Newspoll is keeping up the good work in the post-federal election lull with a series of state polls, today following last week’s Victorian and South Australian polls with a survey showing the Coalition taking a narrow lead in Western Australia. The following charts show how Newspoll has tracked the progress of the Bracks/Brumby, Gallop/Carpenter and Rann governments.

Welcome to episode two in the slower-than-anticipated Post-Match Report round-up of federal electorate results, which today brings us to South Australia.

Of the three seats that were highly marginal for the Liberals going into the election, Kingston emerged with the smallest Labor margin following a relatively subdued 4.5 per cent swing. The swing was reasonably consistent throughout the electorate, though slightly heavier at Morphett Vale and the Liberal-voting suburbs to the north than along the coast. Makin produced the third biggest swing in the state, perhaps boosted by the retirement of sitting member Trish Draper, with the 0.9 per cent margin obliterated by an evenly distributed 8.6 per cent shift to Labor. In Wakefield the swing was 7.3 per cent, which was markedly lower than in the small towns in the north of the electorate than in the low-income outer Adelaide centres of Elizabeth and Salisbury.

Only at four of Boothby’s 42 booths did Nicole Cornes achieve a swing greater than the 5.4 per cent needed to win the seat. All were in strong Liberal areas, including the coast around Brighton and the Adelaide Hills suburb of Flagstaff Hill. Labor’s worst results came in the area closest to the city, with swings to the Liberals recorded at Mitcham, Myrtle Bank, Kingswood and Hawthorn West. The Greens’ vote picked up 3.1 per cent, perhaps benefiting from embarrassment surrounding Cornes’s performance. In Sturt the Labor candidate Mia Handshin picked up a close-but-no-cigar swing of 5.9 per cent that was concentrated in the heavily mortgaged northern end of the electorate, with swings near or above 10 per cent at Dernancourt, Gilles Plains and Windsor Gardens. Pyne now sits on an uncomfortable margin of 0.9 per cent.

The 7.2 per cent swing in Adelaide was slightly higher than the state average of 6.8 per cent, and was driven in remarkable degree by the stronger Labor areas to the north and north-west of the city. The swings in many of these booths cracked double figures, whereas the strong Liberal booths to the north-east and south-east of the city mostly came in at well under half that. Labor’s Hindmarsh MP Steve Georganas also had a much more relaxing election night this time around after prevailing by 108 votes in 2004, picking up a 5.0 per cent swing that was fairly evenly distributed throughout the electorate.

Labor’s biggest swing in South Australia was wasted in the safe Liberal rural seat of Barker, where Liberal member Patrick Secker went to preferences for the first time since 1998 after his primary vote fell from 53.2 per cent to 46.8 per cent. Labor was up 8.6 per cent on the primary vote and 10.4 per cent on two-party preferred. Swings were larger in the bigger centres than the small rural booths: all five Mount Gambier booths produced above average swings, peaking at a remarkable 21.4 per cent at Mount Gambier North. Talk of a swing in Grey big enough to endanger the Liberals was partly borne out by double-digit swings in the seat’s traditional Labor centres of Whyalla, Port August and Port Lincoln. Swings were much more gentle in the many smaller rural and remote booths, dampening the overall shift down to an insufficient but still severe 9.4 per cent.

Alexander Downer’s seat of Mayo followed the statewide trend in swinging to Labor by 6.5 per cent. Particularly heavy swings were recorded at the southern coastal towns of Victor Harbor and Goolwa. Nine years after coming within an ace of winning the seat, the Australian Democrats can now manage only 1.5 per cent. The Greens did well to increase 3.4 per cent to 11.0 per cent, partly assisted by the donkey vote. Another good seat for the Greens was Port Adelaide, where they picked up 3.3 per cent and boosted Labor from a 3.7 per cent increase on the primary vote to 6.8 per cent on two-party preferred. Remarkably, all but one of the 10 booths in Paralowie, Salisbury and Parafield to the east of Port Wakefield Road produced a double digit swing, a trend which carried over into neighbouring Makin. Swings in booths further west varied around the 4 per cent mark.

