ó State Influence in the Presidential Selection Process: Assessing Pennsylvania's Non-Participation in the Frontloading Trend James E. Vike Widener University The frontloading trend ill the presidential primary system has been widely criticizedfor creating disparities among states in their level of influence over the presidential selection process. This article examines the general motivations, hel1pfits, fll1d consequences of state participation in frontloading. It focuses upon Pen11sylvania, which has not joined the trend. Presidential primaries i11 Penn­ sylvania have become increasingly ul1competitive due to frol1tloading in. other states. Comparative analysis with Ohio alldforecastillg ~malysis based 01/ Penn­ sylvania voting history suggest that an earlier presidential primary in Pennsyl­ vania in 2000 would have resulted in not only more candidate competitiveness and campaign activity, but also more voter choice and voter turnout. Yet, fear of increased administrative burdens and disruption to the electoral status quo stopped Pennsylvania's legislators fr0111 adoptill:;? the cha/l:;?e. Since the 1968 presidential election cycle, direct primaries have been thp dominant mp;m~ hy which tJlP major polit1c. Frontloading Decision Factors Several motivations drove particular states to move up on the pri­ mary season calendar in 1996 and 2000. The main reason was the desire to increase the state's influence in determining the outcome of the nomi­ nation process. When the winnowing process concludes prior to a state's election date, state voters are given no choice and are excluded from any role in determining the party's nominee 0 ackson and Crotty 2001; Lengle 1997). The lack of voter choice or influence contributes to citizen apathy and alienation from the political system. These burgeoning levels of alien­ ation are strongly linked to diminished interest in political participation (Nye, King, and Zelikow 1997). From the perspective of the individual state,' the lack of choice and influence can be remedied only by a move forward on the primary calendar. Consistent with this reasoning is the suggestion that an earlier primary date is likely to increase voter turnout as candidates spend more time campaigning in the state and voters be­ lieve they will have an impact on determining the party nominee (Ceasar and Busch 1997; Geer 1989; Lengle 1997). Another motivating factor in the decision to move forward on the primary calendar was the belief that 54 an earlier primary or caucus date would bring greater candidate atten­ tion to and media coverage of state interests and issues (Cook 1992; Jack­ son and Crotty 2001; T.f'neip 1997; Mayer 19(7) The key to this rationale is the understanding that candidate campaign strategies and media at­ tention during the primary season tend to be frontloaded as well. A successful campaign strategy must involve a keen focus on the early stage of the primary season (Cook 1992; Jackson and Crotty 2001; Mayer 1997; Wayne 2001). Candidates engage in a retail style of campaigning in the earliest contests by spending considerable time face-to-face with po­ tential supporters. As the campaign proceeds, candidates must convert to a wholesale· approach in order to reach potential supporters in up­ coming primary states that are spread all across the nation (Haskell 2000; Wayne 2001). Given the speed with which the winnowing process oc­ C'llTS, C';::mninates mnst p')(pend resources in early states because there is little to be gained from hoarding money or building organizational sup­ port in later states when continued survival is not assured. Late primaries and caucuses also receive little media coverage, if any, and there is rarely any attention paid to issues distinct or relevant to these state. The lack of attention is highlighted by the common media practice of discontmuing extensive reports of election results from the later states once the nomination has been determined (Lichter 1992; Mayer 1997). The frontloading effort, as it relates to a desire for increased cam­ paign activity and media coverage, is also linked to hopes that an earlier primary date would provide substantial economic benefits as candidates and the media poured dollars into thfl statp (MaYPT '19(7). Individual states considering moving up on the primary calendar must also weigh the negative potential consequences. If the state or party al­ ters the presidential preference primary date, some uncertainty and up­ heaval will inevitably result. Particularly important is the upheaval caused by disruption to the established electoral calendar and processes in the state. The intensity of the upheaval depends on the date of the relocation and the extent to which the primary date was institutionalized in the minds of voters and election officials.3 The upheaval occurs through aJ­ terations to electoral procedures and timelines; the uncertainty emerges from the unknown impact on turnout and outcomes that may arise from separating or pushing forward primary election events. There can also be substantial administrative costs associated with a significant step forward on the calendar. Most states with early presi­ dential preference primaries separate them from statewide office primary elections. For example, of the first 21 GOP presidential preference pri­ maries held in 2000, only three supported joint congressional primary elections. This separation mandates addItIOnal spending to support mul­ tiple primary events. Separating the primaries and frontloading the presi- 55 dentinl event cnn also cnuse cnpncity problems for elections officinls be cause of the shorter time spans for attracting workers, the increased bur­ den on volunteers, and greater conflict with end of the year obligations during filing and recruitment periods (McCarthy 1999; Race 1999). Apart from administrative costs, some states choose to forgo a move to the top of the presidential primary calendar because moving up does not guarantee influence and might even produce an undesirable out­ come. Just because a state moves to a well-positioned calendar date does not mean that the state will attract a full range of candidates. The actual influence of early states is vulnerable to manipulation by campaign strat­ egists, who can direct their candidates to bypass a state featuring a lfavorite sonl or popular candidate in pursuit of a more appealing future confrontation (Polsby and Wildavsky 2001). In other cases, decision mak­ ers may resist a move forward because they fear significant increases in participation from non-traditional and less partisan voters, which may produce an election outcome far different from what the move's sup­ porters had in mind (Beck and Hershey 2000). The major political parties also weighed into the decision making pro­ cess by offering incentives and issuing restrictions to discourage contin­ ued frontloading for the 2000 primary season. The Democratic Party pro­ tected the early