Following the lead of the new political historians, we will analyze a real historical data set using correlation. Our case study involves the presidential election of 1860. As you may recall, there were four candidates for president in 1860: Abraham Lincoln (Republican), Stephen Douglas (northern Democrat), John C. Breckinridge (southern Democrat), and John Bell (Constitutional Union). We are concerned with the pattern of support for John C. Breckinridge, the most extreme pro-southern candidate, in the state of Mississippi, one of the first states to secede after Lincoln was elected. Statewide Breckinridge received 59.0% of the vote, compared to 4.7% for Douglas, 36.2% for Bell, and 0.0 % for Lincoln. But Breckinridge’s support was not distributed evenly across all of Mississippi’s counties. What kinds of counties do you think most strongly backed Breckinridge? More specifically, do you think there was a relationship between the density of slaves in a county and its level of support for Breckinridge? If so, what kind of relationship do you think there was?

Below you will find a table that lists the 58 counties in Mississippi for which the following data are available: (1) the percentage of voters who cast their ballots for Breckinridge in the county, and (2) the percentage of the county’s total population that was enslaved.

Table C: Percentage of Votes for Breckinridge and Percentage of Population Enslaved, Mississippi Counties, 1860

County	% votes for Breckinridge	% slave
Adams	38.3	70.9
Amite	52.5	64.0
Attala	66.0	35.4
Bolivar	43.0	86.7
Calhoun	65.9	19.2
Carroll	59.8	62.7
Chickasaw	65.2	55.3
Choctaw	66.3	26.7
Claiborne	59.3	78.4
Clarke	68.3	47.1
Coahoma	38.2	77.0
Copiah	65.6	51.7
Covington	77.4	35.5
De Soto	37.4	59.9
Franklin	67.8	57.5
Greene	81.6	31.6
Hancock	84.5	27.3
Harrison	83.9	21.1
Hinds	47.0	71.4
Holmes	55.1	67.3
Issaquena	42.8	92.5
Itawamba	68.1	19.9
Jackson	88.3	26.4
Jasper	65.3	41.3
Jefferson	51.4	80.8
Jones	73.3	12.2
Kemper	54.8	49.1
Lafayette	55.5	44.2
Lauderdale	65.8	38.2
Lawrence	84.7	40.1
Leake	65.1	32.8
Lowndes	56.6	70.8
Madison	53.5	77.5
Marion	88.9	46.6
Marshall	45.7	60.5
Monroe	65.8	59.8
Neshoba	81.0	26.5
Newton	73.5	35.0
Noxubee	58.4	75.0
Oktibbeha	72.8	58.8
Panola	38.3	62.0
Perry	64.3	28.3
Pike	79.0	44.3
Pontotoc	56.1	34.4
Rankin	56.7	52.1
Scott	69.3	36.4
Simpson	72.7	38.2
Smith	68.4	28.7
Sunflower	55.6	78.0
Tallahatchie	48.6	64.1
Tunica	45.0	79.8
Warren	39.2	66.5
Washington	47.2	92.3
Wayne	62.1	52.7
Wilkinson	53.0	82.4
Winston	72.6	43.0
Yalobusha	54.0	56.2
Yazoo	48.1	74.7

Source: Great American History Machine

Now let us look at these data graphed as a scatterplot.

Graph 9: Relationship Between Percentage of Vote for Breckinridge and Percentage of Population Enslaved in Mississippi Counties, 1860


What does the scatterplot suggest about the relationship between support for Breckinridge and slave density in Mississippi counties in 1860? ANSWER

With the help of Microsoft Excel, we can further specify the relationship by calculating the coefficient of correlation. Pearson r is -.73. What does this number mean? ANSWER

Are you surprised? Many students assume that on the eve of the Civil War pro-southern extremism was strongest where slavery was most deeply entrenched. But our analysis of the Mississippi data suggests otherwise. Support for Breckinridge was greatest where slave density was lowest, not highest. By itself the existence of this negative correlation does not explain why Mississippi counties with more slaves were less favorably inclined toward Breckinridge. Nor does it contradict the fact that Mississippi as a state voted overwhelmingly for Breckinridge in 1860. Yet if numbers can speak, this figure cries out for further investigation. As a next step in the research process, we could turn to election statistics for other southern states to see if the same pattern holds for them.* Alternatively we could explore into the Mississippi sources, such as newspapers or political speeches, further to see what more we can discover there. Either way, we now know better than to assume that the prevalence of slavery alone explained the pattern of southern extremism in 1860.

Even if quantitative methods are better at dispelling myths and challenging simple assumptions than they are at proving arguments about historical causation, they can serve as critical tools in the hands of all kinds of historians. As a beginning historian, try to approach numeric data as you would other types of evidence, with seriousness and skepticism, and devote enough time and energy to mastering the quantitative skills you need to accomplish your research goals. Good luck!