While it is true that the informal rate has increased compared to the 2016 election, it is still lower than it was in either 2010 or 2013. And the increasing rate of enrolment means a larger proportion of the Voting Eligible Population (VEP) have cast a vote in 2019 than in any election in the last decade.

In the post below I’ve run through the key stats for formal voting and turnout over the last few elections, and look at a map showing the informal rate by seat.

I should clarify a few terms before I start:

• Informal rate – Informal votes as a proportion of total votes
• Voting Eligible Population (VEP) – Total population eligible to vote
• Enrolment rate – Total roll as a proportion of VEP
• Turnout – Total votes as a proportion of roll
• Effective participation rate – Formal votes as a proportion of VEP

This table shows the relevant stats for every election since 2004. I was only able to find VEP for elections since 2010 so the first two elections have less data.

In my last post I did explain how the turnout rate will likely be higher than in 2016, and that the proportion of VEP who have cast a vote has gone up significantly. But I hadn’t explored how this interacts with the informal rate, which had increased since 2016.

It turns out that the effective participation rate, which calculates what proportion of the eligible population has cast a formal vote, has continued to climb up year on year, increasing by about 1% per election, from 80% in 2010 to 83.2% in 2019 (and it should climb a little bit further by the end of counting).

While the informal rate has gone up slightly in 2019, it is still less than in either 2010 or 2013, and partly looks worse because 2016 had one of the lowest rates in the last two decades. With the exception of the big drop-off in informal votes in 2007, the informal rate has bounced around between 4.8% and 6% at every election since 2001.

Next up, this map shows the informal rate per electorate. The map can also be toggled to show how much the informal rate went up (red) or down (green) in each seat.

The informal rate went up in 90 seats, and down in 61. The trend was particularly bad in New South Wales (up in 77% of seats) and Western Australia (up in 88% of seats). The informal rate dropped in two thirds of seats in Victoria. This table shows the informal rate and the effective participation rate per state:

The informal rate is much higher in New South Wales than in any state. This is not a new thing: the leading explanations for this is the use of optional preferential voting in state elections and the larger non-English speaking populations in Western Sydney compared to similar populations in other states. The latter wouldn’t explain the increase, but it’s possible the closeness of the state and federal elections may have resulted in more informality.

Interestingly this higher informal rate doesn’t put New South Wales at the bottom of the pack when it comes to effective participation. Low enrolment rates and turnout rates in Queensland, Western Australia and particularly in the Northern Territory leads to much lower rates of effective participation in those three jurisdictions. In particular the NT stands out as a place with terrible participation rates despite a significant drop in informal rates.

The other factor which played into rates of informality in different seats is the number of candidates running. 10.7% of votes in Mallee were informal, which was up from less than 5% in 2016. This is surely explained by thirteen candidates running here, causing a significant splintering of the vote upon the retirement of the sitting member and a lot more votes simply not counting.

Of the ten seats with the biggest increase in informal rates, six were in New South Wales: two in Western Sydney, three on the north coast and one in far south-west.

We don’t yet know why people vote informal, but previous AEC analysis suggests about half of all informal votes have a clear first preference vote. I expect we’ll see similar results this time. Sure, some people deliberately vote informal as a protest, but many others attempt to cast a formal vote and fail to do so.

That’s not a problem under optional preferential voting, and I think it’s about time we do something to reform the system so voters aren’t required to fill out (potentially irrelevant) preferences to make their count. Sure, let’s encourage voters to number multiple preferences, but it’s immoral to throw out ballot papers with a clear intent to enforce the rule.

I was planning to write a post explaining why this is misleading, that while it may be true that the proportion of the roll to have cast a vote may have gone down, you can’t spin a story about voters becoming disengaged while the proportion of the eligible population who have voted has been steadily increasing.

Yet it turns out that no such nuance is needed, because even the basic fact at the core of the article is false. Updated statistics tonight reveal that the turnout at this election is about to surpass the 2016 election, with more votes yet to be counted.

Yes there are reasons to be concerned about voters’ “disengagement”, but it is false to claim that the number of Australians who are voting is dropping.

Turnout is traditionally defined as the number of people who voted as a proportion of those enrolled to vote. Yet we should also use a different metric which looks at the proportion of the eligible electorate who are enrolled to vote.

