Wednesday, January 02, 2008

Books that scientists use - an addendum

Most cited references in Nature 2007 - all books

Some years ago when I started my tenure as Science Librarian at the Leddy Library, I worked on a particularly tedious project that really helped me get a better understanding on what books scientists actually use in their work (as opposed to so many other titles).

Using Web of Science, I would search for a year's worth of articles from Science, Nature and PNAS, download these articles' references into an Excel spreadsheet and try to decipher which ones were books. It was a very very mechanical and time-consuming process. But it made me a better science librarian.

Now, I'm pleased to say, that Scopus makes this same task possible in less than 30 seconds:

1. in search for box type in journal name (e.g. Nature) and select Source Title from drop-down menu
2. limit date range to a particular time period (e.g. 2007)
3. hit search button
4. in the Refine Results box, limit your results to just the journal name in question (e.g. Nature and not Nature Biochemistry)
5. in the Results box, click the Select All box and the hit the References button
6. If your initial search results brought more than 2000 hits, you will be informed that only the first 2000 articles will have their references retrieved.
7. review the list of most cited references of that journal from most cited to least

Here's a list of the 10 most cited books by Nature, Science, and PNAS*. The second number in the list represents where in the journal's list of most cited items can the book be found.

Most Cited Books in Nature, 2007**
1. [1] Molecular cloning: a laboratory manual (1989)
2. [2] Diagnostic Statistical Manual of Mental Disorders (1994)
3. [3] Numerical Recipes (1992)
4. [4] Biometry (1995)
5. [5] Biostastical Analysis (1984)
6. [7] The Rat Brain in Stereotaxic Coordinates (1986)
7. [13] Principles of Optics (1980)
8. [15] Physics of Semiconductor Devices (1981)
9. [19] Intermolecular and Surface Forces (1992)
10. [21] Co-Planar Stereotaxic Atlas of the Human Brain (1988)

Most Cited Books in Science, 2007**
1. [1] Diagnostic Statistical Manual of Mental Disorders (1994)
2. [2] CRC Handbook of Chemistry and Physics (2005)
3. [3] Biometry (1995)
4. [4] Biostastical Analysis (1984)
5. [5] The Rat Brain in Stereotaxic Coordinates (1986)
6. [9] Principles of Optics (1980)
7. [10] Computer Simulation of Liquids (1987)
8. [11] Structured Clinical Interview for DSM-III-R (1990)
9. [12] Advanced Organic Chemistry (1988)
10. [14] An Introduction to Probability Theory and its Applications (1971)

Most Cited Books in PNAS, 2007**
1. [1] Molecular cloning: a laboratory manual (1989)
2. [4] Diagnostic Statistical Manual of Mental Disorders (1994)
3. [6] Numerical Recipes (1992)
4. [8] CRC Handbook of Chemistry and Physics (2005)
5. [9] Biometry (1995)
6. [10] Biostatistical Analysis (1984)
7. [12] The Rat Brain in Stereotaxic Coordinates (1986)
8. [13] Handbook of Mathematical Functions (1972)
9. [22] Stastical Power Analysis for the Behavioral Sciences (1988)
10. [23] Statistical Methods (1980)

*Nature and Science published almost 2000 items in 2007 and PNAS published closer to 3000 items in 2007.

I'm planning to use this method to determine if there are any important books my library is missing by reviewing the references of the key journals in various fields for 2007. Its still a largely mechanical process (although the Foxy Leddy LibX toolbar makes book-checking much faster than typing titles into the library catalogue) but its a good task to slowly start the work of the new year.

I've been thinking further about these lists and I think I am in grievous error.

For example: the approximately 2000 articles in Nature evidently produce 24,660 references. That's about 200 items in each item's bibliography - which sounds high but its in the realm of possibility. But what confuses me is the cited by column which says that the first item in the list, Molecular Cloning: A Laboratory Manual, has been cited 103, 825 times. So that must refer to how many times the item has been cited within the Scopus database. The fact that the some of the books in the list appear in the same relative order when performing the same procedure using the journals PNAS and Science, means that these lists reflect the most popular science books within Scopus and not necessarily within each journal.



Anonymous said...

Well- you are also assuming that not citing = not useful. When you write a journal article you don't cite background reading that is usually very influential on your thinking. You tend to be very conservative about what you are citing.

Mita said...

I'm aware of many of the reasons why a citation count alone cannot be used as a means of determining a paper's worth. For example, a paper may be cited many times simply because it has many detractors.

The intention of this post was to highlight a method of finding the most highly cited books in science and in doing so, list the current 'canon' of laboratory and statistical methods that no science library should be without.