The national divide: A social representations approach to US political identity

Extending research on US ideological identity as a social identity, this study employs a social representations approach to capture identity meaning as a form of national attachment. Across two studies ( n = 723), we demonstrate that two novel organizing principles of US ideological identity— national reverence (veneration of the nation in the abstract, its symbols, sacred texts, and founding fathers) and individual support (a focus on the progress of individuals and the change required to bring about equality)—significantly predicted ideological self-categorization, 2016 presidential voting, and affective political polarization over and above right-wing authoritarianism and social dominance orientation. The results suggest an integration of national and conservative identities that places US national identity in opposition to liberal identity and to progress and equality, pointing to the divisive employment of national identity in the current political environment.


INTRODUCTION
The importance of citizens' ideological and partisan selfcategorizations-the extent to which they identify as a Democrat, Republican, conservative, or liberal-to political behaviour is well established (e.g., Campbell, Converse, Miller & Stokes, 1960;Green et al., 2002;Iyengar et al., 2012;Mason, 2015). Central to understanding this link between a person's political identity and political behaviour is the meaning an individual attributes to that identity: 'it is the meaning of [a political identity], not its existence, that determines its political consequences' (Huddy, 2001, p. 130). Meaning may include-but is not bound by-demographics or policy preference alone; it reflects any number of the subjective values, behaviours, and norms that an individual associates with their self-categorization.
In the current study, we offer a novel perspective on political identity meaning by exploring national attachment-anchored representations. These representations were identified as central to US conservative and liberal identity meaning based on a qualitative analysis of interviews on the subject. Specifically, we operationalize two contrasting national attachment narratives theorized to underlie lay representations of US ideological identity; we demonstrate their consequences for two key political behaviours linked to political identity: affective political polarization and voting behaviour; and we investigate their association with right-wing authoritarianism (RWA) and social dominance orientation (SDO). This exploratory work suggests that these contested elements of American national identity capture a substantial degree of political identity meaning and may, therefore, have consequences for American political behaviour. In addition to the identification of substantive measures of US ideological identity, this study contributes to the current literature on US political identity theoretically and methodologically through the application of a social representations approach.

