Designing Evaluation for Complex Multiple-Component Programs Genevieve deAlmeida-Morris National Institute on Drug Abuse National Institutes of Health Dept. of Health and Human Services
Designing Evaluation for Complex Multiple -component Research Programs Research programs are complex – Biomedical research programs impact health and lives for generations - Counting and valuing benefits from the programs is difficult - Programs have time-related effects Benefits from averted incidence of disease are not actual Biomedical research increasing costs industry with ec efficiency only from translation to treatment and prevention
Contextual Dimensions of NIH Programs Multiple components in a program - components working together, providing a service/product for use by other components conducting research - awards as a self-contained component with a set of functions –interdisciplinary research, clinical and translational research, community engagement - development of disciplines from these, research training - components at different stages in function, different progress rates
Contextual Dimensions Context from Administration of the Program - Scientist-managers administering the programs for research conduct and progress for compliance with regulations - With the legal function and authority that evaluation does not have
Contextual Dimensions Contexts from NIH Roadmap Initiative - funded from the common fund co-administration of a component, more than one scientist-manager lead, from different Institutes an over-arching workgroup participating in planning and decision-making, self-selected, from the Institutes - a set of Institute Directors - decisions go through multiple levels of review
Contextual Dimensions In addition - new projects are funded each year - a research component can be added to individual projects funded by an Institute(s), not by the common fund - the RM program must be integrated – integration of components integration of functions
The Stakeholder Context Evaluation Faces Evaluation faces a broader concept of ‘stakeholder’ A much larger group of Program Managers A changing and/or increasing set of PIs conducting the research Independence of Pis in the research
Constraints on the Role of Evaluation The evaluation must remain distinct from administration of the program It has no authority over the conduct of the science It cannot constrain the conduct of the science It must have concern for respondent burden What’s left? - a role of documenting and reporting rather than authoritative program change
Identifying the Challenges for Evaluation Establishing its credibility before a scientific community No equivocation with legal functions and responsibility of the NIH administration Determining what will meet the approval of diverse overseers and stakeholders Methodological requirements, and information collection without respondent burden Levels of assessment tailored to each component
The Challenges for Evaluation The need for everyone to understand evaluation, evaluation terms, in the same way The need for consensus, among evaluation leads, among science program administrators The need for comparability of reported information from one funded project to another
What We Did in Planning the Evaluation for all Contexts Outside of the box of contextual dimensions but still among them, without challenging them …… Putting Wholey, Rutman, Rossi to the test
Evaluation for all Contexts- the requirements Evaluation discussions ‘White papers’, and ‘lessons’ in evaluation to help understanding – - of the total design of an evaluation - to explain each next step - at-a-glance evaluation plans - iteratively among these - or Q & As on rationale/explanation A big effort for understanding, and correct conceptualizing from all
Evaluation for all Contexts The approach in Evaluation - - Evaluation by Objectives - objectives in the Requests for Applications - objectives in the awarded projects
Evaluation for all Contexts….? Conducted Evaluability Assessment – (from the literature) - User surveys – What would be best to show program achievement? What in a program can be subjected to realistic and meaningful evaluation? - Tailored the surveys to the content of components - Or conducted analysis of research applications to categorize them
Evaluation for all Contexts……? To accompany the user survey we developed simple Logic Models of the Program Components to operationalize the program/its components - goals, objectives and activities to achieve them (no time specified for outputs, outcomes)
Evaluation for all Contexts....? We asked (e.g.) : Is this model correct, does it portray the program concept ? What would be best to show program achievement ……… by the required reporting times, what will be ready in the program? We emphasized the need for accountability, but with fairness to the program This generated enthusiasm for a correct program model, an enthusiasm to perfect the model We achieved a crucial step towards ownership of the evaluation A case for ex ante Evaluability Assessment –
Evaluation for all Contexts…..? Persisted with ‘white papers’ or paragraphs, or Q & As on rationale/explanation Were able to guide their participation Were able to develop evaluable models of the Program to share with the program manager stakeholders The models
Evaluation for all Contexts….? Specified definitively phases of the evaluation - Process Evaluation – if reporting time is required before program/funding period completion - Outcome or Goal-oriented Evaluation – at program completion/funding period completion Impact Evaluation Utilization Assessment
Evaluation for all Contexts….? The evaluable models had: - objectives, activities, anticipated outcomes, time to achieve, indicators of achievement - the contexts were not compromised Developed with participation from sc. Pgm staff
Evaluation for all Contexts….? We developed Evaluation Questions – - for the evaluable objectives specified We specified information sources We convinced program staff of the need for the primary source information
So do we have Evaluation Design and Planning for Multiple Contexts? We used the approach and E-A for simpler programs, with much success We introduced these for 4 complex programs National Centers for Biomedical Computing The Roadmap Interdisciplinary Research and Research Training The Roadmap Clinical and Translational Service Awards The Roadmap Epigenomics Program We progressed farthest with the last program Success = the approach in evaluation was accepted; programs were evaluated; evaluation information informative and useful
Evaluation Design and Planning for Multiple Contexts? We have consensus for the evaluation approach Buy-in from the Workgroup Participatory Evaluation and ownership from science program managers No backward steps, and feasible evaluation questions Evaluation that works with the NIH Roadmap concept - can accommodate the different study periods and objectives of new projects from re-issue, and stimulus-funding projects…… …….The contexts are under control ………
Anticipating Some Problems If an ‘integrated program’ is what makes the difference how is it to be measured? – by opinion or objective indicators? - ex post /ex ante? - earliest projects will be conducted before integration – can a ‘begin’ time be specified? - can we show integration to be a program objective or as value added? - integration of components with different functions will be more challenging than components doing the same functions A tailored evaluation with external validity? We may have to ‘pull back’ from some evaluation questions
Designing Evaluation for Complex Multiple-Component Research Programs From our experience, Ex ante Evaluability Assessment Evaluation by Objectives Tailoring the Evaluation Will get Ownership and Participatory Evaluation from the Science Program Managers Accountability with fairness to the program Accommodation of the multiple contexts…without evaluation intervening the contexts
Selected References Rossi, P.H., Freeman, H.E. (1993). Evaluation: A systematic approach, 5th Ed. Sage Publications, Newbury Park, CA. . Rutman, L. (1980) Planning Useful Evaluations: Evaluability Assessment. Beverly Hills:Sage Publications. Roessner, D. (2000) Choice of Measures: Quantitative and qualitative methods and measures in the evaluation of research. Research Evaluation, 8(2) 125-132. Thank you