Rock and a hard place: public willingness to trade civil rights and liberties for greater security.(Report)

Introduction

People in every society grapple with the relationship between freedom and security, often with relatively little say about how the two are balanced. In developed polyarchies (Dahl 1972) the relationship occupies a central place in public policy, and when exogenous or internal forces are perceived to threaten domestic order, the balance is likely to become a matter of substantial public interest. For citizens of the United States, the terrorist attacks of September 11, 2001, led to a seismic shift in policies concerning security and to significant reassessments of the risks of terrorism (Huddy, Khatib, and Capelos 2002; Jenkins-Smith and Herron 2005). This paper focuses on how Americans understand the balancing of civil rights and liberties, on one hand, with security from terrorism on the other. Beliefs about such relationships are fundamentally political, and to better understand the structures of balancing mechanisms, we investigate the following key questions.

* Are preferences and perceptions of the balance of U.S. liberty and security in the contemporary age of terrorism systematically influenced by political beliefs, and if so, how?

* How do belief structures underlying the relationship of liberty and security vary among different political orientations in the United States?

We begin by placing the tensions between rights, liberties, and security in an historical and analytical frame. Despite similarities between modern events and historical clashes between freedom and security, the current dilemma raises distinct challenges in that Americans face a conflict with demonstrably potent but shadowy opponents for which an end to the threat is difficult to discern. We employ recent data from Internet survey interviews with a cross section of U.S. residents to evaluate the ways in which Americans understand the relationships among threats, trust in government, and willingness to curtail rights and liberties for greater security from terror threats. The intent of this paper is to evaluate Americans' current beliefs and policy perceptions concerning the relationship of security and liberty and to better understand how individual political orientations help shape those understandings.

Framing the Security/Freedom Relationship

Tensions between civil liberties and security measures in the contemporary struggle against terrorism are subsets of larger and more enduring relationships between freedom and security. These key dynamics sometimes are mistakenly cast in zero sum terms (particularly by the media), suggesting that gains in security and order necessarily come at the expense of freedom and liberty, or that increasing freedom always diminishes security. But, within some range, the relationship is symbiotic. Freedom and civil liberties are not possible in the absence of some minimum level of security and social order. As Judge Learned Hand put it: "A society in which men recognize no check upon their freedom soon becomes a society where freedom is the possession of only a savage few" (Hand 1952, 191). Conversely, in societies founded on liberal democratic principles, social order and security are unsustainable without the willing support of free citizens who legitimate and empower government to secure society from physical threats. However, at times when social order and public security are under threat, policies that infringe on civil liberties in pursuit of enhanced security become more attractive to policy makers, and it is under these circumstances that freedom and security can become juxtaposed. The United States has experienced many such instances in the past, and we are again facing such conditions in today's response to the threat of terrorism.

Historically, such trade-offs have been justified most often in times of war or impending war in which threats to the social order and political system are understood to pose clear and present dangers. But such circumstances also can be used to justify motivations other than legitimate security requirements, such as the aggrandizement of power, pursuit of ideological beliefs, and the advancement of personal and political ambitions. Brinkley (2003, 23) asserts that: "Every major crisis in our history has led to abridgements of personal liberty, some of them inevitable and justified. But in most such crises, governments have also used the seriousness of their mission to seize powers far in excess of what the emergency requires." Even a brief review of such exigencies seems sufficient to illustrate Brinkley's concerns.

Beginning very early in our national history, President John Adams used the threat of impending war with France (which did not occur) to pressure Congress into passing the Alien and Sedition Acts of 1798, which authorized the president to detain, confine, or deport citizens or subjects of an enemy nation without cause during times of war.

During the U.S. Civil War, President Abraham Lincoln declared martial law and temporarily suspended the writ of habeas corpus (on multiple occasions) under which the government is required to specify in legal proceedings the reasons for detaining an individual.

During World War I, under the Espionage Act of 1917, supplemented by the Sedition Act of 1918, the federal government prosecuted some 2,000 people for opposing the war and military conscription, and those convicted were routinely sentenced to severe terms of 10-20 years in prison. The effect largely was to stifle genuine debate about the merits of the war (Stone 2004).

