Previous Next

Double Disillusion

9

The States and Territories

Ferran Martinez i Coma and Rodney Smith

In November 2015, Bill Shorten declared that, if elected, his government would provide $100 million towards the construction of a new Townsville football stadium. The Queensland Labor government would match the funding. The stadium would primarily serve as the home ground for the newly crowned NRL Premiership winners, the North Queensland Cowboys (Australian Labor Party (ALP) 2015). In the months leading up to the 2016 federal election, Shorten continued to promote his stadium proposal, challenging the Coalition to equal his commitment (Peel 2016). Business analysts criticised Labor’s plan, while the Coalition remained uncommitted (Ludlow 2016). During the fourth week of the election campaign, after the Queensland government announced it would increase its funding to $140 million, Malcolm Turnbull matched Shorten’s stadium promise as part of a broader ‘City Deal’ for Townsville. The State’s Assistant Minister for North Queensland welcomed this new bipartisanship, while criticising the time it took Turnbull to make his promise (Australian Broadcasting Corporation (ABC) 2016; Liberal Party of Australia (LPA) 2016b).

Townsville’s football stadium illustrates some of the ways in which federalism and party competition interact in Australian federal elections. The fact that Labor controlled the State government gave federal Labor the possibility of an initiative that created policy and electoral dilemmas for the federal Coalition. As events transpired, the Queensland government was able to leverage State infrastructure funding from both federal major parties. Had the Queensland government been in Liberal–National Party (LNP) hands, as was the case until early 2015, the dynamics of the stadium decision would have been quite different.

While the interplay between Australian federal and State governments influences the behaviour of political parties in federal elections, the results of this interplay on voters should not be assumed to be defined by State borders.1 Any positive effects of the Townsville stadium deal, for example, may have been limited to voters in the Townsville region and not spread across Queensland (QLD) (supporters of the Brisbane Broncos and Gold Coast Titans may have been indifferent or even hostile to the attention given to the Cowboys). Alternatively, the positive effects may have extended across the border to rugby league–loving voters in New South Wales (NSW).

These may seem obvious points; however, election commentary in Australia often implicitly assumes that political events that occur within a particular State affect patterns of voting only of that State, that the effects stop at the borders, and that the resulting State by State outcomes matter for the results of House of Representative elections. These assumptions are, for example, found in the commentaries on individual States in previous edited collections on federal elections since the late 1990s. There is little evidence for or against these assumptions. No systematic comparative testing of the impact of State voting patterns on federal elections has been conducted during the last two decades.

This chapter attempts to re-open the question of whether States matter for Australian federal elections, by focusing on the 2016 election. Since Senate elections are covered elsewhere in this book (Green, Chapter 8), this chapter focuses on the contest for House of Representative seats. The chapter also primarily focuses on the contest for the Treasury benches between Labor and the Coalition, with some attention paid to the Greens, the only other party to stand candidates in all 150 House of Representative seats in 2016. We begin by reviewing the development of debates in Australian political science over the impact of State voting on federal elections. The chapter then outlines four potential factors that could produce State voting variations. We use three of these factors to identify the States that should have been expected to produce better and worse results for the Coalition and Labor. The chapter presents the results of a detailed analysis of State-by-State voting patterns in the 2013 and 2016 House of Representative elections. These results suggest that the States had a modest impact on the outcome of the election, but that electorate level effects were stronger. None of the possible explanations for State differences in 2016 works particularly well. The chapter concludes with some reflections on the need for future research in this area.

One election or eight? The Australian debate

Until the 1970s, Australian federal elections were overwhelmingly seen as national events. The view that the States might have an effect on federal election outcomes was largely ignored or dismissed (Sharman 1975: 16–18). This view became much more contentious from the late 1970s to the late 1990s. Debate over the impact of the States was sparked by a broader discussion of the concepts and methods of analysis employed by Malcolm Mackerras, then Australia’s leading psephologist (see Goot 2016). Reviewing two-party preferred results for the 1977 federal election, Mackerras (1978a: 135) concluded that ‘one must be impressed by the similarity of response regardless of which part of Australia the voter inhabited. State differences seem to me to be negligible, not important’. Australia, he argued earlier in the same article, ‘is pretty much one nation electorally’, a phrase he repeated in response his critics later in the same year (ibid.: 133; 1978b: 335).

In a rejoinder, Campbell Sharman criticised Mackerras’s focus on the two-party preferred vote, arguing that once the magnitude and direction of State variations in first preference votes were taken into account, the States clearly did matter (1978: 338–39). It was a matter of ‘perception and judgement’ as to whether the effects of the States constituted a ‘glass … half empty or half full’. Several years later, Owen Hughes sought greater precision on the issue of State variation versus national uniformity via a series of tests comparing standard deviations in the vote. The result, he claimed, was that the national uniformity glass was ‘shown to be empty’ (1984: 116).

Clive Bean and David Butler weighed into the debate in 1991. Using two-party preferred figures, they reviewed variations in results and swings for all federal elections from 1961 to 1990, concluding that these elections demonstrated a ‘broad [national] uniformity’ (1991: 135). Sharman’s rejoinder accused Bean and Butler of ‘trivialising the analysis of voting patterns’, ‘deterring further research on an important topic and masking a host of issues that deserve analysis’ (Sharman 1991: 346).

