Script to aggregate PUMS responses to reproduce and extend table characterizing the demographics of mobile home residents found in Esther Sullivan, Manufactured Insecurity Mobile Home Parks and Americans’ Tenuous Right to Place. (Berkeley, CA: University of California Press, 2018). Made for David Tisel for a forthcoming article in the _Review of…

pums_mobile_homes.R

library(tidycensus)
library(dplyr)
library(tidyr)
library(gt)

# vars <- tidycensus::pums_variables %>% 
#   dplyr::filter(year == 2017, survey == "acs5")
# 
# vars_housing <- vars %>%
#   dplyr::filter(level == "housing")
# 
# vars_people <- vars %>%
#   dplyr::filter(level == "person")

fetch_data <- function(year, state){
  tidycensus::get_pums(
    variables = base::c(
      "BLD", # Units in structure (this is how we get mobile homes.)
      "HISP", # Recoded detailed Hispanic origin
      "RAC1P", # Recoded detailed race code
      "AGEP", # Age
      "HHT", # Household/family type ("Living Alone")
      "CIT", # Citizenship status
      "POVPIP" # Income-to-poverty ratio recode
    ),
    state = state,
    survey = "acs5",
    year = year
  )
}

process_data <- function(df, state, people_method = "head_household") {
  # Household vars.
  df %>%
    # This filters for heads of households.
    # TODO: Extend to aggregate individuals in households for demographics.
    dplyr::filter(SPORDER == "1") %>%
    dplyr::mutate(
      mobile = (BLD == "1"),
      alone = HHT %in% c("4", "6"),
      POVPIP = base::as.numeric(POVPIP),
      # Below 100% of the poverty line.
      pov = dplyr::between(base::as.numeric(POVPIP), 0, 100),
      # Over 55.
      over55 = (base::as.numeric(AGEP) >= 55),
      # Citizenship status.
      noncit = (CIT == "5"),
      # Hispanic/Latino
      hisp = (HISP != "1"),
      # Non-Hispanic white
      white = (RAC1P == "1") & !hisp,
      # Non-Hispanic Black
      black = (RAC1P == "2") & !hisp
    ) %>%
    summarize(
      # Total mobile home households.
      `Mobile Home Households` = sum(WGTP[mobile]),
      # Total households.
      hh_all = sum(WGTP),
      # Total population in mobile homes...
      ppl_mobile = sum(PWGTP[mobile]),
      # ...who are living alone.
      ppl_mobile_alone = sum(PWGTP[mobile & alone]),
      `Living alone (%)` = (ppl_mobile_alone / ppl_mobile),
      # ...who are below the poverty line.
      ppl_mobile_pov = sum(PWGTP[mobile & pov]),
      `Poverty (%)` = (ppl_mobile_pov / ppl_mobile),
      # ...who are over 55.
      ppl_mobile_over55 = sum(PWGTP[mobile & over55]),
      `Age 55 or Older (%)`  = (ppl_mobile_over55 / ppl_mobile),
      # ...who are noncitizens
      ppl_mobile_noncit = sum(PWGTP[mobile & noncit]),
      `Non-Citizen (%)` = (ppl_mobile_noncit / ppl_mobile),
      # ...who identify as Non-Hispanic white.
      ppl_mobile_white = sum(PWGTP[mobile & white]),
      `Non-Hispanic White (%)` = (ppl_mobile_white / ppl_mobile),
      # ...who identify as Non-Hispanic Black.
      ppl_mobile_black = sum(PWGTP[mobile & black]),
      `Non-Hispanic Black (%)` = (ppl_mobile_black / ppl_mobile),
      # ...who identify as Hispanic or Latino.
      ppl_mobile_hisp = sum(PWGTP[mobile & hisp]),
      `Hispanic (%)` = (ppl_mobile_hisp / ppl_mobile),
      state = state
    ) %>%
    select(-c(starts_with(c('ppl_', 'hh_'))))
}

va <- fetch_data(state = "VA", year = 2011)
fl <- fetch_data(state = "FL", year = 2011)
tx <- fetch_data(state = "TX", year = 2011)
# national <- fetch_data(state = "all", year = 2011)

va_processed <- process_data(va, state = "Virginia")
fl_processed <- process_data(fl, state = "Florida")
tx_processed <- process_data(tx, state = "Texas")
#national_processed <- process_data(national, state = "National")

results <- bind_rows(va_processed, tx_processed, fl_processed) %>%
  tidyr::pivot_longer(
    -c(state), 
    names_to = 'var', 
    values_to = 'val'
  ) %>%
  tidyr::pivot_wider(
    id_cols = var, 
    names_from = state, 
    values_from = val
  )

gt(results, rowname_col = "var") %>%
  fmt_integer(
    rows = var %in% ("Mobile Home Households")
  ) %>%
  fmt_percent(
    rows = ends_with("(%)"),
    decimals = 1
  ) %>%
  tab_row_group(
    label = "Race and Ethnicity",
    rows = c(
      "Non-Hispanic White (%)", 
      "Non-Hispanic Black (%)",  
      "Hispanic (%)"
    )
  ) %>%
  tab_header(
    title = "Characteristics of Mobile Home Residents in Virginia, Texas, and Florida"
  ) %>%
  tab_source_note(
    source_note = "Source: American Community Survey Public-Use Microdata 2007-2011 five-year sample. Made following Esther Sullivan, Manufactured Insecurity Mobile Home Parks and Americans’ Tenuous Right to Place. (Berkeley, CA: University of California Press, 2018)."
  ) %>%
  tab_footnote(
    footnote = "Demographic characteristics are based on heads of households, not all residents of households.",
    locations = cells_stub(rows = 4:8)
  )