Summary of findings –
- Preferences were not distributed in the NT, the lead Labor and CLP candidates declared elected on the first count.
- At 19.5% the Northern Territory had the highest rate of voters going beyond six preferences above the line, four times the national average. This was helped by there being only nine ballot paper groups in the NT.
- 77.7% of Green preferences reached Labor, but not by following the Green how-to-vote. Of all Green votes, 45.5% went to Labor as a second preference, another 21.3% at the third preference after giving a suggested second preference for HEMP.
- United Australia Party (UAP) preferences favoured Labor, against the party’s how-to-vote recommendation for the CLP, largely because one in five UAP above-the-line votes were donkey votes.
- On how-to-vote concordance, 16.0% of Labor voters followed the how-to-vote exactly compared to 10.3% for the CLP and 10.2% of the Greens. Green concordance rates were lowered by the 2nd preference being given to HEMP rather than Labor.
The Constitutional Basis of Territory Senate Representation
Section 7 of Australia’s Constitution guarantees the six original states equal representation in the Senate. Since 1984 there have been twelve Senators per state, elected for fixed and staggered terms, six positions facing election every three years.
The two territories have Senate representation courtesy of Section 122 that gives the Commonwealth Parliament power to legislatively determine representation for territories. In 1975 the two Territories were granted two Senators each to be elected for a maximum of three years and with terms tied to House of Representative’s elections. (At the time both sides of politics supported ending fixed Senate terms and tying them to terms of the House.) On oddity of Territory Senate terms is that Territory Senators would face election at a separate House election but not at a separate half-Senate election.
Territory Senators were first elected at the 1975 double dissolution election. With only two seats per Territory, the quota for election is 33.3%, and every election in both Territories has produced the same result, one Labor and one Coalition Senator. In the Northern Territory the Coalition is represented by the local Country Liberal Party.
The 2019 Result
The 2019 NT Senate election reprised the previous 16 elections by returning one Labor and one CLP Senator. Labor’s Malarndirri McCarthy was re-elected, joined by new CLP representative Samantha (Sam) McMahon, elected in place of the retiring Nigel Scullion. Both were declared elected on the first count having polled more than the 33.3% quota required for election. There was no need for preferences to be distributed.
There were nine groups on the ballot paper and no ungrouped candidates. One Nation did not contest the Northern Territory. The ballot draw for position saw the United Australia Party allocated column A, Labor column B, the Country Liberals column C, with the Greens at the other end of the ballot paper in column H. Fraser Anning’s Conservative National Party (CNP) drew the final column.
Preferences weren’t needed, but the AEC was still required to scan and data enter all ballot papers. From the data files it is possible to analyse how the preferences for third parties were directed, and measure the impact of how-to-vote material distributed by parties.
The table below sets out the first preference tally. Rather than list groups in the traditional ballot paper order used by the AEC, I’ve re-arranged the groups in descending party vote order.
Table 1 – 2019 Northern Territory Senate Election – First Preferences Result
|Group – Party/Candidates||Votes||% Vote||Quotas||% Ticket|
|B – Labor Party||39,353||37.47||1.1241||92.29|
|McCARTHY, Malarndirri (Re-elected 1)||2,791||2.66|
|C – Country Liberal||38,513||36.67||1.1001||94.51|
|McMAHON, Sam (Elected 2)||1,580||1.50|
|H – Greens||10,752||10.24||0.3071||87.34|
|GILL, Lia Ngaree||169||0.16|
|A – United Australia||6,469||6.16||0.1848||89.09|
|WOLF, Michael J||617||0.59|
|McROBERT, Ross Thomas||89||0.08|
|F – HEMP||4,027||3.83||0.1150||88.83|
|I – Fraser Anning’s NCP||2,207||2.10||0.0630||84.19|
|DICKSON, Mark James||316||0.30|
|WHEELER, James David Richard||33||0.03|
|E – Rise Up Australia||1,955||1.86||0.0558||90.08|
|D – Group D||1,290||1.23||0.0368||67.05|
|G – Citizens Electoral Council||461||0.44||0.0132||79.18|
|Total / Turnout||108,994||0.00|
Ballot Paper Categories
In a previous post (published here) I noted that the NT had recorded a lower than average rate of votes cast using the above the line (ATL) instructions to number at least 6 preferences. Of NT voters, 66.5% of all formal votes counted as 1-6 ATL votes, lower than the national figure of 80%, but higher than in Tasmania and the ACT where local experience with the Hare-Clark electoral system encouraged many more voters to use the below-the-line (BTL) voting option.
