| Popis: |
The study follows up on a survey investigating the applicability of the Poisson model (preregistered on OSF: https://osf.io/z35y7) to the Unrelated Question Model by Greenberg et al. (1969) using drinking and driving as an example (preregistered on OSF: https://osf.io/nh6e9). The data collection phase of the first study was concluded on 5th August 2022. In the first steps of data analysis, we discovered surprising results regarding the validity item, a question about the eye colour of the respondents. While the prevalence for blue eyes seems plausible in the Direct Question (DQ) group (.355) in comparison to previous studies regarding eye colour (around 40% in Germany, see Katsara & Nothnagel, 2019), the estimate of the eye colour question in the Unrelated Question Model (UQM) group is much higher than anticipated (.517). This unexpected difference between DQ and UQM group can be explained in multiple ways. A possible explanation that is easy to examine is the existence of order effects. Since the eye colour question was posed after the first question regarding drinking and driving and after the attention check, some answers to the eye colour question might have been influenced. Some respondents might have been reluctant to think of a different birth date for the second UQM question and thus stuck to their first answer, some might have realized that they will not be filtered out after the attention check and thus answered at random to the following question. Whatever the exact reasons might be, we now want to research if the order of the questions is responsible for the supposed overestimation of eye colour prevalence in the UQM question. If, as we hypothesize, the UQM question regarding eye colour will result in valid estimates when posed before the attention check and the other UQM question, this would mean that order effects are likely. Thus, our UQM estimate for drinking and driving in the first study would not seem invalid just because the UQM estimate of eye colour in the first study was invalid. If, on the other hand, the UQM estimate for eye colour yields invalid results even when posed in the slot of the first question, there are most likely no order effects present. Thus, our UQM estimate for drinking and driving might be flawed and the validity of the UQM might be in question (in general, for non-sensitive topics or at least for this specific question about blue eye colour). So, the central research questions are: - Is the estimation of blue eye colour prevalence via UQM valid if it is posed before the second UQM question? - And thus: Could the prevalence estimate of blue eye colour in the UQM group in the first study be influenced by order effects? |
