Contested Primaries: A Review and Comparison from 2006/2008

advertisement
What Divisive Primary?
Evidence from State Senate Elections in 2006/2008 and 2010
Presented to the State and Local Politics and Policy
Conference 2014
Bloomington, IN
Dr. Ben Tafoya
Walden Univeristy
ben.tafoya@waldenu.edu
(781) 315-5920
Ben Tafoya
What Divisive Primary?
2
Abstract
Conventional political wisdom states that candidates without primary opponents have a
built-in advantage against general election opponents that must run vigorous primary elections.
The existing academic literature is split in this contention. This paper examines the result of state
senate elections from 2010 to test the hypothesis of this assumption. It finds an important
confounding factor with respect to previous research efforts – party of the previous incumbent –
as a more significant predictive factor than competitive primaries. Further analysis compares the
district-based results from 2010 with their previous cycle in either 2006 or 2008. The paper holds
open the possibility that the effect will vary depending on whether we look at district-based or
statewide elections.
Ben Tafoya
What Divisive Primary?
3
Introduction
There is a view, supported by the literature and conventional wisdom, which holds that
primaries negatively impact a candidate’s chances for election (Bernstein, 1977; Johnson, 2010).
Primaries can result in party divisiveness, which means that candidates have a more difficult
challenge to unite their party for general election purposes. An alternative view emerged that
argued, primaries are an opportunity to increase name recognition, fundraise, and refine political
messaging (Lazarus 2005). To test the competing assertions, the researcher examined open seat
state senate races in 2010 that involved competitive primaries for one candidate and no primary
for the other. Open seats have typically more opportunities for primaries than races involving
incumbents. The research included a comparison of party vote totals between the last election
cycle of the previous incumbent (either 2006 or 2008 depending on the state) and the general
election of 2010.
The literature on primary election activities is equally split on the issue of divisiveness.
Piereson and Smith (1975) examined Gubernatorial elections throughout the US from 1956
through 1964 and found no support for the divisive primary hypothesis. In fact they find a small
positive effect for competitive primaries on the prospects of non-incumbents. Their results mirror
those of Hacker (1965) who concluded that a divisive primary bore little relation to the general
election result. However, Piereson and Smith’s methodology was significantly different in that
they measured the nominal impact of primaries on the general election vote, rather than simply
winning and losing, the approach used by Hacker. It is important to point out that both studies
looked at statewide elections while this work concentrates on district-based elections. A study by
Born (1981) on US House of Representatives primary elections found no impact of primaries on
Ben Tafoya
What Divisive Primary?
4
non-incumbent races. Yet V.O. Key (1964) expressed concerns about the impact of primaries on
stable party structures and increased intra-party factions.
Key’s conclusions were carried over into recent work by Makse and Sokhey (2010)
where they examined the outcome of the 2008 Democratic Primary match between Clinton and
Obama. They concluded that the divisiveness of the historically close nomination contest could
have cost Obama the presidency absent the economic crisis. Yet Wichowsky and Niebler (2010)
examined the competitiveness of presidential primary contests and the divisiveness (as a function
of negative advertising) and came to the conclusion that Obama was helped by participation in
close primary contests in the subsequent general election vote.
What the researcher found through the course of the work was surprisingly clear based on
this record; the data suggests that in the examined races (open elections for State Senate), the
presence of a primary had little impact on the outcome of the race. That is, primaries had a
statistically insignificant impact on the outcome of the subsequent general election. Instead, it
was found that the party of the previous incumbent served as the strong indicator of which party
would win the open seat election. The findings from 2010 were tempered somewhat by the
presence of a Republican wave.
Research Methods and Results
The 2010 legislative elections were characterized by a distinctive wave in the direction of
Republican candidates. Republicans gained control of six State Senate chambers that were
previously controlled by Democrats. One upper chamber remained a tie (Alaska Senate). Lower
houses passed into Republican control in 14 states and one moved from Democrats to a tie.
Ben Tafoya
What Divisive Primary?
5
According to the National Council of State Legislators, Republicans gained a total of 675 seats in
both lower and upper chambers and as a result in 2011 Republicans held their highest number of
legislative seats since 1928 (NCSL 2013).
This study examined open seat elections for State Senate across the universe of 2010
elections. State Senate level elections were chosen because of the preponderance of single
member districts, which made measuring the impact of primaries possible. In 2010, 46 states
held State Senate elections including Nebraska’s non-partisan, unicameral legislature (which was
excluded from the data). Analysis focused on those states where there were open seat elections as
a result of the incumbent office holder choosing not to stand for re-election or who were
sidelined because of term limits. Election data came from historical records maintained by
Ballotpedia, The Almanac of State Legislative Races 2000-2006 and various state election
officers. The resulting races were broken down to find those instances where there was one party
nominee with a primary and an opponent that did not have a contested primary. The null
hypothesis was that candidates with a contested primary were at a competitive disadvantage in
their general elections against opponents without a primary opponent. Further screening was
done to determine a “competitive primary” as those where the victor had less than 67% of the
vote. This left 93 races for consideration in this population.
