Unravelling the Referendum: An analysis of the 2023 Australian Voice to Parliament Referendum outcomes across capital cities

doi:10.21203/rs.3.rs-4069107/v1

Download PDF

Research Article

Unravelling the Referendum: An analysis of the 2023 Australian Voice to Parliament Referendum outcomes across capital cities

https://doi.org/10.21203/rs.3.rs-4069107/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The 2023 Australian Voice to Parliament Referendum presented a pivotal moment in the nation's democratic landscape, aiming to enshrine indigenous voices in the constitutional fabric through the establishment of an Aboriginal and Torres Strait Islander Voice. Despite widespread support for indigenous well-being, the referendum did not secure the necessary approval, prompting extensive analysis of its outcome. This paper employs an ecological approach to scrutinize the referendum's dynamics, exploring six hypotheses derived from public discourse.

Findings reveal multifaceted influences on voting behavior. Economic concerns, exemplified by the cost-of-living crisis, seemingly diverted attention from constitutional reform, potentially swaying votes towards maintaining the status quo. Conversely, culturally diverse communities demonstrated heightened empathy towards indigenous issues, aligning with the yes vote. Lower levels of education correlated with support for the no vote, highlighting the impact of political knowledge on decision-making.

Moreover, religious conservatism and political partisanship emerged as influential factors, with Christian values and party affiliations shaping voting patterns. These findings underscore the complexity of referendum dynamics, emphasizing the importance of effective messaging and understanding diverse socio-political contexts in shaping public opinion.

The defeat of the referendum marks a setback in indigenous relations, prompting critical reflection on messaging strategies and the broader socio-political landscape. This analysis provides a foundational empirical framework for understanding the referendum outcome, offering insights crucial for informed discourse and future democratic endeavours.

Other Political Science

referendum

Voice to parliament

Australia

spatial analysis

Within the machinery of democratic governance, referendums occupy a unique and consequential role, serving to directly engage the voices of the people in critical matters of state. The 2023 Voice to Parliament Referendum aimed to enable a change to the Australian Constitution to recognise the First Peoples of Australia by establishing a body called the Aboriginal and Torres Strait Islander Voice. If passed into law, the ‘Voice’ would legislate a structure that would enable Aboriginal and Torres Strait Islander peoples to provide advice to the government and parliament on issues that were likely to impact on the lives of Australian indigenous people and communities.

The impetus for the referendum began in 2017 when representatives of First Nations people met and produced the Uluru Statement from the Heart, which called for a First Nations Voice enshrined in the Australian Constitution. At the time, the ruling conservative Liberal/ National Party coalition rejected the call outright. However, following the election of the Australian Labor Party in 2022 the new Prime Minister Anthony Albanese announced that Australians would have their say in a referendum to include an Aboriginal and Torres Strait Islander Voice to Parliament.

In Australia, for a constitutional change to be passed, the referendum vote needs to be approved by a ‘double majority’ comprising a national majority of electors from all states and territories, together with a majority of electors in a majority of the states (i.e. at least four of the six states). Historically, attempts to amend the constitution in Australia have resulted in many more rejections than approvals. Since 1901 of the 44 constitutional referendums presented to the Australian voting public, only 7 had reached the required threshold, with the remaining being defeated. Following the October 14, 2023, vote, the Voice to Parliament referendum did not meet the required test and hence was not passed. Of the 17,671,784 enrolled voters, approximately 90 per cent cast a vote with 39.9 per cent casting a yes vote and 60.1 per cent casting a no vote (Biddle, Gray, McAllister, & Qvortrup, 2023)[1].

The outcome of the referendum was in contrast to social surveys that have found that Australian generally support improving the well-being of the country’s indigenous population (Levy & McAllister, 2022; Markham & Sanders, 2020) and this contrast became more stark when social commentary began to unpack the possible reasons for the overwhelming no-vote and the uneven support for the yes-vote. Associated with these debates, commentators began to analyse the results looking at why some areas tended towards a no vote, while others tended towards a yes vote. Spatially, commentators were quick to point out the distinct patterns that appeared to have emerged. Within the capital cities, these included distinct clusters of yes votes within inner and near inner-city locations with no votes dominating outer suburban locations.

Associated with the rudimentary spatial analysis, questions began to be asked that focused on the socio-demographic make-up of the electorate and how different ‘types’ of voters impacted on the outcome. The various questions and explanations were wide-ranging but can be distilled into 6 testable hypotheses.

During the campaign leading up to the referendum, a common theme in the media related to the argument that for many Australian voters, the referendum and what it stood for was not front-of-mind due to everyday issues. This can be labelled the ‘concern with everyday issues rather than the referendum’ hypothesis where it was argued that the ‘voice to Parliament can’t compete with cost-of-living crisis in voters’ minds (Chowdhury, 2023). Prosecuting this argument, commentators suggested that in cases where voters are more concerned about everyday issues such as the cost of living, they may be less concerned with bigger-picture issues and vote to maintain the status-quo (i.e vote no). Furthermore, as suggested by Biddle et al. (2023, p. 60) people were

less inclined to support a change to the constitution that would result in benefits to one group over another, and that people were angry with the government for focusing on what is seen as a niche issue when more pressing issues are being ignored.

