In caseybreen/wcensoc:

[^updated]: Last updated: r format(Sys.time(), '%d %B, %Y')

| Page| Variable | Label | |----:|:------------------------|:---------------------------------------------| | \hyperlink{page.2}{2} | \hyperlink{page.2}{HISTID} |Historical Unique Identifier | | \hyperlink{page.3}{3} | \hyperlink{page.3}{byear} |Year of Birth | | \hyperlink{page.4}{4} | \hyperlink{page.4}{bmonth} |Month of Birth | | \hyperlink{page.5}{5} | \hyperlink{page.5}{dyear} |Year of Death | | \hyperlink{page.6}{6} | \hyperlink{page.6}{dmonth} |Month of Death | | \hyperlink{page.7}{7} | \hyperlink{page.7}{death_age} |Age at Death (Years) | | \hyperlink{page.8}{8} | \hyperlink{page.8}{weight} |CenSoc Sample Weight | | \hyperlink{page.9}{9} | \hyperlink{page.9}{Additional IPUMS variables}| Additional 1940 Census variables, including: pernum, perwt, age, sex, bpld, mbpl, fbpl, educd, educ_yrs, empstatd, hispan, incwage, incnonwg, marst, nativity, occ, occscore, ownershp, race, rent, serial, statefip, and urban.|

\vspace{210pt}

Summary: The CenSoc-DMF Version 3.0 Demo dataset (N = 42,421) links the IPUMS 1940 1% census sample to the Death Master File (DMF) dataset, a collection of death records reported to the Social Security Administration. Records were linked using a conservative variant of the ABE method developed by Abramitzky, Boustan, and Eriksson (\textcolor{blue}{2012}, \textcolor{blue}{2014}, \textcolor{blue}{2017}).

We note that this demo dataset is not conducive to high-resolution mortality research. We recommend using this file for exploratory and demonstrative purposes. To best conduct research with CenSoc data, researchers may download the full CenSoc-DMF from the CenSoc website, obtain an extract of the full-count 1940 Census from IPUMS-USA, and merge data using on the individual-level, unique identifier HISTID variable. Please adhere to CenSoc and IPUMS citation guidelines when using this file.

\newpage

\huge HISTID \normalsize \vspace{12pt}

Label: Historical Unique Identifier

Description: HISTID is a unique individual-level identifier. It can be used to merge the CenSoc-DMF file with the 1940 Full-Count Census from IPUMS.

\newpage

\huge byear \normalsize \vspace{12pt}

Label: Birth Year

Description: byear reports a person's year of birth, as recorded in the Social Security Death Master File.

## Library Packages
library(tidyverse)
library(data.table)

## read in censoc_dmf_v2.1 data file

censoc_dmf_v3 <-read_csv("/data/censoc/censoc_data_releases/censoc_dmf_demo/censoc_dmf_demo_v3/censoc_dmf_demo_v3.csv")

byear_plot <- censoc_dmf_v3 %>%
    group_by(byear) %>%
    summarise(n = n()) %>%
    ggplot(aes(x = byear, y = n)) + 
    geom_line() + 
    geom_point() +
    theme_minimal(base_size = 15) + #replace with a different theme (theme_bw()) if the bbplot package isn't downloaded 
  ggtitle("Year of Birth") + 
  theme(legend.position="bottom") +
  xlab("title") + 
  labs(x = "Year", 
       y = "Count") + 
  scale_y_continuous(labels = scales::comma) + 
  scale_x_continuous(breaks = scales::pretty_breaks(n=5)) 

\vspace{75pt}

byear_plot

\newpage \huge bmonth \normalsize \vspace{12pt}

Label: Birth Month

Description: bmonth reports a person's month of birth, as recorded in the Social Security Death Master File.

