Sample Design is part of many evaluation and research efforts, but it is addressed separately here because it requires a special set of skills. Knowledge of statistics and the sample design formula is essential, and experience with a wide variety of sample designs is helpful for minimizing the cost of research studies while ensuring confidence in the precision of the results.

Commercial and Industrial Energy Efficiency Rebate Program Evaluations. Mary used non-proportional stratified sample designs with ratio estimation to minimize the number of project reviews needed for reliable estimation of program impacts. The non-proportional stratification increases the probabilty of selecting large projects for review, which improves accuracy while reducing costs. Using ratio estimation uses pre-existing information, in this case the size of the project, to reduce the overall number of project reviews needed to achieve a required precision and reliability. This sampling method was used for multiple jurisdictions, including Commonwealth Edison, AEP-Ohio, PECO, and Wisconsin Public Power, Inc. (various years)

Northwest Energy Audit Evaluation. Mary created a nested sample design for this study. A large number of participants received a telephone survey, asking them if they had taken action on the recommendations they received from their home energy audit. Then, a smaller number of respondents from the telephone sample received an in-home visit. During the in-home visit, the accuracy of the telephone responses was checked. Ratios of results from in-home visits compared to telephone responses were used to project in-home visit results to the full population. This nested sampling design was more cost-effective than performing an in-home visit for everyone in the sample. (2008)