Data Quality in Student Risk Behavior Surveys and Administrator Training

Abstract

SUMMARY This study examined the influence of survey validity screening on the results from three group-administered school surveys administered to samples totaling approximately 5500 students in 19 schools. The estimated levels of risk behaviors, antisocial behaviors, and victim experiences were substantially reduced when respondents who gave multiple inconsistent or extreme responses to other survey items were screened out of the data. The researchers also observed that the percentage of students giving inconsistent and illogically extreme responses was greater among those surveys given by an untrained administrator, raising the hypothesis that administrator training could be a critical factor in obtaining more consistent and trustworthy survey data. These results indicate that it may be important to train school staff in survey administration and to screen surveys for validity in order to improve the accuracy of student self-report surveys.