## run in the console and copy and paste into documentation
bmonth_tabulated <- knitr::kable(censoc_dmf_v3 %>%
    filter(bmonth != 0) %>% 
    group_by(bmonth) %>%
    tally() %>%
    mutate(freq = signif(n*100 / sum(n), 2)) %>%
    mutate(label = c("January", "February", "March", "April", "May", "June", "July", "August", "September", "October", "November", "December")) %>%
    select(bmonth, label, n, `freq %` = freq))

\vspace{30pt}

bmonth_tabulated

\newpage

\huge dyear \normalsize \vspace{12pt}

Label: Death Year

Description: dyear reports a person's year of death, as recorded in the Social Security Death Master File.

dyear_plot <-
  censoc_dmf_v3 %>%
    group_by(dyear) %>%
    summarise(n = n()) %>%
    ggplot(aes(x = dyear, y = n)) + 
    geom_line() + 
    geom_point() +
    theme_minimal(base_size = 15) +  
  ggtitle("Year of Death") + 
  theme(legend.position="bottom") +
  xlab("title") + 
  labs(x = "Year", 
       y = "Count") + 
  scale_y_continuous(labels = scales::comma, limits = c(0, 2000)) + 
  scale_x_continuous(breaks = scales::pretty_breaks(n=5)) 

\vspace{75pt}

dyear_plot

\newpage

\huge dmonth \normalsize \vspace{12pt}

Label: Death Month

Description: dmonth reports a person's month of death, as recorded in the Social Security Death Master File.

dmonth_tabulated <- knitr::kable(censoc_dmf_v3 %>%
    group_by(dmonth) %>%
    tally() %>%
    mutate(freq = signif(n*100 / sum(n), 2)) %>%
    mutate(label = c("January", "February", "March", "April", "May", "June", "July", "August", "September", "October", "November", "December")) %>%
    select(dmonth, label, n, `freq %` = freq))

\vspace{30pt}

dmonth_tabulated

\newpage

\huge death_age \normalsize \vspace{12pt}

Label: Age at Death (Years)

Description: death_age reports a person's age at death in years, calculated using the birth and death information recorded in the Social Security Death Master File.

death_age_plot <- censoc_dmf_v3 %>%
    group_by(death_age) %>%
    summarise(n = n()) %>%
    ggplot(aes(x = death_age, y = n)) + 
    geom_line() + 
    geom_point() +
    theme_minimal(base_size = 15) + #replace with a different theme (theme_bw()) if the bbplot package isn't downloaded 
  ggtitle("Age at Death") + 
  theme(legend.position="bottom") +
  xlab("title") + 
  labs(x = "Age at Death", 
       y = "Count") + 
  scale_y_continuous(labels = scales::comma) + 
  scale_x_continuous(breaks = scales::pretty_breaks(n=5)) 

\vspace{75pt}

death_age_plot

\newpage

\huge weight \normalsize

\vspace{12pt}

Label: CenSoc Sample Weight[^1]

Description: weight is a post-stratification person-weight to National Center for Health Statistics (NCHS) totals for persons (1) dying between 1975-2005 (2) dying between ages 65-100. Weights are based on age at death, year of death, sex, and race, and place of birth. Please see the \textcolor{blue}{technical documentation on weights} for more information.

[^1]: The IPUMS-USA 1940 1% sample also includes a weight (perweight) to account for the 1940 sampling procedure (thus no weights for the 100% complete count 1940 census). For analysis, we recommend using both sets of weights. A final weight can be constructed by multiplying the two weights together.

weights_tabulated <- censoc_dmf_v3 %>%
  filter(!is.na(weight)) %>% 
  summarize('Min Weight' = round(min(weight),2), 'Max Weight' = round(max(weight), 2)) %>%
  mutate(id = 1:n()) %>% 
  pivot_longer(-id, names_to = "Label", values_to = "Value") %>% 
  select(Value, Label) %>% 
  add_row(Label = "No Weight Assigned", Value = NA) %>% 
  knitr::kable()