positions of Iowa and New Hampshire by restricting states from scheduling binding primaries prior to March 7. The GOP took a less restrictive approach. It tried to discourage frontloading by adding bonus delegate options to states that scheduled later primary dates (Bibby 1999; Busch 2000; Edw;;trds and Wayne 2000; Hargen and Mayer 2000; Jackson and Crotty 2001). Assessing Pennsylvania's Decision to Stay Late To understand fully the frontloading considerations in Pennsylvania prior to the 2000 election, it is important to examine Pennsylvania presi­ dential primaries in a historical context. Pennsylvania has held presi­ dential primary elections for both major political parties since 1968. An examination of primary election outcomes since the early 1980s shows different trends between the parties but consistency in the general lack of competitiveness. Republican Party primaries exhibited one-candidate domination in most years since the introduction of primaries.4 The only close election occurred in 1980 when the race between Ronald Reagan and George H. W. Bush was decided by an eight-point margin of victory. The narrowest win since 1980 was the 45 point spread for Bob Dole in 1996. The recent lack of competitiveness is noteworthy because each pri mary season after 1984 included several serious candidates. The Democratic Party, on the other hand, routinely sponsored mod­ erately close elections through 1984. These elections were moderately 56 competitive because they featured relatively 11ar1'mv margins of victory winners that attained only a plurality of the popular vote, and several serious candidates actively campaigning for the party nomination. The closest Democratic primary eJection occurred in 1980, when the race be­ tween Jimmy Carter and Edward Kennedy was decided by less that 1 %. Although Kennedy beat Carter in the popular vote, Carter gained a nar­ row advantage in state delegates. The close outcome in Pennsylvania indicated significant competitiveness in the state, but it was of little con­ sequence nationally because Carter' searlier electoral successes had placed him well ahead on the pat.h to the nomination. The Democratic primaries in 1972, 1976, and 1984 were also moder­ ately competitive. Each ejection featured multiple active campaigns, plu­ rality victories, and margins of victory under 14%. Hubert Humphrey won the 1972 Democratic primary election with a 13.8% margin of vic­ tory. Despite his relatively decisive victory, the enhanced voter choice and competitiveness of the race is most evident in that four candidates each received more than 20% of the vote. Jimmy Carter's 12% margin of victory over Henry Jackson in 1976 was another moderately decisive vic­ tory. Voters again were offered greater choice, as shown by four candi­ dates each receiving at least 12% of the vote. Pennsylvania additionally. contrihutpd to thp winnowing process in 1976 because Humphrey de­ cided not to enter the primary and Jackson bowed out of the race follow­ ing his loss in Pennsylvania. The 1984 primary featured active campaigns by Walter Mondale, Gary Hart, and Jesse Jackson. Mondale's 12% mar­ gin of victory was moderately decisive, yet the nomination was still in doubt at the time of the primary. The outcome bolstered the Mondale campaign, which had lost seven of the initial 11 primaries (Bartels 1988). Election Day competitiveness and voter choice experienced a dramatic decay following the 1984 primary. Despite mUltiple candidates appear­ ing on the ballot, the Democratic primaries in 1988 and 1992 featured one-candidate dominance. The dominance was exhibited by majority vote victories, minimal campaignine from other candidates, and margins of victory in excess of 30%. The relative competitiveness and enhanced voter choice in early Demo­ cratic party races and the 1980 GOP race is not surprising, given that Pennsylvania used to have a relatively early primary date. Pennsylvania was the seventh of only 27 states to hold a presidential preference pri­ mary in 1976 and thIrteenth of 32 primary states in 1980. As a result, when a nomination actually was contested, Pennsylvania voters were given the opportunity to influence the selection process. Reduced com­ petitiveness in more recent election years is expected because the presi­ dential candidate winnowing process was complete prior to each elec­ tion date. Frontloading activity since the 1980s eroded Pennsylvania's 57 relative calendar position, to the point that it was twenty-third of 39 GOP primary states in 1992 aml Lhirty-fin;i uf 43 GOP ~tates in 1996. Hoping to reverse the non-competitive primary trend, while establish­ ing an influential position in the presidential selection process that might spark voter interest, numerous sponsors introduced legislation in the Penn­ sylvania House and Senate in early 1999 to alter the 2000 state primary date (Cattabiani 1999). Each of the bills called for frontloading by making a permanent change forward from the legally mandated primary date of the fourth Tuesday in April. The bills varied only in exact placement on the calendar and whether or not the presidential preference primary was to be separated from the general primary election. Preferred primary dates in the bills included the first, second, and third Tuesdays in March.5 NUlle uf the~e bills was reported out of committee in either chamber of the state legislature. They all failed to attract sufficient support from com­ mittee chairs, committee members, and the rank and file to compel even a vote atthe committee level (Associated Press State and Local Wire 1999a). Although no single alternative sparked overwhelming support, some in­ terest in examining a date change persisted among lawmakers. The legis­ lature finally established a task force charged with examining the impli­ cations of moving Pennsylvania's annual primary date to September, and examining the cost and consequences of holding separate primaries dur­ ing presidential election years (Durantine 1999a, 1999b; Race 1999). In October 1999; a bill was introduced calling for a one-time calendar shift of the 2000 prim.ary from the fourth Tuesday in April to May 2. Spun­ sors introduced the bill in response to pleas from religious leaders and local election boards to change the date because it conflicted with a reli­ gious hOliday (Hartzell 1999). The change was widely supported. Follow­ ing a floor amendment changing the primary date from May 2 to April 4, the bill was unanimously approved in both chambers and signed into law. Although Pennsylvania's legislators refused to act on numerous frontloading alternatives, two questions make further scrutiny of the impact of this decision worthwhile: what was the impact of staying late in 2000, and what might have happened if Pennsy lvania had moved for­ ward to the first Tuesday of March during the 2000 primary season? The initial answer lo the first queslion is sllai1:5hlforward amI predict­ able. The