4
Residential Energy Consumption Survey Program History and Design
The goal of the Residential Energy Consumption Survey (RECS) is to collect comprehensive information about residential energy use in the United States. Unlike the Commercial Buildings Energy Consumption Survey (CBECS), the RECS is not a building survey, but rather it is a survey of a national sample of housing units. In 2009 the sample represented approximately 113.6 million housing units in the United States. The design of the RECS is more straightforward than that of the CBECS, and the U.S. Energy Information Administration (EIA) is able to benefit from the vast experience of the federal statistical system and of other research organizations with nationwide household data collections. The scope of the survey has not changed significantly since its inception except for the EIA’s efforts to meet an increasing demand for state-level data. The 1993 RECS was the first survey that produced state-level data for the four most populous states in the United States: California, Texas, New York, and Florida. In 2009, EIA was able to triple the RECS sample size and produce state-level data for 16 states, which together account for approximately 65 percent of total household energy consumption in the United States. This expansion also enabled EIA to produce estimates for groups of states that together represent a smaller geographic entity than a census division. Table 4-1 summarizes the evolution of the RECS design.
The RECS collects information about the physical characteristics and energy consumption of the surveyed housing units as well as the demographic characteristics of the households. As is the case with the CBECS,
TABLE 4-1 RECS Data Collection Years and Survey Characteristics
Year |
Completed Household Interviews |
Survey Characteristics |
1978 |
4,081 |
The first survey was called the National Interim Energy Consumption Survey (NIECS). The sample was based on a household survey sample developed by the data collection contractor for a previous survey. |
1979 |
4,033 |
The Household Screener Survey (HSS) was conducted as a follow-up to the NIECS, with the same sample being recontacted. |
1980 |
6,051 |
A new multistage area probability sample was developed based on the 1970 decennial census. The name of the survey was changed to the Residential Energy Consumption Survey (RECS). A rotating panel component was introduced. The smallest level of geographic detail for which data were available was the census division. |
1981 |
6,269 |
No major changes to the sample design. The smallest level of geographic detail for which data were available was the census division. |
1982 |
4,724 |
No major changes to the sample design. The smallest level of geographic detail for which data were available was the census division. |
1984 |
5,682 |
A new sampling frame was developed based on the 1980 decennial census. The smallest level of geographic detail for which data were available was the census division. |
1987 |
6,229 |
No major changes to the sample design. The reference period was changed from April through March to the calendar year. The smallest level of geographic detail for which data were available was the census division. |
1990 |
5,095 |
No major changes to the sample design. The smallest level of geographic detail for which data were available was the census division. |
1993 |
7,111 |
A new multistage area probability sample was developed based on the 1990 decennial census. The rotating panel component was discontinued. State-level data were available for the four most populous states (California, Texas, New York, and Florida) in addition to the data by census division. |
TABLE 4-1 Continued
Year |
Completed Household Interviews |
Survey Characteristics |
1997 |
5,902 |
No major changes to the sample design. State-level data were available for the four most populous states. |
2O01 |
4,822 |
No major changes to the sample design. State-level data were available for the four most populous states. |
2005 |
4,382 |
A new sampling frame was developed based on the 2000 decennial census. State-level data were available for the four most populous states in addition to the data by census division. |
2009 |
12,083 |
Most of the sampling frame was based on a commercial version of the USPS delivery sequence file. Most of the segments were new, but some segments from the 2005 sampling frame were updated for 2009. State-level data were available for 16 states in addition to the data by census division. |
SOURCE: U.S. Energy Information Administration (http://www.eia.gov/consumption/residential/index.cfm [December 2011]).
the RECS is conducted in two data collection stages: a housing unit survey and an energy supplier survey. In the case of some rental units, information is collected from a rental agent in addition to the residents, using a modi-fed version of the questionnaire. And, as is the case with the CBECS, only the energy supplier survey is mandatory. Box 4-1 shows the topics included on the RECS household survey. For more information on the RECS data collection instruments, including the full list of questions included in each of the questionnaires, see http://205.254.135.7/consumption/residential/surveyforms.cfm [December 2011].
The RECS uses a multistage area probability sample of occupied housing units that serve as the primary residence for a household for at least 6 months in a particular year. Under the current design the first step in the sample selection for the RECS involves randomly choosing counties to serve as primary sampling units (PSUs). Over the years, the sample has included anywhere from 103 to 441 PSUs. Typically the counties selected are then subdivided into segments composed of census blocks, and a sample of these segments is randomly selected. Within the selected segments, field staff members list all housing units in order to create a housing unit frame.
BOX 4-1
2009 RECS Household Questionnaire Main Topics
Housing unit characteristics
Kitchen appliances
Home appliances and electronics
Space heating
Water heating
Air conditioning
Miscellaneous topics
Fuels used
Housing unit measurements
Fuel bills
Residential transportation
Household characteristics
Energy assistance (supplement)
SOURCE: U.S. Energy Information Administration (http://205.254.135.7/consumption/residential/surveyforms.cfm [December 2011]).
In 2009 a commercially available version of the U.S. Postal Service (USPS) delivery sequence file (DSF) was used to perform the listing, where feasible. A traditional field listing was used when DSF-based listing was not possible, as was the case in many rural segments. The sampling frame is typically updated from one data collection to the next in order to account for such changes as new construction.
In the early days, the household portion of the RECS was conducted in the form of in-person interviews, using paper and pencil. Beginning with the 1997 data collection, the survey was moved to computer-assisted personal interviewing. The supplier survey was a mail survey before 2009, when the web became the primary mode of data collection.