Housing Characteristics Tables

Specific questions on this product may be directed to:

Chip Berry
RECS Survey Manager
Phone: (202) 586-5543
Fax: (202) 586-0018

Consumption & Expenditures Tables

RECS Consumption and Expenditures data for 2009 are currently being collected and processed. EIA anticipates a first release of these data in early 2012.

See the 2005 RECS Consumption and Expenditures data tables.

Specific questions on this product may be directed to:

Chip Berry
RECS Survey Manager
Phone: (202) 586-5543
Fax: (202) 586-0018

Public Use Microdata File

The Residential Energy Consumption Survey (RECS) is a national sample survey that collects energy-related data for housing units occupied as a primary residence and the households that live in them. First conducted in 1978, the 2009 version represents the 13th iteration of the RECS program. Data were collected from 12,083 households selected at random using a complex multistage, area-probability sample design.The sample represents 113.6 million U.S. households, the Census Bureau’s statistical estimate for all occupied housing units in 2009 derived from their American Community Survey (ACS).

For the first time in its history, EIA offers a preliminary 2009 RECS Public Use Microdata File (PUMF) for users who wish to perform custom statistical tabulations and economic analysis.This preliminary file contains housing unit characteristics based on information collected or derived from answers provided by survey respondents. Square footage, weather data, consumption, and expenditure variables will be released in a future version of the PUMF. Data are available in two formats: a comma delimited file and a SAS data file. The comma delimited data file is accompanied by a corresponding “Layout File”, which contains descriptive labels and formats for each data variable. Users should also refer to the “Response Code Labels” file, which contains the descriptive labels for variables and descriptions of the response codes.

Additional data variables will be appended to the 2009 RECS PUMF in the coming months and include household square footage components and totals, weather data (heating and cooling degree days), and a full complement of energy consumption, expenditures and end-uses data for households that completed a RECS interview.

MAJOR CHANGES for the 2009 RECS

Notable survey design revisions, content changes, and variable updates for the 2009 RECS include:

• An expanded sample size for the 2009 RECS allows EIA to release estimates for household characteristics and energy use for 16 States, 12 more than in past rounds of RECS.The variable REPORTABLE_DOMAIN contains these states and other groups of states for which estimates can be computed.
• Expanded data on the type and usage of consumer electronics, including televisions and related devices, computers, and personal electronic devices.
• An expansion of the 10-19 equipment and appliance age range from previous surveys is split into two responses; it is now split into two age groups, 10-14 and 15-19 years.
• An introductory variable in the space heating section (HEATHOME) indicates whether a household had and used space heating equipment in 2009. Therefore, subsequent questions about space heating, unless otherwise noted, were only asked of those respondents with expected space heating consumption for the reference year. Separate variables were added to account for homes that have heating equipment, but did not use it.Similar variables are used in the Air-Conditioning section.
• New data items for recent energy efficiency actions taken by the household, including caulking, weatherstripping, insulation, and home energy audits are added for 2009.

SAMPLE WEIGHTS

The RECS sample was designed to estimate energy characteristics for the national stock of occupied housing units and the households that live in them. To produce estimates from the RECS sample, base sampling weights, which are the reciprocal of the probability of being selected for a RECS interview, were calculated for each sampled housing unit. For example, a housing unit with a base weight of 10,000 represents itself and 9,999 unsampled housing units in the total stock of occupied housing units. The base weights were adjusted to account for survey nonresponse and ratio adjustments were used to ensure that the RECS weights add up to ACS estimates of the number of households for the survey reference year. The variable NWEIGHT in the data file represents the final weight, accounting for different probabilities of selection and rates of response and being adjusted for the ACS housing unit estimates.

The following examples illustrate proper usage of sample weights (NWEIGHT) to calculate survey estimates.

Example 1: Using NWEIGHT to estimate a single response

The respondent with DOEID = 00001 has NWEIGHT = 2,472. Hence this respondent represents a total of 2,472 households. The respondent used 2 refrigerators (NUMFRIG = 2), thus contributing 4,944 (2,472 x 2) refrigerators to the estimated national total of refrigerators used in US households.

Example 2: Using NWEIGHT to estimate number of households

There were 904 respondents that used fuel oil in their homes (USEFO = 1). By adding the NWEIGHT data for these 904 cases, the estimated number of households that use fuel oil is approximately 7,636,350.

Example 3: Using NWEIGHT to estimate percentage of households

The sum of NWEIGHT over all cases is 113,616,229. This is also an estimate of the total number of occupied primary housing units in 2009. Hence, the estimated percent of households that use fuel oil (for any use in the home) is (7,636,350/113,616,229) times 100 equals 6.7 percent.

CONFIDENTIALITY

These data were collected under the authority of the Confidential Information Protection and Statistical Efficiency Act (CIPSEA), as such EIA, project staff and its contractors and agents are personally accountable for protecting the identity of individual respondents. The following steps were taken to avoid disclosure of personally identifiable information on the PUMF.

• Local geographic identifiers of sampled housing units, such as zip codes, were removed.
• Building America Climate Regions with few sample cases (“Very Cold” and “Mixed-Dry”) were combined with the most similar region.
• The year of construction for sampled housing units (YEARMADE) was bottom coded at 1920.
• Two variables were masked to prevent identification of large multiunit residential buildings sampled in 2009, NUMFLRS (number of floors in a 5+ unit apartment building) and NUMAPTS (number of apartments in a 5+ unit apartment building). Households with NUMFLRS greater than 15 were replaced with the mean of the values above 15 by Census region.To give a very simple example, if there were only three households in a Census region with NUMFLRS greater than 15 with NUMFLRS values of 20, 25, and 30, then the NUMFLRS values for all three would be 25. Similarly, households with NUMAPTS greater than 200 were replaced with the mean of the values above 200 by Census region.
• The variable indicating the type of on-site electricity generation (ONSITETYPE) was removed due to too few responses.
• The variable HHAGE (age of the householder) was top-coded at 85.
• Household member ages other than the householder (AGEHHMEM2-14) were categorized.

