Article contents
Southern Politics Revisited: On V. O. Key's “South in the House”
Published online by Cambridge University Press: 28 September 2015
Abstract
V. O. Key's Southern Politics in State and Nation continues to be a central text in political science, the single most important work in understanding the role of the South in American politics. This article returns to, replicates, and seeks to advance Key's analysis of southern politics in Congress, reanalyzing and extending his account of southern strategies and actions in the House of Representatives. Where Key's text was characterized by an episodic attention to issue substance, we focus directly on how southern representation varied across discrete issue areas. We generate temporally fine-grained issue-specific ideal points for members of Congress that allow us to determine how congressional preferences changed across time, generating a more refined portrait of the process by which southern Democratic members diverged from their northern counterparts. We also thicken and extend Key's account along regional and temporal dimensions, assessing how his findings change when we employ a legal-institutional definition of the South, and include the whole period from the beginning of the New Deal to the close of the Truman administration. The article concludes by detailing the significance of our finding to the study of American politics, particularly American political development.
- Type
- Research Article
- Information
- Copyright
- Copyright © Cambridge University Press 2015
References
1. Aldrich, John H., “Southern Parties in State and Nation,” Journal of Politics 62 (2000): 644 Google Scholar, 643.
2. Alexander Heard, “The Making of Southern Politics,” Perspectives on the American South: An Annual Review of Society, Politics, and Culture, ed. Merle Black and John Shelton Reed (New York: Gordon and Breach Science Publishers, 1984), 6. The president had been notified by Governor Ellis Arnall of Georgia, and in his letter to Key expressed his “opinion that the project is an important one, and [his] hope that you will find it possible to undertake this assignment.” Alexander P. Lamis, “Southern Politics at the Time of V.O. Key,” The Oxford Handbook of Southern Politics, ed. Charles S. Bullock III and Mark J. Rozell (Oxford: Oxford University Press, 2012), 31. Shortly after receipt of the president's letter, Key wrote to Roscoe Martin, saying “I suppose I'll be hearing from the Pope next.” Heard, “Making of Southern Politics,” 6.
3. Gunnar Myrdal, with the assistance of Richard Sterner and Arnold Rose, An American Dilemma: The Negro Problem and Modern Democracy (New York: Harper & Brothers Publishers, 1944). For an examination of Southern Politics’ absences, see the contributions in Angie Maxwell and Todd G. Shields, Unlocking V.O. Key Jr.: Southern Politics for the Twenty-First Century (Fayetteville: University of Arkansas Press, 2011).
4. Holcombe, Arthur N., “Review of Southern Politics in State and Nation, by V.O. Key, Jr.,” American Political Science Review, 44 (1950): 192 CrossRefGoogle Scholar.
5. Valdimir Orlando Key, Southern Politics in State and Nation (New York: Knopf, 1949), 666.
6. Key, Southern Politics, 671–72.
7. Key, Southern Politics, 664.
8. Key, Southern Politics, 671.
9. Key, Southern Politics, 392, 387.
10. Aldrich, “Southern Parties,” 662.
11. Key, Southern Politics, 315–16.
12. By contrast, a recent collection of essays on “Southern Politics for the Twenty-First Century” pays scant attention to the two chapters on Congress. Maxwell and Shields, Unlocking V. O. Key, Jr.
13. There are a number of reasons to limit the analysis to the House. Space limitations mean that an analysis that incorporated both the House and the Senate would be unlikely to go beyond a bare replication of Key. As discussed above, our goal is not simply to replicate Key but to anticipate what he would have done had he had a policy coding scheme and ideal point estimation techniques available. The methodological motivation is that the greater number of House members relative to the Senate allows for greater analytical leverage and enables us to better explore differences between southern members.
14. Katznelson, Ira and Mulroy, Quinn, “Was the South Pivotal? Situated Partisanship and Policy Coalitions during the New Deal and Fair Deal,” Journal of Politics 74 (2012): 604–20Google Scholar.
15. Ira Katznelson and John S. Lapinski, “The Substance of Representation: Studying Policy Content and Legislative Behavior,” in The Macropolitics of Congress, ed. E. Scott Adler and John S. Lapinski (Princeton, NJ: Princeton University Press, 2006), 96–126.