The verdict from the McEwen recount is in: Labor candidate Rob Mitchell’s six vote win has been overturned, and Liberal member Fran Bailey declared re-elected by just 12 votes. This gives the result level pegging with the Liberals’ 1974 win in Stirling as the closest federal electorate result of modern times. Labor is still considering a legal challenge, but it’s an open question as to whether a re-match would really be in their interests. It seems very likely that we can now settle on a final result of 83 seats for Labor and 65 for the Coalition plus two independents. Two other recount demands await adjudication:

• The Greens will reportedy call for a recount for the Victorian Senate, a contentious move given that nearly 3.3 million ballots would need to be rechecked. Antony Green’s projection shows both the Coalition and Labor winning third seats upon the exclusion of eighth placed Family First, the Liberals doing so with a surplus of 21709 votes (0.68 per cent) and Labor with 6088 (0.19 per cent). At this point Greens candidate Richard di Natale is left stranded on 13.4 per cent, 0.9 per cent or 27804 votes short of a quota. This of course assumes that all votes are cast above the line, when there are in fact 65101 (2.05 per cent) below-the-line votes for which we presently have only first preference results. These are unlikely to make much difference, as most are votes for parties whose preference tickets favoured the Greens ahead of Labor. Much of the leakage would come from Liberals going below the line to ensure the Greens did not get their vote. Against this can be weighed Labor voters who gave their first preference to a Labor candidate before switching to the Greens, but past experience suggests this is unlikely to account for more than 10 per cent out of 14123. If the assumption of all votes behaving as ticket votes were to hold, the Greens would need for Labor to finish around 2000 votes below the quota after Family First’s exclusion, which is roughly 8000 less than they presently appear to have. The distribution of the Liberal surplus would then be enough to give di Natale the narrowest of victories. In support of their recount appeal, Greens spokesman Jim Buckell provided The Age with an interesting list of claimed irregularities: “309 Greens Senate votes from one booth were not recorded at all; in Isaacs 150 votes were missed; in Dunkley 173 Greens votes were recorded as 17; and in Gellibrand, some Greens votes were attributed to another minor party”. However, it seems most unlikely that the required average of around 215 votes per electorate would be found to have wrongly favoured Labor over the Greens.

• Labor candidate Jason Young’s request for a recount in Bowman following his 64-vote defeat has been knocked back by the divisional returning officer. Young is continuing to pursue his recount request further up the Australian Electoral Commission hierarchy, but one suspects he is unlikely to find any joy.

On a completely unrelated note, here is a chart I knocked together showing each state’s deviation from the national Labor two-party preferred vote going back to 1949.

The first thing to note is the hyperactivity of Tasmania, which can in large part be put down to its small population of five seats. Nonetheless, the results tell a story of a natural Labor state which turned around temporarily following the Whitlam government’s tariff cuts and Labor’s opposition to the Franklin dam at the 1983 election. The largest state by contrast has stayed within a narrow 5 per cent band on the Labor side of the ledger, dipping below the line only in 1987 and 1998. Victoria’s long-lost standing as the jewel in the Liberal crown looks very much like a symptom of the 1954 Labor split and the party’s subsequent paralysis at state level, and its Labor vote has only once fallen below the national result since 1980. The exception was the 1990 election which also proved aberrant for reliably conservative Queensland, state government factors providing the explanation in each case. It can also be seen that the Coalition’s relative strength in Western Australia at the 2007 election was matched only by 1961, there is nothing new about its conservative leaning.

On another completely unrelated note, I have just had to pay a fee to renew the pollbludger.com domain. This wasn’t hugely expensive ($50 to be precise), but it nonetheless offers a good excuse to pass the hat around among those of you who enjoy giving me money.

Roy Morgan’s first face-to-face poll of the Rudd era shows Labor with a predictably bloated two-party lead of 60.5-39.5. Read all about it here.

With electorate results progressively being declared, I will start appending my election guide entries with overviews of results for each seat. All five seats in Tasmania have been declared, so that seems a good place to start.

Bass provided Labor supporters with cause for nagging doubt during the early part of the count, with the smaller booths outside Launceston delivering a seemingly insufficient swing. In Scottsdale the swing to Labor was below the required 2.6 per cent, and Liberal member Michael Ferguson in fact picked up a small swing in Bridport. The turning point came when the big Launceston booths began to report, with Labor swings as high as 7.6 per cent at Summerhill and 8.1 per cent in Newnham. The other notable feature of the result was a big surge to the Greens who were able to monopolise the anti-pulp mill vote, pushing their support up from 8.1 per cent to 15.3 per cent at the expense of both major parties. This was reasonably consistent throughout the electorate with the interesting exception of Scottsdale, where the increase was only 0.8 per cent. Nothing particularly remarkable happened in George Town, the centre closest to the actual site of the mill.

