Reflecting the broad scope and blind spots of research on VAAs, this Special Issue of Policy & Internet brings together VAA research articles not only from different continents (Europe, North America, and Oceania) but also from different research angles—articles which try to tackle some of the remaining puzzles. Of the five articles in this issue, three can be regarded as “traditional” VAA articles, in that they examine the effects of VAAs on electoral behavior, on general models of electoral choice, and (turning to the question of VAA design) the effects of different answer scales on the advice given to users. These can be considered “traditional” inasmuch they focus on the tools and their effects, as well as on their mode of operation. They are also innovative, as they apply new methods in partly new contexts to address these questions. In her article in this issue, Mahéo (2016) speaks to a consolidated subfield of VAA research—the effect of these tools on users’ voting intentions. She tackles one of the crucial methodological issues in this strand of literature, namely selection bias, through a full-fledged experimental design in the context of the 2014 Quebec Elections. Considering the multidimensional political space of this political entity, Quebec could indeed serve as an instructive context for testing the effects of VAAs. The results of her experiment indicate—at first glance—that the users of VAAs are more likely to form electoral preferences than non-users. However, she specifies this observation along different user groups regarding age, formal educational attainment and political interest. For most of these subgroups, she could detect only short-term effects of using a VAA. That is why she concludes that VAAs involve users cognitively, but do not make them effectively alter their voting decision. Ladner (2016) examines the question of the effects of VAA on voting behavior through the lens of normative democratic theory. Sitting at the crossroads of the literature on democratic representation and that on changing models of electoral choice, his article investigates the possibility that widespread VAA usage can be held responsible for fostering promissory representation and the delegate model of political representation—at the core of which stands the idea that candidates should keep their electoral promises. By relying on an online survey conducted among the users of the Swiss VAA Smartvote, he indeed finds evidence that issue voters are not only more likely to follow the recommendations issued by the VAA, but that they also expect elected candidates to keep their promises to a higher extent compared with partisan voters. The article by Rosema and Louwerse (2016) turns attention to VAA methodology, by analyzing a rather underresearched topic in the literature, namely, the effect of different answer scales on the voting advice provided to users. Their data derive from a VAA developed for the 2014 Dutch local elections. As is the case with statement selection, party coding strategies and the choice of the matching algorithm, their article shows that answer scales matter—except for voters with an extremist response style. On the basis of their findings, Rosema and Louwerse (2016) discuss the practical implications for VAA designers, suggesting that the voting advice should be presented as a preference list, rather than focusing on the “best match.” The final two articles in the issue illustrate how VAA-generated data sets can be used to address general questions of political science. Data on party positioning can be used to map party systems and to identify the dimensionality of the political space. VAA user data can be used in studies of how attitudes to certain policy issues influence political behavior and electoral decisions. The article by Wheatley (2016) addresses the question of how to survey the political space in England, by overcoming the traditional focus on a left–right cleavage. He argues that in many European societies a new ideological cleavage between “cosmopolitans” and “communitarians” (based on the “winners” and “losers” of globalization) has emerged that draws on cultural rather than economic issues. To test this assumption, Wheatley (2016) draws on party position data and user data generated by two VAAs deployed in England in 2014 and 2015. By identifying latent dimensions and mapping the positions of parties and party supporters, he concludes that the political space in England is defined by two main dimensions: an economic left–right dimension and a cultural communitarian–cosmopolitan dimension. The article by Carson, Dufresne, and Martin (2016) uses VAA-generated data to answer the question whether and how attitudes to specific issues have an effect on the voting decision. For their case study, they turn to the 2013 Australian elections and to the topic of asylum and immigration, which played an important role in it. To answer their question they draw on data provided by the Australian Vote Compass, a VAA implemented before the elections. Using these data they were able to include about 438,000 cases (voters) into their calculations—as they argue, a survey of “unprecedented size in Australia.” As well as working with this “big” VAA data, Carson et al. (2016) engaged in media monitoring and content analysis of a party's press releases in order to measure the salience of this issue among the public, the media and the parties. They find that the attitudes of voters toward asylum seekers had an impact on voting intention—particularly in marginal electorates and in the group of voters who care most about this issue. These last two articles in particular show nicely how much potential lies in VAA-generated data—a potential that has yet to be exhaustively exploited. Much more research could be done with existing and future VAA data if it were processed appropriately. For example, if collected, standardized, and compiled systematically, the multitude of party positions documented in VAAs could constitute a valuable database for studies on party positioning over time and across countries, allowing for longitudinal as well as cross-country comparative analyses of the characteristics and developments of parties and party systems. Similarly, the collection and harmonization of user data could create a sound and far-reaching database on public opinion on certain policy issues—again allowing for analyses over time and across countries. In this respect, research on VAAs has just started.
Garzia, D., Marschall, S. (2016). Research on Voting Advice Applications: State of the Art and Future Directions. Hoboken : John Wiley & Sons.