16. While based on legal and institutional measures, the seventeen-state South also has, or at least had, broader cultural significance in the minds of southern representatives. Frank E. Smith, the “Congressman from Mississippi,” wrote in 1964 that “race has thus been the major influence in Southern politics for the last hundred years. ‘Southern’ in this case means more than the former states of the Confederacy—it includes border states like Maryland, West Virginia, Kentucky, and Oklahoma, and parts of states like Missouri and New Mexico.” Frank E. Smith, Congressman From Mississippi (New York: Pantheon Books, 1964), 111. The Southern Governors’ Conference in 2014 is composed of sixteen of the seventeen states we include—the exception being Delaware—as well as Puerto Rico and the U.S. Virgin Islands. The Brown v. Board of Education decision originated in Kansas, which was one of four states that allowed but did not require racial segregation in schools. The constitution of Wyoming prohibited distinction on the basis of race in public schools, although a statute had authorized local authorities to do so when there were more than fifteen black students. This option does not seem to have ever been employed. Arizona required segregated schools until 1951, while in New Mexico and Kansas school segregation was decided at the local level. Sutherland, Arthur E., “Segregation by Race in Public Schools, Retrospect and Prospect,” Law and Contemporary Problems 20 (1955): 169–83CrossRefGoogle Scholar; Charles T. Clotfelter, After “Brown”: The Rise and Retreat of School Desegregation (Princeton, NJ: Princeton University Press, 2004), 18.
17. Farhang, Sean and Katznelson, Ira, “The Southern Imposition: Congress and Labor in the New Deal and Fair Deal,” Studies in American Political Development 19 (2005): 1–30 Google Scholar; Caughey, Devin and Schickler, Eric, “Public Opinion, Organized Labor, and the Limits of New Deal Liberalism, 1936–1945,” Studies in American Political Development 25 (2011): 162–89Google Scholar; Katznelson and Mulroy, “Was the South Pivotal?”
18. Ira Katznelson, “Historical Approaches to the Study of Congress: Toward a Congressional Vantage on American Political Development,” in The Oxford Handbook of the American Congress, ed. Eric Schickler and Frances E. Lee (New York: Oxford University Press, 2012), 115–37.
19. David Mayhew, Divided We Govern: Party Control, Lawmaking and Investigations ( New Haven, CT: Yale University Press, 1991); Gregory J. Wawro and Eric Schickler, Filibuster: Obstruction and Lawmaking in the U.S. Senate (Princeton, NJ: Princeton University Press, 2006), 37.
20. Topkis, Jay, “Review of Southern Politics in State and Nation, by V.O. Key, Jr.,” Yale Law Journal 59 (1950): 1203–6Google Scholar.
21. Ewing, Cortez A., “Review of Southern Politics in State and Nation, by V.O. Key, Jr.,” Journal of Politics 12 (1950): 155 Google Scholar.
22. A cohesion score is calculated as the absolute difference between the number of members of a defined group who favor or oppose a given roll call, divided by the number of voting members of this group. The cohesion scores reported by Key, and by us, are averaged across relevant groups of roll calls. Cohesion and likeness scores were pioneered by Stuart Rice and have been deployed widely since. See Stuart Rice, Quantitative Methods in Politics (New York: Knopf, 1928). For the most pertinent critiques, see Krehbiel, Keith, “Where's the Party?,” British Journal of Political Science 23 (1993): 235–66Google Scholar; and Krehbiel, “Party Discipline and Measures of Partisanship,” American Journal of Political Science 44 (2000): 212–27Google Scholar. The concerns raised by Scott Desposato, that cohesion scores are biased for small groups, should not apply here, as there was always a sizeable contingent of southern Democrats, Republicans, and nonsouthern Democrats. Desposato, Scott, “Correcting for Small Group Inflation of Roll-Call Cohesion Scores,” British Journal of Political Science 35 (2005): 731–44Google Scholar.
23. If the cut-line for a bill—the halfway point between the status quo and the policy proposal on an array of policy preferences—falls right in the middle of a bloc, it is likely that a low cohesion score will result, while a cut-line that appears on an extreme point distant from this bloc will likely result in a high cohesion score. This is true whether the bloc has strongly similar or widely diverse preferences.
24. The slight differences between our scores and Key's are likely the result of our coding members as having voted for or against a measure if they were paired, and different assessments of whether a vote was procedural.
25. These scores are effectively equivalent to party unity scores, in that they are counts of the number of votes on which a majority of one faction voted against a majority of another.
