If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols

Abstract

Real-world tests of the impact of partisan cues on voters are scarce because they require assessing how citizens’ attitudes changed toward an issue from before to after it became politically divisive. During the 2015–2016 New Zealand flag referendums, the leader of the (center-right) National Party and then–Prime Minister, John Key, championed changing the flag—a move strongly contested by the (center-left) Labour Party. Accordingly, we measured New Zealanders’ attitudes toward changing the flag using national longitudinal panel data collected in 2013, before the change was proposed, and again in 2016 at the height of the debate (Ns = 6,793–6,806). Registered voters who supported the National Party were more likely to shift from opposing to supporting the flag change, whereas those who supported the Labour Party were more likely to shift from supporting to opposing the change. These data demonstrate the powerful impact of partisan cues on political attitudes in a real-world setting.

Keywords

partisanship public opinion national symbols politics voting open materials

Democracy rests on citizens’ ability to influence public policy. This may be through voting for political parties on the basis of shared values or voting on contentious referenda. Yet attitudes do not form in a vacuum; rather, myriad factors facilitate attitude formation (Huckfeldt, Mondak, Craw, & Mendez, 2005; Zaller, 1992). One factor that has received longstanding attention is partisanship (i.e., voters’ attachment to political parties; Campbell, Converse, Miller, & Stokes, 1960). Indeed, partisanship has been described as the “unmoved mover,” a social identity that fundamentally shapes policy preferences (see Campbell et al., 1960). Yet partisanship can be problematic, as many citizens may simply echo the visions of their political parties and leaders rather than express their own ideals.

The effects of partisanship on individuals’ attitudes have been researched primarily through experimental cuing and framing studies. Cues subtly communicate the stance that a political party or figure has taken on an issue, whereas frames affect how people think about those issues (Bullock, 2011; Chong & Druckman, 2007; Iyengar & Kinder, 1987). Such studies examine the extent to which cue exposure leads participants to align their views with those expressed by their preferred political faction. Many psychological theories including social identity theory (Greene, 2004; Iyengar, Sood, & Lelkes, 2012), partisan-motivated reasoning (Bolsen, Druckman, & Cook, 2014; Druckman, Peterson, & Slothuus, 2013; Leeper & Slothuus, 2014), and shared reality theory (Hardin & Higgins, 1996; Jost, Ledgerwood, & Hardin, 2008) have been used to explain why partisan effects occur. These approaches generally suggest that people form deep attachments to the political groups with which they identify and are motivated to think positively of, and hold attitudes consistent with, those groups.

Numerous studies have demonstrated partisan effects on peoples’ attitudes toward various policies and issues and across diverse contexts (e.g., Brader & Tucker, 2012; Bullock, 2011; Kam, 2005; Nicholson, 2011; Slothuus, 2010; Slothuus & De Vreese, 2010). For example, liberal students in the United States expressed greater support for harsh welfare policies when told that Democrats supported the policy, and conservative students in the United States expressed greater support for generous welfare policies when told that Republicans supported the policy (Cohen, 2003). Goren, Federico, and Kittilson (2009) also showed that people express greater support for core values (e.g., self-reliance and equal opportunity) after hearing that their partisan in-group supports such values. The effectiveness of partisan cues in shaping public attitudes can also be moderated by numerous factors including the perceived importance of the issue (Ciuk & Yost, 2016), political awareness (Kam, 2005; Slothuus & De Vreese, 2010), and degree of polarization over the issue (Carsey & Layman, 2006; Druckman et al., 2013; Leduc, 2002).

Although they are informative, the external validity of these experimental findings is somewhat inconclusive. Sniderman and Theriault (2004) argued that exposing participants to just one frame or cue is unrepresentative of real-world politics, in which people are exposed to numerous competing cues and frames. Kinder (2007) similarly noted that participants’ near guaranteed exposure to frames in experiments may overestimate their impact and called for more real-world examples of framing effects. Although we know much about the effects of partisanship on individuals’ attitudes in controlled experimental settings, real-world demonstrations are rare.

One exception is research by Slothuus (2010), who found a 13% increase in Danish Social Democrat voters’ support for the abolishment of early retirement benefits after the new Social Democrat Party leader changed the framing of the issue from a social to an economic one. However, Slothuus’s design was cross-sectional and could not assess change in the same individuals over time. Hence, this work cannot speak to within-person change.

Here, we present a rare analysis of partisan effects in a natural setting using national probability longitudinal panel data collected before and after the 2015–2016 New Zealand flag-change referendums. In the first referendum in December 2015, New Zealanders voted on their preferred alternative flag design from a selection of five, which a nonpartisan panel had narrowed down from thousands of public submissions. At the second referendum in March 2016, New Zealanders voted to either retain the flag or change it to a silver fern design. With 67.8% voter turnout, 56.6% of voters opted to retain the flag (Electoral Commission New Zealand, 2016).

Although the referendums allowed New Zealanders to express their personal flag preferences, the referendums quickly became polarized along party lines. John Key, then–Prime Minister and leader of the center-right National Party, openly advocated for changing the flag. Across numerous widely publicized video interviews, Key argued that New Zealand needed a more representative flag that would promote a greater sense of patriotism, one that would stand out from the Australian flag, and that another chance to change the flag would likely not occur for many years (Key, 2015; Moir, 2015). Conversely, Andrew Little, then-leader of the center-left opposition party (namely, Labour), opposed the referendums. Little argued that New Zealanders were uninterested in the process and that the millions it would cost to run the referendums would be better spent elsewhere. Like many pundits, Little suggested that changing the flag was a personal project by which Key wished to be remembered (Jones, 2015; Little, 2016; for further details on the debate, see “Political Context” in the Supplemental Material available online).

The divide between the two party leaders dominated media coverage of the referendums. Figure 1 displays the number of articles referring to flag change published in The New Zealand Herald (the largest newspaper in New Zealand) each month from when the referendums were announced to shortly after they concluded. The flag-change issue first appeared particularly prominent during September 2015, a month in which Key and Little partook in prolonged arguments over the referendum process and whether to include an additional design in the first referendum backed by a public petition (see Trevett, 2015).