During the first "Red scare" of 1919-20, then Attorney General Mitchell Palmer established a General Intelligence Division within the Federal Bureau of Investigation to monitor communist activities in the United States. This department conducted raids in 33 cities, arresting more than 5,000 people suspected of socialist or communist radicalism; more than 1,000 aliens were deported. These activities became known notoriously as the "Palmer Raids" (Stone 2003).

Immediately after the Japanese attack on Pearl Harbor in 1941, martial law was declared in Hawaii, and the writ of habeas corpus was suspended. On the mainland, the Roosevelt administration imposed travel restrictions and nighttime curfews on all enemy aliens and interred without due process approximately 120,000 people of Japanese ancestry in what were, in reality, prison camps.

After World War II and the transition to the Cold War era, a second "Red scare" evolved to near hysteria, with intimidations and civil infringements that are too numerous to recount in detail here. The most egregious excesses by the House Un-American Activities Committee and Senator Joseph McCarthy are widely known. Less remembered are such activities as loyalty oaths required of thousands of government employees and legal restrictions on political activities imposed by Congress.

Government attempts during the 1960s to stifle and control widespread dissent about the war in Vietnam are legion and helped lead, at least indirectly, to the failure of the Johnson and Nixon administrations. During this period, government heavy-handedness in suppressing dissent, infiltrating opposition groups, and spying on citizens was matched by widespread public unrest, riots, and eventually by what may have been the greatest groundswell of support for civil liberties in the country's history (Neier 2004).

Through all these periods, and others, the American people participated in adjustments to the tensions characterizing the relationships between freedom and the protections of its liberties versus society's continuing need for some minimally adequate level of security and order necessary for the exercise of those liberties. The full historical pattern is much more rich and detailed than can be done justice here, but it seems to suggest an equilibrating process in which an existing equilibrium between freedom and security is disturbed by some perceived threat that in most cases has been external. Government responds with initiatives to enhance security, some of which may infringe significantly on civil liberties and individual prerogatives. These actions may include other underlying motivations as well, such as shifts in relative power between the executive and legislative branches of government. The judicial branch plays a role in adjudicating the new dynamics, but because of the temporal dimension of litigation, and because of the courts' reluctance to hinder government's ability to act in crises, the judicial check often has been delayed or deferred. (1) In the end, it is the public at large who must accede to or reject the adjustments to the preexisting equilibrium, and the historical record suggests that the public has most often initially been accepting of efforts to enhance security. Over time, as the original threat subsides, or as public acquiescence wanes, a shift in dynamics occurs. The public begins to decide that the nature of the risk has changed from the original external physical security threat to the internal domestic threat to civil liberties posed by intrusive government (Matthew and Shambaugh 2005). With sufficient public pressure, the balance may eventually be readjusted to a new state of equilibrium.

Some of these historical cases have definite endings followed by readjustments toward pre-event equilibrium. The finality of conclusions to the U.S. Civil War, World War I, and World War II provide clear endings of the threat situations used to justify incursions into civil liberties, but following both world wars, new fears justifying other incursions soon were raised. Other cases are not clearly delineated. Public rebellion against infringements of citizens' rights and liberties during the McCarthy era were not associated with the end of the Cold War or the demise of communism. Similarly, public resistance to government attempts to stifle opposition to the war in Vietnam occurred well before the United States ended its support for that conflict. Indeed, public pressure helped force the war's end. Without clear conclusions, such as wartime victories, public pressures for shifting emphasis from security measures to reasserting liberty priorities forced major adjustments to the freedom-security relationship.

