At the end of the paper, I think it is rightly pointed out that ideology is a “fluid construct”, situated in complex hermeneutic zones. Freeden 2006 identified some of the epistemological issues surrounding ideology as a concept; its susceptibility to the “winds of academic fashion" leads it to be caught in many debates surrounding its slippery definition – many of which persist today.
The supervised learning approach here has shown itself in the study to be a robust forecasting method for specific outcomes. Even if ideology is otherwise defined more broadly, the measurement of latent concepts will still be important in quantitative analysis in political science – and these will no doubt continue to obfuscate interpretation in favor of prediction. Perhaps dynamic causal inference methods like those proposed by Blackwell 2013, Egami et. al 2018, and others might be a good companion to the methods demonstrated in this study.
Thank you for presenting!
At the end of the paper, I think it is rightly pointed out that ideology is a “fluid construct”, situated in complex hermeneutic zones. Freeden 2006 identified some of the epistemological issues surrounding ideology as a concept; its susceptibility to the “winds of academic fashion" leads it to be caught in many debates surrounding its slippery definition – many of which persist today.
The supervised learning approach here has shown itself in the study to be a robust forecasting method for specific outcomes. Even if ideology is otherwise defined more broadly, the measurement of latent concepts will still be important in quantitative analysis in political science – and these will no doubt continue to obfuscate interpretation in favor of prediction. Perhaps dynamic causal inference methods like those proposed by Blackwell 2013, Egami et. al 2018, and others might be a good companion to the methods demonstrated in this study.