The pattern of voting across Braddon was remarkably similar to the 2001 election, with voters reverting to type after the convulsion of Labor’s forestry policy in 2004. A large number of booths have produced double-digit swings first one way and then the other, including Acton in Burnie and East Devonport, along with the smaller town booths of Montague, Latrobe, Smithton. Coastal centres outside of the big towns, such as Wynyard, Somerset, Penguin and Ulverstone, followed relatively small swings to Liberal in 2004 with relatively small swings to Labor this time. However, Sid Sidebottom’s overall margin of 1.4 per cent (from a two-party swing of 2.6 per cent) is substantially lower than his 6.0 per cent from 2001. Predictions that the Mersey Hospital would boost the Liberals in Davenport at the expense of a backlash in Burnie received fairly modest support, Burnie collectively swinging 4.4 per cent compared with 1.2 per cent in Davenport. Despite a quite healthy lift on the Greens’ primary vote from 5.6 per cent to 8.1 per cent, Braddon remains their weakest Tasmanian seat.

Lyons produced a superficially status quo result, except that Liberal renegade Ben Quin gouged 9.6 per cent of the primary vote directly at the Liberals’ expense. However, this obscures big swings to Labor concentrated in the southern part of the electorate, particularly just outside Hobart at Brighton and New Norfolk. The 1.3 per cent lift in the Greens’ vote was the smallest in the state, presumably because much of the pulp mill protest vote was absorbed by Quin. Both major parties were slightly down slightly on the primary vote in Denison, the slack being taken up by a 4.0 per cent lift for the Greens. This converted into a 2.3 per cent two-party swing to Labor. Franklin was one of only four seats in the country to swing to the Coalition, due to the loss of retiring Harry Quick’s personal vote and perhaps also lingering static surrounding Kevin Harkins’ disendorsement. The Labor primary vote was down from 46.4 per cent to 41.4 per cent while the Liberals were up from 37.7 per cent to 41.0 per cent, with the Greens up from 11.1 per cent to 14.4 per cent. The Liberal two-party swing was 3.1 per cent.

A couple of other updates are in order:

• As most of you are aware, a recount began today in McEwen following Labor candidate Rob Mitchell’s six vote win over Liberal member Fran Bailey. Progressive results will not be posted, so I guess we all just have to wait a week until the AEC tells us what has happened.

• In other close result news, rechecking has reduced Liberal member Andrew Laming’s lead in Bowman to just 64 votes, although there does not seem to be any dispute that he has won the seat.

• A definitive result in O’Connor will have to await a full distribution of preferences, which to my limited knowledge is yet to be published in any electorate. There still remains a mathematical possibility that Nationals candidate Philip Gardiner can overhaul Labor’s Dominic Rose with Greens and other preferences and then defeat Liberal member Wilson Tuckey on Labor preferences. However, the possibility has been diminished by a weak Nationals performance on declaration votes, which has reduced their election night total of 18.4 per cent to 17.7 per cent, leaving a 2.7 per cent deficit against Labor that will need to be closed through Greens and other minor party preferences.

• Two other strong performances by independents should be noted. In Calare, Gavin Priestley might overtake Labor on preferences and leave John Cobb of the Nationals with a fairly narrow win on two-candidate preferred. However, Cobb’s 48.5 per cent primary vote is high enough that he does not face a serious prospect of defeat. In neighbouring Parkes, independent Tim Horan has polled 20.7 per cent. This is unlikely to be enough for him to overhaul Labor’s 25.4 per cent on preferences, which is just as well for Nationals candidate Mark Coulton who has pulled up short of a primary vote majority on 46.8 per cent.

Morgan, which ended its recent poor run at the federal level with a 53.5-46.5 result on the eve of the election, has produced the first post-election poll on voting intention. It shows Labor enjoying a honeymoon boost to 58.5-41.5, with a primary vote lead of 49 per cent to 36.5 per cent. Newspoll will presumably return to the fold in the new year.

