Donate
Independent, objective, nonpartisan research

R 514EMR

Authors

R 514EMR

Tagged with:

Publication PDFs

Database

This is the content currently stored in the post and postmeta tables.

View live version

object(Timber\Post)#3711 (44) { ["ImageClass"]=> string(12) "Timber\Image" ["PostClass"]=> string(11) "Timber\Post" ["TermClass"]=> string(11) "Timber\Term" ["object_type"]=> string(4) "post" ["custom"]=> array(5) { ["_wp_attached_file"]=> string(12) "R_514EMR.pdf" ["wpmf_size"]=> string(6) "484679" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(58191) "Voter Turnout in Primary Elections May 2014 Eric McGhee with research support from Daniel Krimm Supported with funding from the S. D. Bechtel, Jr. Foundation Summary In June 2010, California voters passed Proposition 14 to authorize a new “top -two ” primary election system. Previously, those registered with a party could only vote in that party’s primary, though independents were usually allowed to choose a primary they wanted to join for that election. Now under the top -two system , voters can vote for any candidate from any party , and the two candidates who receiv e the most votes , regardless of party affiliation, compete against each other in the fall general election. California ’s first experience with the top-two election system came in 2012. T o the surprise of many, turnout was the second lowest on record. This has raised a host of questions about the new reform and about turnout in primary elections in general . Turnout issues are especially important for the top two. It makes it easier for candi dates to build different coalitions of support, some of them across parties, making the complexion of the primary electorate potentially more important. It also sets the agenda for the fall election much more forcefully than the old system did. Only two ca ndidates will advance, and they are not even guaranteed to be from different parties . As a result, who votes in primary elections and who does not has suddenly taken on a new urgency. In this report, we examine voter turnout in California’s primary electio ns, both over time and in relation to other states. We discuss the factors influencing voter participation, policies that might improve turnout, and the ramifications of voter participation in primary elections, both generally and more specifically in the case of the new top -two primary system. We arrived at the following conclusions:  California’s turnout in primary elections has been declining over time, but remains one of the highest of any state .  California’s primary electorate is not representative of t he general electorate. It is older and less likely to be either Latino or Asian American. It is also typically more Republican, and this partisan bias interacts with the top two to create or prevent same party contests in the fall, where the outcome might be different had the primary electorate looked more like the general . The bias has already strongly contributed to one same -party contest in a competitive seat, and may lead to others in the future.  Compared to turnout in general elections, primary turnou t is driven far more by the dynamics of individual candidate races and the presence or absence of initiatives on the ballot. This helps explain the lower turnout in 2012, since the presidential contests had been decided by the time California held its prim ary. The influence of initiatives also suggests that California’s recent decision to move all citizen initiatives to the general election ballot might depress primary turnout .  Although turnout in 2012 was low, th e top two may increase primary participation on balance since independents now receive the same ballot as everyone else by default . This has made it easy for them to make choices in the legislative and congressional races covered under the top two, since they no longer need specifically request a ba llot that includes those contests. However, in the general election, another key feature of the reform appears to have discouraged vote choice, as many voters skipped the contests on the ballot where two Republicans or two Democrats faced off against each other .  Contrary to some expectations, the new registration reforms in California —online registration and same -day registration —may not increase voter turnout in the primaries; these two reforms do not appear to have boost ed primary participation in other states. T he top -two system currently bans independent or write- in candidacies in the fall election. G iven the partisan and demographic biases of the primary electorate and the aggressive way in which the top -two primary winnows the field, it would seem pru dent to have some option for a candidacy that could serve as a safety valve in the case of strange or unexpected outcomes in the primary . http:// www.ppic.org /main/home.asp Voter Turnout in Primary Elections 2 Contents Summary 2 Figures 4 Introduction 5 California’s Primary Turnout in Perspective 6 Open Primaries and Turnout 9 How Do Primary and General Electorates Differ? 12 Electoral Reform and Primary Turnout 15 Conclusions and Policy Directions 16 References 18 About the Author 19 Acknowledgments 19 T echnical appendi ces to this paper are available on the PPIC website: www.ppic.org/content/pubs/other/514EMR_appendix.pdf Figures 1. The gap between general and primary election turnout has been growing 6 2. California’s primary turnout has been high compared to other states 7 3. Independents have cast more votes under the blanket and top- two primary systems than under the semi -closed system 10 4. Voters casting ballots often skipped same -party contests 11 5. Fall electorates have generally been younger and more diverse than primary electorates 12 6. Fall electorates have typically voted more Democratic than primary electorates 13 http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 4 Introduction California launched a new open primary election system in 2012. Under the old system, voters who were registered with a party received a ballot that included only the candidates of that party, while voters registered as independents (officially known as “decline to state” ) could request a ballot for a particular party if the party decided to admit independents in that election . The downside of this system was that r egistered partisans in primary elections were able to vote only for candidates who were members of the same political party as themselves, and independents were often locked out of the vote entirely . Under the new “top -two” primary system, e very voter receives the same ballot and can vote for any candidate from any party for any office . The two candidates who receiv e the most vote s (again , regardless of party ) mov e on to the fall run-off election. In essence, t he new system convert s the primary into a firs t- stage general election, introducing the novel possibility of competition in the fall between candidates of the same political party. Many had hoped that the additional choices offered to voters in the top two system would encourag e participat ion in prim ary elections . Instead, turnout in the June 2012 primary was one of the lowest on record. Many questions have been asked in the wake of this outcome , while other questions that should have been asked have not been forthcoming . Was this low turnout brought on by the top-two system ? What are the political consequence s of lower turnout in the primary? Will California’s recent voter registration reforms — online registration and same -day registration —encourage participation in the primary stage ? Strong t urnout i n primaries has always been important (for one thing, smaller electorates are presumed to favor ideologues ). Yet in a more traditional primary system —including the one which the top-two system in California replaced—voters are presented with limited choices. For example, a registered Republican can vote only for a member of the Republican party, no matter what the office might be. From a candidate’s perspective, t his makes it difficult to build a coalition that reaches beyond core party identifiers . It also means that any differences between the partisan make- up of the primary electorate earlier in the year and the general electorate in the fall will have no consequence for candidate races, since the top vote -getter within each party always advance s to the f all election. The new top -two primary system has altered t his situation . Voters can now cross party lines to support candidates for reasons of ideology, personality, cultural affinity, or anything else they deem relevant. This greatly expands the number o f theoretically possible coalitions, making the composition of the electorate more important. The partisan composition becomes especially relevant compared to the old system, because the top two candidates always advance, even if they are of the same party . This report addresses t hese issues and others while exploring the topic of turnout in primary elections and the role of the top -two primary system. We begin by examining California’s turnout in primary elections , both over time and in relation to the ex periences of other states. We then discuss the impact of open primaries on turnout in primary elections in other states, and some special effects that the top two has had on participation in California. We also look at t he demographic and partisan differen ces between the primary and general electorates and the problems these differences create for the top-two system . Finally, the report looks at whether the new registration reforms are likely to have any impact on primary turnout moving forward. We conclude with thoughts on policy changes: both those that may be necessary, and those that probably are not. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 5 California’s Primary Turnout in Perspective Was the low turnout in California’s 2012 primary a result of the top -two primary, an aberration, or neither? Fi gure 1 shows the gap between turnout in California’s primaries and general elections as a share of all eligible voters ; larger gaps indicate primary turn out is falling farther behind. We show this information for presidential elections , as well as for midt erm elections when no presidential contest is on the ballot . It is clear that the gap has been growing, at least over the past few election cycles. (F igures B1 and B2 in Appendix B indicate that turnout has been declining in California’s primaries for decades, al though it has only recently grown worse relative to turnout in the general election.) Wh ile it may be too soon to consider this a clear trend, there is no question that voter turnout in primary elections has been heading in the wrong direction in recent years . F IGURE 1 The gap between general and primary election turnout has been growing SOURCE: California Secretary of State. At the same time, California’s primary turnout has been and continues to be among the highest in the nation —always in the top ten among states (Figure 2). And while California’s position slipped some what in 2012, it was still one of the best in the country. Thus, California’s primary turnout would seem to be only a relative problem in terms of voter participation within the state . 0 5 10 15 20 25 30 35 19801982198419861988199019921994199619982000200220042006200820102012 Turnout difference: general minus primary Presidential Midterm http:// www.ppic.org /main/home.asp Voter Turnout in Primary Elections 6   http://www.ppic.org/main/home.asp Voter Turnout in Primary Elections 7 FIGURE 2 California’s primary turnout has been high compared to other states   SOURCES: United States Elections Project (eligible voters, 1980- 2012); National Conference of State Legislatures (ballot measure outcomes); Congressional Quarterly Voting and Elections Collection (election outcomes); various secretaries of state (election outcomes da ta). It is  difficult  to  say  why  turnout  in  the  state’s  primary  has  been  higher  than  in  most  other  states.  In  many  ways,  California’s  demographic  profile  points  toward  lower  turnout:  the  state’s  population  is  highly  diverse   and  mobile  and  includes  significant  pockets  of  poor  and  less ‐educated  residents.  In  fact,  California’s  general   election  turnou t  is  only  average  and,  over  time,  has  been  slipping  compared  to other  states  (McGhee,  2014).   Election  dynamics  may  explain  some of  the  difference.  Research  has  generally  shown  that,  all  else  equal,   closer  general  elections  (including  closer  initiative  campaigns)  generate  higher  turnout  (Geys,  2006; Tolbert   et  al.,  2009;  Childers  and  B inder,  2012).  Our  analysis  suggests  the  same  is  true  for  primary  elections.  In fact,   the  dynamics  of  individual  campaigns  explain more  of the  turnout  in  primary  elections  than  in  general   elections.  In  general  elections,  differences  between  states tend  to  be  relatively  fixed  over  time, with  turnout   rising  and  falling  across  all  states  from  on e election  to the  next,  mostly  in  response  to the  presence  of  a   presidential  contest  on  the  ballot.  Primary  elections  are  far  more  idiosyncratic;  turnout  is harder  to predict  and   varies  more  in response  to  the  competitive  dynamics  of  the  races  on the  ballot  (see  analysis  in  Appendix  A ).   If  compet itive,  high ‐profile  elections  matter,  then  they  ought  to  be  drawing  people  to  the  polls  in  California.   Some  states  have mixed  caucus and  primary  systems  that  limit  voter choices,  or they  do  not  conduct  certain   high ‐profile  state  contests,  such  as the  election  of a  governor,  in the  same  ye ar as  federal  contests  for  the   presidency  or the  U.S.  Senate.  Such is  not  the  case  in  California,  and  our  statewide  primary  elections  for   governor,  U.S.  Senate,  and  president  have  been  slightly  more  competitive  than those  in other  states.  Even   over  time within  California,  turnout  has  been  higher  in years  wh en  our  primary  elections  have  been  more   competitive.  For  example,  turnout  was  high  in  1998,  when  there was a  hotly ‐contested  gubernatorial   primary,  and  in  the  presidential  primary of  2008,  when  both  parties  featured  very  competitive  presidential   contests.  