Residential Energy Consumption Survey (RECS)

A periodic study conducted to provide detailed information about energy usage in U.S. homes.

One table with one row per sampled housing unit.
A complex sample survey designed to generalize to U.S. homes occupied as primary residences.
Released approximately every five years since 1979.
Prepared by the Energy Information Administration, with help from IMG-Crown and RTI International.

Please skim before you begin:

Annual Energy Outlook 2023
Household Characteristics Technical Documentation Summary
A haiku regarding this microdata:

# housing code dogma
# even satan ceased sweat since
# eighth sin: central air

Download, Import, Preparation

Download and import the most recent sas file:

library(haven)

sas_tf <- tempfile()

sas_url <- "https://www.eia.gov/consumption/residential/data/2020/sas/recs2020_public_v2.zip"

download.file( sas_url , sas_tf , mode = 'wb' )

recs_tbl <- read_sas( sas_tf )

recs_df <- data.frame( recs_tbl )

names( recs_df ) <- tolower( names( recs_df ) )

recs_df[ , 'one' ] <- 1

Save Locally

Save the object at any point:

# recs_fn <- file.path( path.expand( "~" ) , "RECS" , "this_file.rds" )
# saveRDS( recs_df , file = recs_fn , compress = FALSE )

Load the same object:

# recs_df <- readRDS( recs_fn )

Survey Design Definition

Construct a complex sample survey design:

library(survey)

recs_design <-
    svrepdesign(
        data = recs_df ,
        weight = ~ nweight ,
        repweights = 'nweight[1-9]+' ,
        type = 'JK1' ,
        combined.weights = TRUE ,
        scale = 59 / 60 ,
        mse = TRUE
    )

Variable Recoding

Add new columns to the data set:

recs_design <- 
    
    update( 
        
        recs_design , 
        
        main_heating_fuel = 
            factor(
                fuelheat ,
                levels = c( -2 , 5 , 1 , 2 , 3 , 7 , 99 ) ,
                labels = 
                    c(
                        'Not applicable' , 
                        'Electricity' , 
                        'Natural gas from underground pipes' , 
                        'Propane (bottled gas)' , 
                        'Fuel oil' , 
                        'Wood or pellets' , 
                        'Other' 
                    )
            ) ,

        rooftype =
            factor(
                rooftype ,
                levels = c( -2 , 1:6 , 99 ) ,
                labels =
                    c(
                        'Not applicable' ,
                        'Ceramic or clay tiles' ,
                        'Wood shingles/shakes' ,
                        'Metal' ,
                        'Slate or synthetic slate' ,
                        'Shingles (composition or asphalt)' ,
                        'Concrete tiles' ,
                        'Other'
                    )
            ) ,
            
        swimpool_binary =
            ifelse( swimpool %in% 0:1 , swimpool , NA )
            
    )

Analysis Examples with the `survey` library

Unweighted Counts

Count the unweighted number of records in the survey sample, overall and by groups:

sum( weights( recs_design , "sampling" ) != 0 )

svyby( ~ one , ~ main_heating_fuel , recs_design , unwtd.count )

Weighted Counts

Count the weighted size of the generalizable population, overall and by groups:

svytotal( ~ one , recs_design )

svyby( ~ one , ~ main_heating_fuel , recs_design , svytotal )

Descriptive Statistics

Calculate the mean (average) of a linear variable, overall and by groups:

svymean( ~ totsqft_en , recs_design )

svyby( ~ totsqft_en , ~ main_heating_fuel , recs_design , svymean )

Calculate the distribution of a categorical variable, overall and by groups:

svymean( ~ rooftype , recs_design )

svyby( ~ rooftype , ~ main_heating_fuel , recs_design , svymean )

Calculate the sum of a linear variable, overall and by groups:

svytotal( ~ totsqft_en , recs_design )

svyby( ~ totsqft_en , ~ main_heating_fuel , recs_design , svytotal )

Calculate the weighted sum of a categorical variable, overall and by groups:

svytotal( ~ rooftype , recs_design )

svyby( ~ rooftype , ~ main_heating_fuel , recs_design , svytotal )

