Political Parties and Campaigning in Australia: Data, Digital and Field

By Glenn Kefford

Big data and microtargeting steal the headlines about campaigning. But how important are they really to the way that political parties campaign? This book provides a fine-grained account of the campaign practices of three Australian political parties. It explores how prevalent data-driven campaigning is, introduces an original theoretical framework to understand these practices, and demonstrates that there is a disconnect between what Australian voters think about these issues and the way that parties campaign in the 21st century. Drawing on 161 interviews, participant observation and original survey data, it shows that the reality of contemporary campaigning is often different to what we are led to believe.

• Language: English
• Publisher: Palgrave Macmillan
• Release date: Feb 15, 2021
• ISBN: 9783030682347

Book preview

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021

G. KeffordPolitical Parties and Campaigning in AustraliaPolitical Campaigning and Communicationhttps://doi.org/10.1007/978-3-030-68234-7_1

1. Introduction

Glenn Kefford¹  

(1)

School of Political Science and International Studies, University of Queensland, St Lucia, QLD, Australia

Glenn Kefford

Email: [e-mail]

Data-driven campaigning and microtargeting are said to not only be changing election campaigns, but democracy itself. Reports of this occurring are widespread. Christopher Wylie, the Cambridge Analytica whistleblower, told the Guardian (Cadwalladr and Graham-Harrison 2018):

We exploited Facebook to harvest millions of people’s profiles. And built models to exploit what we knew about them and target their inner demons.

Writing about the United Kingdom’s referendum to exit the European Union, journalist, Carole Cadwalladr (2017a), summed up what the result meant for the UK:

This is Britain in 2017. A Britain that increasingly looks like a managed democracy. Paid for by a US billionaire. Using military-style technology. Delivered by Facebook.

Claims about the power and significance of data are ubiquitous in the coverage of election campaigns across advanced democracies. In the aftermath of the 2016 United States (US) presidential election, campaign insiders claimed microtargeting was a key reason that Donald Trump won (Pramuk 2017). Likewise, similar assertions have been made about Australian democracy (Preiss 2018; Burns and Morris 2018). According to this view, twenty-first-century election campaigning involves political parties using ‘big data’, targeting voters with incredible precision, changing what we think and how we will vote.

Except, almost none of this appears to be true. In fact, most of what is written in the media about election campaigns seems to be false. Some is the work of public relations teams looking to drum up business for their clients (Ball 2016). Some is campaigns exaggerating their capacities to scare their opponents (Lynch et al. 2018), and the rest are fantasies about how a new software program swung voters one way or the other, or how a new dataset completely revolutionised how we understand women aged between 35 and 45 who live in regional areas (Allen and Vogel 2014). The reality is that we have almost no scholarly evidence to support most of the claims made about data-driven campaigning. So, what is really going on?

With the exception of the US, we have a patchy understanding of how prevalent data-driven campaign practices are. This includes how parties collect data, the way data is used in election campaigns, and how it informs microtargeting strategies online and offline. Similarly, we have a limited understanding of how data-driven campaigning is affecting political parties in parliamentary democracies. While a small but expanding group of scholars argue that the reality of data-driven campaigning is different from the way it is portrayed in the media (Hersh 2015; Baldwin-Philippi 2017; Anstead 2017; Baldwin-Philippi 2019), the breadth and depth of knowledge remains inadequate.

In this book, I provide a fine-grained account of data-driven campaigning in Australia. I examine the campaigning practices of three political parties—the Australian Labor Party (Labor), the Liberal Party and the Greens.¹ In doing so, I detail how prevalent data-driven campaign practices are, outline a new theoretical framework for understanding these practices, analyse the way these practices are affecting political parties and detail what voters think of these practices. What I demonstrate is data-driven practices are extremely uneven. And, even in those parties that are engaging in data-driven campaigning, the effect of these campaigns is debatable. Hence, while discussions of data-driven campaigning practices have become ritualised, and are part of the everyday discourse of political operatives in Australia—like elsewhere—the evidence I present suggests we should remain sceptical about many of the claims made about these practices.

