Dowding, Hindmoor and Martin’s (hereafter DHM) insightful article emphasises what they frame as a limitation of the Comparative Agendas Project (CAP) data collection effort. Specifically, they state that, although CAP collects information about the attention that political parties (and governments) direct towards different issue areas (the economy, environmental issues and so on) and in different venues (parliamentary questions, proposed legislation, government budgets, etc.), CAP does not record the positions parties/governments stake out in these issue areas: “PAP/CAP generally measure policy attention – what is being discussed in various forums – rather than what government is actually doing”. Although this is a fair point, I want to emphasise – from the perspective of someone who studies how voters infer and react to (their perceptions of) parties’ issue positions – that parties’ issue attention and their issue positions are closely intertwined. Thus, the CAP data on party/government issue attention offer promising opportunities to understand how voters infer party positions – in particular parties’ left-right ideologies, and also how voters respond to the party positions they perceive.

To address the second point, first, the standard approach to analysing how voters evaluate parties’ issue positions – one that is associated with the spatial model of elections (e.g. Downs Reference Downs1957) – is that voters evaluate a focal party more positively as that party’s positions more closely resemble the voter’s positions along a set of relevant policy dimensions (all else equal), i.e. that voters employ a proximity-based policy metric.Footnote ² However, conditional on policy proximity, the effect of the party’s position on the voter’s party evaluation increases with the salience of the focal issue dimension: the more salient the issue area, the more the voter approves of parties that share his/her issue position and the more he/she disapproves of parties that do not. Thus, even if voters evaluate parties based on position, the salience voters ascribe to party positions is crucial to these evaluations. Moreover, there is extensive evidence that, although the mass public’s pre-occupation with an issue can prompt party elites to emphasise the issue (e.g. Spoon and Klüver Reference Spoon and Klüver2014; Wagner and Meyer Reference Wagner and Meyer2014), voters reciprocally take cues from parties about which issues to prioritise – in that increasing party attention to an issue prompts increased attention to this issue in the mass public (e.g. Green and Hobolt Reference Green and Hobolt2008; Hobolt et al. Reference Hobolt, Klemmensen and Pickup2009) and especially among the party’s core supporters (e.g. Neundorf and Adams (Reference Neundorf and Adams2014)). In this regard, the CAP data provide valuable cross-national and longitudinal data on how much attention each political party (and government) directs to different issue areas, and how this attention distribution varies across different venues (parliamentary questions, proposed legislation, government budgets and so on) – i.e. the CAP data record the distribution of party (and government) issue attention, which mediates how strongly voters react to party positions.Footnote ³

Second, it seems plausible that the CAP data on parties’/governments’ distributions of issue attention, even without details about the specific policies parties implement (or propose) in each issue area, can be used as a component of a procedure for estimating parties’ ideological positions; moreover, that rank-and-file voters use the distribution of parties’ issue attention as a heuristic to infer party ideologies. In this regard, the well-known Comparative Manifesto Project (CMP) codings of the left-right tone of parties’ election manifestos (Budge et al. 2001; Volkens et al. 2014) consider not only parties’ stated positions along dimensions where parties may directly oppose each other (such as staking out favourable or unfavourable stances vis-à-vis the military, economic protectionism and so on), but also the distribution of attention or emphasis that parties place on universally shared objectives (conditions), such as clean air and law and order that all citizens value, but which left- and right-wing parties may prioritise differently. With respect to the attention-based component of these codings, the CMP coders identify quasi-sentences in the party’s manifesto that endorse specific, universally shared objectives (e.g. clean air, peace, law and order, etc.), and the distribution of the party’s attention to these different objectives is then combined with the parties’ stated positions on issues where parties may disagree (e.g. the value of central economic planning, multiculturalism, etc.) to create a summary measure of the left-right tone of the party’s manifesto. Manifestos that invoke widely shared objectives typically associated with the right (such as law and order) are coded as more right wing than manifestos that invoke objectives associated with the left, such as clean air.Footnote ⁴ Thus, the distribution of party attention to different, universally valued goals is a component of the CMP codings of the left-right positional tone of party manifestos.Footnote ⁵ Although these coding procedures have been criticised (e.g. Benoit et al. Reference Benoit, Laver and Mikhaylov2009), the CMP left-right manifesto codings match up well with both experts’ and rank-and-file voters’ perceptions of party positions (see e.g. Bakker et al. Reference Bakker, Hooghe, Marks, de Vries, Edwards, Steenbergen and Vachudova2010; Dalton et al. Reference Dalton, Farrell and McAllister2011).Footnote ⁶ This suggests that, regardless of whether rank-and-file voters (and experts) should infer parties’ ideological positions from their distribution of attention to different goals, in practice, voters do make such inferences. This raises important questions about voter perceptions of parties’ ideological positions, which can be addressed via analyses of the CAP data on party issue attention, calibrated against survey data on citizens’ party placements (discussed below). In particular, given that voters infer parties’ ideological positions in part from the distribution of their manifesto-based attention to different, universally valued objectives, do voters similarly infer party ideologies from their attention distribution in other forums including parliamentary questions, press releases, proposed legislation, government budgets and so on? And assuming the answer is yes, how do voters then weigh parties’ rhetoric and behaviour across these different forums as they arrive at summary estimates of parties’ ideological positions? This latter question is interesting when considering the relative weight voters attach to forums involving action, such as parties’ legislative proposals and government budgets, as opposed to forums that involve words, such as parliamentary questions, party press releases and manifestos: for while parties’ invocations of universally valued goals like clean air and law and order will invariably be positive, their legislative proposals and government budgets will involve a positional (directional) component of either increasing or scaling back the public resources devoted to achieving such goals, a point that DHM insightfully emphasise.Footnote ⁷

