The ‘nature of belief systems in mass publics’ has captivated (and confounded) political scientists for decades (for example, Abramowitz and Saunders Reference Abramowitz and Saunders1998, Ansolabehere, Rodden and Snyder Reference Ansolabehere, Jonathan and Snyder2008; Bobbio Reference Bobbio1997; Conover and Feldman Reference Conover and Feldman1981; Converse Reference Converse and Apter D1964; Ellis and Stimson Reference Ellis and Stimson2012; Lane Reference Lane1962; Levendusky Reference Levendusky2009). Within this research tradition, several studies have emphasized the centrality of abstract normative priorities (variously referred to as ‘core values’, ‘moral foundations’ or ‘principles’) to liberalism–conservatism (for example, Feldman Reference Feldman1988; Goren, Federico and Kittilson 2009; Graham, Haidt and Nosek Reference Graham, Haidt and Nosek2009; Jacoby Reference Jacoby2006; Peffley and Hurwitz Reference Peffley and Hurwitz1985; Rokeach Reference Rokeach1973; Schwartz Reference Schwartz1992; Tetlock Reference Tetlock1986; Thompson, Ellis and Wildavsky Reference Thompson, Ellis and Wildavsky1990). However, political scientists have yet to comparably appreciate the distinct role of premises, or abstract descriptive beliefs, as ideological ingredients – despite the fact that for centuries, philosophers have used premises regarding human nature and society to ground their religious, political and economic visions (for example, Augustine ~400; Burke Reference Burke1790; Hobbes Reference Hobbes1651; Jefferson Reference Jefferson1776; Locke 1689; Madison Reference Madison1787; Marx and Engels Reference Marx and Engels1848; Rousseau Reference Rousseau1762; Smith Reference Smith1776).
In this article, we address this gap in scholarship. Drawing insights from Polarity Theory (Tomkins Reference Tomkins and White1964), Cultural Theory (for example, Thompson, Ellis, and Wildavsky Reference Thompson, Ellis and Wildavsky1990), Dual-Process Cognitive Motivational Theory, and Reality Theory (Kruglanski Reference Kruglanski1999), and building upon our own theoretical perspective (Marietta Reference Marietta2011), we examine the extent to which three core premise disputes about human nature and society (human benevolence–selfishness, progress idealism–skepticism, social stability–fragility) may account for the variance in liberalism–conservatism – independently from five essential value conflicts that are known to underlie such ideological differences (collectivism–individualism, humanism–theism, sexual progressivism–traditionalism, pacifism–militarism and nurturance–punitiveness). Studies 1 and 2 use nationally representative data to assess these relationships (with respect to both ideological identification and issue-based liberalism–conservatism), and Study 3 uses a randomized controlled experiment to compare how subjects respond to ‘evidence’ regarding human nature which complements their ideological perspective relative to that which contradicts it. In all three studies the data are consistent with our hypotheses, suggesting that premises are distinct and important elements of political ideology.
Ideology, Values and Premises
One useful definition of political ideology is ‘a fairly coherent and comprehensive set of ideas that explains and evaluates social conditions, helps people understand their place in society, and provides a program for social and political action’ (Ball, Dagger and O'Neill Reference Ball, Dagger and O'Neill2016, 5; see Jost, Federico and Napier Reference Jost, Federico and Napier2009 for a review of definitions). The dominant ideological cleavage in the United States pits liberalism against conservatism (for example, Abramowitz and Saunders Reference Abramowitz and Saunders1998; Ellis and Stimson Reference Ellis and Stimson2012; Jacoby Reference Jacoby1991; Knight Reference Knight1985).
Though ‘liberal’ or ‘conservative’ identifications do not usually constrain the majority of Americans’ issue attitudes across policy dimensions (for example, Converse Reference Converse and Apter D1964; Kinder and Kalmoe Reference Kinder and Kalmoe2017), the labels do correspond to a trove of important psychological traits, motives, needs and cognitive styles (for example, Jost Reference Jost2006; Jost et al. Reference Jost2003). Fittingly, then, for well over 50 years, political psychologists have been trying to understand the structure and substance of liberalism–conservatism. Within that line of inquiry, a dense literature highlights the importance of core values as ideological determinants.