results of the 2000 presidential primary indicate a continuation of the recent non-competitive trend for both parties. Even though the official primary date was moved forward three weeks, Peru1sylvania re­ mained near the end of the primary season and did not leapfrog a single state in that jump forward. The 54% margin of victory in the Democratic primary and 51 % margin of victory in the Republican primary were un­ derstandably high because Democratic candidate Al Gore and Republi­ can George W. Bush each had secured a suHicient number of delegates to 58 win his partys nomination three weeks prior to the election. Further con­ sideration of the first question and initial probing into the second re­ quire a comparison of voter choice and campaign activity in Pennsylva­ nia with that in other states facing similar frontloading decisions in 2000. Assessing the Impact of State Frontloading Decisions in 2000 To identify variations in individual state influence on the presidential selection process, I examine a range of variables measuring the actual value of individual state decisions to adjust their position on the 2000 primary calendar. 111 II1t'dtiuring the impact of these state decisions, I fo­ cus on a systematic comparison between states that moved their election date from an earlier position in 1996 and states that rejected frontloading proposals in favor of retaining their position at the later stages of the primary season. Specific measurement variables include voter participa­ tion rates by state/, active candidates on Election Day, candidate cam­ paign stops by state, voter attentiveness at the time of the primary,' com­ petitiveness of elections, and debates by state. Table 2 displays the Republican Party campaign activity and election outcome results for selected states that moved to an earlier primary date and states that resisted the frontloading trend for 2000.8 I refer to the former set of states as early 111.0Ver (EM) sLaLes dnd to the latter tiet as move resistant (MR) states. Early mover states are those states that moved forward on the calendar in search of an exclusive early date or to join a bloc primary with other states. In some instances the bloc primary takes on a regional character, whereas in other instances the bloc is just a union of participating states (Polsby and Wildavsky 2001). Move resistant states are those states that rejected pleas to move up in the 2000 primary sea­ son, knowing that they were retaining a calendar date that was very likely to fo11ow thp completion of the nomination stage winnowing process. The most visible distinguishing characteristic of the early mover states in 2000 was size. Compared with other years when smaller states strate­ gically moved forward in search of a unique daLe, Lhe luOuentiul muvers in 2000 were mostly large states seeking proportional influence over the nomination process similar to what they wielded in the general election. The most prominent early mover states in 2000 were Michigan, Wash­ ington, California, and Ohio. Michigan and Washington each moved up four weeks to establish distinctive early date Republican Party prima­ ries. California and Ohio moved their unified statewide primaries up three weeks and two weeks, respectively, to join the nationwide March 7 bloc primary The big foul" newcomers to the early primary season com­ bined for a hefty 16% of the total GOP delegates in 2000, which goes a long way toward explaining the dramatic increase in the early March delegate counts mentioned previously. 59 Table 2 State Influence on the Republican Presidential Selection Process, 2000* Primary State Campaign Debates Active Turnout Voter Margin Ivlargin Type Visits Campaigns Change mv olvement of Change at Primary from 1996 Index Victory from 1996 California EM 11 1 3 24% 46 24% -24(~ Michigan EM 15 1 3 143% 37 8% -9% Ohio EM 4 0 3 47% 46 21% -23% 0 Washington EM 5 0 3 307% 38 19% -23% \.0 lllinois MR 1 0 1 -10% 27 45% +3% New Jersey MR 2 0 1 "1OoA> 20 66% -5% Pennsylvania MR 3 0 1 -4% 22 50% +4% * EM refers to early mover states and MR refers to move resistant states. Campaign visits were tabulated by the author for the year 2000 only, with the count beginning on January 3,2000 and ending on the date of the state's primary. Debate totals are talied from December 1999 to the date of the state's primary. Voter turnout and margin of victory data were obtained from state election bureaus. The data relating to the big four early mover states dearly exhibit significant beneficial consequences for states moving up on the primary calendar. Each state moved up in search of greater voter choice and can­ didate attentiveness, which is exactly what each state received. The states offered voters greater candidate choice and potential for influence, while also attracting greater attention from the candidates and the media dur­ ing a period when voter attentiveness was at or near its peak.9 Michigan (143%) and Washington (307%) experienced exceptionally strong increases in voter participation, and California (24%) and Ohio (47%) saw sizeable increases as well.lO The frontloading decisions by these states produced competitive elections in 2000, which was a substantial improvement over 1996 when the winnowing process ended prior to each state's primary. Table 2 also contains data regarding campaign activity and election outcomes in Pennsylvania and two other prominent states that rejected appeals to move up significantly their 2000 primary date. The two other move resistant states are Illinois, which dismissed legislation to move its date from the third Tuesday in March, and New Jersey, which rejected legislation to move its date from the first Tuesday in June. Data in Table 2 indicate that these states had no impact on the winnowing process and attracted little campaign attention prior to the election. Low levels of voter turnout, uncompetitive elections, and poor voter attentiveness scores are all due to nominations having been secured prior to these primaries. On the surface, the data from Table 2 indicate that the frontloading decision for the early mover states generated substantial rewards for state voters and the states as a whole. When directly compared to competi­ tiveness and participation levels in move resistant states, the participa­ tory value of frontloading decisions is obvious. The final outcome of the Michigan election, however, highlights the problem of uncertainty that can be associated with a date shift. Despite the record primary turnout, candidate attentiveness to the state, substantial advertising, and media presence, John McCain's victory was in direct contrast with the desired intent of the principal supporters of the legislation to move the primary date forwardY In addition to the basic disparity between early mover states and move resistant states, an examination of aggregate election data from the 2000 GOP primary season further supports the contention that extreme frontloading creates vast disparities among states in regard to candidate attention and the competitiveness of elections. The relative competitive­ ness of the early presidential preference primaries is indicated by an av­ erage 18.6% margin of victory for COP primaries, up to and including the March 7 super primary. This average margin is especially notewor­ thy when compared with the average 60% margin of victory in GOP pri­ maries held after the March 7 primary bloc and the suspension of 61 McCain's campaign. 