Campaigning in the dark: theorising campaign strategies from the 2022 Seanad by-election

ABSTRACT Parties and election candidates are often understood to be rational actors, adapting to conditions to remain electorally competitive. However, despite the rational pursuit of goals requiring enough information with which to update one’s strategies, to our knowledge, no study has sought to examine the precise role of information in influencing party or candidate updating during campaigns. To fill this gap, we use qualitative process tracing to examine the 2022 Irish Seanad by-election, which acts as an extreme case of a low-information electoral environment that can be used for theory-building. From this analysis, we expect that the information level is a function of knowledge about voters’ intentions and the terms of political debate, in the form of an interaction between the effective number of electoral parties and the range of issues discussed by parties. From this, we hypothesise that strategic updating is least likely when information about both voters and the terms of political debate are low. Indeed, the lower the information level, the more likely candidates are to prioritise their preferred issues and ignore their competitors, even when they could acquire some rudimentary knowledge of their competitors.


Introduction
One of the core goals of any electoral actor, whether they are parties or individual candidates, is to secure votes in elections, from which they gain legislative seats and, if they secure enough votes, office and the ability to enact their preferred policy.However, to continually gain enough votes to achieve these other objectives, electoral actors need to be adaptable, willing to change over time to not only survive and remain relevant, but also to succeed in their political goals.One way to respond to changing environments in the pursuit of votes is by updating the saliency attached to important issues.However, here electoral actors are often torn between two different imperatives during campaigns: the need to highlight issues that they 'own' and are seen as competent or trustworthy on, and the need to appear responsive to highly salient issues that matter to voters (Abou-Chadi, Green-Pedersen, & Mortensen, 2020;Seeberg, 2022).In the heat of a campaign, when electoral actors are deciding whether to update the saliency attached to their preferred issue or to increase their focus on a different issue, they are reliant upon information about both the priorities of voters and other competitors to determine the most beneficial strategy.However, in resolving this trade-off, it is often unclear what the most important sources of information are in the context of this trade-off, and how electoral actors will respond to them.Thus, we seek to answer the following research question: how do electoral actors respond to different information levels in election campaigns?
Within the academic literature, there are two main bodies of work that could help us to understand decision-making during elections.Firstly, the extensive literature on party competition has developed a variety of different models to explain how parties will compete.In particular, Downsian spatial analyses expect parties to strategically adjust their issue positions to attract the median voter, without losing their current voters (Adams, Clark, Ezrow, & Glasgow, 2006;Downs, 1957).In contrast, issue ownership models focus on how parties prioritise the issues that they 'own' and are trusted by voters on, while avoiding the issues owned by their rivals (Petrocik, 1996;van der Brug, 2004).Secondly, there is considerable work on party strategies within election campaigns, especially in terms of the allocation of resources to particular contests (Gurian, 1993;Hartman, Pattie, & Johnston, 2017).However, the former strand of literature is often focused on the development of party positions and issue saliency in manifestos prior to the onset of campaigns, given the widespread use of manifestos and the ease with which they can be analysed, whereas the latter strand of literature tends to assume that parties have relatively high knowledge, which then informs their precise campaign strategies.Consequently, the literature tends to ignore how party strategies, including on particular issues, can vary within election campaigns based on how much information is available.This paper thus seeks to fill a gap in the literature by developing a theoretical framework to understand the effect of different information levels on how electoral actors strategically adjust the salience they accord to various issues during campaigns.
To understand how electoral actors respond to different information levels, we examine the 2022 Seanad by-election for the University of Dublin constituency.Seanad Éireann is the upper house of the Oireachtas, and is generally seen as the weaker one (Coakley, 2013), since the government is not required to have its confidence and it can only delay legislation on which it disagrees with the Dáil (lower house).Elections to the Seanad must occur within 90 days of the Dáil's dissolution, and there are 60 Senators, drawn from three stands: 43 are elected from five vocational panels; 11 are appointed by the Taoiseach; and six are elected by university graduates, with three seats going to the University of Dublin and three to the National University of Ireland.In addition to this case being a by-election for one seat in a relatively weak house, this is also an extreme case of low information for a variety of reasons, including it being the first by-election for this constituency since 1979, with an unusually large number of candidates ( 17), no incumbency effect, little party structuring of the vote or campaign, and little information about the diffuse electorate.Given the utility of extreme cases in theory building (Gerring & Seawright, 2007), we analyse this by-election to develop our theory of what the most important determinants of candidates' information levels are, and how this will influence their updating during campaigns.We use qualitative process tracing based on semi-structured candidate interviews and other sources on the campaign, including campaign literature and social media posts.
