Residential Energy Consumption Survey (RECS)
Does EIA have official EIA energy consumption statistics for counties, cities and ZIP codes?
2009 RECS Features
Heating and cooling no longer majority of U.S. home energy use
March 7, 2013
Newer U.S. homes are 30% larger but consume about as much energy as older homes
February 12, 2013
Where does RECS square footage data come from?
July 11, 2012
RECS data show decreased energy consumption per household
June 6, 2012
The impact of increasing home size on energy demand
April 19, 2012
Did you know that air conditioning is in nearly 100 million U.S. homes?
August 19, 2011
Other End Use Surveys
Methodological Research
Implementing a Mixed-Mode Design for Collecting Administrative Records: Striking a Balance between Quality and Burden
RECS 2009 — Release date: September 5, 2012 full paper
Summary
RECS relies on actual records from energy suppliers to produce robust survey estimates of household energy consumption and expenditures. During the RECS Energy Supplier Survey (ESS), energy billing records are collected from the companies that supply electricity, natural gas, fuel oil/kerosene, and propane (LPG) to the interviewed households.
As Federal agencies expand the use of administrative records to enhance, replace, or evaluate survey data, EIA has explored more flexible, reliable and efficient techniques to collect energy billing records. The ESS has historically been a mail-administered survey, but EIA introduced web data collection with the 2009 RECS ESS. In that survey, energy suppliers self-selected their reporting mode among several options: standardized paper form, on-line fillable form or spreadsheet, or failing all else, a nonstandard format of their choosing.
In this paper, EIA describes where reporting mode appears to influence the data quality. We detail the reporting modes, the embedded and post-hoc quality control and consistency checks that were performed, the extent of detectable errors, and the methods used for correcting data errors. We explore by mode the levels of unit and item nonresponse, number of errors, and corrections made to the data. In summary, we find notable differences in data quality between modes and analyze where the benefits of offering these new modes outweigh the "costs".