Calculate the median (50th percentile) of a linear variable, overall and by groups:

svyquantile( ~ totsqft_en , recs_design , 0.5 )

svyby( 
    ~ totsqft_en , 
    ~ main_heating_fuel , 
    recs_design , 
    svyquantile , 
    0.5 ,
    ci = TRUE 
)

Estimate a ratio:

svyratio( 
    numerator = ~ totcsqft , 
    denominator = ~ totsqft_en , 
    recs_design 
)

Subsetting

Restrict the survey design to households that cook three or more hot meals per day:

sub_recs_design <- subset( recs_design , nummeal == 1 )

Calculate the mean (average) of this subset:

svymean( ~ totsqft_en , sub_recs_design )

Measures of Uncertainty

Extract the coefficient, standard error, confidence interval, and coefficient of variation from any descriptive statistics function result, overall and by groups:

this_result <- svymean( ~ totsqft_en , recs_design )

coef( this_result )
SE( this_result )
confint( this_result )
cv( this_result )

grouped_result <-
    svyby( 
        ~ totsqft_en , 
        ~ main_heating_fuel , 
        recs_design , 
        svymean 
    )
    
coef( grouped_result )
SE( grouped_result )
confint( grouped_result )
cv( grouped_result )

Calculate the degrees of freedom of any survey design object:

degf( recs_design )

Calculate the complex sample survey-adjusted variance of any statistic:

svyvar( ~ totsqft_en , recs_design )

Include the complex sample design effect in the result for a specific statistic:

# SRS without replacement
svymean( ~ totsqft_en , recs_design , deff = TRUE )

# SRS with replacement
svymean( ~ totsqft_en , recs_design , deff = "replace" )

Compute confidence intervals for proportions using methods that may be more accurate near 0 and 1. See ?svyciprop for alternatives:

svyciprop( ~ swimpool_binary , recs_design ,
    method = "likelihood" , na.rm = TRUE )

Regression Models and Tests of Association

Perform a design-based t-test:

svyttest( totsqft_en ~ swimpool_binary , recs_design )

Perform a chi-squared test of association for survey data:

svychisq( 
    ~ swimpool_binary + rooftype , 
    recs_design 
)

Perform a survey-weighted generalized linear model:

glm_result <- 
    svyglm( 
        totsqft_en ~ swimpool_binary + rooftype , 
        recs_design 
    )

summary( glm_result )

Replication Example

This example matches the statistic, standard error, and relative standard error shown on PDF page 8 of Using the microdata file to compute estimates and relative standard errors (RSEs)

sas_v1_tf <- tempfile()

sas_v1_url <- "https://www.eia.gov/consumption/residential/data/2020/sas/recs2020_public_v1.zip"

download.file( sas_v1_url , sas_v1_tf , mode = 'wb' )

recs_v1_tbl <- read_sas( sas_v1_tf )

recs_v1_df <- data.frame( recs_v1_tbl )

names( recs_v1_df ) <- tolower( names( recs_v1_df ) )

recs_v1_design <-
    svrepdesign(
        data = recs_v1_df ,
        weight = ~ nweight ,
        repweights = 'nweight[1-9]+' ,
        type = 'JK1' ,
        combined.weights = TRUE ,
        scale = 59 / 60 ,
        mse = TRUE
    )

recs_v1_design <- 
    update( 

        recs_v1_design , 

        natural_gas_mainspace_heat = as.numeric( fuelheat == 1 )
        
    )
    
result <-
    svytotal( 
        ~ natural_gas_mainspace_heat , 
        recs_v1_design 
    )

stopifnot( round( coef( result ) , 0 ) == 56245389 )
stopifnot( round( SE( result ) , 0 ) == 545591 )
stopifnot( round( 100 * SE( result ) / coef( result ) , 2 ) == 0.97 )

Analysis Examples with `srvyr`

The R srvyr library calculates summary statistics from survey data, such as the mean, total or quantile using dplyr-like syntax. srvyr allows for the use of many verbs, such as summarize, group_by, and mutate, the convenience of pipe-able functions, the tidyverse style of non-standard evaluation and more consistent return types than the survey package. This vignette details the available features. As a starting point for RECS users, this code replicates previously-presented examples:

library(srvyr)
recs_srvyr_design <- as_survey( recs_design )