This chart shows the proportion of the total eligible population who have turned out to vote, as well as the proportion of the electoral roll.

The AEC has been getting better at enrolling people, in part thanks to the “direct enrolment” system which allows for potential voters to be enrolled without an explicit application based on other government data. This has seen the roll increase from 90.9% of the eligible population in 2010 to 96.8% in 2019.

(It’s worth pointing out that while the marriage plebiscite may have helped it is probably only a small factor. Most of the growth happened between the 2013 and 2016 elections.)

This has meant that it is harder to meet the same turnout goals due to the size of the roll increasing.

Yet despite these concerns, turnout is still looking set to be a bit higher in 2019 than in 2016. At the moment turnout is 0.15% less than in 2016, with the final result not yet declared in 117 out of 151 seats. More votes will be counted and it will undoubtedly result in a higher turnout.

The authors have made two mistakes: the first is comparing turnout without accounting for the increasing completeness of the roll, and the second mistake is to compare final figures for 2016 to interim figures for 2019, and making conclusions about how low the turnout has been on that basis.

They also include data for five seats they identify as having low turnout, and there are good reasons to be concerned about lower turnout in seats like Lingiari in particular, but the numbers they used are already out of date and in one case turnout is now higher than in 2016.

Their data claims that turnout in Melbourne has dropped from 86.8% to 81.6%. The turnout in Melbourne is actually 87.2%, with some votes left to be counted. The drop-off in turnout has almost closed in Blaxland, Sydney and Lalor, with the turnout jumping 2% since their data was compiled. It’s not necessarily their fault for putting out partial information, but I don’t think you can make conclusions about dropping turnout when so many votes are left to be counted, and you should make your qualifications very clear.

This AEC page identifies how many ballot envelopes are yet to be processed. Only in seven seats have all votes been processed. In 49 seats (almost one third of the country) there are 2000 or more envelopes yet to be processed.

Not every envelope will end up counting as a vote, but there are almost 247,000 envelopes yet to be processed. If half of those ended up counting, it would result in a turnout at 91.6% (0.6% higher than in 2016), and would see 88.7% of the eligible population turning out to vote, up from just 84.7% in 2010.

I actually think that’s great, and I don’t think journalists should be using incomplete data to tell the opposite story.

Update 9:45pm: Since publishing this story another 541 votes were counted and now turnout has exceeded the 2016 figure.

Update 11:00pm: It has been pointed out to me that I was using a slightly incorrect enrolment figure for 2016, which means the turnout was actually 91.01%, not 90.86%. This means the 2019 turnout has not yet surpassed 2019 but it will certainly do so.

It is possible that some areas swung in one direction in 2016 because of the popularity (or unpopularity) of Turnbull, which caused the swing in the opposite direction in 2019 to be larger, and by comparing the swings over two election cycles you could isolate which areas have moved the furthest since the Coalition first won power in 2013.

To do this I have added up the 2016 and 2019 two-party-preferred (2PP) swings in the 134 seats which are comparable (two others are completely new seats, while fifteen others are not Labor vs Coalition contests, so we will need to wait some time for the two-party-preferred figures in these seats). I’ll explain some issues with this method further down in the post, but I think the data is easily comparable.

Unsurprisingly, a string of Queensland seats have shifted towards the Liberal National Party against the national trend, while other seats (particularly those in Victoria) have swung most strongly to Labor.

I have then taken these figures and mapped them out, below the fold.

It’s worth noting that the overall Coalition vote is substantially lower in 2019 than in 2013. But this makes it more interesting to see which places have swung strongly to the Coalition.

And top of the list is Queensland. Labor has lost 11.3% in Capricornia since 2013, and over 5% in Hinkler, Dawson and Petrie. Seven of the ten seats with the biggest swings to the Coalition are in Queensland. The other three are Page (NSW, 6.7%), Cook (NSW, 3.7%) and Barker (SA, 3.7%). 18 out of 28 comparable Queensland seats have moved towards the LNP since 2013, while a majority of seats in every other state have moved towards Labor.

