Party Identification, Voter Attitudes, and Voter Behavior in New Jersey and Pennsylvania 1980–2012: Real Difference or Election Law Mirage? COMMONWEALTH, Volume 20, Issue 1 (2018). © 2018 The Pennsylvania Political Science Association. ISSN 2469-7672 (online). http://dx.doi.org/10.15367/com.v20i1.180. All rights reserved. Party Identification, Voter Attitudes, and Voter Behavior in New Jersey and Pennsylvania, 1980–2012 Real Difference or Election Law Mirage? JAMES E. VIKE Widener University This study examines the extent to which pronounced differences in voter reg- istration statistics between Pennsylvania and New Jersey are truly indicative of significant differences in political attitudes and voter behavior across citi- zens of these neighboring states. The analysis is based upon an examination of cumulative 1980–2012 American National Election Study (ANES) survey data. Although there is evidence that minor differences in partisan attachments exist, there is no evidence that these differences are manifest in diminished interest, efficacy, polarization levels, participation rates, or straight-ticket voting behav- ior. The central conclusion of this study is that the elevated tendency of New Jersey voters to register as unaffiliated is less of a marker of greater independent orientations and more of an election law mirage that furthers a myth of mount- ing citizen independence from partisan affiliations. Furthermore, these find- ings support the conclusion that a follow-up methodology for determining true partisan attachments among ANES respondents is a more accurate predictor of partisan attitudes and behavior than methods relying upon an initial probe of self-identified partisanship or statewide voter registration statistics. A quick look at 2016 voter registration statistics in New Jersey and Pennsylvania highlights a dramatic difference in overt declarations of party affiliation among citizens of these neighboring states. Voter 62 JaMes e. Vike registration data in the Commonwealth of Pennsylvania, a state with a closed primary system that restricts participation in primary elections to only those registered for a particular party, show that over 85% of registered voters affili- ate with one of the two major parties. New Jersey, on the other hand, features a modified closed primary system, where affiliated voters may only vote in their party primary but unaffiliated voters are free to choose in which primary they wish to participate.1 This difference in election participation policies has allowed 43% of New Jersey voters to register as unaffiliated without relin- quishing their option to participate in primary elections. Figure 1 displays a detailed distribution of 2016 voter registration data between the two states. While both states have greater numbers of registered Democrats than registered Republicans, the difference in registration rates for unaffiliated voters is dramatic and worthy of further study. Subsequent analy- sis of this disparity is necessary and important because a substantial volume of research has found meaningful linkages between party attachments and voter attitudes and behavior. 50 40 e 30 agt enc erP 20 10 0 Pennsylvania  Registered  Voters  2016 New  Jersey  Registered  Voters  2016 Republicans Independents/UNA/NE Democrats figure 1. Comparing Voter Registration Statistics: Pennsylvania and New Jersey.* (Source: Voter registration data for the Commonwealth of Pennsylvania were derived from the PA Department of State, Voter Registration Division’s report of official voter registration statistics from November 8, 2016. Voter registration data for New Jersey were derived from the NJ Department of State, Division of Elections’ monthly voter registration summary report from November 7, 2016.) * To ease comparison with ANES party identification data, percentages of individuals registered with other parties are not displayed in this chart. Just over 5% of PA voters and less than 1% of NJ voters were registered under a third-party banner in 2016. Party Identification, Voter Attitudes, and Voter Behavior in NJ and PA 63 This study examines the extent to which these pronounced differences in voter registration actually indicate real and significant differences in voter attitudes and behavior. More specifically, this study examines cumulative 1980–2012 American National Election Study (ANES) survey data from New Jersey and Pennsylvania to address the following question: Is this disparity in unaffiliated party registration a marker of vastly different attitudes and behaviors among citizens of these states or merely a mirage produced by more liberal registration requirements that further a myth of citizen independence from partisan affiliations? The choice of factors for comparison is informed by numerous studies that have outlined a range of tangible differences across attitudinal and behavioral dimensions linked to the strength of partisan attachments (Campbell et al. 1960; Keith et al. 1992; Lewis-Beck et al. 2008; Magleby, Nelson, and Westlye 2011; Dalton 2013). If the differences in actual voter registration levels between the two states are truly indicative of differing partisan attachments, then we can expect the comparative analysis of ANES data to highlight differences along these dimensions—with the more party-oriented citizens of the Commonwealth of Pennsylvania exhibiting higher rates of interest, efficacy, activism, and polar- ization than the supposedly more “independent-oriented” citizens of New Jersey. Methodological Differences in Tracking Partisan Identification There are two basic approaches to tracking party identification across the American electorate. One approach examines state-level voter registration statistics, while the other examines self-identification rates obtained through responses to survey questions. A solid majority of party identification scholars have tended to prefer the latter approach. The stated rationale for adopting this approach includes: (1) self-identification is deemed to be a more fitting and time-sensitive approach for measuring partisanship when defined as an individual’s psychological attachment to a particular party; (2) actual voter registration statistics are thought to potentially mask true attachments due to vastly different laws, traditions, and customs across the states; and (3) sur- vey responses provide rich datasets enabling in-depth comparative analysis of voter affinities, attitudes, and behaviors. Despite overwhelming reliance upon the self-identification approach in the scholarly community, there still remains some disagreement over the ideal method of tabulating voter attachments among survey respondents. At the core of the dispute is the sequence of two partisan self-identification questions posed in ANES surveys, as originally outlined in The American Voter (Campbell et al. 1960). The initial question probing personal partisan 64 JaMes e. Vike attachments sorts respondents into three pools: Democrats, Republicans, and Independents. A follow-up question probes either the strength of the partisan attachment or whether an independent leaned toward a particular party. The two question sequence has the effect of creating a seven-point scale of partisan identification. In-depth analysis of the resulting data led Campbell and his colleagues to determine that partisan attachment was a crucial predictor of voter attitudes, preferences, and behavior. Disagreements over this methodology emerged several decades later when scholars claimed that increasing numbers of self-identifying independents in the 1970s signaled a significant decline in partisanship in America and across other western democracies (Wattenberg 1984; Miller 1991; Dalton and Wat- tenberg 2000). These “party decline” scholars argued that a rise in the number of individuals identifying as independents in response to the initial survey query indicated a declining attachment to political parties within the elector- ate and subsequently signaled a diminished reliance upon partisanship as a cue for political attitudes and behavior. Dalton (2013) more recently argues that an individual’s initial expression of disaffection from party attachments is a meaningful indicator of an enduring independent-oriented political iden- tity and signals a greater propensity to eschew party-line voting over time. McGhee and Krimm (2009) and Dalton (2013) also rank among the small cohort of scholars who champion analysis of actual voter registration statis- tics as a preferred method of gauging partisan identification patterns among American voters. Both point to modest increases in unaffiliated registrations in recent years as a sign that voters are becoming less sympathetic toward the major parties. McGhee and Krimm additionally assert that the new indepen- dent voters are less polarized along partisan and ideological lines. The early claims of declining partisanship were summarily contested by a team of scholars charging that the apparent rise of independents was actually a product of misguided methodological choices (Keith et al. 1992). The “myth of the independent voter” argument contends that partisan attachments have not really declined, because follow-up partisanship questions in ANES sur- veys show that many self-identified independents clearly lean toward one particular party. They purport that true independents are less interested, less informed, express lower levels of polarization, and participate at substantially lower levels, but dispute that the rolls of independents are on the rise. Several updated studies have reinforced this finding for more recent election cycles, showing that contemporary independent leaners exhibit attitudes and behav- iors more in line with partisans than true independents (Kaufmann, Petro- cik, and Shaw 2008; Petrocik 2009; Magleby, Nelson, and Westlye 2011). A thorough reexamination