Today's external threat of terrorism poses a similar challenge because it seems unlikely to end in the foreseeable future, or at least to not have a finite and legal termination. It is not yet clear what would constitute victory, defeat, or even an end to the struggle against terrorism. It also poses the threat of mass casualty acts in which nuclear/radiological or biological weapons could be employed against civilian population centers. While such events may have very low probability of occurrence, they carry such high potential consequences that they raise the question of whether beliefs of how freedom and security should relate might be altered permanently if such weapons are used. (2)

These conditions strengthen and further emphasize the need to better understand how the American people relate security and freedom. If that relationship is fundamentally political, what is the role of political values? How do beliefs about political culture and political ideology help shape the freedom-security relationship? Does political partisanship exert predictable influence? Do beliefs about how liberty and security should relate vary systematically among strong Democrats, moderates, and strong Republicans? How are public assessments of the threat of terrorism balanced against civil-liberty values, such as free speech, due process, and privacy? How does trust in government affect the relationship? Do ordinary members of the public exhibit sufficiently structured patterns of beliefs that support normative preferences for how freedom and security should be balanced, and can they relate current policies to those normative preferences?

Conflicting Views in the Literature

The literature on reasoning and political choice suggests that ordinary citizens employ structured beliefs and a variety of heuristics in organizing views and developing preferences about complex issue domains, such as security and freedom (Herron and Jenkins-Smith 2006a; Hurwitz and Peffley 1987; Jenkins-Smith, Mitchell, and Herron 2004; Peffley and Hurwitz 1985; Sabatier and Jenkins-Smith 1993; Sniderman, Brody, and Tetlock 1991). But the literature on political belief systems also suggests two competing schools of thought about how citizens employ political ideology and party identification as important heuristics for guiding policy choices. The traditionalist school holds that ideological beliefs and "nonattitudes" are insufficiently structured to serve as reliable guides for mass publics (Almond 1956; Campbell et al. 1960; Conover and Feldman 1981; Converse 1964, 1970, 1987; Feldman 1988; Lippmann 1922, 1925, 1955; Zaller 1992). A revisionist school holds that structured beliefs and complex heuristics support and constrain mass opinions in both domestic and foreign policy domains in much the same ways they act on elites (Herron and Jenkins-Smith 2006a; Hurwitz and Peffley 1987; Page and Shapiro 1992; Peffley and Hurwitz 1985; Sniderman, Brody, and Tetlock 1991; Sniderman and Tetlock 1986; Wittkopf 1983, 1987, 1990, 1994). (3)

Other scholars emphasize the utility of political parties and other heuristics for assisting the ideologically challenged, citing the role of parties in helping to simplify and clarify ideological orientations and policy positions that might otherwise be confusing or misleading to voters. As Levitin and Miller (1979, 767) note, "party identification may have a direct influence on perceptions of the ideological meaning of policy alternatives.... It is as though party identification denotes responses to political parties and connotes ideological responses as well." Similarly, Conover and Feldman (1981, 624) assume that "ideological self-placement reflects a 'psychological attachment' to a particular group." And Arian and Shamir (1983, 143) conclude that: "Left and right definitions are profoundly political; irrespective of issue content and attitude existence, they are mainly party loaded and party linked." In the United States, self-characterizations of political ideology may be influenced by perceptions of the liberal or conservative orientations of key opinion leaders and ideological positions associated with the two dominant political parties.

Views prevalent among some analysts of public opinion are that political sophisticates are capable of political-reasoning processes that derive specific policy preferences from abstract principles, such as ideology, but those abilities do not extend to the less sophisticated. (4) Ordinary citizens often are inadequately informed about issues and lack cognitive sophistication needed to make structured connections between political beliefs and policy positions. In other words, even when heuristic pointers and shortcuts are available, political reasoning does not operate among less sophisticated mass publics in patterns similar to the reasoning of better informed and more politically sophisticated elites. From this perspective, factual knowledge and political sophistication provide key contextual requirements that are absent among mass publics (Delli Carpini and Keeter 1996; Luskin 1987, 1990; Zaller 1992).

