Party Registration Deadlines and Hidden Partisanship: An Individual Analysis

Author: Matthew Thornburg, University of South Carolina at Aiken

Abstract

Many voters in states with party affiliation identify with or lean towards one political party but are not registered with it. This sort of “hidden partisanship” may be intentional be it may also result from a combination of changes in a voter’s party identification and the electoral institutions in place. In many states it is difficult to change party registration due to early deadlines intended to prevent crossover voting. Using individual-level survey data, I find that hidden partisanship in a state increases, the further in advance of the primary the deadline to change party affiliation is. This deadline affects primary turnout, with deadlines further in advance of primary elections leading to lower turnout by partisans in their own party’s primary and greater rates of abstention and crossover voting. A difference-in-difference design at the county level shows that moving the deadline to change party registration closer to the primary increases primary turnout.

Introduction

In 2016, Republican Donald Trump won 62.5% of the vote in Kentucky to Hillary Clinton’s 32.7%. At the same time, registered Democrats outnumbered registered Republicans 51.2% to 40.4%. While this disjunction might at first glance reflect Kentuckians’ longstanding ticket-splitting tendencies, aggregate Gallup data on the state in 2016 found the GOP to have a nearly 10-point edge in party identification at the time of the 2016 election (Jones 2017). It is reasonable to conclude from these data that many voters in Kentucky identify or lean Republican but remain registered as Democrats. What explains this phenomenon?

Notably, Kentucky has the second earliest deadline in the country to change party registration. A voter who wishes to vote in primary elections during an election year must have changed their party affiliation by December 31 of the previous year. The “hidden partisanship” (lack of concordance between a voter’s party identification and party registration) (Arrington and Grofman 1999), of Kentucky voters may be due to this significant hurdle to change party affiliation, which requires vigilance and advance planning.

Most states in the country require voters to officially affiliate with a political party when they register to vote. This affiliation manages their participation in primary elections. In the fourteen states with closed primaries, voters must be affiliated with a political party to participate in its primary elections. In order to discourage crossover voting, all of these states require individuals who change political party affiliation to be registered with a political party for a certain period of time (ranging from two weeks to nearly a year) in advance of voting in that party’s primary.

This deadline to change affiliation is notable as a consistent finding in the literature on general election turnout is that individuals are more likely to register when an election is salient. Thus, states with deadlines to register to vote close to Election Day (Kelley et al. 1967; Wolfinger and Rosenstone 1980) or even election day registration (Brians and Grofman 2001) exhibit higher voter turnout.

Thornburg (2019) examines the effect of party registration deadline on hidden partisanship and primary turnout using aggregate data at the county level from 2010 and 2014. The author finds that counties in states with early deadlines to change party registration exhibit more hidden partisanship and that this effect is conditioned on realignment that the county has experienced. Counties that have exhibited the largest changes in aggregate partisanship and are located in states with early deadlines to change party registration (such as Kentucky on both counts) exhibit a large disconnect between aggregate party registration and partisan vote of the county which causal mediation analysis shows then affects primary turnout.

However, the Thornburg (2019) piece uses aggregate units (counties) to reach its conclusions and thus is vulnerable to the ecological fallacy. I extend this research by examining whether deadlines to change party affiliation affect turnout in primaries at the individual level. Based on the existing literature, I predict early party registration deadlines to be associated with lower turnout in primaries as more voters have hidden partisanship, now identifying with one party but still having registration that reflects their previous party identification. While party identification is generally thought to be resistant to change, it is not immutable, especially in the face of large scale realignments like the South.

Analyzing Democratic and Republican identifiers/leaners in closed primary states, I find greater incidence of hidden partisanship—self-identified Democrats and Republicans not registered with the party with which they identify. The misregistered comprise as much as 25 percent of registrants among party identifiers and leaners in states with the earliest party registration deadlines compared with just 10 percent in states with party registration deadlines near the primary. Looking at turnout in the 2010 state and federal primaries at the individual level, I find party registration deadlines to have a modest but statistically significant effect on turnout in the primary of the party a voter identifies with/leans to. I also use a difference-in-difference model, regressing changes in county primary turnout from 2010 to 2014 on changes in party registration closing date during this same period. I find results that strongly support my hypothesis on the effect of party registration deadlines on primary turnout.