In making these arguments, I draw on unparalleled qualitative data and evidence for an Australian political science project. This includes 161 interviews conducted between 2015 and 2020 with party officials, members and supporters as well as third-party actors and consultants who have worked for the parties on election campaigns. I also draw on participant observation inside the field campaigns of two separate parties during the 2019 Australian federal election—Labor and the Greens. In total, I spent approximately 96 hours embedded in these campaigns undertaking orthodox field campaigning activities such as doorknocking and phonebanking. While participant observation has been used previously by scholars studying campaigns (Nielsen 2012; Baldwin-Philippi 2015), to my knowledge no other scholar has undertaken participant observation inside two different parties, and certainly not concurrently. I also draw on original survey data from a nationally representative sample of Australian voters (n = 1019) which explores voter perceptions of campaigning practices, including the collection and use of data and targeting. Finally, I draw on a large secondary scholarship from both Australia and internationally. More on this can be found in Chapter 10—the research appendix—which outlines in detail the methods used in this book and how data was collected and analysed.

Campaign Effects: Fact and Fiction

That tales of data-driven campaigning are catnip to journalists and political commentators should not surprise us. They contain the best parts of any good story. They present a riddle or mystery. There are heroes and heroines. And they provide closure by telling us how elections were won or lost. Quelle surprise, it was only possible because of the unique set of skills and insights our heroes and heroines possess (Karpf 2019). These stories are influential and effect perceptions of campaigning as they appear to shine a light on processes and practices that most laypeople know very little about (Baldwin-Philippi 2020). Like many advanced democracies, Australia has its fair share of these stories (Bourke 2019).² But whether it is Cambridge Analytica, Donald Trump’s 2016 presidential campaign, Barack Obama’s campaigns before that, or any other campaign, the scholarly evidence is clear, campaign effects are small and this has been the orthodox view for decades (Gosnell 1927).

The cutting-edge literature on campaign effects says: mobilisation is easier than persuasion; mobilisation effects are likely to be larger than persuasion effects; the best way to mobilise or persuade voters is through a one-on-one conversation between a campaigner and a prospective voter (Green and Gerber 2015; Kalla and Broockman 2020); and, persuasion effects are likely to be zero except in a very specific set of circumstances (Kalla and Broockman 2018). Thus, while there is plenty of talk about the effects of digital campaigns—and there remains much that we do not know—what we do points in a similar direction: there is little to no effect on persuasion (Haenschen and Jennings 2019; Hager 2019). In fact, for all the talk of the game-changing effects Trump’s digital team had (for example, see Beckett 2017; Ellyat 2017), the most systematic analyses of the 2016 election from Sides et al. (2018: 194) states unequivocally: ‘there is little evidence yet that digital advertising has much impact on voters or consumers’. This literature does come with some caveats. One is that these effects can be hard to measure. Another is that we still have limited knowledge about effects in a range of contexts and institutional setups, including in a country like Australia that uses compulsory voting.³ But for now, the decades long consensus on campaign effects remains.

Laypeople reading this may be surprised. After all, much of the commentary on data and electoral politics suggests it is an existential threat to liberal democracy or human agency (see, for example, Cadwalladr 2017a, b). But much of the media coverage confuses and conflates a range of related but distinct issues concerning data and electoral politics. In particular, it conflates the perceived sophistication of data and analytics operations—such as modelling to build scores and to target voters—with the claimed efficacy of such campaigns. In other words, it assumes that because statistical and computational methods used to analyse data are increasingly sophisticated that this inevitably leads to some enlarged effect on voting behaviour. However, the evidence to support this is flimsy to say the least (Baldwin-Philippi 2017, 2019, 2020).

