Citation Indexing:
ISI's Web of Science

There are three strategies that most of us use in finding materials on some piece of research we are working on:

1.  Following references back to other work: We look through the documents (books, monographs, articles) that we already know about to find references  (footnotes, endnotes, etc., that we just generally call "citations") made by their authors to other, former works.  We get and read those works too, because we assume they will, in some way, are related to the relevant document's content.

2.  Bibliographies and indexing services: We search through bibliographies or abstracting/indexing services, using subject words we think are reflective of what it is we are looking for.

3.  Consult a subject expert: We talk to a researcher or scholar (maybe the professor who taught us in a related course) who knows the area, and who advises us of what authors/researchers we should be familiar with.

The essential problem with the first two strategies above is that they are only going to point you to research literature that is older than they themselves are.  If you are reading a 1995 article on cognitive dissonance, you just won't find references to more recent literature in it than at least a couple of years before the date of the article that referenced it.  It is the same with much of the bibliographic literature too: unless a bibliography is a recurrent publication, or is made available online in a continuously updated fashion, it can not maintain entries for the very, very recently published research materials.

The third method--going to a scholar who keeps up with who is working on what aspects of some topic--is a wonderful way of getting an evaluated assessment of who is now working in the research area, and an assessment of what they are saying in their research.  But, it is also the method that is not generally available to most of us.  This is a form of inquiry that depends on the requestor's professional/academic relationship with the scholar being queried.  A colleague of the scholar will probably be able to get this information, and a student of the scholar might too, but many of us would not: we have no association with the scholar; we are just taking up his or her time.

But there is another strategy, a fourth technique, that can be employed to locate recent research: citation indexing. 

4.  Citation indexing

Citation indexing makes links between books and articles that were written in the past and articles that make reference to ("cite") these older publications.  In other words, it is a technique that allows us to trace the use of an idea (an earlier document) forward to others who have used ("cited") it.  The evidence that we take as indicating this "relationship" between earlier research and subsequent research are the references or footnotes or endnotes (citations) in the more recent work.

If you will, there is an informational "ancestry" of a current idea as it is expressed in the literature.  That ancestry is noted by the author of the current idea (his journal article) through his footnotes--his research report's citations. 

Well, think of the, say, 12 citations made in a current journal article about some research topic.  Just as there might be four "assigned" subject headings to this current journal article for indexing purposes in some indexing database, we could also find the 12 citations to previous journal articles and books "assigned" as headings to "index" this article under as well.  Were we to do that for all of the current journal articles we are adding to our indexing service, we could look up current articles by subject headings or by authors of previously published materials that were cited by more recent articles.  That is the trick: we tread the citations found in recent articles as "index entries" to be put in our indexing database for retrieval purposes.

How does this technique change a searcher's strategy for locating information?  Well, it allows us to follow--forward through time--a concept's or idea's or methodology's use by other scholars.  You see, what the Institute for Scientific Information (ISI)--the company that makes citation indexing services--does is build a database of current articles coming from thousands of journals that it selects to use for this purpose.  For the social sciences, there are something on the order of 8,000 journal titles whose separate issues are analyzed as they are published to see what journal articles are in them.  Those journal articles are represented in the graphic below as the light blue boxes, shown as having been published in one of two current years (2002 or 2003).  Of course, there are thousands upon thousands of actual articles, but we are only showing a small number for illustration purposes.

What ISI does is record a "heading" (sort of like a subject heading, but it isn't a subject at all) for each reference or "citation" that these current authors make in their journal articles back to each previously-published article or book.  We have highlighted two such previous publications:  Smith's 1995 article on some topic, and Jones' 1984 article on some other topic.  Now here is the important, novel feature of citation indexing:  all of the articles or books shown at the bottom of this graphic (the 8 white articles or books) become the indexing access points that you search by in ISI's citation index. 

In other words, if you know that Smith wrote a very important research article on something-or-other in the field of Economics in 1995, you can search ISI's citation index to determine what more recent journal articles have cited her 1995 work. In the example below, we show two journal articles that cited Smith's 1995 work--one from 2002, and the other from 2003.  That is how citation indexing works: you know about an author (or more specifically, an author's particular book or article that has already been published) and you use ISI's citation index to find out who has cited that work.

How ISI's Web of Science works

ISI calls its whole set of online services the Web of Knowledge, but you will find it listed under the OU Libraries LORA collection of database services as Web of Science.  When you are correctly authenticated in the database (your 4+4 and password entered correctly) you are presented with this top menu:

Under the left-side heading "Go directly to a specific product," click on the line labeled "ISI Web of Science."

