exaemplar references and significance of scholarly reviews

Eugene Garfield eugene.garfield at THOMSON.COM
Sat Jan 27 17:59:50 EST 2007

If you export the results of a search in WebofKnowledge into the HistCite software (e.g. your 1000 hits, then the default result after you request an historiograph will be the "exemplar references". 

I would be cautious in describing the significance of review articles in such simplistic terms. I don't recall any studies in which there is an analysis of why people cite reviews. Really good scholarly reviews are a lot more than mere bibliographic surrogates, though they may be useful in that respect as well. Interpretative reviews often play a key role in the historical development of topics. Gene Garfield

-----Original Message-----
From: ASIS&T Special Interest Group on Metrics [mailto:SIGMETRICS at LISTSERV.UTK.EDU] On Behalf Of Morris, Steven (BA)
Sent: Saturday, January 27, 2007 11:55 AM
Subject: Re: [SIGMETRICS] question


I agree that you'd probably only find a weak correlation between number of references cited and citations received if you don't distinguish between the type of paper (review or not) and the way it is used as a reference (well-cited exemplar reference or not).

In my mind the relation is very much tied to the dynamics of specialty growth.  In a recent paper [1] I asserted that after a discovery that prompts the birth of a specialty, there is a period of rapid growth in the specialty where scientists extend the discovery, and present evidence to support those extensions. The discovery paper and other early important papers become heavily cited 'exemplar references' during this growth period. At the end of the growth period, 'consolidation' review papers appear that codify and summarize the newly generated base knowledge in the new specialty. These consolidation papers can become highly cited exemplar references in the sense that they are cited as summaries of collected base knowledge. Some of these reviews become highly cited, some don't,  I suspect it has to do both with timing (written at a point when the newly generated knowledge was ready to be codified), quality and comprehensiveness, and perceived authority of the review author. 

Given the growth and exemplar process described above, you'd expect the

1) Discovery papers, written before all the base knowledge in the specialty is generated, wouldn't cite many references, but would be cited heavily. I think there is evidence out there that discovery papers tend to have few references. I heard Kate McCain mention this once at a
conference ;-),   but I don't have a reference to support that. 

2) Consolidation papers, written to summarize base knowledge immediately after initial growth, would cite many references and be cited heavily. Here, the problem is that only some of the consolidation papers become exceptionally heavily cited exemplar references (the winning reviews that provide the first good consolidation of the new knowledge), while others may just be cited at a 'normal' rate for reviews, which is probably a greater rate than non-review papers.  

Some notes: 

1) There is certainly evidence that the mean number of references per paper increases over time. I've read this in the literature (though I can't recall where) and I've seen this in all specialty specific data sets where I've bothered to check it. I think this is function of specialty growth:  The network of base knowledge in the specialty gets more intricate as the specialty grows and 'fills in the blanks', so authors of later papers have to cite more 'marker references' (Hargens' term [3]) to describe the position of the contribution of their papers
in the network of base knowledge in the specialty...    

2) There is a correlation between the mean number of references per paper and the length of the papers. Evidence for this is given by Abt[2]. So any correlations you find between number of references in the paper and the number of citations it receives may be related to length of papers. 

3) In my experience, I find that the distribution of the number of references per paper is log normally distributed and that the mode of that distribution varies from one specialty to another.  Now, this fact totally baffles me.  What social or cognitive process would cause this
distribution to appear?   Is it tied to the same process that governs
the distribution of length of papers? Some sort of proportional growth process? It's a mystery wrapped in an enigma!  If you figure out what generates that log-normal distribution, I'll send you a one pound bottle of Tupelo honey as a prize....

