Bank error in your favour? How to gain 3,000 citations in a week

Every Thursday I check my new citations in the Web of Science and submit data change reports for citing references that have not been properly linked to their master record. It is not often that a week passes without me having to submit a data change report. This is largely caused by the fact that many of my publications are of a non-standard type (e.g. the Publish or Perish software, the Journal Quality List) and my initials seem to cause no end of confusion (is it A., A.W., A.-W., A.K., A.W.K., A-W.K.?). Web of Science data entry typists seem to see each variation, however slight (e.g. A.W. vs. A.-W.), as a reason to create separate record. Whilst a little frustrating for me, this is understandable as - given their likely "paygrade" - these employees cannot be expected to make judgment calls and thus erring on the conservative side and only merging exact duplicates makes sense.

Hirsch's h-index article attributed to my name

However, my weekly verification also leads to more disconcerting discoveries, such as those documented in Web of Science: How to be robbed of 10 years of citations in one week! [now resolved]. This week though, I experienced one of those rarest of Monopoly events: a bank error in your favour. My citation level was suddenly boosted with nearly 3,000 citations by inaccucrately crediting me with one of the most highly cited articles in bibliometrics: Hirsch's 2005 h-index article.

The screenshot below shows this week's partial result for my normal "Harzing A" search in the Web of Science Cited Reference search function. Between a list of some of my first-authored articles and articles where I am the last author, you can find Hirsh's 2005 PNAS article with 2996 citations.

How could this have happened?

So how could this have happened? I am puzzled, but I think a clue might be found in another result that I had not spotted before: a stray reference - without a linked master record in the Web of Science - that lists me as the first author of the 2005 Hirsch article.

The error can be verified more easily by searching for "Harzing A" as author and "Proceedings*" as cited work. This clearly shows the non-existing article without a linked record in the Web of Science Core Collection and the original Hirsch article with a linked record.

Referencing a non-existing article

There are two articles that are reported to reference this (non-existing) article. The first does indeed do so as can be seen in the two screenshots below.

The authors seem to be fairly sloppy in their write-up of their methods in general. I can identify at least half a dozen factual mistakes in this short section, including a URL linking to a commercial company website selling CRM systems instead of the correct link to Publish or Perish. So this might just be one of their many mistakes.

The second article doesn't seem to reference the non-existing article, neither in the text nor in the reference list. The first screenshot below shows the only occurrence of Harzing in the text of this article and the second shows the only article attributed to me in the reference list. So I am not sure why it was linked to to the non-existing reference in the Web of Science but that's a separate - and in comparison, minor - problem. Small mistakes like this happen in any database.

The million-dollar question

The biggest question to me is: how did we get from a relatively minor problem - sloppy authors creating a non-existing publication - to a situation where Hirsch's seminal article shows up when you search for my name? Ok, it is better than being robbed of citations, but I would prefer an accurate account of my own citation record :-)

Related blog posts

Are referencing errors undermining our scholarship and credibility?

Web of Science: How to be robbed of 10 years of citations in one week!

Health warning: Might contain multiple personalities

Postscript

Less than 10 hours after posting this blog, the error in the WoS database was fixed. This works faster than a data change report ;-)