How stable are self-reported measures of media use?

Abstract

Measuring media exposure is challenging due to apparently unreliable self-reports. As a result, media exposure appears highly stable after correcting for measurement error. The model commonly used to quantify reliability and stability makes untenable assumptions about the data used in communication. Commonly violated assumptions can bias reliability estimates downward and stability estimates upward. Despite high reported stability, media use changes are associated with theoretically expected outcomes, suggesting more meaningful variation than captured by current methods.