26. These are calculated by the equation: 100–|%Yes bloc 1 - %Yes bloc 2|
27. Krehbiel, “Party Discipline and Measures of Partisanship”; Weisberg, Herbert, “Alternative Baseline Models and Their Implications for Understanding Coalition Behavior in Congress,” Journal of Politics 45 (1983): 657–71Google Scholar; Shade, William, Hopper, Stanley, Jacobson, David, and Moiles, Stephen, “Partisanship in the Unites States Senate: 1869–1901,” The Journal of Interdisciplinary History 4 (1973)Google Scholar; Hammond, Thomas and Fraser, Jane M., “What Role Calls Should We Exclude from Conservative Coalition Calculations?” Legislative Studies Quarterly 7 (1982): 423–34CrossRefGoogle Scholar.
28. If southern Democrats and Republicans have a likeness score of 80 for a series of votes, one might interpret that as implying a high degree of similarity between the two blocs. But if the likeness score for southern Democrats and northern Democrats is 100, then what in fact occurred was that the three blocs all voted nearly unanimously.
29. Variations of these graphs were made that also looked at the likeness between northern and southern Republicans, and southern Republicans and southern Democrats. With some exceptions, southern Republicans were all located in the seventeen-state rather than the eleven-state South. The intraparty likeness scores were consistently high, while the intraregional cross-party scores closely mapped on to that of the southern Democrats and the northern GOP. Accordingly, we limit the attention paid here to southern Republicans. While southern Republicans of the period merit much more attention than the dismissiveness with which they were treated by Key, for space reasons we leave that analysis to a future date.
30. Key, Southern Politics, 372.
31. For space reasons, the scheme has been reduced to highlight those areas discussed here. For the full scheme, see Katznelson and Lapinski, “The Substance of Representation,” 112–13.
32. Key, Southern Politics, 360, 345.
33. Of the votes identified by Key, sixteen concerned the tier 1 category of “sovereignty,” including two votes on the tier 3 category “immigration,” ten on “African American civil rights,” and four on “voting rights.” Twenty-four concerned the tier 1 category of “international relations,” including one on the tier 3 category of “conscription,” one on “foreign aid,” four on “international organizations,” nine on “trade,” and an additional nine on “preparation for war,” a category likely to fall under “defense,” but for which we were not able to find which specific roll calls to which he was referring. Twenty-seven votes concerned the tier 1 category of “domestic policy,” including two on the tier 2 categories “agriculture and food,” five on “planning and resources,” twelve on “political economy”—nine of which were in the tier 3 category of “labor markets and taxation,” and seven were on “social policy.” An additional six votes concerned quasi-private bills—such as the deportation of Lazar Limonsky, denying federal pay to specific radicals and union leaders on public works, the relief of “sundry aliens,” payment of Union Iron Works, and a vote on relocating offices away from the District of Columbia. Key also identified an additional two votes on education, which are coded under this scheme as civil rights for African Americans as they concerned nondiscrimination. See Key, Southern Politics, 351–54, 356–59, 371–77.
34. Key selected a specific subset of votes based on whether they crossed a specified threshold. Instead, we use the votes he analyzed to identify discrete issue areas in which all of the relevant votes are taken into account, and we generate measures that are specific to these issue areas.
35. Clinton, Joshua D., Jackman, Simon, and Rivers, Douglas, “The Statistical Analysis of Legislative Behavior: A Unified Approach,” American Political Science Review 98 (2004): 355–70Google Scholar; Clinton, Joshua D. and Jackman, Simon, “To Simulate or NOMINATE,” Legislative Studies Quarterly 34 (2009): 593 Google Scholar.