I find that party registration deadlines near the primary do not lead to more crossover voting. This indicates that the fears of policymakers that a party registration closing date near the election will lead to voters “raiding” the opposition primary do not appear to be justified. In fact it is states with early party registration closing dates that see a modest rise in crossover voting. Ironically, party registration closing dates appear to have the opposite of their intended effect. The only effect early closing dates have on who votes in primary elections appears to be to depress turnout in primaries among identifiers and leaners and lead to greater crossover voting.

Literature Review and Theory

Most studies examining the effect of electoral institutions on voter turnout rely on the theoretical framework of Downs (1957) as well as Riker and Ordeshook (1968) which posit the decision to cast a vote as balancing costs against benefits. While quantifying the benefits of voting has been controversial (e.g. Grofman 1993) most research has held that voters are responsive to the administrative costs of casting a ballot.

This research, beginning with Merriam and Gosnell (1924) and reaching a major milestone with Wolfinger and Rosenstone (1980) identifies aspects of election administration that affect voter turnout. Notably, the latter work finds deadlines to register to vote have the largest single effect on individual turnout. This finding has been confirmed by numerous other studies (Kim et al. 1975, Rhine 1995, Rhine 1996, Rosenstone and Hansen 1993, Squire et al. 1987, Texeira 1992). Related work on the timing of voter registration shows that it is most frequent in the final days before an election (Gimpel, Dyck and Shaw 2007). 

Thornburg (2019) applies this line of research to the effect of party registration deadlines on hidden partisanship at the aggregate (county) level. Examining hidden partisanship and primary turnout in 2010 and 2014, the author finds that counties which have undergone a significant aggregate shift in partisanship over the decade prior and are located in states with early deadlines to change party registration contain many voters who have shifted partisanship with the county but remain registered with their old party. This leads to decreased turnout in the primary of the ascendant party in the county and elevated turnout in the primary of the waning party.

Thornburg (2019) draws on the literature from economics (Frederick et al. 2002), biology (McClure et al. 2004) and psychology (Ainslie 1975, Trope and Liberman 2003) which shows human beings have a tendency to discount the value of future events. The author theorizes that deadlines to change party registration far in advance of a primary lead to fewer people doing so because the lack of salience and proximity to the primary does not provide sufficient motivation to bring one’s party registration in line with one’s new party identification.

While Thornburg’s (2019) results are suggestive, they rely on aggregate data vulnerable to the ecological fallacy and require guesses about the party identification of voters based on their aggregate vote. In this article, I therefore use survey data to observe hidden partisanship at the individual level. The present research also acknowledges the possibility that deadlines to change party registration are not exogenous to primary turnout or the political conditions in a state. Because primary turnout, hidden partisanship and deadlines to change party registration may all be related to the strength of the party system in a state, I use categories for party system strength devised by Morehouse and Jewell (2005) and use a difference-in-difference design to examine how changes in party registration deadlines may affect turnout in primaries.

Another possible spurious correlation may result between timing of primary elections and party registration closing date. States with early party registration closing dates tend to hold primary elections closer to the general election compared to those with later closing dates. When evaluating the effect of party registration closing date on turnout, I therefore also control for the number of days in advance of the general election that the state primary was held.

