Not Surprising, Part Deux
Recently Michael Rodgers reminded me of this post from 2011, in which I pointed out the very tight correlation (beyond -0.9, for y’all stats wonks) between the order in which the states that formed the Confederacy seceded, and both the proportion of slaves and the proportion of slaveholders in their total population. It’s based on a table compiled by David C. Hanson of Virginia Western Community College, and came to my attention via Donald R. Shaffer’s Civil War Emancipation blog. It makes a fundamental point but, in tabular format, the point may not be obvious to all. At the time, Shaffer wrote that “Hanson makes a definite connection between the concentration of slaves and slaveholders in a particular southern states and when it seceded from the Union in 1860-61, and whether it seceded at all. Certainly, there is plenty of anecdotal evidence connecting slavery and Civil War causation, but this table also makes a compelling statistical case.”
It’s more than a “definite connection.” Hanson’s simple data set reveals that for the eleven states that actually seceded, the proportion of free persons owning slaves was an excellent predictor of the order in which they seceded, beginning with South Carolina (highest proportion of slaves and slaveholders) and ending with Tennessee (low proportions of both). The percentage of slaves as part of the state’s overall population was equally reliable as a predictor. In short, each percentage is a nearly perfect predictor of where each state falls in the order of secession. The states with the largest proportions of slaves and slave-holders seceded earliest.
Anyway, back to data presentation: here’s Hanson’s data shown in two simple bar charts. There is absolutely a clear pattern in both cases; if slaves and slaveholding were not connected to the timing of each state’s secession, the tall and short columns would be randomly scattered. They’re not.