Summer workshop report, 2016
[Presentation slides in PDF]
For the HDW this summer we were assigned to Professor Kastor’s project on Creating a Federal Government. The aim of the project is to use quantitative data analysis to track and map the presence of various government officials throughout U.S. soil and abroad. For the main part of the workshop, our work consisted of transcribing into Excel sheets the data regarding the creation of post offices and the appointment of postmasters in the state of Virginia from the 1790s to the 1830s.
For each of the photographs we recorded the full name of the postmaster, the locality of the post office, the county where the post office was located, the date the post office was created, and a source code. After transcribing the data, there were some questions that we wanted to look at, such as the number of postmasters in office per year, the number of postmasters in office per county each year, and the average tenure in office of the postmasters in each state. We were also interested to see how these things changed over time. We used a program called R in order to clean up the data. In order to use the data to find the information that we wanted, we had to get rid of the things that didn’t provide us with useful information. For example, there were times when we had the the source code but not the name of the postmaster or their appointment date. We also had to make sure that all of the county names were standardized because there were times where a county’s name was spelled more than one way. We also had to get rid of duplicate postmaster names because we were interested in how many people served as postmasters at any time during this time period, so we did not want to recount people who were serving for more than one term. Lastly we got rid of the rows of data that were only telling us to look at another post office. Once we finished cleaning the data, we were able to export the information back to Excel and use it to find out what we wanted to know.
The process of cleaning up the data allowed us to standardize the information in order to answer a series of questions on the development of post offices over time and space. Our interest focused over the average tenure time of postmasters in both Virginia and Pennsylvania. We calculated that the average time in office for postmasters in Virginia was about 4.5 years, while in Pennsylvania it was closer to 5 years. We were also able to calculate the total number of postmasters and post offices present in either Virginia or Pennsylvania in any particular point in time. We selected a five-year time span and we then created a series of graphs that helped us visualize the development of post offices every five years from 1800 to 1830, both at the state and county level. This process helped us answer our questions about changes over time for the appointment of postmasters and creation of new post offices, but it also triggered a number of interpretative and methodological questions over the data we had so far collected and analyzed. In particular, it was evident that the process of cleaning up the data we could not use for our calculations implied a level of interference with the original information that had a deep effect on our calculations. Furthermore, the graphs visualizing the development of post offices over time showed a drastic drop in the number of postmasters created after the 1820s, which we were not expecting. We were able to explain this result by the fact that our original sources become very fragmentary after those years.
Overall, collaborating in this project allowed us to critically approach the use of quantitative analysis for the study of the past and made us aware of both the potentials and the limitations of quantitative studies of past sources.