At the core of our theoretical framework is the finding that candidates in this by-election made it clear that they entered to discuss particular issues and that this strategy did not change, which is corroborated by how they approached their campaign in their campaign literature and social media campaigns.This failure to update intuitively seems like an inefficient strategy, but in this context, it acts as a clear demonstration of how the extremely limited information faced by candidates meant that acquiring the information needed for updating was too costly.However, this low information took two main forms, which resulted in two hypotheses for future testing.Firstly, the candidates all agreed that they had very little information about voter preferences, and struggled to find additional information on the voters.Consequently, we hypothesise that lower levels of information about voter intentions preclude updating.Secondly, the candidates also focused on the unusually large number of candidates, and how the difficulty of following so many campaigns resulted in them focusing almost entirely on their own campaigns, and ignoring their competitors.Indeed, even when it was possible for candidates to acquire some information about their competitors, we find that they purposefully chose not to.Consequently, the more candidates there are, the more difficult it is to determine what issues are the main focus of the campaign, or how to strategically alter the saliency of different issues, especially as the range of issues discussed increases.Thus, we hypothesise that larger numbers of effective competitors will reduce updating, especially as the range of issues discussed by candidates increases.Overall, our findings underline the risk-aversion that often characterises electoral actors in campaigns, indicating that they persist with their original strategy in the face of limited information, rather than expending resources in an effort to overcome this hurdle.
While the Seanad by-election was a low-priority campaign, attracting little interest from political parties, this article's theoretical framework can be easily generalised to other electoral contests, including those dominated by parties.The more heavily dominated the election is by individual candidates, the easier it is to directly generalise from our theoretical framework.There are a number of political contexts beyond Ireland where independents maintain an important presence, and where the theoretical framework developed here may provide some useful insights.Several island nations in the South Pacific, such as Kiribati and Palau, are noted for the absence of political parties which leads to a candidate-centric system.At the sub-national level, parties are also absent in the Canadian territories of Nunavut and the Northwest Territories, and islands including American Samoa, Guernsey, the Isle of Man, and the Falklands (Weeks, 2017).Importantly, a key difference between an individualand party-dominated race is that parties provide a minimum level of information and structure to any electoral contest (Aldrich, 2011;Kölln, 2015), which suggests that the range of information levels will be narrower for party-dominated races.However, our theory will still be relevant as it draws on highly general features of electoral competition, and there will be considerable overlap in how both individual candidates and parties will need to track both voters and their competitors.Despite this, we also discuss some potential refinements to our framework when applying it to partydominated contests, and other potential responses to low information environments.
In addition to filling an important gap in the literature, this research also has practical implications for how campaigns operate.Campaign dynamics determine the issues that prevail on the political agenda, and thus to a certain extent the issues that parties prioritise when they get into government.Understanding how parties talk to each other during campaigns is particularly important where running for office is a means to affect the agenda, which it often is for smaller parties and independents that enter a race knowing that they are unlikely to win representation.This research may also uncover potential implications of systems that allow for an abundance of candidates to contest a given race.If, for example, it is the case that the number of candidates decreases parties' responsiveness and updating, this should factor into discussions about electoral systems and their impact on the political agenda and effective democracy.

Literature review
There are two sets of literature that are relevant to understanding campaign strategies: the party competition literature, and the campaign literature.
Within the party competition literature, there are a number of theoretical approaches seeking to explain how parties decide what position to adopt with the aim of attracting voters during an election campaign.However, all of these approaches rely on the assumption that parties use information on voters and their competitors to update their strategy.One of the main approaches is the spatial model, which sees party competition as parties taking a position on an issue, and voters voting for the party that is ideologically closest to them (Downs, 1957).A party seeking to attract new voters, then, will change its ideological position to one that will place them closest to the median voter.Laver (2005) identifies a 'hunter' model of party decision-making, whereby a leader that is unconstrained by obligations to existing party members or any democratic procedures for changing party policy moves within the ideological space based on previously learned information.If a previous move succeeded in attracting more voters, the party will move again in the same direction; if it did not, it will move again in a different direction.Adams et al. (2006) note that Downsian logic implies that parties will change their position in response to ideological moves by other parties as well as voters.This follows logically from a vote-maximisation goal: if voter X is ideologically closest to party A, but party B changes its position in order to 'capture' voter X, party A has an incentive to move closer again to party X, assuming it won't lose other voters in the process.In short, Downsian spatial modelling sees party positions as changing in response to positional changes by voters and other parties.
Political competition has also been characterised as taking the form of issue competition (Abou-Chadi, 2016;Green-Pedersen, 2007).The issue competition approach sees parties competing through the issues that they choose to emphasise over others (Green & Hobolt, 2008).In general, it is advantageous for parties to emphasise issues that they own, or on which they are perceived as being competent relative to other parties (Petrocik, 1996;van der Brug, 2004;Wagner & Meyer, 2014).Given that parties will benefit if their owned issue becomes more salient (Bélanger & Meguid, 2008;Neundorf & Adams, 2018), discussing this issue as much as possible may increase its saliency for voters during the campaign, thus increasing their chances of winning votes based on their competence.The flipside is that parties have an incentive to avoid emphasising issues on which they are perceived as incompetent, or at least less competent than another party: emphasising issues a party doesn't own risks increasing the issue's salience in the minds of voters, and increasing the vote share of the more competent party (Han, 2022).However, parties are not entirely free to determine the electoral agenda.This is because, in any campaign, there will be a hierarchy of issues (Abou-Chadi et al., 2020) that parties will face incentives to respond to.Thus, parties will face incentives to respond to issues that they do not own in order to avoid being viewed as out of touch with the public and ensure that they can contribute to the framing of those issues (Wagner & Meyer, 2014).Indeed, the closer the party is to election day, the more powerful these incentives will become (Seeberg, 2022).Overall, while a party's own competence and ideological priorities affects how issue competition plays out, it is also affected by the preferences of other parties and voters, and each party's knowledge thereof.