One researcher was not deterred. In 1997, Christian Leithner produced the most extensive and sophisticated research on the topic, drawing on analysis of variance techniques applied to American elections (see Kawato 1987). Leithner separated national-, State- and electorate-level variance components in federal elections from 1900 to 1990, grouped by decade. On the impact of States, he concluded that ‘[a]t no time—not even in the 1930s and 1940s, when state effects were strongest—could House of Representative elections be regarded as “state-based” events’ (1997: 219). Equally, voting patterns in federal elections were not the result of nationwide effects. Local electorates were most important: ‘[T]he constituency component of variance has, throughout the twentieth century and as much today as in the past, dwarfed the state and national components’ (ibid.).

Just when Leithner seemed to have settled the debate—there have been no more recent systematic attempts to assess the impact of the States on voting in federal elections—the edited federal election book series began to include chapters on individual States, a practice that persisted until the volume on the 2013 federal election (Bean et al. 1998; Jaensch with Miragliotta and Wear 2015). While these chapters contain interesting contextual material, they assume an importance for the States that Leithner had effectively debunked. Moreover, the practice of various authors focusing on different individual States ran the risk of ad hoc and post hoc explanations of apparent variations in voting across the States. Events in a particular State appeared important only because voters in that State later voted in a particular way, while similar events in other States were implicitly deemed unimportant because voters had voted in a different way.

To avoid these traps, the rest of this chapter will identify some general explanations of possible inter-State variations in voting at the 2016 federal election, and then critically test these using systematic electorate-level data.

Why might the States have mattered in 2016?

International comparative findings on regional voting, along with previous Australian studies, suggest four factors that may have caused the States to have an impact on voting in the 2016 federal election: socioeconomic and/or ethnolinguistic diversity across States; differences in State economic performance; the partisan complexion and popularity of State governments; and direct campaign appeals to voters in specific States.

Socioeconomic and/or ethnolinguistic diversity is unlikely to have been a factor in 2016. Compared with other federations and countries with strong regional identities, Australia is demographically uniform. While the States have some minor peculiarities (see Aroney, Prasser and Taylor 2012), they lack the strong differences associated with regional impacts on national elections. In addition, there is little evidence that any State-based demographic variations that do exist translate into the lasting communal commitments to particular political parties that are found in some other parts of the world (Johnston and Pattie 2006: 83–84). Instead, support for different parties across the Australian States tends to cluster around a national mean (Sharman and Moon 2003: 241–43; Smith 2001: 284–87). As Table 9.1 shows, the average Labor vote at State elections over the past 20 years has varied by just 5.5 per cent, from a low in NSW (36.7 per cent) to a high in Victoria (VIC) (42.2 per cent). The Coalition parties (in their various State guises) have registered a larger 11.5 per cent range; however, if the small Territories are excluded, this is reduced to 5.7 per cent, from Tasmania’s (TAS) low (37.5 per cent) to VIC’s high (43.2 per cent). Table 9.1 also demonstrates that the average differences between the major parties in each State are equally small. There is little indication that voters in any of the States think ‘We are a Labor State’ or ‘We vote Coalition here’.

The second possible explanation for State effects is economic. Given the centrality of economic management to Australian federal elections (Bean and McAllister 2015: 418–21), and the variable economic performance of different Australian States, we might expect voters in States with weaker economic performance to be more likely to vote against the party in government than voters in States experiencing a strong economy (Painter 1993: 136; Johnston and Pattie 2006: 86). This explanation rests on the assumption that significant numbers of voters in each State vote according to their retrospective judgements of the government of the day’s economic performance (Fiorina 1981).

Table 9.1. Average Coalition and Labor first preference votes in lower house State and Territory elections, July 1996 – June 2016 (percentages)

Coalition

Labor

QLD

38.0

41.2

NSW

40.4

36.7

VIC

43.2

42.2

TAS

37.5

42.0

SA

40.1

38.0

WA

43.1

36.8

NT

46.4

42.1

ACT

34.9

38.6

Range

11.5

5.5

Source. Calculated by authors from University of Western Australia n.d.

A third explanation has to do with the impact of State governments. One version of this explanation is that some voters vote in federal elections to gain some protection against the party in government in their State (Painter 1993: 137). Examining the first five elections of the Hawke–Keating era, Martin Painter found that ‘support for a party [at a federal election] is higher where that party is in opposition in the state, and lower where it is in government’ (1993: 135). A variation in this explanation factors in the popularity of State governments. Examining the 1990 federal election, Bean and Butler (1991: 128) observed pro-Labor swings in the three States with popular Labor or unpopular Coalition State governments and anti-Labor swings in the three States with unpopular Labor governments.

Both versions of the State government explanation take the fortunes of the parties contesting federal elections out of their own hands. The same cannot be said for the fourth explanation. The federal parties determine their own policy emphases and choose whether to target specific policies aimed at voters in particular States and regions. Targeted campaigning may reinforce existing patterns of State support; however, comparative experience suggests that parties can also strategically craft their campaign messages to make gains in territory held by their opponents (Johnston and Pattie 2006: 87).