In contrast, others argue that lack of political sophistication and factual knowledge do not prohibit ordinary citizens from holding integrated beliefs, effectively employing various heuristics, and making reasoned political choices (Goren 2004; Hurwitz and Peffley 1987; Lupia and McCubbins 1998). Goren (2004, 462) argues that "political sophistication will neither affect the absorption of domain-specific principles from the broader political environment, nor strengthen the impact these principles have on policy preferences." He finds that political sophisticates do not systematically rely more on ideology and structured beliefs to derive policy preferences than do ordinary citizens. Goren's findings support those of Herron and Jenkins-Smith (2002), who identify similar belief structures among mass publics and elite scientists and legislators underlying the complex policy domain of nuclear security. These scholars do not argue that beliefs are as coherently and systematically structured among mass publics as they are among more sophisticated publics, but that structures similar to those found among elites also help inform the political reasoning of ordinary citizens.

One key to better understanding conflicting perspectives and findings about political reasoning is to consider its dimensionality. Chubb, Hagen, and Sniderman (1991, 141-63) argue that political reasoning has two important dimensions, one relating to position (ideological identification) and the other relating to intensity (ideological feelings). Ideological identifications are functions of more cognitive, abstract, and distal reasoning that yield relatively stable beliefs, while ideological feelings are products of affect and proximate likes and dislikes for such things as social groups, alternative policies, and political parties that yield more volatile beliefs. These authors contend that the more consistent are ideological feelings with ideological identification, and the more intensely feelings are held, the more ideologically coherent is political reasoning. They conclude that analyzing political reasoning requires assessing how the cognitive and affective dimensions combine and interact. (5)

To better understand how beliefs about freedom and security are structured at varying levels of political integration, we follow the Chubb, Hagan, and Sniderman model to investigate the interaction of political ideology (cognitive/ distal) and party affinity (affective/proximate) in helping shape public beliefs about the relationship between liberty and security in an age of terrorism. Does the degree to which members of the public integrate their ideological beliefs and affective feelings about political parties differentiate preferences for balancing freedom and security? Do preferences for policies intended to help secure the public from terrorism vary by the level of integration of ideological beliefs and party preferences?

To address these and related questions, we analyze results from a survey of 3,006 respondents to our values survey administered via the Internet, June 6-18, 2007. (6) Question wordings, survey methodology, and data used in the analyses are provided in the Appendix section. (7)

Differentiating the Integration of Political Beliefs

For comparative purposes, we separate respondents into groups based on responses to three questions. First, respondents are asked with which political party they most identify (none; Democratic; Republican; independent; something else). Next, participants are asked whether they slightly, somewhat, or completely identify with that choice. Finally, each is asked to place themselves on a continuous ideological scale ranging from 1 (strongly liberal) to 7 (strongly conservative). Using responses from these three questions we create the typological matrix shown in Table 1 incorporating both cognitive and affective dimensions of political beliefs.

Vertically, the table separates responses to the ideology question into three groups. Those who identify themselves as strongly liberal (1) or liberal (2) are grouped; those who consider their political ideology to be slightly liberal (3), middle of the road (4), or slightly conservative (5) form a moderate group; and those who are conservative (6) or strongly conservative (7) are combined into a third group. Horizontally, those who identify strongly (1) or somewhat (2) with the Democratic party are grouped; those who identify slightly with the Democratic party (3), those who consider themselves to be politically independent (4), those who identify slightly with the Republican party (5), and those who do not identify with any formal political party (0) are combined to form a nonaligned group; and those who identify somewhat (6) or strongly (7) with the Republican party are grouped. The resulting matrix has nine categories, numbered from left to right and labeled as shown. To simplify comparisons, we combine respondents in matrix 2 (nonaligned liberals) with those in matrix 4 (moderate Democrats) to form a group we term liberal leaning. Similarly, the moderate Republicans in matrix 6 are grouped with nonaligned conservatives in matrix 8 to form a conservative leaning group. The few respondents (?1%) in matrix 3 (liberal Republicans) and (?2%) in matrix 7 (conservative Democrats) are excluded from further analysis because the inconsistencies with which they express ideology and partisanship make typing them problematic. We retain the five remaining and reorganized categories for further analysis. They include the liberal Democrats in matrix 1; liberal leaning respondents combined from matrices 2 and 4; moderate nonaligned participants in matrix 5; those combined from matrices 6 and 8 who are conservative leaning; and conservative Republicans in matrix 9.