It’s also interesting to see which seats have shifted most strongly to Labor. Macarthur and Burt have both swung to Labor by over 11% over the last two elections. Both seats were subject to major redistribution changes prior to the 2016 election. Burt was newly created as a seat in a marginal part of a semi-safe Liberal seat, and has since shifted to become a 5% Labor seat. Macarthur shifted further into the Labor-friendly Campbelltown area in 2013, but it appears the area has swung even further towards Labor since that redistribution, possibly due to changing campaign decisions.

A lot of the seats which have shifted most strongly towards Labor are in Victoria, including Bendigo, Ballarat, McEwen and a lot of suburban Melbourne seats.

Notes on methodology

Kevin Bonham did attempt to get a sense of this a week ago, but he simply looked at how much each seat’s margin has changed between 2013 and 2019. This is problematic since every state and territory has held a redistribution since 2013, and in some places the boundaries have changed dramatically. Macarthur is almost 20% better for the ALP in 2019 than it was in 2013, but part of this is due to the electorate being pulled into the Campbelltown area prior to the 2016 election.

Instead I thought I would add up the two-party-preferred swings in each seat at the 2016 and 2019 elections. This firstly eliminates the influence of the pre-2016 redistributions, since the swings have been adjusted for that redistribution. It does mean the swings are not perfectly comparable in those states which held redistributions prior to the 2016 election. Unfortunately most of the country underwent redistributions in the last 3 years – only New South Wales and Western Australia were spared.

Fortunately the redistributions in Queensland, Tasmania and the Northern Territory were mild. 12 of the 37 seats in these three jurisdictions were completely unchanged, with another 23 experiencing less than 5% change. The Tasmanian seats of Bass and Clark experienced 10-15% change.

Changes were more dramatic in Victoria, South Australia and the ACT. The new seats of Fraser (VIC) and Canberra (ACT) can’t be compared at all, and the average change for the other 49 seats is 17.5%. In particular Hotham and Bruce were changed dramatically.

Despite these changes, I don’t think the scales of the swings have necessarily changed very much, and I think it still gives you a sense of the trends in different regions.

I’ve always found Macquarie to be a fascinating electorate, because it has long been marginal despite the constituent parts of the electorate not being particularly marginal in themselves.

The electorate covers the Blue Mountains and Hawkesbury council areas on the western and north-western fringe of Sydney. About 55% of election-day votes cast in this electorate in 2016 were cast in the mountains.

I divided this electorate into three areas in my pre-election guide, with the Blue Mountains split between lower mountains and upper mountains. I had the Labor two-party-preferred vote at 68.8% in the upper mountains, 57.4% in the lower mountains and just 41.1% in the Hawkesbury area.

I have produced a map showing the two-party-preferred vote in this electorate, as well as neighbouring Lindsay, below the fold.

The electorate of Lindsay recorded a 6.5% swing to the Liberal Party following the removal of local member Emma Husar, and is an interesting comparison to Labor’s results in the Blue Mountains.

You can toggle this map to show the swings at the two-party-preferred level, rather than the totals. In some ways that’s a more interesting map.

The upper mountains are quite distinct from western Sydney and in some ways are similar to more Greens-friendly educated communities on the NSW north coast or in the inner city, while the lower mountains more resemble the western suburbs. Hawkesbury is a Liberal-voting area, but more like Liberal-voting enclaves in Western Sydney such as the Hills district than the Liberal heartland of the North Shore.

The trend in Macquarie reflects what we’ve seen in a lot of places. Labor gained small swings in most of the upper mountains, while the Liberals gained small swings in the lower mountains and some large swings in parts of the Hawkesbury region.

It’s interesting to compare these swings to those in Lindsay, which saw bigger swings than in the lower mountains. I tend to think that is evidence that there was something specific about Lindsay which led to such a large swing, but I admit to not having done the deeper analysis to justify such a claim.

The count in Macquarie will likely end up in a recount, and whoever wins it’s bound to be a close seat at the next election, and will continue to be as long as the Labor-voting mountains and the Liberal-voting Hawkesbury are joined together.

Nationals candidate Anne Webster polled 28.7%, with the Liberal Party’s Serge Petrovich second on 18.4%. Labor’s Carole Hart is on 15.5%, with two independents (Jason Modica and Ray Kingston) polling just over 9%.

