DOI: 10.5176/2251-1865_CBP18.31
Authors: Jian Han, Miaodan Fang , Shenglu Ye, Chuansheng Chen, Xiuying Qian , Qun Wan
Abstract: Response rate has long been a major concern in survey research. Based on 244 published studies on consumer satisfaction, attitude and loyalty that are predictors of customer retention and behavior, this study aimed to identify predictors of response rates. A decision tree analysis (using the C5.0 algorithm on 70{6e6090cdd558c53a8bc18225ef4499fead9160abd3419ad4f137e902b483c465} of the studies as the training set and 30{6e6090cdd558c53a8bc18225ef4499fead9160abd3419ad4f137e902b483c465} as the test set) revealed that a model with seven attributes of the surveys attained an accuracy of 80.52{6e6090cdd558c53a8bc18225ef4499fead9160abd3419ad4f137e902b483c465} in predicting whether surveys had high (> 50{6e6090cdd558c53a8bc18225ef4499fead9160abd3419ad4f137e902b483c465}) or low (< 50{6e6090cdd558c53a8bc18225ef4499fead9160abd3419ad4f137e902b483c465}) response rates. Direct invitation was the most important factor (yes > no), followed by mode of data collection (face-to-face or mail > telephone or online). If it was telephone or online survey, 20 items was the crucial cutoff point for length of survey. The accuracy of the decision tree model was higher than that of the traditional logistic regression.
Keywords: Cognitive load theory; human cognitive architecture; working memory; depletion effect; instructional procedures
Updating...