This article outlines how different types of survey response reliability are estimated and used within the social sciences. Implementing the Classical Test Theory’s concept of reliability in a panel survey exploring opinions and attitude crystallization in a Czech town. Estimates of the test-retest reliability of respondents’ political orientation are estimated using four different approaches. Political orientation is measured using a standard 10 point left-right scale. It also estimates the ‘true score’ of a respondents’ political orientation; and shows the implications of measurement precision for the estimation of this individual level trait. Inter-item reliability is calculated using a 4 item scale composed of dichotomous indicators. This analysis reveals that the estimated reliability of the political orientation scale depends on the number of items used to construct the scale. More concretely the 10 point left-right political orientation scale exhibits a satisfactory level of test-retest reliability demonstrating its validity in making inferences about the target population. The analyses presented in this study were estimated using the R statistical programming language and LISREL – a specialized software package for estimating structural equation models.
This article outlines how different types of survey response reliability are estimated and used within the social sciences. Implementing the Classical Test Theory’s concept of reliability in a panel survey exploring opinions and attitude crystallization in a Czech town. Estimates of the test-retest reliability of respondents’ political orientation are estimated using four different approaches. Political orientation is measured using a standard 10 point left-right scale. It also estimates the ‘true score’ of a respondents’ political orientation; and shows the implications of measurement precision for the estimation of this individual level trait. Inter-item reliability is calculated using a 4 item scale composed of dichotomous indicators. This analysis reveals that the estimated reliability of the political orientation scale depends on the number of items used to construct the scale. More concretely the 10 point left-right political orientation scale exhibits a satisfactory level of test-retest reliability demonstrating its validity in making inferences about the target population. The analyses presented in this study were estimated using the R statistical programming language and LISREL - a specialized software package for estimating structural equation models., Jan Schubert., and Obsahuje bibliografii a bibliografické odkazy