Campaign effects are also frequently conflated with data privacy and a range of other related issues. This includes: the use of tracking data about online activity; the acquisition and use of commercial data based on purchasing or financial history; and, the spread of disinformation (Bennett and Lyon 2019). We should be concerned about these issues (Franz 2013), and I would argue for stronger legislative and regulatory mechanisms in every single democracy, including Australia. But these are separate issues that need to be understood in isolation from one another. There is very little scholarly evidence to suggest that parties or any other campaigner have the types of persuasive effects often alluded to in the commentary. In fact, there is good reason to doubt the capacities of political parties to even execute data-driven campaigns (Baldwin-Philippi 2019). Hence, the journalistic sub-genre that I like to describe as ‘data dystopia’, should be largely ignored when trying to make sense of campaigning in the twenty-first century.

The Study of Political Parties and Campaigning

The scholarly literature on campaigning is enormous, spanning sub-fields such as political science, political communication, journalism and internet studies. The international study of campaigning is also methodologically diverse and rich with empirical and theoretical contributions. Interviews remain one of the primary methods used, especially by many scholars in political science and communication. Quantitative methods are common, as are experimental approaches and designs. However, just like in political science more generally, fine-grained qualitative accounts are increasingly rare.

In contrast to this vast international literature, there is little written on campaigning in Australia, and especially campaigning by political parties. Mills (1986, 2014) has written two seminal books. There have been important contributions on digital from Chen (2010, 2012, 2013, 2015), as well as Gibson and co-authors (2002, 2008, 2011), analyses of advertising campaigns and strategies from Young (2003, 2004, 2015), considerations of various reforms to party organisation which affects campaigning from Gauja (2014, 2016, 2017), and a number of relevant contributions from Ward (1991, 2001, 2006) and Van Onselen (2007, 2008).⁴ Yet for all the virtues of these contributions, we still know little about how campaigns actually function and are organised. We also know next to nothing about how data is collected, analysed and used, and what those involved in the campaigns actually think about these practices.

Unlike previous contributions, this book brings the various threads together so that we can better understand not just data and analytics but also digital and field campaign practices. It does so to make sense of these associated phenomena, including how they affect party organisation. Campaigning is—as I discuss in much more detail in Chapter 2—unlike any other aspect of party organisation for the simple fact that it is subject to disruption technologies and there is no ‘official’ story (Kefford 2018a). Even modes of campaigning which appear to have remained largely the same since the 1970s such as television advertising or direct mail, are no longer simply about market saturation. Instead, these legacy techniques are underpinned by growing datasets. In the words of one former campaign director from the Liberal Party: ‘Better use of data and targeting means that direct mail and TV are now much more like voter contact and social media and are now much more effective in persuading voters’.

While parties in Australia, like elsewhere, are still engaged in broadcast campaign techniques such as television and print advertising (as well as direct mail), these are not the focus of this book. Instead, this book is anchored around three themes: data, digital and field. The reason for this is that while other practices and campaign channels remain important, they are significantly less important than they once were. Moreover, the nature of these campaign activities have far less to do with the parties themselves. Indeed, this was the whole point of much of the scholarly literature which spoke of the professionalisation of politics from the 1960s onwards, including the work of Epstein (1968), Panebianco (1988), and, of course, Katz and Mair (1995). As this book is not just about how parties’ campaign, but what effect these campaigns have on the parties themselves, I focus on what I argue are three core ingredients of contemporary party-based campaigns. Each of these is worth speaking about in more detail.

Scholarly research on data-driven campaigning is a burgeoning field, albeit one dominated by US scholars who almost exclusively write on one case, their own. Many of these contributions are, however, excellent. This includes those from Kreiss (2012, 2016), Hersh (2015) and Baldwin-Philippi (2017, 2019, 2020). There have also been important contributions from a range of scholars who have analysed developments in parliamentary democracies such as the UK (Anstead 2017, 2018), and Germany (Kruschinski and Haller 2017; Papakyriakopoulos et al. 2018). Moreover, as the field has expanded, the theoretical insights have also deepened, and attempts to understand the psychological bases of targeting and the analytics process are increasingly common (for example Dobber et al. 2017; Madsen 2019). Significant gaps do, however, remain.