36. Specifically, they estimate the model y
it
= α
t
+ β
t
X
i
+ ε
it
, with y
it
being legislator i's ideal point at time t, X
i
being a mean-preference parameter—initially the mean score for a member over their entire career—and ε
it
being an error term capturing individual change. They generate “shift” and “stretch” parameters for each unique session—a chamber in a given congress, a legislative session, a bloc of time—and use this to adjust all the scores for this session, with the formula
${\hat y_{it}} = \displaystyle{{\left( {{y_{it}} - {\alpha _t}} \right)} \over {{\beta _t}}}$
where
${\hat y_{it}}$
is the “adjusted” score for member i’ at time t , and α
t
and β
t
are the session specific shift and stretch parameters. The major limitation of the Groseclose and coauthors’ technique, however, is that it assumes members’ movements are idiosyncratic, and that the mean long-run weighted average of congressional members remains the same. The estimates are comparable across time for each issue area; but while the rank orderings across issue areas are comparable, their specific location or the distance between members is not directly comparable across issue areas. Tim Groseclose,
Levitt, Steven D., and Snyder, James M., “Comparing Interest Group Scores across Time and Chambers: Adjusted ADA Scores for the U.S. Congress,” American Political Science Review
93 (1999): 33–50
Google Scholar. See Joshua D. Clinton, Ira Katznelson, and John Lapinski, “Where Measures Meet History: Party Polarization During the New Deal and Fair Deal,” in Governing in a Polarized Age: Elections, Parties, and Representation in America, ed. Alan Gerber and Eric Schickler New York, NY: Cambridge University Press, forthcoming); John Lapinski, The Substance of Representation: Congress, American Political Development, and Lawmaking (Princeton, NJ: Princeton University Press, 2013).
37. For a similar approach see Bonica, Adam, “Punctuated Origins of Senate Polarization,” Legislative Studies Quarterly 39(2014): 5–26 Google Scholar.
38. We find that moving from 50 to 100 does not significantly change the recovered ideal points, but does decrease the temporal granularity. All ideal point techniques suffer from the possibility that the roll calls will not sufficiently discriminate between members, and increasing the number of roll calls is no panacea against this. See Rosenthal, Howard and Voeten, Erik “Analyzing Roll Calls with Perfect Spatial Voting: France 1946–1958,” American Journal of Political Science 48 (2004): 620–32Google Scholar. We find that for tier 1 issue areas, there are sufficient roll calls for either a six- or a twelve-month window. For some tier 2 issue areas there are sufficient roll calls for a twelve-month window, but that for all but a few tier 3 issue areas, we need to arbitrarily set a number of roll calls (rather than a temporal interval) to subset. While this increases the ability to discriminate between members, the problem is that it requires merging across many different congresses—several years in which the politics around an issue might change but only a single score will be generated. To compensate for this, we rely on the inflection points identified with cohesion, likeness, and the tier 1 and tier 2 ideal scores to temporally bound the selection. For instance, as we discuss below, we subset the tier 2 “civil rights” roll calls by merging several congresses, but only those after the inflection that seems to have occurred in the 78th Congress. Note also that the scores are calculated for months in which Congress is in session, and so do not constitute an exact calendar year. Where a specific event or set of debates might be responsible for sudden changes in legislative behavior, this can be confirmed by locating the median at the final date rather than at the center. Merging across several years is certainly not a perfect solution, but it is a widespread practice in estimating across time. DW-NOMINATE estimates scores based on a member's entire tenure, as do the state-level scores estimated by Shor, Boris, Berry, Christopher, and McCarty, Nolan, “A Bridge to Somewhere: Mapping State and Congressional Ideology on a Cross-Institutional Common Space,” Legislative Studies Quarterly 35 (2010): 417–48Google Scholar.
39. In this sense they are equivalent to Common Space scores. Poole, Keith, “Recovering a Basic Space From a Set of Issue Scales,” American Journal of Political Science 45 (1998): 954–93Google Scholar.
40. In estimating issue-specific ideal points, we are seemingly going against the finding of Poole and Rosenthal that most of congressional voting can be reduced to a two-dimensional space. Keith Poole and Howard Rosenthal, Congress: A Political Economic History of Roll Call Voting (New York: Oxford University Press, 1997). We do not see it this way. The two-dimensional model established by Poole and Rosenthal was chosen based on its parsimony in explaining voting across the entirety of American history. In most congresses, they find some gain to estimating additional dimensions, but that they are not worth including relative to the task of explaining voting in the aggregate. Scholars interested in using preference-based measures to understand how specific issues change over time—acknowledging that these preferences will not be unrelated to the party cleavage that provides most of the structure to the first DW-NOMINATE—are left with few options in DW-NOMINATE. Where these issues generate different ordering of preferences but do not constitute a sizeable portion of the agenda, they will simply appear in the two-dimensional model as an increased rate of error. We propose issue-specific scores not as an intervention in the debate as to the number of dimensions, but because they allow for closer inspection of how given preferences on a given issue change over time. We are also persuaded by the reasons as to why scholars might want to subset roll calls, as we have done here, offered by Keith Poole. One is to “uncover the microstructure of the spatial map,” to find those issues that in the short-term might result in distinct shifts in the aggregate location of members. Another is “to uncover what is going on when structural change is occurring.” Keith Poole, Spatial Maps of Parliamentary Voting (Cambridge: Cambridge University Press, 2005), 185. Southern preferences did clearly shift in an important way during this period, which is also the period in which the second dimension becomes clearly relevant in voting patterns, suggesting the possibility of a structural change for southern representatives, if not for the legislature as a whole. Moreover, as we will see when we examine distinct issue areas, the aggregate shift in the southern position was a reflection of the specific shifts that they took on distinct issue areas, and the aggregate spatial map was produced by how these distinctive patterns of shifts and stability occurred at different times. Understanding the microfoundations of the spatial maps that place the southerners to the right of the nonsouthern Democrats requires closer attention to how preferences potentially varied by issue.