However, there may also be limitations to how much parties will adjust their ideological position or issue saliency in response to voter feedback.This is because parties may misdiagnose this feedback, such as by wrongly assuming that the median voter is closer to their position than they actually are (Norris & Lovenduski, 2004), or due to internal constraints from party activists in their ability to respond (Wagner & Meyer, 2014).Indeed, parties may even blame the representation or reception of their message during the campaign, rather than the message itself.Thus, while parties pay attention to voter feedback, they may not do so consistently.
While much of the existing literature focuses on the information that parties use to update their strategies ahead of launching their manifesto at the beginning of an election campaign, some attention has been focused on parties' responsiveness to cues during the campaign itself.As Pereira (2020) notes, drastic changes to a party's platform during a campaign are unlikely, as they may lead to frustration among voters and accusations of pandering.It is unsurprising, then, that studies focus on strategic decisions about resource allocation and issue emphasis.Gurian (1993) demonstrates how candidates for presidential nomination in the United States (US) update the importance attributed to a given state primarily based on the competitiveness of the campaign.In this case, the sequential nature of US presidential primaries means that the results of one primary can inform strategy ahead of another, especially in terms of the allocation of resources and the relative prioritisation of strategic goals.Similarly, Pereira (2020) shows that the prioritisation of goals by parties changes in response to information about the performance of the party during the campaign.In the United Kingdom (UK), Hartman et al. (2017) found that local constituency parties respond to new information during an election campaign by updating decisions about the allocation of resources, with parties increasing spending in constituencies where a constituency-level poll had been carried out.Moreover, spending is higher in more competitive constituencies.In addition, Hersh (2015) finds that the level of publicly available information on voters in the US affects how parties construct voter databases, and thus their microtargeting strategies in campaigns.In short, empirical evidence suggests that the very availability of information about an electoral race can affect campaign dynamics.
Finally, partisanship can provide another important source of information for both voters and candidates.While there is a vast amount of often contrasting evidence on the nature and extent of party identification, there is a broad consensus that it is important in determining vote choice (Campbell, Converse, Miller, & Stokes, 1980;Miller, Shanks, & Shapiro, 1996).One hypothesised mechanism for the link between party identification and vote choice is that partisanship acts as a heuristic for voters: it provides an ideological cue allowing them to make a somewhat informed choice without having to expend time and energy investigating the positions of each candidate (Feldman & Conover, 1983;Kam, 2005).It follows, then, that an election with minimal party involvement is likely to be shrouded in uncertainty.Not only are voters lacking an important heuristic that helps them make informed choices, but candidates themselves are lacking a heuristic that would help them gauge how a voter is likely to vote.Candidates and campaign staff often target certain voters based on the likelihood that they will be swayed by engagement from the campaign, avoiding expending resources on voters that are sure to vote for the candidate in question, or sure to never vote for them (Anstead, 2017;Nickerson & Rogers, 2014).Without clear party structuring of an election, candidates are much less able to campaign as effectively and efficiently as possible.Indeed, while few elections suffer from a complete lack of partisanship, party identification has experienced considerable decline across established democracies (Dalton, 2014;van der Meer, van Elsas, Lubbe, & van der Brug, 2015), which will result in significant variation across elections in how much parties will know about voters.
Thus, there is an extensive literature about the information electoral actors use to guide their platforms, and how they use this information both between and during campaigns.What remains less clear is how electoral actors make decisions about campaigning where cues are ambiguous and information is scarce.In these circumstances, however, electoral actors still have the same basic goals as actors in information-rich environments: attracting votes.This presents the puzzle of how electoral actors pursue their goals when they lack the information on which the literature tells us they base their strategies.It is also unclear precisely how updating varies as a function of the information that is available to campaigns.Consequently, the gap that this article seeks to fill is to understand the effect of low information environments on updating in election campaigns.

Methodology & case selection
To understand the effect of low-information environments on whether electoral actors update their strategies during electoral campaigns, we examine the 2022 by-election for the University of Dublin constituency in the Irish Seanad, which represents an extreme case of low information.Using qualitative process tracing, we use this case to develop our theoretical expectations about what causes low levels of information and how electoral actors respond to this.The choice of an extreme case is important for theory-building as it provides the full range of variation in the variables of interest, thereby allowing us to explore what causes low levels of information and its effect on updating in more detail (Gerring & Seawright, 2007).Indeed, extreme cases make it easier to determine what variables matter in determining the information environment as they can be expected to have a particularly pronounced influence, with the extreme values in these variables causing extremely low levels of information.While our theoretical framework will be most directly applicable to other independent-centric elections, it is still generalisable to party-dominated systems as it relies on general goals of electoral actors, and general features of electoral competition that both parties and candidates need to consider.However, we will also consider some factors that may affect our theoretical framework when it is applied to parties.
