I came across an interesting web article that compared the number of downloads of a scientific article. The article is somewhat dense with a number of figures and formulas. Honestly, I don’t completely understand it all, but I get the gist of the article (I think).
Here’s a quick recap. The authors wanted to know if examining web download statistics can be a (reliable) predictor of an articles citation impact factor (which can take years to determine). Using a set of physics articles, the authors compared web download statistics with the trailing citation impact factor between 2000-2004. After doing a bunch of statis-magic, the authors found a statistically significant association of 0.4 between the 6 month download statistics and the 2 year impact factor.
The authors conclude that “if the baseline correlation for a field is significant and sufficiently large, the download data could be used after 6 months as a good predictor of citation impact after 2 years”. So what does this mean? For articles being published in online journals, you can get a reasonable idea of the ultimate impact factor after about 6 months.
You can read the original article by Brody, T., & Hanard, S. (No Date). Earlier Web Usage Statistics as Predictors of Later Citation Impact. Accessed on March 24, 2005. URL: http://eprints.ecs.soton.ac.uk/10647/02/timcorr.htm.