A social representations approach
As an example of research that is dependent upon political identity meaning, the phenomenon of affective political polarization has attracted a good deal of recent research attention. A prominent theory of this animosity towards the political outgroup is based on the social identity approach (Tajfel, 1974;Turner et al., 1987) and suggests that affective political polarization is driven by citizens' ideological and partisan self-categorizations (Huddy et al., 2015;Iyengar et al., 2012;Mason, 2015Mason, , 2018aMason, , 2018b. In the application of the social identity approach to affective political polarization, the focus has primarily been on the 'natural' ingroup favouritism and outgroup derogation that can be a consequence of identifying with a particular group (Iyengar et al., 2012;Mason, 2018b). This psychological dynamic was considered by Tajfel to be only a starting point because it fails to recognize the contextual features-including meanings-that mediate between social categorization and behaviour (Duveen, 2001;Huddy, 2001Huddy, , 2002Reicher, 2004;Turner, 1999). Without meaning, the conceptualization of the social categories of party, ideology, race, and religion are reified and static, and the current theoretical conceptualization of affective political polarization as driven by social identity can therefore envision only further polarization.
By including meaning in the analysis of identity, we acknowledge that ideological identity-like any social identity-is not an exclusively individual or social phenomenon. It occurs at the space in-between the individual and society, drawing upon both our individual differences and our social influences (Pehrson & Reicher, 2014). To investigate the meaning associated with a political identity then, we drew on the system of knowledge (common sense meanings) that also exists in this space, located in everyday communications. By doing so, we gain insight into the language that justifies outgroup antipathy. These common-sense meanings that people draw upon in order to make sense of the world around them and to act towards it in meaningful ways are referred to as social representations, and were initially outlined by Moscovici in 1961(Sammut et al., 2015. A social representations approach (Elcheroth et al., 2011) holds that the meaning of a social identity is not a given, instead it is 'represented' as an understanding between people.
Social representations are psychological organizations, but are specific to one's social environment (Moscovici, 1961(Moscovici, /2008. Like social identities, they are situated 'at the crossroad between the individual and society' , a space in-between (Bauer & Gaskell, 1999 , p. 167). A social representations approach is therefore in marked contrast to previous work on ideological identity content in which social cognitive approaches focus on the attitudes and beliefs of the individual. Within this body of work, it is widely accepted that US ideological identity is anchored in the contrast between the support versus rejection of inequality and the support versus rejection of the status quo, often measured using the RWA (Altemeyer, 1996) and SDO (Pratto et al., 1994) scales, respectively (Jost et al., 2003). A social representations approach holds that political lay thinking transforms these ideological attitudes into shared representations (Staerklé, 2009 objectification produces shared symbols that incorporate the original attitudes, so that they can be more easily used in everyday communication. These symbolic tools allow group members to make sense of their social world and their relationship to other groups.
Situated in language, it is through shared representations that meaning is elaborated and that social groups-such as political parties and ideologies-establish their identities, differentiating themselves from other groups within society. A social representations approach is a framework that provides a means by which to describe everyday communications and offers insight into the reciprocal process of identity construction. While identity itself is a social representation (Andreouli & Chryssochoou, 2015;Elcheroth et al., 2011), social representations are the building blocks of the structure and content of identity (Breakwell, 1993(Breakwell, , 2001. In this way, social representations may serve to confine thought and actions by determining 'the field of possible communications, of the values and ideas present in the visions shared by groups' (Moscovici, 1961(Moscovici, /2008. By serving to perpetuate and maintain particular meanings of identity, certain social representations may be driving increasing polarization. However, by framing these elements within the social representations approach, the theoretical framework conceptualizes these representations as dynamic elements that, while undergoing negotiation between the individual and society, have the potential to change.
Identifying social representations of US ideological identity is important not only because of their role in ideological identity meaning, but also because they serve as the lens through which unfamiliar information is accommodated, oranchored. Through understanding the social representations of US ideological identity, we gain insight into the language and symbols through which understandings of new information related to things such as political issues, events, or candidates, are likely to be structured. And in the case of groups in conflict such as US conservatives and liberals, understand how these representations may be positioned in opposition to each other.
Organizing principles of a social representation refer to its core structure, focusing on the systematic variation groups give to different dimensions within a field of representations (Doise et al., 1993). In the case of the social representation of US ideological identity, the organizing principles are the content dimensions towards which those who self-categorize as a conservative or liberal take a stance (Clemence et al., 2014). It is with regard to these organizing principles that ideological identity is given meaning and through which individuals may express preferences for a certain social order such as RWA or social dominance. In this study, we therefore aim to identify the organizing principles of US ideological identity, arguing that these principles anchor the meaning of ideological identity. In doing so, this approach offers a novel perspective on the problem of affective political polarization as well as other political behaviours.

The proposed principles
In the interest of capturing the interpersonal nature of social representations, we take as our starting point a qualitative analysis of interviews with US conservatives and liberals regarding the content of US ideological identities . The open-ended nature of interviews allows access to the potential universe of representations. Amongst other themes, this work identified differing national attachment narratives in the content of American liberal and conservative identities. The researchers characterized liberal participants' talk as centred on 'Individuals and Issues' (p. 387): individual characteristics and selected issues that aimed to secure equal rights for more citizens. This group tended to distance their personal identities from the liberal and Democratic groups. In contrast, conservative participants' discourse revolved around 'Ideology and the Nation' (p. 390); they were more apt to embrace their ideological label and spoke in ideological terms. They positioned themselves as defenders of American values and symbols, closely linking their personal, political, and national identities.
The first key distinction made in this work is the left's focus on issues contrasted with the right's more ideological talk. These observations are consistent with the longstanding but recently revived observation that Democrats represent their position in terms of issues while Republicans represent theirs in terms of ideology (Converse, 1964;Grossmann & Hopkins, 2015). The second distinction is the readiness with which the right embraces their political ideological identity in contrast to the left. This well-substantiated outcome is often attributed to negative associations with the liberal identity label (Ellis & Stimson, 2012), but moral foundations theory (Graham et al., 2009) also points to a divide between group-and individual-centric reasoning. Conservatives' morals of authority, purity, and loyalty 'in addition to' liberals' primary morals of fairness and care are described as reflecting group enhancing versus individuating motivations, respectively.
The third distinction is that of a differing relationship with-and perhaps conceptualization of-the nation on the right and left. Where the left appeared to conceptualize the nation as an aggregation of individuals, the right conceived the nation as a symbolic community. Although little work has addressed the differing relationship with the nation between the American left and right, this contrast can be gleaned from earlier works on patriotism and nationalism-concepts that seek to conceptualize differing relationships with the nation. Studies employing these concepts have found that the right is more drawn to symbolic representations of the nation (Kosterman & Feshbach, 1989;Schatz et al., 1999;Sullivan et al., 1992), while the left values the right to criticize and to protest government, seeking progress in individual rights (Huddy & Khatib, 2007;Sullivan et al., 1992). Indeed, the right and Although the three distinctions can be seen to reinforce and support one another through their group-versus individual-centricity, in this study we focus primarily on this third distinction. In particular, we argue that representations of US ideological identity are currently organized by the distinct and contested principles of national reverence (NR) and individual support (IS) ('the principles'), representing national attachments on the right and left, respectively. These proposed principles conceptualize the nation as symbolic community whose common myths, ideologies, and symbols are revered (NR) on one side; and as a collection of individuals whose equal right to thrive is a priority (IS) on the other. Measurement items selected to comprise the measures of NR and IS can be found in Table 2. These item were constructed from previous research that spoke to facets of the hypothesized principles Kosterman & Feshbach, 1989;Schatz et al., 1999;Sullivan et al., 1992), and were refined on the basis of pilot research that assessed construct validity and internal consistency. As social identities are constructed in reference to both the acceptance of ingroup identity content and the resistance to outgroup identity content, both measures are important in understanding the structure of US ideology and the relationship between the identities.
The constructs juxtapose the group and the individual. NR reflects an affective attachment to the nation and its tangible and intangible symbols, including the flag, the anthem, the founding documents, the 'American system' , and the founding fathers. The IS principle attempts to capture the primacy of the individual that is represented by promoting issues that allow for progress towards self-realization through equality. The IS narrative is less about affect towards the nation, than it is about what the nation owes to its people. NR is clearly connected to the traditional idea of American patriotism, thereby potentially imbuing American national identity with conservative identity content.