# weight.plot <- censoc_dmf_v3 %>%
#   filter(!is.na(weight)) %>% 
#   ggplot() +
#     geom_boxplot(aes(x=weight), fill='grey92') +
#     ylim(-1,1) +
#     theme_minimal(15) +
#     theme(axis.title.y=element_blank(), axis.text.y=element_blank(), axis.ticks.y=element_blank()) +
#     ggtitle("Distribution of Weights") +
#     xlab('Weight')

weights_dmf <- censoc_dmf_v3 %>% 
  filter(!is.na(weight)) %>% 
  group_by(death_age, dyear) %>% 
  summarize(weight = mean(weight))

weights_lexis_dmf <- weights_dmf %>% 
  ggplot() +
  geom_raster(aes(x = dyear, y = death_age,
                  fill = weight)) +
  ## Lexis grid
  geom_hline(yintercept = seq(65, 100, 10),
             alpha = 0.6, lty = "dotted") +
  geom_vline(xintercept = seq(1985, 2005, 10),
             alpha = 0.6, lty = "dotted") +
  geom_abline(intercept = seq(-100, 100, 10)-1910,
              alpha = 0.6, lty = "dotted") +
  scale_fill_viridis_c(option = "magma") +
  scale_x_continuous("Year", expand = c(0.02, 0),
                     breaks = seq(1975, 2005, 5)) +
  scale_y_continuous("Age", expand = c(0, 0),
                     breaks = seq(65, 100, 10)) +
  guides(fill = guide_legend(reverse = TRUE)) +
  # coord
  coord_equal() +
  # theme
  theme_void() +
  theme(
    axis.text = element_text(colour = "black"),
    axis.text.y = element_text(size = 10),
    axis.text.x = element_text(size = 10, angle = 45, hjust = 0.5), 
    plot.title = element_text(size = 10, vjust = 2),
    legend.text = element_text(size = 10), 
    axis.title=element_text(size = 10,face="bold")
  ) + 
  labs(X = "Year",
       Y = "Age",
       title = "Average weight by age at death and year of death") 

\vspace{30pt}

weights_tabulated

\vspace{30pt}

weights_lexis_dmf

\newpage

\huge IPUMS 1940 Census Variables \normalsize

\vspace{12pt}

The variables below are from the IPUMS-USA 1940 1% census sample. We recommend looking at the terrific documentation on the IPUMS-USA website: \textcolor{blue}{https://usa.ipums.org/usa/index.shtml}

| Variable | Label | |:-------------|:---------------------------------------------| | pernum |Person number in household| | perwt |IPUMS person weight[^2] | | age |Age on April 1st, 1940| | sex |Sex | | bpld |Place of birth (detailed)| | mbpl |Mother's place of birth[^3]| | fbpl |Father's place of birth[^4]| | educd |Educational attainment (detailed IPUMS codes)| | educ_yrs |Educational attainment in years (constructed)[^5]| | empstatd |Employment status (detailed)| | hispan |Hispanic/Spanish/Latino origin (imputed)[^6]| | incwage | Wage and salary income in 1939 | | incnonwg |Had non-wage/salary income over $50 in 1939| | marst |Marital status| | nativity |Foreign birthplace or parentage| | occ |Occupation| | occscore |Occupational income score| | ownershp |Ownership of dwelling (tenure)| | race |Race| | rent |Montly contract rent| | serial |Household serial number| | statefip |State of residence 1940 (FIPS code)| | urban |Urban/rural status|

[^2]: The IPUMS perweight accounts for the 1940 sampling procedure to construct the 1% sample, and thus is only available in the 1940 1% sample. For analysis, we recommend using both the IPUMS perweight and the CenSoc weight. A final weight can be constructed by multiplying the two weights together [^3]: This variable is only available for sample-line persons (a one-in-twenty sample asked additional questions in the 1940 Census) or those living with their mother. [^4]: This variable is only available for sample-line persons (a one-in-twenty sample asked additional questions in the 1940 Census) or those living with their father. [^5]: educ_yrs is constructed from the IPUMS educd variable but not directly available from IPUMS.
[^6]: The 1940 Census did not directly inquire about Hispanic ethnicity or origin. This variable is determined by IPUMS using information such as one's birthplace or a parent's birthplace.

caseybreen/wcensoc documentation built on Nov. 21, 2024, 5:15 a.m.