A revised version of this paper with Steven A. Kohnke as a third author, was published in The American Review of Politics, vol. 21, Summer 2000

PARTY REALIGNMENT IN THE SOUTH:

A MULTI-LEVEL ANALYSIS

Stephen D. Shaffer

Stacie Berry

Mississippi State University

Prepared for presentation at the annual meeting of the Southern Political Science Association, Savannah, Georgia, Marriott Riverfront Hotel, November 3-6, 1999.

Dramatic change has transformed the South from the V.O. Key (1949) days of a one-party Democratic bastion to a region of dynamic two-party competition for an increasing number of elective offices. Lamis (1990) detailed the increasing two-party nature of southern presidential, U.S. senate, and gubernatorial elections in the 1970s and 1980s in The Two-Party South, and his team of political scientists and journalists extended the analysis through the mid-1990s. In Southern Politics in the 1990s, Lamis (1999) offers the Sundquist (1973) thesis of delayed realignment to account for the gradual nature of GOP gains in southern elective offices below the presidency.

Our paper builds on the exceptional foundation of Lamis' team of scholars and practitioners to address the following issues:

1) Can a theoretical model derived from political science voting behavior studies account for the increasing two-party nature of contests for higher level, more visible offices in the South, such as governor and U.S. senate? Specifically, to what extent do such forces as divisive primaries, candidate ideology, dissatisfaction and scandals, personal attributes such as experience and name visibility, and national forces explain the outcomes of these races in the modern South?

2) Can a related theoretical model, centered around the concept of party identification, help explain GOP growth in less visible offices, such as the state legislature and sub-gubernatorial statewide offices? Can such a model also explain why GOP gains occur at different time points in different states?

3) Do powerful personalities affect the rate of delayed realignment? For example, have dominant Democratic governors leading ideologically and racially unified coalitions retarded the growth of Republicanism in some of the southern states?

4) Finally, can a comprehensive model of delayed realignment in the South provide insight into the future of party politics in that region, particularly for less visible offices such as the state legislature and sub-gubernatorial statewide offices?

LITERATURE REVIEW

Southern politics scholars are quite familiar with the literature of the field, so we briefly acknowledge important works that have guided our study. In identifying the key factors that accounted for the outcomes of U.S. senate and gubernatorial elections in the South over a thirty year time frame, we relied on the two Lamis (1990, 1999) books, the first an expanded version of his dissertation which was a co-winner of the V.O. Key Book Award, and the second an edited volume with each state chapter authored by the most prominent columnist and expert political scientists of each state. Sundquist's (1973) discussion of how the New Deal realignment of the 1930s occurred at later times in non-presidential elections and at different rates in different states is very pertinent to our study, and utilized by Lamis in his concluding chapter as well. One theoretical and empirical effort to examine how political offices differed in their importance and visibility, and how simple vote cues such as party identification might exert greater effects in less visible offices is found in a Shaffer (1982) article drawn from his dissertation. The most comprehensive study of changes in the public's partisan identifications in each southern state is provided in a co-authored book by Swansbrough and Brodsky (1988). Exceptional dissertations, published as books which examine Republican gains in the South and Democratic attempts to blunt those gains, are provided by Aistrup (1996) and Glaser (1996). Finally, we are indebted to the research team of Steed, Moreland, and Baker at the Citadel, whose books on presidential elections in the South have provided vital information about each state.

METHODOLOGY

The sources of our electoral offices information in tables 1-11 are indicated below each table and are readily available. Except for the presidential race, we relied on measures of a party actually winning an office, rather than on popular vote totals. This was due to the importance of an electoral victory to a party's ability to use patronage to attract ambitious rising politicians to their banner. Popular vote totals for governor and U.S. senate were obtained from Hadley and Bowman (1995) and included in our data set for future multivariate analysis. Party identification information was not obtained for every year for each state, suggesting the need for more financial support for such projects as the Network of State Polls at the University of North Carolina. In the multivariate analyses, we provided estimates for all missing years by a process of extrapolation which compared states with other southern states that they were most politically similar to. Additionally, we relied on partisanship data from the Comparative States Election Project of 1968 (Kovenock, Prothro, and Associates, 1973). The partisan composition of state legislatures was derived from two reference books, compared for each state with tables in Hadley and Bowman (1995), and errors were corrected. The correlations between these different sources of state legislative information exceeded .99 in values.

Very little research has been directed toward sub-gubernatorial statewide offices, a problem that has begun to be addressed by Lamis' (1999) team of researchers. This promises to be a rich area for future research, since all except three southern states elect at least six executive offices below governor on a statewide basis. The exceptions are Texas which has five sub-gubernatorial elective offices, Virginia with only lieutenant governor and attorney general (the governor has a cabinet), and Tennessee which elects none (the governor has a cabinet, and the lieutenant governor is chosen by the state senate). The success of state parties at inclusiveness is illustrated by the Georgia Democrats who currently have an elected African-American attorney general (first appointed by a Democratic governor) and labor commissioner, North Carolina Democrats' election of a black auditor, and the Republican governors' of Tennessee and Virginia appointments of two African-Americans to head important agencies (Tennessee- Corrections and Human Services; Virginia- Education and Health-Human Services). Information on sub-gubernatorial offices over the past thirty years was derived from state agency websites accessed through the Citadel's Southern Politics website (http://users.aol.com/thecitadel/sopol/) and obtained from political science scholars in each state.

Our analyses begin in 1970, because we are interested in the more modern period of two-party development in the South, because most Republican gains prior to 1970 were confined to the presidential level, and because a large body of literature exists about the pre-1970 era. It is also important to note that the 1972 presidential election marked the modern period of southern politics, as it constituted the first election in which the Deep and Rim South states behaved in a similar manner, both voting for the same party for president (Republican Nixon). Nevertheless, we find the Rim vs. Deep South dichotomy to remain theoretically interesting, as realignment was especially delayed in certain Deep South states. We now turn to a brief examination of GOP gains in each southern state, with states ranked by how quickly and how thoroughly the gains occurred.