And  California  has  experienced  lower  voter  turnout  in  those  elections  that  have  not  incl uded highly   competitive  contests.  For  example,  the  2008  election  featured  a  separate  June  primary  for  all  contests  except   the  nomination  for president,  and turnout  was  the lowest  on  record.  A  similar  case  can  be made  for  the   http://www.ppic.org/main/home.asp Voter Turnout in Primary Elections 8 June 2012  primary:  Although  the  presidential  contest  was  on the  ballot,  neither  the  Democratic  nor  the   Republican  nominations  were  in  doubt  at  the  time.    More  important,  California  has  been  one  of  only  seven  states  to  regularly  place  direct  democracy  measures   on  the  primary  ballot. 1 It  has  also  used  this  option  far  more  often, averaging  nine  measures  per  primary   election  compared  to  three  or  fewer  for all  other  states.  Over  the  past  30  years,  California’s  primary  turnout   has  been,  on  average,  about  10  points  higher  than  in  the  rest  of  the  nation.  The  presence  of  initiative   campaigns  on  the  ballot  appea rs  to  account  for  about  half  of  this  difference.  Neither  the  closeness  of  the   initiative  contest nor  the  number  of  initiatives  on  the  ballot  seems  to  matter  as  much  as  the  simple  fact  that at  least  one  initiative  is present  on the  ballot. 2 In  contrast,  while candidate  competition  explains  a  lot  of  the  ups   and  downs  across  states  and  over  time, it  does  not  help  explain  California’s  unusually  high  primary  turnout.   California’s  candidate  campaigns  are  only  marginally  more  competitive  than  those  in other  states,  not  enough  to  explain  the  difference.    The  effect  of  initiative  cam p aigns deserves  close  consideration  because  of  a  recent  policy  change.  The   California  Legislature,  through  SB  202  in 2011,  offered  a  new  interpretation  of  the  state  constitution  and   declared  that all  citizen  initiatives  must  appear  on  the  general  election  ballot. 3 As  a result,  the  June  2014   primary  will  be  the  first  in  decades  that  does  not  include  a statewide  citizen initiative.  However,  the  June   ballot  will  not  be  devoid  of  all  ballot  measures:  The  legislature  has  placed  two  of  its  own  proposals  on  the   ballot,  and  it  is  possible  that  these  two will  be  eno ugh to  draw  voters  to  the  polls.  However,  our  analysis   suggests  that  should  future  primary  elections  fail  to  include  any  initiatives  on  the ballot,  primary  turnout  is   likely  to  be  between  three and  seven  percentage  points  lower  than  it  might  otherwise  be. 4   1 The  other  states  are  Maine,  Missouri,  Nebraska,  North  Dakota,  Oklahoma,  and Oregon.  2 These  conclusions  are  based  on  a  regression  of  primary  turnout  on  the  margin  of  victory  in  all  initiative,  presidential,  gubernatorial,  and  U.S.   Senate  campaigns.  The  model  also  included  dummies  for  the  presence  of  each  type  of  campaign,  and  fixed  effects  for states  and  years.  Details  of  the  model  are  available  in  Appendix  B . We  al so tried  running  the  model  with  the  number  of  initiatives  in  addition  to  the  simple  dummy  for   presence  or  absence,  but  this  variable’s  coefficient  was  small,  statistically  insignificant,  and  bearing  the  wrong  sign,  while  the  dummy  remained   strong  and  significant.  This  suggests  that it  is  the  presence  rather  than  th e number  of  initiatives  that  matters.   3 Ballot measures  have  long  appeared  on  primary  election  ballots,  although  prior  to  1972  they would  appear on  the  ballot  as  “special”  elections  that  just  happened  to  coincide  with  the  scheduled  primary  election.  In  1972,  the  Secretary  of State  issued  an  opinion  interpreting  a  key  passage  in   the  constitution  as  allowing  initiatives  to  appear  on  the  prima ry ballot.  This  interpretation  was  occasionally  questioned  in  succeeding  years,   including  by  Chief  Justice  Rose  Bird  in  Brosnahan  v. Eu  [(1982)  31  Cal.3d  1  ],  where  she  argued  in  a  footnote  that  the  matter  deserved  revisiting.   For  a  brief  discussion  of  the  history,  see  the bill  analysis  accompanying  SB  202,  av ailable at   www.leginfo.ca.gov/pub/11 ‐12/bill/sen/sb_0201 ‐0250/sb_202_cfa_20110909_230915_sen_comm.html.   4 This  estimate  is  derived  from  the  95%  confidence  interval  around the  “ballot  measure  election”  coefficient  in  Model  3  of  Table  B1.  The  estimated   effect  of a  ballot  measure  election,  according  to  that  model,  is  5.13  +/‐ 1.76.    Open Primaries and Turnout California’s top-two primary system was adopted in part to encourage more voters to come to the polls. Yet turnout in 2012 was very low —the lowest, in fact, of any presidential primary in 90 years. As noted above, there are reasons to believe that this was at least partly due to low -key contests among the candidates for president and the U.S. Senate. But was it also an indictment of relying on open primaries to increase turnout? It certainly seems unli kely that expanding the primary electorate to include independents would actually discourage turnout. In California’s case, the change in the format of the primary was not as significant as it might at first appear, since independents were allowed some participation in primaries before 2012. Nonetheless, the question remains: Do open primaries have any consistent effect on turnout? Our analysis suggests that open primaries have not been associated with higher turnout in the states that have used them over t he past 30 years, at least in the cases where “open” meant any system that allowed independents to participate in some way in the election. While this curious result is fairly robust, we should be cautious about extrapolating too much from it. Part of the explanation is that independents do not participate as often in primary elections in the first place. 5 In California, where the rate of independent registration is about average for the nation as a whole, independents have never amounted to more than 14 percent of primary voters over the past five election cycles. And independents appear to represent an even smaller share of all registrants in closed primary states, meaning more of the electorate is composed of partisan voters who are more dependable about showing up at the polls.. 6 Thus, in open primary states, the potential electorate is much larger because independents can participate, but those extra potential voters do not turnout to vote at the same rates; in closed primary states, the potential electo rate is limited to partisan voters, but there are relatively more of them and they are more reliable at showing up. The two differences may neutralize each other and suggest no effect on turnout from open primaries. If overall turnout is about the same und er an open primary, turnout volatility is higher. In closed primary states, fixed differences between states explain much of the electorate’s size: States with high turnout in one election tend to have high turnout in the next. By contrast, the turnout in open primary states varies from election to election. The dynamics of competition appear to exp lain a lot of this difference: Once they are factored in, turnout in open primary states becomes about as predictable as in states where primaries are closed. 7 5 Developing a full explanation of the result is complicated by data limitations. While there has been considerable experimenta tion with open primaries at the legislative and congressional level (McGhee et al. 2014), that experimentation has often been concerned with the degree of openness rather than whether independents are allowed to partici pate at all. Moreover, there has been far less experimentation at the presidential level, and often by only one party in a particular state. Thus, the temporal variation necessary to explore the effect of ind ependent participation is less available. Finally, there is always the possibility that political parties have been willing to hold an open primary only when an election is not in doubt, which would likely result in a low turnout even though the system was more inclusive. Pulling apart this sort of se lection effect is ultimately impossible with the data at hand. 6 Many states with open primaries do not report registration by party, so there is no way to determine exactly how many independent registrants there might be if those states suddenly decided to force their citizens to choose. However, the independent registration rate in the open primary states with party registration is almost twice as high as in closed primary states (which need party registration in order to determine eligibility). This is tr ue whether independent registration is calculated as a share of total registrants (31% vs. 16%) or total eligible voters (23% vs. 12%). 7 These conclusions are based on a comparison of the difference in model fit between a null model with only state and y ear fixed effects and one with measures of competition included, and doing so separately for open and closed primary states. For closed primary states, the adjusted R2 was 0.59 for the null model and 0.64 for the full model, while the standard error of the regression was 5.60 for the null model and 5.23 for the full model. For open primary states, the adjusted R 2 was 0.45 for the null model and 0.62 for the full model, while the standard error of the regression was 6.69 for the null model and 5.62 for the f ull model. Thus, the full model explained about the same amount in both cases, but the null model was weaker for open primaries, suggesting that campaign dynamics account for a larger share of the variance in those states. As a placebo test http:// www.ppic.org /main/home.asp Voter Turnout in Primary Elections 9 Do these conclusions about open primaries extend to California’s new top-two primary? While the top -two system is “open” in the sense that it allows independents to participate, it also goes further by giving independents the same ballot as everyone else b y default. This makes it far easier for independents to participate in candidate contests; and in this respect, the open system more closely resembles the “blanket” primary California used in 1998 and 2000 than it does other types of open primary systems. By contrast, the “semi -closed” system California used between 2002 and 2010 required independents to request a ballot from one of the major parties if they wanted to vote for a partisan candidate. Only about half of the independent voters took advantage of this option ( McGhee, 2010 ). Figure 3 illustrates this difference in response: During the “semi -closed” years, t he share of ballots with a vote for president or U.S. House of Representatives was notably lower than in the years of the blanket or top -two primaries. 8 Note also the divergence in 2012, when voting for the House of Representatives shifted to the top -two s ystem while the presidential primary remained in the semi -closed system. Voters in 2012 actually ended up casting fewer votes for president than for the House, a remarkable outcome given the relative profile of the contests. With only one election under th e new system, it is not clear that this pattern will persist, but it is worth noting. FIGURE 3 Independents have cast more votes under the blanket and top-two primary systems than under the semi -closed system SOURCE: California Secretary of State. In sum, there is no clear evidence yet that open primaries produce a significantly higher turnout, partly because independents constitute such a small share of the primary electorate. However, on this first occasion of its use in California, the state’s particu lar approach to the open primary, in which independents received the same ballot as all other voters, did appear to get independents to cast more votes for the races other than president covered under the new system. Thus, the reform’s impact on the number of votes cast in an election may be quite positive. The true effect of the open primary in California will likely grow clearer as voters become more familiar with the reform and campaign consultants find better ways to mobilize voters under the new system . we ran the same analysis on general elections, where independents are always allowed to participate. We found virtually no difference between open and closed primary states. Model specifics are available in Appendix B . 8 Independents received a presidential ballot in 2000 and were allowed to vote for a candidate, but their votes were not counted in determining the outcome of the race . These votes are included in the 2000 presidential calculation fo r Figure 3. 60 65 70 75 80 85 90 95 100 1998 2000200220042006200820102012 Votes cast as a share of total turnout President House Open http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 10 Higher primary turnout was not the only goal of California’s new system. It was also meant to provide a better set of choices for the general election in heavily partisan districts where the fall outcome is typically a foregone conclusion. The new system led to same-party contests in the general election in 27 of these “safe - seat” districts, and many of these races ended up being quite competitive. In fact, about one- third of the same -party races were decided by less than 10 points, while only 15 percent of the races between candidates of different parties had less than a 10 -point spread. But this competition between same -party candidates did not lead to more votes in same- party contests in the fall. Quite the opposite: Between the primary and general el ections, same-party contests gained 8 percent fewer votes than cross -party races. 