A comparison of characteristics and responses for non-responding and responding certified athletic trainers to mail and Web-based surveys on continuing education units

Survey research with the study population of the National Athletic Trainers\u27 Association membership is increasing in prevalence. Unfortunately, many researchers have not achieved optimal response rates necessary for unbiased generalization to the population. The primary purpose of the study was to determine if characteristics and responses of non-respondents are comparable to that of respondents. A secondary purpose is to determine if mail and web-based modes produce different response rates, responses, and characteristics of responders. One thousand Certified Athletic Trainers were asked to complete a survey with 30 Likert scale questions about their attitudes regarding continuing education and seven demographic questions. For the mail group, the first contact included a cover letter, the survey, and a business reply envelope. This was followed by a reminder postcard fourteen days later. After another two weeks, the survey and letter was re-mailed. Finally, an additional reminder postcard was distributed after fourteen more days passed. The web-based group received a message every two weeks for the two-month period via e-mail. Completion of the survey required entering an access code as specified in the first and third e-mails at the identified website. Non-responders were contacted a fifth time and presented with the survey administered via the opposite mode. A percentage of individuals that still did not respond were contacted via telephone to obtain information on non-response. Logistic regression analyses revealed that the treatment condition was the most consistent at predicting version of survey response (early (first contact or reminder), middle (third or fourth contact), mixed (fifth contact via opposite mode), or non-response). The mail group was more likely to respond to the early and middle versions, while the web-based group was more likely to return the mixed method version. Some demographic variables, gender and highest degree earned, played a role in survey version returned, however; the survey responses did not influence survey version response. Overall, response rate is higher for mail surveys compared to Internet surveys, females compared to males, and individuals with terminal or master\u27s degrees compared to bachelor\u27s, however; response rate or time did not influence the survey results.