Although it is an open question whether rank-and-file voters (and experts) should weigh parties’ attention distributions across different, universally valued objectives, when inferring parties’ ideologies – and this question is also well worth studying – I suggest we evaluate whether voters and experts do in fact employ this attention-based heuristic. In terms of the CAP data, this entails a research agenda assessing whether, all else equal, voters and experts perceive political parties as more leftist when their election manifestos, parliamentary questions and executive speeches pertain disproportionately to issues or objectives that traditionally pre-occupy the left (such as full employment, worker safety and protection and environmental protection), and more right wing when parties’ attention pertains to issues such as inflation and law and order that are typically associated with the right.Footnote ⁸ Moreover, of course one can carry out parallel analyses with respect to how voters (and experts) respond to the distribution of party issue attention across venues that involve action, including their legislative proposals and government budgets. Such a study would involve calibrating the CAP data on the distribution of party/government issue attention against data on rank-and-file voters’ party placements – such as data from the Comparative Study of Electoral SystemsFootnote ⁹ (CSES) and data on experts’ left-right party placements, such as the Chapel Hill Expert Survey (CHES) data.Footnote ¹⁰ The fact that both the CHES and CSES data offer longitudinal, cross-nationally comparable measures of experts’ and voters’ left-right party placements across a large set of party systems would enhance the value of this project, for this allows us to evaluate whether, cross-sectionally, voters infer ideological differences between parties based on these parties’ differing attention profiles, and also whether, over time, voters infer temporal changes in parties’ ideologies based on changes in their attention profiles.

The approach outlined above connects with growing scholarly interest in combining studies of parties’ issue positions and their issue emphases (e.g. De Sio and Weber Reference De Sio and Weber2014; Guinaudeau and Perisco Reference Guinaudeau and Perisco2014), along with studies on how voters react to governments’ issue agendas (Bertelli and John Reference Bertelli and John2013). Moreover, this discussion should not detract from DHM’s valuable point that the CAP data record the distribution of party (and government) issue attention across different forums, and that attention is not synonymous with position. As the CAP researchers would surely agree, it would be a worthwhile exercise to devise explicitly positional measures for the stances parties stake out with respect to issues on which parties may disagree (such as their positive or negative stances towards the welfare state, protectionism, the European Union and so on), although as Bryan Jones delineates in his essay in this volume, this exercise raises challenging measurement issues (and would also surely require a vast investment of resources). My point is simply that, in studies of how voters react to parties’ issue-based behaviour (whether this behaviour involves words or deeds), issue attention and position are closely intertwined. In compiling such a rich source of data on the distribution of parties’ and governments’ issue attention, CAP provides us with tools that may greatly enhance our understanding of the effects of party positions.