anes.Rmd

# American National Election Studies (ANES) {-}

[![License: GPL v3](https://img.shields.io/badge/License-GPLv3-blue.svg)](https://www.gnu.org/licenses/gpl-3.0) <img src='https://img.shields.io/badge/Tested%20Locally-Windows%20Laptop-brightgreen' alt='Local Testing Badge'>

A time series recording belief, public opinion, and political participation back to Dewey vs. Truman.

* Most tables contain one row per sampled eligible voter, varying weights like pre- and post-election.

* A complex sample generalizing to eligible voters in the U.S. with some panels to follow individuals.

* Core studies released quadrennially (presidential elections), plus pilot studies (often at midterms).

* Administered by a consortium of universities and funded by the [National Science Foundation](https://www.nsf.gov).

---

## Recommended Reading {-}

Four Example Strengths & Limitations:

✔️ [Time series studies interview both before and after quadrennial elections](https://electionstudies.org/data-center/2024-time-series-study/)

✔️ [Instrument design tested in smaller study prior to inclusion](https://electionstudies.org/data-center/2024-pilot-study/)

❌ [Turnout errors are part of a long-standing problem of turnout over-estimation in surveys](https://electionstudies.org/wp-content/uploads/2022/03/DeBell-Maisel-Benchmark-Report-for-the-ANES-2016-Time-Series-Study.pdf#page=5)

❌ [Prior survey questions not always asked again](https://electionstudies.org/data-tools/anes-continuity-guide/)

<br>

Three Example Findings:

1. [Younger Americans were less politically polarized than older Americans in 2020](https://doi.org/10.1371/journal.pone.0302434).
	
2. [In 2020, 90% of Biden and Trump voters also opted for a congressional candidate of the same party](https://rollcall.com/2024/05/14/will-the-last-ticket-splitting-voter-please-turn-out-the-light/).

3. [Between 1996 and 2016, demographic groups least likely to use the Internet and social media experienced larger changes in political polarization than those more likely to use the Internet](https://doi.org/10.1073/pnas.1706588114).

<br>

Two Methodology Documents:

> [ANES 2020 Time Series Study Full Release: User Guide and Codebook](https://electionstudies.org/wp-content/uploads/2022/02/anes_timeseries_2020_userguidecodebook_20220210.pdf)

> [How to Analyze ANES Survey Data](https://electionstudies.org/wp-content/uploads/2018/04/nes012492.pdf)

<br>

One Haiku:

```{r}
# chez sacrificed queen
# quadrennial bloodless coup
# knight churchill's least worst
```

---

## Function Definitions {-}

Define a function to import a stata file as a data.frame:

```{r eval = FALSE , results = "hide" }
library(haven)

anes_import_dta <-
	function( this_fn ){
		
		this_tbl <- read_dta( this_fn )
		
		this_tbl <- zap_labels( this_tbl )
		
		this_df <- data.frame( this_tbl )
		
		names( this_df ) <- tolower( names( this_df ) )
		
		this_df
	}
```

---

## Download, Import, Preparation {-}

1. Register for the ANES Data Center at https://electionstudies.org/

2. Choose `2020 Time Series Study`

3. Download the `STATA` version of the February 10, 2022 file:

```{r eval = FALSE , results = "hide" }
library(haven)

anes_fn <-
	file.path( 
		path.expand( "~" ) , 
		"anes_timeseries_2020_stata_20220210.dta"
	)

anes_df <- anes_import_dta( anes_fn )
```

### Save Locally \ {-}

Save the object at any point:

```{r eval = FALSE , results = "hide" }
# anes_fn <- file.path( path.expand( "~" ) , "ANES" , "this_file.rds" )
# saveRDS( anes_df , file = anes_fn , compress = FALSE )
```

Load the same object:

```{r eval = FALSE , results = "hide" }
# anes_df <- readRDS( anes_fn )
```

### Survey Design Definition {-}
Construct a complex sample survey design:

```{r eval = FALSE , results = "hide" }
library(survey)

anes_design <-
	svydesign(
		ids = ~ v200010c ,
		strata = ~ v200010d ,
		weights = ~ v200010a ,
		data = subset( anes_df , v200010a > 0 ) ,
		nest = TRUE
	)
```

### Variable Recoding {-}

Add new columns to the data set:
```{r eval = FALSE , results = "hide" }
anes_design <- 
	update( 
		anes_design , 
		
		one = 1 ,
		
		democratic_party_rating = ifelse( v201156 %in% 0:100 , v201156 , NA ) ,

		republican_party_rating = ifelse( v201157 %in% 0:100 , v201157 , NA ) ,
		
		primary_voter = ifelse( v201020 %in% 1:2 , as.numeric( v201020 == 1 ) , NA ) ,

		think_gov_spend_least =
			factor( v201645 , levels = 1:4 , labels =
				c( 'foreign aid (correct)' , 'medicare' , 'national defense' , 'social security' )
			) ,
		
		undoc_kids =
			factor( v201423x , levels = 1:6 , labels =
				c( 'should sent back - favor a great deal' ,
					'should sent back - favor a moderate amount' ,
					'should sent back - favor a little' ,
					'should allow to stay - favor a little' ,
					'should allow to stay - favor a moderate amount' ,
					'should allow to stay - favor a great deal' )
			)

	)
```