Values
For our purposes, values can be defined simply as abstract normative priorities.Footnote 1 The earliest considerations of core values as they relate to politics tended to treat them as broadly accessible alternatives to ideology (for example, Feldman Reference Feldman1988; Feldman and Steenbergen Reference Feldman and Steenbergen2001; Rokeach Reference Rokeach1973). However, over the past 20 years, the conventional wisdom has come to view values – and value conflicts, in particular – as important elements of ideology, especially as they relate to its three basic dimensions: economic, cultural and security/foreign policy (for example Graham, Haidt, and Nosek Reference Graham, Haidt and Nosek2009; Jacoby Reference Jacoby2006; Jost, Federico, and Napier Reference Jost, Federico and Napier2009; McAdams et al. Reference McAdams2008; Pjurko, Shalom and Davidov Reference Pjurko, Shalom and Davidov2011; Tetlock Reference Tetlock1986).Footnote 2
Specific Value-Ideology Linkages
When it comes to the economic dimension of ideology, the intersecting values of liberalism include egalitarianism (Feldman Reference Feldman1988; Jacoby Reference Jacoby2006; Jost et al. Reference Jost2003; McCloskey and Zaller Reference McCloskey and Zaller1984; Swedlow and Wyckoff Reference Swedlow and Wyckoff2009; Thompson, Ellis and Wildavsky Reference Thompson, Ellis and Wildavsky1990), humanitarianism Footnote 3 (for example, Feldman and Steenbergen Reference Feldman and Steenbergen2001), nurturance (for example, Barker and Tinnick Reference Barker and Tinnick2006; Lakoff Reference Lakoff1996; McAdams et al. Reference McAdams2008), collectivism (Hofstede Reference Hofstede2001) and universalism (for example, Schwartz Reference Schwartz1992), whereas the values of economic conservatism include individualism (for example, Feldman Reference Feldman1988; Thompson, Ellis and Wildavsky Reference Thompson, Ellis and Wildavsky1990; Yoon Reference Yoon2014), limited government (for example, Goren Reference Goren2012), support for free enterprise (for example, Feldman Reference Feldman1988; McCloskey and Zaller Reference McCloskey and Zaller1984; Schwartz, Caprera and Vecchione Reference Schwartz, Caprera and Vecchione2010), proportionality or punitive ‘fairness’ (for example, Haidt Reference Haidt2012) and hierarchism (for example, Schwartz, Caprera and Vecchione Reference Schwartz, Caprera and Vecchione2010; Thompson, Ellis and Wildavsky Reference Thompson, Ellis and Wildavsky1990).Footnote 4
As for cultural ideology, intersecting liberal values include civil libertarianism/tolerance (for example, Jacoby Reference Jacoby2006; McCloskey and Zaller Reference McCloskey and Zaller1984; Sullivan, Piereson and Marcus Reference Sullivan, Piereson and Marcus1981; Swedlow and Wyckoff Reference Swedlow and Wyckoff2009), sexual progressivism (for example, Layman and Green Reference Layman and Green2006) and multiculturalism (for example, Schwartz, Caprera and Vecchione Reference Schwartz, Caprera and Vecchione2010), and conservative values include sexual traditionalism (for example, Hunter Reference Hunter1992; Layman Reference Layman2001; Layman and Green Reference Layman and Green2006), moral traditionalism (for example, Goren Reference Goren2012), sanctity (for example, Haidt Reference Haidt2012), in-group loyalty (for example, Haidt Reference Haidt2012) and order (for example, Swedlow and Wycoff Reference Swedlow and Wyckoff2009).
Finally, with respect to security/foreign policy ideology, the values of liberalism include pacifism (for example, Hurwitz and Peffley Reference Hurwitz and Peffley1987), multiculturalism (again) and humanitarianism (again); the values of conservatism include militarism (Hurwitz and Peffley Reference Hurwitz and Peffley1987; Wittkopf Reference Wittkopf1990), strength (Braithwaite Reference Braithwaite1997; Goren Reference Goren2012), nationalism (Hurwitz and Peffley Reference Hurwitz and Peffley1987), punitiveness (for example, McCleary and Williams Reference McCleary and Williams2009), ethnocentrism (Hurwitz and Peffley Reference Hurwitz and Peffley1987) and tough mindedness.
Linking the three dimensions of liberalism–conservatism together, liberalism is associated with an understanding of morality that emphasizes moral compassion, whereas conservatism emphasizes moral discipline (Barker and Tinnick Reference Barker and Tinnick2006).Footnote 5 These competing understandings of morality correspond to conflicting interpretations of the proper role of authority/government: liberals tend to view government's role as that of a helpful Sherpa, whereas conservatives tend to envision a strict schoolmaster (for example, Barker and Tinnick Reference Barker and Tinnick2006; Lakoff Reference Lakoff1996; McAdams et al. Reference McAdams2008).
Other Psychological Constructs of Liberalism–Conservatism
Importantly, though, the psychology of political ideology is about more than what can be objectively dubbed ‘values’. Indeed, some of the normative orientations we have listed above (for example, hierarchism, traditionalism, punitiveness, in-group loyalty, order, tough-mindedness, ethnocentrism) overlap with broader dispositions toward authoritarianism and/or social dominance, each of which undergirds political conservatism – though in somewhat different ways, with authoritarianism being more predictive of cultural conservatism and social dominance orientation being more predictive of economic conservatism (for example, Duckitt Reference Duckitt2001; Feldman and Stenner Reference Feldman and Stenner1997; Haidt Reference Haidt2012; Hetherington and Weiler Reference Hetherington and Weiler2009; Jost et al. Reference Jost2003; Pratto et al. Reference Pratto1994; Sidanius and Pratto Reference Sidanius and Pratto1999; Stenner 3005).
Likewise, some of the value conflicts we have described are also consistent with Cultural Theory's (CT) ‘group’ and ‘grid’ dimensions of social relations (for example, Thompson, Ellis and Wildavsky Reference Thompson, Ellis and Wildavsky1990). The group dimension refers to the extent to which an individual is incorporated into bounded social units. The more one is incorporated in this way, the more likely one is to possess progressive economic values (egalitarianism, humanitarianism, nurturance, etc.). The grid dimension refers to the extent to which an individual's life is circumscribed by externally imposed prescriptions. The more it is, the more likely one is to possess conservative cultural and security values (moral and sexual traditionalism, punitiveness, etc.).
In summary, the correspondence between value priorities and liberalism–conservatism is intuitive, compelling and well established, but it is also incomplete. Liberalism–conservatism also reflects deeper psychological dispositions relating to order, change, certainty, simplicity, cognitive closure, anxiety, threat sensitivity and negativity bias, among other things (for example, Hibbing, Smith and Alford Reference Hibbing, Smith and Alford2014; Jost et al. Reference Jost2003; Jost et al. Reference Jost2007; Pedersen, Muftuler and Larson Reference Pedersen, Muftuler and Larson2017; for good overviews, see Jost Reference Jost2006; Jost, Federico and Napier Reference Jost, Federico and Napier2009; Jost et al. Reference Jost2003). We believe these psychological dispositions are likely related to various abstract beliefs about human nature and society – a.k.a. ‘premises’ – which inform part of what we mean when we describe someone as ‘liberal’ or ‘conservative’. In the remainder of this article, we explore the potential relationships between premise beliefs and liberalism–conservatism more fully than has been attempted to date.
