Think Before You Count: Reflections on Trotti's recent article in the JAH

Being that the school year is still young and that the work is usually weighted towards the end of the semester, I actually had time to prop up my feet at the office and enjoy reading an article in the most recent edition of the Journal of American History that I got in the mail today. (I can already hear my professors lecturing me on how I'm supposed to use the free time I have now to get ahead on all the work that that will be due in December, even though they know damn well they did the exact sort thing when they were in grad school.) Of particular interest to me was Michael Ayers Trotti's article, "What Counts: Trends in Racial Violence in the Postbellum South." Trotti's piece is germane to most graduate students, regardless of their specialty, because it wrestles with the methodological pitfalls surrounding quantitative history. Trotti highlights the great difficulty inherent in quantitative historical analysis, especially when said analysis that has moral implications related to some of the darkest chapters in American history. (A quick disclaimer: this blog post is by no means intended to be an exhaustive summary of the article, but only notes some of the key points that I felt like particularly mattered to grad students.)

Trotti begins by reminding his readers that the very definition of lynching is itself contested and not always the best yardstick with which to measure the effectiveness of racial terrorism in American history. Definitions and statistical analyses of lynching emerged in the early twentieth century, usually through the work of groups like the NAACP, in order to combat the epidemic of violence. Usually, we think of lynchings as having multiple perpetrators. But Trotti points out that simply determining what counts as a "lynching" for the historian is far from simple and, furthermore, all lynchings were not created equal. Sometimes the murder of a solitary black leader by a white citizen provoked far greater terror than a brutal public lynching. Even still, it is impossible to even grasp at the total number of lynchings that occurred between 1880 and 1920 (what we usually think of as the high-point of racial terrorism in the US). Records are shoddy, testimony is often skewed, and ascertaining whether an event should count in the tally of a "lynching" is far from clear. (If three people watched one white person murder an African American, but didn't actually participate in the killing, how should the historian count it?) Trotti insists that scholars broaden their view beyond lynchings to a more expansive notion of racial terrorism because studies of "lynchings" fail to capture the totality of the situation for many black Americans living in this time period.

While few historians are willing to rely solely on statistical data on which to draw their conclusions, Trotti observes that when it comes to lynching, all of the work "is implicitly or explicitly predicated upon an understanding of trends.... And the lynching numbers that reveal these trends are problematic on more than one level. It is a strange subset of murder, and the evidence will never yield more than estimates of the numbers actually lynched." All historians have to start somewhere and when trying to examine lynching, statistics are a necessity for historians to uncover the locations and the quality of violence. It is ultimately up to each individual historian to decide what counts. In a unique twist, Trotti's article includes a  table that compares the overlapping statistics of a number of prominent monographs on lynching. In one case, he compares the data for Georgia lynchings between 1882 and 1920 compiled by Fitzhugh Brundage with that of Stewart Tolnay and E. M. Beck. While both works had similar overall totals, they only shared data for 69% of the total crimes chronicled. This disparity suggests that there are many lynchings yet undiscovered, but Trotti further suggests that numbers of still uncounted lynchings might mean we are also looking in the wrong decades. For years, the patterns these numbers revealed have framed studies of lynching, but under Trotti's careful scrutiny, they seem faulty and worthy of significant revision. With numbers of lynchings still yet to be discovered, the author asserts, historians should expand their their chronological gaze back to the 1860s and 1870s - a time of shocking racial violence that is often overshadowed by the latter decades of the nineteenth century.

For Trotti, an even bigger problem lurks in the methodologies used in lynching studies. Some historians have simply counted lynchings, and found the highest number of incidents. For example, Brundage simply mapped the number of total lynchings within a given geographic area (these historians like their data raw - raw is actually a pretty good word that could be used to describe a lynching). Conversely, Edward Ayers didn't look at the total numbers, but numbers of lynchings relative to the total population (lynchings per 100,000 people). Each approach yielded startlingly differing results with major implications for historians. Using Brundage's model, the Deep South emerges as the center of southern lynching; for Ayers it is the frontier regions of the South - those newly settled areas with growing populations, the home to new migrants of both races - places like Florida and Arkansas. Moreover, Trotti finds that the use of state boundaries in such studies to be problematic. Would a murder just a few miles across the state line in Alabama not mean more to blacks in the western part of Georgia than a lynching clear across the state in a place like Livonia? Finally, if historians account for the relative black population compared to the proportional number of crimes, the violence of Reconstruction dwarfs anything experienced by African Americans in the decades usually associated with racial violence. The number of crimes or the rate of crime will not tell the whole story, but at the same time numbers reveal nothing about the reasons why people perpetuated such violence or how they reacted to it. Statistics are a useful guide in places, but Trotti reminds us that they certainly have their limits.

Trotti concludes by asserting that there are just some things in history that will resist quantification. "Much of the past," he writes, "cannot be set to numbers. Counts, rates, and any calculations built from them necessarily flatten out the particulars, considering each lynching to be an equivalent point of data." Here the moral implications of reducing an event as unspeakable as a lynching to a piece of data should make even most ambitious graduate student pause for serious reflection about just what the hell they are doing. Trotti ends with terrific advise for any historian, but which should be particularly relevant to any grad student who hopes to incorporate statistical data into his or her dissertation. He advises us to "bring our methodological assumptions to the fore, making plain why we are counting in the way that we are counting."

Not only do we, as grad students, need to be aware that the way we count, what we count, and how we interpret it will profoundly effect our conclusions - we must also be aware that we are also actively creating data. Statistics do not exist in an archive just waiting for somebody to come along and interpret them. When it is compiled, statistical data becomes separated from its original primary source and becomes a creation of the person doing the compiling. This is not to say that you shouldn't go the quantitative route in your own research, but remember that any historical investigation worth its salt should have moral consequences. An awareness of exactly what you are asking from a particular set of data and what this data can realistically show you is a good first step as you probe the underlying moral implications of your methodologies.