Abstract

The Integrated Public Use Microdata Series (IPUMS) Complete Count Data include more than 650 million individual-level and 7.5 million household-level records. The microdata are the result of collaboration between IPUMS and the nation’s two largest genealogical organizations—Ancestry.com and FamilySearch—and provides the largest and richest source of individual level and household data.

Before Manuscript Submission

All manuscripts (and other items you'd like to publish) must be submitted to

phsdatacore@stanford.edu for approval prior to journal submission.

We will check your cell sizes and citations.

For more information about how to cite PHS and PHS datasets, please visit:

https:/phsdocs.developerhub.io/need-help/citing-phs-data-core

Documentation

Historic data are scarce and often only exists in aggregate tables. The key advantage of historic US census data is the availability of individual and household level characteristics that researchers can tabulate in ways that benefits their specific research questions. The data contain demographic variables, economic variables, migration variables and family variables. Within households, it is possible to create relational data as all relations between household members are known. For example, having data on the mother and her children in a household enables researchers to calculate the mother’s age at birth. Another advantage of the Complete Count data is the possibility to follow individuals over time using a historical identifier.

In sum: the historic US census data are a unique source for research on social and economic change and can provide population health researchers with information about social and economic determinants.

The historic US 1920 census data was collected in January 1920. Enumerators collected data traveling to households and counting the residents who regularly slept at the household. Individuals lacking permanent housing were counted as residents of the place where they were when the data was collected. Household members absent on the day of data collected were either listed to the household with the help of other household members or were scheduled for the last census subdivision.

Notes

• We provide household and person data separately so that it is convenient to explore the descriptive statistics on each level. In order to obtain a full dataset, merge the household and person on the variables SERIAL and SERIALP. In order to create a longitudinal dataset, merge datasets on the variable HISTID.

• Households with more than 60 people in the original data were broken up for processing purposes. Every person in the large households are considered to be in their own household. The original large households can be identified using the variable SPLIT, reconstructed using the variable SPLITHID, and the original count is found in the variable SPLITNUM.

• Coded variables derived from string variables are still in progress. These variables include: occupation and industry.

• Missing observations have been allocated and some inconsistencies have been edited for the following variables: SPEAKENG, YRIMMIG, CITIZEN, AGE, BPL, MBPL, FBPL, LIT, SCHOOL, OWNERSHP, MORTGAGE, FARM, CLASSWKR, OCC1950, IND1950, MARST, RACE, SEX, RELATE, MTONGUE. The flag variables indicating an allocated observation for the associated variables can be included in your extract by clicking the ‘Select data quality flags’ box on the extract summary page.

• Most inconsistent information was not edited for this release, thus there are observations outside of the universe for some variables. In particular, the variables GQ, and GQTYPE have known inconsistencies and will be improved with the next release.

%3C!-- --%3E

Section 2

This dataset was created on 2020-01-10 18:46:34.647 by merging multiple datasets together. The source datasets for this version were:

IPUMS 1920 households: This dataset includes all households from the 1920 US census.

IPUMS 1920 persons: This dataset includes all individuals from the 1920 US census.

IPUMS 1920 Lookup: This dataset includes variable names, variable labels, variable values, and corresponding variable value labels for the IPUMS 1920 datasets.

This dataset includes all individuals from the 1920 US census.

The data sets in this repository allow users to link people among the U.S. decennial censuses, using the "histid" identifier. The census data sets users will need are indexed by Ancestry.com and are hosted by IPUMS at https://usa.ipums.org/usa-action/samples. Users will need to download the full-count census for each year and be sure to select the "histid" variable that is available under the Person/Historical Technical drop-down menu.As of 7/12/21, links are available between the 1900-1910, 1910-1920, and 1900-1920 censuses.A detailed account of how these links are created and a description of the data and its characteristics are available in the following article:Price, J., Buckles, K., Van Leeuwen, J., & Riley, I. (2021). Combining family history and machine learning to link historical records: The Census Tree data set. Explorations in Economic History, 80, 101391.https://www.sciencedirect.com/science/article/pii/S0014498321000024

Description of sources of names and ancestry data.

This dataset includes all households from the 1920 US census.

Distribution of surname-ancestry in the reference data.

Surname ancestry estimated according to the last or unique surname.

This dataset includes variable names, variable labels, variable values, and corresponding variable value labels for the IPUMS 1920 datasets.

Confusion matrix—observed and predicted values obtained using the Cavnar and Trenkle procedure for the classification of surnames in the test set.

Data Sources- Scanned copies of the U.S. Census for various years (including 1920 and 1930) available from Ancestry Library Edition database.- Sanborn shapefiles were created by Bednar student interns at Penn State's Pattee Library. They are based on the collection of PA Sanborns housed at the same library.

Hourly wages in Brazil (2013)- Analysis of variances.