12 The absence of candidate attention to the voters or issues in these later states is additionally remarkable because onlv one of the 28 states that held primaries on datt~s after March 7, or did not sched­ ule a primary or caucus, had an official candidate visit prior to March 7. Comparing Pennsylvania and Ohio Examining the impact that a decision to move the primary date for­ ward might have had in Pennsylvania requires comparative analysis. Ohio is the ideal state for comparison with Pennsylvania on election matters because of its similar size, voting history, and cycle for senatorial and gubernatorial elections. Table 3 displays the long-standing similari­ ties between the two states in population and presidential voting prefer­ ences. The two states rank among the largest states in the Union, and their diminishing influence in the Electoral College is indicated at nearly identical rates. They also produced identical general election outcomes in presidential elections from 1972 through 1996. Table 3 Pennsylvania and Ohio: Electoral Size and Presidential Election Winners Pennsylvania Ohio Electoral Election Electoral Election Votes Winner Votes Winner and Rank and Rank 1968 29 (3) Humphrey 26 (4) Nixon 1972 27 (3) Nixon 25 (6) Nixon 197fi 27 Ci) Cartf'r 2S (fi) Cartf'r 1980 27 (3) Reagan 25 (6) Reagan 1984 25 (4) Reagan 23 (6) Reagan 1988 25 (4) Bush 23 (6) Bush 1992 23 (5) Clinton 21 (7) Clinton 1996 23 (5) Clinton 21 (7) Clinton 2000 23 (5) Gore 21 (7) Bush 62 Figure 1: Pennsylvania and Ohio General Election Turnouts in Presidential and Gubernatorial Years, 1968 w~o.-----------------------------------------------~ ~~r---------~~~~-------'r---------~~---------'M ~+---------~----~----------~----~----------~--~ OH Turnouts 2~+-------------------------------------------------~ 1~+-------------------------------------------------~ 196819701972 19741976197819801982 1984 198619881990 1992 1994 199619982000 Figure 2: Pennsylvania and Ohio Pt"esidential Primary Turnout, 1968 ~~------------------------------------------------~ 1~+---~~~~----------------------------------~~-; 1~+-~~------------------------------------------'-; 500000 +----------------------------------------------1 1968 1972 1976 1980 1984 1988 1992 1996 More than size and electoral preferences, the voting patterns and turn­ uut data fur tht: twu ::;tatt:::; ::;huw high conelalions for volel- behavior- over time at the general election and primary levels. Figure 1 shows general election voter participation in presidential and gubernatorial years.13 The correlation between the two states from 1968 to 1998 is a statistically sig­ nificant Pearson's r-value of .861 over 16 election cycles. This represents a strong association behveen the two time series, showing substantially similar patterns in historical voter participation. Figure 2 shows an equally strong association between Pennsylvania and Ohio voter participation in presidential primaries from 1968 to 1996. The correlation between the two states is a statistically significant Pearson's r-value of .862 over eight elections. Both correlation coefficients were statistically significant at the .01 level. 63 Table 4 Pennsylvania and Ohio: 2000 Primary Election Impact Pennsylvania Ohio Dem Rep Total Dem Rep Total Active Campaigns 1 1 2 2 ,.., 5 .J at Primary Margin of Victory 53.8% 50.0% 48.9% 21.0(7'0 Pre-March 7 Visits 0 0 0 4 4 8 Total Campaign 7 3 10 7 8 15 Visits Turnout Change -2.8% -6.0% -4.3% 26.0% 45.1% 36.6% from 1996 Voter Involvement Index - 22 46 In addition to the similarities between the two states in size, electoral preferences, and voter participation rates, Ohio is particularly useful for comparison with Pennsylv;:miCl hecCll1se Ohio approved legislation in 1999 to move the state primary date from the third Tuesday in March to the first Tuesday in March. This legislation followed similar legislation in 1993 moving the state primary date forward from a long-standing early May date. The 1999 frontloading legislation was supported by Republi­ can Party leaders in the state legislature and by Republican Governor Robert Taft (Associated Press State and Local Wire 1999b). Many Demo­ crats in the state opposed the move as an attempt to boost the presiden­ tial prospects of Ohio Congressman John Kasich, a Republican (Colum­ bus Dispatch 1999). That Ohio moved its primary to an earlier date in the 2000 presiden­ til'll nominl'ltion cycle, one ('onsicleTecl hllt Tejectecl by Pennsylvania legis­ lators, invites further comparison. Table 4 shows several comparative indicators of electoral choice, competitiveness, participation, and cam­ paign activity for Ohio and Pennsylvania primaries in 2000. Electoral choice and competitiveness differences can be seen in the multiplicity of active campaigns in Ohio and lower margins of victory in the Demo­ cratic and GOP primaries. Dramatic differences in voter turnout, candi­ date visits (especially those coming before victory was assured for both major party nominees), and voter involvement are additionally evident and can easily be attributed to state position on the primary calendar. 64 The most dramatic difference in Table 4 is in voter participation. Penn­ sylvania voter turnout in 2000 declined for both parties and declined statewide bv 4.3% from 1996 turnout. The 2000 Pennsylvania turnout -' . figure represents a 40.8% decline from statewide presidential primary participation in 1992, and it was 52.1 % under the high participation point set during the lYI;O primary. Ohio primary turnout in 2000 improved 36.6% statewide from 1996 and was almost 25% higher than turnout in 1992. The 2000 Ohio participation rate was a 9.1 o,{) increase over the record statewide primary participation level set in 1988. Candidate visits and campaign advertising expenditures also high­ light differences in statewide influence in the selection process and can­ didate attention to state issues and concerns. Although Al Gore and George W. Bush made several trips to Pennsylvania during the official time period of lhe 2000 primary ~ea~on, no official candidates visited the state in 2000 before Bush and Gore had already attained presumptive nominee status. Campaign advertising was additionally telling. Bush and McCain each spent aggressively on campaign ads in Ohio in the two weeks preceding the Super Tuesday primary. Bush spent over $800,000 and McCain spent over $650,000 (Columbus Dispatch 2000). In contrast