The two-candidate-preferred count is between the Nationals and Labor candidates, and shows a clear win for Webster, with 66.4%.

There appears to be a small chance that either of the independents could overtake Labor and then make it to the final count. I won’t go into detail about this possibility but Kevin Bonham has considered this prospect.

After the fold I have some maps showing aspects of the Mallee result, and I’ve analysed the geographic shape of this election result.

There are three layers of voting data in this map.

The first layer shows which candidate topped the primary vote in each booth. The first six candidates each won a booth, including the Shooters, Fishers and Farmers candidate who managed to win a small booth on the eastern edge of the electorate.

Of course, we don’t use a first-past-the-post system so I tried to come up with a slightly different map which gave a sense of whether the Coalition candidates or the other challengers polled a higher total vote, by combining the Labor vote and those for the two stronger independents.

Finally there’s a third map which shows the combined primary vote swing against the Liberal and Nationals candidates since 2016 (when there was a Nationals candidate in most of the seat except for an area in the south-east which was previously in Wannon).

Unsurprisingly the Nationals topped the primary vote in most parts of the seat. Jason Modica topped the poll in most of the booths in the Mildura open area, while Ray Kingston came first in Warracknabeal and a number of booths immediately to the south. Labor came first in a series of booths in the south-east of the seat, with the Liberals also winning some booths in this area.

Another way to look at the diversity of voting patterns is to break down the vote totals by subarea. I broke up the booths of Mallee into five subareas in my pre-election guide: north-east, south-east, south-west, central and Mildura.

The top two candidates were different in almost every area. The Nationals topped the poll in four areas, only falling short in the south-east, which was mostly transferred in from areas previously covered by the Liberal Party. Labor came first in the south-east, followed by the Liberals.

Jason Modica almost outpolled the Nationals in the Mildura areas, but performed very poorly in the rest of the seat. Ray Kingston had a more evenly distributed vote, peaking in the centre of the seat, where he came second. The Liberal Party came second to the Nationals in the Coalition’s stronger areas in the north-east and south-west.

Finally, it’s worth looking at which areas saw the biggest drop in the Coalition vote. The two-candidate-preferred figure masks how much support was lost to the two independents, so instead I’ve looked at a combined primary vote swing for the Nationals and Liberal candidates. It looks like the worst swings were in the north-east (along the Murray) and the south-west, with some pretty big swings in the centre of the seat.

Abbott held his seat by an 11.6% margin against the Greens or 11.1% against Labor as of the 2016 election. Once most of the dust had settled, independent candidate Zali Steggall ended up with 57% of the two-candidate-preferred vote.

Abbott’s primary vote plummeted by more than 12% from 52% to less than 40%.

Today’s map shows the two-candidate-preferred (2CP) vote at each booth in Warringah, and can be toggled to show the primary vote swing against Abbott.

Zali Steggall won the 2CP vote at almost every booth in the seat. Abbott did narrowly win four small booths at the northern end of the seat in the Frenchs Forest area (as well as the pre-poll booth in this area and two hospital booths).

Steggall’s highest vote was in the booths along the Pacific Ocean coast from Manly to Curl Curl, but her support also exceeded 60% in the Neutral Bay area.

We don’t have a 2CP swing, since Steggall did not run in 2016, but it is interesting to examine where Tony Abbott lost the most primary votes. He tended to do best on the northern border of the seat and suffered the biggest swings in the Mosman area.

In my pre-election guide I divided the area of Warringah into three geographic subareas: Manly, Mosman and Warringah. These areas roughly translate into the centre, south and north that I described in the previous paragraph. Abbott suffered an average swing of 8.5% in the Warringah subarea booths, compared to 15.2% in Manly and 16.8% in Mosman.

Possibly the most interesting thing about this race is the lack of variety in the results. Steggall did not win thanks to popularity in one part of the seat. Abbott didn’t hold on to his base in one area while losing ground elsewhere. Mosman had been his strongest part of the seat, but saw the biggest swings. Swings were almost as big in Manly. While Abbott suffered less of a backlash at the northern end of the seat, Steggall still managed over 40% of the primary vote and over 56% of the 2CP vote in this area. There’s no way to look at this result as anything other than a comprehensive dismantling of Abbott’s base in Warringah.

