The unit of analysis (objects of study) in bibliometrics and scientometrics

Slow development of bibliometric theory may be due in part to neglect of the unit of analysis - the objects described by variables and about which inferences are made. Problems include: inferences are often made on units other than those sampled, leading to inappropriate conclusions; units in literature reviews and meta-analysis are often not comparable, thus hindering cumulation of knowledge; confusion when names of sampling units in one study might also be the names of variables in other studies - e.g., no. of citations (variable) to papers (sampling unit) and no. of papers (variable) in journals (sampling unit); loss of information about the unit of analysis, means and variances when data are aggregated. If theory is to advance, scientometrics needs a generic definition of the unit of analysis, a complete list of all known units - classified and structured, meta-analyses, reporting standards - especially when data are aggregated, clear indications of data level (nominal, ordinal, interval, ratio), conventions for including units in titles, abstracts and keyword or subject indexes.