The purpose of this editorial is to discuss problems in gathering and evaluating data in library and information science. Increasingly I find that I advocate an empirical approach to research in our field of library and information science. Others appear to agree and are writing more works based on explicit empirical sources.
The method uses standard social science techniques for examining the reliability and representativeness of data.
The editorial looks at two specific examples, one an experiment with an outlier and another a published source whose data are potentially incomplete. Sometimes the obvious problems are not really problems at all and sometimes completely unexpected issues invalidate results. This is not a reason for avoiding empirical research or for doubting empirical results every time they appear. But it is a reason for authors to engage in a thorough discussion of their data as a part of any scholarly work.
The editorial argues for a broader and more explicit discussion of data issues in journal articles. Some data is quite usable despite its apparent nastiness and some is just hopelessly bad. But a majority of data in our field lie somewhere between these extremes and need to be discussed as part of the analysis process.