9 As just noted, this was not because the same -party contests were less competitive (see Table B5 in Appendix B for more evidence on this point). Nor was it because the type of districts with same -party contests had lower turnout in the fall for other reasons. In fact, the increase in the percentage of ballots cast was identical on average in both same -p arty and cross -party races; the only difference was the number of votes cast in the same -party race s (Figure 4) . Put differently, the electorate consistently voted for candidates running for higher offices, such as the U.S. Senate and the president, but of ten skipped same -party races further down the ballot. Since we don’t have data on individual voter decisions, it’s not possible to say anything about the characteristics of the voters who skipped these contests, but we can confidently say that, on balance, fewer felt the need to offer a vote in those races. FIGURE 4 Voters casting ballots often skipped same -party contests SOURCES: Political Data Incorporated (ballots cast); California Secretary of State (votes cast). NOTE: Results are for races that eith er became or failed to become same- party races by fewer than five percentage points, in order to account for other differences between same- party and cross-party districts. Numbers also reflect changes from the primary to the general election to further account for the possibility of different average turnout in the types of districts more likely to have same -party races. 9 These estimates are derived from a regression discontinuity analysis. Many fall contests became same -party contests by a razor-thin margin in the primary, raising the prospect that the same -party outcome was effectively random in those cases . We leveraged this fact, using the margin by which a seat became same -party (or did not) as the forcing variable which assigned a fall contest to same -party status. The result of this analysis suggested that the number of votes as a share of total registe red voters was 8 percent lower in same-party contests, a statistically-significant difference. We used software from Rocio Titiunik to implement a randomization inference approach to calculating standard erro rs, choosing a margin of five percentage points above or below the threshold for the sake of analyzing the results. Details of these estimations are available from the author on request. For details on the randomization inference method, see Cattaneo et al. (2013). 0 5 10 15 20 25 30 35 40 45 Ballots Cast Votes Cast Participation: general minus primary Cross party Same party http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 11 How Do Primary and General Electorates Differ? Thus far we have shown that turnout in California’s primary elections has been declining o ver time relative to turnout in the general elections, and that the state’s new open primary system may not be able to reverse this trend. Is the smaller electorate that turns out in the primary simply a microcosm of the larger one that turns out in the fa ll? In terms of basic demographics, there is little question that primary electorates have included fewer young people, Latinos, and Asian Americans than the general electorate in the fall (Figure 5). The difference is largest for Latinos, whose share of the electorate has been about seven percentage points higher in the two most recent fall presidential elections. But there are clear differences for young people and Asian Americans as well. 10 To the extent that these groups might support different candidates for office —either of different parties or within the same party —these variations in turnout are important. In fact, there may be many more differences between the primary and general electorates than discussed here, but these are just the differences in dicated by the data available in the voter registration file. 11 Even within demographic groups, those who vote in the primary might differ on policy issues or general ideology. FIGURE 5 Fall electorates have generally been younger and more diverse than pri mary electorates SOURCES: Statewide Database (2004 -2012 primary electorate); Political Data Incorporated (2012 general electorate). Partly because of the demographic differences highlighted above, there are also partisan differences between the primary and general electorates. The fall electorate has been notably more Democratic than the primary 10 These differences in magnitude are mostly a function of the size of each group. Proportional to size, the differences in turnout between primary and general elections are similar across all three groups. However, since the difference i n each group’s share of the electorate is what matters for election outcomes, we illustrate the respective shares in Figure 5. 11 The only other demographic category for which we have data across all five of these election cycles is gender, and it shows no difference between primary and general electorates. Other important demographic variables, such as education, income, or certain racial categorizations (African American, white) are not available in the voter registration file. +3.6 +1.2 +5.5 +2.7 +3.8 +1.3 +0.6 +0.9 +0.6 +2.0 +4.2 +2.0 +7.2 +5.4 +7.3 0 1 2 3 4 5 6 7 8 2004 2006200820102012 20042006200820102012 20042006200820102012 Percent change in share of electorate: primary to general Age 18–24 Asian/Pacific Islander Latino http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 12 electorate in six of the last eleven election cycles, and at least slightly more Democratic in all but two (Figure 6). The difference is about 3 percentage points for a seat balanced perfectly between the two parties in the primary stage. 12 These estimates are based on the actual votes cast, so they incorporate any effects from the growing population of independent registrants. FIGURE 6 Fall electorates have typically voted more Democratic than primary electorates SOURCE: California Secretary of State. This primary-to -general shift in partisan turnout explained about 92 percent of the variation in fall outcomes in 2012, missing the actual result by an average of only 5 percentage points. 13 This is remarkably accurate considering the radical winnowing of the candidate field between the two sets of elections. It suggests party loyalties still played a large role under the top -two system in its debut. In the past, the consequences of this partisan shift have been minimal for candidate races, because every party that ran at least one candidate in the primary could be assured of a place on the fall ballot. 14 But the top two primary upends this logic. There is no guarantee now that any party will be present in the fall—only that the top two vote getters will advance. It is entirely possible that the difference between the primary and general electorates will end up closing off the possibility of a cross -party contest in the fall where one might otherwise have occurred. Since the fall electorate is more Democratic than the primary electorate, this difference in turnout should usually produce more same -party contests between Republicans and fewer 12 The shift is smaller as po litical seats become more heavily Democratic. While there are more voters of every partisan stripe in the general election, the relative increase for Democrats, and especially for Independents, is much higher than for Republicans. As a result, the percenta ge of fall voters who are Democrats is usually about as large as in the primary, while the percentage of Independents is much highe r and the percentage of Republicans much lower. In heavily Democratic seats, there are relatively fewer Republicans to produc e this discrepancy, so the change between the primary and the general elections is smaller. 13 To calculate this number, we regressed the partisan vote in the primary on the party registration of voters in the primary el ectorate, and then predicted fall out comes with this equation, using the party registration from the fall electorate. Thus, the method assumes that the relationship between party registration and party voting is the same in both the primary and general election, and the only difference is the partisan composition of the voters. We then calculated the difference between this prediction and the actual fall outcome, and treated that as the residual for the calculation of an R -squared statistic. For the purposes of this calculation, we included on ly those races with at least one candidate from each major party printed on the ballot in both the primary and fall election, which eliminated 41 races. 14 This was true even for the blanket primary in 1998 and 2000, which was otherwise quite similar to the top - two primary. Indeed, the fact that every party is represented in the fall in a blanket primary is the main difference between the two systems. +0.8 -3.4 +0.7 +3.0 +3.2 +2.2 +0.7 -0.9 +3.7 +3.0 +3.2 -4 -3 -2 -1 0 1 2 3 4 5 1992 1994199619982000200220042006200820102012 Percent Democratic vote shift: primary to general http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 13 between Democrats than would otherwise be the case (although in at least one election in Figure 6, the opposite would have occurred). How large is this effect? We can simulate how the primary electorate in 2012 might have voted had it had the same partisan composition as the general el ectorate in 2012. The result is that there would be very little change in the total number of same -party contests, but there would be a significant shift in their partisan complexion: The simulation produces three same- party Republican races and 26 Democratic races, compared to the nine Republican and 19 Democratic that actually occurred. As in the case of the same- party races, the districts where the same- party status was most likely to change were almost exclusively “safe” for one party, so the partisan o utcome in the fall contest was not in doubt regardless. The one exception is Congressional District 31, where the actual fall contest pitted two Republicans against each other in a seat that was otherwise ripe for a competitive race. In this seat, the simu lation suggested a 97 percent chance that a Democrat would have faced off against a Republican if the primary electorate had been more representative of the fall electorate. We have already noted that the absence of ballot measures from the primary ballot may affect turnout in future primaries. Will it also exacerbate the partisan bias of the primary electorate? It turns out that over the 11 election cycles in Figure 5, there is little or no relationship between the size of the increase in the primary to g eneral turnout and the size or direction of the partisan shift (see Figure B3 in Appendix B ). For example, the primary with the best turnout relative to the general election (1998) saw a partisan difference of 3 percent, compared to 3.7 percent for the primary -to -general with the worst gap (2008). However, for the five recent elections for which we have data on demographics, a higher primary -general increase in turnout does correspo nd with a larger increase in turnout by young people and Latinos (the relationship for Asian Americans is more mixed; see Figure B4 in Appendix B ). Because this evidence is based on such a limited number of election years, we should treat it as tentative. But it does seem as though lower primary turnout corresponds with an older, less -diverse electorate, without necessarily making it more Republican. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 14 Electoral Reform and Primary Turnout California has recently passed two important voter registration reforms meant in part to encourage more turnout in state elections. The first, online registration, was authorized first by SB 381 in 2008, and then expedited by SB 397 in 2011. 15 This re form allows users to enter voter registration information on -line and, with a keystroke, complete the process, eliminating pen and paper transactions. The new system was used heavily in its inaugural election in 2012. The second reform, “conditional” voter registration, was established by AB 1436 in 2012. It allows residents to both register and vote on any day after the traditional registration deadline, up to and including Election Day itself. The main constraint is that valid conditional registration mus t take place at a county registrar office. This option will not be available until 2016 at the earliest, but many observers are eagerly anticipating a significant impact once the system is implemented. Several existing analyses of these reforms suggest th at they have not had, and may not have, much impact on general election turnout, although a variety of administrative effects have been and could be significant ( McGhee, 2014) . We conducted a similar analysis of primary turnout across the nation over the past 30 years and found no effect of these reforms. On line registration might have increased primary turnout about two percentage points for states that have adopted the reform, but the increase is well within the margin of error. Moreover, primary turnout has been, if anything, slightly lower in the states where election-day registration has been used, although this likely says more about the particular states and years where the system has been adopted than the causal effect of election -day registration itself. Thus, it seems doubtful that the new reforms wi ll significantly increase California’s primary election turnout. 15 SB 381 authorized an online registration system, but made it conditional on completion of the state’s Vot eCal database of voter registration records merged with Department of Motor Vehicle files. VoteCal has since become delayed until 2015 at the earliest, so SB 397 authorized the Secretary of State to use a different process to implement an online registrati on system. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 15 Conclusions and Policy Directions This report has looked at the causes and consequences of turnout in California’s primary elections. Turnout in primaries has, in fact, declined over the past few decades, even in comparison with turnout in the general election; and there are reasons to believe that turnout will continue to fall, since initiatives may no longer appear on the primary ballot. Furthermore, key policy interventions intended to b oost primary turnout may not have much effect. California’s new top -two primary failed to produce the increase in turnout that many had hoped for, and there is little evidence that open primaries in other states have fared any better. Independents appear t o be fickle primary voters, inclined to participate only when a ballot includes a close race. Recent efforts to increase turnout by making registration easier —such as online and same -day registration —also appear to have no meaningful impact on primary turn out. On the other hand, the top -two system may encourage those independents who do vote in the primary to make choices in candidate contests for State Legislature and U.S. Congress, since they now receive the same ballot as everyone else by default. When t he weak effect on turnout is combined with the positive effect on down -ballot voting, the net effect of the top -two primary on participation is probably positive. Moreover, it would be counterproductive to increase primary turnout by repealing SB 202 and o nce again permitting citizen initiatives on the primary ballot. California has always placed far more initiatives on the primary ballot than other states. The clear differences between the primary and general electorates documented in this report should ra ise equity concerns about this practice. It seems inappropriate to allow a subset of the electorate to decide major policy questions for the rest of the state. Of course, the cost of a more representative electorate for initiative questions may be a small er primary electorate for candidate races. But does that matter? The state’s primary turnout has been falling, to be sure, yet even with the decline it remains among the highest in the nation. And it is worth noting that research on primary turnout and pol arization has found no real link between the two, at least in the case of U.S. Senate elections ( Hirano et al., 2010). Thus, lower turnout is not necessarily producing more radical representation. That said, lower turnout may produce an older and whiter electorate than would otherwise be the case, affecting the choice of candidates advan cing to the fall contest. And even though lower -turnout primaries do not seem to consistently benefit one party or the other compared to higher -turnout primaries, the lower turnout in primaries as a whole does produce a fairly consistent Republican tilt (a lthough one election cycle in our analysis had a primary electorate that skewed toward the Democratic candidates). This bias creates some problems for the top -two primary system, which winnows the field much more aggressively than the old system, and witho ut regard to party. There has already been one competitive seat with a same -party contest in the fall, and we have shown that the outcome almost certainly would have been different if the fall electorate had shown up earlier in the primary. Although such o utcomes will probably never be common under the top two, the fact that one already happened in the first election under the new system suggests it may easily happen again. 16 Moreover, it appears these same- party contests are of less interest to voters on av erage than are races between two candidates of different parties. The prospect that the skewed primary electorate might lead to more 16 The major parties can certainly try to address this problem more informally by restricting the number of candidates within their party. For example, if only one Democrat had run in Congressional District 31 instead of the four who actually did , the Democratic vote would likely not have been split between so many candidates and there would have been a cross -party contest in the fall. But there are no certain methods of clearing a candidate field in American politics. Encouraging major party dono rs and officeholders to coordinate their efforts on the behalf of one candidate will often work to discourage others, but sometimes it will not. It would be useful to have a safety valve that would allow a party to place a candidate directly on the fall ba llot in case such coordination fails. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 16 same-party contests in competitive seats, or even that such an outcome might occur in a statewide race, should not be dismi ssed. 17 One possibility worth exploring would be to allow more than two candidates in the fall contest, at least in some cases. At present, the top -two primary system bans more than two candidates in the fall, including independent candidacies or write -ins . In other states, this sort of provision is often called a “sore loser” law because it supposedly prevents a losing primary candidate from entering the race in the fall election and wreaking havoc for the sake of ego or revenge. While there may be a publi c interest in preventing too many such candidacies —especially under a system like the top two that is so committed to a two -candidate fall race —it might also be prudent to have a way of avoiding same -party contests when they might not reflect the choices g eneral election voters would like to see. A write -in option, or the option to launch an independent petition- drive candidacy, could provide a sort of “safety valve” in cases of such unexpected outcomes, while also ensuring that most races would still pit o nly two candidates against each other. Indeed, California law already allows for both options in presidential contests (California Elections Code, Sec. 8300 and 8600). The effects of such a change could extend beyond candidate competition. While research suggests that closed primaries do not increase polarization (McGhee et al., 2014 ), recent researc h has suggested that “sore loser” laws have increased polarization by preventing candidates with broader appeal from running directly in the fall ( Burden et al., 2014). We should be cautious about concluding too much from this research, since it did not look strictly at t he top-two system (nor could it, given the data available). But it offers some suggestion that an independent or write -in option (allowing for a safety valve candidacy) might not only enable the system to address quirky outcomes but also advance the cause of moderate representation that the top -two system was meant to promote. 18 The specifics of such a change would be important. The requirements for either a write- in or an independent candidacy option would need to be set quite high to ensure that these options would be used only in cases where the potential demand for another candidate was strong. For example , it currently takes 40 signatures to be a write -in candidate on the primary ballot (and have one’s votes counted), and it takes one percent of all registered voters in the last election to be an independent candidate for president. This second number, though much higher, may still be too low to serve as a proper barrier in the context of the top two. Something closer to the signature requirement for initiatives—for example, 5 percent of the last vote for the office in quest ion (Assembly, State Senate, Congress , or statewide race)— may be closer to the mark. Regardless, further exploration of the idea would be needed to assess its viability. Should the ide a be worth pursuing, however, it would not require the extra hurdles of a state constitutional amendment. T he ban s on write -ins and independent petition candidates are statutory provision s that were added by the legislature through SB 6 (2009) , so they could be changed through another bill passed by the legislature or a statutory initiative placed before the voters . In sum, turnout in California’s primaries is falling but remains high compared to the rest of the nation. The interventions proposed to increas e turnout may not prove particularly effective, although the top -two system appears to have already encouraged more participation by independents in the primary candidate contests. The most serious consideration moving forward is what to do about the inter action between the differences in the primary and general electorates under the top -two system. 17 Note that this possibility differs from an uncontested race, where no candidate comes forward in the first place. Uncontested races are much more a product of political dynamics, while the sort of anomalous same -part y races described here are largely a product of the mechanics of the top -two system. Put differently, an uncontested race will always have the same outcome, regardless of the primary system in use. 18 This approach might also help manage cases where a cand idate for a top-two office dies during the course of the election season. Existing law provides no means of replacing such a candidate; the candidate’s name remains on the ballot, and if that candidate is ultimat ely “elected,” a special election is immedia tely called to fill the vacancy (California Election Code 8803(b), 8805(b)). http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 17 References Burden, Barry C., Bradley Jones, and Michael S. Kang. 2014. “Sore Loser Laws and Congressional Polarization. ” University of Wisc onsin, Madison. Cattaneo, Matias D., Brigham Frandsen, and Rocio Titiunik. 2013. “Randomization Inference in the Regression Discontinuity Design: An Application to Party Advantages in the U.S. Senate. ” Un iversity of Michigan. Childers, Matthew, and Mike Binder. 2012. “Engaged by the Initiative? How the Use of Citizen Initiatives Increases Voter Turnout. ” Pol itical Research Quarterly 65(1): 93 –103. Geys, Benny. 2006. “Explaining Voter Turnout: A Review of the Aggregate-Level Research. ” Electoral Studies 25: 637– 663. Hirano, Shigeo, James M. Snyder Jr., Stephen Ansolabehere, and John Mark Hansen. 2010. “Primary Elections and Partisan Polarization in the U.S. Congress. ” Quarterly Jour nal of Political Science 5(2): 169– 191. McGhee, Eric. 2010. At Issue: Open Primaries. San Francisco: Public Policy Institute of California. Available at www.ppic.org/main/publication.asp?i=904 . McGhee, Eric. 2014. Expanding California's Electorate: Will Recent Reforms Increase Voter Turnout? San Francisco: Public Policy Institute of California. Available at www.ppic.org/main/publication.asp?i=1083 . McGhee, Eric, Seth Masket, Boris Shor, Steven Rogers, and Nolan McCarty. 2014. “ A Primary Cause of Partisanship? Nomination Systems and Legislator Ideology.” American Journal of Political Science forthcoming Tolbert, Caroline J., Daniel C. Bowen, and Todd Donovan. 2009. “ Initiative Campaigns: Direct Democracy and Voter Mobilization. ” American Politics Research 37(1): 155– 192. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 18 About the Author Eric McGhee is a research fellow at PPIC. His work focuses on electio ns, political participation, political polarization, legislative behavior, redistricting, and surveys a nd polling. Before joining PPIC, he was a professor of political science at the University of Oregon . He holds a Ph.D. in political science from the Univ ersity of California, Berkeley. Acknowledgments The author would like to thank several people outside PPIC for their help in making this report possible: Darren Chesin, Ethan Jones, Paul Mitchell, and Nicole Winger all offered expertise and insights. Paul Mitchell deserves special credit for the rich resource of registration data he made available for analysis. An earlier version of the report benefited greatly from a careful reading by Dean Bonner, Benjamin Highton, Tony Quinn, Karthick Ramakrishnan , and Lynette Ubois , and Gary Bjork and Kate Reber helped clean up the final draft for public presentation. As always, thanks to Mark Baldassare for his continuing commitment to both this report and a neutral, nonpartisan agenda of research on California’s polit ical reforms. Any errors are my own. http:// www.ppic.org /main/home.asp Voter Turnout in Primary Elections 19 PUBLIC POLICY INSTITUTE OF CALIFORNIA Board of Directors Donna Lucas, Chair Chief Executive Officer Lucas Public Affairs Mark Baldassare President and CEO Public Policy Institute of California Ruben Barrales President and CEO GROW Elect María Blanco Vice President, Civic Engagement California Community Foundation Brigitte Bren Attorney Walter B. Hewlett Member , Board of Directors The William and Flora Hewlett Foundation Phil Isenberg Vice Chair Delta Stewardship Council Mas Masumoto Author and Farmer Steven A. Merksamer Senior Partner Nielsen, Merksamer, Parrinello, Gross & Leoni, LLP Kim Polese Chairman ClearStreet, Inc. Thomas C. Sutton Retired Chairman and CEO Pacific Life Insurance Company T he Public Policy Institute of California is dedicated to informing and improving public policy in California through independent, objective, nonpartisan research on major economic, social, and political issues. The institute’s goal is to raise public aware ness and to give elected representatives and other decisionmakers a more informed basis for developing policies and programs. The institute’s research focuses on the underlying forces shaping California’s future, cutting across a wide range of public polic y concerns, including economic development, education, environment and resources, governance, population, public finance, and social and health policy. PPIC is a public charity. It does not take or support positions on any ballot measures or on any local, state, or federal legislation, nor does it endorse, support, or oppose any political parties or candidates for public office. PPIC was established in 1994 with an endowment from William R. Hewlett. Mark Baldassare is President and Chief Executive Officer o f PPIC. Donna Lucas is Chair of the Board of Directors. Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source. Research publications reflect the views of th e authors and do not necessarily reflect the views of the staff, officers, or Board of Directors of the Public Policy Institute of California. Copyright © 201 4 Public Policy Institute of California All rights reserved. San Francisco, CA PUBLIC POLICY INSTITUTE OF CALIFORNIA 500 Washington Street, Suite 600 San Francisco, California 94111 phone: 415.291.4400 fax: 415.291.4401 www.ppic.org PPIC SACRAMENTO CENT ER Senator Office Building 1 121 L Street, Suite 801 Sacramento, California 95814 phone: 916.440.1120 fax: 916.440.1121" } ["___content":protected]=> string(102) "

R 514EMR