But the seat stood out for the remarkably high One Nation vote in the lower house. One Nation’s Stuart Bonds polled over 21%. One Nation didn’t poll over 20% in any other seat. The next best seat outside of Queensland for One Nation was the neighbouring seat of Paterson, where they polled 14%. Paterson is another urban-rural fringe electorate in the Hunter, covering Maitland and Port Stephens.

I’ve put together an interactive map showing how One Nation did in the Hunter region, which is below the fold. I’m not planning to assess why the swing was so large in this area, but hopefully the map is useful to others interested in the area.

This map has three layers of voting data. You can toggle between the One Nation House vote in the two seats named above, view the One Nation Senate vote in the four seats in the Hunter region (including Newcastle and Shortland) or see the two-candidate-preferred swing across the region. I really don’t recommend viewing more than one of these layers at the same time.

(You can link on the little chain icon in the bottom right of the map to blow it up into its own window.)

One Nation certainly did better in the more rural parts of Hunter, polling well over 25% in the Muswellbrook and Singleton areas and 23% in the Cessnock area. But they also did very well in the more suburban Lake Macquarie area.

In my pre-election guide, I divided Hunter into six subareas. The Lake Macquarie part of the electorate makes up a majority of the population, and was split into north, central and south. One Nation polled just over 21% in the north and south and they polled 17% in the centre.

One Nation did not do as well in the Senate, but it was still their best area in New South Wales. One Nation polled 14.1% in the Senate in Hunter and 11.4% in Paterson. They just cracked 10% in the deeply rural electorates of Parkes and New England, and nowhere else. They doubled the Greens Senate vote in these seats.

If you look at the two-candidate-preferred swing map, you see a similar trend in microcosm to what we’ve seen in the big cities. The more educated and wealthier inner city areas swung to Labor (in this case, the urban parts of Newcastle) while the suburban outskirts favoured the Coalition. For simplicity I have coded swings to the Coalition as blue, but the Nationals were the coalition party in the seat of Hunter.

That’s it for today. This is the first in a series of posts focusing on an election map in an interesting part of the country. If you have a request please post it as a comment below. Tomorrow: Warringah.

This is a big shift compared to the old voting system, where there were many complex exclusion points and seats in play in pretty much every state. It’s also a change from 2016, where the larger number of seats to be elected and the lower quota created more opportunities for close races.

If my estimates of the seat count are correct, this is how the total number of seats in the new Senate currently looks:

• 34 – LNP
• 26 – ALP
• 8 – GRN
• 1 – ON
• 2 – CA
• 1 – CON
• 1 – Lambie

This leaves three seats in Queensland to be contested between the LNP, ALP, Greens and One Nation. If Labor and the Greens both win, it will create a situation where Labor, Greens and Centre Alliance form a blocking majority, or a full majority with Jacqui Lambie. If the LNP and One Nation both win, the LNP, One Nation and the Conservatives will hold half the seats, with Lambie or Centre Alliance having the votes to give them a majority.

Either way, Centre Alliance and Lambie are important, but Lambie will be more important if the right win two of the remaining seats in Queensland, as this will mean her vote alone can be decisive, whereas Centre Alliance will be in a stronger position if the left wins two of the remaining seats.

I’ll run through the race in New South Wales, Victoria and Tasmania before coming back to Queensland. I won’t touch on Western Australia and South Australia – in both these states the result of 3 Liberal, 2 Labor and 1 Greens seems very clear. I also think the same outcome is pretty much locked in for New South Wales and Victoria but I’ll run through those below.

New South Wales (75.2% counted)

• Liberal/National – 2.7169
• Labor – 2.1149
• Greens – 0.6019
• One Nation – 0.3471

Since last week the Liberal/National position has strengthened, while the Greens have weakened, but there is no real prospect of the Greens being overtaken. There are more preferences from small right-wing parties but the Greens should still benefit from preferences from the Labor surplus and HEMP, Animal Justice and ICAN, as well as smaller parties.

The other point of interest in New South Wales is how Jim Molan is doing with his below-the-line campaign as the #4 candidate on the Liberal/National ticket. It is hard to be sure about this count because most votes remain “unapportioned” – as in the AEC is yet to distinguish between above-the-line and below-the-line votes.

