10 May 2010

PubMed vs. Google Scholar

ResearchBlogging.orgA comment on Twitter about PubMed left me wondering aloud why people use the thing instead of Google Scholar. This idle comment brought a surprising amount of comments.

Before I get to the comments, let me explain my point of view. I’ve never warmed to PubMed, although I know many of my peers use it multiple times daily. I suppose part of it is the “med” moniker. While PubMed does include a lot of the basic biological literature, it’s still fundamentally a medical resource. And I am not, nor ever will be, a medical researcher.

For me, Google Scholar has been the greatest thing since Otto Rohwedder’s invention.* I always found it much more straightforward to search compared to PubMed’s finicky structure, and it was more likely to just give me the PDF reprint. From what I can see, Scholar searches PubMed, so if I like the interface better, and it includes PubMed, why should anyone use it?

Andrew Thaler was the most explicit Google Scholar booster:

I have never been in a situation where PubMed was better than Scholar.

On the PubMed booster comments, Rob Oakes wrote that Google Scholar includes only part of PubMed index. This I hadn’t known, but found that this has indeed been reported in the literature (e.g., Schultz 2007). Indeed, I hadn’t followed closely, but there are a whole series of articles comparing PubMed and Google Scholar. On a related note, Rob noted that Google Scholar is proprietary, so there’s no clear way to know what it’s doing to bring those hit forward. I’m not entirely sure many people could explicitly spell out what criteria PubMed uses for searches, though. Finally, he said:

Scholar doesn't have any drill-down features. That makes doing medication searches a serious PITA.

I don’t do medication searches, so this is not an issue for me.

eskay8 liked PubMed for integrating with the particular database she uses, while for Jason Goldman PubMed was all about “street cred.”

Biochembelle noted that PubMed’s functions have improved in the last year. She also commented that with PubMed, she didn’t have to “wade through patents and books.” Perhaps she hadn’t seen the new-ish checkbox on Scholar front page that allows you to skip patents.

And that’s perhaps the problem with these services: they’re moving targets. I tried PubMed a few years back and didn’t like it. Others tried Scholar and didn’t like it. But both have moved and changed. An even more recent article found only trivial differences between the two of them.

So Scholar users, maybe it’s worth it to go back and give PubMed another shot. PubMed boosters, try playing with Scholar’s new toys. Maybe you opinions will change. Or maybe JATetro is ahead of the game: he preferred the search engine that Falagas and colleagues (2007) found had features that surpassed both PubMed and Scholar: Scopus.

Additional: Speaking of moving targets, I see that Google Scholar has just added an email alerting function – I think that just went in today, in fact.

Reference

Falagas, M., Pitsouni, E., Malietzis, G., & Pappas, G. (2007). Comparison of PubMed, Scopus, Web of Science, and Google Scholar: strengths and weaknesses The FASEB Journal 22(2): 338-342. DOI: 10.1096/fj.07-9492LSF

Shultz, M. (2007). Comparing test searches in PubMed and Google Scholar Journal of the Medical Library Association : JMLA: 95(4): 442-445. DOI: 10.3163/1536-5050.95.4.442

* What? You didn’t know the inventor of sliced bread?

4 comments:

Daemios said...

Perhaps you're right about the kind of people using each one of them. I'm a bioinformaticist and put it simple, I always retrieve precisely what I'm looking for in PubMed, and only use Scholar occasionally to find alternative PDF's or book citations.

James said...

For a truly new way to get more from searching literature, please try NextBio and let us know what you think. For instance, here is a link to ESR1 articles: http://www.nextbio.com/b/search/lit/esr1?type=gene.

NextBio finds over 60,000 articles, while PubMed finds only ~750.

Enjoy the experience.

rpg said...

I'm not convinced that finding more is always better, James! It's all down to relevance. I'd much rather find 750 relevant articles than wade through 60k...

Lit said...

rpg hit the nail on the head. I search for what is relevant and hate getting bogged down in long lists of results. That is why I use Quertle, which finds actual relationships. Here is an example query looking for the role of James' ESR1: http://www.quertle.info/v2/search?s=r&query=ESR1+%24BiologicalProcess&so_a=&so_j=. (This search uses one of Quertle's Power Terms.)