Sir

The newly revised Publication Manual of the American Psychological Association ranks among the most important books in the behavioural sciences and is one of Amazon.com's best-sellers in all book categories. Yet many research psychologists ignore its stipulations on the importance of data sharing.

The manual prescribes structure and style for scientific manuscripts, and deals with the ethics of publication. In this and earlier versions, authors are instructed not to withhold their raw data from other researchers who wish to verify the conclusions through reanalysis (citing standard 8.14 of the APA Ethics Code). It has been found, however, that 73% of psychologists publishing in high-impact APA journals failed to meet this obligation (J. M. Wicherts et al. American Psychologist 61, 726–728; 2006).

Unfortunately, the revised APA guidelines in the new edition aggravate the situation. They stress that data should be shared by written agreement: the agreement must specify conditions relating to the proposed use of the data (for example, for verifying results or for secondary analysis), limits on dissemination of re-analysis results, and authorship expectations. Although imposing such conditions on other researchers seeking to use the data for their own ends is reasonable, it will hinder data-sharing for verification purposes.

Suppose that statistical flaws become apparent in a published paper. Under the new guidelines, these can't be investigated by an independent researcher unless the author agrees the terms for sharing the raw data. For instance, the author may demand to be a co-author on ensuing publications. It seems to us that the new APA guidelines will impede, rather than advance, the critical assessment of the quality of data analyses in psychology research.