41. The decline in the northern Democratic median reflects the electoral loss of approximately 30 percent of their seats in November 1942, as well as the loss of approximately 40 percent of their seats in November 1946. But these losses persisted through the 81st and 82nd Congresses, suggesting a transformation in the northern faction of the party during this period.
42. For a similar finding of low southern preference heterogeneity, see Devin Caughey, “Congress, Public Opinion, and Representation in the One-Party South, 1930s–1960s” (Ph.D. diss., University of California, Berkeley, 2012), 49.
43. As an empirical matter, the roll calls for the period did indeed cluster at the center of the chamber: 74 percent of roll calls saw cut-lines within the two standard deviations surrounding the chamber mean.
44. In the days immediately preceding the attack, the South and the GOP had supported the Carl Vinson's (D-GA) Anti-Strike Bill, which would limit strikes in defense industries, limit the establishment of new closed-shop defense factories, and require industry to take an oath of all persons that they were not Communists or Bund members, and to discharge those who were. Southern opposition to labor unions, which was an important component of their rightward movement, preceded the attacks in December. In the months following Pearl Harbor, southern Democrats allied with Republicans on amending the Employment Stabilization Act (2/19/1942), on the Agricultural Appropriations Bill (6/9/1942), on the sale of grain and wheat (6/26/1942 and 7/15/1942), and on the Emergency Price Control Act (9/23/1942). The next year, southern Democrats and Republicans voted together on a version of the Hobbs Act targeting labor unions (4/9/1943), on a bill regulating wartime use of production plants (6/4/1943), on the Commodity Credit Corporation (11/23/1943), and on a host of other issues falling in the category of domestic politics.
45. All ideal-point measures share the problem that changes in these can be a product of changes in the agenda or in preferences, and that there is no perfect means of isolating which is responsible for shifts in the estimated scores. This problem, however, is potentially greater when scores are estimated over short time windows. With longer time horizons, short-term variation in the agenda is likely to be averaged out, although the problem then becomes one of interpreting the meaning of the recovered dimension as the mix and content of issues changes over time. The cost of longer time horizons, however, is that they can result in misleading assessments of political conflict at particular moments and obscure potentially important differences in preferences across issue areas. There is no perfect balance between fine-grained measures more sensitive to subtle changes in the mix of votes and stable measures that potentially obscure the microstructure of the spatial map. Poole, Spatial Maps, 185. Researchers should instead be attentive to the particular dangers of each and exercise the appropriate caution in their interpretations.
46. It is important to note that this might call into question whether sovereignty constitutes a coherent issue category, or whether the change in the agenda leads to different dimensions being compared. If so, then the ideal points will not be directly comparable. We are less concerned with this possibility, for a few reasons. For one, while it is certainly possible that by aggregating the votes into the tier 1 category of sovereignty we are effectively comparing two noncomparable dimensions rather than identifying changes in members’ preferences, this measure allows us to clearly identify when this occurred and, by extension, on which votes and policies. Our confidence in the scores as being truly comparable measures of preferences might be reduced, but the gain will be a much more pinpointed identification of when this dimensional divergence began to occur. Second, both before and after 1942—and before and after the New and Fair Deals—member preferences on the different categories aggregated under sovereignty policy show a consistent relationship. The change that we see is the result primarily of northern Democrats, and to a lesser extent Republicans. The positions of southern Democrats—both substantively and in terms of their estimated positions—remain consistent across this period. This suggests that nonsouthern preference change rather than changed dimensionality—on the set of votes aggregated under sovereignty policy—is responsible for the change that becomes evident around 1942.