" ["_permalink":protected]=> string(77) "https://www.ppic.org/publication/voter-turnout-in-primary-elections/r_514emr/" ["_next":protected]=> array(0) { } ["_prev":protected]=> array(0) { } ["_css_class":protected]=> NULL ["id"]=> int(8909) ["ID"]=> int(8909) ["post_author"]=> string(1) "1" ["post_content"]=> string(0) "" ["post_date"]=> string(19) "2017-05-20 02:42:00" ["post_excerpt"]=> string(0) "" ["post_parent"]=> int(4355) ["post_status"]=> string(7) "inherit" ["post_title"]=> string(8) "R 514EMR" ["post_type"]=> string(10) "attachment" ["slug"]=> string(8) "r_514emr" ["__type":protected]=> NULL ["_wp_attached_file"]=> string(12) "R_514EMR.pdf" ["wpmf_size"]=> string(6) "484679" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(58191) "Voter Turnout in Primary Elections May 2014 Eric McGhee with research support from Daniel Krimm Supported with funding from the S. D. Bechtel, Jr. Foundation Summary In June 2010, California voters passed Proposition 14 to authorize a new “top -two ” primary election system. Previously, those registered with a party could only vote in that party’s primary, though independents were usually allowed to choose a primary they wanted to join for that election. Now under the top -two system , voters can vote for any candidate from any party , and the two candidates who receiv e the most votes , regardless of party affiliation, compete against each other in the fall general election. California ’s first experience with the top-two election system came in 2012. T o the surprise of many, turnout was the second lowest on record. This has raised a host of questions about the new reform and about turnout in primary elections in general . Turnout issues are especially important for the top two. It makes it easier for candi dates to build different coalitions of support, some of them across parties, making the complexion of the primary electorate potentially more important. It also sets the agenda for the fall election much more forcefully than the old system did. Only two ca ndidates will advance, and they are not even guaranteed to be from different parties . As a result, who votes in primary elections and who does not has suddenly taken on a new urgency. In this report, we examine voter turnout in California’s primary electio ns, both over time and in relation to other states. We discuss the factors influencing voter participation, policies that might improve turnout, and the ramifications of voter participation in primary elections, both generally and more specifically in the case of the new top -two primary system. We arrived at the following conclusions:  California’s turnout in primary elections has been declining over time, but remains one of the highest of any state .  California’s primary electorate is not representative of t he general electorate. It is older and less likely to be either Latino or Asian American. It is also typically more Republican, and this partisan bias interacts with the top two to create or prevent same party contests in the fall, where the outcome might be different had the primary electorate looked more like the general . The bias has already strongly contributed to one same -party contest in a competitive seat, and may lead to others in the future.  Compared to turnout in general elections, primary turnou t is driven far more by the dynamics of individual candidate races and the presence or absence of initiatives on the ballot. This helps explain the lower turnout in 2012, since the presidential contests had been decided by the time California held its prim ary. The influence of initiatives also suggests that California’s recent decision to move all citizen initiatives to the general election ballot might depress primary turnout .  Although turnout in 2012 was low, th e top two may increase primary participation on balance since independents now receive the same ballot as everyone else by default . This has made it easy for them to make choices in the legislative and congressional races covered under the top two, since they no longer need specifically request a ba llot that includes those contests. However, in the general election, another key feature of the reform appears to have discouraged vote choice, as many voters skipped the contests on the ballot where two Republicans or two Democrats faced off against each other .  Contrary to some expectations, the new registration reforms in California —online registration and same -day registration —may not increase voter turnout in the primaries; these two reforms do not appear to have boost ed primary participation in other states. T he top -two system currently bans independent or write- in candidacies in the fall election. G iven the partisan and demographic biases of the primary electorate and the aggressive way in which the top -two primary winnows the field, it would seem pru dent to have some option for a candidacy that could serve as a safety valve in the case of strange or unexpected outcomes in the primary . http:// www.ppic.org /main/home.asp Voter Turnout in Primary Elections 2 Contents Summary 2 Figures 4 Introduction 5 California’s Primary Turnout in Perspective 6 Open Primaries and Turnout 9 How Do Primary and General Electorates Differ? 12 Electoral Reform and Primary Turnout 15 Conclusions and Policy Directions 16 References 18 About the Author 19 Acknowledgments 19 T echnical appendi ces to this paper are available on the PPIC website: www.ppic.org/content/pubs/other/514EMR_appendix.pdf Figures 1. The gap between general and primary election turnout has been growing 6 2. California’s primary turnout has been high compared to other states 7 3. Independents have cast more votes under the blanket and top- two primary systems than under the semi -closed system 10 4. Voters casting ballots often skipped same -party contests 11 5. Fall electorates have generally been younger and more diverse than primary electorates 12 6. Fall electorates have typically voted more Democratic than primary electorates 13 http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 4 Introduction California launched a new open primary election system in 2012. Under the old system, voters who were registered with a party received a ballot that included only the candidates of that party, while voters registered as independents (officially known as “decline to state” ) could request a ballot for a particular party if the party decided to admit independents in that election . The downside of this system was that r egistered partisans in primary elections were able to vote only for candidates who were members of the same political party as themselves, and independents were often locked out of the vote entirely . Under the new “top -two” primary system, e very voter receives the same ballot and can vote for any candidate from any party for any office . The two candidates who receiv e the most vote s (again , regardless of party ) mov e on to the fall run-off election. In essence, t he new system convert s the primary into a firs t- stage general election, introducing the novel possibility of competition in the fall between candidates of the same political party. Many had hoped that the additional choices offered to voters in the top two system would encourag e participat ion in prim ary elections . Instead, turnout in the June 2012 primary was one of the lowest on record. Many questions have been asked in the wake of this outcome , while other questions that should have been asked have not been forthcoming . Was this low turnout brought on by the top-two system ? What are the political consequence s of lower turnout in the primary? Will California’s recent voter registration reforms — online registration and same -day registration —encourage participation in the primary stage ? Strong t urnout i n primaries has always been important (for one thing, smaller electorates are presumed to favor ideologues ). Yet in a more traditional primary system —including the one which the top-two system in California replaced—voters are presented with limited choices. For example, a registered Republican can vote only for a member of the Republican party, no matter what the office might be. From a candidate’s perspective, t his makes it difficult to build a coalition that reaches beyond core party identifiers . It also means that any differences between the partisan make- up of the primary electorate earlier in the year and the general electorate in the fall will have no consequence for candidate races, since the top vote -getter within each party always advance s to the f all election. The new top -two primary system has altered t his situation . Voters can now cross party lines to support candidates for reasons of ideology, personality, cultural affinity, or anything else they deem relevant. This greatly expands the number o f theoretically possible coalitions, making the composition of the electorate more important. The partisan composition becomes especially relevant compared to the old system, because the top two candidates always advance, even if they are of the same party . This report addresses t hese issues and others while exploring the topic of turnout in primary elections and the role of the top -two primary system. We begin by examining California’s turnout in primary elections , both over time and in relation to the ex periences of other states. We then discuss the impact of open primaries on turnout in primary elections in other states, and some special effects that the top two has had on participation in California. We also look at t he demographic and partisan differen ces between the primary and general electorates and the problems these differences create for the top-two system . Finally, the report looks at whether the new registration reforms are likely to have any impact on primary turnout moving forward. We conclude with thoughts on policy changes: both those that may be necessary, and those that probably are not. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 5 California’s Primary Turnout in Perspective Was the low turnout in California’s 2012 primary a result of the top -two primary, an aberration, or neither? Fi gure 1 shows the gap between turnout in California’s primaries and general elections as a share of all eligible voters ; larger gaps indicate primary turn out is falling farther behind. We show this information for presidential elections , as well as for midt erm elections when no presidential contest is on the ballot . It is clear that the gap has been growing, at least over the past few election cycles. (F igures B1 and B2 in Appendix B indicate that turnout has been declining in California’s primaries for decades, al though it has only recently grown worse relative to turnout in the general election.) Wh ile it may be too soon to consider this a clear trend, there is no question that voter turnout in primary elections has been heading in the wrong direction in recent years . F IGURE 1 The gap between general and primary election turnout has been growing SOURCE: California Secretary of State. At the same time, California’s primary turnout has been and continues to be among the highest in the nation —always in the top ten among states (Figure 2). And while California’s position slipped some what in 2012, it was still one of the best in the country. Thus, California’s primary turnout would seem to be only a relative problem in terms of voter participation within the state . 0 5 10 15 20 25 30 35 19801982198419861988199019921994199619982000200220042006200820102012 Turnout difference: general minus primary Presidential Midterm http:// www.ppic.org /main/home.asp Voter Turnout in Primary Elections 6   http://www.ppic.org/main/home.asp Voter Turnout in Primary Elections 7 FIGURE 2 California’s primary turnout has been high compared to other states   SOURCES: United States Elections Project (eligible voters, 1980- 2012); National Conference of State Legislatures (ballot measure outcomes); Congressional Quarterly Voting and Elections Collection (election outcomes); various secretaries of state (election outcomes da ta). It is  difficult  to  say  why  turnout  in  the  state’s  primary  has  been  higher  than  in  most  other  states.  In  many  ways,  California’s  demographic  profile  points  toward  lower  turnout:  the  state’s  population  is  highly  diverse   and  mobile  and  includes  significant  pockets  of  poor  and  less ‐educated  residents.  In  fact,  California’s  general   election  turnou t  is  only  average  and,  over  time,  has  been  slipping  compared  to other  states  (McGhee,  2014).   Election  dynamics  may  explain  some of  the  difference.  Research  has  generally  shown  that,  all  else  equal,   closer  general  elections  (including  closer  initiative  campaigns)  generate  higher  turnout  (Geys,  2006; Tolbert   et  al.,  2009;  Childers  and  B inder,  2012).  Our  analysis  suggests  the  same  is  true  for  primary  elections.  In fact,   the  dynamics  of  individual  campaigns  explain more  of the  turnout  in  primary  elections  than  in  general   elections.  In  general  elections,  differences  between  states tend  to  be  relatively  fixed  over  time, with  turnout   rising  and  falling  across  all  states  from  on e election  to the  next,  mostly  in  response  to the  presence  of  a   presidential  contest  on  the  ballot.  Primary  elections  are  far  more  idiosyncratic;  turnout  is harder  to predict  and   varies  more  in response  to  the  competitive  dynamics  of  the  races  on the  ballot  (see  analysis  in  Appendix  A ).   If  compet itive,  high ‐profile  elections  matter,  then  they  ought  to  be  drawing  people  to  the  polls  in  California.   Some  states  have mixed  caucus and  primary  systems  that  limit  voter choices,  or they  do  not  conduct  certain   high ‐profile  state  contests,  such  as the  election  of a  governor,  in the  same  ye ar as  federal  contests  for  the   presidency  or the  U.S.  Senate.  Such is  not  the  case  in  California,  and  our  statewide  primary  elections  for   governor,  U.S.  Senate,  and  president  have  been  slightly  more  competitive  than those  in other  states.  Even   over  time within  California,  turnout  has  been  higher  in years  wh en  our  primary  elections  have  been  more   competitive.  For  example,  turnout  was  high  in  1998,  when  there was a  hotly ‐contested  gubernatorial   primary,  and  in  the  presidential  primary of  2008,  when  both  parties  featured  very  competitive  presidential   contests.  And  California  has  experienced  lower  voter  turnout  in  those  elections  that  have  not  incl uded highly   competitive  contests.  