Premises
To reiterate, premises are not about how people believe things should be, but rather how people believe things are.Footnote 6 Theoretically speaking, it stands to Aristotelian reason that one pathway through which humans form specific attitudes is by extrapolating them from a set of abstract premises to which they subscribe (for example, Kruglanski Reference Kruglanski1999). And indeed, since more or less the beginning of recorded time, premises about human nature have been a common preoccupation of political, religious and economic philosophers (for example, Augustine ~400; Hobbes Reference Hobbes1651; Jefferson Reference Jefferson1776; Locke 1689; Madison Reference Madison1787; Plato 375 BCE; Rousseau Reference Rousseau1762; Smith Reference Smith1776; Thucydides 404 BCE).
Yet in their empirical accounts of political attitudes and ideology, political scientists typically ignore premise disputes or conflate them with value conflicts (for example, Converse Reference Converse and Apter D1964; Feldman Reference Feldman1988; Sears, Huddy and Schaffer Reference Sears, Huddy and Schaffer1986). This standard practice is certainly understandable; it is difficult in practice to delineate where premises stop and values begin, and the causal relationship between them is certainly not clear.Footnote 7 Nevertheless, conceptually and theoretically speaking, it is important to distinguish premise disputes from value conflicts, and to give the former greater empirical attention.
A few lines of extant research do point to premises as important undercurrents of liberalism–conservatism. For example, Tomkins (Reference Tomkins and White1964) argues that conservatives are more likely than liberals to presume that people are essentially bad and that the world is a dangerous place – premises that are consistent with (or perhaps emerge from) heightened psychological needs for security, order and so on (see Jost et al. Reference Jost2003 for an excellent overview of the way psychological needs inform liberalism–conservatism). In the foreign policy realm, Brewer and Steenbergen show that support for isolationist or co-operative policies is connected to views of human nature (Reference Brewer and Steenbergen2002).
Moreover, the CT scholars to which we referred earlier distinguish between ‘crowning postures’ that are descriptive and those that are prescriptive (though they use different terms than most other political psychologists have). They posit that different ‘myths’ about human and physical nature (what we would call premises) combine with values and ‘grid’ versus ‘group’ social relations to produce four different cultures (hierarchism, egalitarianism, individualism and fatalism; for example, Coughlin and Lockhart Reference Coughlin and Lockhart1998; Douglas Reference Douglas1982; Ellis Reference Ellis1993; Grendstad and Selle Reference Grendstad and Selle2000; Swedlow Reference Swedlow2011; Thompson, Ellis and Wildavsky Reference Thompson, Ellis and Wildavsky1990; Wildavsky and Dake Reference Wildavsky and Dake1990; see Swedlow et al. Reference Swedlow2016 for a for an overview of this literature and the measurement scales that CT scholars have used).Footnote 8
Finally, John Duckitt and his colleagues make some empirical progress, using what we would term premise beliefs (they call them ‘worldviews’) to predict authoritarianism and social dominance orientation – which they in turn use to predict political liberalism–conservatism (for example, Duckitt Reference Duckitt2001; Sibley, Wilson and Duckitt Reference Sibley, Wilson and Duckitt2007, also see Altemeyer Reference Altemeyer1988; Altemeyer Reference Altemeyer1998; Federico, Hunt and Ergun Reference Federico, Hunt and Ergun2009).
However, scholars have yet to systematically examine the independent role of specific premise disputes as ideological ingredients, which differentiates them empirically from value conflicts in models of liberalism–conservatism. We make such an effort here. In the next sub-section, we provide the theoretical scaffolding needed to ground our hypotheses with respect to three specific premise disputes about human nature and society.
Specific Premise Disputes and Liberalism–Conservatism
In examining premises and liberalism–conservatism, there are any number of specific premises on which we might have focused. We focus on three foundational premise disputes that have grounded philosophical arguments for centuries: Are humans intrinsically benevolent or selfish? Is significant social progress attainable without God's fingerprints? Is society intrinsically stable or fragile? These questions, though distinct, are interrelated in that they each contrast a sanguine impression of human nature and society with a cynical one.
Intrinsic Human Benevolence–Selfishness
Premise disputes regarding human benevolence–selfishness undergird the predominant schools of religious, economic and political philosophy in the United States. With regard to religious philosophy, many Christian traditions (especially evangelical and fundamentalist ones) begin from the Calvinistic premise that humans are fundamentally depraved, while others (most mainline Protestant and Catholic traditions) view human nature as being more complicated (for example, Williams Reference Williams2008). When it comes to economic philosophy (which has been central to left–right orientations for as long as that distinction has had meaning), laissez-faire capitalism assumes that the inherent selfishness of humankind must be channeled (for example, Mandeville Reference Mandeville1732; Smith Reference Smith1776), whereas socialists and welfare state advocates place much greater faith in the co-operative instincts of humans (for example, Marx and Engels Reference Marx and Engels1848; Rawls Reference Rawls1971). Finally, as for patently political philosophy, classical proponents of popular democracy assumed that human nature is broadly altruistic and co-operative (for example, Jefferson Reference Jefferson1776; Rousseau Reference Rousseau1762), whereas those who favored a republic (or stronger forms of government) assumed that human nature is either selfish or that it varies greatly across individuals (for example, Hobbes Reference Hobbes1651; Madison Reference Madison1787; also see Golding Reference Golding1954).
