Please select the starting and final year of your desired linked dataset. You will then be able to download a ZIP file containing the crosswalk data, associated codes, and documentation.

Starting year:

Final year:

Download crosswalk and documentation:

Abramitzky, Ran; Boustan, Leah; Eriksson, Katherine; Rashid, Myera; Pérez, Santiago, 2022, "Census Linking Project: 1850-1860 Crosswalk",, Harvard Dataverse, V2.
Joseph Price, Kasey Buckles, and Mark Clement. Census Tree Data: Version 1.0 [dataset]. 2021.
| Citation copied!
author = {Abramitzky, Ran and Boustan, Leah and Eriksson,
Katherine and Rashid, Myera and Pérez, Santiago},
publisher = {Harvard Dataverse},
title = {{Census Linking Project: 1850-1860 Crosswalk}} ,
year = {2022},
version = {V2},
doi = {10.7910/DVN/KO5J44},
url = {}
author = {Price, Joseph and Buckles, Casey and Clement Mark},
title = {{Census Tree Data}} ,
year = {2021},
version = {1.0},
url = {}

The data found above contains the crosswalks for each Census wave linked to every other (1850-1940). There are 36 crosswalks in total, representing every combination of Census years1.

The crosswalks also give users the option to select the linking method with which matches were created. Users can then merge into these crosswalks a wide set of individual- and household-level variables provided publicly by IPUMS, thereby creating a historical longitudinal dataset for analysis.

In the above menu, “starting year” refers to one particular wave of the Census and “final year” refers to another wave of the Census.

For instance, if you choose 1900 as your starting year and 1940 as your final year, then the crosswalk will provide you with histids of individuals in 1900 matched to histids of individuals in 1940. You will then be able to download the data (in .dta and .csv formats), codes used to create the crosswalks, and further codes that will allow you to merge in information from IPUMS and to reweight the data. Extensive documentation is provided and should be consulted before using the data. The linking methods have varying accuracy based on race, birth place and time period. We encourage checking the links based on the context of your project, utilizing multiple methods, and using your best judgement to determine whether a particular method is right for your context.

1 Note: The microdata for the 1890 Census is no longer extant.