zenourn:In my area I focus particularly on mathematics/statistics (was my initial background) and I have to suggest substantial revisions to most papers that I review.
I've run into that as well, proper statistical analysis of results in CompSci papers in the area I review were so rare in the past that some years ago I was actually shocked to see a paper submitted that did rigorous analysis of the data. OTOH management science papers are the exact opposite, they'll take three diverging data points and apply 20 pages of statistical mangling to identify a trend. So the CompSci stuff was lots of data, little to no analysis (although it's gotten better in recent years), the MgtSci stuff was little to no data and infinite analysis for which changing any parameter anywhere would completely change the outcome. And don't even get me started on the quality of the data and analysis of a lot of Social Sciences papers.
If only it were possible to combine the CompSci empiricism with the MgtSci analysis...