Table 1
Ben Tafoya
What Divisive Primary?
6
As shown in Table 1, Republicans won 56 of the 93 seats up for election from the
population. Of those 56 victors, 45 of them had contested primaries. Of the 37 victories by
Democrats, 32 had primaries with the other five seats won by candidates not engaged in
contested primaries. Overall 82.8% (77/93) of the winners of general elections in open seat
contests in 2010 had competitive primaries leaving those without a primary and facing an
opponent who had a primary only winning 17.2% (16/93) of the general elections. Based on this
result we rejected the null hypothesis and concluded that having a contested primary in an open
seat election where the winning candidate’s opponent did not have a primary is not a
disadvantage.
The research then focused on what factors could play a role in this result. Why do
candidates with primaries fare better in general elections than their opponents who do not have a
primary challenge? The research examined a potential salient factor that could determine the
outcome of general elections despite the presence or absence of a primary election. Is the district
drawn in such a way that one party is favored to retain the position in the event of an open seat
election? The 93 districts were re-examined based on the party of the previous incumbent and the
results are shown in Table 2.
Ben Tafoya
What Divisive Primary?
7
Table 2
The data shows that the party of the previous incumbent is a strong predictor of the
results of an open seat election. Democrats only won one seat that was previously held by a
Republican and 36 of their 37 victories were in districts previously held by a Democrat.
Democrats won 31 out of 41 seats where they were defending a seat previously held by a
Democrat and there was a primary. Democrats lost 7 out of 12 seats where they were defending a
seat previously held by a Democratic incumbent and there was no primary on the Democratic
side.
The Republican wave was demonstrated in this analysis as they held all 35 seats vacated
by a Republican incumbent and where they had a primary and the Democrats had none. They
also won 4 out of 5 seats where they had no primary and there was a primary on the Democratic
side for a seat previously held by a Democrat. The conclusion is that being a member of the same
party as the previous office-holder is a better predictor of electoral success than whether there
was a primary election, overall 75 out of 93 state Senate general election contests were won in
2010 by a candidate from the same party as the previous incumbent.
Ben Tafoya
What Divisive Primary?
8
Taking a broader universe of contests the researcher added in all races for open seats
regardless of primary status. This universe including the previous contests (primary in one party
and no primary in the other) and also included those where there were contested primaries in
both parties and contested primaries in neither party. A linear regression on the 182 contests
showed in Table 3 present a strong result for the party of the previous incumbent as a factor and
the primary as insignificant.
Table 3
Regression Model for Open Seat Outcomes
Model
Unstandardized Coefficients
B
1
Standardized
Coefficients
(Constant)
.434
Std. Error
.040
Beta
t
10.837
Sig.
.000
Incumbent Party
.504
.062
.532
8.114
.000
Primary
.063
.073
.057
.870
.385
a. Dependent Variable: Winner Party
!
The relationship between election cycles is shown in Appendix 1 and 2. The results from
the previous election cycle are depicted (either 2006 or 2008 depending on the state versus
2010). In the first set the population is the subset (N = 134) of the 182 races where there was
actual competition in the previous election cycle. This eliminated 48 races which were
uncontested by one party or another. The graph shows the two-party vote percentage going to the
Republican candidate in the general election. As can be seen by inspection there is a Republican
Ben Tafoya
What Divisive Primary?
9
advantage in 2010 over the previous election cycle. In a paired samples test the difference
between the Republican vote in both sets of cycles is 7.5%. The correlation is .753. Looking at
the Republican vote, in those districts previously held by a Republican the average vote across
all districts in 2010 was 64%. In districts previously held by a Democrat the total was 46%.
In Appendix 2 the same pattern is shown but for only those contests where there is a
primary in one party and not the other. With an N = 78 after winnowing to those contests that
had valid results in both cycles and where there was a primary in only one party the vote
difference widens to 8.3% and the correlation is .84.
Threats to Validity and Further Research
There are concerns of internal and external validity with this study. The greatest internal
threat was deemed to be historical as all observations were of races in 2010. This last election
cycle was a decided wave election in favor of Republican candidates. If the results were viewed
in elections with countervailing waves (such as 2006 or 2008 looking back against 2004, etc…)
or no waves then there is a possibility of differing results. Moreover, because of the filter and the
sample from one election cycle, the sample is small. External validity is also a concern as this
study examined district level elections for state offices (State Senate) and may be difficult to
apply to other district level elections (such as US Congress) or statewide, non-district elections.
Previous research showed the impact of an intra-party primary on incumbent officeholders was
negative, but that the situation described here, open seat elections, attracts more primary
opportunities (Hogan 2003). Moreover, the 2010 election cycle was the last under existing
Ben Tafoya
What Divisive Primary?
10
district configuration prior to re-apportionment. Will new districts after 2012 change the
outcomes of analysis? It is certainly possible that the effect of partisan line drawing
(gerrymandering) could deepen after this latest round of reapportionment.
In an attempt to isolate some of concerns of validity in the study the research looked at
whether the election results from the previous election cycle (either 2006 or 2008 depending on
the state) influenced the relationship between the variables (party of the previous incumbent and
primary). This study provides a good baseline on the performance of candidates with competitive
primaries. Not all states had State Senate elections in 2010 so research can examine the results
based on a set of elections in the new apportionment decade starting 2012 with elections in 2014.