of the original American voter study reaffirmed the contention that strength of partisan identification impacts party-line voting, Party Identification, Voter Attitudes, and Voter Behavior in NJ and PA 65 candidate assessments, party assessments, interest in campaigns, and concern for election outcomes (Lewis-Beck et al. 2008). A common practice arising from this methodological dispute is to rou- tinely track two different three-point scale measures of partisan identifica- tion: one relying upon the initial expression of partisan identification that places independent-leaners in the category of independent voters; and the other relying upon the follow-up responses that places independent-leaners with their party of preference. Party Identification Variations in Pennsylvania and New Jersey Of particular relevance for this study is the extent to which current Pennsylva- nia and New Jersey voter registration statistics highlight the dispute over the ideal manner for tabulating voter attachments in the United States. Examina- tion of official voter registration statistics shows that these two states closely track with nationwide trends across the dual partisan identification method- ologies. Figure 2 shows that actual voter registration rates in Pennsylvania t enc erP 20 10 0 Republicans Independents/UNA/NE Democrats ANES  Party  ID  With  Leaners  2012 50 40 e 30 ag Pennsylvania  Registered  Voters  2016 figure 2. Comparing Partisan Identification: ANES 3-Point Scale with Leaners and Official Pennsylvania Registration Statistics. (Source: ANES party identification data were derived from the ANES 2012 Time Series Study [available at www.electionstudy.org]. Voter registration data for the Commonwealth of Pennsylvania were derived from the PA Department of State, Voter Registration Division’s report of official voter registration statistics from November 8, 2016.) 66 JaMes e. Vike closely reflect nationwide partisan distribution rates from the 2012 ANES data when independent leaners are included as partisans. Figure 3, on the other hand, shows that contemporary New Jersey voter registration data are similar to the 2012 ANES data when all self-identified independents are clas- sified as independents. Recognizing the clear distinction in partisan registration patterns across these two states and their similarities to the dual approaches to interpreting self-identified party affiliation offers an intriguing opportunity to re-examine the veracity of the “myth of the independent voter” thesis at the state level. Comparative Findings The following tables display findings across a range of party attachment, atti- tude, and behavioral factors examined in the statistical comparison between Pennsylvania and New Jersey citizens. Partisan measures include the dis- tribution of partisan identification in response to the initial partisanship probe, the partisanship distribution after independent leaners are merged figure 3. Comparing Partisan Identification: ANES 3-Point Scale without Leaners and Official New Jersey Registration Statistics. (Source: ANES party identification data were derived from the ANES 2012 Time Series Study [available at www.electionstudy.org]. Voter registration data for New Jersey were derived from the NJ Department of State, Division of Elections’ monthly voter registration summary report from November 7, 2016.) 50 40 e 30 agt enc erP 20 10 0 Republicans Independents/UNA/NE Democrats ANES  Party  ID  Without  Leaners  2012 New  Jersey  Registered  Voters  2016 Party Identification, Voter Attitudes, and Voter Behavior in NJ and PA 67 with partisans, and the strength of partisan attachments. The examination of potential differences across attitudinal measures includes interest in elections, feelings of efficacy toward electoral outcomes, and levels of affective polarized thinking toward the major parties and presidential candidates. Behavioral comparisons include attainment of political information, voter turnout rates, overall political activism levels, and party-line voting. Table 1 collectively displays the empirical results across the range of com- parisons. The initial intriguing finding is the apparent disconnect among the differences across the three dimensions. Each of the three partisan attachment measures indicated statistically significant differences between the states, while only one of eight attitudinal and behavior measures indicated a statisti- cal difference. At first glance, the significant differences in partisan attach- ments offers support for the view that New Jersey’s more independent-friendly voter registration data is actually a marker of a less partisan state. Levels of self-identified affinity to a particular political party are lower in New Jer- sey when measured by the initial question and also once independent leaners are merged with party identifiers. Although statistically significant, the chi- square values are actually quite