THE GROWTH OF REPUBLICANISM IN THE MODERN SOUTH

Rim South states, except for Arkansas, exhibited competitive party politics for the U.S. senate as early as 1970, and usually evidenced significant competition for governor and the U.S. house as well. Yet until the 1980s, Republicans comprised less than 40% of those who identified with the two major parties in each Rim and Deep South state. Hence, GOP victories in senate and gubernatorial contests reflected their advantage on such short-term forces as candidate personal qualities like experience or name visibility (often enhanced by a previous election bid), performance (such as scandals or other public dissatisfactions associated with their opponent), divisive Democratic primaries, ideologically extreme opponents, and favorable national tides.

Tennessee

In Tennessee, both Baker and Brock were elected to the senate in 1966 and 1970 after divisive primaries hurt their Democratic opponents; Baker was advantaged from a previous race for the senate, while Brock successfully tagged Democratic incumbent Gore as a liberal. Democrat Sasser, running as a moderate and benefitted by Carter's presidential coattails, re-gained Brock's seat in 1976. Republicans scored a breakthrough for governor in 1970 with Winfield Dunn's election, after he attacked his opponent as a "liberal" and benefitted from the Democrat's poor business judgment. Blanton regained the governorship for the Democrats in 1974, benefitting from a previous statewide race (senate in 1972), a national recession during a Republican presidency, and a populist image. Blanton's conviction while in office and a bitter Democratic primary elevated Republican Lamar Alexander, who benefitted from his previous campaign against Blanton and from a folksy image with his walk across the state, to two terms as governor beginning in 1978.

By 1984 Republicans had closed the partisan identification gap in Tennessee to such an extent that short-term forces would exert an even more decisive impact over the vote. Democrat Al Gore Jr., benefitting from his father's name, a united party, a divided GOP, and an incumbency-style campaign, downed an inexperienced opponent to replace Baker in the senate. In 1986 Democrat Ned McWherter, an experienced state house speaker sporting a folksy conservative image that played well in rural areas, began a two-term stint as governor. Republican lawmakers in 1991 and 1992 successfully torpedoed the governor's income tax proposal (for education). Republicans scored a triple play in the 1994 year of voter dissatisfaction with Washington. A well-funded Frisk backed by groups alienated by Sasser, targeted his Democratic opponent as a liberal and Washington insider. The actor Thompson, sporting a red pickup truck, campaigned against Congressman Cooper, the professional politician, who won the Democratic nomination for Gore's old seat. In addition to these two GOP senate pickups, 12-year Republican congressman Sundquist won the governorship after a bitter Democratic primary and the nomination of a business-oriented Democrat who alienated liberals in the party (table 1).

In the less visible offices of U.S. house and state legislature, a competitive two-party situation during the Nixon years reverted back to a Democratic dominance of over sixty percent until 1988, at which time the GOP made staggered gains. In 1988 Republicans began to generally occupy 40% of the state house seats, in 1990 the GOP consistently held over 40% of the state senate seats, and in the 1994 landslide year the party gained a majority of U.S. house seats which

persists today. It is likely that the GOP closing of the party identification gap by 1984 finally translated into the party's gains in these offices that are less visible to voters.

Florida

The Rim South state of Florida also shows a trickle-down effect of party realignment, with two-party competition emerging at different time points depending on the visibility of the office. Florida backed Republicans in the 1952-1960 presidential elections, but not until the mid-1960s was the GOP competitive in gubernatorial and U.S. senate elections, benefitting from short-term forces (table 2). Republican Claude Kirk won the governorship in 1966, benefitting from a divisive Democratic primary and public dissatisfaction with national conditions that produced GOP congressional gains nationally. In 1968 Republican Ed Gurney won a senate seat after a divisive Democratic primary and after branding his opponent as a "liberal." Senate and gubernatorial elections have remained fairly competitive since then. One senate seat saw an ideologically-inclusive Democrat Richard Stone beating Gurney in 1974, losing to Republican Paula Hawkins in 1980, who was herself beaten by current senator Bob Graham (a two-term governor). The other seat was won by Democrat Lawton Chiles in 1970, a member of the majority Democratic party, after a bruising Republican primary and Chiles' walk-across-the-state. Chiles was re-elected twice, and replaced by Republican Connie Mack in 1988, a telegenic candidate benefitting from a fund-raising advantage, a popular presidential nominee for Florida, and a divisive Democratic primary. While contested, gubernatorial races were won by Democrats until 1986, when a two-term Tampa mayor and Catholic, Bob Martinez, was elected. Veteran senator Lawton Chiles succeeded him in 1990, and was narrowly re-elected over Jeb Bush in 1994. A bitterly divided Democratic party, plagued by defecting blacks after white Democratic lawmakers unseated the key African-American legislative leader, handed the 1998 governorship to Republican Jeb Bush, who pursued a racially-inclusive campaign.

Once again, party identification of the populace appears to be a key force in shaping party competition for elective offices. The year 1982 saw Republicans reach 45% of the two-party adherents, while the GOP broke the half-way mark only four years later. By 1986 gubernatorial races had become even more competitive than in previous years. Two years later Republicans gained a majority of the U.S. house delegation, a majority that they have continually retained. The trickle-down effect of partisan gains to less visible offices was evident in the legislature and sub-gubernatorial executive offices as well, where the GOP made historic gains after their gains in the public's partisan ties. The GOP broke the 40% mark in legislative seats in 1988 in the upper chamber and in the lower chamber in 1992. Among the six state-wide elected offices (omitting lieutenant governor, who runs on the ticket with the governor), Republicans won their first office in 1986 with Treasurer-Insurance Commissioner Tom Gallagher, picked up the Secretary of State (Jim Smith) when he switched parties in 1988 and was re-elected in 1990 (along with Gallagher), and by 1994 were winning half of these executive offices (Today, Republicans control secretary of state, comptroller, and education commissioner, while Democrats hold the attorney general, treasurer/insurance commissioner, and agriculture commissioner.).

