Measuring smoking cessation: problems with recall in the 1990 California Tobacco Survey

Abstract

Measures of quitting activity are important both to evaluate public health interventions and to predict the likelihood of future quitting in the individual smoker. In population surveys, such measures are generally based on recall and thus may lack validity. In this article, we present the results of a 1-year quitting history of 8924 persons from a random-digit dialed population survey conducted in California in 1990. Respondents often forgot dates of quit attempts other than the most recent. Also, our results suggest that respondents may not recall short quit attempts at all, especially if they took place more than a few months before the interview. Thus, accurate population measures of relapse rates or quitting activity should rely on recall only for the last few months before the interview. Using data from quit attempts that took place within 4 months of the interview, actuarial analysis showed that 71.1% of attempts lasted at least 2 days, 58.5% at least 3 days, but only 39.2% lasted a week or more; this rate dropped to 19.6% at 1 month and to 14.1% at 3 months. As a predictive measure, a quit attempt that lasted a week or longer in the last year appears less biased by recall than any attempt of a day or longer in the last year.