While I discuss this more in Chapter 2, I define data-driven campaigning as a set of interlocking practices and processes which includes collecting data, building models of the electorate, creating supporter and persuadability scores, segmenting and targeting voters at the individual level, and the use of testing to refine messaging and campaign strategy (Baldwin-Philippi 2017). The data I talk about above and in the title of this book, therefore, is the information collected and analysed so that political parties and other organisations can contact voters online and offline. Discussed more in Chapter 3, the practices associated with collecting and analysing this information are often referred to as data and analytics. Depending on the sophistication of the data and analytics operations, targeting may be at the individual level—which is commonly seen as a key characteristic of data-driven campaigning—or at the demographic or segmented level—which I describe as a characteristic of narrowcasting.

Of the three key components of campaigning that will be the focus of this book, digital receives the most attention from commentators. I define digital as online communications which aim to go directly from parties to voters. This includes: (a) the use of social media such as Twitter, Facebook, and Instagram; (b) online advertising via Google, YouTube and other similar sources; and (c) email to message, advertise and fundraise. While data may have underpinned the Cambridge Analytica scandal, or Trump’s Facebook campaign in 2016 (Lapowsky 2016), digital steals the headlines. We now have over a decade of sustained in-depth scholarly attention devoted to digital from a variety of sub-fields and approaches. These contributions are too many to note but some worth pointing to include: Vaccari’s (2013) comparative study of digital in seven Western democracies, and Baldwin-Philippi’s (2015) account of digital campaigning and the construction of citizenship in the US. Each of these, in their own way, points to how digital affords campaigns new ways to organise and create opportunities for supporters to engage and participate in democracy.

Field campaigns, or the ‘ground war’, are an essential component of the campaigns that many political parties in advanced democracies employ. In simple terms, field campaigns are the doorknocking and phonebanking operations that political parties conduct. In describing field campaigns in the US, Nielsen (2012: 7) argues that American political campaigns are pursuing ‘personalised political communication’. These practices are increasingly informed by data and analytics to allow field organisers to mobilise and co-ordinate volunteers⁵ for maximum impact. In Australia, canvassing has long been a part of the ‘ground war’ (Mills 2015: 123), but we still know very little about how field campaigns are affecting party organisation and the relationship between parties and citizens.

While data and analytics, digital and field are distinct areas of a campaign, the goal of many political parties is to integrate relevant insights across other areas of the campaign. Therefore, studying them together is important. Each of these has, of course, been the focus of scholarly books. However, no monograph that I know of has pulled these strands together in this way. Moreover, the focus of most books is on the US. While hugely influential in regard to campaigning practices and approaches, the US is an outlier due to its unusual institutional architecture and campaign finance regime. Much more evidence of how these practices are affecting parliamentary democracies is required. We also need more in-depth fine-grained analyses of how prevalent data-driven campaigning practices are within party systems. This book does just that.

Cases and Macro-Political Context

The primary unit of analysis for this book are the three parties that contest the most elections in Australia. These three parties provide useful contrasts with one another organisationally and ideologically, while confronting the same set of institutional, regulatory and contextual factors. The Australian party system, the way these parties organise themselves, and the institutional architecture of Australian democracy are complex and need to be explained, especially for non-Australian readers who may be unfamiliar with the players or the ‘rules of the game’.⁶

The defining feature of Australian democracy is the use of compulsory voting.⁷ Unsurprisingly, this has a significant effect on how political parties’ campaign. The Australian Electoral Commission (AEC) estimates that as of December 2019, 96.9 per cent of eligible Australians were registered to vote (AEC 2020). Turnout for Australian federal elections also remains high. From those registered to vote, between 91 and 95 percent have done so since 2001 (AEC 2019). Put another way, out of almost 17 million eligible voters, over 15 million cast a ballot in the 2019 federal election.

