Any Data or None at All?

Publisher Website
Google Scholar
Abstract
Accuracies of self-reports of household energy consumption were examined among a sample of 2090 electricity and 1699 fuel oil and natural gas consumers. Information obtained from self-reports allowed an increase in response of 37% over what could be obtained from utility company files. Errors in measures as a result of self-reporting occurred, ranging between 10.5% and 12.6% and 24.6% and 29.3%, depending upon fuel type and respondents' use of household receipts. There was no apparent tendency toward systematic over-or underreporting, with the exception of underreporting among larger users. Selfreports correlated more highly with actual consumption than did scores computed to replace missing data. Relationships among variables did not vary significantly between subgroups of respondents for which utility records were and were not obtained. It is concluded that no great error is introduced into findings by researchers' use of self-reports of household energy consumption, either in terms of parameter estimation or in the calculation of relationships among variables. At the sametime, use of self-reports allows the number of missing observations to be reduced.