Some other notes: 

If you want to study the correlation of references per paper to citations received, I suggest the following:

1) Gather specialty-specific collections of papers for your studies. The heterogeneity in a large multiple-specialty study will totally screw up
the statistics...   You should get  about 1000 papers citing about
20,000 references for each specialty study...
2) Separate your references in the collection into 'exemplar' and 'non-exemplar', you can do this by applying a citation threshold, see [1].
3) Arrange the exemplar references serially by the order of their appearance in the specialty.  I have some SQL queries I can send you for doing this. 
4) Look for 'discovery' references at the beginning of this sequence, and 'consolidation' references at the end of the sequence. 
5) Study the correlation for 6 classes of reference: 1- general references, 2- general references less exemplar references, 3- discovery exemplar references, 4- consolidation exemplar references, 5- general review references, 6- general review references less exemplar references. 



[1] Morris, S. A., 2005,  "Manifestation of emerging specialties in journal literature: a growth model of papers, references, exemplars, bibliographic coupling, cocitation, and clustering coefficient distribution" , JASIST, 56(2) 1250-1273 [2] Abt, H. A., 2000,  "The reference-frequency relation in the physical sciences", Scientometrics, 49(3), 443-451. 
[3] Hargens, L. L., 2000, "Using the literature: Reference networks, reference contexts, and the social structure of scholarship"  American Sociological Review, 65(6), 846-865


Steven A. Morris, Ph.D
Electrical Engineer V, Technology Development Group Baker-Atlas/INTEQ Houston Technology Center 2001 Rankin Road, Houston, Texas 77073
Office: 713-625-5055, Cell: 405-269-6576

-----Original Message-----
From: ASIS&T Special Interest Group on Metrics [mailto:SIGMETRICS at LISTSERV.UTK.EDU] On Behalf Of Stephen J Bensman
Sent: Saturday, January 27, 2007 8:30 AM
Subject: Re: [SIGMETRICS] question

It is well known that review articles summarizing research receive on the average more citations than other types of articles.  Your question is considered in the book below:

Narin, F.  (1976).  Evaluative bibliometrics: The use of publication and citation analysis in the evaluation of scientific activity.  Cherry Hill,
NJ: Computer Horizons, Inc.

Here Nariin write:

CHI (Narin, 1976, pp. 183-219) developed its "influence" method in a report prepared for the National Science Foundation.  In this report it criticized Garfield's impact factor as suffering from three basic faults (p. 184). First, although the impact factor corrects for journal size, it does not correct for average length of articles, and this caused journals, which published longer articles such as review journals, to have higher impact factors.

My guess is that you would find no or low correlation between length of references and number of citations, but, if you used a chi-squared test of independence,  you a strong positive association with review articles dominant in the high reference/high citation cell.  As usual,It would be best to do this test with well-defined subject sets than globally to avoid the influence of exogenous subject variables.  However, Narin seems to have been of a different opinion in respect to correlation, so you might look at what he did.


Ronald Rousseau <ronald.rousseau at KHBO.BE>@listserv.utk.edu> on 01/27/2007 07:33:34 AM

Please respond to ASIS&T Special Interest Group on Metrics
       <SIGMETRICS at listserv.utk.edu>

Sent by:    ASIS&T Special Interest Group on Metrics
       <SIGMETRICS at listserv.utk.edu>

To:    SIGMETRICS at listserv.utk.edu
cc:     (bcc: Stephen J Bensman/notsjb/LSU)

Subject:    [SIGMETRICS] question

Dear colleagues,

Is there a positive correlation between the length of a reference list of a publication and the number of citations received? Is this true (or not) in general, i.e. considering all types of publication? And what if one only considers 'normal articles', this is when reviews and letters (and other short
communications) are not taken into account?

Can someone point me to a reference?



Ronald Rousseau
KHBO (Association K.U.Leuven)- Industrial Sciences and Technology
Zeedijk 101    B-8400  Oostende   Belgium
Guest Professor at the Antwerp University School for Library and Information
   Science (UA - IBW)
E-mail: ronald.rousseau at khbo.be
web page:  http://users.telenet.be/ronald.rousseau

This message was sent using IMP 3.2.8, the Internet Messaging Program.

No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.432 / Virus Database: 268.17.12/653 - Release Date: 1/26/2007 11:11 AM

No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.5.432 / Virus Database: 268.17.12/653 - Release Date: 1/26/2007 11:11 AM

More information about the SIGMETRICS mailing list