Background: Assessment of authorship contribution is often based on unreliable questionnaires.
Objective: To assess if the use of different formats for the disclosure of authorship contributions influences authors' compliance with the criteria of the International Committee of Medical Journal Editors (ICMJE).
Design: Randomized study.
Participants and measurements: Eight hundred sixty-five authors of 181 manuscripts submitted to the Croatian Medical Journal from January to July 2005 were randomly allocated into 2 groups: 456 authors (94 manuscripts) received an ordinal rating form to rate their contributions to the submitted manuscript in 12 categories on a scale from 0 (none) to 4 (full), whereas 409 authors (87 manuscripts) received a binary rating form to tick the categories in which they made a contribution.
Results: The ordinal rating form identified twice as many authors (87.9%) as meeting the ICMJE criteria than the binary rating form (39.2%, P < .001). The group answering the ordinal rating form also had 5 times more manuscripts (71.6%) with all authors meeting the ICMJE criteria than the binary rating form group (15.5%, P < .001). The fraction of authors who reported contributions on each item on the binary rating form was similar to the fraction of authors who reported at least moderate participation to the same items on the ordinal rating form except "Final approval of the article."
Conclusions: Ordinal scales for reporting authors' contributions to manuscripts were more sensitive than tick boxes for assessing the appropriateness of authorship. The exception is "Final approval of the article," which should be considered a dichotomous variable and may not be appropriate for the ICMJE definition of authorship.