European Social Survey (ESS)

The barometer of political opinion and behavior across the continent.

One table per country with one row per sampled respondent.
A complex sample designed to generalize to residents aged 15 and older in participating nations.
Released biennially since 2002.
Headquartered at City, University of London and governed by a scientific team across Europe.

Please skim before you begin:

Findings from the European Social Survey
Wikipedia Entry
This human-composed haiku or a bouquet of artificial intelligence-generated limericks

# pent up belief gauge
# open border monarchists
# survey for your thoughts

Download, Import, Preparation

Register at the ESS Data Portal at https://ess-search.nsd.no/.
Choose ESS round 8 - 2016. Welfare attitudes, Attitudes to climate change.
Download the integrated file and also the sample design (SDDF) files as SAV (SPSS) files:

library(foreign)

ess_int_df <- 
    read.spss( 
        file.path( 
            path.expand( "~" ) , 
            "ESS8e02_2.sav" 
        ) ,
        to.data.frame = TRUE ,
        use.value.labels = FALSE
    )

ess_sddf_df <-
    read.spss(
        file.path(
            path.expand( "~" ) ,
            "ESS8SDDFe01_1.sav"
        ) ,
        to.data.frame = TRUE ,
        use.value.labels = FALSE
    )
    

ess_df <-
    merge( 
        ess_int_df , 
        ess_sddf_df , 
        by = c( 'cntry' , 'idno' ) 
    )

stopifnot( nrow( ess_df ) == nrow( ess_int_df ) )

Save locally

Save the object at any point:

# ess_fn <- file.path( path.expand( "~" ) , "ESS" , "this_file.rds" )
# saveRDS( ess_df , file = ess_fn , compress = FALSE )

Load the same object:

# ess_df <- readRDS( ess_fn )

Survey Design Definition

Construct a complex sample survey design:

library(survey)

options( survey.lonely.psu = "adjust" )

ess_df[ , 'anweight' ] <-
    ess_df[ , 'pspwght' ] *
    ess_df[ , 'pweight' ] *
    10000

ess_design <- 
    svydesign(
        ids = ~psu ,
        strata = ~stratum ,
        weights = ~anweight ,
        data = ess_df ,
        nest = TRUE
    )

Variable Recoding

Add new columns to the data set:

ess_design <- 
    update( 
        ess_design , 
        
        one = 1 ,
        
        gndr = factor( gndr , labels = c( 'male' , 'female' ) ) ,
        
        netusoft =
            factor(
                netusoft ,
                levels = 1:5 ,
                labels = c( 'Never' , 'Only occasionally' ,
                    'A few times a week' , 'Most days' , 'Every day' )
            ) ,
            
        belonging_to_particular_religion = as.numeric( rlgblg == 1 )
    )

Analysis Examples with the `survey` library

Unweighted Counts

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

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

svyby( ~ one , ~ cntry , ess_design , unwtd.count )

Weighted Counts

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

svytotal( ~ one , ess_design )

svyby( ~ one , ~ cntry , ess_design , svytotal )

Descriptive Statistics

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

svymean( ~ ppltrst , ess_design , na.rm = TRUE )

svyby( ~ ppltrst , ~ cntry , ess_design , svymean , na.rm = TRUE )

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

svymean( ~ gndr , ess_design , na.rm = TRUE )

svyby( ~ gndr , ~ cntry , ess_design , svymean , na.rm = TRUE )

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

svytotal( ~ ppltrst , ess_design , na.rm = TRUE )

svyby( ~ ppltrst , ~ cntry , ess_design , svytotal , na.rm = TRUE )

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

svytotal( ~ gndr , ess_design , na.rm = TRUE )

svyby( ~ gndr , ~ cntry , ess_design , svytotal , na.rm = TRUE )

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

svyquantile( ~ ppltrst , ess_design , 0.5 , na.rm = TRUE )

svyby( 
    ~ ppltrst , 
    ~ cntry , 
    ess_design , 
    svyquantile , 
    0.5 ,
    ci = TRUE , na.rm = TRUE
)

Estimate a ratio:

svyratio( 
    numerator = ~ ppltrst , 
    denominator = ~ pplfair , 
    ess_design ,
    na.rm = TRUE
)

Subsetting

Restrict the survey design to voters:

sub_ess_design <- subset( ess_design , vote == 1 )

Calculate the mean (average) of this subset:

svymean( ~ ppltrst , sub_ess_design , na.rm = TRUE )

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( ~ ppltrst , ess_design , na.rm = TRUE )

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

grouped_result <-
    svyby( 
        ~ ppltrst , 
        ~ cntry , 
        ess_design , 
        svymean ,
        na.rm = TRUE 
    )
    
coef( grouped_result )
SE( grouped_result )
confint( grouped_result )
cv( grouped_result )

Calculate the degrees of freedom of any survey design object:

degf( ess_design )

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

svyvar( ~ ppltrst , ess_design , na.rm = TRUE )

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

# SRS without replacement
svymean( ~ ppltrst , ess_design , na.rm = TRUE , deff = TRUE )

# SRS with replacement
svymean( ~ ppltrst , ess_design , na.rm = TRUE , deff = "replace" )

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

svyciprop( ~ belonging_to_particular_religion , ess_design ,
    method = "likelihood" , na.rm = TRUE )

Regression Models and Tests of Association

Perform a design-based t-test:

svyttest( ppltrst ~ belonging_to_particular_religion , ess_design )

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

svychisq( 
    ~ belonging_to_particular_religion + gndr , 
    ess_design 
)

Perform a survey-weighted generalized linear model:

glm_result <- 
    svyglm( 
        ppltrst ~ belonging_to_particular_religion + gndr , 
        ess_design 
    )

summary( glm_result )

Replication Example

This example matches statistics and confidence intervals within 0.1% from the Guide to Using Weights and Sample Design Indicators with ESS Data:

published_proportions <- c( 0.166 , 0.055 , 0.085 , 0.115 , 0.578 )

published_lb <- c( 0.146 , 0.045 , 0.072 , 0.099 , 0.550 )

published_ub <- c( 0.188 , 0.068 , 0.100 , 0.134 , 0.605 )

austrians <- subset( ess_design , cntry == 'AT' )

( results <- svymean( ~ netusoft , austrians , na.rm = TRUE ) )

stopifnot( all( round( coef( results ) , 3 ) == published_proportions ) )

( ci_results <- confint( results ) )

stopifnot( all( abs( ci_results[ , 1 ] - published_lb ) < 0.0015 ) )

stopifnot( all( abs( ci_results[ , 2 ] - published_ub ) < 0.0015 ) )

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 ESS users, this code replicates previously-presented examples:

library(srvyr)
ess_srvyr_design <- as_survey( ess_design )