Research on Voting Advice Applications: State of the Art and Future Directions
Diego Garzia;
2016
Abstract
Reflecting the broad scope and blind spots of research on VAAs, this Special Issue of Policy & Internet brings together VAA research articles not only from different continents (Europe, North America, and Oceania) but also from different research angles—articles which try to tackle some of the remaining puzzles. Of the five articles in this issue, three can be regarded as “traditional” VAA articles, in that they examine the effects of VAAs on electoral behavior, on general models of electoral choice, and (turning to the question of VAA design) the effects of different answer scales on the advice given to users. These can be considered “traditional” inasmuch they focus on the tools and their effects, as well as on their mode of operation. They are also innovative, as they apply new methods in partly new contexts to address these questions. In her article in this issue, Mahéo (2016) speaks to a consolidated subfield of VAA research—the effect of these tools on users’ voting intentions. She tackles one of the crucial methodological issues in this strand of literature, namely selection bias, through a full-fledged experimental design in the context of the 2014 Quebec Elections. Considering the multidimensional political space of this political entity, Quebec could indeed serve as an instructive context for testing the effects of VAAs. The results of her experiment indicate—at first glance—that the users of VAAs are more likely to form electoral preferences than non-users. However, she specifies this observation along different user groups regarding age, formal educational attainment and political interest. For most of these subgroups, she could detect only short-term effects of using a VAA. That is why she concludes that VAAs involve users cognitively, but do not make them effectively alter their voting decision. Ladner (2016) examines the question of the effects of VAA on voting behavior through the lens of normative democratic theory. Sitting at the crossroads of the literature on democratic representation and that on changing models of electoral choice, his article investigates the possibility that widespread VAA usage can be held responsible for fostering promissory representation and the delegate model of political representation—at the core of which stands the idea that candidates should keep their electoral promises. By relying on an online survey conducted among the users of the Swiss VAA Smartvote, he indeed finds evidence that issue voters are not only more likely to follow the recommendations issued by the VAA, but that they also expect elected candidates to keep their promises to a higher extent compared with partisan voters. The article by Rosema and Louwerse (2016) turns attention to VAA methodology, by analyzing a rather underresearched topic in the literature, namely, the effect of different answer scales on the voting advice provided to users. Their data derive from a VAA developed for the 2014 Dutch local elections. As is the case with statement selection, party coding strategies and the choice of the matching algorithm, their article shows that answer scales matter—except for voters with an extremist response style. On the basis of their findings, Rosema and Louwerse (2016) discuss the practical implications for VAA designers, suggesting that the voting advice should be presented as a preference list, rather than focusing on the “best match.” The final two articles in the issue illustrate how VAA-generated data sets can be used to address general questions of political science. Data on party positioning can be used to map party systems and to identify the dimensionality of the political space. VAA user data can be used in studies of how attitudes to certain policy issues influence political behavior and electoral decisions. The article by Wheatley (2016) addresses the question of how to survey the political space in England, by overcoming the traditional focus on a left–right cleavage. He argues that in many European societies a new ideological cleavage between “cosmopolitans” and “communitarians” (based on the “winners” and “losers” of globalization) has emerged that draws on cultural rather than economic issues. To test this assumption, Wheatley (2016) draws on party position data and user data generated by two VAAs deployed in England in 2014 and 2015. By identifying latent dimensions and mapping the positions of parties and party supporters, he concludes that the political space in England is defined by two main dimensions: an economic left–right dimension and a cultural communitarian–cosmopolitan dimension. The article by Carson, Dufresne, and Martin (2016) uses VAA-generated data to answer the question whether and how attitudes to specific issues have an effect on the voting decision. For their case study, they turn to the 2013 Australian elections and to the topic of asylum and immigration, which played an important role in it. To answer their question they draw on data provided by the Australian Vote Compass, a VAA implemented before the elections. Using these data they were able to include about 438,000 cases (voters) into their calculations—as they argue, a survey of “unprecedented size in Australia.” As well as working with this “big” VAA data, Carson et al. (2016) engaged in media monitoring and content analysis of a party's press releases in order to measure the salience of this issue among the public, the media and the parties. They find that the attitudes of voters toward asylum seekers had an impact on voting intention—particularly in marginal electorates and in the group of voters who care most about this issue. These last two articles in particular show nicely how much potential lies in VAA-generated data—a potential that has yet to be exhaustively exploited. Much more research could be done with existing and future VAA data if it were processed appropriately. For example, if collected, standardized, and compiled systematically, the multitude of party positions documented in VAAs could constitute a valuable database for studies on party positioning over time and across countries, allowing for longitudinal as well as cross-country comparative analyses of the characteristics and developments of parties and party systems. Similarly, the collection and harmonization of user data could create a sound and far-reaching database on public opinion on certain policy issues—again allowing for analyses over time and across countries. In this respect, research on VAAs has just started.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