Party Registration Deadlines and Hidden Partisanship

If an early party registration closing date increases the peripheral costs of bringing one’s party registration in line with one’s party identification, decreasing turnout, we should expect to observe a relationship between closing date and likelihood of “hidden partisanship”—identifying with or leaning to one party but not registered with it. The further out the party registration deadline, the more likely that hidden partisanship is observed. The most effective way to test this hypothesis is to observe party registration and party identification in individual-level data to judge their concordance. The Cooperative Congressional Election Study (CCES) is one of the few election studies to provide validated party registration. The CCES is a large opt-in survey conducted online by YouGov Polimetrix. The sample size for the 2008 CCES, which is used here, is 7,918 respondents residing in closed primary states with validated party registration. Using the 2008 CCES, I am able to measure political attitudes, demographic characteristics, party registration and state characteristics.

Primary turnout at the aggregate level has traditionally been measured as the percentage of supporters of a political party that vote. This level of support at the state or county level has been measured mainly by votes for candidates for high office, either in the most recent election or over time as a “normal vote” (Norrander 1986). For individual analysis of primary turnout and party misregistration, there are a greater number of options for deciding who to include in the pool of Democratic and Republican supporters. I use a measure of party identification—including Democratic and Republican identifiers. Independent leaners are included in the analysis shown here. However, the substantive conclusions of the analysis in this paper are replicated when only identifiers are examined.

My dependent variable is validated party registration. Because of their small numbers, minor party registrants are included with unaffiliated voters. The dependent variable has three values: registration as a Democrat, Republican and unaffiliated/minor party.

My independent variable is a transformation of party registration closing date. Many states have different party registration closing dates for those switching from major party registration compared to those switching from being unaffiliated. Because I believe the negative effects of closing date are felt most keenly by those registered with a major party who shift party identification, I use the date for shift from major party registration. The independent variable transforms the number of days this deadline lies in advance of the primary. States with no party registration closing date are excluded from the analysis due to their qualitative difference as de facto open primary states. My theory is partially based on the psychological predilection to discount the benefits of far off-events. Scholarship evaluating the functional form of this discounting rate has consistently found it to be nonlinear and either exponential (Lancaster 1963, Meyer 1976) or hyperbolic (Madden et al. 1999). I thus use a natural log transformation to convert party registration closing date. 

I also control for education (whether the respondent graduated from four-year college), age, race (dummy variables for African-American or Hispanic), whether the respondent had a high level of political interest, gender, and the percentage of the state’s population identifying as Democrats minus the percentage identifying as Republicans based on 2008 Gallup data. Finally, as described previously, I include dummy variables for moderate and weak state party strength. Separate analyses are run for Democrats and Republicans using a multinomial logit model with robust standard errors clustered by state.

The results are shown in Table 1. For the Democratic identifiers/leaners, the coefficients in the Republican and unaffiliated/minor party comparisons are both positive. They approach statistical significance in both cases and a plot of the predicted probabilities and their confidence intervals (Figure 1) shows a significant decrease in the number of Democrats registered with their party against the transformation of party registration closing date. For Republicans, in the Democratic registration comparison the coefficient is also positive. The variable is statistically significant at p < 0.005. Figure 1 uses these models to plot our quantity of interest—probability of being registered as a Democrat, unaffiliated/minor or Republican—against the transformation of party registration closing date for Democrats and Republicans. Probabilities and confidence intervals are generated using the “observed value” approach rather than the “average case” approach because the former leads to less bias in the estimates 

of marginal effects (Hanmer and Kalkan 2013). Both Democrats and Republicans show a significant decrease in the percentage of voters registered with their party as the transformation of party registration closing date increases.

Table 1: Party Registration of Democratic and Republican Voters, 2008

Figure 1: Party Registration Probabilities, 2008 CCES

The theorized relationship between party registration closing date and misregistration holds for both Democrats and Republicans. The change in the percentage of Democrats and Republicans voters who are not registered with their party when one goes from shortest to longest party registration closing dates analyzed (fifteen days to ten months, respectively) is substantial. Among both Democrats and Republicans, going from a closing date fifteen days in advance of the primary to one ten months in advance leads to a roughly fifteen percentage point drop in the percentage of voters registered with their party in closed primary states.