low and will be examined more closely below. table 1. Differences in Party attachments, Political attitudes, and Voter Behavior—Pennsylvania and new Jersey, 1980–2012 Difference Value Degrees of freedom statistical significance Partisan Measures Party Identification (Initial) χ2 = 24.754 2 0.000*** Party Identification (Leaners as Partisans) χ2 = 9.589 2 0.008** Strength of Party Attachments χ2 = 30.717 3 0.000*** attitudinal Measures Interest in Elections χ2 = 2.061 2 0.357 Concern over Election Outcomes χ2 = 0.144 1 0.704 Presidential Candidate Affect Polarization χ2 = 8.000 4 0.092 Political Party Assessment Polarization F = 5.168 1 0.023* Behavioral Measures Political Information Acquisition χ2 = 4.917 4 0.296 Voter Turnout χ2 = 3.309 1 0.069 Political Activism Levels χ2 = 0.613 3 0.894 Party-line Voting χ2 = 3.030 1 0.082 Source: American National Election Study, Cumulative Data File 1980–2012. *p < 0.05; **p < 0.01; ***p < 0.001 68 JaMes e. Vike The absence of statistical significance across three of the four attitudinal measures and all four behavioral measures is a startling finding given the findings relating to party identification. The nearly uniform absence of differ- ence across attitudinal and behavioral measures is inconsistent with scholarly expectations if New Jersey were actually to be considered a less partisan state than Pennsylvania. Distribution patterns relating to concern over the out- comes of the election and political activism levels are nearly identical between the states. While some minor differences are visible across state distributions of interest levels, candidate assessment polarization, voter turnout, and will- ingness to cast a split-ticket ballot, these differences do not meet the signifi- cance standard. These findings support a conclusion that there are virtually no differences between Pennsylvania and New Jersey citizens pertaining to political attitudes or electoral behavior. A more detailed look at the limited range of statistically significant find- ings is warranted to explore the nature and intensity of partisan and party polarization differences between Pennsylvania and New Jersey. Table 2 high- lights the partisan distributions of state respondents to the ANES surveys over time. The table indicates comparative differences between the states, but also highlights the volatile impact that methodological choice has on efforts to track partisan identification. The percentage value in each cell represents the difference in the partisan identification rates between the initial breakdown based upon responses to the first partisanship question (Initial) and the sub- sequent interpretation that counts partisan identification after independent leaners have been grouped with their party of preference (Follow-up). This table shows that the methodological distinction is particularly impactful on New Jersey, since the initial plurality of independent self-identifiers is attrib- utable to the fact that over one-quarter of respondents can be classified as independent leaners. Table 3 offers a more detailed breakdown of the statistically significant differences in the strength of self-classified partisan attachments between table 2. Methodological impact on Party identification rates, 1980–2012 new Jersey Pennsylvania Party iD (initial) Party iD (follow-up) change Party iD (initial) Party iD (follow-up) change Democrats 29.1% 40.4% +11.3% 39.5% 48.5% +9.0% Independents 43.5% 16.7% –26.8% 29.6% 11.8% –17.8% Republicans 27.5% 42.9% +15.4% 30.9% 39.7% +8.8% Source: American National Election Study, Cumulative Data File 1980–2012. Party Identification, Voter Attitudes, and Voter Behavior in NJ and PA 69 table 3. Distribution of strength of Party identification, 1980–2012 nJ Pa total Independents / Apoliticals 87 16.7% 75 11.8% 159 14.0% Independent Leaners 136 27.0% 116 18.3% 252 22.2% Weak Partisans 161 32.0% 202 31.9% 363 32.0% Strong Partisans 122 24.3% 240 37.9% 362 31.9% Total 503 100.0% 633 100.0% 1136 100.0% χ2 = 30.717, df = 3, Sig = 0.000*, Cramer’s V = 0.167 Source: American National Election Study, Cumulative Data File 1980–2012. *p < 0.001 New Jersey and Pennsylvania citizens. This finding appears to reinforce dis- parate partisan inclinations represented through voter registration data. New Jersey appears more authentically inclined toward independents, featuring a higher rate of pure independents (5% more) and independent leaners (9% more) than does Pennsylvania. In turn, only one-quarter of New Jersey citi- zens self-identify as strong partisans, compared to 38% of Pennsylvanians. Although the resulting chi-square value denotes a statistically significant dif- ference in party affinity between the two states, the strength of the association is quite weak. The low (0.167) Cramer’s V value indicates that party identi- fication differences are certainly less dramatic than the initial scan of party registration statistics might