Open science? Responsiveness to requests for data in a review of smoking cessation interventions.

Abstract

Objective: Little research has examined rates or correlates of adherence to Open Science practices such as data sharing. We investigated how often researchers share data for inclusion in a meta-analysis and their reasons for not sharing data, and tested factors that could be associated with data sharing.

Methods: We requested data for 189 studies (167 authors) as part of a National Cancer Institute-funded meta-analysis of quit intentions and smoking cessation. Authors were contacted via email up to 4 times. We tracked responses, reasons for not sharing data, and coded 23 features of the author team (eg, number of authors and h-index), the request (eg, amount of information requested), and the study (eg, year of publication and preregistration).

Results: Thirty-five percent of authors provided the requested data, 21% responded but did not provide data, and 44% never responded to our request. Of the 37 reasons offered for not sharing data, the most common were loss of access to data (76%) and lack of time (11%). More recent trials, fewer citations, publication in medical (vs. behavioral) journals, and study preregistration were each associated with providing the requested data (Ps < .05).

Conclusions: Contacting authors for unpublished data resulted in a moderate response rate (56%) and modest provision of data (35%). Barriers to data sharing such as access and time constraints highlight challenges faced by behavioral health researchers in promoting transparency. The factors associated with responsiveness underscore the importance of journal policies and Open Science practices in enhancing data sharing.