Fig. 1.

Number of articles published per month on The New Zealand Herald website related to changing the flag from 2014 to 2016. Data were obtained on March 1, 2018, from a Google search of “flag change National OR Labour OR Key OR Little site:nzherald.co.nz.”

Table 1 displays a timeline of polling data collected on New Zealanders’ flag attitudes over the years. Opposition to change appeared high (72%) following the initial announcement of the referendums and subsequently fluctuated around 55% to 65%. Polls that included voter-specific data reveal that National Party voters were particularly supportive of change in late August 2015 (shortly after Key further reinforced his support for change in the media; see Key, 2015; Moir, 2015), whereas Labour voters were evenly split at the time. From October 2015, approximately 70% of Labour Party voters opposed change (considerably higher than the aggregate), whereas National Party voters continued to express greater support than the aggregate. These shifts appear to correspond with the increased prominence of the flag-change issue seen in Figure 1, suggesting that the salience of party divides encouraged shifts in Labour voters’ flag attitudes. These data are indicative of partisan shifts in a real-world setting (and broadly consistent with the findings of Slothuus, 2010); however, the same individuals were not tracked over time, and the use of different measures over time further complicates interpretation of these data.

Table 1.

Polling Data on New Zealanders’ Attitudes Toward Changing the Flag From 2010 to 2016

Poll	Date	Change	Retain	Unsure	Additional information and voter-specific data
Herald Digipoll (Cheng, 2010)	January 29–February 10, 2010	52%	44%	4%
Herald Digipoll (Cheng, 2011)	December 3–19, 2010	22%	72%	6%
Research New Zealand (2014)	August 2011	19%	52%	1%	Included neutral option: 30%
Research New Zealand (2014)	February 2014	22%	37%	1%	Included neutral option: 39%
One News Colmar Brunton (Curia Market Research, 2014)	February 15–19, 2014	28%	72%
Herald Digipoll (Davison, 2014)	March 6–16, 2014	40.6%	52.6%	6.8%
Research New Zealand (2014)	March 2014	18%	37%	2%	Included neutral option: 43%
Research New Zealand (2014)	September–October 2014	19%	43%	1%	Included neutral option: 37%
Herald Digipoll (Trevett, 2015)	April 17–26, 2015	25%	70%	5%	Taken during coverage of ANZAC day
Milne (2015)	July 8–November 30, 2015	40%	60%		Change responses include “yes depending on design”National: 46% change, 54% retainLabour: 34% change, 66% retain
Herald Digipoll (Trevett, 2015)	August 14–24, 2015	47%	53%		24% support change, depending on the alternative23% support change in principle
UMR Research (UMR Research, 2015a)	August 25–31, 2015	52%	48%		Current flag compared with favorite alternative designNational: 60% change, 40% retainLabour: 51% change, 49% retainFor current flag compared with what became the final alternative design: 56% change, 44% retain overall
One News Colmar Brunton (Curia Market Research, 2015)	August 29–September 2, 2015	28%	66%	6%
3News Reid Research (Gower, 2015)	September 8–16, 2015	25%	69%	6%
Research New Zealand (2015)	September 2015	28%	61%	11%
UMR Research (UMR Research, 2015b)	October 1–15, 2015	35%	65%		Final alternative vs. current flagNational: 47% change, 53% retainLabour: 28% change, 72% retain
3News Reid Research (Gower, 2015; Trevett, 2015)	November 9–17, 2015	28%	65%	7%	National: 39% change, 54% retainLabour: 20% change, 77% retain
Horizon Poll (Horizon Poll, 2015)	Published December 8, 2015	34%	58%	7%
UMR Research (UMR Research, 2016a)	January 29–February 15, 2016	35%	65%		Final alternative vs. current flagNational: 50% change, 50% retainLabour: 30% change, 70% retain
Newshub Reid Research (George, 2016; Gower, 2016)	Published February 1, 2016	30%	61%	9%	Final alternative vs. current flagNational: 41% change, 48% retainLabour: 20% change, 73% retain
Curia Market Research (Voxy, 2016)	January 28–February 2, 2016	36%	56%	9%
One News Colmar Brunton (Curia Market Research, 2016; One News, 2016)	February 13–17, 2016	26%	63%	7%	National: 40% changeLabour: 76% retain
UMR Research (UMR Research, 2016b)	February 25–29, 2016	32%	59%	9%	Final alternative vs. current flagNational: 45% change, 45% retainLabour: 19% change, 73% retain
UMR Research (UMR Research, 2016c)	March 10–15, 2016	35%	58%	7%	Final alternative vs. current flagNational: 52% change, 43% retainLabour: 25% change, 72% retain

Note: Poll references are available in Supplemental References in the Supplemental Material available online. ANZAC = Australian and New Zealand Army Corps. In each row, the boldface percentage represents the most common response for the given poll.

Complementing previous research on partisanship in experimental settings, our findings provide rare insight into within-person effects of partisanship in a real-world setting. Our analyses capitalized on longitudinal panel data tracking registered voters’ attitudes toward changing the flag both before and after the referendums were introduced and politicized. We examined whether New Zealanders’ flag-change attitudes were swayed by political partisanship, measured through levels of support for the National and Labour Parties (as well as the party they voted for in the 2011 general election). Given the impact of partisan cues on policy support in past research, we expected that National Party supporters would have a higher probability, and Labour Party supporters a lower probability, of shifting over time to support changing the flag. Additionally, we expected more National Party voters, but fewer Labour Party voters, to support a flag change in 2016, compared with 2013.

Method

Sampling procedure

Data come from the 2012–2013 (referred to here as 2013) and 2015–2016 (2016) waves of the New Zealand Attitudes and Values Study (NZAVS), a national longitudinal panel of New Zealand adults sampled from the New Zealand electoral roll. The electoral roll contains contact details of all registered voters; registration from age 18 years is compulsory.