Key Metrics

In addition to the measures of political beliefs previously described, we also include questions to provide the following additional metrics which we hypothesize may be importantly related to preferences about relating freedom and security: (1) security from terrorism; (2) support for freedom of speech; (3) support for due process; (4) support for privacy; (5) measures of trust in government; (6) measures of political culture; (7) a normative or preferred balance between liberty and security; and (8) perceptions of the existing balance between liberty and security. Next we briefly describe how each metric is formed.

Conceptualizing and Measuring Beliefs about Security from Terrorism

The term "security" is associated with contextual meanings that are so broad and variable that some scholars consider it to be an "essentially contested concept" (Buzan 1991; Freedman 1992; Gallie 1962; Rothschild 1995). While a detailed examination of the concept of security is beyond the scope of this discussion, it is useful make a few key points. Arnold Wolfers' (1952, 485) enduring perspective on security may provide the best definition for our purposes: "security, in an objective sense, measures the absence of threats to acquired values, in a subjective sense, the absence of fear that such values will be attacked." It follows that perhaps security can be best understood as the inverse of risk/threat. Because security takes its meaning from the absence of risk/threat, and because it is impossible to prove why something did not occur, attributing the sources and causes of security is problematic. We cannot know for sure why large-scale acts of terrorism have not occurred in the United States from September 11, 2001 to the time of this writing. We can make assumptions about the effectiveness of preventive measures and about terrorist capabilities and motivations, but we cannot prove why another act of the scale of 9/11 has not yet occurred. This becomes key when considering how to measure and track security from terrorism.

Accordingly, we think it is preferable methodologically for purposes of opinion survey research to conceive of security and measure it as the inverse of risk or threat. Because threats can be more discretely defined and specified, we hypothesize that respondents are better able to compartmentalize and separately assess threats of different types acting at different levels of analysis than they are to assess conceptual questions about the more difficult to specify and more variable concept of security. Thus we explore public perceptions of threats and risks, the inverse of which can be used more reliably to represent feelings of security.

To provide measures of perceived threats relating to terrorism, we ask participants to rate the threat of terrorism using eight inquiries, each of which is answered on a scale from 0 (no threat) to 10 (extreme threat). In the lead-in to this series, respondents are instructed when assessing threat to consider both the likelihood of terrorism and its potential consequences. The questions ask the threat of: (1) terrorism of all types throughout the world today; (2) terrorism of all types in the United States today; (3) nuclear terrorism in the United States today; (4) use of a dirty bomb in the United States today; (5) biological terrorism in the United States today; (6) chemical terrorism in the United States today; (7) suicide bombings in the United States today; and (8) overall threat of terrorism in the United States in the next ten years. (8) Weaverage equally weighted responses to the eight assessments, then reverse the scale to create a security index for which a scale value of 0 represents judgment that we are not at all secure from terrorism, and a value of 10 represents judgment that we are extremely secure from terrorism. Cronbach's alpha value for this scale is 0.944. (9)

Measuring Beliefs on Freedom of Speech

Speech is one of the broadest areas of civil liberties because it is multidimensional, including overt and symbolic speech, and overlaps other groups of civil liberties, such as freedom of religion, assembly, and association--the full exercise of which depends on free speech. To provide a broad index of such measures, we ask participants to respond to 14 statements about multiple dimensions of free speech using a scale from 1 (strongly disagree) to 7 (strongly agree). They include assertions about written, spoken, and demonstrative forms of expression as they relate to religion, assembly, association, teaching, immigration, classified information, the Internet, and advocating terrorism. (10) Equally weighted responses are averaged to form a composite speech index where a scale value of 1 represent the lowest level of support for free speech, and 7 represents the highest level of support. Cronbach's alpha for this scale is 0.804.