This time yesterday, Liberal member Fran Bailey held on to a 32-vote lead in McEwen which, though rapidly diminishing, was calculated by Antony Green to be 77 per cent likely to hold after the few remaining votes were counted. Those votes are now in: the last few absent votes broke 100-93 in favour of Labor’s Rob Mitchell, postals went 37-21 his way, pre-polls favoured him to the tune of 33-23, and further rechecking of booth votes cost Bailey 14 and Mitchell eight. All of which leaves Mitchell seven votes ahead. This is apparently the final result, pending the final recount, which could certainly turn up enough anomalies to overturn a lead as small as this. Adam Carr further argues that with a margin of fewer than 20 votes, “the Liberal Party’s lawyers will be able to scrape up some pretext or another for a court challenge”. He also states: “Unfortunately for Labor, most of the precedents are that the incumbent government loses the subsequent by-election (Nunawading, Mundingburra, Greensborough).”

Mundingburra of course was the Queensland by-election in February 1996 that cost the Goss government the one-seat majority it retained after the 1995 election. The other two are from Carr’s home patch of Victoria. There are probably about five people in the country who can tell you about the 1985 by-election for the state upper house province of Nunawading, and I am not of their number. UPDATE: Scratch that – the result cost the Cain government its short-lived control of the upper house, so probably quite a few people know about it, including me from now on. What’s more, it followed an initial tied result and a win for Labor decided by a draw from a hat. The Greensborough by-election refers not to the one Sherryl Garbutt won in 1989, but rather to the one Poll Bludger commenter Chris Curtis ran in as DLP candidate in 1977, which produced a massive swing to the then Labor opposition. ANOTHER UPDATE: A correction in comments from Brian McKinlay (of McKinlay case fame), who says Carr was in fact referring to yet another by-election for Greensborough which took place in 1973, which saw a Liberal win overturned by the court before being re-confirmed by the electorate. One might respond that the 1996 Lindsay by-election demonstrates that voters do not take kindly to initiators of legal challenges, but perhaps the 5.0 per cent Liberal swing on that occasion had more to do with Labor’s generally poor performance at re-matches than is generally realised.

Anyway, let’s assume now for the sake of argument that this result stands. We now have a new modern standard for close federal electorate results to beat Liberal candidate Ian Viner’s 12-vote win in Stirling in 1974. The historians among you are invited to relate other famous close shaves in comments. We also have Labor on 84 seats and the Coalition on 64, with two independents. This is pleasing from a personal perspective as it’s exactly what I predicted early in the campaign for New Matilda, although I did underestimate Queensland’s contribution to the Labor total. Unfortunately, the day before the election I upped the ante to 87 in a prediction for Crikey, which looked very good on election night but became progressively less good as counting proceeded.

This prediction was highlighted today in The Australian, which has promoted me from confuser of fact with opinion and baser of opinion on ignorance and prejudice to the slightly more elevated title of “pundit”. I suspected at first that The Australian compiled this list as a subtle dig at an online commentariat that had leaned a little too heavily to Labor in its predictions, but that doesn’t explain the inclusion of Malcolm Mackerras. In any case, Brad Norington bails me out in the accompanying article by trying on the line that Labor owes its win to “fewer than 12,000 people across nine electorates”. Those of you marvelling over the seven-vote margin in McEwen are invited to consider an election in which the Coalition held on to power after retaining each of Bass, Bennelong, Braddon, Corangamite, Deakin, Flynn, Hasluck, Robertson and Solomon by one solitary vote. On this basis, I hereby declare that my prediction of 87 seats was only out by 595 votes out of 12,350,549. It would in fact be far more accurate to say it was 0.2 per cent out, which isn’t so bad either I suppose.

UPDATE: Adam Carr on historical close results:

In terms of numbers of votes, the closest result in a House of Representatives contest was 1 vote (13,569 to 13,568), when Edwin Kirby (Nationalist) defeated Charles McGrath (ALP) in Ballarat (Vic) in 1919. The result was declared void in 1920. In 1903 Robert Blackwood (FT) defeated John Chanter (Prot) in Riverina (NSW) by 5 votes (4,341 to 4,336). This result was also declared void. The closest result allowed to stand was 7 votes (13,162 to 13,155), when John Lynch (ALP), defeated Hon Alfred Conroy (Lib) in Werriwa (NSW) in 1914. In terms of percentages of the vote, the closest result was Kirby’s voided win in Ballarat in 1919: he polled 50.002% of the vote. The closest result allowed to stand was that in the Griffith (Qld) by-election of 1939, when William Conelan (ALP) defeated Peter McCowan (UAP), after preferences, with 50.007% of the vote. The closest winning margin in recent times has been 50.011%, polled by Ian Viner (Liberal) in Stirling (WA) in 1974 and by Christine Gallus (Liberal) in Hawker (SA) in 1990.