The current study
Extending research on US ideological identity as a contributor to political behaviour, this study employs a social representations approach to capture identity meaning. We gain insight into the drivers behind political behaviour by exploring the extent to which NR and IS, as ordinary people's view of their identities, act as organizing principles of US ideological identity representations. Across two studies, we operationalize the two organizing principles and assess their reliability and validity; we also demonstrate that positions towards NR and IS serve as significant descriptors of these identities, and that they significantly predict voting behaviour and animosity towards the outgroup. As lay representations, NR and IS may suggest the means by which individually located conceptualizations-such as the desire for a particular societal structure reflected in RWA and SDO measures-are expressed in everyday language. Therefore, in Study 2 we specify the principles by comparing their predictive ability to that of RWA and SDO, demonstrating the predictive power of the principles over and above these measures. The implications of these findings, including opportunities for application and future research, are discussed in the final section.

STUDY 1
Study 1 explores the proposal that, in the current US context, NR and IS are conceptually distinct factors that describe lay persons' understandings of US liberal and conservative identity content. We first determined that the items comprising each principle were empirically distinct (using an exploratory factor analysis) and were internally reliable.
We then assessed the validity of the resultant factors, finding that the principles were significantly and substantively associated with liberal and conservative self-categorizations. Finally, we employed the identified principles to predict political behaviours, finding them to be significant and substantive predictors of affective political polarization and voting behaviour.

Participants
We collected the survey data for the main study on-line using Qualtrics; participants were recruited through the on-line participation platform, Prolific https://prolific.ac. MacCallum et al. (1999) suggest that the necessary sample size in common factor analysis depends on several aspects of the study, including the level of communality of the variables (the proportion of the variance of the variable that is accounted for by the common factors) and the level of overdetermination of the factors (the degree to which the factor is clearly represented by a sufficient number of variables). Highly overdetermined factors are considered to be factors that have high loadings on at least three to four variables and exhibit good simple structure. We also considered that, for wide com-munalities, sample size has little difference for samples with at least five variables per factor and 20 participants per variable (Hogarty et al., 2005;Osborne et al., 2008). In addition, a ratio of 20 variables to three factors produces no increase in accuracy in sample sizes over 200, even with low communality (MacCallum et al., 1999). Without knowing the commonalities until data were collected, we estimated a wide communality, and given our 8:1 and 10:1 variable-to-factor ratios for the original principles, it was reasonable to project five variables per factor loading. We therefore selected a conservative initial sample size of 360 (20 participants × 18 variables).
Because we were interested in describing the social representations of ideological identity, we have restricted our participants to those who (as registered with the recruiting platform) indicated that they considered themselves to be either liberal or conservative. To ensure the authenticity of this filter, the question as to participants' ideological identification was asked again within the survey. Any participants who indicated an ideological identity that did not match their profile were removed and replaced (a total of one 'liberal' and 10 'conservative' participants

