Data were extracted in two steps. First, PB read all full texts and extracted the following basic data into an Excel (Microsoft) spreadsheet: study characteristics (authors, year, title, aims/objectives, policy focus, theory used, study design, methods, data sources, funding source); setting (focal nutrition issue, geographical level, jurisdiction name, income-level); outcomes (study conclusions/key findings, commitment outcome). Second, studies were coded in ATLAS.ti (Scientific Software GmbH) using a coding schema derived from the initial framework and refined abductively using constant comparative analysis, whereby the coded concepts were confirmed, integrated, modified and/or added to through iterations of data analysis.21
Data were then synthesised. First, text associated with each code was read in situ by PB and summarised, including: (i) a definition of each factor, identified as what influenced commitment; (ii) the mechanism(s) associated with it, identified as underlying entities, structures or processes that transmitted a causal force between the factor and political commitment (either stated in the study or inferred)22 (link) and (iii) cofactors that amplified, diminished and/or sustained the mechanism. On this basis, we defined ‘context’ as ‘underlying social, economic and physical phenomena’ influencing how the mechanism functioned to generate an outcome.23 Second, any cofactors missed in the first step were identified using the ATLAS.ti code cooccurrence tool.
Data were then synthesised. First, text associated with each code was read in situ by PB and summarised, including: (i) a definition of each factor, identified as what influenced commitment; (ii) the mechanism(s) associated with it, identified as underlying entities, structures or processes that transmitted a causal force between the factor and political commitment (either stated in the study or inferred)22 (link) and (iii) cofactors that amplified, diminished and/or sustained the mechanism. On this basis, we defined ‘context’ as ‘underlying social, economic and physical phenomena’ influencing how the mechanism functioned to generate an outcome.23 Second, any cofactors missed in the first step were identified using the ATLAS.ti code cooccurrence tool.
