckland.
Helen Clark remains a clearly preferred Prime Minister in the South Island and Wellington.
The Greens enjoy higher levels of support in the South Island and Wellington grouping than their national average, registering 8.1 per cent of the party vote in those areas.
New Zealand First leader Winston Peters, who has been in the spotlight discussing Air New Zealand's troop-carrying flights, registers 6.9 per cent in the preferred Prime Minister stakes.
That may suggest that despite his party polling below the crucial 5 per cent threshold, on just 2.8 per cent support, Mr Peters is popular enough to potentially lift New Zealand First back into Parliament.