The top 100 linked-to pages on UK university web sites: high inlink counts are not usually associated with quality scholarly content

Abstract

High citation counts are usually indicators of high-quality research, but is the same true for counts of links to academic web pages? The belief that links can be used to indicate quality or utility on the Web in general drives the highly successful search engine Google, lending additional pertinence to this question. The 100 highest linked-to pages on university web sites were obtained from link counts between institutions in the UK, excluding same university links. The results split into two parts. (1) It is discovered that 23 of the original top pages owed their high counts to the automatic inclusion of links in web pages on other sites, usually on one other university site. It is concluded that simple link counts are highly unreliable indicators of the average behaviour of scholars. (2) After excluding these 23 pages only one of the top 100 contained scholarly content equivalent to a journal article, although several were e-journals or databases of academic publications. Aside from the 45 university home pages, the most common targets were collections of external links and home pages of collections of subject-specific information or resources. It is concluded that the most highly linked-to pages are those that facilitate access to a wide range of information, rather than providing specific content. Implications of these two findings for the newly emerging area of cybermetrics are discussed.