Measures
National reverence (NR) and individual support (IS). A survey of 29 statements (see Online Appendix) was newly constructed to reflect the theoretical organizing principles of ideological identity representations. These items were subjected to construct validity testing whereby five individuals (PhD students and a staff) were asked to match the statements with the principles' definitions. These statements also provided the basis for a survey pilot of 50 participants who were asked to comment on the wording of the questions and the questionnaire's overall ease of use. Based on the construct validity feedback, the pilot participants' feedback, and analysis of item internal consistency, the survey items were pared down and wording was amended to more appropriately reflect the principles under enquiry.
In particular, questions intended to capture the perspective that political positions stem from the individual (e.g., 'my political positions are a reflection of my personal character') proved inconsistent and were eliminated from the IS measure to increase internal reliability.
Eighteen items (eight IS and ten NR items) were retained to form the survey for the main study (refer Table 2).

RESULTS
Unless otherwise noted, all analyses were conducted in JASP (JASP Team, 2019).
The national reverence and individual support items were submitted to an unconstrained exploratory factor analysis. Parallel analysis with Promax rotation indicated a two-factor model. Because we anticipated that the factors would be correlated, an oblique rotation was selected. To confirm this extraction, an exploratory factor analysis in SPSS using maximum likelihood estimation produced two factors with eigenvalues greater than 1.00 (9.3 and 2.4). A scree-plot under both analyses clearly suggested a two-factor model. We retained items with loadings of at least .32 on the specified factor, no cross-loading (Osborne et al., 2008), and a difference in factor loading across the group and individual perspective factors of at least .3. According to these criteria, all 10 of the NR and five of the original eight IS items were retained, refining IS to primarily represent support for progress (change), equality, and action ('doing more') for citizens, broadly reflecting the egalitarianism on which liberal ideology is based.

As shown in
As expected, the two factors were highly correlated (r = −.58).

Construct validity: Ideological identity
To assess construct validity, we calculated relationships between the principles and participants' ideological self-categorizations. The extent to which the identified perspectives may provide description of political self-categorization was assessed through a linear OLS regression that included the two principles along with controls for operational ideology, race (white = 1), age, income, and education. Pearson correlations and OLS regression estimates for these identities are in Table 3.
Both the NR and IS principles were highly and significantly correlated with ideological identity, though NR was the stronger of the two. IS was highly correlated with the more liberal (r = −.63) selfcategorizations, while NR was highly correlated with the more conservative (r = .80) self-categorizations. The principles also significantly predict ideological identity (adjusted R 2 = .70, p < .001), with NR contributing more than twice the predictive value of IS. The principles are therefore considered to be useful descriptors of ideological identity content. Identity measures scaled such that more conservative is more positive. *p < .05, **p < .01, ***p < .001.

TA B L E 3 Regression of principles on ideological identities
The second model demonstrates that the principles contributed descriptive value over and above operational identity and demographics. NR was the most significant predictor of ideological identification (β = −0.50, p < .001), followed by operational identity (β = .30, p < .001), and IS (β = −.19, p < .001). Income was the only other significant, though minor, predictor of ideological identity (β = .07, p = .016).
An analysis of this model within ideological groups is set out in Table 4. The results affirm the significant descriptive value of NR and IS identified in the first analysis within both conservative (adjusted R 2 = .23) and liberal (adjusted R 2 = .23) identities. This analysis also indicates two primary ideological asymmetries. In the first instance, the relative predictive values of the principles differ. For conservatives, the predictive value of NR (β = .44, p < .001) was substantially greater than IS (β = −.16, p = .015), while the predictive values for liberals were much more similar (β NR = −.30, p < .01, β IS = .33, p < .001). In addition, the relative predictive value of operational identity and principles varies substantially between the ideological groups. For conservatives, NR was the largest contributor (β = .47, p < .001) and operational identity was non-significant (β = −.10, p = .07). In contrast, for liberals, operational identity was the largest (β = −.29, p < .001) and is similar in predictive value to IS (β = .24 p < .001) and NR (β = −.22 p < .01).
The prominent role of issues in the liberal identity is consistent with previous literature (Grossmann & Hopkins, 2016; and captures meaning outside of the proposed principles. This analysis also supports the assertion that the operational and symbolic elements of identity between liberals and conservatives may not be symmetric (Ellis & Stimson, 2012).