VARIABLE CODING

Standard Coding for “Don’t Know”, “Refuse”, and “Not Applicable”

Variables that were not imputed use the response codes -9 for “Don’t Know” and -8 for “Refuse”.Variables that are not asked of all respondents use the response code -2 for “Not Applicable”.For example, if a respondent said they did not use any computers at home (COMPUTER = 0) then they were not asked what type of computer is most used at home, thus PCTYPE1 = -2.

Indicator Variables for Fuels and End-Uses

The public microdata file contains variables to indicate the use of major fuels and specific end-uses within each housing unit for 2009. These variables are derived from answers given by each respondent and indicate whether the respondent had access to and actually used the fuel and engaged in the end-use. All indicators are either a 0 or 1 for each combination of major fuel and end-use.  For example, a respondent who says they heated their home with electricity in 2009 will have the derived variable ELWARM = 1. If a respondent says they have equipment but did not use it the corresponding indicator will be 0. As an example, a respondent in a cool climate might have air-conditioning equipment but did not use it in 2009. For this case, ELCOOL would be 0.

Imputation Flags

The "Z variables" are also referred to as "imputation flags." Imputation is a statistical procedure used to fill in missing values for items that are otherwise considered to be complete. Missing values for many, but not all, of the variables were imputed. The imputation flag indicates whether the corresponding non-Z variable was based upon reported data (Z variable = 0) or was imputed (Z variable = 1). There are no corresponding "Z variables" for variables from the RECS questionnaire that were not imputed, variables where there was no missing data, and variables that are not from the questionnaire.

Specific questions on this product may be directed to:

Chip Berry
RECS Survey Manager
Phone: (202) 586-5543
Fax: (202) 586-0018

How does EIA estimate energy consumption and end uses in U.S. homes?

RECS 2009 — Release date: March 28, 2011

EIA administers the Residential Energy Consumption Survey (RECS) to a nationally representative sample of housing units. Specially trained interviewers collect energy characteristics on the housing unit, usage patterns, and household demographics. This information is combined with data from energy suppliers to these homes to estimate energy costs and usage for heating, cooling, appliances and other end uses – information critical to meeting future energy demand and improving efficiency and building design.

RECS uses a multi-stage area probability design to select sample

A multi-stage area probability design ensures the selection
of a representative sample ofhousing units in the United
States.

All housing units in the 50 States and the District of Columbia that are occupied as primary residences are eligible to be included in the RECS sample.

Sample selection begins by randomly choosing counties. The selected counties are then sub-divided into groups of Census blocks called segments and a sample of segments is randomly drawn from the selected counties. Within each selected segment, a list of housing units (sample frame) is created by field listing.¹

The final sample of housing units is randomly selected from the housing unit frame constructed from the selected area segments. This type of sampling is called a multi-stage area probability design. Its proper application ensures that the selected sample represents the entire population of occupied housing units in the United States.

The number of counties, segments, and housing units to be selected are carefully controlled so that RECS produces estimates of average energy consumption at specified levels of precision within the following geographic levels, called domains: National, Census Region, Census Division, and individual states or group of states within Census Divisions.

Two surveys capture energy characteristics for sampled housing units: the Household Survey and the Rental Agent Survey

In the Household Survey, trained interviewers use a standardized questionnaire to collect data from the selected housing units. The field interviewer uses a laptop to record the householder's responses to the survey. This method of collecting data is called Computer-Assisted Personal Interview (CAPI).

Questions in the Household Survey are designed to collect energy-related characteristics of the housing unit ("What is the main fuel used for heating your home?"), as well as energy usage patterns of the household members ("How often is your dishwasher used?").

Where respondents in rental housing units are less sure of their housing unit's energy characteristics, EIA uses the Rental Agent Survey. Those data are collected by phone or in person from the unit's landlord or his/her representative.

All of the data collected from the Household and Rental Agent Surveys go through a series of rigorous statistical processes to ensure the highest possible data quality. These processes include:

• editing
• validation and quality control
• imputation of missing data

EIA produces estimates of end uses of energy by modeling the data from the Household and Energy Supplier Surveys

The flagship product of RECS is the estimate of how much energy is used within the home for heating, cooling, refrigeration, and other end uses. EIA uses RECS to estimate end-use consumption through a non-linear statistical model applied to data from the Household and Energy Supplier Surveys, which disaggregates total energy consumption into end-use components.

These estimates of energy end uses make RECS uniquely important: it is the only survey that provides reliable, accurate and precise trend comparisons of energy consumption between households, housing types, and areas of the country.

Learn More

• American Community Survey
• Precision of RECS estimates in terms of Relative Standard Errors (RSEs)

¹For the first time in its 30-year history, residential addresses from the U.S. Postal Service were used to construct a substantial portion of the 2009 RECS housing unit frame. ²The use of the internet as a primary data collection method for ESS was initiated in the 2009 RECS.

Specific questions on this product may be directed to:

Chip Berry
RECS Survey Manager
Phone: (202) 586-5543
Fax: (202) 586-0018