The 2013 sample consisted of 12,182 participants, of whom 10,071 answered the flag-change question. Data collection for this wave took place from late November 2012 and throughout 2013. Following successful booster sampling, 13,944 participants completed the 2016 NZAVS, of whom 13,561 answered the flag-change question. Notably, 8,261 (67.8%) participants in the 2016 sample were retained from the 2013 sample. Data collection for this wave took place from November 2015 and throughout 2016.

Further details on the NZAVS sampling procedure can be found in the work by Sibley (2014). Analyses by Sibley et al. (2017) demonstrated that data from the 2014 NZAVS closely tracked New Zealanders’ voting intentions (as well as the election results) as measured by one of the leading New Zealand random-digit-dial political polls (i.e., the Colmar-Brunton TVNZ poll). The NZAVS thus provided an accurate picture of New Zealanders’ voting intentions throughout the election cycle.

Participants

All analyses were limited to participants who completed the measures relevant to the main analyses (N = 6,806, for the largest main analysis, of which we report the demographics here). Of those participants, 92% identified as New Zealand European, 10% as Ma¯ori, 3.4% as Pacific Islander, 3.7% as Asian, and 2.2% as other (because participants could identify with multiple ethnicities, percentages do not sum to 100); 62% of participants were women, 40% were religious, 80% were born in New Zealand, and 73% were employed. The NZAVS sample thus overrepresented women by approximately 10%, as women form approximately 52% of the New Zealand population. The mean age for the sample was 51.01 years (SD = 14.51). The percentage of participants voting for a named political party in the sample analyzed here were as follows: National (47.1%), Labour (24.9%), Greens (18.3%), New Zealand First (3.7%), Ma¯ori (1.9%), Conservative (1.1%), Association of Consumers and Taxpayers (ACT; 0.8%), Mana (0.6%), and United Future (0.5%).

Sample size and statistical power

The possibility of changing the New Zealand flag had not been contested by political parties or publicly debated in depth until the referendums were announced in 2014–2015. The 2013 NZAVS sample size was thus determined well before we formulated our hypotheses, as the flag referendum had yet to be initiated. Similarly, the 2016 NZAVS sample size was determined by our retention of participants from the 2013 wave. Because sample size was predetermined, power analyses were not conducted. Relatedly, the NZAVS is a longitudinal panel study. Thus, we did not have a stopping criterion for data collection in the same way that an experiment might. Instead, we made every effort to retain existing participants by sending postal letters, making phone calls, and sending e-mails (when available) to nonrespondents who were enrolled in the study.

Measures

Participants’ support for changing the New Zealand flag was measured with one item at both time points: “Should the design of the New Zealand flag be changed?” (yes, no, or don’t know). To assess support for the National and Labour Parties in 2013, participants were asked to “rate how strongly you oppose or support each of the following political parties” (1 = strongly oppose, 7 = strongly support). Participants were also asked, “Did you vote in the last (2011) New Zealand general election?” (yes or no) and “If yes, to what party did you give your party vote?” (open ended).

Results

Table 2 displays the percentage of total respondents, National voters, and Labour voters who wanted to change, wanted to retain, or were unsure about changing the flag in 2013, before party leaders’ opinions were voiced, and in 2016, after opinions were voiced. Although participants reported voting for numerous political parties, we examined participants who stated that they voted for either the National Party or Labour Party. These are by far the largest political parties in New Zealand, and the respective party leaders expressed the strongest (opposing) stances on the flag-change debate.

Table 2.

Percentage of Participant Responses to the Item “Should the Design of the New Zealand Flag Be Changed?” in 2013 and 2016

Group	2013			2016
Group	Don’t know	No	Yes	Don’t know	No	Yes
Total sample (N = 6,908)	16.4%	50.0%	33.6%	10.9%	50.5%	38.6%
National voters (N = 2,916)	13.0%	59.2%	27.8%	10.0%	43.7%	46.3%
Labour voters (N = 1,552)	17.5%	45.9%	36.6%	10.7%	59.5%	29.8%

Note: Cells that are boldface within the same row reflect the focal effect of partisan cues on flag-change support.

Among the total sample (see Table 2), the percentage of participants wanting to retain the flag was similar at each time point, although the percentage who supported the flag change increased overall. Notably, different patterns of change emerged for National and Labour voters. Although only 27.8% of National voters wanted to change the flag in 2013, this percentage rose to 46.3% in 2016 (i.e., following the politicization of the issue), slightly more than the percentage wanting to retain the flag (43.7%). Labour voters also tended to oppose change in 2013 (45.9%) but expressed even greater opposition by 2016 (59.5%). Thus, whereas New Zealanders overall were slightly more favorable toward change in 2016, shifts in the attitudes of National and Labour voters appeared to polarize along partisan lines.

These changes are further distilled in Table 3, which displays the percentage of total respondents, National voters, and Labour voters who fall into each paired-response category. As shown here, the percentage of National and Labour voters who consistently wanted to change, wanted to retain, or were unsure about changing the flag were roughly comparable. Clear differences emerge, however, when we examine respondents who changed their views over time. Indeed, 18.9% of National voters moved from wanting to retain to wanting to change the flag, compared with only 6.3% of Labour voters. Conversely, 12.4% of Labour Party voters supported change before the referendum process and yet wanted to retain the flag after the process, whereas only 4.4% of National Party voters moved from supporting to opposing change. Overall, 24.5% of National Party voters and 23.2% of Labour Party voters changed their previously stated attitude to adopt a position consistent with their party leader. These figures rise to 30.5% and 27.5%, respectively, when we include respondents who initially held an opinion contrary to their leader but became unsure in 2016. Substantial proportions of National and Labour Party voters changed their attitudes to align with their preferred political party following the flag-change discussion.

Table 3.