Mitchell has 50.003%, so his percentage is lower than both Conelan’s in 1939, Viner’s in 1974 and Gallus’s in 1990.

On election night and the following day, the best bet seemed to be that Labor would emerge with between 86 and 88 seats. After that, Labor watched leads disappear in one seat after another. Liberal candidates took the lead in McEwen and La Trobe on the Monday after polling day, followed by Dickson and Swan on Tuesday, Herbert on Friday and Bowman on Wednesday of this week. Corangamite, Flynn and Robertson were also on the critical list at various stages after looking secure for Labor on election night: Robertson arguably still remains there. In Solomon, Labor’s Damian Hale watched nervously as his 860-vote lead on booth votes was whittled down to 89 on Wednesday, before he was saved by a late rally yesterday that widened the gap to 194. The entirely one-way nature of this traffic raises the question of what has happened and why. Here at least I will limit myself to the first half of the equation.

The first table shows the size of the swings to Labor for each type of vote in all seats which look to have margins of less than 1 per cent, barring the new seat of Flynn where any swing calculations would be hypothetical (an unfortunate omission as it would have cut the Labor swing on postals still further). Provisionals are excepted because too many of them are still to be counted, and they are few in number in any case. The outstanding feature is the Coalition’s strong performance on postal votes, which cost Labor dearly in McEwen, Dickson, Herbert and La Trobe. I read one newspaper report (I can’t remember where) suggesting this was because most postal votes were cast before the Lindsay pamphlet scandal broke, but the pattern would surely have been reflected in pre-polls if this was the case.

The second table shows the number of votes cast for each type over the past three elections. Here as elsewhere it must be remembered that a small number of 2007 votes still remain to be counted. It can be seen that this election has maintained a trend of sharply increasing numbers of postal votes, exacerbating the impact of the Coalition’s strong performance, along with the more neutral pre-polls.

The final point to note is how lucky the Coalition has been. Present indications suggest it will win five of seven seats determined by margins of less than 0.3 per cent. Assuming no further changes, the bottom end of the Mackerras pendulum will look as follows:

Elsewhere, the chances of a National Party boilover in O’Connor have been reduced as the slowly progressing late count has widened the gap between Labor and the Nationals from 2.08 per cent to 2.70 per cent. It will take an extremely high level of obedience to the how-to-vote card from Greens voters if that gap is to be closed, which seems an unlikely prospect in a sparsely populated electorate where the party would have had a hard time finding volunteers to cover each of the booths. Any vague chance that independent Gavin Priestley might win Calare has probably been laid to rest by late counting which has increased Nationals candidate John Cobb (formerly member for Parkes) from 47.71 per cent to 48.47 per cent, close enough to an absolute majority that the question of who comes second out of Priestley and Labor is probably academic. In the Victorian Senate, the Greens’ hopes rested on what would have been an out-of-character boost from declaration votes, which have in fact reduced their vote from 10.1 per cent to 9.7 per cent. The Labor vote has also faded enough that third Liberal candidate Scott Ryan has overtaken Labor’s number three David Feeney, so that Ryan looks likely to take the fifth seat and Feeney the sixth. Greens candidate Richard di Natale is 1.67 per cent behind Feeney after preferences CORRECTION: I wasn’t factoring in the Liberal surplus, which actually makes the gap more like 0.9 per cent.

UPDATE: One other thing – it is clear that dramatically fewer provisional votes are being allowed through this year. In 2004, any given electorate ended up with about 400 to 600 provisional votes counted. This time it’s more like 100 to 200. I suspect the answer to this mystery lies somewhere in the Electoral and Referendum (Electoral Integrity and Other Measures) Act. Can any wise heads out there point me in the right direction?

UPDATE 2: Comments respondents note that provisional voters must now show photo ID either at the booth or by emailing or faxing a copy to the AEC in the following week. Peter Brent: “Presumably the number of people who took ID to the AEC in the next week was about zero”. Grace Pettigrew: “Many voters who are likely to need a provisional vote do not carry ID around with them (aboriginal voters, the homeless, for example) are also most likely not to vote for the Coalition”. Adam Carr also takes issue with my description of O’Connor as “sparsely populated”: I would argue that this is sort of accurate, but Carr says the real point is that O’Connor is “the most agricultural seat in Australia, where most people live in or near small farming towns”, and consequently has “more booths than any other seat”.