Concurrent validity: Political behaviour
Affective polarization. For the sample as a whole and then by group, the correlation and regression coefficients for the principles and feelings towards conservatives and liberals are set out in Table 5.
For the whole sample, both principles were significantly correlated with and were significant predictors of, feelings towards conservatives and liberals. NR was significantly and strongly correlated with feelings towards both ideological groups; the principle was positively correlated with feelings towards conservatives (r = .78, p < .001) and negatively correlated with feelings towards liberals (r = −.68, p < .001).
Similarly, NR was a better predictor of these thermometer ratings than was IS. On a full sample basis, the two principles significantly predict feelings towards conservatives and liberals (adjusted R 2 = .64, p < .001; adjusted R 2 = .52, p < .001; respectively), with both principles contributing to the equations in partial opposition. The principles are therefore considered significant descriptors of affective political polarization.

STUDY 2
Study 1 established the descriptive and predictive value of NR and IS with a sample of conservative and liberal identifiers. In Study 2, we aim to: confirm the factor structure of the principles, assess the general predictive utility of the measures by replicating the findings on political behaviour from Study 1 in an unrestricted sample, and demonstrate that the principles have predictive validity beyond typically used measures of ideological identity such as RWA and SDO (e.g., as used in the widely cited study by Jost et al. [2003]

Participants
As in Study 1, participants were recruited through Prolific. For the confirmatory factor analysis, we sought to match the exploratory factor analysis sample in Study 1 and therefore aimed to recruit 362 participants (363 participants were recruited). Unlike Study 1, the participant group was not restricted to only self-identified liberals and conservatives; recruitment was instead opened to include those of any ideological identification. This restriction was lifted to more accurately reflect the American voting population, of which (according to the 2016 ANES) approximately 40% identify as neither liberal nor conservative. The sample was, however, filtered to include only those whose pre-screening information indicated that they had voted in the 2016 presidential election. Ideally, a 'registered voters' screen (to simulate the ANES) would have been employed, but such a filter was not available on Prolific at the time of the study. Restricting our sample to previous voters results in a more constrained population than is reflected in the ANES, and by definition, more politically active one. However, the 'previous voter' filter is considered appropriate to eliminate unregistered voters from the sample.
The sample comprised 131 (36%) participants who indicated that they identified as 'liberal' or 'strong liberal' on the US ideological spectrum, 157 (43%) who marked themselves as 'moderate' , 'lean liberal' or 'lean conservative' , and 75 (21%) who identified as a 'conservative' or 'strong conservative' . One hundred and ninety participants (52%) voted for Hillary Clinton, 108 (30%) for Donald Trump, and 65 (18%) for a third party. Forty nine percent of the sample were men and 52% were women, average age was 35.11 (SD = 11.60). The demographic breakdown of the participants, along with benchmark measures, can be found in Table 1.

Measures
The measures of NR and IS, as well as measures of ideological identity, feeling thermometers, operational identity, and demographic information were the same as in Study 1. In addition, seven-point Likert scales were included in the current study to measure (R indicates that items was reverse scored): Right-wing authoritarianism (ω = .80, α = .80). A 6-item scale measured from strongly agree to strongly disagree (Bizumic & Duckitt, 2018).
Items included: 'it's great that many young people today are prepared to defy authority' (R), 'what our country needs most is discipline, with everyone following our leaders in unity' , 'God's laws about abortion, pornography, and marriage must be strictly followed before it is too late' , 'there is nothing wrong with premarital sexual intercourse' (R), 'our society does NOT need tougher government and stricter laws' (R), 'the facts on crime and the recent public disorders show we have to crack down harder on troublemakers, if we are going preserve law and order.' Social dominance orientation (ω = .91, α = .91). The SDO7 scale (Ho et al., 2015) consists of 6-items measured from very positive to very negative. Items included: 'Some groups of people are simply not the equals of others' , 'some people are just more worthy than others' , 'to get ahead in life, it is sometimes necessary to step on others' , 'Increased equality' (R), 'if people were treated more equally, we would have fewer problems in this country' (R), 'it is important that we treat other countries as equals' (R).