Such a hypothesis is aligned with a range of empirical research which has reported that anxiety may stimulate preferences for protective policies (Albertson & Gadarian, 2015) and drive voters to consider their choices more carefully (MacKuen, Marcus, Neuman, & Keele, 2007) or that voters move toward the status quo under times of threat (Bisbee & Honig, 2022).

The second hypothesis that emerged relates to views that ‘multicultural support offers hope for a Yes outcome’ (Gunstone, 2023). This might be referred to as the multi-cultural empathy hypothesis and suggests that multi-cultural communities may have empathy for indigenous issues and the message behind the voice to Parliament and hence are more likely to vote yes. Alluding to the potential empathy impact on the referendum outcome Jakubowicz (2023) noted the support received for the Black Lives Matter movement by members of various ethnic groups arguing that

These events may have heightened the awareness in immigrant communities of the prevalence of racism in Australia. They may also have enhanced empathy for Indigenous people’s struggles, and potentially, support for the Voice (para 24).

Such a view is supported in the academic literature where it has been suggested that viewing an issue through the perspective of ethnic minorities or other disadvantaged groups can reduce the impact of prejudice towards these groups (Galinsky & Moskowitz, 2000) and by extension increase support for policy issues impacting these groups.

A third hypothesis established following the referendum related to a perceived lack of information or knowledge, best expressed by the phrase used by the no-campaign ‘If you don’t know, vote no’. The ‘if you don’t know, vote no’ hypothesis related to the confusion around the intent of the referendum and was summed up by statements such as the ‘voice referendum was too 'complicated' for 'less educated' Australians to understand’ (Collins, 2023). Regarding this argument, it may well be the case, as Crisp (1983) suggests ‘from all that we know about voting behaviour in Australia, it is clear that some of the voters will cast a vote in ignorance of what it is all for—and what seems true for elections seems to apply to constitutional referenda’ (Bennett, 1985, p. 27). Within the political science literature, there is a range of empirical material that has investigated the links between education and political engagement and motivations. A number of researchers have found that the broader education curriculum and school experience can influence political attitudes and awareness and help create politically informed and engaged voters (Boden & Nedeva, 2010; Hillygus, 2005; Mishra, Klein, & Müller, 2023; Schofer, Ramirez, & Meyer, 2021), or provide important social capital and social networks which aid and encourage greater engagement and participation (Evans, Rees, Taylor, & Fox, 2021; Mishra et al., 2023; Putnam, 2000).

The fourth hypothesis relates to headlines such as ‘Voice Referendum: Old-fashioned racism driving 'No' campaign’ (Duffield, 2023). Some posit that the referendum has tapped into ‘a deep well of historical racism’ which has outweighed other considerations (Anderson, Paradies, Langton, Lovett, & Calma, 2023). Indeed, racism was an important issue in the 1967 Referendum where Australians voted to remove references in the Constitution that discriminated against Aboriginal and Torres Strait Islander people. In the aftermath of the 1967 referendum, it was suggested that there was a strong inverse relationship between the percentage of electors agreeing with the proposals and the ratio of Aboriginal to European populations (Mitchell, 1968). As such it was thought that the no vote was influenced by a proximity or contact hypothesis (Bennett, 1985; Ray, 1983) whereby voters living in communities with higher numbers of indigenous people were more likely to vote no.

A fifth possible hypothesis relates to the link between conservative views and political voting behaviour. The ‘conservatism and fear of change’ hypothesis is reflected in the reported links between the no campaign and conservative politics where the media reported ‘Indigenous voice: no campaign’s deep links to conservative Christian politics revealed’ (Butler, 2023). The conservatism and status-quo nexus has been well established in the voting patterns of Australians. During the 1967 referendum, it was argued that the no vote was driven in part by conservatism with rural electoral sub-divisions that were thought to be more conservative, voting against change (Bennett, 1985). Similarly, with reference to climate change policy Colvin and Jotzo (2021) analysed the 2019 Australian Federal election and found that conservative voters were more likely to vote to maintain the status quo on climate change policy rather than vote for change. Within the broader political science literature, researchers have investigated the links between conservatism and voting for the status-quo. For example, researchers including Jost, Sterling, and Stern (2017), Federico and Malka (2018) and Thorisdottir, Jost, Liviatan, and Shrout (2007) have found that typically, people with a strong desire to diminish insecurity and minimize uncertainty tend to be drawn to the political right (i.e. conservatives), which prioritizes stability and hierarchy and increases the likelihood of voting for the status quo. With reference to these links Thorisdottir et al. (2007, p. 179)

[T]here is a special resonance or match between motives to reduce uncertainty and threat, and the two core aspects of right-wing ideology, resistance to change and acceptance of inequality.