This by-election is an extreme case for four main reasons: the lack of a clear territorial nature; the lack of polling and a ground campaign; the limited influence of parties; and the lack of any incumbency advantage.Firstly, the electorate was composed of all graduates of the university aged 21 and over who are Irish citizens and are registered as electors (Trinity College Dublin, 2022).Consequently, this is not a territoriallybounded constituency, with the electorate instead being spread across the world, even if there is a Dublin bias to the constituency, with around 47 percent of the 67,000 registered voters being based in Dublin (Coakley, 2013).This makes the constituency exceptional compared to other Irish constituencies, which are either territorial (Dáil, local, and European Parliament elections) or based on a small electorate composed mostly of professional politicians (the Seanad's panel seats).Secondly, as a low saliency election across a diffuse constituency, there is little polling information about voter preferences.The previous point concerning the constituency's geographical spread also means that candidates find it difficult to carry out a ground-war campaign due to the difficulty of isolating high concentrations of registered voters.
Thirdly, parties have exceptionally limited influence in structuring vote choice as this constituency has a long track record of electing independents, with the major exceptions being Mary Robinson and Ivana Bacik, both of whom were formally aligned with Labour.Thus, most candidates are formally independent, even if they may have links to parties.These university constituency contests potentially remain independent-centric as they are of very low saliency to parties, given that the added value of winning an additional Seanad seat is low, relative to the difficulty of competing in these constituencies.Finally, this history of electing independents also enables a strong incumbency advantage as incumbency provides a public profile that candidates can exploit during the campaign (Reidy, 2008).This name recognition is something that challengers are often lacking.Indeed, the defeats of Catherine McGuinness in 1982 and Sean Barrett in 2016 are the only instances of an incumbent for this constituency being defeated since 1982 (Murphy, 2016).However, this by-election was held to replace Ivana Bacik, who had won a Dáil seat in the 2021 Dublin Bay South by-election.This was the first by-election for the constituency since 1979, and the lack of any incumbents made it even more uncertain than usual, as did the unusually large number of candidates -17 in total.
The uniqueness of this by-election is underlined when compared with previous University of Dublin Seanad elections.Figure 1 below plots the effective number of candidates (ENC) across University of Dublin Seanad elections, and it highlights the unusually large number of candidates, and lack of clear dominance of a small number of candidates in the 2022 by-election.Particularly interesting is the comparison between the ENC in 2022 and 2011: 2011 saw more candidates with a total of 19 contesting the election, 1 but the dominance of incumbents David Norris and Ivana Bacik, who together won 55.3 per cent of first preference votes resulted in an ENC of 5.4 compared to 9.6 in 2022.The distribution of votes at the 2022 by-election compared to previous elections thus highlights the additional uncertainty introduced by the absence of any incumbency bias.
To develop our theoretical framework, we utilise qualitative process tracing, based on the triangulation of evidence from multiple sources, including candidates' campaign literature and social media posts, media coverage of the election, and semi-structured interviews with the candidates themselves.Please see the Supplementary Materials for more details about the interview methodology, interview schedule, and list of interviewees.Interviewees were given the option to be anonymous or for the information they provide to be attributed to them and are cited accordingly. 2The Seanad by-election: building a theoretical framework Overall, there were 17 candidates in the by-election, and while we did contact all of these candidates, we secured interviews with 6 candidates.Despite not all candidates agreeing to be interviewed, our interviewees are representative of the campaign overall, given that it includes candidates with the highest vote shares (including the successful candidate, Tom Clonan), those with the among the lowest vote shares (e.g.Michael McDermott), and those near the middle of the vote distribution (e.g.Ray Bassett).The breakdown of the by-election results can be found in the Supplementary Materials.Our interviewees are also representative of candidate experience of Seanad campaigns, as they include candidates that have run in this constituency before (e.g.Tom Clonan) and those who have not (e.g.Ursula Quill).This variation ensures that we are able to understand how the candidates generally understood the election and sought to campaign, without our results being biased by the strategies of either the most or least successful and/or experienced candidates.
To start with, the evidence suggests that there was little updating when it came to issue saliency as all candidates entered to focus on a particular issue throughout the campaign.One respondent stated that they entered the race to focus on mental health, and throughout the interview they made it clear that they did not amend any aspect of their strategy (Interview A).Similarly, other candidates made it clear that they entered the race to speak about certain issues, and did not seek to change what they focused on during the campaign (Interview with Quill).The winning candidate, Tom Clonan, was motivated to run in the by-election due to his son's disability, while also highlighting his experience in pushing for reforms in multiple areas (Interview with Clonan).While candidates prioritised certain issues from the beginning, they also faced considerable uncertainty as to the final result, with Tom Clonan being confident that he would not win, given his experience in the 2016 and 2020 Seanad elections (Interview with Clonan).