That will take you to this page:

As indicated in the graphic above, you should next click on the "Full Search" button. 

That takes you to another display, where you will indicate that you are going to be using the part of the ISI database that specifies its social sciences database, "Social Sciences Citation Index," and that you wish to carry out a "Cited Ref Search"-- a cited reference search. 

After clicking on "Cited Reference Search," this display comes up:

You are going to use two areas on this display:

  1. the "Lookup" button

  2. the "CITED AUTHOR" fill-in box

The first thing you should do is formulate the cited author specification, and we will talk about that process in a moment.  But, lets say, without explanation, that you already know of an author and a work that you wish to "trace forward" through citation indexing to see what other journal articles have used this earlier work.  As an example, there is a book by the American pollster, Daniel Yankelovich, called Coming to Public Judgement

We can use ISI's Web of Science citation indexing service to see if there are any later journal articles that have made reference to (cited, used, etc.) his work, which we know was published in 1991 (Syracuse University Press).

What we would do, then, is enter this expression in the "CITED AUTHOR" box:

yankelovich d*

What this expression tells the ISI system is to look for all cited references that have the author whose last name is "yankelovich" and whose first name begins with a "d."  You see, ISI doesn't construct author names in the more standard format of last name, comma, first name, (and sometimes) middle name or initial.  Instead, it uses last name, a space, and a first initial, (and sometimes) a middle initial too.  So by using the asterisk (*), we are denoting truncation or the "wild-card symbol" of any other letter that might occur in authors' names after the "d."

At any rate, what we get as a response from the Web of Science system is several pages of results that all fit our search specification.  The first page of the results looks like this:

If you notice, the citations are organized alphabetically by the third column, the "name" of the cited work.  This is the column we are interested in looking in to find something similar to "coming to public judgement," the title of the book Yankelovich authored in 1991. 

Before clicking on the next display page (the "2" along the display page bar across the top and bottom of each display page), take a look at the separate entries on each line of this display page.  The "Hits" column records the number of journal articles in the ISI database that have cited a particular publication written by Yankelovich.  So, for example, the ISI database has 9 articles indexed that cited something by Yankelovich which is indicated by the shortened title, "Beltway Engaging Pub," whatever that is.

Also note that the amount of information per entry given to you in the ISI database is very, very minimal (skimpy, shortened, abbreviated).  As you will see when you use this service, you must know enough about what a previously published work's author name is, and what the title was of the work, and when it appeared, in order to make effective use of the service.  The entries just doesn't overwhelm the user with detail: the user has to know those details.

Lastly, note that there is a check box on the left side of each line.  If you decide you are interested in seeing what the journal article's are that cited a particular Yankelovich entry (book or article), click that box on.  Later, when you have finished noting what you want to see, you will click the "SEARCH" button, "to find articles that cite selected reference."

First, though, we need to click through to the next page (page 2) to see if our Yankolovich publication, Coming to Public Judgment, appears there.  As it turns out, it does:

You can see that there are a variety of entries that resemble our book, Coming to Public Judgement.  Why the differences?  Well, ISI hires staff members to input this information directly from the new journal articles going into the ISI database, with no "authority control" work being done.  So, if an author's name is spelled out with a middle initial in one journal article, but without in another, that is the what the staff members of ISI use in inputting the author's name.  If an author makes a mistake on citing Yankolovich's book, saying it was published in 1974, that is what the ISI staffers will use.  Notice, if you will, that on this page alone, ISI has what looks like the Coming to Public Judgement book as being published in 1991, 1988, and 1994!  And we haven't even looked at the next page, where there are probably more entried indicating some more journal articles that were citing Yankelovich's book, Coming to Public Judgement.  Just be forewarned: ISI is very, very bad about quality control over its entries.

Without looking on page 4 to see if there are more entries referring to the Yankelovich book, lets now decide what entries we wish to have the ISI system work with for us.  As you can see below, we have clicked on 8 entries, one of which was clearly the usually-used (or "correct") format of citing this book: it was used 28 times.  At any rate, we have indicated that we wish for ISI to show us 35 journal article records that all have cited the Yankelovich book, Coming to Public Judgement.

To do that, all we need do is click on the "SEARH" button.

When we do that, this is the beginning results page that ISI gives us:

These article titles may each be clicked on to reveal, in many cases, an abstract of the article, as well as all of its citations, as well as whether other later journal article authors have cited it:

 Of course, the more recent the article is, the less likely it is going to have been cited already!