Measuring Beliefs on Due Process

Providing due process protections to those accused of terrorism is both controversial and an evolving area of law in the United States. Of particular relevance are two dimensions of due process: the jurisdiction where suspects are apprehended, and the combatant and citizenship status of suspects. Applying a similar methodology to that previously described, we ask participants to respond to eleven statements about various dimensions of due process as it applies to those suspected of supporting or conducting acts of terrorism. Included are U.S. citizens and noncitizens apprehended or captured inside and outside U.S. legal jurisdiction, issues of what kinds of courts are appropriate, and the use of torture. (11) Again, each response is on a scale from 1 (strongly disagree) to 7 (strongly agree). Responses are given equal weight and averaged to form a composite due process index where 1 reflects the lowest level of support for due process and 7 reflects the highest level of support. Cronbach's alpha for the due process index is 0.804.

Measuring Beliefs on Privacy

Our measures of privacy reflect respondents' tolerance for intrusive measures taken by government officials that are designed to enhance security from terrorism, and they include issues associated with collecting personal data, monitoring communications and behavior patterns, and conducting searches. Reactions to each of ten related policies affecting privacy are provided on a scale from 1 (strongly oppose) to 7 (strongly support). (12) Equally weighted responses are averaged and the scale is reversed to form a composite privacy index where 1 represents the lowest valuation of privacy protections, and 7 represents the highest valuation of privacy. Cronbach's alpha value for the privacy index is 0.880.

Estimating Public Trust

Trust in government has multiple dimensions. We employ one omnibus question about how much respondents trust the government in Washington to "do the right thing for the American people" and nine specific questions across each of the three branches of federal government about providing security from terrorism, protecting civil liberties, and balancing security and liberty today. (13) Equally weighted responses are averaged to provide a trust index having a Cronbach's alpha value of 0.903.

Assessing Political Culture

There are three other important dimensions of mass belief structures relating to world views that we hypothesize may figure importantly in understanding preferences for how freedom and security are balanced. Each derives from political culture theory (Douglas 1970; Douglas and Wildavsky 1982; Thompson 1997). By overlaying the degree to which individuals understand themselves to be incorporated into bounded units or social groups (termed "group") with the degree to which patterns of individuals' interactions are circumscribed by externally imposed prescriptions (termed "grid"), four primary cultural types (or solidarities) can be identified: individualists; hierarchists; egalitarians; and fatalists. Three of these cultural types seem particularly relevant to the relationship of freedom and security.

Hierarchists exhibit high group identity and binding prescriptions (high group, high grid). They place high value on order, security, procedures, clear lines of authority, and social stability, and are predisposed to trust officials in positions of authority (Ellis and Coyle 1994; Jenkins-Smith and Smith 1994). For many hierarchists, we hypothesize that security and social order can be expected to be valued over protections for civil liberties.

Egalitarians seek strong group identities and prefer minimal external prescriptions (high group, low grid). They fear concentrations of power and distrust experts and those in positions of authority. Egalitarians prefer equality of outcome, and distrust the "establishment" to ensure it (Ellis and Coyle 1994; Jenkins-Smith and Smith 1994). We hypothesize that for many egalitarians, civil liberties can be expected to be valued over security.

Individualists have little if any group identity and feel bounded by few structural prescriptions (low group, low grid). They prefer a libertarian society without many rules and regulations, and they feel little obligation to define themselves in terms of group memberships. Individualists prefer to transact their own terms for social relations through bidding and bargaining (Ellis and Coyle 1994; Jenkins-Smith and Smith 1994). We hypothesize that individualists will strongly value individual liberties, but also will value the security and order that allows them to pursue their own goals without threat of interference. As a group, they may be more divided than either hierarchs or egalitarians.

On a scale from strongly disagree (1) to strongly agree (7), we record responses to three assertions for each index designed to measure beliefs about hierarchism, egalitarianism, and individualism. (14) For each set, equally weighted responses to the three associated questions are averaged to form corresponding indices for the three cultural types. (15) For each, a scale value of 1 represents the lowest level of affinity for the measured cultural dimension, and 7 represents the highest respective valuation. Cronbach's alpha values for each index are as follows: hierarchism, 0.696; egalitarianism 0.735; and individualism, 0.679.