no official campaign ads ran in major Pennsylvania markets before the winnowing process was completely over. Although the data in Table 4 show significant differences between the two states in competitiveness and participation, the comparison requires additional empirical analysis. I employ forecasting analysis to determine the impact lhal a firsl Tuesday ill March primary date may have had on Pennsylvania voter participation in 2000. The process involves model­ ing historical voter participation rates in Pennsylvania and Ohio, and then projecting voter turnout rates for 2000 by using forecasting applica­ tions provided in Autobox version 4.0.14 Figure 3 depicts 2000 presidential primary forecasts derived after modeling participation rates in Pennsylvania from 1968 through 1996. The 2000 forecast called for a 16% increase over 1996 turnout. The actual 2000 vote came in 4.3% under 1996 and 286,341 votes, or 17.5%, under the forecast. The actual 2000 vote was within the parameters of the lower limit of the forecast. Figure 4 depicts the 2000 presidential primary fore­ cast derived afler modeliI 19 participatiun ri:lte~ in Ohio from 1968 through 1996. The 2000 forecast called for a 15.9% increase over the actual 1996 turnout. The actual 2000 Ohio vote came in 36.6% over 1996 and 360)51 votes, or 17.9%, over the 2000 forecast. The actual participation figures from 2000 were 12.4% over the upper limit forecast. Because the dramatic increase over forecast participation in Ohio was due to the frontloaded date change and the resulting improvements in voter choice, competitiveness, interest, and attention, it is logical to ex- 65 Figure 3: Pennsylvania Presidential Primary Voting 1968-2000: with 2000 Forecasts 15000){) /"" / ".- ..... - ---...., .... ./ "'" ---e- PA Primary Votes ".... / --.- 2000 forecast ~ 1 ... upper forecast limit " * lower forecast limit 2500000 2000000 500000 1968 1972 1976 1980 1984 1988 1992 1996 2000 Figure 4: Ohio Presidential Primary Voting 1968-2000 with 2000 Forecasts ~ ~-----------------------------------------------, 1~J-~~--~============~---------------j --e- OH Primary Votes 1000000 1---------1 --.- 2000 forecast ... upper forecast limit 500000 r-______ ~L_-=*_·_·~lo~w:e~r~fu:r:ec~as:t~1i~m~it~------------------~ 1968 1972 1976 1980 1984 1988 1992 1996 2000 pect that a similar increase would have resulted from such a change in Pennsylvania. By applying the 12.4% increase over the upper limit fore­ cast in Ohio to the Pennsylvania forecast T projPct thM 2.3 million Penn­ sylvanians would have participated in the hypothetical first Tuesday in March contest. This figure represents a dramatic 63.9% increase over ac­ t:ual voter participation in the 1996 Pennsylvania presidential primary and suggests that 961,000 more voters would have participated in 2000 than actually did. Although this predicted participation is a dramatic increase from 1996, it is only a 1.5% increase over 1992 turnout and is less than actual voter turnout in 1988. One difference betw·een Pennsylvania and Ohio is that Pennsylvania's primaries are closed, whereas Ohio's are open. The early date in Ohio might have encouraged a larger number of non-affiliated voters to go to 66 the polls than Penn5ylvania's closed primary would have allowed. Al­ though the format difference may have contributed to some variation between the states in turnout, closed primaries in P€lU1sylvania would not have prevented an upsurge in overall turnout similar to Ohio's. His­ torically, a strong similarity in primary election participation rates has existed between the two states (see Figure 2). Despite the difference in primary formats, the states exhibited very similar voter participation patterns in prior events. Pennsylvania's political culture and registration rules also encourage major partisan affiliation. Over 90% of registered voters were enrolled as Democrats or Republicans at the time of the 2000 primary, indicating a high level of party affiliation. This high rate of party affiliation, along with state law allowing voters to declare or change party affiliation up to thirty days prior to the election, left few potential voters excluded from participation on primary day if they were inclined to sup­ port one of the major candidates. Increased participation rates in the Republican and Democratic prima­ ries in Ohio fell short of the Secretary of State's projections (Brazaitis 2000). Unlike claims from election officials and pundits in New Hampshire and Michigan, the post-election analysis of voter turnout rates in Ohio was de­ void of references to dramatic increases in participation due to John McCain's appeal to non-traditional primary voters. A range of states holding early primaries in 2000 experienced increases in voter participation. The turnout increases were not exclusive to open primary states but included equally dramatic increases in participation from closed primary states such as Con­ necticut, Maine, Maryland, Rhode Island, and South Carolina. Potential Impact on Other ~tatewide l.Jrimaries Although a decision to frontload the 2000 Pennsylvania presidential primary date likely would have led to enhanced competitiveness and substantially increased actual voter turnout it prompts another ques­ tion. What effect would dramatically improved participation rates have had on other races in the state if Pennsylvania had held an early unified primary in 2000? Critics of the various legislative proposals in Pennsyl­ vania cited concerns about cost, uncertainty, and disruption to the status quo when speaking out against frontloading and splitting the primaries (Durantine 1999a; McCarthy 1999; Race 1999; Rubinkam 1999). Many Pennsylvania legislators and party officials were willing to consider fu­ ture changes to the primary date, but they feared the impact that changes to the primary date might have on state legislative races. Concerns over potential disnlption and uncertainty were compelling in 2000 given the very narrow partisan distribution in the House and the looming specter of redistricting following the 2000 census. The prospect of significant in­ creases in voter partiCipation during the 2000 primary, particularly among 67 non-traditional primary voters, did not arouse support from the state­ wide political community. My analysis of the potential disruption to the status quo due to in­ creased turnout and an earlier primary date involves scrutiny of Ohio and Pennsylvania state senate and state house election results in 2000. To assess the potential impact of a frontloaded primary, I first determine the extent to which state legislative seats were competitive and contested in 2000. I then examine election outcomes in the context of voter participa­ tion rates in both states. Potential impact calculations in Ohio assess the potential for change in electoral outcomes if turnout had not increased significantly in 2000, while potential impact calculations in Pennsylva­ nia