project. Table 4 displays a more detailed comparison of the only attitudinal mea- sure showing a statistically significant difference between New Jersey and table 4. Party assessment Polarization comparison 1980–2012 Mean Party standard Polarization rating Deviation f Value significance eta Value New Jersey 28.95 25.65 (N = 496) 5.168 0.023* 0.068 Pennsylvania 32.69 28.60 (N = 626) Source: American National Election Study, Cumulative Data File 1980–2012. *p < 0.05 70 JaMes e. Vike Pennsylvania citizens. The mean polarization value derived from the feel- ing thermometer ratings of the two major political parties shows that New Jersey citizens are slightly less polarized in their assessments of the parties than Pennsylvanians. The significance and eta values, however, indicate that this difference is substantially limited. When considered in tandem with the absence of significant difference in affect assessments of presidential candi- dates, it can safely be concluded that the pronounced differences in the overt registration patterns of citizens of the two states are not indicative of substan- tial differences in polarization levels. Conclusions Although statistical analysis does demonstrate that partisan affiliation dif- ferences in voter registration statistics between the states are representative of small variances in party attachment, the anticipated associations with dimin- ished partisanship do not hold firm across attitudinal and behavioral mea- sures. In particular, evidence of lesser partisan attachments in New Jersey is not manifest through diminished interest, efficacy, polarization, participation rates, or straight-ticket voting behavior. The fact that disparities in partisan registrations between these two states do not foreshadow meaningful differences in political attitudes or voter behavior offers clear support for one perspective on the central question of this study. The elevated tendency of New Jersey voters to register as unaffili- ated is less of a marker of greater independent orientations and more of an election law mirage that furthers a myth of mounting citizen independence from partisan affiliations. Furthermore, these findings support the conclu- sion that the follow-up methodology for determining true partisan attach- ments among ANES respondents is a more accurate predictor of partisan attitudes and behavior than either the probe of self-identified partisanship or statewide voter registration statistics. NOTE 1. The National Conference of State Legislatures provides an excellent accounting of variations in state primary election laws and policies. See http://www.ncsl.org/research/ elections-and-campaigns/primary-types.aspx for additional information about the cat- egorization of state election systems. REFERENCES Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. New York: Wiley and Sons. Party Identification, Voter Attitudes, and Voter Behavior in NJ and PA 71 Dalton, Russell J., and Martin Wattenberg, eds. 2000. Parties Without Partisans: Political Change in Advanced Industrial Democracies. Oxford: Oxford University Press. Dalton, Russell J. 2013. The Apartisan American: Dealignment and Changing Electoral Politics. Thousand Oaks, CA: CQ Press. Kaufmann, Karen M., John R. Petrocik, and Daron R. Shaw. 2008. Unconventional Wis- dom: Facts and Myths About American Voters. New York: Oxford University Press. Keith, Bruce, David Magleby, Candice Nelson, Elizabeth Orr, Mark Westlye, and Ray- mond Wolfinger. 1992. The Myth of the Independent Voter. Berkeley: University of California Press. Lewis-Beck, Michael, William Jacoby, Helmut Norpoth, and Herbert Weisberg. 2008. The American Voter Revisited. Ann Arbor: University of Michigan Press. Magleby, David, Candice Nelson, and Mark Westlye. 2011. “The Myth of the Independent Voter Revisited.” In Facing the Challenge of Democracy: Exploration in the Analysis of Public Opinion and Political Participation, eds. Paul Sniderman and Benjamin High- ton. Princeton, NJ: Princeton University Press, 238–263. McGhee, Eric, and Daniel Krimm. 2009. “Party Registration and the Geography of Party Polarization.” Polity 41 (July): 345–367. Miller, Warren E. 1991. “Party Identification, Realignment, and Party Voting: Back to Basics.” American Political Science Review 85 (June): 557–568. Petrocik, John. 2009. “Partisan Identification: Leaners Are Not Independents.” Electoral Studies 28 (December): 562–572. Wattenberg, Martin. 1984. The Decline of American Political Parties. Cambridge, MA: Harvard University Press. James Vike is an associate professor of political science and director of the Master of Public Administration program at Widener University. He also serves as campus program director for Widener’s affiliation with Project Pericles. His research interests range from the study of bureaucratic politics to political participation and engagement. His publica- tions include articles in COMMONWEALTH and Politics & Policy.