Using Entropy-Based Information Theory to Evaluate Survey Research

Abstract

An assumption underlying most survey-based research studies is that respondents provide equally informative responses. While researchers regularly examine the validity and reliability of survey data, few assess whether the quantity of information collected is consistent across respondents. This manuscript uses information theory to evaluate whether certain types of respondents are more (or less) likely to give equally informative responses. A simple method to evaluate the quantity of information inherent in survey items and sampling designs is presented. We apply this technique to a survey conducted for a large, amateur sporting event and find that the survey design was generally appropriate.