Assessing Open Science practices in physical activity behaviour change intervention evaluations

Abstract

Objectives: Concerns on the lack of reproducibility and transparency in science have led to a range of research practice reforms, broadly referred to as 'Open Science'. The extent that physical activity interventions are embedding Open Science practices is currently unknown. In this study, we randomly sampled 100 reports of recent physical activity randomised controlled trial behaviour change interventions to estimate the prevalence of Open Science practices. Methods: One hundred reports of randomised controlled trial physical activity behaviour change interventions published between 2018 and 2021 were identified, as used within the Human Behaviour-Change Project. Open Science practices were coded in identified reports, including: study pre-registration, protocol sharing, data, materials and analysis scripts sharing, replication of a previous study, open access publication, funding sources and conflict of interest statements. Coding was performed by two independent researchers, with inter-rater reliability calculated using Krippendorff's alpha. Results: 78 of the 100 reports provided details of study pre-registration and 41% provided evidence of a published protocol. 4% provided accessible open data, 8% provided open materials and 1% provided open analysis scripts. 73% of reports were published as open access and no studies were described as replication attempts. 93% of reports declared their sources of funding and 88% provided conflicts of interest statements. A Krippendorff's alpha of 0.73 was obtained across all coding. Conclusion: Open data, materials, analysis and replication attempts are currently rare in physical activity behaviour change intervention reports, whereas funding source and conflict of interest declarations are common. Future physical activity research should increase the reproducibility of their methods and results by incorporating more Open Science practices.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 15 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.