The availability of the COMPUSTAT tapes and other computerized data bases has increased greatly the quantity of financial research, but some critics question whether it has led to a concomitant increase in the quality of research findings. They note that the COMPUSTAT tapes contain incorrect data and that studies based on the tapes may be biased because they include only large, well known firms or only firms that have been in existence over a specified period.
Although the selection of only large firms may tend to bias the research sample in favor of firms that are more successful and possibly less risky than the general population, this should not be a problem as long as users of such research realize that its conclusions may only apply to this particular type of firm. The effects of survivorship bias—resulting from the exclusion of firms that are delisted during the study period—are less clear. In order to validate results, the researcher may have to determine the number of such firms and attempt to appraise the sensitivity of his results to the problem.
Erroneous data may create major problems, since even small error rates can have a significant effect on results. The researcher should attempt to detect errors through statistical methods and should take extreme care to make sure that he understands the definitions used by his data source and adjusts data from two or more sources so that they are absolutely consistent.