Percentage of National and Labour Voters, Compared With the Whole Sample, in Each of the Paired (2013–2016) Response Categories

Group	Consistent change	Consistent retain	Consistent unsure	Change 2013–retain 2016	Unsure 2013–retain 2016	Change 2013–unsure 2016	Retain 2013–unsure 2016	Retain 2013–change 2016	Unsure 2013–change 2016
Total sample (N = 6,908)	21.7%	33.5%	3.3%	8.8%	8.1%	3.1%	4.5%	12.0%	4.9%
National voters (N = 2,916)	21.7%	34.2%	2.4%	4.4%	5.0%	1.6%	6.0%	18.9%	5.6%
Labour voters (N = 1,552)	20.0%	36.3%	3.1%	12.4%	10.8%	4.3%	3.4%	6.3%	3.6%

Markov models of flag attitudes

To formally assess changes in registered voters’ flag attitudes over time, we employed Markov modeling to estimate the probability of transitioning among (or remaining in) each of the three response states of the categorical flag-attitude (retain, unsure, change) variable in 2013 and 2016. This approach thus estimated the probability that registered voters maintained a consistent attitude over time, as well as the probabilities of moving from each given response state (or attitude) to each of the other response states (retain, unsure, change).

We conducted three Markov models with differing sets of covariates. Rather than using a single reference category for the models, we varied the reference categories so that they aligned with each state in 2013. In other words, the regression coefficients always represent the log odds of transitioning to a different state, relative to remaining stable in the same state over time. These coefficients represent the same model, but this allowed us to examine the most relevant aspects of each model (i.e., how party support influences the tendency to transition between states vs. remain stable in a state, over time). Model 1 included only ratings of support for the National Party and the Labour Party as covariates.¹ Table 4 displays the results of the multinomial logistic regression for this model. As expected, National Party support was significantly associated with flag attitudes in all but one case. However, support for the Labour Party was not significantly associated with flag attitudes after adjusting for National Party support. This is likely because of redundancy in the use of both variables to assess attitudes toward the two parties, which tend to reflect different poles of a single National–Labour continuum. Indeed, levels of party support were highly negatively correlated, r = −.60, p < .001.

Table 4.

Results of the Multigroup Multinomial Logistic Regression Examining the Log Odds of Transitioning to a Different State Relative to Remaining in the Same State Associated With Ratings of Support for the National Party and the Labour Party (Model 1)

Flag-change response and predictor variable in 2013	Unsure in 2016					Change in 2016					Retain in 2016
Flag-change response and predictor variable in 2013	b	SE	b/SE	OR [95% CI]	p	b	SE	b/SE	OR [95% CI]	p	b	SE	b/SE	OR [95% CI]	p
Retain in 2013
National Party support	0.153	0.038	4.03	1.165 [1.082, 1.255]	< .001	0.291	0.029	10.06	1.337 [1.264, 1.415]	< .001
Labour Party support	0.007	0.040	0.17	1.007 [0.932, 1.088]	.868	0.002	0.029	0.06	1.002 [0.946, 1.061]	.950
Unsure in 2013
National Party support						0.302	0.058	5.17	1.353 [1.206, 1.517]	< .001	0.020	0.051	0.38	1.020 [0.923, 1.127]	.701
Labour Party support						0.087	0.068	1.28	1.091 [0.955, 1.246]	.201	0.092	0.060	1.53	1.096 [0.974, 1.233]	.127
Change in 2013
National Party support	−0.176	0.046	−3.83	0.839 [0.766, 0.918]	< .001						−0.287	0.033	−8.83	0.750 [0.704, 0.800]	< .001
Labour Party support	0.063	0.055	1.15	1.065 [0.957, 1.186]	.249						−0.053	0.038	−1.37	0.949 [0.880, 1.023]	.170

Note: N = 6,768. Unstandardized coefficients (bs) represent log odds of transitioning to a different state in 2016, relative to remaining in the same state. OR = odds ratio; CI = confidence interval.

Because of this multicollinearity, our subsequent models focused on how support for either the National Party (Model 2) or the Labour Party (Model 3) separately moderated the probability that registered voters changed their flag attitudes. In both cases, these two models also included (and hence adjusted for) gender (0 = women, 1 = men), age, and ethnicity (1 = New Zealand European, 0 = other). These models are complex relative to standard multiple regression or structural equation models more typically seen in psychological research. Therefore, we restricted our models to adjust for only these focal covariates because we deemed them to be the most plausible third variables that might also account for the observed effect of political party support on flag-change attitudes. As shown in Tables 5 and 6, when support for the National Party and the Labour Party was modeled separately, both significantly moderated change and stability in registered voters’ flag attitudes over time, albeit in different directions.

Table 5.

Results of the Multigroup Multinomial Logistic Regression Examining the Log Odds of Transitioning to a Different State Relative to Remaining in the Same State With Support for the National Party as the Focal Moderator (Model 2)