The final hypothesis relates to political party alignment. Throughout the referendum campaign, clear political party lines were evident with the conservative coalition largely supporting the no vote, while the ruling Labor party, together with independent and green party members of parliament leaning towards the yes-vote. Opinion surveys leading up to the referendum supported this hypothesis, with voters for the major parties (labor and the collation) falling in line with the stated party views. Reviewing a number of opinion polls (Markham & Sanders, 2020, p. 17) note that

in 2018 and 2019, levels of support for a Voice among committed Coalition voters fell below 50%, likely due to the unsupportive positions of Coalition Prime Ministers.

Interestingly, they also note

this move among Coalition voters could well have been in the opposite direction had their leadership chosen a supportive position on a Voice, at which point a national majority vote in a referendum would have been very likely.

The suggestion of party partisanship in voting outcomes in the referendum should not come as a surprise given the existing literature on the impact of party policy position on the opinions of citizens. Deriving from early work by Campbell, Converse, Miller, and Donald (1960) who referred to political elites / parties as ‘an opinion-forming agency of great importance’ (p. 128), arguments have focused on understanding the ways in which citizens use prompts about the position of their preferred party as an information shortcut to reach an informed opinion (Lupia, 2006, 2018; Sniderman & Stiglitz, 2012) or citizens simply follow the ‘party-line’ of their choosing to remain consistent with their identify and stay loyal to the partisan group (Huddy, Bankert, & Davies, 2018). Although there appears to be significant debate regarding the strength and veracity of such linkages (Slothuus & Bisgaard, 2021), there is a significant body of research literature which supports, to a greater or lesser extent, such opinion forming linkages (Barber & Pope, 2019; Chong & Mullinix, 2019; Leeper & Slothuus, 2014; Peterson, 2019; Slothuus & Bisgaard, 2021; Slothuus & De Vreese, 2010).

While these diverse arguments have represented the mainstream discussion of the referendum outcome, we have only just begun to see an emerging academic literature addressing the referendum outcome. A recent report by Biddle et al. (2023) undertook an analysis using survey data and reported findings that go some way to adding empirical rigour to the public debate. Interestingly, in the context of the arguments set out above, the authors undertook an analysis of outcomes at the electoral district level and found that districts with higher proportions of indigenous population, higher proportion of people born overseas, and districts held by the coalition or green / independent candidates were more likely to return a no-vote. Conversely, districts with higher levels of individuals with a bachelor’s degree or with incomes equal to or lower than the median were more likely to record a yes vote. A broad interpretation of these findings suggests that some of the arguments that have emerged in the media and in social commentary do have some relevance to understanding the voice referendum outcomes.

This current paper adds to a burgeoning empirical literature by presenting an ecological-based analysis of the referendum outcome, focusing on the association between voting outcomes at small spatial levels across the major capital cities and a range of socio-economic and demographic variables that capture the essence of the arguments put forward within the public discourse.

[1] Of the 15,895,231 voters who cast a vote, 155,545 recorded an informal vote, represent less than 1 per cent of all votes.

Across the literature empirical analysis of voting outcomes has been undertaken using a number of approaches ranging from the use of general and specialist surveys (Biddle et al., 2023; Cameron & McAllister, 2019a, 2019b), voting exit polls (Larcinese, Snyder, & Testa, 2013) and aggregate-level ecological data (Baum, 2023; Keir, 2009; Stimson & Shyy, 2009). The current paper is set within an ecological approach and presents an analysis of the referendum outcome focusing on the likelihood of a no-vote given a range of independent variables. The unit of analysis used in the study is the Australian Bureau of Statistics Statistical Local Area 2 (SA2). SA2s represent a community that interacts together socially and economically. In large cities, they can be thought of as largely representing one or a few small suburbs (ABS, 2016). Accounting for missing data, 1393 SA2s across the eight capital cities were used in the analysis.

Data

The dependent variable (no-vote) is dichotomous taking the value of 1 if the outcome in a particular SA2 was no and 0 if the outcome was yes. Data for individual polling booth outcomes was obtained from the Australian Electoral Commission and geocoded to match Australian Bureau of Statistics SA2 boundaries. Data was then extracted using QGIS so that each SA2 was given a total for the number of no votes cast at a polling booth within the SA2 boundary and the total number of formal votes. In cases where SA2s contained more than one polling booth the sum of no and total formal votes was obtained. This data was then used to calculate the percentage of no-votes and the dichotomous independent variable was coded based on the final percentage of no-votes.