Virginia

Virginia is yet another Rim South state that voted Republican for president as early as the 1950s and 1960. Republicans won their first senate seat in 1972 with U.S. Representative William Scott, who benefitted from national GOP support that financed his blasting of the incumbent Democrat as a liberal. Senate races have been competitive since then, as the parties have tended to split the two seats. Republicans won their first governorship in 1969, as liberals seized control of the Democratic party to such an extent that some Democratic officeholders began running as independents (table 3). By 1981 the state Democrats had moved back to the ideological center and regained control of the governorship and held it for twelve years, after which Republicans regrouped and have elected two GOP governors. Since 1968, when the U.S. House delegation was evenly split between the parties, control for house seats has been very intensely fought.

Republicans reached a level of 43% of the major party identifiers in the Virginia population in 1982, though presumably because of Democratic incumbency the GOP did not become a significant presence in the state legislature until the early 1990s. The GOP gained 45% or more of the legislative seats in 1992 for the senate and 1994 for the house. Only two sub-gubernatorial offices are elected statewide in Virginia-- lieutenant governor and attorney general. Republicans held one of these offices during much of the 1970s when their party held the governorship, but then lost both offices during the time of Democratic gubernatorial control. In the recent era of Republican governors, one or both of the offices have gone to the GOP. Virginia's subgubernatorial electoral situation differs from those of other states. With only two offices contested, both offices become more visible to voters than in states with six or seven contests, making it more likely that voters will be affected by short-term forces rather than dependent on the low-information cue of party identification. Subgubernatorial elections also appear to be influenced in Virginia by the parties' fortunes in the governor's race.

North Carolina

The state of North Carolina has followed a similar trickle-down, delayed realignment pattern as have other Rim South states. While Republicans did not win the state for president until 1968, the next presidential year they not only helped re-elect Nixon but also elected Republicans to the Senate and governorship for the first time in modern history, marking the start of intense partisan competition for both of these highly visible offices. While Republican Helms has continued to hold one of the state's senate seats, the other has switched back and forth between the parties (table 4). Gubernatorial elections have been equally competitive with Republican James Holshouser elected in 1972 with the help of Nixon's coattails, the moderate lieutenant governor Democrat James Hunt elected in 1976 and re-elected, Republican James Martin serving from 1984 thru 1992, and a moderate, pro-business Hunt returning for another two-term stint in 1992.

Republicans passed the 40% mark among those identifying with one of the major parties in 1990, later than in the other Rim South states, and the GOP didn't achieve at least a majority of U.S. house seats until 1994. However, the GOP landslide year of 1994 was also a breakthrough year in state legislative seats, where for the first time the party has been able to maintain control of at least 40% of the seats in both chambers. One disappointing puzzle for the GOP has been in the state's nine sub-gubernatorial executive offices, where the party has elected only one person since 1970 (Lieutenant Governor Jim Gardner from 1988-1992 in a year when Republican Martin was re-elected governor). Hunt's domination of state government with his moderate and inclusive appeal may have helped delay the GOP realignment at this final level of office. The inclusiveness of the Democratic party is reflected in its success in elevating African-American Ralph Campbell as auditor and female Elaine Marshall as secretary of state.

Texas

In the final Rim South state of Texas, Republican Eisenhower was victorious in both 1952 and 1956, but the GOP lost the state in succeeding presidential races until 1972. John Tower's special election to the U.S. Senate in 1961 and his re-election in 1966 marked the start of intense partisan competition for U.S. senate races. Democrats dominated the governor's office until 1978, when millionaire Republican Bill Clements was elected after a divisive Democratic primary. The governorship has rotated since then, with centrist Democrat Mark White elected in 1982, Republican Clements returning to office in 1986, Democrat Ann Richards elected in 1990, and Republican George W. Bush Jr. elected in 1994 and then re-elected easily (table 5).

In party identification terms, Republicans achieved virtual parity with the Democrats in 1990 and have maintained it since then. Once again, GOP gains in less visible offices followed the party's breakthrough in this fundamental political attitude of voters. Republicans achieved over 40% of the state senate in 1992, over 40% of the state house in 1994, and over 40% of the U.S. house in 1996, and have maintained their competitiveness for all three offices. As related in Lamis (1999), the first Republicans elected to the state's five sub-gubernatorial executive offices (six before the 1996 merger of treasurer and comptroller) were in 1990 with Kay Bailey Hutchison as treasurer and Rick Perry as agriculture commissioner. Hutchison's elevation of the U.S. senate caused a dip in GOP strength until Bush's landslide re-election as governor in 1998 swept in an all-GOP slate of executive officers. The GOP presence also reflects significant gender diversity with a female comptroller, agriculture commissioner, and an appointed secretary of state. Texas today joins Virginia as only the second southern state to have ever had an all-GOP slate of sub-gubernatorial elective executive offices, though Virginia's achievement of this feat in 1998 is less impressive in that only two offices below the governor are elected statewide.

South Carolina

Turning to the Deep South state of South Carolina, we find a pattern of Republican partisan gain and economic strength that resembles that of the Rim South states. South Carolina began voting Republican for president in the 1964 Goldwater election, a year that prompted Senator Thurmond to switch to the GOP, producing the even partisan split in senators that persists today. Gubernatorial elections were competitive beginning in 1970 despite Democrat John West's victory. The 1974 election began a cycle of alternating partisan control of governor, with Republican James Edwards elected after a divisive Democratic primary (table 6). Democrat Dick Riley served two four-year terms as governor beginning in 1978, and Republican Carroll Campbell served two terms starting in 1986. Partisan turnover heightened in the 1990s with party switching Republican David Beasley elected in 1994 and Democrat Jim Hodges riding the popular education issue to victory in 1998.

