14.2.3 Stray citations are very common in ISI

Stray citations are citations that refer to a publication that is also listed correctly in ISI, but have small errors in their referencing. This can be caused either because of the sloppiness of the referring academic or because of data entry errors by ISI staff.

The problem of stray citations in ISI can be quite significant. One of the most-cited academics in the field of Management Geert Hofstede has published a book called Culture's Consequences. This book was first published in 1980, reprinted in 1984, and published in a 2nd revised edition in 2001. These three versions of the same book respectively have 4677, 664 and 1422 citations.

However, there are also some 150 additional stray citations in ISI's Cited Reference search, all referring to the same book. Some of these entries refer to specific page numbers in the book and hence have been entered as separate entries. Others refer inaccurately to different publication years or misrepresent the books title as Cultural Consequences. In contrast Google Scholar has less than 30 variations of this title.

Many stray entries in ISI are simple misspellings of the title, with some of the weirdest being “CULTURES CIONSEQUENC”, “CULTUES UCULTURES CO”, and “CULTURES OCNSEQUENCE”. In many of these cases, the references were correct in the referring works, but spelling errors were made by ISI data entry staff. Most of these inaccurate references only occur once or twice, but a substantial number has a double-digit number of citations, with three accumulating more than 50 citations. Ironically, none of the more than 7000 citations to this work was correctly entered as the title is Cultures Consequences, not Cultures Consequences. However, as we have seen above, ISI does not seem to be able to deal with apostrophes.

Fortunately, for most authors the problem of stray citations is more modest, but nevertheless annoying. The screenshots below refer to two articles by Margaret Abernethy, with the last three columns referring to the volume of the journal, the first page of the article and the number of citations respectively. Please note that these screenshots were specially constructed for this book. In reality, the different incarnations of the same paper would not normally appear neatly listed together.

The first screenshot refers to a paper in Accounting, Organizations & Society, which is an ISI listed journal. There is a master record which with 27 citations has the bulk of the total number of citations. The second line shows the same article, but without the issue of the journal and with journal name not following the ISI standard notation for this journal, a data entry error. The third line also refers to the same article, but includes a wrong volume and only lists one of the authors initials. The fourth and fifth line respectively are identical to the master record, safe for the fact that the referring author included the wrong year.

Abernethy #1

The second screenshot refers to an article in Accounting & Finance, a journal that is not ISI listed. Hence, there is no official master record for this article in the ISI data-base and citations are spread rather evenly over different versions of the paper, which differ mainly in terms of the journal volume and issue that are listed. The first line shows a very interesting variant of the paper. In addition to missing the authors second initial, the referrer seems to have accidentally included the volume and issue number of one of Abernethy's other papers.

Abernethy #2

If all of these stray citations were correctly merged into one record, the number of citations for the first paper would be 33 instead of 27. The second paper would have a total of 23 citations, even though currently the record with the largest number of citations is 9. So what does it matter? That depends on whether the journal in question is ISI listed or not. If a journal is ISI listed, the stray citations are not reported in ISI's (General) Search function or ISI's citation report. This means that:

If the journal in question is not ISI listed, the consequences are not as serious as these citations are not counted in any of ISI's other databases or analytical tools anyway. However, having your citations to a particular article spread over many different records which are not always easy to match makes it rather difficult to assess the citation impact of individual articles. Therefore, it underestimates the unique contribution that a particular author has made to the field by publishing impactful articles, rather than publishing a lot of articles with only a small of citations.

Data change request

Fortunately, there is a way to get your ISI Cited Reference citation report cleaned up. You can submit a Data Change Request to Thomson Reuters with the references to be corrected at http://science.thomsonreuters.com/techsupport/datachange/. Please make sure you indicate the master record that the stray references need to be merged into. If the article in question is published in an ISI listed journal, the master record will be clickable link highlighted in blue. If the article was not published in an ISI listed journal, you can designate the master record yourself. It is usually best to choose the record with the largest number of citations, as that would normally be the most “correct” reference to the paper.

Requests normally take 3-5 weeks to be processed. In my experience, Thomson Reuters nearly always makes the changes that are requested, provided of course that the records do indeed refer to the same paper. You might think this is quite a lot of work for little reward. However, submitting data change requests is fairly quick, especially if you use the back button on your browser to avoid having to enter all the general information (name, university, email address etc.) again.

I cleaned up my own record by spending about an hour to submit various data changes and now spend five minutes once a month to submit any further changes if necessary. The result is that rather than having citations to my work spread over three full pages with some publications found on three different pages (because referrers used one, two or three initials when referring to the work), my publications now nearly fit on one page. Some publications had more than a dozen different incarnations.

For instance the Journal Quality List that can be downloaded from my website (http://www.harzing.com/jql.htm) was referred to in 16 different ways, with 1-4 citations each. Collecting all these citations into one master record now show 38 citations for this “publication”. Likewise, the numerous stray citations to the Publish or Perish program were collated to 17 citations. As a result, it is much easier for me to convince research evaluators of the significant contribution that these resources make to the academic community. Whether or not cleaning up your ISI Cited Reference citation record is beneficial in your case is something only you can tell. However, I do encourage you to at least check it out.