Summary report on the methodological framework for counterfactual development

Janne Artell, Jyrki Aakkula, Heini Toikkanen
2015 Zenodo  
The main issue in constructing the counterfactual is programme participants self-selecting to the programme. In such a case, a simple comparison between participants and non-participants gives a biased estimate of the environmental effects of the programme. The logic model for constructing a counterfactual in evaluating the environmental public good impact determines the possible routes of analysis with piece-by-piece evaluation of the available data. The logic model identifies methods for
more » ... ructing a counterfactual, and explicitly forces the evaluator to assess the possibilities and weaknesses of the available data. There are three general level options for counterfactual construction: evaluation options without comparison groups, qualitative and naïve quantitative evaluation options with an ad-hoc approach to sample selection, and statistics-based evaluation options with an explicit approach to sample selection. A comparison of experiences from the ENVIEVAL public good evaluation case studies shows that the logic model serves the purpose of identifying issues and methodological options opened by existing data. Further, the experiences show that despite the general level guidance of the logic model, data-related issues, and in some cases problems in the methods themselves, can complicate evaluation. As problems are case specific, so are the solutions. However, this report serves to give examples of different cases in constructing a counterfactual. The report does find, however, that current data can be used for evaluation, and can be improved to allow rigorous evaluation procedures - Data availability, timing of monitoring data and its spatial resolution all need actions. Generally, the case studies using direct impact indicators (i.e. the actual environmental impact) had more often data availability related problems. Monitoring is expensive, leading to relatively small data repositories which may not cover non-participant farms or areas. Further, actual environmental impacts are often slow processes which ma [...]
doi:10.5281/zenodo.7058192 fatcat:5ih5cdgggvep7mnbyax2wbfn7y