Self-reported frequency of drinking assessed with a closed- or open-ended question format: a split-sample study in Switzerland.

OBJECTIVE: This study was designed to determine whether reporting of drinking frequency differed according to whether the question asked was open or closed ended. METHOD: In a split-sample design with 994 participants responding to a questionnaire, the question on usual frequency of drinking for one sample (n = 457) was open ended, and for the other (n = 537) was closed ended (participants were offered seven response alternatives, ranging from "three times a day" to "never"). RESULTS: The closed-ended question elicited higher frequencies of drinking than the open-ended question. No significant interactions could be found between sociodemographic background variables and question format. CONCLUSION: This study does not support the general recommendation of survey methodologists that questions about frequency of behavior be open ended.