Republican competitiveness heightened beginning in 1980, when the GOP attained the adherence of 45% of major party identifiers in the state population. The same year, a majority of the U.S. house delegation became Republican for the first time in modern history. The GOP made more gradual gains in lower level offices. It was not until the Republicans in 1990 reached a strength of slightly above 50% of partisan identifiers that the party become a significant force in the legislature or sub-gubernatorial executive offices. That year saw the landslide re-election of Republican governor Campbell, and Republican capture of three of the state's eight sub-gubernatorial offices (secretary of state, agriculture commissioner, and education superintendent). The next six years saw new GOP landmarks with control of over 40% of state house seats in 1992, capture of a majority of sub-gubernatorial offices in 1994, and control of over 40% of state senate seats in 1996. South Carolina's current status of intense partisan competition at all levels of government is very similar to that of the Rim South states already discussed. This is a sharp historical change for a state that had voted with its Deep South neighbors for Democratic presidents as late as the 1950s when Rim South states were voting more like the rest of the nation.

Alabama

Alabama is a second Deep South state that has become as intensely competitive as the Rim South states. Backing Democratic presidential candidates as late as the 1950s, Alabama finally voted Republican for president in 1964 and 1972. U.S. senate elections became more competitive beginning in 1980 when Republican war hero Jeremiah Denton was elected with the help of Reagan's coattails (table 7). Democrats gained the seat back in 1986 with conservative Richard Shelby, but lost both seats in the 1990s with Shelby's switch to the GOP and Jeff Session's 52% popular vote victory in 1996. George Wallace's domination of state politics as a convert to the inclusive black-white Democratic coalition-building strategy of New South Democrats held Republican gubernatorial candidates to less than 40% of the popular vote until 1986. Wallace's retirement and the bizarre Democratic primary in which the winner was disqualified because of his encouraging GOP cross-over voting in the runoff primary produced a Republican victory by Primitive Baptist minister Guy Hunt. The volatility of gubernatorial politics since then is reflected in Hunt's re-election in 1990, his replacement by Democratic Lieutenant Governor Jim Folsom in 1993 after the governor's resignation for a criminal conviction, the narrow victory of Democrat-turned-Republican Fob James in 1994 and his defeat four years later by Democratic Lieutenant Governor Don Siegelman, who pledged a lottery for education.

In the long-term force of party identification, Republicans passed the 45% mark of support among Alabama identifiers of the two major parties in 1988. While the GOP had held over 40% of U.S. house seats during the 1970s, it achieved that mark once again in 1992 and four years later had achieved a significant majority of seats. In the 1994 Republican landslide year, GOP representation in the state legislature passed the 30% mark for the first time. Most

impressive in 1994 was that the party won four of the state's six sub-gubernatorial executive offices (auditor, agriculture commissioner, attorney general, and secretary of state), and despite losing the governorship in 1998 continued to hold three statewide offices (winning Siegelman's lieutenant governorship post while losing the auditor and agriculture commissioner's races). For a state that until 1986 had not elected a Republican governor and that until 1994 had not elected a Republican to any statewide office below governor, Alabama politics today is as intensely competitive as most Rim South states, except in state legislative seats where Democrats retain a significant two-to-one advantage.

Georgia

Georgia is a third Deep South state which has recently developed a competitive two-party system at both the state and federal levels. Reflecting Deep South presidential voting patterns, Georgia didn't vote Republican for president until 1964 and 1972. With one exception GOP support in Senate races didn't exceed 40% until the 1980 defection of many African-Americans from Democratic senator Talmadge, which elected Republican Mack Mattingly. Senate elections have been more competitive since then, with Republican Paul Coverdell's and Democrat Max Cleland's election totals during the 1990s held below 53%. Unlike the other southern states, Georgia has still failed to elect a Republican as governor, though beginning in 1990 the GOP vote total has finally remained significantly above 40% (table 8). Lamis (1990) credits Democratic governors such as Carter (1970-74) and George Busbee (1974-82) with creating inclusive black-white coalitions by straddling the race issue. A conservative pledging to resist take hikes, Joe Frank Harris was elected governor in 1982 and re-elected in 1986, followed by pro-education Lieutenant Governor Zell Miller who in 1990 began two terms, culminating in long-time Democratic state legislator Roy Barnes (elected on a pro-education platform in 1998).

Regarding the long-term force of party identification, Republicans had achieved 43% or more of the two-party adherents by 1992, whereupon the party began making striking gains in lower level offices. In the 1994 GOP landslide year, the party achieved a first-time majority of U.S. house seats and increased their majority two years later. Another 1994 breakthrough was in sub-gubernatorial executive offices, where Republicans won three of the seven offices (school superintendent, insurance commissioner, and a party switching attorney general), losing the attorney general office in 1998 after the incumbent's unsuccessful bid for the GOP gubernatorial nomination. In 1996 and 1998 the GOP passed the 40% mark in legislative seats in the state house and state senate, respectively. Georgia Democrats have made an impressive effort in maintaining party unity by elevating African-Americans to sub-gubernatorial executive offices elected statewide. African-American legislator Thurbert Baker was appointed attorney general by the Democratic governor after the GOP attorney general's resignation, and then elected to a full term in 1998. African-Americans achieved the Democratic nominations in 1998 for two other offices-labor secretary and insurance commissioner with the latter victorious (Michael Thurmond). The bi-racial Democratic coalition is also evident in the congress, where all three House Democrats are African-American (with Sanford Bishop and Cynthia McKinney easily re-elected in 1998 in redrawn white majority districts) and the Senate Democrat a white male.

Mississippi

Mississippi is one of two Deep South states that has shown the least Republican growth in less visible state offices, despite impressive gains in more visible offices where short-term forces are potent. Beginning in 1972 state voters have backed GOP presidential candidates in every election except 1976, when the South minus Virginia backed fellow southerner Carter. Nixon's 1972 landslide in Mississippi instantly made the GOP competitive in U.S. house races, where Republicans Thad Cochran and Trent Lott were elected (table 9). U.S. senate elections began to become competitive between the two parties in 1978 when Democratic incumbents Eastland and later Stennis retired. A racial split within the Democratic party which produced a black independent candidate helped elect Cochran to the senate in 1978. Cochran's effective use of non-ideological constituency service themes have won him repeated easy re-elections. Trent Lott's massive campaign warchest, spent on visually-attractive television ads that portrayed him as a "progressive," produced in 1988 another GOP senate pickup. Except for one election (contested by progressive businessman Gil Carmichael), Republicans failed to break 40% of the gubernatorial vote until 1987, when progressive GOP businessman (and Board of Education member) Jack Reed held Democrat Ray Mabus to only 53% of the vote. Dissatisfaction with budget cuts during the economic recession saw voters oust Mabus for conservative Republican businessman Kirk Fordice in 1991, who was re-elected four years later during an economic boom. The current campaign between pro-education and racially-inclusive Lieutenant Governor Ronnie Musgrove and Democrat-turned-Republican former Congressman Mike Parker is viewed as a toss-up.