Party Registration Closing Date and Primary Turnout in 2010, Individual Analysis

Because closed primary states prohibit voters not registered with a party from voting in its primaries, the conclusions in the preceding section have implications on primary turnout. Misregistered voters are prevented from voting in the primary of the party they support. To test whether party registration closing date influences turnout in primary elections through the mechanism of misregistration, I examine turnout in the 2010 state and federal primaries at the individual level.

Because presidential primaries introduce several complications to comparing primary turnout across states, such as caucuses as well as presidential primaries in some states linked to state and federal primary elections, I restrict myself to analysis of state and federal primaries during midterm elections. The CCES survey is the only one to measure turnout in midterm primaries and the 2010 version of the CCES is the only one to explicitly ask which party primary the voter participated in. This is then validated with Catalist turnout data. This makes primary turnout in a midterm election easy to measure accurately.

I separately analyze Democratic and Republican identifiers/leaners with a validated active registration in closed primary states. The dependent variable is primary turnout in the 2010 state and federal primaries and has three possible values—(1) abstention if Catalist validated this individual as not voting, (2) Democratic primary if Catalist validated the voter as participating and the voter reported she voted in the Democratic primary and (3) Republican primary if Catalist validated the voter as participating and the voter reported she voted in the Republican primary. All other combinations of these variables are removed from the analysis.

As in the previous section, my primary independent variable of interest is a natural log transformation of party registration closing date. I also include the same controls for race, age, gender, education, state partisanship, state party strength and voter political interest. In addition, I include the primary margin of victory for the gubernatorial or senatorial race in the state for the voter’s party.² I also include a variable for the number of days in advance of the 2010 general election that the state’s state and federal primary took place.

The results for Democrats and Republicans are shown in Table 2. For Republicans, the natural log transformation of party registration closing date is positive and statistically significant at p < 0.01 for the Democratic comparison and in the positive direction in the abstention comparison. Among Democrats, the coefficients in the Republican and abstention comparisons are both positive and the Republican coefficient is significant at p < 0.05. The positive coefficients indicate that an earlier party registration closing date makes Republican voters more likely to either abstain from voting in a primary or vote in the Democratic primary than with a closing date closer to the election, and analogous results among Democrats.

Table 2: Primary Election Turnout for Democratic and Republican Voters, 2010 State and Federal Primaries

		Democratic Voters				Republican Voters
		Abstention		Republican Primary		Abstention		Democratic Primary
		Coeff.		Coeff.		Coeff.		Coeff.
Variable		(Std. Error)	p-value	(Std. Error)	p-value	(Std. Error)	p-value	(Std. Error)	p-value
Closing Date	Ln(Days Between CD and Primary)	0.198	0.095	1.485	0.038	0.132	0.400	0.578	0.003
		(0.119)		(0.717)		(0.157)		(0.193)
S. Partisanship	D% – R% by Gallup in 2008	-0.059	0.000	-0.068	0.337	0.012	0.260	0.035	0.002
		(0.011)		(0.071)		(0.011)		(0.011)
Statewide MOV	MOV in Gubernatorial/Senatorial Race	0.013	0.000	-0.020	0.298	0.010	0.076	-0.005	0.261
		(0.002)		(0.019)		(0.006)		(0.004)
Party Sys. Str.	Medium	-0.929	0.000	0.741	0.278	-0.797	0.000	2.358	0.000
(Ref. Cat.:		(0.089)		(0.683)		(0.191)		(0.319)
Strong)	Weak	-1.426	0.000	-1.902	0.045	-0.768	0.000	2.599	0.000
		(0.194)		(0.947)		(0.157)		(0.129)
Days to Gen.	Number of days between pri. and gen.	-0.001	0.376	-0.000	0.536	-0.002	0.182	0.005	0.004
		(0.001)		(0.001)		(0.001)		(0.002)
Political Interest	High Political Interest	-1.041	0.000	0.023	0.953	-1.000	0.000	-0.681	0.068
		(0.088)		(0.397)		(0.066)		(0.373)
Hispanic	Hispanic	0.225	0.277	-1.136	0.017	0.372	0.009	1.196	0.041
		(0.207)		(0.476)		(0.143)		(0.587)
Black	African-American	0.015	0.886	-1.912	0.120	0.691	0.000	0.017	0.928
		(0.107)		(1.231)		(0.109)		(0.189)
Age	Age	-0.039	0.000	-0.011	0.223	-0.029	0.000	0.034	0.003
		(0.003)		(0.009)		(0.004)		(0.011)
College	College Graduate	-0.218	0.005	-0.172	0.412	-0.117	0.004	-0.179	0.519
		(0.077)		(0.210)		(0.041)		(0.278)
Female	Female	0.022	0.761	-0.603	0.146	-0.069	0.318	-0.272	0.174
		(0.072)		(0.415)		(0.069)		(0.277)
Constant	Constant	3.270	0.000	-7.313	0.011	1.781	0.001	-7.723	0.000
		(0.455)		(2.879)		(0.529)		(1.178)
Number of Observations		5,021				6,382
Log Likelihood		-3236.414				-3908.605
Pseudo R 2		0.126				0.116