Flag-change response and predictor variable in 2013	Unsure in 2016					Change in 2016					Retain in 2016
Flag-change response and predictor variable in 2013	b	SE	b/SE	OR [95% CI]	p	b	SE	b/SE	OR [95% CI]	p	b	SE	b/SE	OR [95% CI]	p
Retain in 2013
Gender	−0.193	0.133	−1.45	0.824 [0.635, 1.070]	.146	0.174	0.086	2.02	1.190 [1.005, 1.409]	.043
Ethnicity	0.210	0.223	0.94	1.233 [0.796, 1.910]	.348	0.236	0.163	1.45	1.266 [0.920, 1.744]	.148
Age	−0.022	0.004	−5.15	0.978 [0.970, 0.986]	< .001	−0.011	0.003	−4.12	0.989 [0.983, 0.994]	< .001
National Party support	0.168	0.033	−5.04	1.183 [1.108, 1.263]	< .001	0.294	0.025	11.96	1.342 [1.279, 1.409]	< .001
Unsure in 2013
Gender						0.287	0.191	1.51	1.333 [0.917, 1.938]	.132	−0.102	0.176	−0.58	0.903 [0.640, 1.276]	.564
Ethnicity						−0.503	0.334	−1.50	0.605 [0.314, 1.165]	.133	−0.853	0.310	−2.75	0.426 [0.232, 0.782]	.006
Age						0.006	0.006	1.03	1.007 [0.944, 1.019]	.303	0.016	0.006	2.82	1.017 [1.005, 1.028]	.005
National Party support						0.262	0.048	−5.41	1.299 [1.182, 1.429]	< .001	−0.029	0.042	−0.68	0.972 [0.894, 1.056]	.498
Change in 2013
Gender	−0.510	0.163	−3.13	0.600 [0.436, 0.827]	.002						−0.141	0.101	−1.39	0.869 [0.712, 1.060]	.166
Ethnicity	−0.482	0.301	−1.60	0.618 [0.342, 1.114]	.109						−0.598	0.202	−2.96	0.550 [0.370, 0.817]	.003
Age	−0.016	0.006	−2.78	0.984 [0.973, 0.995]	.005						0.002	0.004	0.43	1.002 [0.994, 1.009]	.664
National Party support	−0.177	0.036	−4.96	0.838 [0.781, 0.898]	< .001						−0.254	0.026	−9.75	0.776 [0.737, 0.816]	< .001

Note: N = 6,806. Unstandardized coefficients (bs) represent log odds of transitioning to a different response state in 2016, relative to remaining in the same state. OR = odds ratio; CI = confidence interval.

Table 6.

Flag-change response and predictor variable in 2013	Unsure in 2016					Change in 2016					Retain in 2016
Flag-change response and predictor variable in 2013	b	SE	b/SE	OR [95% CI]	p	b	SE	b/SE	OR [95% CI]	p	b	SE	b/SE	OR [95% CI]	p
Retain in 2013
Gender	−0.168	0.132	−1.27	0.845 [0.652, 1.095]	.203	0.181	0.085	2.12	1.199 [1.014, 1.417]	.034
Ethnicity	0.202	0.219	0.92	1.224 [0.797, 1.882]	.356	0.284	0.157	1.81	1.328 [0.977, 1.805]	.070
Age	−0.021	0.004	−4.78	0.979 [0.971, 0.988]	< .001	−0.010	0.003	−3.69	0.990 [0.985, 0.995]	< .001
Labour Party support	−0.105	0.035	−3.05	0.900 [0.841, 0.963]	.002	−0.172	0.024	−7.28	0.842 [0.804, 0.882]	< .001
Unsure in 2013
Gender						0.299	0.190	1.58	1.349 [0.930, 1.957]	.115	−0.082	0.178	−0.46	0.921 [0.650, 1.304]	.643
Ethnicity						−0.487	0.344	−1.42	0.614 [0.313, 1.206]	.157	−0.844	0.311	−2.71	0.430 [0.234, 0.792]	.007
Age						0.009	0.006	1.44	1.009 [0.997, 1.021]	.150	0.017	0.006	2.89	1.017 [1.005, 1.029]	.004
Labour Party support						−0.112	0.058	−1.95	0.894 [0.798, 1.001]	.052	0.083	0.053	1.56	1.086 [0.979, 1.205]	.120
Change in 2013
Gender	−0.527	0.163	−3.24	0.590 [0.429, 0.812]	.001						−0.189	0.100	−1.88	0.828 [0.680, 1.008]	.060
Ethnicity	−0.535	0.304	−1.76	0.585 [0.323, 1.062]	.078						−0.681	0.202	−3.38	0.506 [0.341, 0.752]	.001
Age	−0.017	0.006	−2.97	0.983 [0.972, 0.994]	.003						0.000	0.004	−0.09	1.000 [0.992, 1.007]	.929
Labour Party support	0.158	0.043	3.63	1.171 [1.075, 1.275]	< .001						0.135	0.031	4.28	1.144 [1.076, 1.217]	< .001

Note: N = 6,793. Unstandardized coefficients (bs) represent log odds of transitioning to a different response state in 2016, relative to remaining in the same state. OR = odds ratio; CI = confidence interval.

Because flag attitudes were assessed in the context of the referendums at the second, but not the first, measurement, aspects of the referendums (e.g., knowledge of the alternative flag designs or support for the actual referendums), rather than partisanship, may have affected how participants responded to our flag-attitudes measure. The time span of data collection at the second measurement occasion meant that some responses were received after the referendums had concluded and were therefore less (if at all) relevant to responding to our measure. Thus, Models 2 and 3 were reanalyzed using survey responses received before the conclusion of the flag referendums and after. The results showed minimal differences in changes in attitudes over time between responses received before and after the referendums, suggesting that participants’ responses to our flag-change measure were not unduly affected by the different context at the second measurement occasion. Regression parameters for these models are available in “Analyses by Response Time” in the Supplemental Material.

Markov model transition probabilities

Figure 2 displays the probability of staying in, or transitioning among, each of the three response states (retain, unsure, change) of the categorical flag-attitude variable in 2013 and 2016 for registered voters with high or low levels of support for the National Party (top row) or the Labour Party (bottom row). These transition probabilities represent the focal test of our predictions and were derived from logistic regression parameters (see Muthén & Asparouhov, 2011). Similarly, log-odds and p values (presented in Tables 5 and 6) are directly based on transition probabilities, whereby the odds are obtained from P(transition)/P(stay), and the log-odds represent the log of the change in odds associated with a 1-unit increase in party support. The transition probabilities in Figure 2, however, were solved at high (+1 SD; left) and low (−1 SD; right) levels of support (National: M = 4.08, SD = 1.99; Labour: M = 4.12, SD = 1.72), derived from Models 2 and 3, respectively.

Fig. 2.

Transition probabilities for National (top row; Model 2) and Labour (bottom row; Model 3) support at high (left) and low (right) levels of party support. Note that all other model covariates were solved at mean levels.