But there is a sample of about 13,000 votes for the Coalition in booths where every vote has been apportioned and some of these votes are below-the-line. In this sample Jim Molan has 1108 votes, or 8.5% of the total Coalition vote. This probably exaggerates slightly, but even still if it was extrapolated to the whole state it would mean that Molan would be sitting on about 0.23 of a quota, with the #3 candidate Perin Davey on about 0.48 of a quota. Bear in mind that Davey would be able to accumulate above-the-line preferences while Molan would not. There’s no way Molan would come close in this scenario.

Victoria (71.8% counted)

• Liberal/National – 2.5372
• Labor – 2.2263
• Greens – 0.6963
• Derryn Hinch – 0.1977
• One Nation – 0.1970
• DLP – 0.1766
• United Australia – 0.1761

Since last weekend the Liberal/National position has solidified. The Greens have gone backwards, but should still be fine. Hinch will have a lot of trouble coming first out of the group of minor parties and even then won’t have anyone to help him with preferences.

Tasmania (76.0% counted)

• Labor – 2.2074
• Liberal – 2.1670
• Greens – 0.8768
• Jacqui Lambie – 0.6128
• One Nation – 0.2456

The first five seats are entirely locked and I can’t see anyone challenging Jacqui Lambie for the final seat.

Lisa Singh is doing better than Jim Molan but doesn’t appear to have a chance. She is polling 16.7% of the total Labor vote in booths where votes have been fully apportioned. This would translate to about 0.37 of a quota, leaving the second Labor candidate with over 0.8 of a quota. This would knock third Labor candidate John Short out of the race right away but there just aren’t enough Labor votes for her to win. Bear in mind that Singh can’t gain any above-the-line preferences until #2 candidate Bilyk is elected, so she will be seriously hindered in the race for preferences, and will likely be overtaken by One Nation.

Queensland (71.7% counted)

• Liberal National – 2.7646
• Labor – 1.6038
• Greens – 0.6809
• One Nation – 0.7211
• United Australia – 0.2408

Since my last update the LNP and One Nation have strengthened their positions while Labor and the Greens have weakened.

Ross Leedham on Twitter has been tracking this race closely, and in this tweet below estimated that the Greens and Labor’s final primary vote position will weaken further once the remaining votes are counted.

Accounting for further postals and other declaration votes, remaining percentage projection:

Greens – 9.5% (~0.66 quotas)
ALP – 8.5% (~0.59)

Greens will have the lead but it is close. LNP projects to 11%+ remaining (~0.77+ quotas) and ON are around ~10.3% (~0.72).

— Ross Leedham (@rossleedham) May 25, 2019

I won’t dive into the detailed preference projections but Kevin Bonham and Ross have both done analysis on this topic. One Nation in particular did quite well from preferences in 2016. Overall there are a lot more votes for minor parties of the right than of the left. Considering that the LNP and One Nation are leading in the race, this suggests they are likely to win these seats, with the Greens and Labor’s second candidate competing for the final seat.

If Labor ends up missing out in this seat, Labor, the Greens, Centre Alliance, One Nation and the Australian Conservatives will end up with the same number of seats as they held before the election. Fraser Anning, Brian Burston (UAP), Derryn Hinch, Duncan Spender (LDP) and Tim Storer (who chose to retire) will have all lost their seats, with the Coalition gaining four seats plus Lambie. Most of those departing crossbenchers were relatively right-wing, so it doesn’t massively shift the balance of the Senate, but effectivly Hinch and Storer have been replaced with Lambie and a Liberal in the centre of the Senate, which probably does strengthen the government’s Senate position.

Update: Kevin Bonham has also published a post about the Queensland count.

Last Sunday I identified seven seats still in play. Since then most of these seats have become clearer. The Liberal Party has clearly won in Boothby, Chisholm and Wentworth with margins of over 1000 votes. Four other seats are deserving of attention. I’ll run through the state of the race in those four closest seats below the fold. I’ve called two of these seats (Bass for Liberal, Lilley for Labor), while Cowan and Macquarie are still in play.

