Date of Award
Doctor of Philosophy
The field of smoking cessation research is plagued by inconsistent findings, caused in part by inadequate methodological or statistical analysis procedures. These problems are more serious in the case of smoking cessation treatments administered to groups of individuals;The effectiveness of two treatments is compared: an aversive smoking procedure followed by a maintenance phase (AM); and a nicotine fading procedure followed by the same maintenance phase (NM). Also, the benefits of being randomly selected to treatment (RS) versus self-selecting treatment (SS) are compared. Treatment was administered to groups of individuals, and groups differed in size from 2 to 17 participants. Information on subjects' subsequent smoking status (abstinent or smoking) was collected at several follow-up times;Two statistical analysis procedures are applied to these data. Repeated measures analysis of variance (ANOVA) was carried out using as the unit of analysis the proportion of subjects abstinent at each of the several follow-up intervals. Survival analysis was carried out using individual subject's time to return to smoking as the unit of analysis;Results using the ANOVA methods revealed no significant differences between the two treatments or the methods of selection to treatment. In contrast, a survival analysis procedure using a limited failure population Weibull model, while not demonstrating any difference in effectiveness between the AM and NM treatments, did reveal a significant difference between the methods of selection to treatment in favor of self-selection to treatment. However, other differences between the RS and SS methods make interpretation of this finding difficult;Several factors point to the greater appropriateness and statistical power of the survival analysis methodology over that of analysis of variance. However, there is a need to adapt survival analysis methodology to take account of the possible presence of group effects. Possibilities such as bootstrapping are discussed;In the psychological arena, both the oversmoking and nicotine fading treatment strategies appear equally effective. The indications that self-selection of treatment is beneficial are potentially important and invite further study. It also appears that future improvements in success rate will depend on improved relapse prevention.
Digital Repository @ Iowa State University, http://lib.dr.iastate.edu/
Paul Gerard McGovern
McGovern, Paul Gerard, "Two statistical analysis procedures applied to multivariate smoking cessation data " (1984). Retrospective Theses and Dissertations. 8193.