wcensoc:

Summary: Run the codes in this subdirectory sequentially to create the (1) CenSoc-DMF and (2) CenSoc-Numident datasets. The objecrtive of each code is described below:

01_run_split_by_bpl.R

Time: ~ 1 hour
Goal:
- Split Census & Numident files by birth state
Combine three original Social Security Death Master files into one DMF file.
Notes
- The 1940 Census collected ages at the time of the survey (i.e., 1940 April) and did not ask for exact birth year and date. Therefore, we make a new variable (census_age) using birthdate records in Numident and DMF files (i.e., 1940 minus birth year for those whose birth month is January through March; otherwise, 1939 minus birth year), based on which we match Census data to Numident and DMF files. For DMF, this variable is automatically generated when applying the function load_dmf_deaths.

2_run_parse_titles_and_nicknames.R

Time: ~ 1 minute
Goal:
- Make dictionaries for cleaning errors in first names and converting nicknames to formal first names.

3_run_clean_names.R

Time: ~12 hours
Goal: Clean first names in Census, Numident, and DMF files.
Notes
- This code is to apply the function clean_names to Census, Numident, and DMF files.
- It takes the longest time among the five application codes since it deals with large string variables.

4_run_match_records.R

Time: ~ 6 hours
Goal:
- Perform ABE matching using split and cleaned Census and Numident (or DMF) files.
Note
- This code and the source code match_records are the most complicated ones among the ABE matching application codes.
- match_records consists of four functions
- preprocess_A / preprocess_B: keeping necessary variables and generating new variables for matching
- append_A_to_B: counting rows with the same first & last name and age within each dataset / finding exact matches (e.g., each dataset has only one person with the same first & last name and age) / making variables (e.g., uniquestub_file1, uniquestub_file2) for conservative matches
- find_nonexact_matches: finding non-exact matches (e.g., no person matched by first & last name with the same age, but some matched within ±1 or ±2 years) / making variables (e.g., uniquestub_match1, uniquestub_match2) for conservative matches
- fix_ids: storing Census IDs of matched observations to the corresponding Numident (or DMF) data (or vice versa)
- match_records matches two datasets twice forward (i.e., Census to Numident/DMF) and backward (i.e., Numident/DMF to Census), and takes only those identified as matched observations in both directions
- When matching Census to Numident, (i) males, (ii) ever-married females, and (iii) never-married females (using father's last name in Numident) are matched in turn.
- Conservative matches are defined by two variables, uniquestub_fileN (i.e., first & last names are unique within ±N age within the dataset) and uniquestub_matchN (i.e., matched to one person within ±N age in the corresponding dataset). For example, you can identify conservative matches within ±N years by taking those with uniquestub_file2==1 and unique_stubmatch==1.