Paper presented at Energy Program Evaluation: Conservation and Resource Management Conference, Chicago, Illinois
Paper Dated: August, 1989
By: Michael Blasnik
In the process of conducting a pilot weatherization program, the Grass Roots Alliance for a Solar Pennsylvania (GRASP) noticed substantial differences between houses which met the data requirements for evaluation (the evaluation sample) and those which did not (the attrition sample).
GRASP compared the evaluation sample with the attrition sample and an unscreened sample in order to verify and quantify some of these differences and investigate the potential for bias caused by sample attrition.
GRASP discovered significant differences between the evaluation sample and the other groups in terms of initial measured air leakage rate, reduction in leakage rate due to program measures and pre-period fuel consumption.
The evaluation sample houses had significantly tighter envelopes than the unscreened sample and had much smaller air leakage rate reductions from weatherization measures. A rough comparison of fuel usage showed significantly lower consumption for the evaluation sample than the attrition sample.
There appears to be a correlation between the quality of billing data and the thermal integrity of the house. These results imply that low-income weatherization programs may be achieving greater savings than a standard high attrition evaluation would indicate.
GRASP's findings demonstrate attrition bias and call into question the generalizability of many low-income fuel savings evaluations which have comparable sample attrition. Further exploration is needed in evaluation methods which reduce this bias such as cruder billing data analysis techniques, statistical bias reduction techniques, and methods such as short-term submetering which create their own data.
Also see Impact/Process Evaluations
Bound copies of this entire paper are available for $25 each.
For details about obtaining a copy see How To Order Project Reports