As shown in Table 5, among respondents who wanted to retain the flag in 2013, support for the National Party was associated with increased odds of being unsure (p < .001) and wanting to change the flag (p < .001) in 2016, relative to remaining in the retain state. As Figure 2 indicates, the probability of becoming unsure about changing the flag (.099) relative to remaining in the retain state (.586), among registered voters with high levels of support for the National Party, was 1.95 times that of those with low levels of support for the National Party, P(unsure|retain) = .069, P(retain|retain) = .798. That is, (.099/.586)/(.069/.798) = 1.95. Furthermore, the probability of becoming supportive of change (.315) relative to remaining in the retain state among respondents with high levels of support for the National Party was 3.22 times that of those with low levels of support for the National Party, P(change|retain) = .133, P(retain|retain) = .798. This indicates that people who supported the National Party were less likely to maintain a view that contrasted the position advocated by John Key (the leader of the National Party) and more likely to shift their views to align with Key’s or become unsure about changing the flag.

Among respondents who were unsure about changing the flag in 2013, support for the National Party was associated with increased odds of wanting to change the flag in 2016 (p < .001) but not significantly associated with the odds of wanting to retain the flag (p = .498), relative to remaining unsure (see Table 5). The probability of becoming supportive of change (.449) relative to remaining unsure (.166) among registered voters with high levels of support for the National Party was 2.83 times that of those with low levels of support for the National Party, P(change|unsure) = .210, P(unsure|unsure) = .220. Thus, registered voters who were unsure about changing the flag in 2013 were more likely to support change in 2016 if they had high (vs. low) support for the National Party, but there was no evidence to suggest that they were more or less likely to want to retain the flag in 2016 than those with low support.

Finally, among respondents who wanted to change the flag in 2013, support for the National Party was associated with lower odds of wanting to retain the flag (p < .001) and being unsure (p < .001) in 2016, relative to wanting change. The probability of transitioning to wanting to retain the flag (.164) relative to remaining supportive of change (.766) among respondents with high support for the National Party was about 0.36 times that of those with low support for the National Party, P(retain|change) = .331, P(change|change) = .564. Furthermore, the probability of transitioning to being unsure about the flag (.070) relative to remaining supportive of change for respondents with high support for the National Party was about 0.49 times that of those with low support for the National Party, P(unsure|change) = .105. Thus, respondents with high support for the National Party who supported change in 2013 were less likely to become unsure or want to retain the flag and more likely to remain supportive of change, compared with those with low support for the National Party.

The opposite pattern emerged when we examined Labour Party support. Table 6 shows that increased support for the Labour Party among respondents who wanted to retain the flag in 2013 was associated with lower odds of being unsure or wanting to change the flag in 2016, relative to wanting to retain the flag (p = .002 and p < .001, respectively). As shown in the lower half of Figure 2, the probability of becoming unsure about changing the flag (.078) relative to remaining in the retain state (.737) among registered voters with high levels of support for the Labour Party was about 0.70 times that of those with low levels of support for the Labour Party, P(unsure|retain) = .094, P(retain|retain) = .623. Furthermore, the probability of becoming supportive of change (.185) relative to remaining in the retain state for respondents with high levels of support for the Labour Party was 0.55 times that of those with low levels of support for the Labour Party, P(change|remain = .282, P(remain|remain) = .623. This indicates that respondents who supported the Labour Party were less likely to become unsure or supportive of change if they previously wanted to retain the flag and more likely to maintain that view.

By contrast, among respondents who were unsure about changing the flag in 2013, Labour Party support was not significantly associated with the odds of wanting to retain the flag and wanting to change the flag in 2016 relative to remaining unsure (p = .120 and p = .052, respectively). In other words, there was no evidence to suggest that the tendency to change views on the flag, relative to remaining unsure, differed between high and low Labour Party supporters who were unsure in 2013.

Finally, among respondents who wanted to change the flag in 2013, increased support for the Labour Party was associated with greater odds of wanting to retain the flag and being unsure in 2016 relative to remaining supportive of change (ps < .001). The probability of transitioning to the retain state in 2016 (.300) relative to the probability of remaining supportive of change (.595) for respondents with high support for the Labour Party was 1.45 times that of those with low support for the Labour Party, P(retain|change) = .224, P(change|change) = .704. The probability of becoming unsure (.105) relative to remaining supportive of change for respondents with high support for the Labour Party was also 1.73 times higher than for those with low support for the Labour Party, P(unsure|change) = .072. Thus, Labour Party supporters initially supportive of change were less likely to maintain that view and more likely to want to retain the flag or feel unsure in 2016, becoming more congruent with the stance of Andrew Little and the Labour Party.

Discussion

Our analyses suggest that partisan cues shifted New Zealanders’ attitudes on changing the flag. After both party leaders publicly expressed contrasting views on the flag-change referendums, approximately one quarter of National and Labour voters changed their opinion to align with their preferred party. Furthermore, compared with respondents with low levels of support for the respective party, National Party supporters had a higher probability of shifting attitudes to support changing the flag in 2016, whereas Labour Party supporters had a higher probability of shifting attitudes to oppose changing the flag in 2016. Whereas most studies show absolute levels of policy endorsement after exposure to a party cue, our research highlights that even previously held attitudes toward an issue can be changed—and sometimes reversed—after exposure to partisan cues. Furthermore, our participants’ responses changed in relation to actual experiences in New Zealand between completing each survey (i.e., the political discussion of the issue) rather than to artificially manipulated conditions. Knowing that partisanship could have such a considerable effect on previously held attitudes reinforces the importance of considering psychological explanations of political partisanship.

Although our results demonstrate partisanship effects in the real world, some questions remain. For example, did Labour supporters vote in line with their party leader or against their political opponents? Past research shows that other-party cues influence attitudes, sometimes to a larger degree than own-party cues (see Goren et al., 2009). It is likely that Labour voters changed their opinions to oppose John Key, given that he was particularly vocal throughout the referendum process. However, Little and the Labour Party vocally opposed the referendum process and thereby potentially reinforced this effect. Furthermore, we were unable to examine the extent to which New Zealanders thought that changing the flag was an important issue, a factor associated with the strength of partisan effects in past research (e.g., Ciuk & Yost, 2016). Many commentators (including the Labour Party) suggested that New Zealanders were uninterested in changing the flag. Thus, partisanship may have been a stronger determinant of attitude change for this issue than it would be for other issues.