While short-term forces produced GOP victories in presidential and senate races before 1990, it wasn't until the GOP achieved more than 40% of the public's partisan identifications that the party finally broke the 30% mark in state legislative seats. This milestone was achieved between 1996 and 1998 through party switching, the 1995 state elections, and special elections. Another modest GOP milestone was the pickup of the party's first sub-gubernatorial executive office with party switching auditor Pete Johnson, who was beaten by Fordice in 1991 for the gubernatorial nomination. In the face of voter discontent with recession-induced budget cuts, Democrat-turned-Republican lawmaker Eddie Briggs unseated three-term Democratic Lieutenant Governor Brad Dye, but then lost re-election in 1995. The all-Democratic sub-gubernatorial contingent (dubbed "The Magnificent Seven" by Attorney General Mike Moore) lost an auditor to resignation for allegedly lying about a tax matter involving a car tag, whereupon Republican Fordice appointed GOP lawmaker Phil Bryant, who in 1999 was the one Republican running for a sub-gubernatorial office who was favored to win. The continued Democratic domination of the state legislature (as is the case in Alabama), and the failure of the party to win more than one statewide executive office below governor illustrates that a truly competitive two-party system has yet to extend to all levels of Mississippi government. Likely explanations for continued Democratic dominance involve a Republican party that fails to contest many elections (The GOP offered fewer legislative candidates in 1999 than in 1995, when their "throw-the-bums-out" campaign fizzled.), GOP candidates that may be too conservative for many voters, and an ideologically eclectic Democratic party that backs popular issues such as education and tough-on-crime issues (Shaffer and Johnson, 1996).

Louisiana

Perhaps the most unique Deep South state has been Louisiana, where Republicans have yet to achieve 40% of the public's partisan identifications. The GOP became competitive in presidential elections in the 1950s and 1960s, and by being the only southern state besides Arkansas and Tennessee to back Clinton twice has resisted other states' move towards Republican dominance at this level. While the GOP has never elected a senator in modern times, elections have been closer since 1986. A state GOP effort to promote "ballot security" by purging non-voters in black precincts in 1986 helped elect Democrat John Breaux by a narrow 53% vote total, and treasurer Mary Landrieu kept the other seat Democratic in 1996 by only four-tenths of one percent of the vote. The election of reformist Republican Dave Treen in 1979, after Democratic power Edwin Edwards was barred from a third term, marked the start of a more competitive era in gubernatorial politics. Edwards regained the governorship by easily defeating Treen in 1983. A three-way race in 1987 produced a win for conservative Democrat Buddy Roemer, who switched to the GOP for an unsuccessful re-election bid that instead produced the bizarre 1991 runoff between Edwin Edwards and defrocked Republican David Duke. Facing a liberal African-American Democrat, Republicans regained the governorship in 1995 with party switcher Mike Foster, who remains politically popular today (table 10).

Despite continued Democratic advantage in state partisanship, Republicans have become quite competitive in U.S. house races, winning half of Louisiana's seats in 1988, and controlling five of the seven seats since 1996. Republicans registered some legislative gains after Foster's gubernatorial victory, but still retained control of only about one-third of the state senate seats and less than 30% of house seats. Republicans gained only one of the state's seven subgubernatorial executive offices in 1987 with the election of lieutenant governor Paul Hardy and briefly held the secretary of state's office as well after Fox McKeithen's party switch. However, in the wake of the Duke disaster the GOP lost all statewide races in 1991 except secretary of state (McKeithen remained the only sub-gubernatorial Republican after his re-election in 1995). Louisiana hence remains in the same class with Mississippi as a state where Democrats remain dominant in non-gubernatorial state elections, though the explanation may be as simple as the continued Democratic advantage in the public's partisan identifications.

Arkansas

Arkansas is the lone Rim South state that still remains very Democratic in less visible elective offices and in party identification. In presidential races Arkansas has sometimes behaved like a Deep South state, sticking with Democrat Stevenson in the 1950s while many Rim South states defected, and being the only non-Deep South state backing Wallace in 1968. While other southern states have trended Republican in presidential elections since 1972, Arkansas has split between the parties. Barth, Blair, and Dumas (1999) in the Lamis book argue that Republican growth was stymied by three Democratic officials who dominated the state's politics through their charisma, personal politicking skills, and organizations. Dale Bumpers served as governor from 1970 through 1974, and as U.S. Senator from 1974 through 1998; David Pryor was governor from 1974 through 1978, and U.S. Senator from 1978 through 1996; and Clinton

was governor from 1978 through 1992 (except for two years). Except for the gubernatorial victories of Winthrop Rockefeller in 1966 and 1968, and the victory and subsequent defeat of gubernatorial hopeful Frank White in 1980 and 1982, Republicans never broke 45% of the popular vote for either governor or the U.S. senate until the 1990s (table 11). The departure of the "Big Three" from Arkansas politics has clearly produced a more competitive situation for these highly visible offices. While Democrat Jim Guy Tucker was easily elected governor in 1994 after his ascension to the governorship upon Clinton's election as president, his resignation upon a felony conviction in 1996 elevated Republican lieutenant governor Mike Huckabee to the state's top office, who won easy re-election in 1998. Meanwhile, Republican congressman Tim Hutchison narrowly won Pryor's vacant Senate seat in 1996, though congresswoman Blanche Lincoln more easily retained Bumpers' Senate seat for the Democrats in 1998.

