The Impact of Incorrect Responses to Reverse-Coded Survey Items

The impacts of incorrect responses to reverse-coded survey items were examined in this simulation study by reversing responses to traditional Likert-format items from 700 administrators in randomly selected schools in a 7-county region in central Arkansas that were obtained from an archival dataset. Specifically, the number of reverse-coded items (1 to 5) and percentage of participants (5%, 10%, 15%, or 20%) responding incorrectly were varied to examine impacts on score reliability and scale means. Results suggested that incorrect responses to reverse-coded items can statistically significantly impact scale means. Impacts on estimates of internal consistency reliability of scores were small and might go undetected by researchers. The results support advice by researchers to use reverse-coded items with caution and subject the items to additional scrutiny to detect systematic measurement error arising from incorrect or confused responses. (Contains 4 tables.)