To test the proposed hypotheses, a number of independent variables are included in the analysis (Table 1). Hypothesis 1 (everyday issues) is tested using an indicator of financial resilience developed by Baum and Mitchell (2023). The Financial Resilience Barometer is an index that measures financial resilience for SA2s. Lower scores on the index indicate that an SA2 has lower levels of financial resilience (higher financial vulnerability). Hypothesis 2 (multi-cultural empathy) is tested using a measure of multiculturalism developed by the authors. The measure combines an indictor accounting for the per cent of people in an SA2 who were Australian-born and spoke English poorly and an indicator of country of birth diversity. Hypothesis 3 (confusion about the issues) is tested using the percentage of people in a SA2 working in a low skilled occupation. This is taken to be a broad measure of human capital that accounts for both formal education and informal education such as life experience. Hypothesis 4 (proximity thesis) is tested using the percentage of the population within a SA2 who identified as Aboriginal or Torres Strait Islander in the 2021 census. Hypothesis 5 (Conservatism) is tested using two variables, the percentage of households who own their home outright and the percentage of people who identify as Christian. Hypothesis 6 (political party) is tested using a dummy variable accounting for the political party holding the seat in which the SA2 is located.

Table 1

Independent variables and associated hypotheses included in the analysis
Hypothesis	Variable	Measure	Source
1.Everyday issues	Financial Resilience	Financial Resilience Barometer	Baum and Mitchell (2023)
2.Multi-cultural empathy	Cultural diversity	Cultural diversity index	Author developed
3.Confusion about issues	Human capital	Per cent of people in low skilled occupations	Australian Bureau of Statistics 2021 Census of Population and Housing
4.Proximity thesis	Indigenous population	Per cent of people identified as aboriginal or Torres Straits Islander	Australian Bureau of Statistics 2021 Census of Population and Housing
5. Conservatism	Homeowners	Per cent of household who own their home outright	Australian Bureau of Statistics 2021 Census of Population and Housing
	Christian religion	Per cent of people who identify as being Christian	Australian Bureau of Statistics 2021 Census of Population and Housing
6. Political party hypothesis	Political Party	Dummy variable accounting for the political party holding the SA2’s federal electorate	Author calculated
City control variable	City	Dummy variable accounting for city	Author calculated

Analytical approach

In testing the hypotheses proposed in this paper an approach that takes account of the binary nature of the dependent variable and the potential for spatial autocorrelation introduced by the use of spatial boundaries (SA2) as the unit of analysis was required. The need to account for potential spatial autocorrelation is important in the type of modelling undertaken here. Spatial autocorrelation refers to the statistical dependence between observations in space, meaning that nearby observations tend to have similar values. In particular, the failure to account for spatial autocorrelation jeopardises the validity, accuracy, and reliability of the model's estimates and predictions, particularly in spatially correlated data sets where spatial relationships play a significant role in determining the binary outcome.

Although there are a significant number of approaches for modelling spatial data sets that have continuous dependent variables, the availability of models for binary dependent variables is more limited. The approach used in this paper was to compute a series of spatial autoregressive binary dependent variable models (SARB) using the spldv package in R (Piras & Sarrias, 2023). The spldv package uses generalised methods of moments (GMM) estimators to fit several possible one-step or two-step spatial autoregressive models.

Fitting SARB models provides familiar modelling outputs (regression coefficients, standard errors) and additionally provides a test statistic (λ (lambda)) that measures the spatial autocorrelation of the model's residuals. In the context of spatial regression, λ is often associated with spatial weight matrices that capture the spatial relationships between observations. These matrices assign weights to pairs of observations based on their spatial proximity. A value of λ = 0 indicates no spatial autocorrelation, meaning that residuals are independent of spatial location. On the other hand, λ = 1 implies strong spatial autocorrelation, indicating that residuals at a particular location are highly correlated with those of neighbouring locations (Anselin, 2013; Piras & Sarrias, 2023).

An important decision guiding the use of spatial models is the choice of spatial weights. Spatial weights play a fundamental role in determining the interdependence or connectivity between different geographical units within a spatial analysis. The choice of spatial weights encapsulates the essence of relationships among these units and are generally based on contiguity or distance. In building the analysis, several different spatial weights were used. For the final analysis, a queen contiguity-based spatial weight matrix is employed. Queen contiguity-based spatial weights define neighbours as entities that share any common boundary point, rather than exclusively sharing an edge or border. In other words, two geographic units are considered neighbours if they share any part of their border, even if it's just a single point.

Table 2 presents the means for the independent variables compared with the no and yes vote outcomes and the results of ANOVA and chi-square tests. The simple bi-variate analysis suggests some support for the hypotheses posed. On average, SA2s recording a no vote had a lower level of financial resilience, and a higher proportion of indigenous population and a higher proportion of people identifying as Christians. Higher percentages of low skills (human capital) are associated with the no vote. There are differences between the vote outcomes for cultural diversity and homeownership, but these are not statistically significant. For the political party dummy, SA2s represented by the Labor party, or the coalition were associated with higher proportions of no votes, while SA2s represented by the Greens and independents were associated, on average, with the yes-vote. Across the cities, Canberra, Hobart, and Melbourne were associated with the yes vote, with the other cities being associated with the no vote.