What would each of the last three explanations predict about the pattern of votes across States in the 2016 federal election? Table 9.2 provides some relevant data about economic performance and government characteristics for each of the States. The economic data include State per capita domestic product, economic growth and unemployment. Taken together, these economic indicators suggest that Coalition support was likely to suffer most in TAS, SA and QLD, where gross state product and economic growth were relatively low, and unemployment relatively high. The strength of the NSW, NT and ACT economies would suggest comparatively good results for the Coalition in those jurisdictions. WA and VIC presented more mixed cases. The Western Australian economy was rapidly slowing following the end of the decade-long resources boom, although its gross state product and unemployment rates were still relatively good. The Victorian economy, by contrast, was picking up after a period of relatively poor economic performance. If voters in those States were aware of the economic trends, VIC should have produced a stronger result for the Coalition than WA.

The State government indicators presented a somewhat different set of predictions. On Painter’s incumbency measure, voters in QLD, VIC, SA and the ACT may have wanted to balance their Labor governments with a Coalition vote at the 2016 federal election, while federal Labor would have benefited from the Coalition governments in NSW, TAS, WA and the NT. The final four columns of Table 9.2 allow comparison of the support for State governments in the first half of 2016 against the votes they gained at their most recent election victories. Taking popularity into account confirms the expectation that the Coalition would suffer greater-than-average vote losses in WA and TAS, where the respective governments of Colin Barnett and Will Hodgman were unpopular. By contrast, Mike Baird’s Coalition government in NSW continued to hold a commanding lead over the Opposition in the first half of 2016, indicating that Baird’s incumbency may not affect the federal Coalition vote. QLD and VIC, the two States with the most recent changes of government, retained much the same even balance of party support as at the time of their previous elections, suggesting no moderation either way of the disadvantage that federal Labor might have accrued due to the incumbency of the governments of Annastacia Palaszczuk and Daniel Andrews.

Table 9.2. The States and Territories in 2016

State economic factors

State government factors

GSP per capita 2014–15 ($000)*

Change in GSP per capita 2015–16 (%)*

Unemployment rate, June 2016 (%)**

Party in office

Length of party incumbency

LNP/Lib primary vote at most recent State or Territory election (%)

ALP primary vote at most recent State or Territory election (%)

2016 LNP/Lib intended State primary vote (%)***

2016 ALP intended State primary vote (%)***

QLD

65.4

2.0

6.5

ALP

1 year 5 months

41.3

37.5

40

36

NSW

69.3

3.5

5.3

LNP

5 years 3 months

45.6

34.1

46

28

VIC

62.3

3.3

5.7

ALP

1 year 7 months

42.0

38.1

40

40

TAS

50.3

1.3

6.6

Lib

2 years 2 months

51.2

27.3

40

34

SA

59.4

1.9

7.0

ALP

14 years 3 months

44.9

35.8

30

26

WA

98.0

1.9

5.7

LNP

7 years 9 months

53.1

33.1

42

35

NT

96.9

2.7

3.7

Ctry Lib

3 years 10 months

50.6

36.5

N/A

N/A

ACT

92.2

3.4

3.6

ALP

14 years 7 months

38.9

38.9

N/A

N/A

Source. Created by authors. Sources of specific data are as follows:

* GSP = gross state product (Australian Bureau of Statistics (ABS) 2016a)

** Seasonally adjusted figures (ABS 2016b)

*** Average of two State by State Morgan Polls conducted 25–28 March and 20–22 May 2016 (Roy Morgan Research 2016a, 2016b)

The popularity of Jay Weatherill’s South Australian government in the first half of 2016 is difficult to gauge. On the one hand, Labor surprisingly won a 2014 by-election in a former Liberal-cum-Independent seat that took the government from minority to majority status. On the other hand, by 2016 both Labor and the Liberals had lost significant ground at State level, particularly to the Nick Xenophon Team. No published polls were conducted in the first half of 2016 for the NT, governed by the County Liberal Party, or the ACT, governed by Labor with the support of the Greens. Labor had a landslide victory in the NT election in August, and retained minority government at the ACT election in October, suggesting contrasting levels of support for the territory governments in the first half of 2016 that might both have made the federal competition more difficult for the Coalition.

The final explanation of State differences concerns the parties’ campaign appeals to particular States. The parties are not the only organisations to campaign in particular electorates or States in federal elections (see Halpin and Fraussen, Chapter 17, this volume; Vromen, Chapter 18, this volume); however, they are the most prominent and ubiquitous campaign organisations. If campaigns have an effect on State-by-State voting patterns, we would expect the effects of party campaigns to be visible. To our knowledge, the possible influence of party appeals has not previously been systematically explored in Australia. The approach we take here is twofold: first, to look at how much time the major party leaders spent in different States during the election campaign; and second, to compare the campaign policies directed at particular States. Our rules of thumb are that, other things being equal, leaders who spend more time in a State, and parties that address more specific campaign promises towards a State, are likely to gain an advantage in that State. They may not win the most votes or seats in that State, but they are likely to do better in those States than in States that they neglect.