For  example,  the  2008  election  featured  a  separate  June  primary  for  all  contests  except   the  nomination  for president,  and turnout  was  the lowest  on  record.  A  similar  case  can  be made  for  the   http://www.ppic.org/main/home.asp Voter Turnout in Primary Elections 8 June 2012  primary:  Although  the  presidential  contest  was  on the  ballot,  neither  the  Democratic  nor  the   Republican  nominations  were  in  doubt  at  the  time.    More  important,  California  has  been  one  of  only  seven  states  to  regularly  place  direct  democracy  measures   on  the  primary  ballot. 1 It  has  also  used  this  option  far  more  often, averaging  nine  measures  per  primary   election  compared  to  three  or  fewer  for all  other  states.  Over  the  past  30  years,  California’s  primary  turnout   has  been,  on  average,  about  10  points  higher  than  in  the  rest  of  the  nation.  The  presence  of  initiative   campaigns  on  the  ballot  appea rs  to  account  for  about  half  of  this  difference.  Neither  the  closeness  of  the   initiative  contest nor  the  number  of  initiatives  on  the  ballot  seems  to  matter  as  much  as  the  simple  fact  that at  least  one  initiative  is present  on the  ballot. 2 In  contrast,  while candidate  competition  explains  a  lot  of  the  ups   and  downs  across  states  and  over  time, it  does  not  help  explain  California’s  unusually  high  primary  turnout.   California’s  candidate  campaigns  are  only  marginally  more  competitive  than  those  in other  states,  not  enough  to  explain  the  difference.    The  effect  of  initiative  cam p aigns deserves  close  consideration  because  of  a  recent  policy  change.  The   California  Legislature,  through  SB  202  in 2011,  offered  a  new  interpretation  of  the  state  constitution  and   declared  that all  citizen  initiatives  must  appear  on  the  general  election  ballot. 3 As  a result,  the  June  2014   primary  will  be  the  first  in  decades  that  does  not  include  a statewide  citizen initiative.  However,  the  June   ballot  will  not  be  devoid  of  all  ballot  measures:  The  legislature  has  placed  two  of  its  own  proposals  on  the   ballot,  and  it  is  possible  that  these  two will  be  eno ugh to  draw  voters  to  the  polls.  However,  our  analysis   suggests  that  should  future  primary  elections  fail  to  include  any  initiatives  on  the ballot,  primary  turnout  is   likely  to  be  between  three and  seven  percentage  points  lower  than  it  might  otherwise  be. 4   1 The  other  states  are  Maine,  Missouri,  Nebraska,  North  Dakota,  Oklahoma,  and Oregon.  2 These  conclusions  are  based  on  a  regression  of  primary  turnout  on  the  margin  of  victory  in  all  initiative,  presidential,  gubernatorial,  and  U.S.   Senate  campaigns.  The  model  also  included  dummies  for  the  presence  of  each  type  of  campaign,  and  fixed  effects  for states  and  years.  Details  of  the  model  are  available  in  Appendix  B . We  al so tried  running  the  model  with  the  number  of  initiatives  in  addition  to  the  simple  dummy  for   presence  or  absence,  but  this  variable’s  coefficient  was  small,  statistically  insignificant,  and  bearing  the  wrong  sign,  while  the  dummy  remained   strong  and  significant.  This  suggests  that it  is  the  presence  rather  than  th e number  of  initiatives  that  matters.   3 Ballot measures  have  long  appeared  on  primary  election  ballots,  although  prior  to  1972  they would  appear on  the  ballot  as  “special”  elections  that  just  happened  to  coincide  with  the  scheduled  primary  election.  In  1972,  the  Secretary  of State  issued  an  opinion  interpreting  a  key  passage  in   the  constitution  as  allowing  initiatives  to  appear  on  the  prima ry ballot.  This  interpretation  was  occasionally  questioned  in  succeeding  years,   including  by  Chief  Justice  Rose  Bird  in  Brosnahan  v. Eu  [(1982)  31  Cal.3d  1  ],  where  she  argued  in  a  footnote  that  the  matter  deserved  revisiting.   For  a  brief  discussion  of  the  history,  see  the bill  analysis  accompanying  SB  202,  av ailable at   www.leginfo.ca.gov/pub/11 ‐12/bill/sen/sb_0201 ‐0250/sb_202_cfa_20110909_230915_sen_comm.html.   4 This  estimate  is  derived  from  the  95%  confidence  interval  around the  “ballot  measure  election”  coefficient  in  Model  3  of  Table  B1.  The  estimated   effect  of a  ballot  measure  election,  according  to  that  model,  is  5.13  +/‐ 1.76.    Open Primaries and Turnout California’s top-two primary system was adopted in part to encourage more voters to come to the polls. Yet turnout in 2012 was very low —the lowest, in fact, of any presidential primary in 90 years. As noted above, there are reasons to believe that this was at least partly due to low -key contests among the candidates for president and the U.S. Senate. But was it also an indictment of relying on open primaries to increase turnout? It certainly seems unli kely that expanding the primary electorate to include independents would actually discourage turnout. In California’s case, the change in the format of the primary was not as significant as it might at first appear, since independents were allowed some participation in primaries before 2012. Nonetheless, the question remains: Do open primaries have any consistent effect on turnout? Our analysis suggests that open primaries have not been associated with higher turnout in the states that have used them over t he past 30 years, at least in the cases where “open” meant any system that allowed independents to participate in some way in the election. While this curious result is fairly robust, we should be cautious about extrapolating too much from it. Part of the explanation is that independents do not participate as often in primary elections in the first place. 5 In California, where the rate of independent registration is about average for the nation as a whole, independents have never amounted to more than 14 percent of primary voters over the past five election cycles. And independents appear to represent an even smaller share of all registrants in closed primary states, meaning more of the electorate is composed of partisan voters who are more dependable about showing up at the polls.. 6 Thus, in open primary states, the potential electorate is much larger because independents can participate, but those extra potential voters do not turnout to vote at the same rates; in closed primary states, the potential electo rate is limited to partisan voters, but there are relatively more of them and they are more reliable at showing up. The two differences may neutralize each other and suggest no effect on turnout from open primaries. If overall turnout is about the same und er an open primary, turnout volatility is higher. In closed primary states, fixed differences between states explain much of the electorate’s size: States with high turnout in one election tend to have high turnout in the next. By contrast, the turnout in open primary states varies from election to election. The dynamics of competition appear to exp lain a lot of this difference: Once they are factored in, turnout in open primary states becomes about as predictable as in states where primaries are closed. 7 5 Developing a full explanation of the result is complicated by data limitations. While there has been considerable experimenta tion with open primaries at the legislative and congressional level (McGhee et al. 2014), that experimentation has often been concerned with the degree of openness rather than whether independents are allowed to partici pate at all. Moreover, there has been far less experimentation at the presidential level, and often by only one party in a particular state. Thus, the temporal variation necessary to explore the effect of ind ependent participation is less available. Finally, there is always the possibility that political parties have been willing to hold an open primary only when an election is not in doubt, which would likely result in a low turnout even though the system was more inclusive. Pulling apart this sort of se lection effect is ultimately impossible with the data at hand. 6 Many states with open primaries do not report registration by party, so there is no way to determine exactly how many independent registrants there might be if those states suddenly decided to force their citizens to choose. However, the independent registration rate in the open primary states with party registration is almost twice as high as in closed primary states (which need party registration in order to determine eligibility). This is tr ue whether independent registration is calculated as a share of total registrants (31% vs. 16%) or total eligible voters (23% vs. 12%). 7 These conclusions are based on a comparison of the difference in model fit between a null model with only state and y ear fixed effects and one with measures of competition included, and doing so separately for open and closed primary states. For closed primary states, the adjusted R2 was 0.59 for the null model and 0.64 for the full model, while the standard error of the regression was 5.60 for the null model and 5.23 for the full model. For open primary states, the adjusted R 2 was 0.45 for the null model and 0.62 for the full model, while the standard error of the regression was 6.69 for the null model and 5.62 for the f ull model. Thus, the full model explained about the same amount in both cases, but the null model was weaker for open primaries, suggesting that campaign dynamics account for a larger share of the variance in those states. As a placebo test http:// www.ppic.org /main/home.asp Voter Turnout in Primary Elections 9 Do these conclusions about open primaries extend to California’s new top-two primary? While the top -two system is “open” in the sense that it allows independents to participate, it also goes further by giving independents the same ballot as everyone else b y default. This makes it far easier for independents to participate in candidate contests; and in this respect, the open system more closely resembles the “blanket” primary California used in 1998 and 2000 than it does other types of open primary systems. By contrast, the “semi -closed” system California used between 2002 and 2010 required independents to request a ballot from one of the major parties if they wanted to vote for a partisan candidate. Only about half of the independent voters took advantage of this option ( McGhee, 2010 ). Figure 3 illustrates this difference in response: During the “semi -closed” years, t he share of ballots with a vote for president or U.S. House of Representatives was notably lower than in the years of the blanket or top -two primaries. 8 Note also the divergence in 2012, when voting for the House of Representatives shifted to the top -two s ystem while the presidential primary remained in the semi -closed system. Voters in 2012 actually ended up casting fewer votes for president than for the House, a remarkable outcome given the relative profile of the contests. With only one election under th e new system, it is not clear that this pattern will persist, but it is worth noting. FIGURE 3 Independents have cast more votes under the blanket and top-two primary systems than under the semi -closed system SOURCE: California Secretary of State. In sum, there is no clear evidence yet that open primaries produce a significantly higher turnout, partly because independents constitute such a small share of the primary electorate. However, on this first occasion of its use in California, the state’s particu lar approach to the open primary, in which independents received the same ballot as all other voters, did appear to get independents to cast more votes for the races other than president covered under the new system. Thus, the reform’s impact on the number of votes cast in an election may be quite positive. The true effect of the open primary in California will likely grow clearer as voters become more familiar with the reform and campaign consultants find better ways to mobilize voters under the new system . we ran the same analysis on general elections, where independents are always allowed to participate. We found virtually no difference between open and closed primary states. Model specifics are available in Appendix B . 8 Independents received a presidential ballot in 2000 and were allowed to vote for a candidate, but their votes were not counted in determining the outcome of the race . These votes are included in the 2000 presidential calculation fo r Figure 3. 60 65 70 75 80 85 90 95 100 1998 2000200220042006200820102012 Votes cast as a share of total turnout President House Open http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 10 Higher primary turnout was not the only goal of California’s new system. It was also meant to provide a better set of choices for the general election in heavily partisan districts where the fall outcome is typically a foregone conclusion. The new system led to same-party contests in the general election in 27 of these “safe - seat” districts, and many of these races ended up being quite competitive. In fact, about one- third of the same -party races were decided by less than 10 points, while only 15 percent of the races between candidates of different parties had less than a 10 -point spread. But this competition between same -party candidates did not lead to more votes in same- party contests in the fall. Quite the opposite: Between the primary and general el ections, same-party contests gained 8 percent fewer votes than cross -party races. 