The candidates' statements that they did not change their strategies or issue priorities during the campaign is corroborated by their own campaign literature, which is contained in the Supplementary Materials.For example, prioritising his status as an independent candidate, with no links to a political party, was a major focus for Ray Bassett in both his campaign literature and interview, while Tom Clonan's campaign literature also focused on disability rights and his involvement in pushing for reform.Similarly, Ursula Quill focused on her positions on the European Union (EU) and direct democracy in both her interview and campaign literature.Since the campaign literature was developed before the campaign began, while the interviews took place after the election, the consistency in their answers across these different mediums provides confidence that the candidates did not update their issue saliency during the campaign.This can also be identified in candidates' social media campaigns.Hugo McNeill's campaign was centred on three core issues, and these were highlighted at the beginning and the end of the campaign in much the same way: 'If elected to the Seanad, my top priorities are: . To help transform the lives of people with intellectual disabilities .To use my experience to be a strong voice for business .To help repair British-Irish relations' (MacNeill, 27 February 2022) 'If elected I want . To help transform the lives of people with intellectual disabilities .To use my experience to be a strong voice for business .To help repair British-Irish relations' (MacNeill, 23 March 2022) Ade Oluborode, whose campaign highlighted her status as a minority candidate, described herself as the 'embodiment of Equality, Diversity, and Inclusion' in an early campaign posts, and continued to emphasise this point in the days before polls closed (Oluborode, Facebook, 28 February 2022;Oluborode, Facebook, 29 March 2022).In both of these cases, what is notable is not only that the issues prioritised by these candidates did not change, but also that the wording is also almost identical, highlighting an absence of updating of both the substantive content and its delivery.
Furthermore, the candidates' campaign leaflets in the Supplementary Materials suggest that they focused on one or two key issues and their own suitability for raising their saliency.In particular, candidates often discussed climate change, healthcare, and education in their leaflets.However, the discussion of these issues generally focused on demonstrating why that issue is important to themselves and to society more generally.Consequently, the leaflets tended towards valence competition (Petrocik, 1996) in that they attempted to demonstrate the candidates' ability to use the Seanad as a platform to raise the saliency of certain important issues.In contrast, it was rare for candidates to develop specific positions on the economic leftright or cultural liberal-conservative dimensions, and to compete over those positions.While this was due to the low salience of economic issues in this by-election, the lack of positional competition on the highly salient liberal-conservative dimension is likely due to the liberal orientation of the constituency.
Overall, a clear picture emerges of candidates failing to update either their campaign strategies or issue focus.Updating one's campaign strategies is a potentially costly and risky strategy as resources, which may not be perfectly transferable, will need to be diverted to undertake this new strategy, and it may take time to implement.For example, switching between ground wars and air wars can be costly, especially since they will require different sets of resources, with ground wars being more activist-intensive and air (or social media) wars benefiting from higher levels of capital (Gibson & McAllister, 2015).Furthermore, in addition to the potential financial cost of updating campaign leaflets, updating one's issue focus may also carry the risk of creating uncertainty around the party's message, or causing voters to perceive the party as inconsistent (Janda, Harmel, Edens, & Goff, 1995;Somer-Topcu, 2009).However, updating one's issue focus may be less costly for independents than parties, given their control over their own message and lack of party coherence and branding requirements.This is especially the case if independents only need to adapt their online and social media to convey this new message.Thus, updating is likely to only occur when the electoral actor has enough information to suggest that the likelihood of securing victory increases more than the costs of this change.However, in an environment with as little information as the Seanad by-election, candidates had no way to acquire the additional information that updating would require, and thus failing to update was actually a rational strategy for candidates to pursue.
However, a lack of information can be driven by a variety of factors, and we find that it was driven by a lack of information about both voters and the other candidates.Firstly, candidates tended to decide on their strategy to attract voters and media attention at the beginning of the campaign, and stuck to that strategy throughout.Overall, most candidates decided to prioritise their social media campaign, with some relying extensively on social media accounts, personal websites, and blogposts (Interview A; Interview with Bassett; Interview with McDermott; Interview with Quill).Some candidates also sought to convey their message through local newspapers (Interview A) or press releases (Interview with Quill), and nearly all candidates posted leaflets to voters.Indeed, Tom Clonan attempted to physically canvass voters in 2016, but the low odds of finding an eligible voter meant that he decided it was not a 'particularly efficient way of trying to connect with' voters (Interview with Clonan).Somewhat differently from the other candidates, Ray Bassett decided to place considerable focus on attempting to engage the diaspora and persuade them to vote (Interview with Bassett).