On the first measure, Turnbull and Shorten spent roughly as much time in four States—QLD, SA, TAS and the ACT—as their proportions of House of Representative seats would dictate (see Table 9.3). They both visited VIC less often than might have been expected, given the number of seats at stake there. Perhaps this reflected VIC’s electoral geography, which meant that very few seats were realistically up for grabs across the State. Equally, they both visited the NT more than its two seats might have deserved. The main differences between the two leaders occurred in Shorten’s relatively high presence in NSW electorates and relatively low presence in WA. This suggests that Labor was hoping to win seats in NSW, but had less optimism about picking up WA seats.

Table 9.3. The major party leaders’ campaign visits to States and Territories

Turnbull

Shorten

Seats (%)

Visits (%)

Visits (%)

NSW

31

29

36

VIC

25

18

18

QLD

20

22

20

WA

11

13

7

SA

7

10

8

TAS

3

4

5

NT

1

3

4

ACT

1

1

2

Total

100

100

100

Source. Compiled by authors from data in Doran and Liddy (2016).

Examination of the policy announcements made during the campaign by the Coalition, Labor and the Greens indicates that, while States have a place in the party’s strategic campaigning, they do not occupy a central place. We categorised all policy announcements of the three parties posted on their federal election websites as either national or regional. Policies were categorised as having a national focus if they were addressed generically to Australian voters, or described initiatives targeted at the country as a whole, such as the Coalition’s general ‘jobs and growth’ policies or Labor’s ‘save Medicare’ policies. We included policies as national rather than regional, even if they included specific locations simply to indicate where wider policies would apply. Thus, for example, the Coalition’s ‘funding for facilities to support children needing palliative care, such as Hummingbird House in Queensland, Bear Cottage in New South Wales, and Very Special Kids in Victoria’ (LPA 2016c) and the Greens’ commitment to ‘stop runaway tree-clearing across Australia, including in Queensland, New South Wales and Western Australia’ (Australian Greens 2016b) were all classified as national policy initiatives. In some cases, such as Labor’s commitment to fast-track ‘national infrastructure projects’, these indicative lists seemed carefully crafted to include an example from every State (ALP 2016c).

Policies were classified as regional only if they included commitments exclusively targeted at particular subnational locations. This included policies that had a national focus, but also explicit regional elements, such as the Coalition’s policy on the Great Barrier Reef, which referred to the Reef’s value for ‘millions of Australians’ and the ‘Australian economy’, but also included promises of specific projects in local north Queensland communities (LPA 2016d).

The majority of policy announcements in the 2016 federal election were geographically generic. Of 36 Coalition policy statements during the campaign, 26 were national and 10 regionally specific (LPA 2016a). Roughly the same proportion of Labor’s ‘100 Positive Policies’—25 out of the 100—included a specific regional focus (ALP 2016a). The regional targets of the Australian Greens’ policies were more difficult to quantify; however, Greens’ policies addressing specific regions seemed to be even less common than those of the major parties (Australian Greens 2016a).

The targets of regional policy announcements in 2016 were rarely individual States. In some cases, the targets covered wider regions, as in Labor’s ‘Northern Australia—A Tourism Powerhouse’, which promised initiatives for parts of QLD, the NT and WA (ALP 2016b). More commonly, regional policies concerned infrastructure or environmental projects targeted at specific locations within States, such as the Coalition’s ‘City Deal for Western Sydney’ (LPA 2016e). To the extent that the parties addressed voters as citizens who had State-wide loyalties and interests, they mostly did so via tailored versions of their national policy announcements. These documents were based on generic templates, but included claims about the impact of national policies on the particular State to which they were targeted, using State-based data on economic indicators, education participation and the like (see e.g. ALP 2016c).

In numerical terms, the Coalition paid particular policy attention to QLD, which was the target of five of its 10 regional policy announcements. Labor also paid more attention to QLD than the other States, including it in nine of 25 announcements. Both parties spread their remaining State policy appeals relatively evenly among the States. Unlike the Coalition, Labor released policies aimed directly at NT and ACT voters. Counting policy announcements says something about the emphases of the competing parties; however, it ignores the content and significance of particular announcements. In SA, for example, the Coalition and Labor presented quite similar plans for new defence-related shipbuilding, transport improvements and steel manufacture (LPA 2016f; ALP 2016c). Given the State’s precarious economic position, these plans were likely to be more salient to SA voters than any number of other regionally based policies. By the same token, the fact that the two major parties closely matched each other’s policies for SA probably effectively neutralised any electoral advantage either might have gained.

The points made above about campaigning and the States suggest great caution about predicting their likely effects on voting patterns. Specific appeals to voters as members of one State or another were relatively rare. National policy appeals with differential State effects, along with policy appeals to cross-State regions such as ‘northern Australia’ and sub-State regions within States such as ‘western Sydney’ were all likely to muddy the effects of the State policy appeals made by parties. Labor possibly gained some advantage in NSW from Shorten’s greater campaign presence in that State. Labor’s more visible policies for the ACT and the NT relative to the Coalition might also have helped its cause. On the other hand, Shorten’s relative absence from WA may have cost Labor votes.

Table 9.4. Potential State effects on Coalition voting at the 2016 federal election

Economy

State government party

State government popularity

Federal party campaigning

NSW

+

+

VIC

=

+

=

=

QLD

+

=

=

WA

=

+

SA

+

=

=

TAS

=

NT

+

=

ACT

+

+

=

=

Note. ‘+’ likely to be positive for the Coalition; ‘–’ likely to be negative for the Coalition; ‘=’ likely to be neutral for the Coalition.

