Abstract:
|
Low quality responses (selection of response options by respondents that do not represent their true position on the trait being measured) are ubiquitous in survey research and their extent is estimated to be around 3.5% - 60% of the collected sample. This forms a major threat to validity as low-quality responses (LQR) have an adverse effect on statistical results and inferences. It can lead to spurious within-group variability, lower reliability, and potential type II errors during hypothesis testing. While there has been a surge in the literature on detecting LQR, there are very few comprehensive studies that detail the impact of LQR on the psychometric properties of the survey. In this study, we present the results of a simulation under varying degrees and kinds of LQR. We examine different properties of the scale - dimensionality, reliability, and model fit under realistic survey assessment scenarios.
|