Somehow in the last week or so I have reviewed three papers on totally different subjects. Two were in areas which I know the literature, techniques, etc. The third, I know generally of the field (my advisor knows as much as I do). The latter paper was much harder to review since I am not up on the field and there are only so many papers I can read to review one paper. Experimental design was where I focussed my efforts on that paper, making sure the proper controls were done, etc. Hopefully another reviewer works in that field and was able to comment more on the discussion.
The first two papers I had to sit back a few times from reviewing and ask other people if I was being fair in what I was asking the authors to do since I know the subfields pretty well & can think of plenty of experiments on the subjects that would strengthen the papers. I try and be positive in my reviews, trying to get any critiques to come across as making the paper even better. It is not an easy task when controls are missing. Nobody likes getting a review back and having to do more experiments. I know I don't. Once I have submitted a paper, the most I want to do is revise what is written and when it comes to bench work I want to move on to new things. Controls though must be done. Who wants to be the reviewer of a paper that has to be retracted?
At least at this stage in my career my advisor goes over the paper and my review, so it isn't really on my shoulders yet. I am still training, still learning.
One things I must say, report errors. Don't try to tell me a 10 to 20% increase in activity is significant without them especially if other papers report errors using the same assay anywhere from +/- 10 to 40%. Also, do a literature search prior to submitting especially if you make statements of the nature it has only been done once or not at all. If I can do the search as a reviewer then you can as an author. Basically, don't try to oversell your paper.