Application of a Realistic Evaluation Approach on Three Public Health Programs Aiming at Prevention of Obesity
Table 1
The research process.
Step 1
A topic guide was developed according to the principles of realist evaluation, aiming to explore in three discrete sets of questions the mechanisms, the context, and the outcomes that were produced by the interplay of mechanisms and context.
Step 2
The topic guide was refined for its appropriateness and interview procedures were tested with the paideiatrofi pilot project in Greece by means of in-depth semi-structured interviews with three employees of the project.
Step 3
Methods included participant observation, document analysis, and in-depth semi-structured interviews with 26 employees who delivered the three projects in Scotland, Germany, and England in the native language of each country.
Step 4
Coding strategies from grounded theory were used as a guide for the analysis of the qualitative data, helping to produce program theories in the form of context-mechanisms-outcomes (CMO) configurations. The “coding paradigm” from the grounded theory was combined with the CMO realistic evaluation paradigm, resulting in ‘realistic grounded theories’ to evaluate which mechanisms are in action and which contexts produce effects.
Step 5
Soon after the first two interviews were taken, the analysis of the raw data began. The interviews were translated verbatim into English and the coding was performed with the NVivo qualitative software, version 9.2.81.0. As soon as the initial threads of the theories started to emerge following the analysis of a few interviews, the topic guide was slightly directed at uncovering those respective aspects of the theories. When a new explanation was discovered, it was integrated as a question in the next topic guide, to understand and figure out whether it applies in other contexts as well. When data saturation began to manifest itself in the analysis and the constant refining of the theories, which was informed by the process of further rounds of interviews, it did not produce any new patterns, and the analysis resulted in grounded “context-mechanisms-outcome” configurations.