Table 2

Descriptive statistics, ANOVA and Chi-squared tests
	Yes vote	No vote	Test statistics
Financial resilience	1453.7	962.9	F = 423.15**
Cultural diversity	674.9	648.6	F = 1.69
Indigenous population (%)	1.19	2.51	F = 100.79**
Human capital (%)	11.5	16.1	F = 417.55**
Homeowners (%)	27.5	27.9	F = 0.661
Christian (%)	38.5	45.0	F = 186.0**
Political party			χ² = 169**
Labor (%)	37.3	62.7
Coalition (%)	22.2	77.8
Green (%)	77.3	22.7
Independent (%)	86.7	13.3
City			χ² = 190.5**
Adelaide (%)	20.0	80.0
Brisbane (%)	30.3	69.7
Canberra (%)	92.2	7.8
Darwin (%)	8.0	92.0
Hobart (%)	63.6	36.4
Melbourne (%)	54.1	45.9
Perth (%)	21.7	78.3
Sydney (%)	33.0	67.0

To obtain a better understanding of the association between the independent and dependent variables, a series of spatial and non-spatial Probit models were produced (Table 3). The result from the non-spatial Probit model (column 1) provides insights into the hypothesises presented in this paper. Considering the significant coefficients, on average, an increase in an SA2's financial resilience score correlates with a decreased likelihood of recording a 'no' vote. Similarly, higher levels of cultural diversity are associated, on average, with a reduced tendency for an SA2 to record a no vote, although the size of the coefficient suggests that the impact is small relative to other variables. In contrast, an increase in the percentage of individuals with low skills within an SA2 leads to a higher likelihood of a no vote. As the percentage of people identifying as Christian increases, so does the likelihood that an SA2 will record a no-vote. While an increase in the proportion of people from an indigenous background within an SA2 suggests a lower inclination to vote no, this outcome lacks statistical significance. Similarly, although a higher percentage of homeowners is associated with a no vote, this association is insignificant. Regarding the political party dummies, SA2s represented by the coalition exhibit a greater likelihood of voting no compared to those represented by members of the Australian Labor Party. Conversely, SA2s represented by members of the Australian Greens party or independents demonstrate a lower likelihood of voting no, although only the dummy variable for independents is significant. Analysing city dummies, SA2s in Darwin and Perth exhibit a higher likelihood of voting no compared to Adelaide, while SA2s in Canberra, Hobart, and Melbourne display a lower propensity to vote no. Although SA2s in Brisbane show a tendency to vote no, this finding lacks statistical significance.

As outlined in the methodology, an important consideration in modelling the voting outcomes using Statistical Area 2 as the unit of analysis was to account for the potential for the presence of spatial autocorrelation. All three spatial models record a significant Lambda (λ) which is indicative of the presence spatial autocorrelation within the data. Importantly, controlling for spatial autocorrelation in each of the models has reduced the size of the individual coefficients, compared to the non-spatial model. Of the three models presented, the two-step model has the highest Lambda value suggesting a positive spatial autocorrelation in the likelihood of recording a 'no' vote. Across all three models, the coefficients generally align in direction with the Probit model (non-spatial), lending support to the hypotheses presented earlier.

Table 3

Regression Results, No-vote
	Non-spatial Probit model	Linearised spatial GMM model (LGMM)	One-step spatial GMM model	Two-step spatial GMM model
(Intercept)	-1.290*	-1.665***	-1.472***	-1.357**
Financial resilience	-0.002***	-0.001***	-0.001***	-0.001***
Cultural diversity	-0.000***	-0.000**	-0.000***	-0.000***
Human capital	0.104***	0.079***	0.081***	0.080***
Indigenous population	-0.036	-0.027	-0.035	-0.031
Homeowners	0.002	0.002	0.002	0.002
Christians	0.074***	0.058***	0.055***	0.054***
Political party (ALP contrast)
Coalition	0.521***	0.325***	0.373***	0.401***
Green	-0.331	-0.144	-0.196	-0.163
Independent	-0.536***	-0.338*	-0.346**	-0.302*
City (Adelaide contrast)
Brisbane	-0.397*	-0.364	-0.289*	-0.254
Canberra	-1.618***	-0.933***	-0.924***	-0.868***
Darwin	2.396***	1.392***	1.603***	1.689***
Hobart	-1.668***	-0.955**	-0.955***	-0.961***
Melbourne	-1.203***	-0.719***	-0.747***	-0.741***
Perth	0.166	0.019	0.084	0.13
Sydney	-0.137	-0.135	-0.096	-0.065
Lambda λ		0.403***	0.429***	0.405***
N	1393	1393	1393	1393

This paper has tested several hypotheses relating to the outcomes of the 2023 Australian Voice to Parliament Referendum. Using voting outcomes at the aggregate Statistical Area 2 (SA2) and targeted independent variables, the paper has applied a socio-ecological approach to test six hypotheses relating to voting outcomes that emerged within the broad public debate following the resounding support for the no case. Based on outcomes of the spatial autoregressive binary dependent variable models (SARB) a number of narratives regarding the referendum outcome can be considered.