While polling data for Arkansas are unavailable between 1986 and 1999 (the political science department at the University of Arkansas conducted the state's first academic poll in 1999), residents were overwhelmingly Democratic up until the late 1980s, and remain largely Democratic even today. Republicans have failed to ever capture more than 23% of either state legislative chamber's seats. Prior to 1993 the GOP had won only one of the state's six subgubernatorial executive offices-a lieutenant governor elected and re-elected with Governor Rockefeller in the 1960s. As already noted, Republican Mike Huckabee, a former president of the state's Southern Baptist convention and the party's unsuccessful 1992 U.S. senate nominee, won the 1993 special election for lieutenant governor and was easily re-elected in 1994. Upon Huckabee's elevation to governor, Winthrop Paul Rockefeller, the son of the former GOP governor, won the lieutenant governorship, and gained re-election in 1998. Despite the GOP's retention of one sub-gubernatorial executive office since 1994, Democrats have retained all five other state offices.

A MULTIVARIATE EXPLORATION OF GOP GAINS ACROSS ELECTIVE OFFICES

The principal purpose of this exploratory paper was to mathematically document the extent to which the GOP has made gains at numerous levels of federal and state offices in each of the eleven southern states, and the time sequence when gains were most evident. Other key purposes of the paper were to document GOP gains in sub-gubernatorial statewide elected executive offices (a subject so overlooked by researchers that it necessitated communication with scholars in a number of the states to obtain historical information), and to identify changes in partisan identification in each of the states, another subject overlooked after the Swansbrough and Brodsky (1988) book. We now turn to an exploratory analysis of several factors that may help to explain GOP gains in the South, and to explain the different sequence of the gains in different states.

The dependent variables are the GOP proportion of the state senate seats, state house seats, and sub-gubernatorial elective executive offices. Four key factors that will be examined as correlates of state legislative seats and sub-gubernatorial offices are party identification, the presidential vote, the governor's party, and popular Democratic figures. Party identification is operationalized as the GOP percentage of the major party identifiers in the state. Where data are not available, party identification is estimated from the time series trend of party identification for that state, after comparing the pattern with neighboring states that are politically similar. The presidential coattail effect is measured with the GOP proportion of the two-party vote for president. In midterm years, the previous presidential year's vote is employed, though future studies may wish to include another theoretically important variable measuring the president's job rating in the state, assuming that the data could be found for each state and each midterm year. The party of the governor (in non-gubernatorial election years) or of the victorious gubernatorial candidate helps measure the possible coattail effects of the office of the governor.

Popular Democratic figures measures the concept popularized by Lamis (1990) and Barth, Blair, and Duman (1999) that some governors (and senators) may be so dominant that they retard the growing political prospects of the other party. In other words, they hypothesize that there are political figures at the state level that may have as much of an impact on state politics as national figures such as Franklin D. Roosevelt had on national politics. Reviewing the Lamis (1990, 1999) books led to a variable that ranged from -3 to +3, with -3 being a dominant Democratic figure that actually hurt his party (the liberal Howell in Virginia from 1970 through 1980, inclusive), +3 being a dominant figure that helped his party, and 0 being no dominant figure. The +3 was assigned to: Alabama from 1970 thru 1984 for the Wallace influence; and Louisiana from 1970 through 1994 for the Edwin Edwards influence. Barth, Blair, and Duman (1999) argue that Arkansas has three dominant Democratic figures, so the Arkansas scores measured how many of those figures were on the political scene. North Carolina politics was shaped by two powerful figures of different parties-Helms and Hunt- so that state received a score of 0 after assuming that the two influences canceled each other out. Other states appeared to lack dominant figures, and therefore received scores of 0.

We employed a pooled, time-series, cross-sectional research design (Johnson, 1984: 396-407). The unit of analysis is the state-year (each line in tables 1-11). Regarding possible methodological problems, multicollinearity was ruled out by examining correlation matrices for each analysis. Correlations among predictors never exceeded .36 (for party identification and party of the governor). Auto-correlation did not appear to be a factor, as the sample was divided into presidential and midterm years and separate analyses were conducted for each. The results were very similar to the analysis including all years. Heteroscedasticity was identified by plotting the regression residuals by each of the predictors, and also by calculating the variance of the residuals for the categories of each of the predictors. In some cases heteroscedasticity may have been a factor, as residual variance appeared to be highest in states lacking dominant Democratic figures and in states and years where the party identification gap between the parties had disappeared. These results suggest some improvement in the specification of the model may be required, a task that a future paper shall more completely address.

The results of testing these models for state senate and state house seats are very similar. In both instances, the most important of the four predictors is the party identification of the state population (table 12). In other words, as the Republicans closed the gap with the Democrats in terms of the public's partisan identifications in nearly all of the southern states, the Republicans gained a greater share of seats in both chambers of the state legislatures. One is tempted to subtitle this paper after a modification of the Clinton campaign slogan of 1992: "It's party identification, stupid!" The delayed realignment at the state legislative level, compared to the earlier GOP gains in more prominent elections such as president and U.S. senate, is quite understandable. Prominent offices elected on a statewide or national basis are more visible to voters, so they learn more about the candidates and their relationships to issues. Therefore short-term forces have a greater chance of counteracting the long-term force of party identification than in less visible offices. Hence, the GOP was able to win presidential, senate, and gubernatorial races in many states at an earlier time point, than they were able to win less visible contests such as for state legislative seats. While winning a senate seat required the GOP to recruit only one high quality candidate, winning control of a state legislative chamber might require the recruitment of sixty times that number of high quality candidates. Not until the long-term force of party identification shifted in a more Republican direction in each southern state did the GOP make significant inroads in attaining a competitive status in controlling state legislative seats.