Source: 2010 CCES. Reference Category: Democratic primary vote             

Reference Category: Republican primary vote

Figure 2 plots the predicted probability of each primary turnout choice against natural log of party registration closing date for Democrats and Republicans and the corresponding confidence intervals. Both Democrats and Republicans find turnout in their respective primaries decreasing while the prevalence of crossover voting (and in the case of Republicans, abstention) increases. The trend is stronger among Republicans but a notable increase in crossover voting is also seen among the Democrats.

Figure 2: Primary Turnout Probabilities, 2010 CCES

Party Registration Closing Dates and Primary Turnout, Difference in Difference Model

Individual level results with the 2010 CCES offer strong support for the deleterious effect of party registration closing date on primary turnout. However, individual level estimates of turnout based on self-report are fraught with over-report bias (e.g. Karp and Brockington 2005). I thus take the additional step of estimating the effect of changes in party registration closing date on changes in aggregate primary turnout at the county level for the 2010 and 2014 state and federal primaries.

One concern in studies of the effect of electoral institutions on turnout is endogeneity. In measuring these effects, it is possible that the implementation of the institution may in fact be shaped by turnout rate. To deal with this concern, scholars examining the effect of electoral institutions on turnout have frequently used difference-in-difference models to establish causality (e.g. Burden et al. 2014, Erikson and Minnite 2009, Giammo and Brox 2008). In such models, the dependent variable is the change in turnout measured between two close elections. This is regressed on changes in electoral institutions within the interval between the two elections that compose the dependent variable. This model offers two major advantages (Burden et al. 2014). First, it alleviates concerns about endogeneity because the independent variable precedes the dependent variable. It is unlikely that in this short term, there is an endogeneity bias. In addition, since the dependent variable is not raw turnout but change within the same unit, a difference-in-difference approach addresses concerns about omitted variable bias.

I regress the change in voting age population (VAP) turnout between the midterm primaries of 2010 and 2014 on changes in the natural log of party registration closing date.³ In the interval between the 2010 and 2014 primaries, two states for which turnout was measured moved their party registration closing date nearer the primary election date, three measured states moved their party registration closing date further in advance of the primary election and eight did not change the party registration closing date vis-à-vis the primary election date.⁴

Among the counties in party registration states, 568 have Democratic primary turnout data (measured as above) for both elections and 626 have Republican primary turnout data. I analyze Democratic and Republican primary turnout separately. In addition to evaluating the effect of change in natural log of primary closing date, I also control for the change in the competitiveness of the primary from 2010 to 2014 (as defined above). This variable is interacted with the normal vote for the respective party’s presidential candidate in the state. Other control variables include the change in the proportion of the county’s VAP that was Hispanic from 2010 to 2014 as well as a similar variable for African-Americans. Change in the number of days in advance of the general election that the primaries were held in 2010 and 2014 is included. Finally, the proportion change in the overall VAP of the county is included as a control. The observations are weighted by 2014 VAP times the proportion of the county voting for the party’s presidential candidate in 2012.