Source. Authors.

The potential effects of State-based factors on voting in the 2016 federal election are summarised in Table 9.4. It is immediately clear that economic strength, State politics and federal campaigning pointed in inconsistent directions for each State. TAS is the State in which the factors pointed most consistently in one direction—trouble for the federal Coalition. The other States were more mixed. New South Wales voters, for example, may have been pulled in one direction by the State’s economic strength and the popularity of the State government, but in another by the State government being Coalition and by the Labor leader’s campaign presence. These inconsistencies point to the dangers of focusing on individual States for explanations of federal election results. Factors that appear important in producing a particular set of election results in one State are likely to be similar to those in another State in which the result was different.

State voting patterns at the 2016 federal election

Arguments about whether or not federal election results and swings are uniform or vary between States are often intertwined with arguments about whether to use two-party preferred votes or first preference votes as the relevant measure (see Goot 2016). As Sharman pointed out almost four decades ago (1978: 337), the two measures may give very different impressions of the election result. In this section, we present both two-party preferred and first preference results, first comparing the mean results for each State and then augmenting these with some measures of the dispersion of the vote across electorates within each State. The results discussed in the following paragraphs have been calculated using official Australian Electoral Commission (AEC) results (AEC 2016).

Table 9.5. House of Representatives two-party preferred vote by State and Territory, 2016

LNP

ALP

LNP swing

NSW

50.5

49.5

–3.8

VIC

48.2

51.8

–1.6

QLD

54.1

45.9

–2.9

WA

54.7

45.3

–3.6

SA

47.7

52.3

–4.6

TAS

42.6

57.4

–6.1

NT

42.9

57.1

–7.4

ACT

38.9

61.1

–1.2

Total

50.4

49.6

–3.1

Source. Compiled by authors using data from AEC (2016).

Table 9.5 shows that, on a two-party preferred basis, the Coalition achieved its best results in WA (54.7 per cent) and QLD (54.1 per cent), and its worst results in the ACT (38.9 per cent), TAS (42.6 per cent) and the NT (42.9 per cent), a range of 15.8 per cent (WA to the ACT). If the ACT’s two seats are ignored for the moment, the range falls to 12.1 per cent (WA to TAS).

The two-party preferred swings in every State were towards Labor; however, they also varied considerably, from 1.2 per cent in the ACT to 7.4 per cent in the NT. Excluding the small Territories for the moment, the swing ranged between 6.1 per cent in TAS to 1.6 per cent in VIC. In three States (NSW, QLD and WA), the result of the swing was to reduce the Coalition’s majority across the State, while the results in VIC, TAS and the NT increased slim two-party preferred majorities achieved by Labor at the 2013 federal election. The ACT swing built on an already strong Labor base from 2013. Thus, the only State to move from having a notional Coalition majority in 2013 to a notional Labor majority in 2016 was SA.

As Table 9.6 shows, however, the two-party preferred result in SA hides a rather startling loss of votes by all three of the parties that contested all House of Representative seats across Australia. Calculating the size of the swings as a proportion of their 2013 primary votes in SA, the Coalition lost 21.1 per cent of its primary vote, Labor lost 11.8 per cent and the Greens 25.3 per cent. Without good survey research, it is hard to know exactly where all of these votes went, but many of them appear to have transferred to the Nick Xenophon Team, which ran candidates in every SA seat at its first House of Representative election and secured 21.3 per cent of the primary vote in that State (see also Raue, Chapter 7, this volume; Kefford, Chapter 15, this volume).

In all the other States, the first preference swings involve losses for the Coalition and increased support for both Labor and the Greens. Labor’s gains ranged from a modest 0.8 per cent in VIC to 3.7 per cent in WA, while the Greens’ gains varied from 1.0 per cent in NSW up to 2.6 per cent in QLD. Even the rejuvenation of the Pauline Hanson’s One Nation Party as a federal political force in 2016—particularly in QLD, where it achieved 5.5 per cent of the State-wide first preference vote—did not disrupt this basic pattern of a leftward shift in first preference votes from the Coalition towards Labor and the Greens. This was possibly because the entry of the Palmer United Party had already shaken voters away from the Coalition and Labor at the 2013 federal election in States such as QLD. With Palmer’s party not contesting the 2016 federal election, many of those voters may simply have switched to Pauline Hanson’s One Nation.

Table 9.6. House of Representatives first preference vote by State and Territory, 2016

LNP vote

LNP swing

ALP vote

ALP swing

Green vote

Green swing

NSW

42.3

–5.0

36.9

2.4

8.9

1.0

VIC

41.8

–0.9

35.6

0.8

13.1

2.3

QLD

43.2

–2.5

30.9

1.1

8.8

2.6

WA

48.7

–2.5

32.4

3.7

12.1

2.3

SA

35.1

–9.4

31.5

–4.2

6.2

–2.1

TAS

35.4

–4.8

37.9

3.1

10.2

1.9

NT

33.2

–8.4

40.4

3.0

9.1

1.2

ACT

34.6

–0.1

44.3

1.3

15.1

1.7

Total

42.0

–4.1

34.7

1.3

10.2

1.6

Source. Compiled by authors using data from AEC (2016).

