Measurement Refinement of Disgust Instruments

[Stage: Conducting exploratory data analysis. Funds for data collection secured]

In collaboration with Josh Tybur and Yoel Inbar, I am conducting exploratory analysis on Disgust Instruments with Item Response Theory. The three disgust instruments are: Three-Domain Disgust Scale (TDDS),  The Disgust Scale - Revised (DS-R), and Perceived Vulnerability to Disease (PVD). We believe that carrying out such analysis on the key instruments used in the behavioral immune system literature could lead to the development of a measure that captures the latent construct measured by all three scales - i.e., a new "Behavioral Immune System" scale. After the exploratory analysis is performed, we intend to collect new data as to confirm the findings and validate the new instrument.

The structural complexity & validity of conservatism scales

[Stage: receiving quotations from pooling agencies, assembling instruments for survey, funds secured]

Part One

Together with Professor Dr. Tobias Rothmund, we investigate the internal structure and complexity of conservatism scale. Briefly, the [preliminary] purpose of this enterprise is to (1) validate the original articles' findings, (2) assess comparatively the predictive validity of these scales, (3) assess the structural complexity of conservatism, (4) explore a variety of psychometric properties at both scale- and item-level and (5), propose the first adaptive conservatism test via computerized adaptive testing (CAT-IRT). Please see below for a list of conservatism scales being considered, as well as a more detailed list of objectives.

List of conservatism scales being considered

Scales being strongly considered

• Everett's Social and Economic Conservatism scale, 2013
• Zell & Bernstein's 12-item scale, 2014
• Feldman & Johnston's Social & Economic Scale, 2014
• Inbar, Pizarro & Bloom's Political issue statements scale, 2008

Controls

• Altemeyer's Right-wing Authoritarianism (RWA) Scale, 1988-1996
• Sidanius & Pratto's Social Dominance Orientation (SDO-7), 2015
• Jost's Political and Economic System Justification, 2010

Other scales being considered

• Adorno's F-Scale, 1947
• Sidanious's S-scale, 1976
• Altemeyer's Right-wing Authoritarianism (RWA) Scale, 1988
• Dukitt's Authoritarianism, Conservatism, Traditionalism scale (ACT), 2005

Objectives of the study
 

1. Investigate whether a representative sample would reproduce the results reported in the scale's original article, which (almost always) used much lower sample sizes.
2. Assess the predictive validity of these scales and rank them from most accurate to least accurate via validating questions such as "who did you vote for in the last Presidential election?" and "who do you intend to vote for in the coming Presidential election." There are many ways to do this, from simple counts or odds of correct predictions to using scale's scores as predictors at a multilevel (ordinal, IRT) models, while controlling for the political sophistication, political participation, and other classical variables in political behavior.
3. Analyze the psychometric properties of the pooled items as well the individual scales. In specific, we are interested in four aspects: 
   • Pattern of responses on the pooled items (e.g., LCA) 
   • Measurement Invariance for gender, (e.g., Differential Item Functioning, DIF)
   • Dimensionality of Conservatism (e.g., FA, iCLUST, mIRT)
     • To provide an indication of the dimensionality and information of both items and scales. 
     • With pooled data, and assuming a true representative sample, the data may give indications of the dimensionality of conservatism - i.e., inform the debate between the 1-item political orientation (1-Dimenion construct) vs. social-economic conservatism (2-Dimension).
     • Another idea, depending on the response pattern of subjects, is to investigate whether the GAL-TAN party structure (2-D) found using macro-level data on party-politics literature can also be found in individual-level political preferences (via conservatism scales). For example, in case we find GAL-TAN structure with response patterns, we could conceive GAL-TAN as either a consequence or as a probable cause of individual's preferences on political issues.
   • Information yielded by individual items as well as scales (statistical/psychometric property gauged with Item Response Theory and/or Multi-Group CFA). 
     • Here, the idea is to map scales and items to a (1-D or 2-D) conservatism space. 
     • Ultimately, with this analysis using IRT, it would be possible to recommend items for an optimized conservatism scale which maximizes the accuracy, internal reliability, and information for the American public. 
     • Moreover, and contingent upon the 'found' dimensionality of the data (i.e., the study could either find evidence for the support for a uni- or bi-dimensional theoretical paradigms; or alternatively, given the subjectivity of such techniques suggest discriminative items for each both these paradigms,
     • Lastly, the natural consequence of the above IRT analysis is the development of the first adaptive conservatism test via computerized adaptive testing (CAT), which is a form of computer-based test that adapts to the individual's trait level. Since this study is exploratory in nature, it allows us to pursue ambitious goals.

Part Two

For more than half a century, a voter’s political ideology the most used method to predict one’s political choices. However, there is as yet no accepted, validated, and widely used ideological instrument that is concise, applicable to wide range of countries and political contexts. Furthermore, while the basic integrity of such instruments depend on having an appropriate measurement of political beliefs, several attempts in previous research relied on a 70-year-old statistical paradigm. To address both substantive and methodological gaps, we proposed a modern and innovative statistical method and developed an ideological instrument that gives emphasis on unobservable political-psychological processes unfolding in the minds of political actors. The resulting instrument is going to be tested in the US and Germany – countries whose political environments vary widely. Our objective is that this newly created political instrument is able to provide an in-depth investigation of the Australian ideological spectrum and appraise cross-cultural differences. 

Measuring populist attitudes across nine European countries

Populism is on the rise in Europe. In the past decade - with the emergence of more volatile electorates and protracted global economic crisis - radical and right-wing populist parties have garnered unprecedented political power at national and supra-national elections. While UKip, Front National in France and Party of Freedom in the Netherlands account for most of the news headlines, in Hungary and Poland the government is spearheaded by populist representatives. In order to understand this phenomenon, we focus on the measurement and empirical evaluation of populist attitudes, which are thought to explain – at least, partially – predilections for Populist parties. As of recent, a number of prominent studies have converged on a set of six items which have been to gauge this latent construct. Two original items were developed and added to this scale as a mean to validate and further examine populist attitudes. In 2015, as part of the LIVEWHAT project, the ensuing scale was used to survey populist attitudes across nine European countries (n = 18370). The objective of this article is twofold. Firstly, while drawing insights from both classical test and item response theories, we assess the scale’s psychometric properties – e.g., internal consistency, dimensionality, reliability and validity – as well as item’s difficulty, discrimination and information. Secondly, cross-national differences and similarities are discussed. Results indicate that the nine surveyed are roughly divided into two sub-groups. While Italy, Spain, Greece, France and Poland showed comparable high levels of populism endorsement, Sweden, Switzerland, Germany and the UK consistently exhibited lower levels of it. Moreover, while the scale was conceived to measure across the entire continuum of populist attitudes, its items were found to be informative only at a limited range of the theorized latent construct. Implications to the construct- and predictive-validity are discussed, as well as recommendations for future operationalizations.

[Stage: Conference paper - presented at Team Populism & MPSA 2016]