The second most important predictor in the model for both state legislative chambers is the presence of a dominant and popular Democratic figure in the state. This factor was negatively related to the GOP presence in each state legislative chamber. In other words, states that possessed dominant Democratic figures such as Wallace in Alabama and Edwards in Louisiana had fewer Republicans in the state legislature, even after taking into account the partisan ties of the state populace. It is interesting that the more journalistic approach of Lamis (1990, 1999) has helped to identify a concept that has not received systematic attention. Much as Franklin D. Roosevelt helped make the Democrats the majority party in the nation in the 1930s, there may be Democratic "giants" at the state level that have retarded the growth of Republicanism in some southern states.

Both of the coattail variables, the presidential coattail measured by the presidential vote in the state and the gubernatorial coattail measured by the party of the governor, were unrelated to the party margin in the state legislature. The unimportance of presidential coattails reflects the puzzle of delayed realignment that we have sought to address-despite GOP strength in presidential elections in the South, the party's gains at the state legislative level have been far more modest. These results are not a function of the Nixon landslide in the South at the start of our time period, when fewer Republicans occupied legislative seats, since similar results are obtained when we repeat the analyses after excluding the first three election years of the 1970s. The unimportance of winning the governorship in terms of increasing a party's share of legislative seats is another interesting finding. This suggests that legislative incumbents may be divorcing themselves from more controversial candidates for executive offices, by employing the same kinds of incumbency-related campaign themes that characterize many congressional races.

Sub-gubernatorial election outcomes produce a somewhat different model (table 12). The party identification of the populace is again the most important factor, but not quite as important as in state legislative election outcomes. This suggests that voters are more knowledgeable about the candidates, or that incumbency or experience is a greater factor in shaping their votes, unlike the two state legislative models. The party of the governor predictor is statistically significant, suggesting a gubernatorial coattail effect for sub-gubernatorial executive offices. In other words, states electing Republican governors or having elected a Republican governor two years earlier are more likely to have a greater proportion of sub-gubernatorial executive offices that are Republican, even when controlling for the party identification makeup of the populace. One caveat, however, is that the adjusted R-squared is lower for the sub-gubernatorial model than for the two state legislative models, suggesting that more theorizing and hypothesis testing is necessary in this area.

CONCLUSIONS

We believe that simple voting behavior models applied to presidential elections and to gubernatorial and senate elections nationally can indeed explain GOP gains in senate and gubernatorial elections in the South since the 1960s. As shown in the test cases of Tennessee and Florida, where we recounted the short-term forces identified by the two Lamis books, candidate personal qualities like experience or name visibility (often enhanced by a previous election bid), performance factors (such as scandals or other public dissatisfactions associated with their opponent), divisive Democratic primaries, ideologically extreme opponents, and favorable national tides helped Republicans to be elected to the senate and governorships of these states. Similar factors were at work in Mississippi and other southern states.

Party identification appears to be a major force shaping GOP gains in less visible offices, such as state legislative and sub-gubernatorial executive offices. Not only is party identification statistically related to the partisan composition of the state legislatures, but in each southern state the GOP contingent of the state legislature reached a critical mass after a significant proportion of the state population had become Republican in party ties (see bold face numbers in each table). The two states where Republicans have failed to achieve at least 30% of the seats in both legislative chambers are also two states where Republicans appeared to be weakest in terms of the public's partisan identifications.

We do believe that states may possess political personalities who are as dominant in their state's political system as were presidents such as Andrew Jackson and Franklin D. Roosevelt dominant at the national level. The Lamis books provide convincing journalistic evidence regarding the presence of such southern governors, and our statistical analysis suggests that such dominant Democratic officials retarded the growth of the GOP in some of the southern states. A possible next step would be to investigate the possibility of this variable being more of an interval variable rather than essentially a dichotomous one (with some modification). On the other hand, the creation of new party systems at the national level in 1828, 1860, 1896, and 1932 suggests that our more compressed indicator has some theoretical and empirical support.

Our paper suggests that the two-party South is here to stay, and that it will likely manifest itself in each of the southern states, and at each electoral office level. As party identification remains or becomes competitive in each state, the GOP should become as competitive in the Deep South state legislatures as it is today in the Rim South states. The departure of prominent Democratic figures from the public scene in Louisiana and Arkansas should eventually translate into greater partisan competition in those most Democratic state legislatures as well. As party identification becomes competitive in each of the states, the short-term forces become dominant. One of those short term forces, as noted by Aistrup (1996), is incumbency, and the heavily Democratic nature of such legislatures as Mississippi's explains why the Mississippi GOP is so strongly backing the term limits referendum that is on this November's ballot. However, even if the Democratic advantage on incumbency was reduced by passage of such a measure, it is nevertheless critical for the state GOP in such states to place a stronger emphasis on fully contesting such lower level offices. If state GOPs fail to contest all districts, as is currently the case in Mississippi, or if they nominate candidates who are too ideologically extreme or parochial for the state's voters, these would be additional factors retarding the ability of the GOP to become fully competitive in less visible offices. Identifying and including indicators of such concepts would undoubtedly provide a more comprehensive model of less visible offices.

REFERENCES

Aistrup, Joseph A. (1996) The Southern Strategy Revisited: Republican Top-Down Advancement in the South. Lexington: The University Press of Kentucky.

Barone, Michael; and Grant Ujifusa (1973-1983) The Almanac of American Politics, 1972-1984. Washington, D.C.: National Journal.

Barth, Jay; Diane D. Blair; and Ernie Dumas. (1999) "Arkansas: Characters, Crises, and Change," in Southern Politics in the 1990s, ed. Alexander P. Lamis. Baton Rouge: Louisiana State University Press.

Bass, Jack; and DeVries, Walter. (1977) The Transformation of Southern Politics: Social Change and Political Consequence Since 1945. New York: Meridian, New American Library.

Congressional Quarterly Staff. (1985-1999) Politics in America, 1986-2000. Washington, D.C.; Congressional Quarterly Inc.

Glaser, James M. (1996) Race, Campaign Politics, & the Realignment in the South. New Haven: Yale University Press.

Hadley, Charles D.; and Lewis Bowman, eds. (1995) Southern State Party Organizations and Activists. Westport, Conn: Praeger.