A party’s electoral competitiveness in any State relies not just on its total State vote, but also on how its vote is distributed across the State’s electorates. Figures 9.1 and 9.2 provide data on the distribution of first preference votes for the Coalition, Labor and the Greens in each electorate in the different States in 2013 and 2016. The data are presented as Tukey box plots. They summarise four pieces of information for each party in each State. First, the horizontal lines within the shaded boxes represent the median electorate percentage vote for the relevant party in each State. The higher the line, the better a party has done overall in a State. Second, the boxes themselves show the party’s middle results in a State—the quartile immediately above and below the median. Third, the vertical whisker lines show the ranges of the party’s remaining electorate results above and below the middle quartiles, up to a statistically determined range (1.5 times the height of the box). Taller boxes and longer whiskers indicate more uneven results for a party within a State, while shorter boxes and whiskers, indicate a party’s vote is relatively evenly spread across electorates in a State. Finally, the dots indicate outlier electorates; that is, electorates in which a party did much better or worse than its middle results suggest.

Figure 9.1. Distribution of first preference votes, 2016 federal election (Tukey box plots)

Figure 9.1. Distribution of first preference votes, 2016 federal election (Tukey box plots)

Source. Constructed by authors using data from AEC (2016).

Figure 9.2. Distribution of first preference votes, 2013 federal election (Tukey box plots)

Figure 9.2. Distribution of first preference votes, 2013 federal election (Tukey box plots)

Source. Constructed by authors using data from AEC (2016).

What do these plots suggest about the electoral competition between parties in the 2016 election and how this competition changed from 2013? The two Territories and TAS contain too few seats for detailed analysis, although the plots for all three indicate a degree of homogeneity in party support in 2013 and 2016. The remaining five plots reflect Labor’s generally improved position across all States in 2016 compared with 2013. Labor’s median vote is higher and its boxes and whiskers extend higher than in 2013.

There are clear differences in the voting patterns for WA and QLD, on the one hand, and NSW, VIC and SA, on the other. In WA and QLD, the range of votes across electorates in both 2013 and 2016 was relatively small compared with the more dispersed pattern in the other three States. Not only do the Coalition parties in WA and QLD enjoy higher median electorate votes than their counterparts in the other States, their votes in particular electorates are reasonably tightly clustered around those higher medians. The top three quarters of all Coalition electorate results in WA and QLD in 2013 and 2016 were better than all but the top quarter of Labor’s results in the same States. In the other States, Labor’s results were more competitive compared with the Coalition’s, particularly in 2016. These patterns suggest that the same improvements in Labor’s State-wide votes will produce different results in WA and QLD and in the other States. In the latter States, every small improvement in Labor’s vote is likely to reap some rewards, whereas, in WA and QLD, large swings need to be achieved for Labor to hope to gain any seats.

A final point worth noting from Figures 9.1 and 9.2 concerns the Greens. Comparison of the boxes and whiskers between 2013 and 2016 suggests that the Greens did not manage to improve their general electoral competitiveness at the 2016 federal election. Their votes generally remained clustered in the same bands as in 2013. Nonetheless, they managed to achieve more positive outlier results—results in which their vote was much better than the expected range—in 2016 than they did in 2013, particularly in VIC. While the party has found it difficult to create broad momentum across any State, the 2016 federal election results identified specific electorates in which the party could perform well above expectations (see also Raue, Chapter 7, this volume).

Figure 9.3. Distribution of two-party preferred swing, 2013–16 (Tukey box plots)

Figure 9.3. Distribution of two-party preferred swing, 2013–16 (Tukey box plots)

Source. Constructed by authors using data from AEC (2016).

How uniform was the swing within and between States in 2016? Figure 9.3 presents more box plots for each State, this time summarising variation in the two-party preferred swing. It shows two patterns. In TAS, SA, the NT and the ACT, the swings in every seat were against the Coalition, albeit to varying degrees. In the remaining States, most seats swung to Labor, but at least some recorded shifts to the Coalition. In VIC, one quarter of seats did so.

Consideration of the interplay between individual electorates and overall State results raises the fundamental and much discussed question of the relative contribution of electorate-level factors and State-wide forces to variations in federal election results. In order to gain an approximate measure of the relative importance of States versus electorates, we conducted a principal components analysis on different aspects of the vote. Because we were focused on the 2016 election results, rather than long-term trends, we did not attempt to replicate Leithner’s (1997) approach of grouping elections by decade in order to test for national-level effects. Thus, our results are concerned only with the 150 electorates, and eight States and Territories.

Table 9.7. Results of principal components analysis of State- and electorate-level contribution to variations in the vote, 2013 and 2016

2013

2016

Labor

State

13.6

9.6

Electorate

86.4

90.4

Coalition

State

12.0

8.3

Electorate

88.0

91.7

Greens

State

21.2

25.6

Electorate

78.8

74.4

Two-party preferred swing, 2013–16

State

29.5

Electorate

70.5

Source. Compiled by authors using data from AEC (2016).