It appears that for voters in many areas, day-to-day issues including the cost-of-living crisis impacted voting outcomes. This was a common agreement during the lead up to the day of the referendum vote and in the aftermath of the successful no vote. As individuals grappled with rising expenses, housing affordability challenges, and stagnant wages, their attention and priorities may have shifted towards immediate concerns affecting their livelihoods. Amidst such economic pressures, the discourse surrounding constitutional reforms, including the proposed Voice to Parliament, may have taken a backseat for many voters. With limited resources and bandwidth to engage with complex political issues, individuals may have been less inclined to actively engage in the referendum process or to support constitutional changes, instead focusing on more immediate economic matters. Consequently, the referendum's outcome may have reflected a populace preoccupied with day-to-day survival, impacting the level of engagement, and ultimately shaping the decision-making process regarding the voice to parliament vote.

Even prior to the referendum day the cultural diversity of Australian communities was expected to have an impact on the voting outcome and in particular in garnering support for the yes vote (Gunstone, 2023). In communities where diversity thrives, there is often a heightened awareness of the importance of inclusivity and representation. Members of culturally diverse communities may have recognized the value of a Voice to Parliament as a mechanism for ensuring indigenous voices are heard and their perspectives are represented in the decision-making process. Moreover, individuals from culturally diverse backgrounds may have firsthand experiences of marginalization or discrimination within the existing political framework, making them more receptive to reforms aimed at addressing systemic inequalities. Therefore, the acknowledgment of diverse voices and the desire for more inclusive governance structures likely propelled support for the "yes" vote among culturally diverse communities during the referendum.

There is significant debate around the associations between political knowledge, awareness and engagement and education (Hillygus, 2005; Mishra et al., 2023; Schofer et al., 2021), with many arguing that lower education or human capital is associated with lower levels of engagement and political knowledge, which in turn results in voters being unable to cut through political spin and arguments or simply not engage fully with voting processes. In the context of the referendum results, and the findings of the analysis, low levels of human capital in certain communities may have contributed to support for the no vote. In communities where access to education and information is limited, individuals may be less equipped to fully understand the implications and benefits of constitutional reforms such as the Voice to Parliament. Without sufficient knowledge and understanding of the proposed changes, residents may be more inclined to stick with the status quo out of fear of the unknown or skepticism towards unfamiliar political structures. Additionally, low levels of human capital can correlate with socioeconomic challenges, such as unemployment or poverty, which may lead individuals to prioritize immediate concerns over broader constitutional issues. As a result, communities with limited human capital may have been more susceptible to misinformation or disinformation campaigns that framed the referendum in a negative light, ultimately influencing their decision to vote no.

A fourth narrative expressed both during the referendum campaign and in the aftermath was that voter conservatism may have influenced the no vote. Given the significant variable presented in the analysis a case could be made that religious values associated with Christianity in certain communities could have influenced support for the "no" vote in the Voice to Parliament referendum. Within Christian teachings, there may be an emphasis on traditional authority structures and societal order, which aligns with the status quo upheld by the existing political system. Some adherents may view proposals for constitutional reform, such as the introduction of a Voice to Parliament, as potentially disruptive to established norms and values. Additionally, interpretations of religious teachings regarding governance and authority may lead individuals to prioritize a centralized decision-making process rather than endorsing decentralized power structures proposed by the Voice to Parliament.

A final narrative relates to the impact of political partisanship in impacting on voter behaviour. Given the distribution of yes and no outcomes across communities in seats held by competing political parties that the views espoused by different members likely played a significant role in shaping support for the "no" vote in the Voice to Parliament referendum. In many cases, individuals' voting decisions are heavily influenced by their allegiance to a particular political party rather than the merits of the referendum itself. Party leaders and officials may have framed the referendum as a partisan issue, aligning their messaging with their party's stance and urging supporters to vote in line with party interests. This could lead to widespread opposition to the proposed reforms among supporters of parties that officially opposed the Voice to Parliament, regardless of the individual's personal views on the matter. Additionally, partisan media outlets and campaign strategies may have further reinforced party lines, making it difficult for voters to consider the referendum objectively. As a result, political party partisanship likely contributed to a significant portion of the "no" votes cast in the referendum.

The outcome of the voice to parliament referendum reflected an intricate combination of factors which played into the strategies for both the yes and no camp. Taken at face value the outcome is in contrast with the general high level of support that Australians show towards improving the well-being of the indigenous population (Levy & McAllister, 2022; Markham & Sanders, 2020). But understanding the outcome is not as simple as considering voters viewpoints around indigenous issues per se. The findings in this paper suggest that there was a myriad of complex factors that swayed the referendum outcome. For example, the suggestion that a large number of voters were more concerned about day-to-day issues says a lot about the inability of the the yes campaign’s messaging to cut-through other issues that individuals were living on a daily basis. The importance of conveying the appropriate message in political campaigns cannot be overstated. Messages are the primary means through which candidates and parties communicate their policies, values, and visions to the voting public. Crafting messages that resonate with voters' concerns, aspirations, and values is essential for building trust and garnering support for a particular outcome. Moreover, the way messages are framed and delivered can shape public perception, influence voter behavior, and even define the entire narrative surrounding an issue. Effective messaging requires careful consideration of the target audience, an understanding of prevailing sentiments and attitudes, and the ability to convey complex ideas in a clear, concise, and compelling manner.

