Measuring Parties’ Ideological Positions With Manifesto Data

Abstract

A B S T R A C TWithin the rapidly growing literature on positioning political parties along policy dimensions, the rich data series collected by the Comparative Manifestos Project (CMP) has been widely considered as the most systematic and objective source of information. For estimating parties' positions on the Left-Right dimension alone, several different methods have been proposed which make use of the CMP data. However, unless a new method is proposed, there will seldom be any attempt to check the robustness of the findings across different measurement strategies. In this article, we focus on the parties in Greece, which have been notoriously incorrectly positioned by the 'standard' method proposed by the CMP. We contrast the 'standard' method with various proposed alternatives and show that the latter outperform the former both in terms of face and convergent validity and in terms of reliability. In addition, we show that this cross-checking is essential, since different methods often lead to diametrically opposite results.KEY WORDS Ⅲ ideology Ⅲ left-right Ⅲ party manifestos Ⅲ reliability Ⅲ validity P A R T Y P O L I T I C S V O L 1 6 . N o . 4 pp. 427-450

Aggregation and Representation in European Parliament Party Groups

“…Comparisons of individual candidates and party programmes find that party programmes tend to determine the views of their candidates, thus justifying making the programme the unit of analysis . After reviewing different methods for ascertaining policy dimensions, Dinas and Gemenis (2010) conclude that differences in data sources are justifiable because of differing theoretical and analytic purposes.…”

Estimating parties’ policy positions through voting advice applications: Some methodological considerations

“…Each of these approaches uses different data sources and makes different implicit or explicit assumptions regarding the dimensionality of political space, whereas some methods are considered to be more suitable than others depending on the political context and the particular research question. These differences notwithstanding, there is no apparent consensus regarding the validity and reliability of party position estimates when alternative methods are compared with one another (Volkens, 2007;Dinas and Gemenis, 2010).…”

“…It should be rather evident that the relative emphasis political parties place on certain words or quasi-sentences associated with particular issues may or may not reveal their positions on these issues (Laver, 2001). Words or quasisentence frequencies can be a useful way to measure issue salience but can be problematic in the context of measuring positions (Lowe, 2008;Dinas and Gemenis, 2010). Conversely, Likert scales are concerned with measuring positions, but cannot measure the salience or the importance that parties attach to the issues in question.…”

Private Debt and the Anglo-Liberal Growth Model

“…This makes the data better suited to my purpose here than perhaps many other applications, where criticism of the data has been quite widespread (Gemenis, 2013). Specifically, the appropriateness of using the manifesto data to measure party positions-particularly on the `left-right' dimension, has been widely criticized (Dinas and Gemenis, 2010). However, the theoretical claim of interest here does not concern this generic location, but rather the `raw' data of the CMP: how many times policy areas associated with liberal growth versus those associated with export-oriented growth (as outlined by the economic theory) are mentioned in the two types of country.…”