Rather than voting BTL, many NT voters went beyond six preferences ATL, no doubt encouraged by there being only nine groups. 19.5% of NT voters went beyond six ATL preferences, more than four times the national figure.
Table 2 (below) reports the different categories of ballot papers for each party/group. Groups are listed in descending order of percentage 1-6 ATL voting.
There are two observations that can be made about Table 2 –
- Voters for the two major parties were the most likely to vote only 1-6 above the line and the least likely to vote below the line. The pattern is reflected in every state. This reflects how major party voters used to vote under the old group voting ticket system with the highest rate of ATL voting and lowest of BTL voting.
- Group D, who had no party name above the line, had by far the lowest rate of 1-6 votes (37.8%) and the highest rate of BTL voting (32.9%). As I will show over the next fortnight, this pattern repeats in every state. It appears that voters may be confused by groups with no party name above the line, causing them to shun ATL voting and to vote BTL.
Table 2 – 2019 NT Senate – Ballot Paper Categories
|% by ATL Preference Sequence|
|Country Liberal (C)||38,513||2.2||2.6||69.4||20.3||5.5|
|Rise Up Australia (E)||1,955||4.2||4.8||59.0||22.1||9.9|
|United Australia (A)||6,469||3.9||5.8||55.8||23.6||10.9|
|Fraser Anning’s CNP (I)||2,207||1.8||2.6||53.8||26.0||15.8|
|Citizens Electoral Council (G)||461||2.4||5.0||52.1||19.7||20.8|
Preference Flows and Effective Preferences
In assessing how preferences flow at Senate elections, I have always analysed what I call effective preferences. An effective preferences is one that chooses between the remaining parties at the end of the count. In the case of electing two Senators for the Northern Territory, the effective preference for every ballot paper is in the choice between Labor and the Country Liberals.
The data in Table 3 has been arranged in descending ATL vote order.
Table 3 – 2019 NT Senate – Effective Preferences by Party
|Party (Group)||ATL Votes||% to ALP||% to CLP||% Exhaust|
|United Australia (A)||5,763||48.7||43.2||8.1|
|Fraser Anning’s CNP (I)||1,858||20.6||69.0||10.4|
|Rise Up Australia (C)||1,761||30.7||57.4||11.9|
|Citizens Electoral Council (G)||365||36.7||46.6||16.7|
|Total Excluded Parties||23,580||55.3||36.7||7.9|
The table looks very much like the sort of two-party preferred preference flows you would see in the House of Representatives. Green preferences heavily favoured Labor over the Country Liberals (77.7% to 18.7%) , UAP preferences slightly favoured Labor due to the ‘donkey’ vote (see below), HEMP preferences split evenly, and other party preferences favoured the CLP. Generally, the lower the vote for a party, the higher the rate of preferences that exhausted before reaching either Labor or the CLP.
A ‘donkey vote‘ is a ballot paper that is numbered sequential down or across the ballot paper, seemingly without attention paid to who the candidates and parties are. It shows up as the first candidate having a small but statistically higher vote than candidates in any other ballot position, and also as a higher preference drift down (or left across) the ballot paper.
That candidates at the top of a ballot paper, or on the left in countries where people read left to right, have a small advantage compared to candidates in other positions has been measured in most democracies, especially as ballot paper size increases. However, it does appear to be more noticeable in Australia, almost certainly due to the combination of compulsory voting and compulsory preferences.
The donkey vote tends to show up most clearly in a drift down a House ballot paper, or across a Senate ballot paper, when the lead candidate is excluded. It is very evident in UAP preferences in the Northern Territory.
The UAP issued a how-to-vote with a second preferences for the CLP and a fourth preference for Labor (see available how-to-votes here). Of the 5,763 UAP ATL votes, 1,142 or 19.8% can be classed as donkey votes. 735 numbered 1-6 and stopped, 53 were 1-7, 50 1-8 and 304 ballot papers numbered all nine squares in order left to right. Some of the shorter sequences may reflect duplicated and missing numbers in preference sequences.