47. The likeness scores for liberty votes were initially quite high across the three different factions, at 69, 73, and 72 for southern Democrats/northern Democrats, southern Democrats/GOP, and northern Democrats/GOP, respectively, for the pre-1942 period. After 1942, intra-Democratic similarity declined to 69, while the southern Democrat and Republican similarity score increased to 90. The likeness score between the northern and seventeen-state Democratic Party on civil rights votes was 34, compared with 86 on other votes; the score for northern Democrats and the Republican Party was 61, as compared to 47 on non–civil rights votes.
48. This is supported by the qualitative and historical literature as well, which identifies 1937–38 as the period in which a conservative coalition came into being, with southern Democrats being the most important faction of Democrats to oppose the administration. James T. Patterson, Congressional Conservatism and the New Deal: The Growth of the Conservative Coalition in Congress, 1933–1939 (Lexington: University of Kentucky Press, 1967); Susan Dunn, Roosevelt's Purge: How FDR Fought to Change the Democratic Party (Cambridge, MA: Belknap Press, 2010).
49. The dates at which points these shifts occurred are approximate, as the fifty-roll-call interval can cover successive congresses. For instance, the northern Democratic and eleven-state South medians on planning and resources cross during an interval centered on July 18, 1939, but included roll calls from June 20, 1936, through June 25, 1942. The average span of time included varies across issue area, from 1.6 congresses in the issue area of political economy to four congresses in agriculture, with an overall average of two congresses. This limits our ability to pinpoint the timing of change, although this limitation is true also for measures such as DW-NOMINATE, which estimate scores across a member's entire career, which are generated for an individual Congress, and which are constrained to move in a straight line between congresses. The advantage of more temporally fine-grained measures is that they have clear bounds, which allows for the content of the agenda to be inspected and compared. And an advantage of GLS is that it maintains the rank ordering of members for each interval.
50. On their own, likeness scores cannot tell us whether southern preferences were changing, although we have evidence that this was occurring for tier 1 and tier 2 issue areas. But they do allow us to identify the specific votes and policy measures on which a conservative coalition began to appear, and thus provide a guide for identifying and more closely exploring the substantive policies and distinctive set of regional concerns underlying this coalition.
51. We inspected the likeness scores for all tier 3 issue areas, and selected these as showing the greatest divergence from a pattern of greater intrapartisan likeness than cross-partisan likeness. These, we suggest, are the most likely candidates for being issues on which a conservative coalition began to emerge.
52. The similarity between Republicans and southern Democrats on immigration predates the New Deal, and from the 67th to the 72nd Congress the GOP/southern Democratic likeness score was 62, against 72 for the northern Democrats and GOP and 55 for the two wings of the Democratic Party. From the 73rd Congress on, the respective similarity scores would be 82, 60, and 52.
53. Maury Maverick, Democrat from Texas and staunch New Dealer, opposed the creation of the committee, arguing that it would “be the greatest fishing expedition that Congress ever undertook in the history of the United States,” and would inflame prejudice and enable “the Republicans on the committee [to] investigate Mr. Roosevelt and the Democrats [to] investigate Mr. Landon, Mr. Browder, Mr. Thomas—everybody who ran against Roosevelt.” Maverick, Congressional Record, May 26, 1938, 7576. The painting of Plymouth Rock red (or pink, depending on the source) occurred on May Day 1938. The Communist Party denied involvement. “Plymouth Rock Gets Red Paint Covering,” Lewiston Daily Sun, May 9, 1938, 12.
54. Forty-five percent of southern Democrats (52% of eleven-state representatives) voted, with 89 percent of Republicans and against 15 percent of nonsouthern Democrats in favor of investigating the “sit-down” strikes. While 97 percent of northern Democrats supported the Fair Labor Standards Act of 1938, only 59 percent of southern Democrats and 43 percent of eleven-state representatives did so, against 48 percent of Republicans who opposed it when it passed the House. Southern support had increased, and Republican support declined, when it came to the conference report.
55. The conservative coalition did more than simply vote with Republicans on the floor. For instance, Schickler and Pearson find that the Rules Committee, during much of the period examined here, provided support for several conservative coalitions that were opposed by the Democratic Party leadership. In addition to holding back liberal legislation—which might have resulted in even more southern opposition in roll call voting—the Rules Committee frequently advanced conservative legislation either that had not been reported by the responsible committee or that was attached to “must-pass” legislation. Many of these in turn did not come up for a roll call or saw conservative priorities advanced on legislation supported by most liberals on account of it being “must-pass.” Schickler, Eric and Pearson, Kathryn, “Agenda Control, Majority Party Power, and the House Committee on Rules, 1937–1952,” Legislative Studies Quarterly 34 (2009): 455–91Google Scholar.