The 2023 Voice to Parliament referendum was set to be a significant turning point in Indigenous relations within Australia. The defeat of the referendum has been seen by many as a serious blow to the mission of improving the lives of indigenous people across the country. Understanding the outcomes of the referendum vote has been a common thread within the popular media and commentary since the results were announced. Until now, much of the discussion has been based on supposition and rudimentary consideration of the data. This paper provides, what we believe to be, one of the first analysis of its kind on the referendum outcome and provides a sound empirical starting point within which to frame meaningful discussion.

ABS. (2016). Australian Statistical Geography Standard (ASGS): Volume 1 – Main Structure and Greater Capital City Statistical Areas, Australia, July 2016. Canberra: Australian Bureau of Statistics
Albertson, B., & Gadarian, S. K. (2015). Anxious politics: Democratic citizenship in a threatening world: Cambridge University Press.
Anderson, I., Paradies, Y., Langton, M., Lovett, R., & Calma, T. (2023). Racism and the 2023 Australian constitutional referendum. The Lancet, 402(10411), 1400-1403.
Anselin, L. (2013). Spatial econometrics: methods and models (Vol. 4): Springer Science & Business Media.
Barber, M., & Pope, J. C. (2019). Does party trump ideology? Disentangling party and ideology in America. American Political Science Review, 113(1), 38-54.
Baum, S. (2023). A socio-economic analysis of polling booth catchments at the 2019 Australian federal election. Papers in Applied Geography, 9(1), 104-123.
Bennett, S. (1985). The 1967 referendum. Australian Aboriginal Studies(2), 26-31.
Biddle, N., Gray, M., McAllister, I., & Qvortrup, M. (2023). Detailed analysis of the 2023 Voice to Parliament Referendum and related social and political attitudes ANU Centre for Social Research and Methods.
Bisbee, J., & Honig, D. (2022). Flight to safety: COVID-induced changes in the intensity of status quo preference and voting behavior. American Political Science Review, 116(1), 70-86.
Boden, R., & Nedeva, M. (2010). Employing discourse: universities and graduate ‘employability’. Journal of Education Policy, 25(1), 37-54.
Butler, J. (2023, 13 July). Indigenous voice: no campaign’s deep links to conservative Christian politics revealed. The Guardian Australia. Retrieved from https://www.theguardian.com/australia-news/2023/jul/13/indigenous-voice-no-campaigns-deep-links-to-conservative-christian-politics
Cameron, S., & McAllister, I. (2019a). 2019 Australian federal election: results from the Australian Election Study.
Cameron, S., & McAllister, I. (2019b). Trends in Australian Political OpinionResults from the Australian Election Study1987– 2019. Retrieved from https://australianelectionstudy.org/wp-content/uploads/Trends-in-Australian-Political-Opinion-1987-2019.pdf
Campbell, A., Converse, P. E., Miller, W. E., & Donald, E. (1960). Stokes. The american voter. In: New York: Wiley.
Chong, D., & Mullinix, K. J. (2019). Information and issue constraints on party cues. American Politics Research, 47(6), 1209-1238.
Chowdhury, I. (2023, 10 October). Intifar Chowdhury: Voice to Parliament can’t compete with cost-of-living crisis in voters’ minds. The West Australian. Retrieved from https://thewest.com.au/opinion/intifar-chowdhury-voice-to-parliament-cant-compete-with-cost-of-living-crisis-in-voters-minds-c-12150138
Collins, P. (2023, 16 October). Waleed Aly says Voice referendum was too 'complicated' for 'less educated' Australians to understand - after only the most elite suburbs voted Yes. Daily Mail. Retrieved from https://www.dailymail.co.uk/news/article-12635301/Waleed-Aly-Voice.html
Colvin, R., & Jotzo, F. (2021). Australian voters’ attitudes to climate action and their social-political determinants. PloS one, 16(3), e0248268.
Crisp, L. (1983). Australian National Government. Melbourne: Longmans.
Duffield, L. (2023, 15 September). Voice Referendum: Old-fashioned racism driving 'No' campaign. The Independent Australia. Retrieved from https://independentaustralia.net/politics/politics-display/voice-referendum-old-fashioned-racism-driving-no-campaign,17901
Evans, C., Rees, G., Taylor, C., & Fox, S. (2021). A liberal higher education for all? The massification of higher education and its implications for graduates’ participation in civil society. Higher Education, 81, 521-535.
Federico, C. M., & Malka, A. (2018). The contingent, contextual nature of the relationship between needs for security and certainty and political preferences: Evidence and implications. Political psychology, 39, 3-48.
