This paper is based on an alternative approach to standard quantitativ
e analyses in research on behavior and health. Theory and methods focu
sed on the elaboration of complex situational and behavioral influence
s on health are used in secondary analyses of data from a population h
ealth survey in Spain. Findings showing fundamental differences in the
relationships among the behavioral and health variables within variou
s 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 id
entifying global statistical correlations in cross-sectional and longi
tudinal data. The impact that statistical-methods can have on the find
ings from analyses that are not guided by theory and logic based on su
bstantive 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.