---

## Analysis Examples with the `survey` library \ {-}

### Unweighted Counts {-}

Count the unweighted number of records in the survey sample, overall and by groups:
```{r eval = FALSE , results = "hide" }
sum( weights( anes_design , "sampling" ) != 0 )

svyby( ~ one , ~ undoc_kids , anes_design , unwtd.count )
```

### Weighted Counts {-}
Count the weighted size of the generalizable population, overall and by groups:
```{r eval = FALSE , results = "hide" }
svytotal( ~ one , anes_design )

svyby( ~ one , ~ undoc_kids , anes_design , svytotal )
```

### Descriptive Statistics {-}

Calculate the mean (average) of a linear variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svymean( ~ republican_party_rating , anes_design , na.rm = TRUE )

svyby( ~ republican_party_rating , ~ undoc_kids , anes_design , svymean , na.rm = TRUE )
```

Calculate the distribution of a categorical variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svymean( ~ think_gov_spend_least , anes_design , na.rm = TRUE )

svyby( ~ think_gov_spend_least , ~ undoc_kids , anes_design , svymean , na.rm = TRUE )
```

Calculate the sum of a linear variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svytotal( ~ republican_party_rating , anes_design , na.rm = TRUE )

svyby( ~ republican_party_rating , ~ undoc_kids , anes_design , svytotal , na.rm = TRUE )
```

Calculate the weighted sum of a categorical variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svytotal( ~ think_gov_spend_least , anes_design , na.rm = TRUE )

svyby( ~ think_gov_spend_least , ~ undoc_kids , anes_design , svytotal , na.rm = TRUE )
```

Calculate the median (50th percentile) of a linear variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
svyquantile( ~ republican_party_rating , anes_design , 0.5 , na.rm = TRUE )

svyby( 
	~ republican_party_rating , 
	~ undoc_kids , 
	anes_design , 
	svyquantile , 
	0.5 ,
	ci = TRUE , na.rm = TRUE
)
```

Estimate a ratio:
```{r eval = FALSE , results = "hide" }
svyratio( 
	numerator = ~ republican_party_rating , 
	denominator = ~ democratic_party_rating , 
	anes_design ,
	na.rm = TRUE
)
```

### Subsetting {-}

Restrict the survey design to party id: independent:
```{r eval = FALSE , results = "hide" }
sub_anes_design <- subset( anes_design , v201231x == 4 )
```
Calculate the mean (average) of this subset:
```{r eval = FALSE , results = "hide" }
svymean( ~ republican_party_rating , sub_anes_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:
```{r eval = FALSE , results = "hide" }
this_result <- svymean( ~ republican_party_rating , anes_design , na.rm = TRUE )

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

grouped_result <-
	svyby( 
		~ republican_party_rating , 
		~ undoc_kids , 
		anes_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:
```{r eval = FALSE , results = "hide" }
degf( anes_design )
```

Calculate the complex sample survey-adjusted variance of any statistic:
```{r eval = FALSE , results = "hide" }
svyvar( ~ republican_party_rating , anes_design , na.rm = TRUE )
```

Include the complex sample design effect in the result for a specific statistic:
```{r eval = FALSE , results = "hide" }
# SRS without replacement
svymean( ~ republican_party_rating , anes_design , na.rm = TRUE , deff = TRUE )

# SRS with replacement
svymean( ~ republican_party_rating , anes_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:
```{r eval = FALSE , results = "hide" }
svyciprop( ~ primary_voter , anes_design ,
	method = "likelihood" , na.rm = TRUE )
```

### Regression Models and Tests of Association {-}

Perform a design-based t-test:
```{r eval = FALSE , results = "hide" }
svyttest( republican_party_rating ~ primary_voter , anes_design )
```

Perform a chi-squared test of association for survey data:
```{r eval = FALSE , results = "hide" }
svychisq( 
	~ primary_voter + think_gov_spend_least , 
	anes_design 
)
```

Perform a survey-weighted generalized linear model:
```{r eval = FALSE , results = "hide" }
glm_result <- 
	svyglm( 
		republican_party_rating ~ primary_voter + think_gov_spend_least , 
		anes_design 
	)

summary( glm_result )
```

---

## Replication Example {-}

This example matches statistics and standard errors in the Age rows of the `ANES respondents (weighted)` column of `Table 1A` from [Benchmark and Attrition Report for the ANES 2016 Time Series Study](https://electionstudies.org/wp-content/uploads/2022/03/DeBell-Maisel-Benchmark-Report-for-the-ANES-2016-Time-Series-Study.pdf#page=4):