Grounding their arguments to some degree in those classical philosophies, CT researchers (for example, Grendstad and Selle Reference Grendstad and Selle2000; Swedlow Reference Swedlow2011; Thompson, Ellis and Wildavsky Reference Thompson, Ellis and Wildavsky1990) posit that (1) egalitarianism (which corresponds to liberalism, both economically and socially) naturally follows from the premise that human beings are intrinsically well intentioned, whereas (2) individualism (which corresponds to economic but not social policy conservatism) grows out of the presumption that humans are immutably selfish and (3) hierarchism (which corresponds to cultural but not economic policy conservatism), stems from the premise that human character is naturally depraved but can (and indeed must) be redeemed by the right institutions.Footnote 9
Additionally, Silvan Tomkins’ (Reference Tomkins and White1964) polarity theory posits that left-wing ideologies presume that people are basically good and that the purpose of society is to facilitate human growth, whereas right-wing ideologies presume people are essentially bad and that the function of society is to set limits to prevent irresponsible behavior (also see de St. Aubin Reference de St. Aubin1996; Stone and Schaffner Reference Stone and Schaffner1988).
Moreover, John Duckitt and his colleagues (for example, Duckitt Reference Duckitt2001; Sibley, Wilson and Duckitt Reference Sibley, Wilson and Duckitt2007) have observed that those who believe the world is a competitive jungle (and by inference that people are intrinsically competitive rather than co-operative) tend to be oriented toward social dominance, which is itself associated with economic conservatism (for example, Pratto et al. Reference Pratto1994; Sidanius and Pratto Reference Sidanius and Pratto1999).
Consistent with these perspectives, we posit that conservatism grows (in part) out of the conviction that human character is inclined toward self-interest, aggression and sloth, which is why conservatives perceive the need for institutions such as the market, the church and the family to counter such natural tendencies, and why they view the proper role of authority as disciplinary, albeit limited (for example, Will Reference Will2019). By contrast, we posit that liberalism stems in part from the belief that human character is either intrinsically well intentioned or a blank slate, meaning that neither a highly competitive economic system nor a strongly punitive justice system is necessary to encourage people to act benevolently and responsibly, and that the proper role of government is therefore to provide nurturance (for more theoretical elaboration (but no empirical tests), see Marietta Reference Marietta2011). In sum:
Hypothesis 1: Belief in the premise that human character is inherently and immutably benevolent (selfish) is associated with ideological liberalism (conservatism).
Progress Idealism–Skepticism
Premise disputes regarding human nature are not only about inherent human decency; they also relate to inherent human capacity – including the capacity for improvement. Even if humanity's goals are magnanimous, how capable are humans of achieving them? Again, many Christian traditions offer a skeptical response, unless God (or the agents of God) step(s) in. Indeed, the classical Lutheran and Calvinistic presumptions of human depravity were about weakness as well as wickedness. This cultural premise was paramount across Protestant Europe in the 1600s–1700s until philosophers like Rousseau challenged it, promoting a humanistic vision of relentless human progress instead (see Pinker Reference Pinker2018 for a recent overview). A backlash followed, with British conservative intellectual forbear Edmund Burke (Reference Burke1790) reasserting a classic questioning of secular human capacity as justification for his skepticism toward popular democracy and broader Utopian ambitions.Footnote 10
Here, like before, CT has something to contribute. According to the CT typology, egalitarians believe that human beings have great capacity for change (and thus to achieve progress) as long as heavy-handed institutions do not hold them back (Thompson, Ellis and Wildavsky Reference Thompson, Ellis and Wildavsky1990). This understanding is consistent with a liberal vision of humankind that is progressing steadily toward a free, equal and prosperous Utopia. By contrast, per CT, individualists (who, again, are economically conservative) presume that human selfishness is immutable, and that progress is therefore possible if and only if social/economic structures encourage and harness that selfishness into competition (that is, as long as they do not get in the way). Sharing individualists’ pessimism regarding human nature but not its capacity for change, CT's hierarchists (who may lean left economically because of their high ‘group’ orientation but are culturally conservative because of their high ‘grid’ orientation) presume the opposite – that human weakness can be fortified but only by coercive institutions, such as the church or a legal/governmental framework that is designed to impose discipline. If the heirarchists are right, then perhaps some definitions of ‘progress’ might be achievable, but secular Utopian visions (in which communities are both co-operative and free from coercion) would appear laughably naïve.Footnote 11
Moreover, Jost et al. (Reference Jost2003) review several studies that treat liberalism as an ideology that welcomes change, which is consistent with an idealistic view of the potential for social progress through human efforts, and conservatism as an ideology that resists change, which suggests a cynical view of the potential for social progress through human efforts.
By making the proposed relationship between progress idealism–skepticism and ideology explicit, we argue that regardless of one's presumptions about inherent human decency, if one thinks that government-sponsored attempts to engender progress will inevitably lead to ruin instead – because ordinary people are bound to muck things up – then one is not going to support such efforts, preferring instead to trust ‘the invisible hand’ to maintain equilibrium. This is the essence of free market conservatism (also see Will Reference Will2019). Economic liberals believe the public, via the government, can (and must) try to make things better through direct market interventions. If liberals were to share conservatives’ faith that such efforts are all futile anyway, even they would presumably put their energies elsewhere (again, see Marietta Reference Marietta2011 for more theoretical elaboration). In sum:
Hypothesis 2: Idealism (skepticism) with respect to the notion that human efforts can produce social progress is associated with ideological liberalism (conservatism) – especially economic liberalism (conservatism).