The results from the regressions are shown in Table 3. The effect of the party registration closing date change is strong and in the expected direction when examining Democratic and Republican primary turnout. The variable is statistically significant at p < 0.01. Counties in states moving party 

registration closing dates further from the primary had lower turnout in the Democratic and Republican primaries while counties in states moving party registration closing dates closer to the primary had the opposite effect.

Table 3: Difference-in-Difference Model for Change in Primary Turnout, 2010-2014

		Δ Democratic Turnout		Δ Republican Turnout
		Coeff.		Coeff.
Variable		(Std. Error)	p-value	(Std. Error)
Δ Closing Date	Δ Ln(Days Between CD and Primary)	-7.357	0.002	-4.899	0.000
		(2.335)		(1.320)
Normal Vote	Avg. State Pres. Vote Share, 2008 & 2012	-0.083	0.356	0.423	0.000
		(0.090)		(0.058)
Δ Primary Comp.	Δ MOV for Sen. or Gov. Race in Primary	-0.056	0.396	-0.149	0.201
		(0.066)		(0.116)
Normal*Δ Primary Comp.	Interaction Term	0.001	0.471	(0.116)	0.350
		(0.001)		(0.002)
Δ VAP	Proportion  Δ in VAP, 2010-2014	56.762	0.000	20.628	0.249
		(12.940)		(17.865)
Δ Black	Δ in AA proportion of VAP	-70.661	0.330	87.388	0.084
		(72.494)		(50.505)
Δ Hispanic	Δ in Hispanic proportion of VAP	103.009	0.019	30.328	0.573
		(43.692)		(53.761)
Δ Days to Gen.	Δ in days to General election from Primary	-0.084	0.047	-0.008	0.575
		(0.042)		(0.014)
Constant	Constant	0.337	0.942	-25.730	0.000
		(4.666)		(3.130)
Number of Observations		568		626
R 2		0.192		0.528

Numbers in Percent

Conclusion

Most of the discussion of the effect of institutions on voter turnout follows the basic logic of Downs (1957) and Riker and Ordeshook (1968) that voter turnout balances costs against benefits. Deadlines to register to vote that are well in advance of the election lead to lower turnout because they increase the marginal costs of registering by forcing voters to plan ahead and decrease the perceived benefits through the human tendency to discount the value of future events. Thus, voter registration closing dates well in advance of an election lead to significantly lower turnout. The present research has extended this literature on closing dates to the institution of party registration in states with closed and semi-closed primaries. While voter registration closing dates are limited by federal law from extending more than one month prior to an election, no such restriction exists for the deadline a voter has to change from one political party to another. This means that in many states with party registration, a voter must reregister with a new political party months in advance if he or she wishes to vote in the party’s primary.

I find that over 25% of Democrats and Republicans are not registered with the party they identify with/lean to. These voters may have initially identified differently, but over time have moved to their current party identification. In contrast, states that have party registration closing dates closer to the primary election have only approximately 10% of Republicans and Democrats not registered with their party.

Examining 2010 primary turnout at the individual level, I find a change in party registration closing date from ten months to fifteen days leads to a ten percentage point change in Republican primary turnout. Among Democrats, a change from five months to fifteen days leads to an eight percentage point change in turnout in the Democratic primary.

One major finding of this research is that party registration closing dates close to the election do not increase the probability of either Republicans or Democrats “raiding” opposition primaries. This has long been a justification for early party registration closing dates but party registration states uniformly have little in the way of crossover voting except among voters where party registration closing dates are early.

For party registration to function as intended, a voter’s party in the records must match his or her internal party identification. However, because party registration is a government record, it must be updated when a voter’s party loyalty changes. This is an arduous task with little reward relative to initially registering to vote. In order to facilitate participation in primary elections, voters need to have the ability to change their party registration quickly and easily.

