I had a look at the scopus data and I don't think you can trust the numbers for the current year. I don't know how often these are updated but you can see for example that PLoS ONE has fewer number of articles in the database so far (112) as expected from the previous year (total year=2737) - Pedro Beltrao

PLoS Pathogens for example does not appear to have data for the current year. - Pedro Beltrao

More importantly I think this trend line measure tends to give more weight to journals that publish a lot of papers. Try putting PNAS on the same graph for example. Its the total number of citations received by the journal that year (not specifically to the articles of one year) over the number of papers published in one of the years. So you can't really compare PLoS ONE with the other PLoS Journals in this way - Pedro Beltrao

Yes I agree with the fact data in Scopus is not uniform or let say not complete, but not bad either, I agree with the fact trend lines could be biased to journals that publish a lot of papers because it illustrates the total number of citations each journal received (regardless of the publication date of the cited document) in a given year, divided by the total number of documents published in that year, but if we look the trend line with % Not Cited Parameter it definitely gives a clear picture - Abhishek Tiwari

Probably a nicer number to get is the number of citations from 2008 to papers in 2007. This avoids the problems I mentioned. This number does say that PLoS one is getting more or less the same number of citations per paper than PLoS Comp. and both have about half of PLoS Bio. So I think you are right, PLoS ONE might be get an impact factor similar to PLoS Comp Bio. - Pedro Beltrao

@Pedro - curious. Scopus does list an appropriate number of PLoS ONE articles for 2009 (819 articles, which is too low, but understandable with an indexing lag time) but only lists 112 in the analytics engine (which implies their analytics lag their actual data by about a couple of months). - Peter Binfield

I tried to predict the impact factor of PLoS ONE based on current data. There were 1167 articles published in 2007. These had 3259 citations in 2008. We don't the full results in already for 2009 but I looked at a few cases and you can usually guess that these will be around 1.2 times the number of citations in 2008 so approximately 3900. These would give PLoS ONE a first impact factor of 6.UPDATE: This is actually not the correct formula I will add the correct value when I re-do this. - Pedro Beltrao

I agree with Pedro calculation about possible impact factor of 6 something for PLoS One (around same as PLoS CB ), which is also reflected with Scopus 2009 trends (although data is incomplete). - Abhishek Tiwari

Also as Pedro suggested "tends to give more weight to journals that publish a lot of papers" but lets not forget this is also affected with time line of journals, older journals will have more better trend line which means PLoS One is certainly doing better - Abhishek Tiwari

So, do we use IF or not? Or do we just use with the journals we want? - Paulo Nuin

Ok, I re-did the analysis with the correct formula. Also, I noticed that Scopus has multiple entries per articles for some journals. PLoS Comp for example has more than double the number of correct published articles. So, the predicted impact factor for PLoS ONE will be around half of PLoS Comp. Details here in blog post http://pbeltrao.blogspot.com/2009... - Pedro Beltrao

@Paulo well thats interesting questions actually, IF is good measure when it is used for the comparison of the journals, but it is equally bad when it comes to quality of individual research. We definitely need IF or any such thing otherwise it will be never easy to find quality article out of 20,000 journals. - Abhishek Tiwari

Scopus data for PLoS One is nearly same as ISI. as u can see PLoS One 2007- 1,168(Scopus)1,229(ISI) PLoS One 2008- 2,745(Scopus)2,760(ISI) PLoS One Total- 4,869(Scopus)4,962(ISI) - Abhishek Tiwari

If we use any sort of journal-level metric, we best not use ISIs IF, because it just plainly sucks. Isn't Scimago ranking PLoS One, yet? Wouldn't we all fail students who use averages on such obviously non-normally distributed data (or reject any paper for that matter)? - Björn Brembs

Actually I raised a different kind of question but it unknowingly it became IF game, I just wanted to suggest that PLoS One is getting more attractive and it may overshadow many journals in PLoS family. - Abhishek Tiwari

Actually I think IF is a decent measure to compare journals even if the citations per paper is not normally distributed. All journals have the same distribution so this number of citations per article is a fair measure. What is not fair is estimate the value of single articles from the IF of the journal where it is was published. - Pedro Beltrao

I agree also that we should not decide what articles to read based on IFs of journals but instead based on article level measures but .... we generally don't have these in place yet. - Pedro Beltrao

For that purpose I used Scopus data and Scimago ranking is based on same data and same parameters. Scimago dont rank PLoS One yet - Abhishek Tiwari

@pedro: interesting statement that calculating e.g. the median would not change journal rank as compared to arithmetic mean. Conjecture or have you tested that? I'm not sure one way or another, hence, I'd be really interested in such a comparison. - Björn Brembs

@Bjorn , assuming that the distribution of citations per article will have a similar log normal distribution for most journals I think that is correct. I think the ranking can change but not drastically so. I looked at 10 journals from the same publisher (Nature) from the same year (2007) and the citation information from Scopus (all citations since publication). The correlation between average and median values was 0.96 for these 10 journals. - Pedro Beltrao