1. Log in to the ANES Data Center at https://electionstudies.org/

2. Choose `2016 Time Series Study`.

3. Download the `DTA` version of the September 4, 2019 file

4. Download the `DTA` version of the `Methodology File December 10, 2018`

```{r eval = FALSE , results = "hide" }
anes2016_fn <-
	file.path( 
		path.expand( "~" ) , 
		"anes_timeseries_2016.dta"
	)

anes2016_df <- anes_import_dta( anes2016_fn )

method2016_fn <-
	file.path( 
		path.expand( "~" ) , 
		"anes_timeseries_2016_methodology_dta.dta" 
	)

method2016_df <- anes_import_dta( method2016_fn )

before_nrow <- nrow( anes2016_df )
anes2016_df <- merge( anes2016_df , method2016_df , by = 'v160001' )
stopifnot( nrow( anes2016_df ) == before_nrow )

anes2016_df[ , 'age_categories' ] <- 
	factor(
		findInterval(
			anes2016_df[ , 'v161267' ] , 
			c( 18 , seq( 30 , 70 , 10 ) ) 
		) ,
		levels = 1:6 ,
		labels = c( '18-29' , '30-39' , '40-49' , '50-59' , '60-69' , '70 or older' )
	)

anes2016_design <-
	svrepdesign(
		data = subset( anes2016_df , v160101f > 0 ) ,
		weights = ~ v160101f ,
		repweights = 'weight_ftf_rkwt([0-9]+)' ,
		type = 'JK1' ,
		scale = 32 / 33 
	)

( results <- svymean( ~ age_categories , anes2016_design , na.rm = TRUE ) )

published_results <- c( 0.21 , 0.158 , 0.156 , 0.2 , 0.147 , 0.129 )

published_standard_errors <- c( 0.0091 , 0.009 , 0.0094 , 0.0122 , 0.0069 , 0.0083 )

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

stopifnot( all( round( SE( results ) , 4 ) == published_standard_errors ) )
```

This example matches statistics and standard errors in the Age rows of the `Design-consistent, with published strata` column of `Table 1` from [How to Analyze ANES Survey Data](https://electionstudies.org/wp-content/uploads/2018/04/nes012492.pdf#page=25):

1. Log in to the ANES Data Center at https://electionstudies.org/

2. Choose `2004 Time Series Study`4

3. Download the `DTA` version of the Full Release August 16, 2005 file

4. Choose `2006 Pilot Study`

5. Download the `DTA` version of the April 26, 2007 file

```{r eval = FALSE , results = "hide" }
anes2004_fn <-
	file.path( 
		path.expand( "~" ) , 
		"anes2004TS.dta"
	)

anes2004_df <- anes_import_dta( anes2004_fn )

pilot2006_fn <-
	file.path( 
		path.expand( "~" ) , 
		"anes2006pilot.dta" 
	)

pilot2006_df <- anes_import_dta( pilot2006_fn )

before_nrow <- nrow( pilot2006_df )
pilot2006_df <- merge( pilot2006_df , anes2004_df , by.x = 'v06p001' , by.y = 'v040001' )
stopifnot( nrow( pilot2006_df ) == before_nrow )

	
pilot2006_df[ , 'age_categories' ] <- 
	factor(
		findInterval(
			pilot2006_df[ , 'v043250' ] , 
			c( 18 , seq( 30 , 70 , 10 ) ) 
		) ,
		levels = 1:6 ,
		labels = c( '18-29' , '30-39' , '40-49' , '50-59' , '60-69' , '70 or older' )
	)

pilot2006_design <-
	svydesign( 
		id = ~v06p007b , 
		strata = ~v06p007a , 
		data = pilot2006_df , 
		weights = ~v06p002 , 
		nest = TRUE 
	)

( results <- svymean( ~ age_categories , pilot2006_design , na.rm = TRUE ) )

published_results <- c( 0.207 , 0.162 , 0.218 , 0.175 , 0.111 , 0.126 )

published_standard_errors <- c( 0.0251 , 0.024 , 0.022 , 0.0149 , 0.0125 , 0.0287 )

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

stopifnot( all( round( SE( results ) , 4 ) == published_standard_errors ) )
```

---

## Analysis Examples with `srvyr` \ {-}

The R `srvyr` library calculates summary statistics from survey data, such as the mean, total or quantile using [dplyr](https://github.com/tidyverse/dplyr/)-like syntax. [srvyr](https://github.com/gergness/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](https://cran.r-project.org/web/packages/srvyr/vignettes/srvyr-vs-survey.html) details the available features. As a starting point for ANES users, this code replicates previously-presented examples:

```{r eval = FALSE , results = "hide" }
library(srvyr)
anes_srvyr_design <- as_survey( anes_design )
```
Calculate the mean (average) of a linear variable, overall and by groups:
```{r eval = FALSE , results = "hide" }
anes_srvyr_design %>%
	summarize( mean = survey_mean( republican_party_rating , na.rm = TRUE ) )

anes_srvyr_design %>%
	group_by( undoc_kids ) %>%
	summarize( mean = survey_mean( republican_party_rating , na.rm = TRUE ) )
```