While candidates tended to develop one strategy and stick with it, all our interviewees openly discussed the difficulties they faced in trying to determine what voters' preferences were or how to campaign for their vote.Indeed, one of the first points that some respondents raised when asked how this campaign differed from other campaigns was the difficulty of physically canvassing voters by knocking on doors (Interview A; Interview with Quill).Given the small numbers registered to vote, and the trend of recent graduates putting down their parents' house and failing to update the register, it is 'very rare that you are actually talking to a voter, and then you have huge distances to travel if you're not in certain parts of Dublin' (Interview A).This difficulty in accessing voters and having conversations with them translates into a lack of understanding about voters' concerns, especially compared to normal territorial-based constituencies, where 'you would naturally know what people are interested in.Because even if people aren't voting, as long as they answer the door, they'll tell you what they're interested in, in that area' (Interview A).Indeed, one respondent stated that the electorate is 'incredibly nebulous in every single way' (McDermott Interview), while another described attempting to determine the voters' preferences as 'shooting in the dark' (Interview with Quill).Thus, candidates' knowledge of the voters seems to have been limited to what they could infer about the fact that they were all university graduates -namely, that the electorate was more liberal and affluent than the electorate at a general election.Beyond this, candidates were dealing with an electorate that was geographically diverse and whose political priorities, particularly in terms of what they expected from the Seanad, were unclear.
This was compounded by the lack of media coverage or any polling on the election.For example, one respondent said that the election was 'very low in the pecking order for the media' (Interview A).Similarly, others suggested that there was very little media coverage, and the media coverage that did exist focused on Hugo MacNeill as the probable winner (Interview with Bassett; Interview with Clonan).Indeed, some candidates suggested that media attention only began to increase as this campaign drew to a close and the vote counting began (Interview with Quill).Thus: . Hypothesis 1: The less information around voting intentions, the less likely electoral actors are to update their preferred issue's saliency.
Finally, candidates knew relatively little about other candidates and their preferred issues, with the number of candidates reducing the ability to track one's competitors.This lack of knowledge was especially due to the unusually large number of candidates, with new candidates continually emerging over the long run up to the campaign, and candidates being unsure as to how this would impact voting behaviour and the dynamic of the campaign (Interview with Quill).Indeed, while some candidates knew little about the other candidates (Interview A), others were somewhat familiar with the other candidates, but found trying to follow that many other campaigns overwhelming.For example, Clonan chose to ignore the other campaigns as 'trying to monitor their campaigns induced panic and anxiety, and stressed me' (Interview with Clonan), while Quill also decided to prioritise her own message as attempting to follow and respond to so many campaigns means 'you could get caught up in the noise of that instead' of promoting your own message.One candidate noted that a member of their campaign team prepared profiles on the other candidates, but they did not read them, choosing instead to focus on their own issues (Interview B).This lack of interest in the other candidates was also reflected by the other interviewees (Interview A; Interview with Bassett).
Thus, even if this information was easily available by examining the social media profiles of other candidates, it was often disregarded by candidates who did not want to sacrifice their own message to directly engage with others in the context of a very crowded field.While some candidates may have had information about their competitors, the sheer number of candidates and issues raised by the campaigns made it difficult to use this information effectively and determine how to best respond.Indeed, it was striking that all of our interviewees raised the same point about paying minimal attention to other campaigns.This gains additional plausibility from the consistency in candidate messages throughout the campaign, which we would not expect if candidates were reacting to each other's campaigns.This also highlights an additional effect of the lack of polling data: without knowing who the biggest electoral threats were, it is difficult to establish which campaigns should be engaged with when time and resources are limited.Again, we see a picture emerge of a campaign where candidates were each dedicated to a few core issues and did not update their campaign in predictable ways, due to the difficulty of establishing priorities.
Reflecting this lack of engagement with other candidates, one of the main features of this race was the range of issues addressed by the different campaigns.Indeed, a number of them had at the core of their campaign issues that were hardly, if ever, mentioned by the other candidates.Hugo MacNeill rigidly structured his campaign around his three key priorities of improving the lives of people with intellectual disabilities, advocating for the interests of businesses, and working to repair British-Irish relations (Hugo MacNeill, Facebook, 15 February 2022).None of these issues featured in any of the other campaigns.Paula Roseingrave's campaign had a strong focus on tackling sexism, with a particular focus on the problem of sexual harassment of university students: 'Trinity and all universities in Ireland have to prohibit sexual/intimate relationships [between students and faculty] outright' (Roseingrave, Facebook, 21 March 2022).Again, neither of these issues were the focus of any of the other campaigns.Overall, then, this was a campaign in which candidates were largely talking past each other.Thus: . Hypothesis 2: The higher the number of effective competitors, the less likely electoral actors are to update their preferred issue's saliency, especially as the diversity in issues prioritised by competitors rises.
It should be noted that the election was conducted under the alternative vote (AV) system.Given the crowded field, candidates would not have expected anyone to win on the first count, and thus transfers would be crucial in determining the outcome.This proved to be the case, as Tom Clonan was not elected until the 16th count, while Hugo MacNeill was eliminated on the 14th count despite leading for the first nine.In this case, the rational strategy would be for candidates to run a broad campaign that was capable of attracting transfers.Thus, our findings of issue divergence, static campaigns, and failing to seek out information on other campaigns may appear surprising.However, as previously discussed, the lack of information available about voters beyond their universal characteristics of being university graduates meant that it would have been unclear that issues adopted by other campaigns would be successful in attracting transfers.It also meant that the best way to attract transfers was to develop a catch-all campaign that focused on valence issues, as we have observed.
