[Edited question:]

Sorry, I wasn't being specific enough in my original question (but thanks for your responses so far!). Things like personality and genetics don't really change from day to day, so they can't directly correlate with / predict / cause someone to drink on one day but not another. I'm looking for antecedent causal variables that can fluctuate from day to day, and thereby cause fluctuating behavior from day to day. There are some daily-drinking-diary studies out there on this kind of thing (e.g., daytime experiences of negative social interactions leading to more drinking that evening), and to achieve greater specificity I want to parse "more drinking" into the binary variable "if drank" and the interval variable "how much drank, provided drinking occurred." Different daytime events may be differentially stronger predictors of these two criterion variables, and I'm looking to learn what people know about this possibility. In any research, have these variables been parsed before with respect to the causes of a single drinking episode? Aside from established research, what are your best guesses? Thanks!

[Original question:]

I'm aware of the frequency/quantity literature on characterizing global drinking traits, but would like to hone in on what causes whether or not a person will drink on a given day versus how much they will drink on a given day, provided they have at least one drink. Studies on daily determinants of other kinds of potentially problematic behavior or experiential avoidance would also be useful (e.g., drug use, binge eating); I'm primarily interested in functional and methodological approaches to this kind of distinction. Thanks!   

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2717559/