Confirmatory factor analysis
We performed confirmatory factor analysis to affirm the two-factor structure of ideological identity identified in the exploratory analysis in Study 1. The fit of a one-factor structure to the data, which would represent unidimensional continuum from left to right, was compared to the fit of a two-factor structure. The item loadings and primary fit indicators are shown in Table 7.
In columns two and four, the models were modified to allow the covariance of selected residual errors. This modification is to recognize that the factor has indicator variables that share components, a practice that is appropriate to the extent that it is theoretically supported and within a factor (Landis et al., 2009). In keeping with these criteria, the correlations were limited to those items within an indicator variable that were aimed at measuring similar attitudes. For example, residuals related to the items 'honouring the flag says a lot about who I am' and 'the flag and anthem should be respected as the sacred symbols they are' were allowed to correlate. In columns 2 and 4, we allow for correlated measurement error within NR between these 'flag' items (items 10 and 13), between 'self-reliance' items 8 and 15, between 'American system' items 7 and 11, and between 'founding' items 9, 12, and 14, and within IS between 'improving equality' items 4 and 5. The

Replication of study 1: Predicting outcome variables
We first assessed whether the outcome variables employed in Study 1 were significantly predicted by the principles using Study 2′s broader sample. Results of these regression analyses are in Table 8.

3.2.4
Incremental predictive ability An exploratory factor analysis found a four-factor solution when loading NR and IS along with RWA and SDO. All items loaded on their respective constructs with the exception of Item 4 on the RWA scale which loaded on the IS factor. There were also two cross-loadings between SDO and IS (SDO item 8 and IS item 4). NR items loaded on this construct without exception. A confirmatory factor analysis found a four-factor model (X 2 /df = 4.22, CFI = .84, RMSEA = .09, SRMR = .07) to be superior to a two-factor model that loaded NR and RWA on one factor and SDO and IS on the other (X 2 /df = 5.18, CFI = .78, RMSEA = .11, SRMR = .08). The four constructs are therefore considered to be distinct.
To ascertain whether NR and IS have predictive validity beyond the primary existing measures of ideological identity we ran the same regressions with RWA and SDO as predictor variables in Table 8.
RWA and SDO together significantly predicted all of the outcome variables under analysis, with the combined effect sizes ranging from β RWA = −.36, β SDO = −.35, R 2 = .32, p < .001 for the liberal feeling thermometer to β RWA = .56, β SDO = .28, R 2 = .55, p < .001 for ideological identity. A comparative analysis of effect sizes is set out in Table 9.
Comparing the effect sizes for regressions that pair NR with IS and RWA with SDO (both on their own and as a contribution to RWA and SDO), the combination of NR and IS provided a greater level of explanation for these dependent variables, particularly for ideological identity (ΔR 2 = .13, p < .001) and voting (Nagelkerke's ΔR 2 = .17). The significant incremental predictive value over and above RWA and SDO attests to the utility of these principles, and supports the caution issued by social identity theorists against essentializing political psychology motives without examining the cultural and social context on which these psychological processes are dependent (Reicher, 2004).
In addition, the significant overlap of the combined measures suggests that NR and IS may capture, in the measurement items, the means by which people who score high or low on RWA and SDO understand and express their political identities and the nature of the language used to enforce and promote these attitudes and mobilize the electorate.