Johnson, John. (1984) Econometric Methods, 3^rd edition. New York: McGraw-Hill.

Key, V.O. (1949) Southern Politics in State and Nation. New York: Random House.

Kovenock, David M.; Prothro, James W.; and Associates. (1973) Explaining the Vote: Presidential Choices in the Nation and the States, 1988. University of North Carolina at Chapel Hill; Institute for Research in Social Science.

Lamis, Alexander. (1990) The Two-Party South: Second Expanded Edition. New York: Oxford University Press.

Lamis, Alexander, ed.. (1999) Southern Politics in the 1990s. Baton Rouge; Louisiana State University Press.

Moreland, Laurence W.; Robert P. Steed; and Tod A. Baker. (1991) The 1988 Presidential Election in the South: Continuity Amidst Change in Southern Party Politics. New York: Praeger.

Moreland, Laurence W.; and Robert P. Steed. (1997) The 1996 Presidential Election in the South: Southern Party Systems in the 1990s. Westport, Conn; Praeger.

Network of State Polls. (1999) Website address: http://www.irss.unc.edu/nnsp/nnsp.htm. Chapel Hill: University of North Carolina.

Shaffer, Stephen D. (1982) "Voting in Four Elective Offices: A Comparative Analysis," American Politics Quarterly. Vol. 10, no. 1 (January), pp. 5-30.

Shaffer, Stephen; and Monica Johnson. (1996) "A New Solid South? The Drama of Partisan Realignment in the Deep South State of Mississippi," The American Review of Politics, vol. 17, Summer, pp. 171-192.

Steed, Robert P.; Laurence W. Moreland; and Tod A. Baker. (1994) The 1992 Presidential Election in the South: Current Patterns of Southern Party and Electoral Politics. Westport, Conn: Praeger.

Sundquist, James L. (1973) Dynamics of the Party System: Alignment and Realignment of Political Parties in the United States. Washington, D.C.: Brookings Institution.

Swansbrough, Robert H.; and David M. Brodsky (1988) The South's New Politics: Realignment and Dealignment. Columbia, S.C.: University of South Carolina Press.

Wright, Gerald C.; Robert S. Erikson; and John P. McIver. (1985) "Measuring State Partisanship and Ideology with Survey Data," Journal of Politics, vol. 47, no. 2, pp. 469-489.

TABLE 1

REPUBLICAN GROWTH IN TENNESSEE

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; Wright, Erikson, and McIver's "Measuring State Partisanship and Ideology with Survey Data," in May 1985, Journal of Politics; the Network of State Polls at the University of North Carolina.

NA indicates not available or no election held; in Tennessee, no sub-gubernatorial offices are elected statewide.

TABLE 2

REPUBLICAN GROWTH IN FLORIDA

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; Moreland and Steed's The 1996 Presidential Election in the South.

NA indicates not available or no election held.

TABLE 3

REPUBLICAN GROWTH IN VIRGINIA

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; the Network of State Polls at the University of North Carolina; Lamis' The Two-Party South.

NA indicates not available or no election held.

* Virginia statewide state offices are elected in the year after a presidential election; only governor, lieutenant governor and attorney general are popularly elected.

TABLE 4

REPUBLICAN GROWTH IN NORTH CAROLINA

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; Wright, Erikson, and McIver's "Measuring State Partisanship and Ideology with Survey Data," in May 1985, Journal of Politics; the Network of State Polls at the University of North Carolina.

NA indicates not available or no election held.

Note: 1970 and 1972 party identification data are from 1968 and 1971 polls, respectively.

TABLE 5

REPUBLICAN GROWTH IN TEXAS

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Hadley and Bowman's Southern State Party Organizations and Activists; communication from Professor Richard Murray.

NA indicates not available or no election held.

* Republican Hutchison was elected in a special election in 1993.

TABLE 6

REPUBLICAN GROWTH IN SOUTH CAROLINA

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; Wright, Erikson, and McIver's "Measuring State Partisanship and Ideology with Survey Data," in May 1985, Journal of Politics; communication from Professor Glen Broach.

NA indicates not available or no election held.

TABLE 7

REPUBLICAN GROWTH IN ALABAMA

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; Moreland, Steed, and Baker's The 1988 Presidential Election in the South; Steed, Moreland, and Baker's The 1992 Presidential Election in the South; and communication from Professor Patrick Cotter.

NA indicates not available or no election held.

TABLE 8

REPUBLICAN GROWTH IN GEORGIA

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment.

NA indicates not available or no election held.

TABLE 9

REPUBLICAN GROWTH IN MISSISSIPPI

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; Wright, Erikson, and McIver's "Measuring State Partisanship and Ideology with Survey Data," in May 1985, Journal of Politics; Bass and DeVries, The Transformation of Southern Politics.

NA indicates not available or no election held.

* Mississippi statewide state offices are elected in the year before a presidential election.

TABLE 10

REPUBLICAN GROWTH IN LOUISIANA

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; Moreland and Steed's The 1996 Presidential Election in the South; communication from Professor Wayne Parent.

NA indicates not available or no election held.

* Louisiana statewide state offices are elected in the year before a presidential election.

TABLE 11

REPUBLICAN GROWTH IN ARKANSAS

Source: The Almanac of American Politics, 1972-1984; CQ's Politics in America, 1986-2000 for all information except party identification and sub-gubernatorial offices. These data drawn from: Lamis' Southern Politics in the 1990s; Swansbrough and Brodsky's The South's New Politics: Realignment and Dealignment; Wright, Erikson, and McIver's "Measuring State Partisanship and Ideology with Survey Data," in May 1985, Journal of Politics; the Arkansas Poll, Dr. Janine Parry Director, the University of Arkansas.

NA indicates not available or no election held.

TABLE 12

THE CORRELATES OF PARTY MARGINS IN STATE OFFICES

........................................................STATE SENATE................................... STATE HOUSE............................................SUB-GUBERNATORIAL............

* indicates statistical significance at the .05 level.