The present research demonstrates that early party registration closing dates serve as a barrier to voters updating their party registration when it is obsolete. This has a negative effect on primary turnout among Democrats and Republicans. Moving closing dates for party registration closer to the election would serve to raise turnout. The research demonstrates that such a change would not appreciably increase crossover voting, or raiding, of opposition primaries.

Footnotes

¹ These dates are derived primarily from Paradis (2009) for the 2008 election.

² States without a gubernatorial or senatorial primary were not included in the analysis.

³ While the VAP turnout has been shown to be an imperfect measure of voter turnout (McDonald and Popkin 2001), voting eligible population data are not available at the county level.

⁴ California switched to a top-two primary for its 2014 midterm primary, leading to its omission. Several states, such as New York, did not have a state-wide primary one year, leading to their omission.

References

Ainslie, G. (1975). Specious Reward: A Behavioral Theory of Impulsiveness and Impulse Control. Psychological Bulletin, 82(4), 463–469.

Ansolabehere, S., & Hersh, E. (2012). Validation: What Big Data Reveal About Survey Misreporting and the Real Electorate. Political Analysis, 20(4), 437–459.

Burden, B. C., Canon, D. T., Mayer, K. R., & Moynihan, D. P. (2014). Election Laws, Mobilization and Turnout: The Unanticipated Consequences of Election Reform. American Journal of Political Science, 58(1), 95–109.

Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960). The American Voter. Chicago: University of Chicago Press.

Downs, A. (1957). An Economic Theory of Democracy. New York: Harper.

Erikson, R. S., & Minnite, L. C. (2009). Modeling Problems in the Voter Identification-Voter Turnout Debate. Election Law Journal, 8(2), 85–101.

Frederick, S., Loewenstein, G., & O’Donoghue, T. (2002). Time Discounting and Time Preference: A Critical Review. Journal of Economic Literature, 40(2), 351–401.

Gerber, A. S., Huber, G. A., Biggers, D. R., & Hendry, D. J. (2016). Why Don’t People Vote in U.S. Primary Elections? Assessing Theoretical Explanations for Reduced Participation. Electoral Studies, 45, 119–129.

Giammo, J. D., & Brox, B. J. (2010). Reducing the Costs of Participation: Are States Getting a Return on Early Voting? Political Research Quarterly, 63(2), 295–303.

Gimpel, J. G., Dyck, J. J., & Shaw, D. R. (2007). Election-Year Stimuli and the Timing of Voter Registration. Party Politics, 13(3), 351–374.

Green, D. P., & Palmquist, B. (1994). How Stable is Party Identification? Political Behavior, 16(4), 437–466.

Green, D. P., Palmquist, B., & Schickler, E. (2002). Partisan Hearts and Minds: Political Parties and the Social Identities of Voters. New Haven: Yale University Press.

Grofman, Bernard. (1993). “Is Turnout the Paradox that Ate Rational Choice Theory?” Information, Participation and Choice: An Economic Theory of Democracy in Perspective, 93-103.

Hanmer, M. J., & Kalkan, K. O. (2013). Behind the Curve: Clarifying the Best Approach to Calculating Predicted Probabilities and Marginal Effects from Limited Dependent Variable Models. American Journal of Political Science, 57(1), 263–277.

Highton, B., & Wolfinger, R. E. (1998). Estimating the Effects of the National Voter Registration Act of 1993. Political Behavior, 20(2), 79–104.

Jones, Jeffrey M. (2017). GOP Maintains Edge in State Party Affiliation in 2016. New York: Gallup. https://news.gallup.com/poll/203117/gop-maintains-edge-state-party-affiliation-2016.aspx

Karp, J. A., & Brockington, D. (2005). Social Desirability and Response Validity: A Comparative Analysis of Overreporting Voter Turnout in Five Countries. Journal of Politics, 67(3), 825–840.

