The Journal of Information Science has published some interesting research into academic usage of e-journals (Note: subscription required). A reasonable nutshell summary of it would be that journal reading is now overwhelmingly conducted online, with the focus being on dealing with the enormous amount of information that is available, not only from traditional sources but from relatively new formats, like blogs.
The general trend appears to be that the amount of reading is increasing overall, at the expense of the amount of time spent reading individual articles. Not surprisingly, users are tending to skim articles or target specific sections, as well as beginning their research from a general scholarly gateway or one of Google’s sites. It seems that the old usability adage, that online users scan rather than read, has as much force for researchers as anyone else.
A minority of users now view reading an article in its entirety as a rare or even non-existent occurrence. As such, filtering content and facilitating ease of access appear to be primary requirements for users. This would also suggest that journal publishers need to concentrate on search engine optimisation and improving article page layouts for users, especially those who have come from Google or a gateway site, rather than from the site’s homepage.
It is interesting to note that the study's participants ranked blog content as being more important than reports, primary sources, conference papers or working papers. Publishers therefore need to understand the sort of content time-pressed researchers actually consider most valuable, rather than just what they’ve always produced. By the same token, one of the things the participants appreciated about Google was its inclusion of general content (e.g. books) that cover the entirety of the literature in a field, rather than just journal content.
I should perhaps end with a caveat; the research is principally oriented around survey results and supplemented by direct observation, interviews and analytics data. One of the problems with emphasising survey results to this extent is that it violates the cardinal usability principle that studies should pay attention to what users say rather than what they do, with the result that some of their different methodologies conflict. So, for example, the authors conclude that the responses to a question on the amount of time spent ‘reading’ an article were overstated, given that many of the participants did not necessarily regard skimming and reading as distinct. Similarly, observation and search logs contradicted user claims that they used advanced search features as a matter of course. According to their observations, 40% of searches began with simple terms, with filtering only being done after viewing the first set of results. Given that ‘advanced search’ appears to have been rather broadly defined, it’s quite probable that this figure is overstated too – and that’s without starting on the problem of whether users wish to use advanced search options or simply have to.