DISCUSSION
Drawing on a social representations approach to explore the meaning of US ideological identity, we constructed two measures to reflect .48 .17 .63 .02 Note: the larger incremental predictive contribution is indicated in bold.
the organizing principles of its social representation (NR and IS), and found them to be valid and reliable. NR reflects a regard for the nation as symbolic community whose common myths, ideologies, and symbols are revered; while IS captures a perspective that the nation is a collection of individuals whose equal right to thrive is a priority. The principles were employed by both groups-alternately strongly supported by one and resisted by the other-in predicting ideological selfcategorizations and the political behaviours of 2016 presidential voting and affective political polarization. The principles were therefore recognized by both sides as being content dimensions towards which they take a stance and appear to be important discourses in mobilizing political support. In addition, the predictive contribution of these measures exceeded those of RWA and SDO for the outcomes under consideration. This work supports and extends previous literature that asserts that Democrats and Republicans are different kinds of parties (Grossmann & Hopkins, 2016), that conservatives tend towards binding and liberals towards individuating morals (Graham et al., 2009) Because national identity is one of the most accessible and powerful social identities in the influence of political behaviour (Reicher & Hopkins, 2001;Sullivan et al., 1992), an alignment of conservative and national identity representations is potent. It reinforces conservative identity and gives conservatives the power to shape representations of national identity, casting the political left as outsiders. From an ideological identity perspective, an alignment with national identity imbues the conservative identity with the moral project of preserving the nation, a highly motivating and a putatively selfless objective.
The sacred nature of NR items demonstrate the quasi-religious regard in which the nation is held for those scoring high on this measure.
These sacred convictions not only galvanize the ingroup, but can also be used to cast doubt on the moral standing of the outgroup (Marietta, 2008).
Inherent in the power to shape national identity representations is the power to influence who is perceived as the outgroup. The corollary of a national identity anchored in conservative identity is a liberal identity in conflict with national identity. The consequence is that both the liberal identity and the related representations of equality and progress captured in IS are positioned as being unpatriotic. The left, with a different conception of patriotism Huddy & Khatib, 2007;Yougov, 2018), and lacking a similar influence over national identity, remains to defend themselves against a conservative definition of patriotism. Casting the opposition as unpatriotic has consequences for the prospects for conflict resolution. Common identities such as national identity can foster cooperation, and conflict resolution has long sought common ground in common identity; but if the representations related to the common identity becomes divisive, the power of national identity to ameliorate conflict is diminished.
Although our work suggests that RWA and SDO attitudes are to a large extent captured by the NR and IS organizing principles-suggesting that support for inequality and status quo is to a large degree captured in the right's reverence of the nation-conservatives avoid a direct response to the liberal ideals of progress and equality by constructing an ideological identity that is positioned as a defender of the nation.
Consequences are not limited to the ideological group level, however: this alignment also provides the political right with the power to propagate a conservative worldview through shaping representations of American national identity. Because social representations are not elaborated in isolation but in dialogue with other social representations, conservatives' representations of themselves as the 'true Americans'  impacts the social representation of American national identity. Citizens for whom national identity is important-regardless of political engagement-receive cues that being conservative is being a good American and will strive to act in accordance with what they believe are the prevailing norms of national identity (Hogg, 2006;Theiss-Morse, 2009), and in line with what they believe others believe (Elcheroth et al., 2011).
Viewed as a social representation that is subject to the dynamic process of anchoring, the current politically polarized environment can be reframed beyond a conflict between issues, morals, values, or demographics. As social representations that take into account both individual differences and social influences, our observations of NR and IS as key variables in affective political polarization may be useful in developing causal models of political polarization. Additional work may look at the relationship between particular items associated with patriotism, equality, ideological identification, and political polarization over the years to identify drivers of polarization. Further research is necessary to not only understand the development of this phenomenon, but also to begin to create more effective means of communication with those for whom national reverence is particularly high or low. And finally, this work suggests that, although liberals' social representation of their ideological identity is in line with academic measures, conservatives represent their identity differently. It is only through understanding these representations that we may become cognizant of the process by which they are formed, changed, maintained and employed by entrepreneurs of identity for political mobilization.
More work is required to fully understand the components of the principles, their interaction with and relationship to other measures of identity, and how the messaging identified in these principles may influence political decision-making. Our results are limited to the extent that our sample was not a true representation of the American voting public, but a less ethnically diverse, less affluent, more politically involved, and more educated group. In particular, because the principles assume a level of political understanding (e.g., 'Today's Constitutions says. . . ' and 'Our country's policies. . . '), we may reasonably expect the descriptive power of the principles to be weaker in a less politically engaged sample, perhaps to the extent that the individual difference measures of RWA and SDO exceed the descriptive power of NR and IS. The strength of a social representation approach is that it captures identification influences that are context-dependent; this dependency on context does however also limit the use of these particular measures to the current US environment, although the approach itself may prove useful in other national or international contexts. The concepts of national reverence and individual support, as well as the validation of a social representations approach to ideological identification can be expected to provide researchers with new perspective on political identity processes and outcomes.
The results paint a picture of the nature of the political conflict in the US whereby conservatives appear to have successfully aligned themselves with a particular type of patriotism, leaving the left to respond.
The correlational nature of the analyses does not allow for causality to be inferred for the behaviours predicted in these studies, but given the strength and unique predictive contributions of these principles, the measures may lend themselves to future experimental work. In so far as the alignment of conservative and national identities may contribute towards polarizing behaviour, it is in the interest of long-term depolarization, and perhaps electoral success, for the left to attempt to provide a definition of patriotism in their own image.