Social Stability–Fragility
Finally, we suggest that premise disputes regarding society's inherent social stability–fragility are also an important ideological determinant. Specifically, we expect that those who perceive the nation's defense, economy or culture to be necessarily precarious are more likely to also see threats looming around every corner – a tendency that often corresponds to authoritarianism and political conservatism more broadly (for example, Feldman and Stenner Reference Feldman and Stenner1997; Hibbing, Smith and Alford Reference Hibbing, Smith and Alford2014; Jost et al. Reference Jost2003; Jost et al. Reference Jost2007).Footnote 12 That is, differences in assumptions regarding the degree to which the world is dangerous – and thus that civilized society is inherently precarious – may prompt greater security concerns and suspicion of outsiders, thereby priming socio-cultural conservatism (for example, Duckitt and Fisher Reference Duckitt and Fisher2003; Federico, Hunt and Ergun Reference Federico, Hunt and Ergun2009).Footnote 13 As with the previous two hypotheses, see Marietta (Reference Marietta2011) for a more detailed theoretical argument.
It is worth considering, though, whether this premise dispute regarding social stability–fragility might itself be partially derivative of the views about human nature that we have already discussed. If an individual believes humankind is inherently malevolent – whether redeemable or not – then he or she is more likely to view the world as a dangerous place filled with bad people (reflected in other nations, other cultures and so on) who pose a threat to American culture and society. In other words, to worry that one's own nation or culture is precarious, one has to presume that foreign nations and cultures will inflict harm if given the opportunity. And to presume that, one has to also believe that basic human nature is wicked or at least highly selfish and competitive. However, if human beings were – on the whole and in roughly equal proportion to one another – decent and well intentioned, then there would be no reason to distrust their motives, and there would therefore be less reason to feel threatened by changes to the culture, economy or world.Footnote 14
Despite such theoretical overlap, for now we treat the social stability–fragility premise dispute as distinct from the premise disputes that pertain directly to human nature. Stated formally:
Hypothesis 3: Belief in the premise that the nation and culture are stable (fragile) is associated with ideological liberalism (conservatism) – especially cultural and security liberalism (conservatism).
We conducted three studies to test our hypotheses. Studies 1 and 2 use observational data from 2013 and 2014 to examine the degree to which the three premise disputes described above predict ideological identification and issue-based political ideology, while also accounting for prominent value conflicts and other confounds. Study 3 uses a priming experiment to observe the extent to which (a) liberals respond favorably to ‘evidence’ that paints a sanguine portrait of human nature and society while rejecting information to the contrary (Hypothesis 4a) and (b) conservatives react precisely the opposite way (Hypothesis 4b).
Studies 1 and 2: Observational Evidence
In these studies, we estimated a series of generalized ordered-logit and ordinary least squares regression models predicting liberalism–conservatism with three premise disputes (human benevolence–selfishness, progress idealism–skepticism and societal stability–fragility), five value conflicts (collectivism–individualism, humanism–theism, sexual progressivism–traditionalism, pacifism–militarism and nurturance–punitiveness) as well as other potential confounds.
Data and Measurement
We collected nationally representative survey responses as part of the 2013 and 2014 Cooperative Congressional Election Studies (CCES).Footnote 15 For efficiency's sake, we focus our discussion on the 2014 data and results; we present the 2013 results – which are highly similar to those we discuss below – in Appendix Section III, 10–14.
Measuring liberalism–conservatism
To assess the empirical robustness of our findings, and in recognition of the fact that liberalism–conservatism resonates in American politics both as an identity and as a constellation of policy preferences, we measure it using five distinct dependent variables: (a) a single survey item of self-identified liberalism–conservatism (three-point scale, converted to 0–1; ‘conservative’= 1), (b) a policy preference factor score index (maximum-likelihood extraction) on the following issues: poverty, education, health care, minimum wage, the environment, racial inequality, undocumented immigration, gun control, abortion, gay marriage, defense, terrorism, affirmative action and the Iraq War (Cronbach's α = 0.86; Eigenvalue = 5.11); (c) an economic policy preference factor score index (poverty, the minimum wage, education, health care, minimum wage and the environment; Cronbach's α = 0.82; Eigenvalue = 2.77); (d) a ‘culture war’ policy preference index (abortion, gay marriage, immigration, affirmative action and gun control; Cronbach's α = 0.63; single factor extracted; Eigenvalue = 1.51) and (e) a security policy preference factor score index (defense, terrorism and the Iraq War; Cronbach's α = 0.66; Eigenvalue = 1.38).Footnote 16 The specific question wording, factor loadings and communalities for each indicator appear in the Appendix. We deleted missing observations using the listwise procedure (in alternative specifications we imputed missing observations in the issue-based liberalism–conservatism measure, which improves the statistical significance of some relationships. See Appendix Section I, Table A5, 6).
Measuring premise disputes
We again used factor analysis (maximum-likelihood extraction) to measure the three premise disputes we discussed earlier in order to generate multi-item indices of each premise dispute and to attenuate random error. First, we presented respondents with a series of paired statements. For each paired comparison, we asked them to indicate, on a four-point scale, which statement they agreed with more (0 = clear preference for the first statement and 3 = clear preference for the second statement). To capture human benevolence–selfishness, we used three paired comparisons:
• ‘People naturally compete rather than cooperate’ vs.
‘People naturally cooperate rather than compete’
• ‘People will seek their own interest, rather than what is morally right’ vs.
‘People will do what is morally right, rather than seek their own interest’
• ‘People try to get as much for themselves as they can, even if it is morally wrong’ vs. ‘People try to do what is morally right, even if they have to sacrifice’
To measure progress idealism–skepticism, we used four paired comparisons:
• ‘The world can be made better if we work together’ vs.