Kelley, S., Ayres, R. E., & Bowen, W. G. (1967). Registration and Voting: Putting First Things First. The American Political Science Review, 61(2), 359–379.

Kim, J.-O., Petrocik, J. R., & Enokson, S. N. (1975). Voter Turnout Among the American States: Systemic and Individual Components. The American Political Science Review, 69(1), 107–123.

Knack, S. (1995). Does “Motor Voter” Work? Evidence from State-Level Data. Journal of Politics, 57(3), 796–811.

Knack, S., & White, J. (2000). Election-Day Registration and Turnout Inequality. Political Behavior, 22(1), 29–44.

Lancaster, K. (1963). An Axiomatic Theory of Consumer Time Preference. International Economic Review, 4(2), 221–231.

Madden, G. J., Bickel, W. K., & Jacobs, E. A. (1999). Discounting of Delayed Rewards in Opioid-Dependent Outpatients: Exponential or Hyperbolic Discounting Functions? Experimental and Clinical Psychopharmacology, 7(3), 284–293.

McClure, S. M., Laibson, D. I., Loewenstein, G., & Cohen, J. D. (2004). Separate Neural Systems Value Immediate and Delayed Monetary Rewards. Science, 306(5695), 503–507.

McDonald, M. P., & Popkin, S. L. (2001). The Myth of the Vanaishing Voter. The American Political Science Review, 95(4), 963–974.

Merriam, C. E., & Gosnell, H. F. (1924). Non-Voting: Causes and Methods of Control. Chicago: University of Chicago Press.

Meyer, R. F. (1976). Preferences Over Time. In R. Keeney & H. Raiffa (Eds.), Decisions with Multiple Objectives (pp. 473–489). New York: Wiley.

Mitchell, G. E., & Wlezien, C. (1995). The Impact of Legal Constraints on Voter Registration, Turnout and the Composition of the American Electorate. Political Behavior, 17(2), 179–202.

Morehouse, S.M., Jewell, M.E. (2005). The Future of Political Parties in the States. In Council of State Governments (Ed.), The Book of the States 2005 (pp. 331-345). Lexington: Council of State Governments.

Nagler, J. (1991). The Effect of Registration Laws and Education on U.S. Voter Turnout. The American Political Science Review, 85(4), 1393–1405.

Nickerson, D.W. (2014). Do Voter Registration Drives Increase Participation? For Whom and When? The Journal of Politics, 77(1), 88-101.

Norrander, B. (1986). Measuring Primary Turnout in Aggregate Analysis. Political Behavior, 8(4), 356–373.

Rhine, S. L. (1995). Registration Reform and Turnout Change in the American States. American Politics Quarterly, 23, 409–426.

Rhine, S. L. (1996). An Analysis of the Impact of Registration Factors on Turnout in 1992. Political Behavior, 18(2), 171–185.

Riker, W. H., & Ordeshook, P. C. (1968). A Theory of the Calculus of Voting. The American Political Science Review, 62(1), 25–42.

Rosenstone, S. J., & Hansen, J. M. (1993). Mobilization, Participation and Democracy in America. New York: MacMillian.

Squire, P., Wolfinger, R. E., & Glass, D. P. (1987). Residential Mobility and Voter Turnout. The American Political Science Review, 81(1), 45–65.

Teixeria, R. (1992). The Disappearing American Voter. Washington DC: Brookings.

Thornburg, M. P. (2014). Party Registration and Party Self-Identification: Exploring the Role of Electoral Institutions in Attitudes and Behaviors. Electoral Studies, 36(1), 137–148.

Trope, Y., & Liberman, N. (2003). Temporal Construal. Psychological Bulletin, 110(3), 403–421.

Ware, A. (2002). The American Direct Primary: Party Institutionalization and Transformation in the North. Cambridge: Cambridge University Press.

Wolfinger, R. E., & Rosenstone, S. J. (1980). Who Votes? New Haven: Yale University Press.