The impact of the response format on the left-right self-placement


Description: Social scientists deal with data about the left-right self-placement measured with different types of scales. Two main problems arise: firstly, we are assuming that results are equivalents between scales (Kroh, 2007) and, secondly, we do not dispose of enough evidences to consider how the design of the response format is impacting the results and, therefore, the quality of the data. This study develops its research question in this field of knowledge: which is the impact of the design of the left-right scales on the results and its quality? Using a study conducted by the Centre d’Estudis d’Opinió in its second round of the Omnibus de la Generalitat 2016, we dispose of the results of two different scales that have not been compared yet within the same sample. Both scales are, concretely, the partially labeled eleven-points scale and the fully labeled seven-points scale To analyze the impact of both response formats, we conducted three analyses: (1) a comparison of the reliability, the validity and the quality predictions of both scales using the SQP 2.1 software, (2) an external validity test to complement the quality estimates and, finally, (3) an analysis of the individuals placing themselves on the middle points in each scale. Firstly, we found that the seven-points scale has half of the respondents placing at the middle-point (16.8%) than the eleven-points scale (34.8%). However, which is better capturing reality? Evidences drawn demonstrate that, from a methodological point of view, both scales have similar levels of reliability, validity and quality. Secondly, regarding the analyses of the central value, we have found that the eleven-points scale incentivizes respondents with lower reliability and political consistency levels to tend to the center. Finally, since the seven-points scale rise as a better voting predictor, its external validity is higher.