The election was ultimately won by independent candidate Tom Clonan, who had previously run for the same constituency in 2016 and 2020.Furthermore, the other leading candidates had all previously been active in politics and/or media.For example, Hazel Chu had been the Lord Mayor of Dublin, while Hugo MacNeill was a former Irish rugby international player and had run in this constituency before.However, while name recognition was clearly important, it is difficult to infer much about the effectiveness of the strategies adopted by the candidates, for two main reasons.Firstly, the significant similarities in campaign strategies, with most choosing to focus on a handful of specific policy areas and their own personal attributes, meant there was little basis on which to conclude that aspects of certain campaign strategies were more effective than others.Secondly, given the absence of any voter surveys or opinion polls in this election, there is no data indicating the bases for voters' choices, or how their preferences changed across the course of the campaign.

Refining our theory
Our theoretical expectation is that the less information electoral actors possess, the less likely they are to update their issue saliencies.In particular, the information environment faced by electoral actors is primarily shaped by how much information they possess on both voters and their competitors.However, we now seek to probe other ways electoral actors may respond to low information environments and how this theoretical framework could be refined to apply to party-dominated electoral contests.
Firstly, there are some alternative ways for both electoral actors and voters to respond to low information environments.To start with, whenever actors are public figures with past policy positions (as most electoral actors tend to be) and they know little about their audience, then they may face incentives to simplify their communication and thought processes in the pursuit of consistency and the development of justifications for their positions, thereby reducing their cognitive effort (Tetlock, Skitka, & Boettger, 1989).Thus, the less information electoral actors possess, the more likely they are to simplify their communications.However, while this may influence how electoral actors communicate with voters, the lack of information also means that they will still face the strategic incentive to be risk-averse and avoid altering the saliency of particular issues.This is because, where information is limited, changes to issue saliencies could just as easily harm the actor's electoral prospects as aid them, thereby creating incentives to retain current saliencies.
Alternatively, electoral actors could respond to low information environments by increasing their focus on heuristics relating to the candidate's socio-demographic profile (e.g.their gender, class, race, etc.) or personal traits (e.g.competence, trustworthiness, etc.).Indeed, there is considerable evidence that these heuristics can significantly influence vote choice (Bernhard & Freeder, 2020;Johns & Shephard, 2007).However, while electoral actors may increasingly prioritise these factors in low information contests to engage voters, this does not influence the choice facing these actors about what issues to prioritise.Indeed, the increasing prioritisation of heuristics reduces voters' cognitive effort, but does not necessarily reduce the difficulty of the decision facing electoral actors about what issues to prioritise.
Overall, both cognitive simplification and an increasing focus on heuristics may be particularly useful in reaching voters and persuading them to vote for that electoral actor in low information contests.However, neither directly affect the central problem facing electoral actors: whether they should stick with their current issue saliencies, or adjust them in an effort to be closer to the voters' issue saliencies.Indeed, it may be that electoral actors can pursue all three strategies simultaneously: keep their current issue saliencies; simplify how they discuss each issue; and increase their focus on heuristics.Thus, these two alternatives do not undermine our theory, but future tests can examine if they also occur.
Secondly, focusing on parties as the relevant actors necessitates some refinements to our theoretical framework.Given their role as preference aggregators and imposing both ideological and organisational coherence (Aldrich, 2011;Kölln, 2015), the introduction of parties to our theoretical framework will necessarily narrow the range of information levels facing electoral actors and voters.This is because parties will impose a minimum information level on the electoral contest, in two ways.Firstly, by adopting positions on important cleavages, especially the left-right cleavage (Green-Pedersen & Little, 2023), parties can simplify the contest by defining what issues the contest is about and what their position on those issues are.Secondly, parties also create information through the use of both their own resources and affiliated groups (e.g.think tanks, trade unions, business groups, etc.) to carry out polling, research their competitors, and develop policy initiatives.Thus, in a party-dominated contest, the central party and its candidates are going to have more information about voters and their competitors than independents.However, there is still considerable scope for variation in the information environment, especially given the declining links between parties and civil society (Best, 2011;Mair, 2013;van der Meer et al., 2015) and the emergence of new issues and cleavages & Marks, 2018; Kriesi et al., 2006).
Furthermore, in a party-dominated contest, different parties will have different goals, which may influence what information they seek and how they respond to it.In particular, the more policy-seeking the party, the less important the information environment as it may wish to avoid any compromising on its principles.This incentive may be most important for small and relatively extreme flag-bearers, which seek to ensure that their ideology gains some representation, but may not want to compromise on their ideology beyond that.Thus, any test of our theory in a party-dominated contest would need to account for party size and ideological extremity, with low information potentially having a greater effect on larger and more centrist parties.
Finally, an important caveat is that the precise electoral arena will influence the information level, since low saliency arenas will inevitably result in low information levels as both voters and electoral actors reserve their energies for other, more important contests.For example, European Parliament elections are often seen as low salience, second-order elections (Hobolt & Wittrock, 2011;Reif & Schmitt, 1980), and we may thus expect these elections to be consistently lower salience than national elections.Consequently, our theory would predict different strategies depending on the electoral arena given the variation in information levels and any test of the theory will need to account for the electoral arena in which the contest is occurring.