evaluate the potential for change in electoral outcomes if turnout had increased significantly in 2000. I operationalize potential impact for nar­ row margins of victory (less than 20%) and close races (less than 10%). Potential impact for narrow margins is determined by counting the num­ ber of candidates from narrow margin of victory primaries who won or were defeated by less than 15% of the vote on general election day. Po­ tential impact for close margins is delennlned by counting the number of candidates from close margin of victory primaries who either won or were defeated by less than 10% of the vote on general election day. The post-election analysis detailed in Table 5 shows that the frontloaded primary date and the increased turnout in Ohio had a sizeable potential impact on the outcome of races for the state House of Representatives. The analysis for narrow margins of victory shows that 28 of the 99 Ohio State House races conceivably could have ended with a different victor on general election day if voter turnout had been significantly dimin­ ished on the date of the primary. The calculations still show a dramatic potential impact of 21 seats out of 99 when the close margin of victory ::;tamlanl i::; applit!u. Tht! hi~h ratt! uf l:hallt!ugt!u ral:t!o anu lIlultituut! uf competitive elections in Ohio is partly attributable to the high number of open seat races in the Ohio House. Because of state imposed term limits and attrition, only 57 of the 99 races featured incumbents in 2000. The potential impact on the Ohio House of Representatives was re­ stricted to the individual candidate level and did not show significant pros­ pects for impact in the partisan make-up or control in the state legislature. The pre- and post-election partisan distribution figures showed identical levels of GOP domination in each chamber of the legislature. Although the close race analysis shows that 21 seats conceivably could have had a different winner, only six of the 21 would have resulted in a different party huluiIl~ tht! uffil:t!. Similar aIlaly::;b uf tht! Ohiu St!Ilait! raet!::, ohuwo ::;ig! Lifj­ cantly diminished potential impact. When the close margin of victory stan­ dard is applied, only one of the 16 Ohio Senate races would qualify as potentially affected if voter turnout were significantly depressed. 68 Table 5 Contested Primaries and Close Elections in Pennsylvania and Ohio, 2000 State Senate State House U.S. Congress PA OH PA OH PA OH Total Seats Up 25 16 203 99 21 19 for Election Total Potential 50 32 406 198 42 38 Primaries Total Contested 1 5 32 56 7 10 Primaries (8.0%) (15.6'1'0 ) (7:9%) (28.3%) (16.7%) (26.3%) Primaries with 3 3 12 36 4 5 N arrow Margins (6.0%) (9.4%) (3.0%) (18.2%) (9.5%) (13.2%) «20%) Primaries with 1 1 5 25 3 3 Close Margins (2.0%) (3.1%) (1.2%) (12.6%) (7.1%) (9.4%) « 10%) Potential Impact 1 of 25 2 of 16 8 of 203 28 of 99 2 of2l 1 of 19 on General seats seats seats seats seats seats Election affected affected affected affected affected affected Outcomes for (4.0%) (12.5%) (3.9%) (28.3%) (9.5%) (5.3%) N arrow Margins The analysis of the impact on Pennsylvania state legislature races con­ tained in Table 5 shows an entirely different story. Although partisan control of the House of Representatives was up in the air going into the 2000 election, the limited number of contested primary races and the widespread absence of competitiveness were staggering. Of the 406 po­ tentially contested primaries for the 203 seats in the House of Represen­ tatives, only 32 races involved multiple candidates for the party desig­ nation. Of the 32 contested races, only 12 had narrow margins of victory and only five involved close races. Potential impact calculations based upon the narrow margin of victory standard show that substantial in­ creases in voter participation may have led to a different candidate pre­ vailing in eight of the 203 races. The close margin of victory standard shows that only two seats may have had a different outcome if voter participation were significantly increased from 0 frontloodcd primary date. Table 5 also displays a similar absence of challenges at the primary level for the Pennsylvania Senate races, and it shows very limited poten­ tial impact. Even when the narrow margin standard is applied, only one 69 of the 25 Pennsylvania Senate races would qualify as potentially affected if votPT hlTnOut .crignificantly increased. Given the pre-election partisan distribution of 30 Republicans and 20 Democrats, there is no chance that increased turnout would have threatened partisan control of the Penn­ sylvania Senate. The absence of challenges and the limited competitiveness of Pennsyl­ vania legislative races highlight the importance of incumbency in deter­ mining election competitiveness and outcomes. The entrenchment of in­ cumbents in the Pennsylvania political system (only 11 of the 203 House districts featured open seat races) makes voter participation rates some­ what irrelevant, for incumbents are afforded advantages that discourage serious challengers and create significant imbalances in campaign re­ sources. It is also unlikely that a legislative change to the 2000 primary date would have produced any significant increase in competitive chal­ lenges for House and Senate seats in Pennsylvania. A late decision to frontload the primary to early March would most likely have had the op­ posite effect by creating a shorter candidate recruitment period, a narrower fund raising window, and compressed petition drive deadlines that would have further discouraged challengers and disfavored outsider candidates. Table 5 also includes data regarding the competitiveness and potential impact of turnout changes on congressional races in Ohio and Pennsylva­ nia. Analysis of the primary outcomes in Ohio shows that significant re­ ductions in voter participation would not have affected the outcome of any of the 19 races for the u.s. House of Representatives .. Contested pri­ maries were few and competitiveness was limited because incumbents were seeking reelection in 18 of the 19 districts. Scrutiny of the congres­ sional primary races in Pennsylvania shows that significant increases in voter participation potentially would have affected only one of the 21 cam­ paigns for the u.s. House of Representatives. The absence of competitive­ ness in Pennsylvania is attributable to incumbents seeking reelection in 19 of the 21 districts.IS Additional scrutiny of the u.s. Senate races in Penn­ sylvania and Ohio shows that dramatic changes in voter participation rates would not have affected the ultimate election outcome in either state. Increased voter turnout from a frontloaded primary date in Pennsyl­ vania would have had a little impact on the outcome of elections across the state. The paucity of contested primaries and competitive races of­ fered little chance for upheaval from a significant increase in voter turn­ out. Thus, legislators' fears of the potential impact on non-presidential races in the state were largely unfounded. Despite the limited potential impact, the closeness of the partisan make-up in the House (deadlocked at 100 to 100 with three vacancies at the time of the 8pnpra 1 plpction) made the possibility of disruption to even one campaign worthy of con­ cern to both political parties. 