Japanese Internment at Poston, AZ during WWII

Research Questions

After first visiting the Isamu Noguchi museum in Queens and later coming across a dataset about interned Japanese during World War II, I wanted to explore the lives of interned Japanese and Japanese-Americans in internment camps. Noguchi was a Japanese-American artist from New York who voluntarily self-interned at the Poston, Arizona camp to promote unity and the arts to his fellow internees. He was met with suspicion from both the internees and the white Americans running the camp. His letters from that time describe a rich civic life within the camp. Despite this, there were divisions between Japanese aliens and Japanese-Americans, people with differing political ideologies, and the second (Nisei) generation and first generation. I wanted to look at the gender, age, citizenship, and arrival and departure reasons of the Poston residents, as well as explore any other information in the dataset.


Example Datasets

After selecting my data, I discovered a Tableau dashboard with a related set of data about interned Japanese. While the dataset is similar to mine, it uses WRA forms from all of the camps, while I use data from FAR forms from Poston, AZ. Though they record some overlapping information, there is also unique data in each set. The dashboard by Deron Hayes-Hirschy (Fig. 1) is pleasingly laid out, with a nice use of color gradient to allow exploration of detail by camp name without over-complicating the image. I liked that it showed a variety of chart types, with bar charts, a map, and a line graph used appropriately. The interactivity worked to emphasize the individual camps versus the aggregate.

Fig. 1 Hayes Hirschy dashboard

I thought that data about the present-day movements and demographic makeup of refugees would be analogous to the data about internment, so I found two visualizations related to the current refugee crisis. The first was a very simple bar chart (Fig. 2) from the Foundation for Economic Education; because I have little experience with Tableau, I thought I would start by trying to make something similarly simple, before working towards more complicated visualizations. I liked that the ages were grouped and that genders were broken into separate bars instead of stacked, as well as color coded. The second was a more complex bar chart (Fig 3) from the European Commision and Statista that broke total refugees into bars based on destination country and further color coded based on the original refugee camp in Italy, Greece, and Hungary. I liked this way of displaying origin and destination, as well as the totals on the right and map layered underneath. Though I knew I could not achieve all of this in Tableau, I wanted to get close.

Fig. 2 FEE bar chart


Fig. 3 European Commision and Statista bar chart

Data and Technologies

The data set I used was featured in the Data is Plural newsletter by Jeremy Singer-Vine. It is from the National Archive’s collection of Final Accountability Roster forms made during the internment period. The forms were transcribed by volunteers, many themselves Japanese-American and descendents of the original internees, through the Densho project. It has 19,915 records, described by twenty eight elements. Of interest to me were the birthdates, entry and exit type, entry and exit dates, city of origin, gender, citizenship, and notes dimensions.

I brought the .csv file into OpenRefine, where I faceted all of the relevant fields in order to check for consistency, reduce spelling errors by combining terms, and add new columns to decode the entry and departure fields. I also added a new column called “Married in Camp” that I marked “Yes” or “No” based on the notes section where it appeared marriages had been recorded.

I then brought my data into Google Sheets so that I could have an easily updated source for my Tableau visualization. In Google sheets I added a column where I calculated age based on the year the person entered the camp and the year they were born. I created an additional element to group ages into ranges for simplified analysis.

From there I moved to Tableau Public 10.3.0. Unfortunately, I later had to move my visualizations into 10.4 but this did not appear to affect the visualizations.



For my first visualization, I placed the age groups in ascending order as columns and summed the number of records in each group for rows. I displayed the results in a visual bar chart. I added gender as both another column element to get two distinct bars per age group, as well as a color detail. I selected a pinkish-purple for women and a green-teal for men. These colors are evocative of the (American) tradition of pink for women and blue for men, encoding some preconscious meaning into the visualization without relying too heavily on traditional gender color stereotypes. I removed the label from the y-axis as the numbers seemed self-evident and left the Age and gender labels on the x-axis.

For my second visualization I made a similar bar chart with age groups as columns but with citizenship status as the second column element. Citizens were coded red and Aliens were coded blue. I chose these colors because they would work for colorblind viewers. I excluded the much smaller number of records that were unclearly coded. I also excluded the label from the y-axis.

The third visualization is a map showing the cities that Poston internees arrived from. I had Tableau calculate longitude and latitude of each city, which it then plots on a map. I added number of records as a size element for each dot and had the map display city names. Unfortunately, it does not display names where clusters are dense, but zooming allows greater detail.

In a fourth visualization I plotted dots representing individual internees against their arrival date on the x-axis and their departure date on the y-axis, with departure and arrival reasons as labels while hovering over individual dots. I used number of records (per day of release) to size the dots. I filtered to find Isamu Noguchi in the data, made a label for him, and then unfiltered to show the entire scatterplot.

I then grouped those for visualizations into a dashboard (Fig 4).

I also tried two charts that I left off the final dashboard. One was a bar chart showing number of internees per origin city, which was then color-coded by destination city. So many internees were from California, and so many returned to the city that they came from that this was not easily read at the scale it displayed. I also tried plotting the number of people in different age groups that were married in camp, only to discover that the original data had only noted when women were married, making the visualization less interesting to me.



Fig. 4 Poston Arizona Internment Camp, 1942 – 1946 (Screenshot because the embed wasn’t working)

Through the Gender by Age Group visualization I saw that the majority of internees were young, between one and thirty years old. There were also many more older men than women in the camp.

In the Citizenship Status by Age Group visualization I found that younger internees were more likely to be citizens while older internees were overwhelmingly Alien residents.

In the Cities of Origin visualization I had originally wanted to make a spider map showing paths to and away from Poston. However, this required additional data manipulation and skill. However, I think it conveys the geographic spread of residents, with the majority coming from Los Angeles and California, more broadly much better than the bar chart I had originally envisioned.

In the Entries and Departures chart, I was pleased with how the layout of individual records shows that the majority of residents entered early in the time period and left in a steady stream, with a cluster of departures in the middle and end years. I also liked that in hovering over the records that trailed into later arrivals showed that the majority were newborns, with some transfers. I then found some records where arrival dates are later than entry dates, which I would exclude in future. I liked that pointing out Noguchi emphasized the relative shortness of his stay and located him among thousands of other internees.


Future Directions


In future I would remove the gender and citizenship labels from the bar charts and add a color key instead. I would also change the color of the citizenship chart, with red for Alien internees to evoke the Japanese flag and yellow or another colorblind-appropriate color for Citizens. As it is now, the blue for Citizens is the same as the blue in both the map and scatterplot, implying undue relation.

For the scatterplot, I would like to point out more people who self-interned voluntarily, like Noguchi, perhaps marking them with a uniform color and adding their names when hovering. I would like to group some of the reasons for internment and departure into broader categories and perhaps show those summed up in bar charts.

I would like to learn more about making spider maps or other visualizations that show relationships in order to show the flow of internees from origin cities, to camp and back out to destinations.

I would also like to go deeper into the notes section and extract the names of the husbands (listed in many cases) for women marked married in camp, in order to show the full set of people who were married and compare their ages by gender.