Galinsky, A. D., & Moskowitz, G. B. (2000). Perspective-taking: decreasing stereotype expression, stereotype accessibility, and in-group favoritism. Journal of personality and social psychology, 78(4), 708.
Gunstone, A. (2023). Multicultural support offers hope for a Yes outcome. National Indigenous Times. Retrieved from https://nit.com.au/10-10-2023/8030/multicultural-support-offers-hope-for-a-yes-outcome-on-october-14
Hillygus, S. (2005). D, The missing link: exploring the relationship between higher education and political engagement. Political Behavior Vol. 27 No. 1 March.
Huddy, L., Bankert, A., & Davies, C. (2018). Expressive versus instrumental partisanship in multiparty European systems. Political psychology, 39, 173-199.
Jakubowicz, A. (2023, February 8th). Will multicultural Australians support the Voice? The success of the referendum may hinge on it. The Conversation. Retrieved from https://theconversation.com/will-multicultural-australians-support-the-voice-the-success-of-the-referendum-may-hinge-on-it-199304
Jost, J. T., Sterling, J., & Stern, C. (2017). Getting closure on conservatism, or the politics of epistemic and existential motivation. In The motivation-cognition interface (pp. 56-87): Routledge.
Keir, W. N. (2009). Voter behaviour and constitutional change in Australia since 1967. Queensland University of Technology,
Larcinese, V., Snyder, J. M., & Testa, C. (2013). Testing models of distributive politics using exit polls to measure voters’ preferences and partisanship. British Journal of Political Science, 43(4), 845-875.
Leeper, T. J., & Slothuus, R. (2014). Political parties, motivated reasoning, and public opinion formation. Political psychology, 35, 129-156.
Levy, R., & McAllister, I. (2022). Public opinion on Indigenous issues and constitutional recognition: three decades of liberalisation. Australian Journal of Political Science, 57(1), 75-92.
Lupia, A. (2006). How elitism undermines the study of voter competence. Critical Review, 18(1-3), 217-232.
Lupia, A. (2018). How elitism undermines the study of voter competence. In The Nature of Belief Systems Reconsidered (pp. 263-278): Routledge.
MacKuen, M., Marcus, G. E., Neuman, W. R., & Keele, L. (2007). The third way: The theory of affective intelligence and American democracy. The affect effect: Dynamics of emotion in political thinking and behavior, 124-151.
Markham, F., & Sanders, W. (2020). Support for a constitutionally enshrined First Nations Voice to Parliament: Evidence from opinion research since 2017.
Mishra, S., Klein, D., & Müller, L. (2023). Does the higher education experience affect political interest, efficacy, and participation? Comparing dropouts to graduates and ‘non-starters’. European Journal of Higher Education, 1-18.
Mitchell, I. S. (1968). Epilogue to a referendum. Australian Journal of Social Issues, The, 3(4), 9-12.
Peterson, E. (2019). The scope of partisan influence on policy opinion. Political psychology, 40(2), 335-353.
Piras, G., & Sarrias, M. (2023). GMM Estimators for Binary Spatial Models in R. Journal of Statistical Software, 107, 1-33.
Putnam, R. D. (2000). Bowling alone: The collapse and revival of American community: Simon and schuster.
Ray, J. J. (1983). Racial attitudes and the contact hypothesis. The Journal of social psychology, 119(1), 3-10.
Schofer, E., Ramirez, F. O., & Meyer, J. W. (2021). The societal consequences of higher education. Sociology of Education, 94(1), 1-19.
Slothuus, R., & Bisgaard, M. (2021). How political parties shape public opinion in the real world. American Journal of Political Science, 65(4), 896-911.
Slothuus, R., & De Vreese, C. H. (2010). Political parties, motivated reasoning, and issue framing effects. The Journal of politics, 72(3), 630-645.
Sniderman, P. M., & Stiglitz, E. H. (2012). The reputational premium: A theory of party identification and policy reasoning: Princeton University Press.
Stimson, R. J., & Shyy, T.-K. (2009). A Socio-spatial Analysis of Voting for Political Partries at the 2007 Federal Election. People and Place, 17(1), 39.
Thorisdottir, H., Jost, J. T., Liviatan, I., & Shrout, P. E. (2007). Psychological needs and values underlying left-right political orientation: Cross-national evidence from Eastern and Western Europe. Public Opinion Quarterly, 71(2), 175-203.

The authors declare no competing interests.

Download PDF

Version 1

posted

You are reading this latest preprint version

Unravelling the Referendum: An analysis of the 2023 Australian Voice to Parliament Referendum outcomes across capital cities

Status:

Version 1

Abstract

1. Introduction

2. Data and Approach

3. Findings

4. Conclusion and discussion

References

Additional Declarations

Status:

Version 1