56. George Brown Tindall, The Emergence of the New South, 1913–1945 (Baton Rouge: Louisiana State University Press, 1967), 602; Samuel, Howard D., “Troubled Passage: The Labor Movement and the Fair Labor Standards Act,” Monthly Labor Review 123 (2000): 32–37 Google Scholar, 36.
57. While a majority of southern representatives accepted the conference committee report (60% versus 96% of nonsouthern Democrats and 52% of Republicans), they had secured a considerable loosening of the standards and coverage of the bill. On passage of the Fair Labor Standards Act, see Grossman, Jonathan, “Fair Labor Standard Acts of 1938: Maximum Struggle for a Minimum Wage,” Monthly Labor Review 101 (1978): 22–30 Google Scholar, 22. The public opinion poll, reported in Grossman, was published in the New York Times on February 16, 1938. According to Grossman, about 90 percent of the jobs lost in the first year of the FLSA's operation were in southern industries such as bagging, pecan shelling, and tobacco stemming. Grossman, “Fair Labor Standards Act,” 28.
58. Katznelson, Fear Itself, 330; August Raymond Ogden, The Dies Committee: A Study of the Special House Committee for the Investigation of Un-American Activities, 1938–1943 (Washington, DC: Catholic University of America Press, 1945); Nancy Lynn Lopez, “Allowing Fears to Overwhelm Us: A Re-Examination of the House Special Committee on Un-American Activities” (Ph.D. diss., Rice University, 2002).
59. Roosevelt vetoed the restriction on strikes at war factories, although this was quickly overridden, with 90 percent of southern Democrats and 78 percent of Republicans voting against 65 percent of nonsouthern Democrats.
60. Harvard Sitkoff, A New Deal For Blacks: The Emergence of Civil Rights as a National Issue, vol. 1 of The Depression Decade (New York: Oxford University Press, 1978), 103–4.
61. Caughey and Schickler, “Public Opinion, Organized Labor.”
62. Philip S. Foner, Organized Labor & the Black Worker, 1619–1973. (New York: International Publishers, 1974), 228.
63. Michael Honey, “Industrial Unionism and Racial Justice in Memphis,” in Organized Labor in the Twentieth-Century South}, ed. Robert Zieger (Knoxville: University of Tennessee Press, 1991), 138; Sitkoff, A New Deal for Blacks, 105; Robert Rodgers Korstad, Civil Rights Unionism: Tobacco Workers and the Struggle for Democracy in the Mid-Twentieth-Century South (Chapel Hill: University of North Carolina Press, 2003).
64. These numbers changed relatively little over time. There were thirty-two representatives from the Black Belt in the 73rd Congress, and thirty-four in the 81st; forty-five Democratic representatives from the six non-Confederate states (none of which met the threshold of more than 35% black) in the 73rd, and thirty-seven by the 81st; and seventy-three and seventy Democratic representatives of the eleven-state South outside the Black Belt in the 73rd and 81st Congresses, respectively.
65. The analyses of marginal effects are performed on ideal points estimated across adjacent congresses, in order to allow us to examine tier 3 as well as tier 2 preferences.
66. The marginal effects were calculated from a regression that estimated the effect on a member's issue-specific ideal point associated with being one of four different categories of Democrat: a nonsouthern Democrat (the intercept), a southern Democrat from the six states of the full region, a southern Democrat from outside the Black Belt but inside the eleven states of the former Confederacy, or a southern Democrat from within the Black Belt. There were no Black Belt districts—districts where at least 35 percent of the population was African American—outside the eleven-state South. It is necessary to account for both measurement uncertainty and sampling variability in the ideal points. This problem can be addressed using standard multiple imputation approaches. We sampled twenty iterations from the posterior distribution, ran the analysis on each, and then combined the estimates, taking into account the within and between imputation variation. The effect was ultimately to slightly increase the confidence intervals. We would like to thank an anonymous reviewer for this suggestion.
