Schenk, Paulina Margarete; (2022) Developing an ontology of mechanisms of action in behaviour change interventions. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Schenk_10153697_Thesis_sig_removed.pdf Download (12MB) | Preview

Abstract

Background: Behaviour change interventions can influence behaviours central to health and sustainability. To design better interventions, a strong evidence base about ‘why’ interventions work is needed, i.e., their mechanisms of action (MoAs). MoAs are often labelled and defined inconsistently across intervention reports, creating challenges for understanding interventions and synthesising evidence. An ontology can address this problem by providing a classification system that labels and defines classes for MoAs and their relationships. Aims: To develop an ontology of MoAs in behaviour change interventions, and to explore challenges in understanding MoAs and their links to behaviour change techniques (BCTs) Methods: Behavioural scientists’ challenges to understanding MoAs and BCT-MoA links were investigated using a thematic analysis (Study 1 [S1]). To help better understand MoAs, Studies 2-7 developed the MoA Ontology: (S2) Identifying and grouping MoAs from 83 behavioural theories; (S3) Converting the groupings into an ontology by drawing on relevant ontologies; (S4) Restructuring the ontology to be more usable and ontologically correct; (S5) Applying and refining the ontology to code MoAs in 135 intervention reports; (S6) Nine behavioural scientists reviewing the ontology’s comprehensiveness and clarity, informing revisions; (S7) Investigating the inter-rater reliability of researchers double-coding MoAs in reports using the ontology, informing changes to the ontology. Results: Study 1 suggested challenges to understanding broad and underspecified MoAs. To form the basis of a detailed ontology, Study 2 identified 1062 MoAs and formed 104 MoA groups. Building on these groups, Studies 3-7 created the MoA Ontology, which had 261 classes (e.g., ‘belief’) on seven hierarchical levels. Inter-rater reliability was ‘acceptable’ for researchers familiar with the ontology but lower for researchers unfamiliar with the ontology (Study 7). Conclusions: The developed ontology captured MoAs with greater detail than previous classification systems. With its clear class labels and definitions, the ontology provides a controlled vocabulary for MoAs.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item