This page is referenced by:

• 1 2022-10-17T15:00:52+00:00 Analyzing Complexity: A Granular, Data-Driven Approach 55 plain 2024-02-27T15:32:05+00:00 Developing an analysis of the complexity of the disorder requires reconstructing and interpreting the details of the events. This study employed a data-driven method to comprehensively examine all the events of the disorder rather than selectively focus on only some events. I created data about an event from a variety of sources and categorized it as one of eight types of event in order to find patterns. While many of those sources feature in other studies of racial disorder, they had not been used in previous studies of the disorder in Harlem in 1935: docket books that record cases before the Magistrates Court; case files from the District Attorney’s office for those sent to the grand jury; case files from the Probation Department for those convicted in the Court of General Sessions; transcripts of the public hearings held by the Mayor’s Commission on Conditions in Harlem (MCCH); records of the investigations undertaken by the MCCH contained not just in the records of Mayor La Guardia but also in the papers of E. Franklin Frazier, the Black sociologist who led the group's study of Harlem, and in the papers of Arthur Garfield Hays, the MCCH member who chaired the subcommittee responsible for studying the events of the disorder; photographs of every building in Harlem taken by the city’s Tax Department four to six years after the disorder; and twenty-two major newspapers available on microfilm or digitized and the photographs of the disorder that they published. Data-driven analysis typically employs computationally generated data, as Lincoln Mullen did in America’s Public Bible. By contrast, the data in this project was handcrafted, manually curated by cross referencing and reconciling information from a variety of different sources. All the data was included, not just what could be fitted to arguments made in the narrative: events without information on timing, so not in the chronology; events that have no location, so not on the map; and events that appear only as prosecutions in court, so neither in the chronology nor on the map. (The sources themselves are not part of the site; the published sources are still under copyright and access to the legal records and other archival sources is restricted).
  
  The granular approach of this study allows it to counter the tendency of data to dehumanize its subjects. As Harmony Bench and Kate Elswit point out, “The process of manually curating data from archival materials draws us closely into the lived experiences they index, as we grapple with the multiple and conflicting stories behind each data point, and what each signifies.” Black digital practice, as exemplified by Jessica Marie Johnson, emphasizes the need to carry that engagement forward, to “infuse the drive for data with a corresponding concern with and for the humanity and souls of the people involved.” The attention to individual events of this study allowed all the men and women involved to be named, many to be identified, and a small group to have their experience in the disorder put in the context of a larger picture of their lives. The names of participants, where possible, appear in the labels for an event: the owner or staff member of a looted or damaged store; or the individual killed, injured, or arrested. Although fragmentary, such information offers some counterbalance to the reliance on aggregate numbers to describe the disorder in the existing literature on Harlem in 1935. In an additional effort to foreground the humanity of the data, charts visualizing those involved in the events of the disorder, and in the legal proceedings in its aftermath, were made using Wee People (https://github.com/propublica/weepeople), a typeface of people silhouettes developed by Alberto Cairo for ProPublica.
  
  In line with the developing practices of humanities data analysis, the process of interpretation involved in creating data has been made transparent, taking advantage of the digital medium to include pages discussing each event. The goal is to focus on historians' interpretations, not sources; with ever more sources available online, it is important, as historian Karin Wulf captured in a tweet, that people “understand why history isn’t obvious, and why we know and understand new things about the past.” The individual event pages provide the data’s backstory, in greater detail than the summary traces of the process of data creation that can be left in a dataset. Each of the sources of information on an event are discussed in those pages, agreement and disagreement among sources are identified, and decisions about what information to use and why, and about how to categorize an event, are explained. The analysis on event pages also highlights gaps in the information, when the categorization, timing, or location of an event are uncertain, to counter the apparent certainty with which the event appeared in the chronological narrative and on the map.
  
  To analyze patterns in the events of the disorder, they were classified into one of eight categories. Arrests are a separate category of event to counter the tendency to assume those arrested are guilty or at least involved in the related event. For Harlem in 1935, that is a particularly questionable assumption given the practices of the predominantly white police officers responsible for making those arrests. Events are grouped into additional subcategories based on different features of those categories, who was involved, and whether an arrest or prosecution followed. The Assaults page, for example, includes links to the fifty-four events grouped as assaults. It also features fifteen related tag pages for the subcategories into which those events are grouped. Six tags are forms of assault, six tags are based on the identities of the alleged victim, and two tags are based on the police response. There is also a tag for assaults by police, to highlight a gap in the data: the absence of specific incidents of violence by police notwithstanding widespread statements about police beating and shooting at people on the streets. Additional tags group those who were injured in alleged assaults, alleged assaults that resulted in arrests, and arrests that resulted in prosecutions. In combination, these subtags create a dense web of relationships between the events grouped as assaults. Employed in that way, categories bring patterns in the disorder into focus at multiple scales. While classifying data necessarily simplifies it, a more complex, disaggregated view of the events is accessible in the maps and in the layer of pages discussing individual events linked to each page analyzing a category.
  
  Event data also provided the basis for a spatial analysis of the disorder. For 315 events there was information on location that allowed a map of where the disorder happened to be created. Identifying the individuals involved in those events made it possible to map the residences of many of them, providing a sense of where those on Harlem's streets that night had come from. However, the maps are a partial picture, including only a portion of the events of the disorder. For forty-nine of the 128 arrests, 38% of the total, no information on location was found. Of the seventy-nine that can be mapped, information on the timing of only forty-seven could be found. Overall, the timing of only half of the events (158 of 315) that can be mapped was known. With those qualifications, the map allowed a spatial analysis that explores how the disorder changed in character, intensity and concentration over time, and compares the spatial patterns of different categories of events.
  
  To analyze the central role that businesses played in the disorder, the spatial data was visualized on a map that represents the footprints of the buildings in Harlem. The historical map layer was created from the Atlas of the City of New York, Borough of Manhattan, Volume 4, published by G. W. Bromley & Co. in 1932. Included on the atlas maps are information on the material from which buildings were constructed, the number of floors and whether they included a basement and stores, and information on the width of streets, presence of elevated railroad and subway lines, and stations and trolley lines. This granular attention to Harlem's business landscape was extended using the 1940s Tax Department photograph collection. Intended as the basis for a more equitable property tax assessment, the images offer frontal views of over 720,000 properties in the five boroughs of New York City taken between 1939 and 1941. As photographic historian Gabrielle Esperdy puts it, the images serve as a "visual parallel to census data." Although the photographs were taken four to six years after the disorder, many of the businesses involved in the events of the disorder appear in them. Even when the businesses have closed or the photographs were taken from too great a distance for individual businesses to be identified, the buildings and storefronts in which they were located are pictured, providing a street-level view of the varying physical environment of the disorder. What the photographs do not convey is a sense of the crowds of people on Harlem's streets during the disorder. The photographers tended to work at times when few people were on the street or to wait until there were no people in the frame to take a photo, unsurprisingly given that their task was to capture images of the structures.