70 Conclusion A review ot multiple comparisons between states that participated in the frontloading trend for 2000 and states that resisted frontloading clearly shows that individual state decisions to select art early date offer greater prospects for enhanced voter choice, competitiveness of elections, and higher voter participation. Despite this potential, legislators considering a move forward on the primary calendar should be aware that frontloading does not guarantee successful early influence. The effec­ tiveness of date selection decisions are subject to minimizing influence by other frontloading states and avoidance strategies by savvy campaign managers. The recent frontloading trend has compressed the early sec­ tion of the primary season so greatly that frontloading states must either separate their statewide primaries or hold early unified primaries. When they separate their primaries they incur greater administrative and fiscal costs and pOlentially diminished voler Lurnoul for non-presidential pri­ maries. When they hold unified primaries they create greater adminis­ trative burdens and face the perceived risk that large increases in voter turnout would disrupt the status quo of expected outcomes in non-presi­ dential races. Comparative analysis and forecasting analysis between Ohio and Pennsylvania show that a significant move forward on the 2000 primary calendar by Pennsylvania would have generated significant increases in voter participation. Non-participation in the frontloading trend in 2000 assured that Pennsylvania would miss out on an opportunity for greater voter choice, competitiveness, and campaign activity. The politically prag­ matic decision to retain the state's late primary date was a risk aversive action that displayed a clear preference for avoiding uncertainty and upheaval over pleas for greater political influence and participation. Given the 1im.iled nalw"e of compeLitiveness in Pennsylvania primary elections, a concern with uncertainty and upheaval is unwarranted. 71 Notes 1. In response to encroachment from states establishing early GOP primrlrips or non-binding caucuses, New Hampshire and Iowa moven their dates forward on the calendar prior to the 1992 and 2000 primaries. The February I, 2000 New Hampshire primary date was a full three weeks earlier than the traditional fourth Tuesday of February date of the 1970s and 19805. Since the official 2000 primary season still ran into June, the jump forward had the effect of lengthening the overall primary season. 2. I use this format rather than the weekly tally format of Mayer (2001) and Morton and Williams (2001) because the movement forward by New Hampshire in 1996 and 2000 disrupts the weekly tally comparison between primary sea­ sons. This format more effectively highlights the compressed nature of the del­ egate selection process in the early weeks and months of the primary season. 3. For example, local elections officials in Pennsylvania and Ohio expressed opposition to frontloading proposals for 2000. They criticized the proposals be­ cause the early calendar dates would disrupt established filing deadlines, hinder campaign worker recruitment, and place additional stress on election officials seeking to complete administrative tasks in a shorter time frame and during the holiday season. Pennsylvania election officials were additionally opposed to pro­ posals that would split the primaries. They stated that running two primaries would increase costs by over 50% and would minimize essential preparation time for the general election (Duran tine 1999a; McCarthy 1999; Race 1999). 4. The lack of competitiveness in the 1970s and 1980s was mostly due to an ab­ sence of challengers, as the GOP selection process was routinely ceded to incumbent presidents running for reelection or incumbent vice-presidents seeking the party nod. 5. HB 1610 PN 1965 and SB 233 PN 242 each proposed to move the presiden­ tial preference primary to the first Tuesday in March. The Senate version of the bill proposed to move the rest of the primaries to September; whereas the House bill left the remaining primaries in April. These two bills attracted more co-spon­ sors than any other frontloading bills considered in 1999. SB 91 PN 86 proposed moving all primaries up to the second Tuesday in March and SB 258 PN 259 proposed moving only the presidential primary to the second Tuesday. FinallYI lIB 653 f'N 691 proposed moving Lhe pre::>idelltial primary tu the third 1Uesday in March but holding all others in September. 6. My decision to focus on changes in aggregate vote totals between election years is predicated on the understanding that determinations of voler lurnuul vary substantially across states and even across elections within a single state. Using simple vote counts for GOP candidates provides a helpful and basic sum­ 'mary of the number of individuals taking part in the election while avoiding issues of asymmetry created by using non-uniform measures of voter turnout rates. See Beck and Hershey (2001) for an informative discussion of the problems inherent in creating uniform determinations of voter turnout across states. 7. I determine voter attentiveness by the value of the voter involvement in­ dex at the time of the primary (Vanishing Voter Project 2000). The index repre­ sents a compilation of responses to survey questions measuring nationwide voter attentiveness and involvement. The index was updated weekly throughout the campaign and began the new year at a low of 12% voter involvement. 72 8. Since President Clinton competed in the 1996 Democratic primary without opposition, the Democratic turnout and competitiveness numbers are mislead­ ing. I use only GOP data in this table to produce more reliable comparisons. 9. Republican campaign activity in these states pales in comparison to candi­ date activity in Iowa, New Hampshire, and South Carolina. These states moved up their early primary dates from 1996 in an attempt to preserve their early in­ fluence. Iowa had 46 candidate visits and two debates, New Hampshire had 74 visits and three debates, and South Carolina benefited from 44 candidate visits and two debates. 10. Despite Democratic Party rules restricting states from holding binding primary elections prior to March 7, the Democratic party of Washington state scheduled a non-binding primary for the same day as the GOP primary. This drew the attention of the event-starved Democratic candidates (who scheduled 10 visits to the state prior to the non-binding primary) and increased statewide and national attention to the primary. 