Two late polls from the NSW state election have maintained the satisfying continuity that has been evident throughout the campaign. In the six statewide polls published during March (two apiece by Newspoll, Essential Research and Galaxy), the Coalition’s two-party lead has varied from 63-37 at the bottom end to 66-34 at the top — a range perfectly accounted for by the standard 3% margin of error.

There remains an assumption that things can’t possibly be as bad for Labor as that, and that a latent sympathy vote awaits to be absorbed by them. I suspect this is a misreading of the public mood. Labor is in fact being run down by a bandwagon effect, with the election looming as a public celebration of the government’s demise — a bit like a Mexican wave at the cricket where only a few curmudgeons in the members stand decline to take part. As difficult as the polling figures may be to process, history suggests they should be taken at face value. Neither Newspoll nor Galaxy has been more than 2% astray on two-party preferred in an election eve poll since 2007, and there is no persuasive reason to expect different this time.

On the primary vote, which is generally more telling under optional preferential voting, Labor has ranged from 22% to 26%. The Newspoll and Galaxy phone polls have unfailingly had the Coalition on 51% or 52%, whereas the online Essential Research polls have had it at 54% and 55%. A complication here is that the two Essential polls are in fact largely the same poll, as the company publishes three-week rolling averages. This makes it all the more tempting to cast its higher figure as the odd man out.

The Greens’ best result was 14% in a Galaxy poll conducted on March 1-2, but it’s notable that this was the earliest of the bunch: the five since have had it at 11% or 12%. It must also be noted that the Greens vote was overstated in each of the six late polls from the federal election as well as the seven from the year’s three state elections, by margins of between 0.2% and 3.8%. This is generally thought to indicate that those who decide while walking into the polling booth tend to go with the major parties.

It is interesting to note that that the only poll last year which hit the bull’s-eye with the Greens was the Essential Research poll at the federal election, as this company has likewise had a 10.3% average for the Greens in this year’s federal polling compared with 13.5% for Newspoll. However, Essential’s state results have the Greens in much the same place as the phone pollsters, so maybe the latter are nearer the money this time.

Unfortunately, the data provided in published polling offers little scope for burrowing deeper. Local polling has been few and far between: there has been a Galaxy poll of Marrickville showing Carmel Tebbutt trailing the Greens candidate 57-43; IRIS Research polling of electorates in the catchment area of the Illawarra Mercury, which have had Labor heading for devastating defeats in normally unassailable seats; and one effort from an unheralded outfit showing independent Port Macquarie MP Peter Besseling neck-and-neck with his Nationals challenger.

The statewide polling has likewise offered little in the way of geographic and demographic breakdowns, though presumably we can expect more on this front from Newspoll in tomorrow’s Weekend Australian. Only Essential Research has provided separate figures for Sydney and the bush, publishing results in mid-February and the final week of the campaign. These confirm the impression that it is in Sydney that Labor’s collapse has taken on extraordinary dimensions. Whereas the two “rest of state” results from Essential Research have pointed to anti-Labor swings there of 6%, in Sydney the swing went from 20% in February to 26% in March. Given the overall 66-34 result from the latter poll was at the ceiling of the overall polling range, it is tempting to speculate that an unusually cranky sample from Sydney helped put it there.

If Labor’s seats were split 30/70 between Sydney and the bush and not the other way round, this disparity might offer some sort of comfort to them. For the most part though, it means they face light swings in safe conservative rural seats where they don’t have much vote to lose.

A crude exercise which averages Essential’s two figures for Sydney and applies preferences as per the 2007 result suggests Labor will go into the election with 35 of the city’s 52 seats and come out with nine: Auburn, Bankstown, Blacktown, Cabramatta, Canterbury, Heffron, Lakemba, Liverpool and Mount Druitt. To this Labor might hope to add Campbelltown, Fairfield, Maroubra, Strathfield and especially Toongabbie, where Nathan Rees is thought likely to pick up an anti-Labor vote of his own.

Then there are the two realistic prospects for the Greens, Balmain and Marrickville. Despite the Galaxy poll a fortnight ago, Greens enthusiasts are advised to believe a win in either seat when they see it. Balmain but not Marrickville appears very much a three-way contest, with the Liberals in the hunt for a once-in-a-lifetime win if Labor and Greens voters neglect to preference each other.

Beyond Sydney, the IRIS Research polls bring Labor further torment. The Illawarra’s five seats cover industrial Labor heartland — enormously different electoral terrain from the country seats that dominate the “rest of NSW” figure. The IRIS figures provide a consistent impression that the area will following the pattern of Sydney with swings of at least 20%. Only Shellharbour looks safe — Heathcote and Kiama are almost certain Liberal gains, and Keira and Wollongong are 50-50 propositions, being respectively at risk from the Liberals and an independent.

Although there has been no specific public polling data, the consensus is that swings in Labor’s other non-Sydney stronghold, the Hunter region, will be only slightly less severe. Newcastle is rated all but certain to fall to lord mayor and independent candidate John Tate. Other independents are said to be running strongly in Wallsend, Charlestown and Swansea, which are also at risk from the Liberals, while Cessnock seems as likely as not to go to the Nationals.

In the remainder of non-Sydney New South Wales, where Labor might hope to benefit from an apparently more subdued public mood, they are defending a grand total of five seats. The only one in which they have real cause for confidence is Monaro, where sitting member Steve Whan is thought likely to survive off his personal vote. There is no such luck for Labor in Bathurst and The Entrance, where the retirement of sitting members has Coalition candidates in the box seat. The odds also appear stacked against Labor incumbents in Wyong and Maitland.

On that reading, should everything fall against Labor they really will end up with that parliamentary cricket team, minus a 12th man. On the rosiest scenarios they will be able to double that, and thereby avoid the added indignity of winning fewer seats than the Nationals.

The Greens could score anything from zero to two. The likely number of independents is hard to pick — as well as the five identified earlier, Janet Mays is said to be as likely as the Liberals to take Blue Mountains from Labor, and there are other dark horses around the place who might yet come through at the finish. Against that, independent incumbents in Dubbo, Tamworth and perhaps Port Macquarie look set to lose their seats to the Nationals, with only Richard Torbay in Northern Tablelands likely to go untroubled.

Not that crossbenchers will be having much of an influence in the new parliament. The Liberals appear all but certain to wield a parliamentary majority in their own right, with the added buffer of a National Party holding perhaps as many as 18 seats.