‘The world is what it is and will remain so’
• ‘Humans can change and improve with each new generation’ vs.
‘People in each generation are essentially the same’
• ‘Efforts to improve our society have generally led to good outcomes’ vs.
‘Efforts to improve our society have generally led to bad outcomes’
• ‘Human society can be perfected’ vs.
‘Human society is essentially fixed and cannot be perfected’
To measure social stability–fragility, we used three paired comparisons:Footnote 17
• ‘Our society is vulnerable to losing our place of strength to other nations’ vs.
‘Our society does not need to worry about declining strength compared to other nations’
• ‘The American Way is in danger’ vs.
‘The American Way is changing but there is nothing to worry about’
• ‘The decline of personal character in our society is weakening us’ vs.
‘The personal character of Americans is as strong as ever’
Measuring value conflicts
We chose to include four sets of value conflicts as predictors in the models. In keeping with our discussion in the previous section, the first three correspond directly to the economic, cultural and foreign policy dimensions of liberalism–conservatism: collectivism–individualism, humanism–theism and pacifism–militarism. We also measured nurturance–punitiveness as an additional ‘catch-all’ value, because it is highly associated with political attitudes across the board and with ideological constraint (Barker and Tinnick Reference Barker and Tinnick2006). Thus its inclusion represents an effort on our part to go ‘above and beyond’ in accounting for other confounds that may be related to our premise disputes.
To measure the first three value conflicts, we presented respondents with sets of competing values and asked them to indicate ‘Which is more important?’ The sets of competing norms were as follows: ‘Cooperation & Helping Others … or Self-Reliance & Personal Responsibility’ (collectivism–individualism), ‘Peace … or Protecting America with Force, if Necessary’ (pacifism–militarism), ‘Scientific Knowledge … or Religious Faith’ (humanism–theism).Footnote 18 To measure nurturance–punitiveness, we asked respondents, ‘If you had to define what it means to be a ‘good person,’ which quality would you say is more important? Discipline & Self-Control or Kindheartedness?’ We captured responses on a four-point scale (0 = strong preference for the first option, 3 = strong preference for the second option).
Measuring other confounds
Finally, to round out our models, we include covariates for Race (1 = non-White), Gender (1 = Female), Year of Birth (1920–1995), Gross Family Income (0 = 0; <10k = 1; 10–19k = 2; 20–29k = 3; 30–39k = 4; 40–49k = 5; 50–59k = 6; 60–69k = 7; 70–79k = 8; 80–99k = 9; 100–119k = 10; 120–149k = 11; 150–199k = 12; 200–249k = 13; 250k + = 14) and Educational Attainment (‘What is the highest level of schooling you have completed?’; 0 ≤ HS; 1 = HS only; 2 = some college; 3 = 2-year college degree; 4 = 4-year college degree; 5 = post-graduate).Footnote 19
Validating the Model: Confirmatory Factor Analysis
To evaluate our chosen model specification – and the theory underlying it that treats premise disputes and value constructs as related but distinct elements of political cognition – we performed a series of confirmatory factor analyses within a structural equation modeling environment (maximum-likelihood estimation) to ascertain the degree to which our theoretically derived measurement model fits the data compared to alternative theoretical specifications (see Bentler Reference Bentler1990). Specifically, we compared the theoretical measurement model that we described above to (a) a model that uses all of the premise indicators to form a single Human Nature Optimism–Pessimism index and (b) a model that uses all of the premise indicators and all of the value indicators to form a single belief system index. As the Comparative Fit and Tucker-Lewis Fit Indices reveal below, our theoretical measurement model fits the data better than the others (LR Test p < 0.001).
Predictive Modeling Results
Tables 2 and 3 display the results of our general ordered logit and linear regression analyses. Table 2 displays the results as they pertain to the single-dimension ideological constructs, whereas Table 3 displays the results related to different dimensions of ideology.Footnote 20 For simplicity, we do not include the demographic covariates in the table; the full models are available in Appendix Section I, Table A1, 1).
Table 1. Confirmatory factor analyses: model comparison
Note: comparison of model fit statistics (on 0–1 scales) between (a) the theoretical model we have described here, which distinguishes between three premise disputes (Model 1), (b) a second model that presumes all premise dispute indicators contribute to a single latent concept of premise beliefs but that distinguishes value conflicts (Model 2), and (c) a third model that presumes all premise and value indicators contribute to a single underlying belief structure (Model 3). All differences are statistically significant (p < 0.001).
Table 2. Premise disputes, value conflicts and liberalism–conservatism
Note: the left-hand model is a generalized ordinal logistic regression estimate; coefficients are differences in the odds of identifying as a ‘conservative’ relative to a ‘liberal’, associated with minimum-to-maximum differences in the explanatory variables. The right-hand model is an ordinary least squares regression estimate. Coefficients are differences in policy liberalism–conservatism (conservatism coded high and converted to 0–1 scales) associated with minimum-to-maximum differences in each independent variable. In both models, the p-values correspond to two-tailed tests.
Table 3. 2014 Linear regression analyses of ideological sub-dimensions (Economic, Cultural, Security)
Note: the coefficients are differences in economic, cultural and security liberalism–conservatism (conservatism coded high and converted to 0–1 scales) associated with minimum-to-maximum differences in each independent variable. The probabilities that the null hypothesis in question is true in the population are in parentheses (two-tailed tests).
