ahs.Rmd

# American Housing Survey (AHS) {-}

[![License: GPL v3](https://img.shields.io/badge/License-GPLv3-blue.svg)](https://www.gnu.org/licenses/gpl-3.0) <a href="https://github.com/asdfree/ahs/actions"><img src="https://github.com/asdfree/ahs/actions/workflows/r.yml/badge.svg" alt="Github Actions Badge"></a> 

The nationwide assessment of housing stock, with information on physical condition and neighborhood, costs of financing and maintenance, owner and renter characteristics, and changes over time.

* Nationally-representative and metropolitan flat files with one row per household, plus relational files.

* A complex sample survey of occupied and vacant housing units designed to generalize to all structures in the United States, both nationally and also for about [thirty-five metropolitan areas](https://www.census.gov/content/dam/Census/programs-surveys/ahs/tech-documentation/AHS%20Metropolitan%20Area%20Oversample%20Histories%202015%20and%20Beyond.pdf).

* Released more or less biennially since 1973, with longitudinal samples redrawn in 1985 and 2015.

* Sponsored by [the Department of Housing and Urban Development](https://www.hud.gov/), run by the [Census Bureau](https://www.census.gov/).

---

## Recommended Reading {-}

Four Example Strengths & Limitations:

✔️ [National, state, and metro area geographies](https://www.census.gov/programs-surveys/ahs/research/publications/State_Level_Design_Based_Estimates_for_National_Surveys.html)

✔️ [Housing unit-focused questionnaire provides greater detail on housing stock](https://www.census.gov/topics/housing/guidance/topics.html)

❌ [Housing unit-focused questionnaire asks fewer detailed questions of occupants on some topics](https://www2.census.gov/programs-surveys/ahs/2023/2023__AHS_Items_Booklet.pdf)

❌ [Underreported estimate of adjustable rate mortgages](https://www.census.gov/content/dam/Census/programs-surveys/ahs/publications/Carter_VII-B.pdf)

<br>

Three Example Findings:

1. [In 2017, 21% of residences nationwide did not have adequate space for COVID-19 isolation](https://www.acpjournals.org/doi/10.7326/M20-4331).

2. [From 1991 to 2017, single men earned 1.5% higher housing investment returns vs. women](http://www.nber.org/papers/w26914).

3. [More than a quarter of a million households receiving HUD rental assistance lacked accessibility features but had a member using a mobility device (like a wheelchair or walker) in 2019](https://www.gao.gov/products/gao-23-106339).

<br>

Two Methodology Documents:

> [Getting Started with the Public Use File: 2015 to 2023](https://www.census.gov/content/dam/Census/programs-surveys/ahs/tech-documentation/2015/Getting%20Started%20with%20the%20AHS%20PUF%202015%20and%20Beyond.pdf)

> [Wikipedia Entry](https://en.wikipedia.org/wiki/American_Housing_Survey)

<br>

One Haiku:

```{r}
# real estate supply
# half bath addition, raised roof
# vent, rent too damn high
```

---

## Download, Import, Preparation {-}

Download and import the national 2023 flat file:

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

tf <- tempfile()

this_url <-
	paste0(
		"https://www2.census.gov/programs-surveys/ahs/" ,
		"2023/AHS%202023%20National%20PUF%20v1.0%20Flat%20SAS.zip"
	)

GET( this_url , write_disk( tf ) , progress() )

ahs_tbl <- read_sas( tf )

ahs_df <- data.frame( ahs_tbl )

names( ahs_df ) <- tolower( names( ahs_df ) )
```

### Save Locally \ {-}

Save the object at any point:

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

Load the same object:

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

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

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

ahs_design <- 
	svrepdesign(
		weights = ~ weight ,
		repweights = "repweight[1-9]" ,
		type = "Fay" ,
		rho = ( 1 - 1 / sqrt( 4 ) ) ,
		mse = TRUE ,
		data = ahs_df
	)
```

