This paper is based on an alternative approach to standard quantitative analyses in research on behavior and health. Theory and methods focused on the elaboration of complex situational and behavioral influences on health are used in secondary analyses of data from a population health survey in Spain. Findings showing fundamental differences in the relationships among the behavioral and health variables within various age, gender and social groups illustrate the importance of studying interacting influences in relevant subgroups of the population. Quite meaningful findings can be hidden in behavioral research limited to identifying global statistical correlations in cross-sectional and longitudinal data. The impact that statistical methods can have on the findings from analyses that are not guided by theory and logic based on substantive questions derived from the research literature is discussed. The results point to the need to study patterns of behavior in their contexts of occurrence in research on lifestyles and health.