It would be instructive to examine results in light of a regional classification such as those
proposed by Hogan (2003).
Discussion of Results
Early in the study the researcher rejected the null hypothesis regarding the importance of
contested primaries in predicting the outcome of open seat elections for the State Senate. In the
38 states with open seat elections included in the study it was found that the stronger determining
factor in success was not whether the candidates had primaries but rather the party of the
previous incumbent. The cause of this factor was thought to be two-fold; under some
circumstances the district lines are drawn to favor specific parties because of known voting
behaviors of the residents in precincts, wards and towns (so-called gerrymandering) or because
certain districts have unavoidable tendencies under contemporary political arrangements to vote
for one party (ie…urban districts vote Democratic, rural districts vote Republican). This latter
Ben Tafoya
What Divisive Primary?
11
effect accentuated in smaller districts such as State Senate where only in California are they
bigger than or the US House of Representatives.
It is also possible that primaries do have some impact on the prospects of the party’s
nominee but that effect is hidden, as it does not change the outcome but rather impacts the
overall vote total. Based on the research this would suggest adopting a version of the conclusions
supported by Piereson and Smith (1975), that under the circumstances of open seat, district based
elections there is no difference in effect between looking at nominal outcome or nominal vote
percentage.
These findings were thought to be significant particularly in a reapportionment year. If
contested primaries do not drain resources from candidates but rather provide the forum for the
“true” election in certain districts then states may want to concentrate on drawing districts in
such a manner as to minimize the gerrymandering of lines as general elections lose importance
under the current scenario. Regionally based voting tendencies may be unavoidable (the urban
and rural generalization above) as a consequence of contiguous districts. The question remains as
to whether two-party competitive districts provide a stronger basis for the representation of the
diversity of political opinions inside and outside the major parties. Experiments in non-political
reapportionment such as California and Iowa and (19 other states) bear watching, as do the socalled non-partisan primaries in Washington and California for their impact on the strength of
gerrymandering of districts
Ben Tafoya
What Divisive Primary?
12
Bibliography
Bernstein, Robert A. "Divisive Primaries Do Hurt: Us Senate Races, 1956-1972." American Political
Science Review 71, no. 2 (June 1977 1977): 15.
Born, Richard. "The Influence of House Primary Election Divisiveness on General Election Margins,
1962?76." The Journal of Politics 43, no. 03 (1981): 640-61.
Gregg B. Johnson, Meredith-Joy Petersheim and Jesse T. Wasson. "Divisive Primaries and Incumbent
General Election Performance: Prospects and Costs in U.S. House Races." American Politics
Research 38, no. 5 (2010): 24.
Hacker, Andrew. "Does a “Divisive” Primary Harm a Candidate’s Election Chances?". American
Political Science Review 1 (1965): 5.
Hogan, Robert E. "The Effects of Primary Divisiveness on General Election Outcomes in State
Legislative Elections." American Politics Research 31, no. 7 (January 2003): 20.
Key, V. O. Jr. Politics, Parties, and Pressure Groups. 5th ed. New York: Thomas Y. Crowell, 1964.
Lazarus, Jeffrey. "Unintended Consequences: Anticipation of General Election Outcomes and Primary
Divisiveness." Legislative Studies Quarterly XXX, no. 3 (August 2005): 26.
Legislators, National Council of State. "A Wave or a Ripple." http://www.ncsl.org/legislatureselections/elections/a-wave-or-a-ripple.aspx.
Makse, Todd, and Anand E. Sokhey. "Revisiting the Divisive Primary Hypothesis: 2008 and the
Clinton—Obama Nomination Battle." American Politics Research 38, no. 2 (March 1, 2010
2010): 233-65.
Norrander, Barbara. "Primary Elections." In The Oxford Handbook of American Elections and Political
Behavior, edited by Jan E. Leighley. 515-30. Oxford, UK: Oxford University Press, 2010.
Smith, James E. Piereson and Terry B. "Primary Divisiveness and General Election Success: A ReExamination." The Journal of Politics 37, no. 2 (May 1975 1975): 7.
Wichowsky, Amber, and Sarah E. Niebler. "Narrow Victories and Hard Games: Revisiting the Primary
Divisiveness Hypothesis." American Politics Research 38, no. 6 (November 1, 2010 2010): 105271.
Dr. Ben Tafoya is the Undergraduate Program Director at the Walden University School of
Public Policy and Administration. He has previously been an Assistant Professor of Political
Science at New England College in Henniker, NH. He has taught and supervised a range of
undergraduate and graduate courses in political science, public policy and economics. His
research interests include state and local public policy, economic policy and elections. He is a
local elected official in Massachusetts.
The author would like to thank his colleague Prof Wayne Lesperance of New England College
for his help on an earlier version of the article, Jason Bloxham, a graduate of the political science
program at New England College and Andrew Kluge a Political Science student at Suffolk
University for their assistance with the research in the article.
Download