9 As just noted, this was not because the same -party contests were less competitive (see Table B5 in Appendix B for more evidence on this point). Nor was it because the type of districts with same -party contests had lower turnout in the fall for other reasons. In fact, the increase in the percentage of ballots cast was identical on average in both same -p arty and cross -party races; the only difference was the number of votes cast in the same -party race s (Figure 4) . Put differently, the electorate consistently voted for candidates running for higher offices, such as the U.S. Senate and the president, but of ten skipped same -party races further down the ballot. Since we don’t have data on individual voter decisions, it’s not possible to say anything about the characteristics of the voters who skipped these contests, but we can confidently say that, on balance, fewer felt the need to offer a vote in those races. FIGURE 4 Voters casting ballots often skipped same -party contests SOURCES: Political Data Incorporated (ballots cast); California Secretary of State (votes cast). NOTE: Results are for races that eith er became or failed to become same- party races by fewer than five percentage points, in order to account for other differences between same- party and cross-party districts. Numbers also reflect changes from the primary to the general election to further account for the possibility of different average turnout in the types of districts more likely to have same -party races. 9 These estimates are derived from a regression discontinuity analysis. Many fall contests became same -party contests by a razor-thin margin in the primary, raising the prospect that the same -party outcome was effectively random in those cases . We leveraged this fact, using the margin by which a seat became same -party (or did not) as the forcing variable which assigned a fall contest to same -party status. The result of this analysis suggested that the number of votes as a share of total registe red voters was 8 percent lower in same-party contests, a statistically-significant difference. We used software from Rocio Titiunik to implement a randomization inference approach to calculating standard erro rs, choosing a margin of five percentage points above or below the threshold for the sake of analyzing the results. Details of these estimations are available from the author on request. For details on the randomization inference method, see Cattaneo et al. (2013). 0 5 10 15 20 25 30 35 40 45 Ballots Cast Votes Cast Participation: general minus primary Cross party Same party http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 11 How Do Primary and General Electorates Differ? Thus far we have shown that turnout in California’s primary elections has been declining o ver time relative to turnout in the general elections, and that the state’s new open primary system may not be able to reverse this trend. Is the smaller electorate that turns out in the primary simply a microcosm of the larger one that turns out in the fa ll? In terms of basic demographics, there is little question that primary electorates have included fewer young people, Latinos, and Asian Americans than the general electorate in the fall (Figure 5). The difference is largest for Latinos, whose share of the electorate has been about seven percentage points higher in the two most recent fall presidential elections. But there are clear differences for young people and Asian Americans as well. 10 To the extent that these groups might support different candidates for office —either of different parties or within the same party —these variations in turnout are important. In fact, there may be many more differences between the primary and general electorates than discussed here, but these are just the differences in dicated by the data available in the voter registration file. 11 Even within demographic groups, those who vote in the primary might differ on policy issues or general ideology. FIGURE 5 Fall electorates have generally been younger and more diverse than pri mary electorates SOURCES: Statewide Database (2004 -2012 primary electorate); Political Data Incorporated (2012 general electorate). Partly because of the demographic differences highlighted above, there are also partisan differences between the primary and general electorates. The fall electorate has been notably more Democratic than the primary 10 These differences in magnitude are mostly a function of the size of each group. Proportional to size, the differences in turnout between primary and general elections are similar across all three groups. However, since the difference i n each group’s share of the electorate is what matters for election outcomes, we illustrate the respective shares in Figure 5. 11 The only other demographic category for which we have data across all five of these election cycles is gender, and it shows no difference between primary and general electorates. Other important demographic variables, such as education, income, or certain racial categorizations (African American, white) are not available in the voter registration file. +3.6 +1.2 +5.5 +2.7 +3.8 +1.3 +0.6 +0.9 +0.6 +2.0 +4.2 +2.0 +7.2 +5.4 +7.3 0 1 2 3 4 5 6 7 8 2004 2006200820102012 20042006200820102012 20042006200820102012 Percent change in share of electorate: primary to general Age 18–24 Asian/Pacific Islander Latino http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 12 electorate in six of the last eleven election cycles, and at least slightly more Democratic in all but two (Figure 6). The difference is about 3 percentage points for a seat balanced perfectly between the two parties in the primary stage. 12 These estimates are based on the actual votes cast, so they incorporate any effects from the growing population of independent registrants. FIGURE 6 Fall electorates have typically voted more Democratic than primary electorates SOURCE: California Secretary of State. This primary-to -general shift in partisan turnout explained about 92 percent of the variation in fall outcomes in 2012, missing the actual result by an average of only 5 percentage points. 13 This is remarkably accurate considering the radical winnowing of the candidate field between the two sets of elections. It suggests party loyalties still played a large role under the top -two system in its debut. In the past, the consequences of this partisan shift have been minimal for candidate races, because every party that ran at least one candidate in the primary could be assured of a place on the fall ballot. 14 But the top two primary upends this logic. There is no guarantee now that any party will be present in the fall—only that the top two vote getters will advance. It is entirely possible that the difference between the primary and general electorates will end up closing off the possibility of a cross -party contest in the fall where one might otherwise have occurred. Since the fall electorate is more Democratic than the primary electorate, this difference in turnout should usually produce more same -party contests between Republicans and fewer 12 The shift is smaller as po litical seats become more heavily Democratic. While there are more voters of every partisan stripe in the general election, the relative increase for Democrats, and especially for Independents, is much higher than for Republicans. As a result, the percenta ge of fall voters who are Democrats is usually about as large as in the primary, while the percentage of Independents is much highe r and the percentage of Republicans much lower. In heavily Democratic seats, there are relatively fewer Republicans to produc e this discrepancy, so the change between the primary and the general elections is smaller. 13 To calculate this number, we regressed the partisan vote in the primary on the party registration of voters in the primary el ectorate, and then predicted fall out comes with this equation, using the party registration from the fall electorate. Thus, the method assumes that the relationship between party registration and party voting is the same in both the primary and general election, and the only difference is the partisan composition of the voters. We then calculated the difference between this prediction and the actual fall outcome, and treated that as the residual for the calculation of an R -squared statistic. For the purposes of this calculation, we included on ly those races with at least one candidate from each major party printed on the ballot in both the primary and fall election, which eliminated 41 races. 14 This was true even for the blanket primary in 1998 and 2000, which was otherwise quite similar to the top - two primary. Indeed, the fact that every party is represented in the fall in a blanket primary is the main difference between the two systems. +0.8 -3.4 +0.7 +3.0 +3.2 +2.2 +0.7 -0.9 +3.7 +3.0 +3.2 -4 -3 -2 -1 0 1 2 3 4 5 1992 1994199619982000200220042006200820102012 Percent Democratic vote shift: primary to general http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 13 between Democrats than would otherwise be the case (although in at least one election in Figure 6, the opposite would have occurred). How large is this effect? We can simulate how the primary electorate in 2012 might have voted had it had the same partisan composition as the general el ectorate in 2012. The result is that there would be very little change in the total number of same -party contests, but there would be a significant shift in their partisan complexion: The simulation produces three same- party Republican races and 26 Democratic races, compared to the nine Republican and 19 Democratic that actually occurred. As in the case of the same- party races, the districts where the same- party status was most likely to change were almost exclusively “safe” for one party, so the partisan o utcome in the fall contest was not in doubt regardless. The one exception is Congressional District 31, where the actual fall contest pitted two Republicans against each other in a seat that was otherwise ripe for a competitive race. In this seat, the simu lation suggested a 97 percent chance that a Democrat would have faced off against a Republican if the primary electorate had been more representative of the fall electorate. We have already noted that the absence of ballot measures from the primary ballot may affect turnout in future primaries. Will it also exacerbate the partisan bias of the primary electorate? It turns out that over the 11 election cycles in Figure 5, there is little or no relationship between the size of the increase in the primary to g eneral turnout and the size or direction of the partisan shift (see Figure B3 in Appendix B ). For example, the primary with the best turnout relative to the general election (1998) saw a partisan difference of 3 percent, compared to 3.7 percent for the primary -to -general with the worst gap (2008). However, for the five recent elections for which we have data on demographics, a higher primary -general increase in turnout does correspo nd with a larger increase in turnout by young people and Latinos (the relationship for Asian Americans is more mixed; see Figure B4 in Appendix B ). Because this evidence is based on such a limited number of election years, we should treat it as tentative. But it does seem as though lower primary turnout corresponds with an older, less -diverse electorate, without necessarily making it more Republican. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 14 Electoral Reform and Primary Turnout California has recently passed two important voter registration reforms meant in part to encourage more turnout in state elections. The first, online registration, was authorized first by SB 381 in 2008, and then expedited by SB 397 in 2011. 15 This re form allows users to enter voter registration information on -line and, with a keystroke, complete the process, eliminating pen and paper transactions. The new system was used heavily in its inaugural election in 2012. The second reform, “conditional” voter registration, was established by AB 1436 in 2012. It allows residents to both register and vote on any day after the traditional registration deadline, up to and including Election Day itself. The main constraint is that valid conditional registration mus t take place at a county registrar office. This option will not be available until 2016 at the earliest, but many observers are eagerly anticipating a significant impact once the system is implemented. Several existing analyses of these reforms suggest th at they have not had, and may not have, much impact on general election turnout, although a variety of administrative effects have been and could be significant ( McGhee, 2014) . We conducted a similar analysis of primary turnout across the nation over the past 30 years and found no effect of these reforms. On line registration might have increased primary turnout about two percentage points for states that have adopted the reform, but the increase is well within the margin of error. Moreover, primary turnout has been, if anything, slightly lower in the states where election-day registration has been used, although this likely says more about the particular states and years where the system has been adopted than the causal effect of election -day registration itself. Thus, it seems doubtful that the new reforms wi ll significantly increase California’s primary election turnout. 15 SB 381 authorized an online registration system, but made it conditional on completion of the state’s Vot eCal database of voter registration records merged with Department of Motor Vehicle files. VoteCal has since become delayed until 2015 at the earliest, so SB 397 authorized the Secretary of State to use a different process to implement an online registrati on system. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 15 Conclusions and Policy Directions This report has looked at the causes and consequences of turnout in California’s primary elections. Turnout in primaries has, in fact, declined over the past few decades, even in comparison with turnout in the general election; and there are reasons to believe that turnout will continue to fall, since initiatives may no longer appear on the primary ballot. Furthermore, key policy interventions intended to b oost primary turnout may not have much effect. California’s new top -two primary failed to produce the increase in turnout that many had hoped for, and there is little evidence that open primaries in other states have fared any better. Independents appear t o be fickle primary voters, inclined to participate only when a ballot includes a close race. Recent efforts to increase turnout by making registration easier —such as online and same -day registration —also appear to have no meaningful impact on primary turn out. On the other hand, the top -two system may encourage those independents who do vote in the primary to make choices in candidate contests for State Legislature and U.S. Congress, since they now receive the same ballot as everyone else by default. When t he weak effect on turnout is combined with the positive effect on down -ballot voting, the net effect of the top -two primary on participation is probably positive. Moreover, it would be counterproductive to increase primary turnout by repealing SB 202 and o nce again permitting citizen initiatives on the primary ballot. California has always placed far more initiatives on the primary ballot than other states. The clear differences between the primary and general electorates documented in this report should ra ise equity concerns about this practice. It seems inappropriate to allow a subset of the electorate to decide major policy questions for the rest of the state. Of course, the cost of a more representative electorate for initiative questions may be a small er primary electorate for candidate races. But does that matter? The state’s primary turnout has been falling, to be sure, yet even with the decline it remains among the highest in the nation. And it is worth noting that research on primary turnout and pol arization has found no real link between the two, at least in the case of U.S. Senate elections ( Hirano et al., 2010). Thus, lower turnout is not necessarily producing more radical representation. That said, lower turnout may produce an older and whiter electorate than would otherwise be the case, affecting the choice of candidates advan cing to the fall contest. And even though lower -turnout primaries do not seem to consistently benefit one party or the other compared to higher -turnout primaries, the lower turnout in primaries as a whole does produce a fairly consistent Republican tilt (a lthough one election cycle in our analysis had a primary electorate that skewed toward the Democratic candidates). This bias creates some problems for the top -two primary system, which winnows the field much more aggressively than the old system, and witho ut regard to party. There has already been one competitive seat with a same -party contest in the fall, and we have shown that the outcome almost certainly would have been different if the fall electorate had shown up earlier in the primary. Although such o utcomes will probably never be common under the top two, the fact that one already happened in the first election under the new system suggests it may easily happen again. 16 Moreover, it appears these same- party contests are of less interest to voters on av erage than are races between two candidates of different parties. The prospect that the skewed primary electorate might lead to more 16 The major parties can certainly try to address this problem more informally by restricting the number of candidates within their party. For example, if only one Democrat had run in Congressional District 31 instead of the four who actually did , the Democratic vote would likely not have been split between so many candidates and there would have been a cross -party contest in the fall. But there are no certain methods of clearing a candidate field in American politics. Encouraging major party dono rs and officeholders to coordinate their efforts on the behalf of one candidate will often work to discourage others, but sometimes it will not. It would be useful to have a safety valve that would allow a party to place a candidate directly on the fall ba llot in case such coordination fails. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 16 same-party contests in competitive seats, or even that such an outcome might occur in a statewide race, should not be dismi ssed. 17 One possibility worth exploring would be to allow more than two candidates in the fall contest, at least in some cases. At present, the top -two primary system bans more than two candidates in the fall, including independent candidacies or write -ins . In other states, this sort of provision is often called a “sore loser” law because it supposedly prevents a losing primary candidate from entering the race in the fall election and wreaking havoc for the sake of ego or revenge. While there may be a publi c interest in preventing too many such candidacies —especially under a system like the top two that is so committed to a two -candidate fall race —it might also be prudent to have a way of avoiding same -party contests when they might not reflect the choices g eneral election voters would like to see. A write -in option, or the option to launch an independent petition- drive candidacy, could provide a sort of “safety valve” in cases of such unexpected outcomes, while also ensuring that most races would still pit o nly two candidates against each other. Indeed, California law already allows for both options in presidential contests (California Elections Code, Sec. 8300 and 8600). The effects of such a change could extend beyond candidate competition. While research suggests that closed primaries do not increase polarization (McGhee et al., 2014 ), recent researc h has suggested that “sore loser” laws have increased polarization by preventing candidates with broader appeal from running directly in the fall ( Burden et al., 2014). We should be cautious about concluding too much from this research, since it did not look strictly at t he top-two system (nor could it, given the data available). But it offers some suggestion that an independent or write -in option (allowing for a safety valve candidacy) might not only enable the system to address quirky outcomes but also advance the cause of moderate representation that the top -two system was meant to promote. 18 The specifics of such a change would be important. The requirements for either a write- in or an independent candidacy option would need to be set quite high to ensure that these options would be used only in cases where the potential demand for another candidate was strong. For example , it currently takes 40 signatures to be a write -in candidate on the primary ballot (and have one’s votes counted), and it takes one percent of all registered voters in the last election to be an independent candidate for president. This second number, though much higher, may still be too low to serve as a proper barrier in the context of the top two. Something closer to the signature requirement for initiatives—for example, 5 percent of the last vote for the office in quest ion (Assembly, State Senate, Congress , or statewide race)— may be closer to the mark. Regardless, further exploration of the idea would be needed to assess its viability. Should the ide a be worth pursuing, however, it would not require the extra hurdles of a state constitutional amendment. T he ban s on write -ins and independent petition candidates are statutory provision s that were added by the legislature through SB 6 (2009) , so they could be changed through another bill passed by the legislature or a statutory initiative placed before the voters . In sum, turnout in California’s primaries is falling but remains high compared to the rest of the nation. The interventions proposed to increas e turnout may not prove particularly effective, although the top -two system appears to have already encouraged more participation by independents in the primary candidate contests. The most serious consideration moving forward is what to do about the inter action between the differences in the primary and general electorates under the top -two system. 17 Note that this possibility differs from an uncontested race, where no candidate comes forward in the first place. Uncontested races are much more a product of political dynamics, while the sort of anomalous same -part y races described here are largely a product of the mechanics of the top -two system. Put differently, an uncontested race will always have the same outcome, regardless of the primary system in use. 18 This approach might also help manage cases where a cand idate for a top-two office dies during the course of the election season. Existing law provides no means of replacing such a candidate; the candidate’s name remains on the ballot, and if that candidate is ultimat ely “elected,” a special election is immedia tely called to fill the vacancy (California Election Code 8803(b), 8805(b)). http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 17 References Burden, Barry C., Bradley Jones, and Michael S. Kang. 2014. “Sore Loser Laws and Congressional Polarization. ” University of Wisc onsin, Madison. Cattaneo, Matias D., Brigham Frandsen, and Rocio Titiunik. 2013. “Randomization Inference in the Regression Discontinuity Design: An Application to Party Advantages in the U.S. Senate. ” Un iversity of Michigan. Childers, Matthew, and Mike Binder. 2012. “Engaged by the Initiative? How the Use of Citizen Initiatives Increases Voter Turnout. ” Pol itical Research Quarterly 65(1): 93 –103. Geys, Benny. 2006. “Explaining Voter Turnout: A Review of the Aggregate-Level Research. ” Electoral Studies 25: 637– 663. Hirano, Shigeo, James M. Snyder Jr., Stephen Ansolabehere, and John Mark Hansen. 2010. “Primary Elections and Partisan Polarization in the U.S. Congress. ” Quarterly Jour nal of Political Science 5(2): 169– 191. McGhee, Eric. 2010. At Issue: Open Primaries. San Francisco: Public Policy Institute of California. Available at www.ppic.org/main/publication.asp?i=904 . McGhee, Eric. 2014. Expanding California's Electorate: Will Recent Reforms Increase Voter Turnout? San Francisco: Public Policy Institute of California. Available at www.ppic.org/main/publication.asp?i=1083 . McGhee, Eric, Seth Masket, Boris Shor, Steven Rogers, and Nolan McCarty. 2014. “ A Primary Cause of Partisanship? Nomination Systems and Legislator Ideology.” American Journal of Political Science forthcoming Tolbert, Caroline J., Daniel C. Bowen, and Todd Donovan. 2009. “ Initiative Campaigns: Direct Democracy and Voter Mobilization. ” American Politics Research 37(1): 155– 192. http://www.ppic.org /main/home.asp Voter Turnout in Primary Elections 18 About the Author Eric McGhee is a research fellow at PPIC. His work focuses on electio ns, political participation, political polarization, legislative behavior, redistricting, and surveys a nd polling. Before joining PPIC, he was a professor of political science at the University of Oregon . He holds a Ph.D. in political science from the Univ ersity of California, Berkeley. Acknowledgments The author would like to thank several people outside PPIC for their help in making this report possible: Darren Chesin, Ethan Jones, Paul Mitchell, and Nicole Winger all offered expertise and insights. Paul Mitchell deserves special credit for the rich resource of registration data he made available for analysis. An earlier version of the report benefited greatly from a careful reading by Dean Bonner, Benjamin Highton, Tony Quinn, Karthick Ramakrishnan , and Lynette Ubois , and Gary Bjork and Kate Reber helped clean up the final draft for public presentation. As always, thanks to Mark Baldassare for his continuing commitment to both this report and a neutral, nonpartisan agenda of research on California’s polit ical reforms. Any errors are my own. http:// www.ppic.org /main/home.asp Voter Turnout in Primary Elections 19 PUBLIC POLICY INSTITUTE OF CALIFORNIA Board of Directors Donna Lucas, Chair Chief Executive Officer Lucas Public Affairs Mark Baldassare President and CEO Public Policy Institute of California Ruben Barrales President and CEO GROW Elect María Blanco Vice President, Civic Engagement California Community Foundation Brigitte Bren Attorney Walter B. Hewlett Member , Board of Directors The William and Flora Hewlett Foundation Phil Isenberg Vice Chair Delta Stewardship Council Mas Masumoto Author and Farmer Steven A. Merksamer Senior Partner Nielsen, Merksamer, Parrinello, Gross & Leoni, LLP Kim Polese Chairman ClearStreet, Inc. Thomas C. Sutton Retired Chairman and CEO Pacific Life Insurance Company T he Public Policy Institute of California is dedicated to informing and improving public policy in California through independent, objective, nonpartisan research on major economic, social, and political issues. The institute’s goal is to raise public aware ness and to give elected representatives and other decisionmakers a more informed basis for developing policies and programs. The institute’s research focuses on the underlying forces shaping California’s future, cutting across a wide range of public polic y concerns, including economic development, education, environment and resources, governance, population, public finance, and social and health policy. PPIC is a public charity. It does not take or support positions on any ballot measures or on any local, state, or federal legislation, nor does it endorse, support, or oppose any political parties or candidates for public office. PPIC was established in 1994 with an endowment from William R. Hewlett. Mark Baldassare is President and Chief Executive Officer o f PPIC. Donna Lucas is Chair of the Board of Directors. Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source. Research publications reflect the views of th e authors and do not necessarily reflect the views of the staff, officers, or Board of Directors of the Public Policy Institute of California. Copyright © 201 4 Public Policy Institute of California All rights reserved. San Francisco, CA PUBLIC POLICY INSTITUTE OF CALIFORNIA 500 Washington Street, Suite 600 San Francisco, California 94111 phone: 415.291.4400 fax: 415.291.4401 www.ppic.org PPIC SACRAMENTO CENT ER Senator Office Building 1 121 L Street, Suite 801 Sacramento, California 95814 phone: 916.440.1120 fax: 916.440.1121" ["post_date_gmt"]=> string(19) "2017-05-20 09:42:00" ["comment_status"]=> string(4) "open" ["ping_status"]=> string(6) "closed" ["post_password"]=> string(0) "" ["post_name"]=> string(8) "r_514emr" ["to_ping"]=> string(0) "" ["pinged"]=> string(0) "" ["post_modified"]=> string(19) "2017-05-20 02:42:00" ["post_modified_gmt"]=> string(19) "2017-05-20 09:42:00" ["post_content_filtered"]=> string(0) "" ["guid"]=> string(50) "http://148.62.4.17/wp-content/uploads/R_514EMR.pdf" ["menu_order"]=> int(0) ["post_mime_type"]=> string(15) "application/pdf" ["comment_count"]=> string(1) "0" ["filter"]=> string(3) "raw" ["status"]=> string(7) "inherit" ["attachment_authors"]=> bool(false) }