Conclusion
To remain electorally relevant, electoral actors in the form of either parties or candidates need to be adaptable, willing to change over time to 'win' the competitive struggle for votes.This requirement is most intense during election campaigns, when electoral actors are trying to convince voters of the merits of voting for them, and face the most immediate rewards or penalties for competing effectively.One way that electoral actors may seek to adapt during campaigns is by adjusting the saliency they attach to issues.In the literature, parties are commonly seen as torn between two different imperatives during campaigns: the need to highlight issues that they 'own' and are seen as competent or trustworthy on; and the need to appear responsive to highly salient issues that matter to voters (Abou-Chadi et al., 2020;Seeberg, 2022).However, to choose which strategy is most appropriate, electoral actors need to have enough information about their electoral environment to determine whether they need to update the saliency they attach to different issues.Despite the importance of this underlying information level, it is unclear what forms of information matter most for electoral actors, and how they will respond to it.Thus, in this article we sought to answer the following research question: how do electoral actors respond to different forms of information in election campaigns?
Answering this question is important because, while there is a considerable literature on parties' strategic behaviour, both in terms of the party competition (spatial and issue ownership) and campaign literatures, no study, as far as we know, has sought to examine what influence low information environments will have on electoral actors' strategic behaviour during election campaigns.Indeed, given that electoral actors are torn between the desire to prioritise their preferred issue and appearing responsive to voter concerns, this lack of information can have significant implications for campaign strategy.Furthermore, this article is also important in highlighting the potential effect of campaign dynamics on issues discussed during campaigns and subsequently those addressed in government, insofar as campaign promises made as a result of inter-party dialogue affect subsequent policy.To the extent that electoral actors seek to represent certain interests during the campaign separately from any direct influence on policy if elected, the effect of information in a campaign thus matters for the quality of democratic representation.
This article uses qualitative process tracing to examine the 2022 Seanad by-election for the University of Dublin constituency.This by-election counts as an extreme case due to the unusually large number of candidates, the lack of party influence or even incumbency effects, and the lack of information about voter preferences.Given the utility of focusing on extreme cases for theory building (Gerring & Seawright, 2007), we use this case to probe what features of the campaign were most determinative of the information level, and how candidates responded to this.This analysis was based on interviews with the candidates themselves, which was triangulated using the candidates' campaign literature and social media posts.Overall, we develop two hypotheses, based on the premise that parties are least likely to update issue saliencies at low levels of information.Firstly, we expect that an important form of information concerns voters, such that lower information around voters' intentions result in lower levels of updating.Secondly, we expect that another important form of information concerns the party's competitors, with a more crowded field of competitors creating more uncertainty and thereby reducing updating, especially as the range of issues discussed by competitors increases.Overall, the less information electoral actors have, the more likely they are to continue prioritising their 'owned' issues and ignore their competitors.This study underlines the risk-aversion that often characterises electoral actors in campaigns, indicating that they persist with their original strategy in the face of limited information, rather than expending resources in an effort to overcome this hurdle.
However, by definition, extreme cases are unrepresentative of the wider population of cases, which has significant implications for the external validity associated with the findings drawn from these cases.While our theoretical framework is most immediately applicable to other independent-centric races, it can also be extended to party-dominated races, given that it focuses on general goals of electoral actors and any form of electoral actor will be required to consider both voters and their competitors to be successful.However, we developed how our theoretical framework can be refined to make it more applicable to party-dominated contexts, and some other ways that electoral actors and voters may respond to low information environments.
There are, however, some limitations to this article, which provides avenues for further research.Firstly, the lack of party structuring of the election is not a common feature of elections.The presence of parties and the affiliation of candidates with parties provides a minimum level of information by providing a heuristic for voters, meaning that even where voters are unsure about individual candidates' position, they can infer some information from their party affiliation.More importantly for the study of how parties and candidates deal with uncertain environments, party structuring often makes it clearer what the dividing lines of the campaign are, and where candidates generally sit on particularly salient issues.Thus, our theory needs to be tested on other cases, where parties are more prominent.Secondly, our research focuses explicitly on the campaign period itself, but parties may also face uncertainty outside of these periods.Outside the campaign period, parties may face stronger incentives to prioritise the issues that they own in order to shift the political debate onto their preferred ground, while trying to ignore other issues.Consequently, it would be important to test our theory outside of the campaign period, as parties may act differently in these contexts.

Notes
1. 20 names appeared on the ballot, as Karin Dubsky was deemed ineligible after the ballots had been printed upon discovery that she was not an Irish citizen.
Ballots listing Dubsky as a first preference were excluded, while those listing her as a lower preference were transferred to the next highest preference.Frances Donnelly also effectively (but not formally) withdrew as he was concentrating on his campaign for the Labour Panel and appealed for people not to vote for him.2. These interviews were also given ethical approval by the relevant School.