Our use of longitudinal panel data is important given concerns over the external validity of experimental framing studies (e.g., Kinder, 2007; Sniderman & Theriault, 2004) but is not without its own caveats. For example, it is possible (albeit unlikely) that some participants changed their attitudes in relation to other events that occurred between measurements. Our analyses in the Supplemental Material help rule out the possibility that the effects seen here were contingent on aspects of the referendums themselves that occurred at the second, but not the first, measurement. Additionally, prior research has shown that National and Labour voters had identical top-four alternative flag design preferences before the flag designs were narrowed down for the first referendum (UMR Research, 2015). Thus, our results are unlikely to be explained by differences in flag design preferences.

Conclusion

Given the increasingly partisan nature of contemporary political discourse, it is important to investigate the consequences of partisan cues on vote choice. Here, we show that partisan cues can powerfully influence public opinion and change voters’ previously held views. Our results thus demonstrate that the divisive stances taken by political leaders can entice voters to follow their partisan ties rather than their personal preferences (also see Iyengar et al., 2012). Furthermore, we show that these effects occur in real-word contexts during times in which voters are presumed to have the most influence over political outcomes. By increasing understanding of this critical topic, we hope to shed light on the potentially subversive nature of partisan cues in order to mitigate their effects on voters.

Supplemental Material

SatherleyAnalysesbyResponseTime – Supplemental material for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols

Supplemental material, SatherleyAnalysesbyResponseTime for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols by Nicole Satherley, Kumar Yogeeswaran, Danny Osborne and Chris G. Sibley in Psychological Science

Supplemental Material

SatherleyLeaderEffects – Supplemental material for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols

Supplemental material, SatherleyLeaderEffects for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols by Nicole Satherley, Kumar Yogeeswaran, Danny Osborne and Chris G. Sibley in Psychological Science

Supplemental Material

SatherleyOpenPracticesDisclosure – Supplemental material for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols

Supplemental material, SatherleyOpenPracticesDisclosure for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols by Nicole Satherley, Kumar Yogeeswaran, Danny Osborne and Chris G. Sibley in Psychological Science

Supplemental Material

SatherleyPoliticalContext – Supplemental material for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols

Supplemental material, SatherleyPoliticalContext for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols by Nicole Satherley, Kumar Yogeeswaran, Danny Osborne and Chris G. Sibley in Psychological Science

Supplemental Material

SatherleyReferences – Supplemental material for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols

Supplemental material, SatherleyReferences for If They Say “Yes,” We Say “No”: Partisan Cues Increase Polarization Over National Symbols by Nicole Satherley, Kumar Yogeeswaran, Danny Osborne and Chris G. Sibley in Psychological Science

Footnotes

Acknowledgements

This research forms part of the PhD thesis of N. Satherley

Action Editor

Ayse K. Uskul served as action editor for this article.

Author Contributions

All the authors conceived and designed the study. C. G. Sibley and D. Osborne acquired the data, and all the authors analyzed and interpreted the data. N. Satherley drafted the manuscript. All the authors provided critical revisions and approved the final manuscript for submission.

ORCID iD

Nicole Satherley

Declaration of Conflicting Interests

The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Funding

Collection of the New Zealand Attitudes and Values Study data analyzed in this article was supported by a grant from the Templeton Religion Trust (No. TRT0196) awarded to C. G. Sibley and a University of Auckland faculty research grant (No. 3709123) awarded to D. Osborne. N. Satherley was supported by a University of Auckland doctoral scholarship.

Supplemental Material

Additional supporting information can be found at

Open Practices

The data described in this article are part of the New Zealand Attitudes and Values Study (NZAVS). Full copies of the NZAVS data files are held by all members of the NZAVS management team and advisory board. A deidentified data set containing the variables analyzed in this article is available on request from the corresponding author or any member of the NZAVS advisory board for the purposes of replication or checking of any published study using NZAVS data. The Mplus syntax used to test all models reported in this article is available on the NZAVS website (https://www.nzavs.auckland.ac.nz). All materials have been made publicly available via the Open Science Framework and can be accessed at osf.io/75snb. The design and analysis plans for this study were not preregistered. The complete Open Practices Disclosure for this article can be found at http://journals.sagepub.com/doi/suppl/10.1177/0956797618805420. This article has received the badge for Open Materials. More information about the Open Practices badges can be found at .

Notes

References

Bolsen

Druckman

J. N.

Cook

F. L.

(2014). The influence of partisan motivated reasoning on public opinion. Political Behavior, 36, 235–262. doi:10.1007/s11109-013-9238-0

Brader

Tucker

J. A.

(2012). Following the party’s lead: Party cues, policy opinion, and the power of partisanship in three multiparty systems. Comparative Politics, 44, 403–420. doi:10.5129/001041512801283004

Bullock

J. G

. (2011). Elite influence on public opinion in an informed electorate. American Political Science Review, 105, 496–515. doi:10.1017/S0003055411000165

Campbell

Converse

P. E.

Miller

Stokes

(1960). The American voter. Chicago, IL: John Wiley & Sons.

Carsey

T. M.

Layman

G. C.

(2006). Changing sides or changing minds? Party identification and policy preferences in the American electorate. American Journal of Political Science, 50, 464–477. doi:10.1111/j.1540-5907.2006.00196.x

Chong

Druckman

J. N.

(2007). Framing theory. Annual Review of Political Science, 10, 103–126. doi:10.1146/annurev.polisci.10.072805.103054

Ciuk

D. J.

Yost

B. A.

(2016). The effects of issue salience, elite influence, and policy content on public opinion. Political Communication, 33, 328–345. doi:10.1080/10584609.2015.1017629

Cohen

G. L

. (2003). Party over policy: The dominating impact of group influence on political beliefs. Journal of Personality and Social Psychology, 85, 808–822. doi:10.1037/0022-3514.85.5.808

Druckman

J. N.

Peterson

Slothuus

(2013). How elite partisan polarization affects public opinion formation. American Political Science Review, 107, 57–79. doi:10.1017/S0003055412000500

10.

