Interactions Of Issue Salience And Issue Ownership

While socioeconomic background is traditionally considered as reliable predictors of election outcome,the attention of election study has been gradually attracted to another determinant:voters' perceptions of important issues in electoral periods.Research in this field has become exceptionally convenient and meaningful with the complexity of electoral environment in recent decades and the increasing tendency of political parties to conduct electoral campaign using issues as both assets and ammunitions.One of the most influential research results in the literature of issue voting is Budge and Farlie's issue saliency theory.Building on Down's assumption of a uni-dimensional spatial model.Budge and Farlie structured bidimensional models using linear regression with two elements identified in the behavior of issue voting:issue ownership and issue salience.The effect an issue has on election outcome,according to Budge and Farlie,depends not only on the way to which the issue favors(issue ownership),but also on the magnitude of influence the issue is able to induce(issue salience).The theoretical framework of Budge and Farlie heralded a line of research that started to acknowledge the effect of issue salience.However,scholars have been contributing to the literature of issue saliency in an unordered and somehow whimsical way.The definition and measurement of ownership and saliency were vastly disagreed in practice,which was further complicated by a stealthy transition from aggregate-level research to individual-level analysis in election studies.The advancement of statistical method enabled more well-grounded,factual analysis,but also led to amorphicity and divergence in literature.In response to the paucity of originalism in existing literature,this dissertation advocates a return to Budge and Farlie's theory of issue saliency and extrapolates that logistic regression provides a tenable reproduction of Budge and Farlie's model.with special attention to the use of interaction effect between issue ownership variables and issue salience variables;the interaction represents the interplay between saliency and ownership,and how they jointly affect vote choice.In the original models of Budge and Farlie,issue salience aflccts the vote share gained by each party through some kind of mediating effect on the magnitude ofvarious issues.Logistic regression arguably provides similar structure,with issue saliency variables interacting with ownership variables and mediating its effect on the probability of vote choice.In this dissertation,four logistic regression models are built based on Britain's general elections of 2005,2010,2015 and 2017(2019 not included for want of information),with interaction effects between issue ownership and issue saliency,and demographic variables as the proxy of basic vote-the non-issue-related determinants of vote choice.The data used are provided by British Election Study,a series of reputed surveys conducted regularly each year to enquire the political opinions and social attitudes of Britons.The results support the hypothesis that interaction effect in logistic regression provides moderately tenable reproduction of issue ownership and issue salience,while demographic variables fail to form predictable basic vote.To be specific,interaction terms in logistic regression could model the pattern that the effect of issue ownership becomes stronger,provided that the issue is considered salient:positive evaluation of a party's performance on an issue would become more influential in vote choice probability if the issue is identified as the most important issue(Mil)facing the country.Most of the issue ownership variables in the models predictably achieved statistical signifcance,while only some of the interaction effects proved statistically significant.This is understandable as the binary saliency variables are not fine-grained enough,and some of the corresponding ownership variables do not contribute to the models.The interactive pattern is especially significant on the issue chosen by the highest number of respondents as the Mil facing the country in each election Statistics in this dissertation is done using R,an open-source software environment for statistical computingThe findings of this dissertation also include the changing landscape of Britain's issue voting at aggregate level from 2005 to 2017.In 2005 when the issue of asylum seekers was the MII and owned by Conservatives,Labour won the election by grasping a host of only slightly less-salient issues related to cost of living and welfare.The 2010 election saw the rise of'Conservatives,who formed a coalition government with Liberal Democrats for owning the issue of immigration and economy-related financial crisis.while issues traditionally owned by Labour lost salience.In 2015,Conservatives was elected with a shared ownership of immigration issue with UK Independence Party,and a monopoly of the issue of economy;the issues owned by Labour,again,witnessed declined saliency with no sign of revival.In the 2017 general election which ended up with a hung parliament,Conservatives enjoyed the monopoly of Brexit issue,who somewhat eclipsed all other issues,while Labour's monopoly of basically all other welfare issues including NHS is also noticeable.For 2019,preliminary data shows that Conservatives won the election for owning the most emphatic MIIs,which are possibly Europe-or immigration-related.Reproducing Budge and Farlie's classic model of issue saliency,this dissertation enriches the literature of saliency theory,and reflects on the issue voting in Britain.The use of interaction effect is especially meaningful,as it sheds light on a potentially important element in the application of logistic regression on issue saliency theory.