11. Michigan Act 71, of the Public Acts of 1999, was supported by the Gover­ nor, the Lieutenant Governor, the Speaker of the Michigan House of Representa­ tives, and the Michigan Secretary of State. Each of these officials endorsed - Bush for the Republican nomination, as did numerous official sponsors of the legisla­ tion. Mid-1999 polling reported overwhelming support for Bush among likely Michigan voters (Ceasar and Busch 2001). 12. Even after a candidate withdraws from the race, in most cases his name re­ mains on subsequent ballots. Non-active candidates still attract a measurable, and at times substantial, number at votes, thus preventing unanimous election results. 13. Total votes cast is superior to voter turnout as a measurement of compari­ son between Pennsylvania and Ohio for this time period. Although both states experienced modest population increases between 1960 and 2000, the popula­ tion swings are similar and are much less substantial than changes in numbers of registered voters. Voter registration numbers in both states varied dramatically over time, causing the turnout calculation as a percentage of registered voter~ to be unstable and less reliable than actual participation. That Ohio did not require registration or maintain voter registration statistics prior to 1978 further impedes a turnout-based comparison. 14. Autobox Version 4.0 is distributed by Automated Forecasting Systems, Inc. of Hatboro, Pennsylvania. 15. All the illcLllllb~nls seeking reelection in Ohio and Pennsylvania retained their seats in the 2000 general election. For a thorough breakdown of incum­ bency advantages that discourage serious challengers and promote high reelec­ tion rates in the u.s. Congress and in state legislatures see Jacobson (2001) and Maisel (1999). 73 References Associated Press State and Local Wire. 1999a. "Lawmaker Wants to Move Up Pennsylvania Primary." May 3. __ . 1999b. "Taft Signs Primary Election Bill Into Law." June:l7. Bartels, Larry M. 1988. Presidential Primaries and the Dynamics ~f Public Choice. Princeton: Princeton University Press. Beck, Paul Allen, and Marjorie Randon Hershey. 2001. Party Politics in America. 9th ed. New York: Addison Wesley Longman. Bibby, John F. 2000. Politics, Parties, alld Elections in America. 4th ed. Belmont, CA: Wadsworth. Brazaitis, Torn. 2000. "Record Turnout of Ohio Voters Predicted Today." The Plain Dealer, March 7, SA. Buell, Emmett H. Jr. 2000. "The Invisible Primary. " In 111 Pursuit of the White House 2000: How We Choose Our Presidential Nominees, ed. William Mayer. New York: Chatham. - Busch, Andrew. 1997. Outsiders and Openness in the Presidential Nominating Sys­ tem. l-'ittsburgh: University ot Pittsburgh Press. __ .2000. "New Features of the 2000 Presidential Nominating Process: Repub­ lican Reforms, Frontloading's Second Wind, and Early Voting." In In Pursuit of the White House 2000: How We Choose Our Presidential Nominees, ed. Will­ iam Mayer. New York: Chatham. Cattabiani, Mario F. 1999 "Legislators Seek to Build Interest in Primary Elec­ tium;." Murning Cull (Allentuwn), May 18, B2. Ceasar, James W., and Andrew E. Busch. 2001. The Perfect Tie: The True Story of the 2000 Presidential Election. Lanham, MD: Rowman & Littlefield Publishers. Columbus Dispatch. 1999. "Senate OK Means Ohio Likely to Have March 2000 Primary." May 26, lA. . __ . 2000. "Candidates Pull Out Stops For Weekend." March 4, 1A. Cook, Rhoades. 1992. "The Nomination Process." In The Quest jor National Office, ed. Stephen J. Wayne and Clyde Wilcox. New York: St. Martin's Press. Durantine, Peter. 1999a. "Legislative Task Force to Consider Changing Primary Date to September." Associated Press State and Local Wire, June 24. __ . 1999b. "Panel Will Report in One Year on Whether to Change Primary Election Date." Associated Press State and Local Wire, August 18. Edwards, George R, and Stephen J. Wayne. 2000. Prl'."idential l.eadership: Politics and Policy Making. 5th ed. New York: St. Martin's / Wadsworth Publishing. Geer, John G. 1989. Nominating Presidents: An Evaluation of Voters and Primaries. New York: Greenwood Press. Bargen, Michael G., and William G. Mayer. 2000. "The Modern Politics of Presi­ dential Selection: How Changing the Rules Really Did Change the Game." In In Pursuit of the White House 2000: How We Choose Our Presidential Nomi­ nees, ed. William Mayer. New York: Chatham. Hartzell, Dan. 1999. "Jews Ask Change in Election Date." Morning Call (Allen­ town), September 2, B1. Haskell, John. 2000. "A Quarter Century of Direct Democracy in Presidential Nomination Campaigns: What's the Verdict?" In Political Parties, Campaigl1s, and Elections, ed. Robert E. DiClerico. Upper Saddle River, NJ: Prentice Hall. 74 Jackson, John 5., and William Crotty. 2001. The Politics o.fPresidel1tial Selection. 2nd ed. New York: Addison Wesley Longman. Jacobson. Gary C. 2001. The Politics of Congressional Elections. 5th ed. New York: Addison Wesley Longman Lengle, James. 1997. "Evaluating and Reforming the Presidential Nominating Process." In Understanding the Presidency, ed. James Pfiffner and Roger Davidson. New York: Addison Wesley Longman. Lichter, S. Robert. 1992. "How the Press Covered the Primaries." In The Quest for National Office, ed. StephenJ. Wayne and Clyde Wilcox. New York: St. Martin's Press. Maiset 1. Sandy. 1999. Parties and Elections i11 American: The Electoral Process. 3rd ed. Lanham, MD: Rowman and Littlefield Publishers, Inc. Mayer, William G. 1997. "The Presidential Nominations." In The Election of1996: Reports and Interpretations, ed. Gerald Pomper. Chatham, NJ: Chatham House Publishers. __ .2001. "The Presidential Nominations." In The Election of La au: Reports and Interpretatiol1S, ed. Gerald Pomper. Chatham, NJ: Chatham House Publish­ ers. McCarthy, John. 1999. "House Votes to Move Primary, Despite Local Protests." Associated Press State and Local Wire, April 21. Morton, Rebecca, and Kenneth C. Williams. 2001. Learning by Voting: Sequential Choices in Presidential Primaries and Other Elections. Ann Arbor: University of Michigan Press. Nye, Joseph Jr., Philip Zelikow, and David King. 1997. VV71Y People Don't Trust Government. Cambridge, MA: Harvard UniverSity Press. Patterson, Thomas. 1994. Out of Order. New York: Alfred A. Knopf. Polsby, Nelson W., and Aaron Wildavsky. 2000. Presidential Elections: Strategies and Structures of American Politics. 10th ed. New York: Chatham House Pub­ lishers. Race, Michael. 1999. "Pennsylvania Ponders Moving Primary Date." Morning Call (Allentown), August 18, A3. Rubinkam, Michael. 1999. "Republicans Push Against Early Primaries, Unveil New Internet Technology." Associated Press State and Local Wire, July 9. Vanishing Voter Project. 2000. "Public Involvement and the 2000 Nominating Campaign: Implications for Electoral Reform." A report presented to the National Press Club April 27. Wayne, Stephen J. 2001. The Road to the White House, 2000. New York: St. Martin's / Wadsworth. 75