The results in Table 9.7 suggest that electorate-level variations are more important than State variations for both the results of particular elections and for swings between them. State variations explained less than 10 per cent of the major parties’ first preference votes in 2016. State variations accounted for a little more of the variation in votes in 2013 but, in both elections, they were dwarfed by electorate variations. Interestingly, while electorate variations were also most important for the Greens, State variations had a stronger impact than for the major parties. Perhaps the explanation is that the Greens are primarily associated with the Senate rather than the House of Representatives, so that the Greens’ House of Representative vote is influenced by the varying profiles and performance of the Green Senators from different States. State variations also lay behind some of the two-party preferred swing recorded between 2013 and 2016, although once again electorate variations seemed a stronger influence by some margin. These results suggest that, at the very least, State differences should not be dismissed out of hand as a factor in the 2016 federal election.

Conclusion

At the 2016 federal election, all the States registered two-party preferred swings against the Coalition. This apparent homogeneity masks a range of variations between them. These included the different sizes of the swing between States, the uniformity of the swing within them, the peculiarity of SA’s first preference swings, and inter-State differences in both the two-party preferred and first preference outcomes once the dust had settled.

These differences are difficult to explain in a coherent and parsimonious way. This is partly because Australian political science has not paid attention in recent decades to systematic State-based differences in federal election results and their possible causes. The potential causes of variation outlined earlier in this chapter can only be applied in a loose way to the results. The comparatively large swing against the Coalition in TAS could be seen as resulting from a combination of the State’s economic woes and its unpopular State Coalition government. The much smaller Victorian swing might be seen as a result of that State’s improving economy and the presence of a State Labor government. The other, less consistent State cases are impossible to fit into such a framework. None of the suggested factors—State economic fortunes, State politics or federal campaigning—appears to provide sufficient explanation on its own. Eight outcomes from one election is too small a number to develop a solid understanding of how the factors might interact, or which other factors might be missing from the analysis. More analysis needs to be done to explain the sorts of similarities and differences between States that this chapter has identified.

References

Aroney, Nicholas, Scott Prasser and Alison Taylor. 2012. ‘Federal diversity in Australia: A counter-narrative’. In Gabrielle Appleby, Nicholas Aroney and Thomas John (eds), The Future of Australian Federalism: Comparative and Interdisciplinary Perspectives, Cambridge: Cambridge University Press, pp. 272–300. doi.org/10.1017/CBO9780511902550.020

Australian Broadcasting Corporation (ABC). 2016. ‘Election 2016: Coalition matches Labor promise to fund Townsville sports stadium’. ABC News, 13 June. Available at: www.abc.net.au/news/2016-06-13/coalition-to-match-labor-funding-promise-for-townsville-stadium/7504276

Australian Bureau of Statistics (ABS). 2016a. 5220.0 — Australian National Accounts: State Accounts, 2015–16. Available at: www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/5220.02015-16?OpenDocument

——. 2016b. 6202.0—Labour Force, Australia, June 2016. Available at: www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/6202.0Jun%202016?OpenDocument

Australian Electoral Commission (AEC). 2016. ‘House of Representatives results’. Available at: results.aec.gov.au/20499/Website/HouseResultsMenu-20499.htm

Australian Greens. 2016a. ‘Our complete policy platform’. Available at: greens.org.au/policy-platform

——. 2016b. ‘Protecting our precious places’. Available at: greens.org.au/places

Australian Labor Party (ALP). 2015. ‘Labor will deliver a New Townsville stadium’. Media release: 4 November. Available at: www.billshorten.com.au/labor-will-deliver-a-new-townsville-stadium

——. 2016a. ‘100 positive policies’. Originally available at: www.100positivepolicies.org.au/ (accessed 10 August 2016, site discontinued).

——. 2016b. ‘Northern Australia—A tourism powerhouse’. Originally available at: www.100positivepolicies.org.au/northern_australia_a_tourism_powerhouse (accessed 10 August 2016, site discontinued).

——. 2016c. ‘Labor’s positive plan for South Australia’. Originally available at: www.100positivepolicies.org.au/labors_positive_plan_for_south_australia (accessed 10 August 2016, site discontinued).

Bean, Clive and David Butler. 1991. ‘Uniformity in Australian electoral patterns: The 1990 election in perspective’. Australian Journal of Political Science 26(1): 127–36. doi.org/10.1080/00323269108402140

Bean, Clive and Ian McAllister. 2015. ‘Documenting the inevitable: Voting behaviour at the 2013 Australian election’. In Carol Johnson and John Wanna with Hsu-Ann Lee (eds), Abbott’s Gambit: The 2013 Australian Federal Election. Canberra: ANU Press, pp. 411–24. doi.org/10.22459/AG.01.2015.24

Bean, Clive, Marian Simms, Scott Bennett and John Warhurst (eds). 1998. The Politics of Retribution: The 1996 Federal Election. Sydney: Allen & Unwin.

Doran, Matthew and Matt Liddy. 2016. ‘Election 2016: 5 things we’ve learned from where the leaders have been’. ABC News, 1 July. Available at: www.abc.net.au/news/2016-06-29/election-2016-turnbull-shorten-maps/7552114

Fiorina, Morris. 1981. Retrospective Voting in American Elections. New Haven: Yale University Press.