Starting with Table 2, the first results column focuses on the differences in the odds of identifying as a ‘conservative’, vs. a ‘liberal’, associated with minimum-to-maximum differences in each premise belief or value priority (we omit the estimates relating to the likelihood of identifying as a moderate, relative to either of the other two categories). We see that two of the three premise disputes account for symbolic liberalism–conservatism to a highly substantial degree. Those who were most inclined to believe that American society is fragile and those who were most skeptical that society can be improved through human efforts both tended to be over seven times more likely to identify as a conservative, relative to a liberal, than those who are most convinced that society is stable (p < 0.001) or those who are the most idealistic about society being perfectible (p < 0.001). The only premise that is not statistically related to ideological identification is human benevolence or selfishness: those who strongly indicated that human nature is intrinsically selfish did not appear any more conservative, on average, than those who believe it is intrinsically benevolent.Footnote 21
The right-hand column shows that, with respect to policy-based ideology, respondents who were most inclined to believe that American society is fragile tended to score about 24 points higher (more conservative) on the 0–1 scale than those who were most convinced that society is stable (p < 0.001). Moreover, we see that those who were most skeptical that society can be improved through human efforts also tended to be about 24 points more conservative on the 0–1 scale than those who are the most idealistic (p < 0.001). Finally, those who are most convinced that human nature is intrinsically selfish tended to express about five points greater conservatism on the 0–1 scale than those who are the most sanguine about human nature (p < 0.09).
As anticipated, Table 3 demonstrates that when it comes to different dimensions of policy liberalism–conservatism, the predictive capacity of perceived social fragility appears to be stronger than that of skepticism regarding human progress with respect to cultural conservatism and security conservatism, but the opposite is true for economic conservatism. Finally, a pessimistic view of human nature is associated with policy conservatism in the economic and security realms, but not the cultural realm.Footnote 22
Collectively, these models suggest that premise disputes are powerful correlates of liberalism–conservatism, comparable in both substantive and statistical significance to prominent value conflicts. Both the 2014 analysis (presented above) and the 2013 study (presented in the Appendix) illustrate the same conclusions.Footnote 23
In our final study, which we describe below, we attempt to augment the internal validity of these findings by revealing the latent premises of liberals and conservatives through a priming experiment.
Study 3: Revealing Liberal and Conservative Premises through Priming
To gain greater purchase on the question of whether premises about human nature are relevant ingredients of liberalism–conservatism, we conducted a randomized controlled experiment using a non-probability sample of politically interested respondents that we drew from Amazon's online Mechanical Turk (MTurk) labor market (n = 1,588).Footnote 24 We prompted the survey respondents with claims from a (fake) research study purporting to demonstrate that human nature is either good or bad. We then observed the relative propensities of liberals and conservatives to agree with the claim to which they had been exposed, as reflected by their own expressed views on human nature.Footnote 25
We lean on priming theory, which demonstrates that latent beliefs can be made more or less salient to judgments depending on whether they have been ‘primed’ to the front of someone's consciousness. Importantly, priming is different from persuasion: whereas persuasion changes a person's mind, priming makes some existing beliefs or attitudes more salient than others when people are forming a judgment.Footnote 26 Thus priming only works if the beliefs or attitudes in question are already present in the back of a person's mind (for example, Tesler Reference Tesler2015).
If our logic and evidence from Studies 1 and 2 are correct, and premises regarding human nature are indeed ideologically infused, then the experimental treatments should produce a priming effect, but not a persuasive effect. That is, a priming effect would indicate that people react favorably to new ‘information’ that appears to validate their ideological perspective, but not to information that seems to contradict that perspective (for an extended summary of the vast literature on motivated reasoning, confirmation bias and cognitive dissonance, see Marietta and Barker Reference Marietta and Barker2019).Footnote 27 By contrast, a persuasion effect would indicate that respondents change their minds in response to any new information provided by an experimental treatment, regardless of their ideological point of view, which would suggest that premises are not necessarily related to ideology. Likewise, null results would also suggest that premises are not necessarily related to ideology. Either a persuasion effect or null results would weaken our theory.
To summarize and reiterate:
Hypothesis 4a (Liberals): When prompted with information pointing to a positive appraisal of human nature, ideological liberals tend to express greater sanguinity toward human nature than control group respondents. By contrast, progressives are not persuaded by information pointing to a cynical view of human nature.
Hypothesis 4b (Conservatives): When prompted with information pointing to a cynical appraisal of human nature, ideological conservatives tend to express greater cynicism about human nature than control group respondents. By contrast, conservatives are not persuaded by information pointing to a sanguine view of human nature.
Experimental Procedures and Measurement
We randomly assigned survey respondents to one of three messages. Respondents in the control group received the following message:
Researchers from the Rand Corporation, Stanford University and Brigham Young University recently published a ten-year study of 30,000 infants and toddlers across four continents. Several publications (including The New York Times, The Wall Street Journal, MSNBC and Fox News) have praised the research for its rigor, transparency and lack of bias.Footnote 28
Respondents in the Humanity Good treatment group saw the same thing, but also saw the following sandwiched in between the first and second sentences: ‘The results consistently suggest that human beings are, more often than not, naturally generous, disciplined, and trustworthy.’ Finally, respondents in the Humanity Bad treatment group saw a different sentence sandwiched between the control group sentences: ‘The results consistently suggest that human beings are, more often than not, naturally selfish, undisciplined, and devious.’