Liberal candidate Bridget Archer is leading by 699 votes in Bass. Labor did surprisingly well on the postal votes in Bass, with 49.5% of the total postal vote, but it wasn’t enough to overcome a narrow Liberal lead on the ordinary vote.

There are now just 106 postal votes waiting to be processed, with as many as 800 others yet to arrive (although some of these will never arrive). There’s only 104 absent votes yet to be counted.

The only substantial batch of votes yet to be counted are the 1538 declaration (out of area) pre-poll votes, and it is pretty much impossible for Labor to catch up on this batch – they’d need over 70%. So I’ve called this seat for the Liberal Party.

Labor’s Anne Aly is leading by 799 votes in Cowan. There are relatively few postal votes yet to be counted, but there is a lot of absent votes (over 4500) and declaration pre-poll votes (over 5000). Aly is leading in the absent votes and won both of these categories in 2016, so it seems likely she will win, but it’s still too early to call this seat.

Labor is leading by 879 votes in Lilley. There are not that many postal votes yet to be counted, but there are over 3500 absent votes and over 3000 declaration pre-poll votes. The absent votes are breaking strongly towards Labor. I can’t see the Liberal Party winning here.

Finally the race is extremely close in Macquarie. Liberal challenger Sarah Richards is leading by 125 votes.

Most postal votes have been counted but there are over 4000 absent votes and over 2000 declaration pre-poll votes. Either party could easily win, and there’s a real chance the seat could end up with a recount if the final margin is less than 100 votes.

So I was interested in zooming out to get a sense of how many seats had swung in each direction, and how they fit into the respective “marginal”, “reasonably safe” and “safe” categories.

I’ve defined these categories as follows:

• Marginal – 0-6% margin
• Reasonably safe – 6-12% margin
• Safe – 12%+ margin

Overall I have found a reduction in the number of marginal and reasonably safe seats and an increase in the number of safe seats, mostly on the Coalition side, but also that both major parties have seen seats moving in both directions.

I should note that I have included Cowan and Lilley as Labor seats, and Bass, Chisholm and Macquarie as Coalition seats post-election. I have also treated Wentworth as a safe Coalition seat and Chisholm as a marginal Coalition seat before the election.

This first table shows the number of seats in each category before and after the election.

You can firstly see a net increase in Coalition seats of three, and a drop of two Labor seats, and a gain of one for the crossbench. That reflects the latest expected numbers of 78-67-6.

The total number of seats considered “marginal” has dropped from 49 to 43, with both parties holding less marginal seats than they did before the election.

Labor previously needed a uniform swing of 1% to win the 76 seats necessary for a bare majority (an increase of 5). They now need nine seats (assuming no change in who is leading), which requires a uniform swing of 3.4% on the latest figures.

The Coalition previously needed a swing of 0.02% to regain their majority. A swing of 0.8% would now see them lose their majority.

Next up, this chart shows the seats by their pre-election category and how many seats in that category saw a swing to the incumbent or away from the incumbent.

There is no broader trends of seats in certain marginality categories swinging one way or the other. In fact it appears that both major parties gained support in at least half of their marginal seats, with the Coalition doing particularly well.

The Coalition also gained swings towards them in their safer seats, while the opposite was true for Labor, which will tend to help the Coalition national two-party-preferred figure without having an impact on the seat count. Yet in every category at least one third of seats are bucking the trend.

Finally I decided to check how many seats are now in a different category to what they were before the election.

63 seats out of 151 have changed category. This includes 28 marginal seats. In eight of these seats (6 Labor and 2 Coalition) the seat changed hands. In the other twenty the seat became safer for the incumbent party.

12 reasonably safe Coalition seats moved up a category, while it was more common for Labor seats to move down the category.

While there has been some general shift towards the Coalition, and a reduction in the breadth of the marginal seat battleground, it is more interesting to see where these most marginal seats are.

Prior to this election, the ten most marginal Coalition seats on the pendulum included five Queensland seats, as well as three in New South Wales, one in Victoria and one in South Australia.

Now there is only one Queensland seat on this list (Longman), as well as three seats in New South Wales (Macquarie, Reid and Wentworth), two in Victoria (Chisholm and Higgins), two in Tasmania (Bass and Braddon) and one each in South Australia (Boothby) and Western Australia (Swan). That makes for quite a different election next time around.