Electoral Commission New Zealand. (2016). Second referendum on the New Zealand flag final result. Retrieved from http://www.electionresults.govt.nz/2016_flag_referendum2/

11.

Goren

Federico

C. M.

Kittilson

M. C.

(2009). Source cues, partisan identities, and political value expression. American Journal of Political Science, 53, 805–820. doi:10.1111/j.1540-5907.2009.00402.x

12.

Greene

. (2004). Social identity theory and party identification. Social Science Quarterly, 85, 136–153. doi:10.1111/j.0038-4941.2004.08501010.x

13.

Hardin

C. D.

Higgins

E. T.

(1996). Shared reality: How social verification makes the subjective objective. In Higgins

E. T.

Sorrentino

R. M.

(Eds.), Handbook of motivation and cognition: The interpersonal context (Vol. 3, pp. 28–84). New York, NY: Guilford Press.

14.

Huckfeldt

Mondak

J. J.

Craw

Mendez

J. M.

(2005). Making sense of candidates: Partisanship, ideology, and issues as guides to judgment. Cognitive Brain Research, 23, 11–23. doi:10.1016/j.cogbrainres.2005.01.011

15.

Iyengar

Kinder

D. R.

(1987). News that matters: Television and American opinion. Chicago, IL: The University of Chicago Press.

16.

Iyengar

Sood

Lelkes

(2012). Affect, not ideology: A social identity perspective on polarization. Public Opinion Quarterly, 76, 405–431. doi:10.1093/poq/nfs038

17.

Jones

. (2015, July 29). Second flag referendum should be scrapped if voter apathy continues - Labour. The New Zealand Herald. Retrieved from http://www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=11488745

18.

Jost

J. T.

Ledgerwood

Hardin

C. D.

(2008). Shared reality, system justification, and the relational basis of ideological beliefs. Social & Personality Psychology Compass, 2, 171–186. doi:10.1111/j.1751-9004.2007.00056.x

19.

Kam

C. D

. (2005). Who toes the party line? Cues, values, and individual differences. Political Behavior, 27, 163–182. doi:10.1007/s11109-005-1764-y

20.

Key

. (2015, August 22). John Key: Why my vote will go to flag with the silver fern. The New Zealand Herald. Retrieved from http://www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=11500935

21.

Kinder

D. R

. (2007). Curmudgeonly advice. Journal of Communication, 57, 155–162. doi:10.1111/j.1460-2466.2006.00335.x

22.

Leduc

. (2002). Opinion change and voting behaviour in referendums. European Journal of Political Research, 41, 711–732. doi:10.1111/1475-6765.00027

23.

Leeper

T. J.

Slothuus

(2014). Political parties, motivated reasoning, and public opinion formation. Political Psychology, 35, 129–156. doi:10.1111/pops.12164

24.

Little

. (2016, March 24). Andrew Little: John Key’s flag has cost New Zealand $26m. The New Zealand Herald. Retrieved from http://www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=11611452

25.

Moir

. (2015, August 20). John Key lists his reasons for a new flag. Stuff. Retrieved from https://www.stuff.co.nz/national/politics/71285531/john-key-lists-his-reasons-for-a-new-flag

26.

Muthén

Asparouhov

(2011). LTA in Mplus: Transition probabilities influenced by covariates (Mplus Web Notes: No. 13). Retrieved from http://www.statmodel.com/examples/LTAwebnote.pdf

27.

Nicholson

S. P

. (2011). Polarizing cues. American Journal of Political Science, 56, 52–66. doi:10.1111/j.1540-5907.2011.00541.x

28.

Sibley

C. G

. (2014). Sampling procedure and sample details for the New Zealand Attitudes and Values Study (NZAVS Technical Document e01). Retrieved from https://cdn.auckland.ac.nz/assets/psych/about/our-research/nzavs/NZAVSTechnicalDocuments/NZAVS-Technical-Documents-e01-Sampling-Procedure-and-Sample-Details.pdf

29.

Sibley

C. G.

Robertson

Osborne

Huang

Milojev

Greaves

L. M.

. . . Barlow

F. K.

(2017). Bias and tracking accuracy in voting projections using the New Zealand Attitudes and Values Study. Political Science, 69, 16–34. doi:10.1080/00323187.2017.1321589

30.

Slothuus

. (2010). When can political parties lead public opinion? Evidence from a natural experiment. Political Communication, 27, 158–177. doi:10.1080/10584601003709381

31.

Slothuus

De Vreese

C. H.

(2010). Political parties, motivated reasoning, and issue framing effects. The Journal of Politics, 72, 630–645. doi:10.1017/S002238161000006X

32.

Sniderman

P. M.

Theriault

S. M.

(2004). The structure of political argument and the logic of issue framing. In Saris

W. E.

Sniderman

P. M.

(Eds.), Studies in public opinion: Attitudes, nonattitudes, measurement error, and change (pp. 133–165). Princeton, NJ: Princeton University Press.

33.

Trevett

. (2015, September 16). John Key and Andrew Little’s back and forth on Red Peak. The New Zealand Herald. Retrieved from http://www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=11514059

34.

UMR Research. (2015, September). Flag panel not far away from public choice. Retrieved from https://web.archive.org/web/20180130172004/http://www.umr.co.nz/updates/flag-panel-not-far-away-public-choice

35.

Zaller

. (1992). The nature and origins of mass opinion. New York, NY: Cambridge University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.18 MB

0.41 MB

0.58 MB

0.26 MB

0.15 MB