Goot, Murray. 2016. ‘The transformation of Australian electoral analysis: The two-party preferred vote—origins, impacts and critics’. Australian Journal of Politics and History 62(1): 59–86. doi.org/10.1111/ajph.12208

Hughes, Owen. 1984. ‘Uniform swing revisited: Further comments on Mackerras’. Politics 19(2): 111–18.

Jaensch, Dean, with Narelle Miragliotta and Rae Wear. 2015. ‘The electoral geography of the 2013 Election: Voting patterns in the States and regions for the lower House’. In Carol Johnson and John Wanna with Hsu-Ann Lee (eds), Abbott’s Gambit: The 2013 Australian Federal Election. Canberra: ANU Press, pp. 225–47. doi.org/10.22459/AG.01.2015.14

Johnston, Ron and Charles Pattie. 2006. Putting Voters in Their Place: Geography and Elections in Great Britain. Oxford: Oxford University Press. doi.org/10.1093/acprof:oso/9780199268047.001.0001

Kawato, Sadafumi. 1987. ‘Nationalization and partisan realignment in congressional elections’. American Political Science Review 81(4): 1236–250.

Leithner, Christian. 1997. ‘Electoral nationalisation, dealignment and realignment: Australia and the US, 1900–88. Australian Journal of Political Science 32(2): 205–22. doi.org/10.1080/10361149750904

Liberal Party of Australia (LPA). 2016a. ‘2016 Election policies’. Available at: www.liberal.org.au/our-policies

——. 2016b. ‘The Coalition’s policy to revitalise jobs and growth in Townsville’. Available at: www.liberal.org.au/coalitions-policy-revitalise-jobs-and-growth-townsville

——. 2016c. ‘The Coalition’s policy to help families with diabetes’. Available at: www.liberal.org.au/coalitions-policy-help-families-diabetes

——. 2016d. ‘The Coalition’s policy to invest in our Great Barrier Reef’. Available at: www.liberal.org.au/coalitions-policy-invest-our-great-barrier-reef

——. 2016e. ‘The Coalition’s policy for smart cities’. Available at: www.liberal.org.au/coalitions-policy-smart-cities

——. 2016f. ‘The Coalition’s policy for jobs and growth in South Australia’. Available at: www.liberal.org.au/coalitions-policy-jobs-and-growth-south-australia

Ludlow, Mark. 2016. ‘Election 2016—Townsville Stadium a “Boondoggle”’. Australian Financial Review, 10 May. Available at: www.afr.com/news/politics/election-2016--townsville-stadium-a-boondoggle-20160510-goqjsp

Mackerras, Malcolm. 1978a. ‘No change: Analysis of the 1977 election’. Politics 13(1): 131–38.

——. 1978b. ‘On Elaine Thompson and Tom Wheelwright’. Politics 13(2): 334–36. doi.org/10.1080/00323267808401678

Painter, Martin. 1993. ‘March 13 1993: National Poll or Federal Election?’ In Kathy Dempsey (ed), Beyond the Headlines: Australia and the World. Sydney: Public Affairs Research Centre, pp. 133–44.

Peel, Charlie. 2016. ‘Shorten reaffirms stadium pledge and urges Townsville to pressure all sides of politics’. Townsville Bulletin, 23 April. Available at: www.townsvillebulletin.com.au/news/shorten-reaffirms-stadium-pledge-and-urges-townsville-to-pressure-all-sides-of-politics/news-story/e32becc5a4378d0c5b69c5b6a7815569

Roy Morgan Research. 2016a. ‘ALP increases support in all states’. Article No. 6737, 6 April. Available at: www.roymorgan.com/findings/6737-morgan-poll-state-voting-intention-april-2016-201604010044

——. 2016b. ‘L-NP in front in NSW & WA and ALP well in front in Victoria but parties dead level in Queensland after LNP elect new leader Tim Nicholls’. Article No. 6836, 1 June. Available at: www.roymorgan.com/findings/6836-morgan-poll-state-voting-intention-june-2016-201606010708

Sharman, Campbell. 1975. ‘Federalism and the study of the Australian political system’. Australian Journal of Politics and History 21(3): 11–24. doi.org/10.1111/j.1467-8497.1975.tb01149.x

——. 1978. ‘Swing and the two party preferred vote: A comment on Mackerras’. Politics 13(2): 336–39. doi.org/10.1080/00323267808401679

——. 1991. ‘Forward to the past: A reply to Bean and Butler’. Australian Journal of Political Science 26(2): 342–47. doi.org/10.1080/00323269108402154

Sharman, Campbell and Jeremy Moon. 2003. ‘One system or nine?’ In Jeremy Moon and Campbell Sharman (eds), Australian Politics and Government: The Commonwealth, the States and the Territories. Melbourne: Cambridge University Press, pp. 239–62. doi.org/10.1017/CBO9780511756061.011

Smith, Rodney. 2001. Australian Political Culture. Sydney: Pearson Education.

University of Western Australia (UWA). n.d. Australian Politics and Elections Database. Available at: www.elections.uwa.edu.au/


1 For stylistic reasons, we propose to avoid the expressions ‘State and Territory’ and ‘States and Territories’. Unless otherwise indicated, references to States in this chapter should be taken to include the Australian Capital Territory (ACT) and the Northern Territory (NT).


Previous Next