Next, we asked all respondents to answer four questions regarding the nature of human character/capacity, which corresponded to the information in the treatments:Footnote 29
• Do you tend to think that human nature is naturally selfish or generous (generous or selfish), generally speaking? (0 = definitely generous; 4 = definitely selfish)
• Do you tend to think that human nature is naturally devious or trustworthy (trustworthy or devious), generally speaking? (0 = definitely trustworthy; 4 = definitely devious)
• Do you tend to think that human nature is naturally disciplined or undisciplined (undisciplined or disciplined), generally speaking? (0 = definitely disciplined; 4 = definitely undisciplined)
• Given human nature, would you say that collective public efforts to improve society tend to eventually succeed or fail (fail or succeed)? (0 = definitely succeed; 4 = definitely fail)
Using factor analysis (maximum-likelihood extraction), we created a Human Nature Sanguinity–Cynicism index of the four human nature questions listed above (Cronbach's α = 0.69; Eigenvalue = 1.49), and converted it to a 0–1 scale for analysis (mean = 0.47; standard deviation = 0.37). This index serves as the outcome variable in our model (the results from binary probit models using each individual premise dispute as the outcome variable (collapsing the definitely and probably categories and eliminating the ‘don't know’ middle categories) are available in Appendix Section IV, Table A10, 17).Footnote 30
Then we asked respondents to report their Ideological Identification (three-point, rescaled to 0–1; 0 = liberal; 1 = conservative; mean = 0.38; s.d. = 0.42). The model's explanatory variables are (a) both experimental treatments (the control group is the reference category), (b) Ideological Identification and (c) interaction terms that multiply Ideological Identification by each of the treatments.Footnote 31
We rounded out our models with the following demographic covariates: White (77 per cent of sample), Female (50 per cent of sample), Age (18–85; mean = 42; s.d. = 13.25), Annual Household Income (mean = $39,900; s.d. = $27,000) and Education (6-point ≤ high-school; 6 = post-graduate degree (converted to 0–1; mean = 0.66; s.d. = 0.25)).Footnote 32
Results
As Figure 1 displays (Appendix Table A3 contains the full statistical results), among liberals (illustrated in the lower half of the figure), the Humanity Bad treatment produced no boost in cynicism relative to the control group, but the Humanity Good treatment generated a drop in cynicism (or a boost in sanguinity) from 0.46 to 0.30 on the 0–1 scale (p < 0.001). Likewise, among conservatives, the Humanity Good treatment did not reduce respondents’ stated cynicism regarding human nature, but the Humanity Bad treatment boosted such cynicism (or reduced sanguinity) from 0.5 to 0.68 on the 0–1 scale (p < 0.001).Footnote 33 In other words, ideologues readily embraced information that comported with what we and others have suggested is their instinctual point of view, but resisted information that contradicted that point of view.Footnote 34 This evidence is consistent with a priming effect but not a persuasion effect, and thereby suggests that liberals and conservatives have different latent premises about human nature that they may not necessarily think about until they are primed to do so, but they are not easily persuaded to think otherwise. Another interpretation is that the liberals and conservatives in our sample applied different confirmation biases when they encountered the treatment (that is, they respond more favorably to information that is consistent with their prior beliefs), irrespective of priming theory.
Figure 1. Premise prompts and cynicism toward human nature among liberals and conservatives
Note: differences in human nature sanguinity-cynicism associated with exposure to the experimental treatments, among liberals and conservatives.
Conclusion
In this investigation, we have considered the role that premises – abstract descriptive beliefs – play in the formation of political ideology. We have described three studies. The first and second studies use two rounds of observational data to analyze the capacity of premise disputes relating to human benevolence–selfishness, progress idealism–skepticism and social stability–fragility to predict ideological identification and policy ideology, independent of prominent value conflicts. The third study uses a randomized controlled priming experiment to identify the degree to which citizens respond favorably to ‘evidence’ regarding human nature that is consistent with their ideological perspective while rejecting such information when it contradicts that ideological perspective.
The results suggest that these premises are important components of ideology, which should be incorporated into standard models of attitudinal and ideological formation – alongside party identification, identities and values. These observations are important because they (1) provide overdue empirical validation to centuries-old theories from classical philosophy, (2) empirically complement the theoretical arguments of several contemporary theories, (3) explain variance in ideology that has heretofore been present in residual terms and (4) boost the theoretical leverage in models of ideology since premises are positive, rather than normative, dispositions – and are thus immune to criticism that the constructs on either side of the equation are the same.
There are at least two fruitful avenues for future research. First, sorting out the causality between values and premises was beyond the scope of this article, but doing so is an important next step towards understanding the potential for premises to influence ideology indirectly via values (and vice versa).
Secondly, several additional premise disputes remain to be explored as ideological determinants, including the extent to which (1) human nature varies across individuals (regardless of whether it is, on balance, ‘good’ or ‘bad’), (2) various human characteristics are grounded in ‘nature’ vs. ‘nurture’, (3) fate determines outcomes, (4) the ‘invisible hand’ of the free market produces efficient outcomes in the long run, (5) karma is operative (that is, whether ‘what goes around comes around’ is empirically accurate), as well as premises about God and religion such as (6) whether God exists, and if so, (7) how active God is in human events, (8) the extent to which God is merciful or punitive and (9) whether Heaven and/or Hell exists. We also think it could be useful to explore more specific premises such as (10) when human life begins (at conception? at birth?) and (11) the nature/dynamics of gender and sexual differences (to what extent are they intrinsic or socially constructed, and to what extent are gender and sex related?) We encourage others to examine the extent to which these premises predict liberal–conservative issue attitudes, identification and constraint, and how they relate to various psychological needs and motives that underscore ideological differences.
Supplementary material
Data replication sets are available in Harvard Dataverse at: https://doi.org/10.7910/DVN/2GWLWP and online appendices are available at https://doi.org/10.1017/S0007123420000460.
Acknowledgements
The authors are grateful to Justin Gross, Elizabeth Suhay, and the members of the Political Communication Working Group at UMass Lowell (John Cluverius, Joshua Dyck, Mona Kleinberg, and Jenifer Whitten-Woodring) as well as three anonymous reviewers for their helpful comments on earlier drafts.