It is almost certain that donkey votes explain why UAP preferences favoured Labor, against the UAP’s how-to-vote recommendation and against the flows against Labor in other states. One factor that is difficult to measure is whether the structure of the UAP how-to-vote influenced this, with six boxes, columns A-E plus G, displayed with an out of order number sequence. Did some voters perhaps see the boxes rather than the numbers and simple write 1-6 left to right?
Of all UAP ATL votes, 37.2% had a second preference for Labor as against only 25.5% for the Country Liberals.
Effective Green preference flows were 77.7% to Labor, reflecting the effective destination of the Green’s preference recommendation. The Green how-to-vote (again see how-to-votes here) listed a second preference for HEMP and a third preference for Labor. But most of the 9,391 Green ATL votes did not follow this sequence, with 4,271 or 45.5% reached Labor with a second preference, where only 2,000 or 21.3% reached Labor at the third preference after HEMP.
Of Labor’s ATL votes, 45.9% had a second preference for the Greens and 21.0% for the CLP. Measuring an effective preference for Labor in choosing between the Greens and CLP, 60.3% opted for the Greens and 34.8% the CLP.
Of CLP votes, 41.0% followed the how-to-vote with a second preference for the UAP, with 18.3% of CLP votes having a second preference for Labor. It has always been the case that a noticeable proportion of major party voters preference the other major party ahead of minor parties.
Table 4 below sets out statistics on how many ATL votes were in concordance with each party’s how-to-vote, that is that voters exactly followed a recommendation on a how-to-vote. The table sets out the proportion that matched the how-to-vote to a second preference, a third preference, and so on to the 6th preference and beyond. Again, you can inspect the how-to-votes here.
Table 4 – 2019 NT Senate – HTV Concordance by Party
|% of ATL Votes Matching Party HTV by Preference Number|
|Party (Group)||ATL Votes||2||3||4||5||6||>6|
|Country Liberal (C)||36,398||41.0||18.2||11.4||10.9||10.2||0.5|
|United Australia (A)||5,763||25.5||1.5||0.5||0.3||0.3||0.0|
|Fraser Anning’s CNP (I)||1,858||18.8||4.1||1.3||1.2||0.9||0.3|
|Rise Up Australia (E)||1,761||23.1||5.1||4.1||3.5||2.6||0.1|
Note: How to votes for all groups were not located.
The above table indicates that very few voters followed party how-to-votes. Labor had the highest concordance rate with 16.0% of Labor voters giving six preferences in the same order as the party recommended. The rate was 45.9% at the second preference. Even without a how-to-vote, many Labor voters were clearly able to guess the Greens would be placed second, and overall 25.9% matched the how-to-vote to the third preference for HEMP.
As already mentioned, 45.5% of Green voters gave their second preference to Labor, and only 30.5% matched the recommended second preference for HEMP, and 21.3% matched 2-HEMP and 3-Labor.
As already mentioned, around a fifth of United Australia Party ATL votes were donkey votes, only 25.5% getting the correct second preference for the CLP. UAP concordance disappeared at the third preference for the Citizens Electoral Council. Rates of concordance disappeared after the second preference for all small parties.
Past research on lower house how-to-vote concordance by the Victorian and South Australian Electoral Commissions has shown major parties achieving concordance rates of up to 45%. The South Australian data reveals that the concordance declines as the number of candidates increases. This makes sense as a voter who does not receive a party’s how-to-vote might correctly guess a sequence with three or four candidates, but would struggle to correctly guess a sequence as the number of candidates increases.
I think the lower Senate rates might be a problem created by the structure of party how-to-votes. With lower house ballot papers, parties go to some effort to make their how-to-votes look like the ballot paper.
The larger size of Senate ballot papers makes it impossible for a party how-to-vote to look like the ballot paper. Parties either show six boxes in left to right order with numbers, or list an order of preferences along with the associated column code and party name.
In the Northern Territory, the Labor Party provided a how-to-vote with the nearest approximation to the ballot paper. (again see here) Whether Labor got the strongest concordance because of this, or because they handed out the most how-to-votes, is difficult to say.
What is clear is that preference flows under the new Senate electoral system are nothing like in the pre-2016 era of group voting tickets. Minor parties used to deliver 90% of preferences in a controlled flow via group voting tickets, where voters making up their own minds on preferences create aggregate preference flows that are almost random. And the lower a party’s vote, probably due to there being fewer how-to-votes distributed, the more random the preference flow.