Start Date

16-8-2018 12:00 AM

Description

In survey research, it is well known that the quality of responses is significantly altered by apparently trivial variations in the linguistic or grammatical properties of survey items. Yet numerous seemingly minor changes are made to survey items in the course of the scale development process so that they comply with other requirements (e.g., content validity). As a result, researchers may inadvertently introduce systematic measurement error that is not accounted for in the final model. Remedies to biased items are widely known, but reliable methods to diagnose the problem do not readily exist. In an effort to address this shortcoming, we develop a quantitative method to detect biased items and reinforce the reliability of IS measurement instruments. In this paper, we provide step by step implementation guidelines and show how to apply the method and interpret the output results.

Share

COinS
 
Aug 16th, 12:00 AM

Detecting Biased Items When Developing a Scale: a Quantitative Approach

In survey research, it is well known that the quality of responses is significantly altered by apparently trivial variations in the linguistic or grammatical properties of survey items. Yet numerous seemingly minor changes are made to survey items in the course of the scale development process so that they comply with other requirements (e.g., content validity). As a result, researchers may inadvertently introduce systematic measurement error that is not accounted for in the final model. Remedies to biased items are widely known, but reliable methods to diagnose the problem do not readily exist. In an effort to address this shortcoming, we develop a quantitative method to detect biased items and reinforce the reliability of IS measurement instruments. In this paper, we provide step by step implementation guidelines and show how to apply the method and interpret the output results.