PEER D5.3 USAGE STUDY Randomised controlled trial results

This report reviews the findings of an experiment to measure the effect of exposing early article versions in repositories on downloads of the version of record at various publishers' web sites. The research design used was a randomised controlled trial (RCT) involving 18,062 articles from 135 journals and 254,721 publisher downloads. 7,183 preprints were randomly allocated to a control group and temporarily withdrawn from four PEER repositories for a period of three months. Articles were allocated randomly on a title-­‐by-­‐title basis to the control group ('PEER-­‐hidden') and to the treatment group ('PEER-­‐visible'). The purpose of the experiment was to see whether this allocation made any difference to use at the publisher web site?