Australian federal election results are decided by a mixed electoral system.⁸ The Alternative Vote system is used for the House of Representatives and the Single Transferable Vote system is used for the Senate.⁹ This mixed system—the former a preferential system and the latter a proportional system with preferential elements—means that election contests are never just about the major parties. Despite what many continue to claim, Australia is not and has not been a two-party system for decades (Kefford 2017; Kefford et al. 2018).¹⁰ The nature and shape of electoral competition in Australia is multi-party and multi-dimensional. The major parties are in competition with one another, as well as an array of insurgent forces. The preferences of minor parties are significant, and the major parties need to not only persuade voters to support them but also to preference them above their competitors. Westminster legacies still cut deep and affect how Australian politics is conducted, but this legacy is eroding, and the shape of the electoral contest is fought in new and surprising ways in each cycle. The declining vote for the major parties is both symptom and cause in this (Kefford 2018b, 2020).

The three parties I focus on are worth elaborating on further. The first of these is Labor, which is Australia’s oldest surviving party. This longevity means the organisation has had to evolve and respond to multiple changes in the political landscape. Labor was a classic mass party. What it became next is, like so many other parties, a source of serious debate. Electoral-professional, catch-all and cartel have all been used to describe it (Warhurst and Parkin 2000; Jaensch 2006; Ward 2006). Labor is organised along federal lines, meaning that there is a degree of power-sharing that occurs between the state-based parties and the national organisation. The party has been extremely proficient and successful at winning subnational elections, but has been the party of opposition at the federal level (Moon and Sharman 2003).

The Liberal Party, at the federal level at least, is an election-winning machine. Since 1949, they have held office for 49 of the 71 years. In that time, they have won 19 out of the 28 elections. Their dominance is multicausal but has meant they are often second best when it comes to engaging with new technologies and approaches to campaigning. The organisation of the Liberal Party, like Labor, is along federal lines (Brett 2003, 2006). However, at the federal level—and in most states—they are in a formal and long-standing coalition with the Nationals, the dominant party in regional Australia. In Queensland, a key battleground state in federal elections, the two parties merged into one in 2008, becoming the Liberal National Party (LNP).

The Australian Greens sit at the apex of a confederation which consists of the various state-based green parties. Yet, like most parties in multilevel jurisdictions, federal (or national) issues and actors have increasingly shaped the organisational structure of the party (Jackson 2015; Miragliotta and Jackson 2015). While they had previously been more successful at the subnational level, the Greens have maintained a presence in the Australian Senate since 1998.¹¹ In that time, they have increased their representatives in the Senate from one to ten (currently nine), and since 2010 have held the seat of Melbourne in the House of Representatives. The rise of the Greens as the third force in Australian politics places them in direct competition with Labor and the battle between the two over who best represents the interests of progressives, reflects a critical schism in Australian electoral politics (Crowe 2018).

A key reason for studying campaigning in Australia is that it is a near-perfect—though not unique—laboratory for understanding a critical component of many modern campaigns: persuasion. Australia’s use of compulsory voting means parties devote almost no resources to mobilising voters to turnout. Like most advanced democracies, Australia is suffering from declining trust and satisfaction in our political institutions. Yet turnout remains high because we are incentivised to vote. We are socialised into a political culture where voting is normalised, uncontroversial and election days have a carnival-like atmosphere (Brett 2019).¹² Hence, while there are often issues in determining whether a strategy is effective at mobilising voters as opposed to persuading them in systems which use voluntary voting, in Australia that is largely irrelevant.

On many of the measures used to gauge how healthy democracies are, Australia performs relatively well. In fact, Australia is close to the median on many of these measures (Cameron and McAllister 2019: 15). However, this obscures the precipitous decline in the way Australian voters feel about their democracy. One proxy we can use to test democratic health is voter satisfaction with democracy. On this measure, satisfaction has dropped dramatically in the last decade or more. Hitting an all-time high in 2007 of 86 per cent, since then a sustained decline has been evident, with only 59 per cent of respondents to the Australian Election Study in 2019 saying they were satisfied.

Another aspect of the macro-political environment that Australian parties are