### Variable Recoding {-}

Add new columns to the data set:
```{r eval = FALSE , results = "hide" }
ahs_design <- 
	update( 
		ahs_design , 

		one = 1 ,

		tenure = 
			factor( 
				ifelse( tenure %in% c( -6 , 'N' ) , 4 , tenure ) , 
				levels = 1:4 , 
				labels = 
					c( 'Owned or being bought' ,
					'Rented for cash rent' ,
					'Occupied without payment of cash rent' ,
					'Not occupied' )
			) ,
			
		lotsize =
			factor( 
				lotsize , 
				levels = 1:7 ,
				labels = c( "Less then 1/8 acre" , 
				"1/8 up to 1/4 acre" , "1/4 up to 1/2 acre" ,
				"1/2 up to 1 acre" , "1 up to 5 acres" , 
				"5 up to 10 acres" , "10 acres or more" ) ) ,
				
		below_poverty = as.numeric( perpovlvl < 100 )
				
	)
```

---

## 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( ahs_design , "sampling" ) != 0 )

svyby( ~ one , ~ tenure , ahs_design , unwtd.count )
```

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

svyby( ~ one , ~ tenure , ahs_design , svytotal )
```

### Descriptive Statistics {-}

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

svyby( ~ totrooms , ~ tenure , ahs_design , svymean , na.rm = TRUE )
```

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

svyby( ~ lotsize , ~ tenure , ahs_design , svymean , na.rm = TRUE )
```

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

svyby( ~ totrooms , ~ tenure , ahs_design , svytotal , na.rm = TRUE )
```

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

svyby( ~ lotsize , ~ tenure , ahs_design , svytotal , na.rm = TRUE )
```

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

svyby( 
	~ totrooms , 
	~ tenure , 
	ahs_design , 
	svyquantile , 
	0.5 ,
	ci = TRUE , na.rm = TRUE
)
```

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

### Subsetting {-}

Restrict the survey design to homes with a garage or carport:
```{r eval = FALSE , results = "hide" }
sub_ahs_design <- subset( ahs_design , garage == 1 )
```
Calculate the mean (average) of this subset:
```{r eval = FALSE , results = "hide" }
svymean( ~ totrooms , sub_ahs_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( ~ totrooms , ahs_design , na.rm = TRUE )

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

grouped_result <-
	svyby( 
		~ totrooms , 
		~ tenure , 
		ahs_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( ahs_design )
```

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

# SRS with replacement
svymean( ~ totrooms , ahs_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( ~ below_poverty , ahs_design ,
	method = "likelihood" , na.rm = TRUE )
```

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

Perform a design-based t-test:
```{r eval = FALSE , results = "hide" }
svyttest( totrooms ~ below_poverty , ahs_design )
```

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

Perform a survey-weighted generalized linear model:
```{r eval = FALSE , results = "hide" }
glm_result <- 
	svyglm( 
		totrooms ~ below_poverty + lotsize , 
		ahs_design 
	)

summary( glm_result )
```

---

## Replication Example {-}

This example matches the estimate and margin of error of the `Total` row of the `General Housing` tab from the [AHS 2023 Table Specifications and PUF Estimates for User Verification](https://www2.census.gov/programs-surveys/ahs/2023/AHS%202023%20Table%20Specifications%20and%20PUF%20Estimates%20for%20User%20Verification.xlsx):

```{r eval = FALSE , results = "hide" }
result <- svytotal( ~ as.numeric( intstatus == 1 ) , ahs_design )

stopifnot( round( coef( result ) / 1000 , 0 ) == 133231 )

ci_results <- confint( result , level = 0.9 )

stopifnot( round( ( ci_results[ 2 ] - coef( result ) ) / 1000 , 0 ) == 381 )
```

---

## 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 AHS users, this code replicates previously-presented examples:

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

ahs_srvyr_design %>%
	group_by( tenure ) %>%
	summarize( mean = survey_mean( totrooms , na.rm = TRUE ) )
```