Japanese American Postwar Resettlement


Imagine being able to construct an interactive map of the migrations of more than 100,000 people.

Map of United States with locations of Japanese American incarceration camps

A model explaining how family relationships, education, and occupation impacted migration opportunities and choices.

My new project, ”A Community in Motion” will do this, mapping the movement of all former Japanese American inmates.

Some of the inmates—especially young, single, and highly educated—were permitted to leave already in 1943 and 1944, as long as they stayed away from the “excluded” West Coast states. The majority of the population, however, couldn’t or didn’t want to move.

In January 1945, the US government permitted the return to the West Coast and started pushing inmates out of the incarceration camps, and this is where my project begins. I will not only map the first destinations out of the camps but all the way up to 1950—the latest publicly available census year.

The map above only tracks some of the most common routes of Heart Mountain inmates but I aim to do the same for all of the 10 camps operated by the War Relocation Authority, as seen in the opening map of this page.

The US census records include such information as address, nationality, education, and occupation for each individual. The census will be queried through variables focusing on race and birthplace (either birthplace of the individual in question or parent), resulting in a comprehensive list of Japanese Americans in the United States.

The ability to do this with these two search terms only is due to two characteristics of the population: a) the limited period of Japanese immigration to the United States resulting in very few people whose parent(s) were not Japanese, and b) the miscegenation laws preventing the intermarriage of Japanese with Whites, resulting in very little racial mixing.

As is the case with all methods, the use of historical data and network analysis do not offer a shortcut to rigorous interpretation of the results, nor do they produce automatic answers to questions. I am aware that census records do not capture all individuals living in a given place—or sometimes a person might accidentally respond twice in different places. Similarly, spelling conventions sometimes make two people out of one, or report incorrect birth dates. Nevertheless, running tests on several datasets have proven that the margin of error does not significantly increase even when the dataset grows. It is precisely in this critical assessment of one’s data, algorithms, and tools, where the human researcher’s ability to contextualize is still needed.