The impact of publications is currently measured by the average number of citations received for a journal article. But there are 3 different models for this in the biggest repositories. The classic 2-year JCR model, the new model implemented by Scopus last year (citations to 3-year publications during those same 3 years) and the ScimagoJR method, using a Google-style algorithm to decide the most prestigious sources.
Which one do you think is more suitable, or in your case, what are the strengths and weaknesses of each one?