Estimating the magnitude of method bias on account of text similarity using a natural language processing-based technique

摘要: A number of potential biases have been identified that may contribute a spurious component to the observed correlation between variables. One such bias is manner in which items are worded. This paper presents technique and program research estimate magnitude method arising from item wording. We hypothesize greater level text similarity employed capture predictor criterion variables, effect size them. Two samples will be employed; one investigating perceived usefulness-use correlations reported TAM literature other attitude behavior (physical activity) drawing upon Theory Planned Behavior (TPB). An NLP-based developed rate predictorcriterion pairs on similarity. The hypothesis tested by meta-regressing predictor-criterion against their respective scores. Implications for practice discussed.