67. The poll tax and soldier voting bills saw some southerners from outside the Black Belt support changes to a voting regime that disfranchised many whites along with blacks. The only eleven-state South- representative who might qualify as racially liberal by the tier 2 estimates was Tennessee Republican Dayton Phillips, elected from the first congressional district—held almost continuously by the Republican Party since 1859, and located in the eastern portion of the state where unionism, antislavery, and support for free black voting rights had found support in the antebellum and Civil War eras. He was located one-third standard deviations to left of the chamber median. The most “liberal” eleven-state South- Democrat was North Carolinian Representative Charles Deane, elected to the eighth congressional district, a district whose black community constituted approximately 25 percent of the population. He was located approximately two-thirds to the right of the chamber median, but nearly a full standard deviation away from the South-eleven median. Deane would be defeated in the Democratic primary in 1957 for refusing to sign the Southern Manifesto. He would be honored for this by a Joint Resolution of the General Assembly of North Carolina on July 17, 2003. Resolution 2003–25/House Joint Resolution 231.
68. We sampled legislators’ ideal points from the joint posterior distribution, ranked the legislators by their ideal point, identified which member was in the pivotal position, and repeated this eight hundred times, reporting the proportion of times that a set of legislators was pivotal. See Clinton, Jackman, and Rivers, “Statistical Analysis of Roll Call Data,” 360.
69. These are kernel densities, with fifty estimates generated for each subgroup.
70. Separating out the Black Belt shows that these representatives were distributed even further from their nonsouthern copartisans than the non–Black Belt southerners. This was especially true on the tier 2 issue areas of political economy and planning and resources, as well as the tier 3 issue area of labor markets and unions.
71. Graves, John Temple, “Revolution in the South,” The Virginia Quarterly Review (1950): 190 Google Scholar. It also was the subject of one of the first major behavioral studies of Congress, David Truman's The Congressional Party: A Case Study (New York: Wiley, 1959). His chapter on “Division and Cohesion: The Structure of Party Voting in the House of Representatives” reported findings of particular interest to our extension of the work of Key. Three findings are perhaps most relevant. He found that most of the time, Democrats voted together, but that for low-cohesion votes “the cleavage was essentially sectional.” Truman, Congressional Party, 150. But he also found that the southern faction exhibited far more “shifting and unstable alignments,” ready to enter into coalition on specific questions with the Republicans. And while the South was especially “multinuclear in form,” with “a good deal of fluidity” across issue areas, the nonsouthern Democrats were by the 81st Congress an ever more united and left-oriented party wing that behaved with considerable independence. Truman, Congressional Party, 158–59, 161. Like Key, who along with Robert Dahl and Richard Neustadt is thanked as part of a small group who read and commented on the manuscript, Truman did not have a policy classification with which to work. Issues either appear one by one, as illustrations of the trends in which he was most interested, or are deployed in a very brief treatment of the degree of party cohesion classified by membership in specific substantive House committees.
72. The 81st Congress robustly debated civil rights initiatives, especially an anti–poll tax bill and a controversial effort to create a permanent Fair Employment Practices Committee (FEPC), and extensions to the welfare state. Only one of these, the Housing Act of 1949, became law. But it also produced a number of highly significant enactments concerned with geopolitics, notably the North Atlantic Treaty, the Foreign and Economic Assistance Act of 1950, and the National Security Amendments of 1949.
73. Richard Franklin Bensel, Sectionalism and American Political Development (Madison: University of Wisconsin Press, 1984); Richard Franklin Bensel, The Political Economy of American Industrialization (New York: Cambridge University Press, 2000); Elizabeth Sanders, Roots of Reform: Farmers, Workers, and the American State, 1877–1917 (Chicago: University of Chicago Press, 1999).
74. Truman, Congressional Party, 59.
75. Keith Poole and Howard Rosenthal, Ideology & Congress (New Brunswick, NJ: Transaction Press, 2011), 54.
76. Byron E. Shafer and Richard Johnston, The End of Southern Exceptionalism: Class, Race, and Partisan Change in the Postwar South (Cambridge: Harvard University Press, 2009).
77. Daniel Kryder, Divided Arsenal: Race and the American State During World War II} (New York: Cambridge University Press, 2000); Ira Katznelson, When Affirmative Action Was White: An Untold History of Racial Inequality in Twentieth-Century America (New York: W.W. Norton, 2005).
78. David Brion Davis, Challenging the Boundaries of Slavery (Cambridge, MA: Harvard University Press, 2003), 90.
- 4
- Cited by