Assessing Preferences for Balancing Freedom and Security

As previously discussed, the relationship between freedom and security is not always nor necessarily zero-sum, but policy choices sometimes are framed in terms of trade-offs between prospective enhancements to security purchased (at least temporarily) at the cost of reductions or intrusions on various liberties. Because such trade-offs affect a preexisting equilibrium between freedom and security, it is useful to have a rough indication of the abstract balance preferred by respondents. We pose the following two inquiries to gain tentative insights into how participants think freedom and security should be balanced in general terms and how they perceive current policies to be affecting that relationship.

* Q48: For this question, assume that black marbles represent the level of emphasis placed on the security of Americans and white marbles represent the level of emphasis placed on liberties of Americans. How many of each color would you place in a total combined mix of 100 marbles?

* Q49: Again, using the marbles example where black marbles represent the level of emphasis placed on the security of Americans, and white marbles represent the level of emphasis placed on liberties of Americans, how many of each color do you think represents the way the U.S. government is balancing considerations of security and liberty today?

The difference between responses to the preferred or normative balance and the perceived current balance represents a measure of satisfaction/dissatisfaction with current balancing policies. Table 2 compares mean differences among our five groups.

Clearly, liberal Democrats exhibit the largest difference, with preferred emphasis on liberty being almost 20 percent higher than current policies (in 2007) were perceived to provide. Conversely, conservative Republicans judge current policies to be overemphasizing liberty by more than 7 percent above preferences. Our three middle groups align as expected between the extremes. Table 2 illustrates that not only do mean preferences for balancing liberty and security differ predictably by political beliefs, but cognitive assessments of the existing relationship also are differentiated.

Measuring Key Policy Preferences

We also ask participants to respond to a variety of statements reflecting alternative beliefs and policy preferences about the relationship of liberty and security in the United States today. Mean responses to selected questions used in our analyses are described below.

Comparing Central Tendencies

Before analyzing the data, it is useful to hypothesize about what policy differentiations we might expect political beliefs to provide. Taking cues both from relevant literature on belief system structures and from contemporary political debate, we expect liberal Democrats to judge the threat of terrorism lowest, and thus to score highest on our security index; to place the highest value or importance on protecting civil liberties, such as those measured by our speech, due process, and privacy indices; to exhibit the lowest levels of trust in government; and to be the most concerned that crucial civil liberties are being compromised by the struggle against terrorism. Conversely, we expect conservative Republicans to judge security from terrorism lowest, to assign the lowest values to protecting civil liberties, to exhibit the highest levels of trust in the current government, and to be least concerned about possible intrusions into civil liberties for the purposes of preventing terrorism. Our middle three groups should align monotonically between the two extremes.

Note the consistency in Table 3 with which mean views on the relationship between freedom and security are differentiated by political beliefs. For each measure, intergroup differences are in the expected direction and all means are significantly different. As expected, the largest differences in central tendencies are between the two groups demonstrating more highly integrated ideological and party affinities, but also note the consistency with which those who are liberal leaning, moderate and nonaligned, and conservative leaning orient themselves between the two more polarized groups. These marbles questions (Q48/Q49) are of particular note for differences in group preferences for a normative equilibrium between liberty and security versus group perceptions of how those competing values currently are being balanced. Liberal Democrats express a preference for weighting liberty over security by about 13 percent more than conservative Republicans. Conversely, perceptions of how that relationship currently is being balanced show about a 14 percent gap in the opposite direction between the two most polarized groups. In other words, liberal Democrats judge that liberties of Americans should receive stronger emphasis than was being provided by security policies in 2007, while conservative Republicans prefer less emphasis on protecting liberties than government was providing. Central tendencies among our three middle groups align predictably between the two end groups. On average, mean measures of how liberty and security should relate, and mean assessments of how they were being balanced differ substantively and predictably by political beliefs.

Modeling Underlying Causal Structures

While central tendencies are useful for gauging the directions and magnitudes of intergroup differences, they do not illuminate underlying structures. To gain better insight into how supporting beliefs are structured among liberal Democrats, nonaligned moderates, and conservative Republicans, we employ comparative causal modeling. For purposes of this analysis, we follow the literature on hierarchical belief structures detailed by Fisk and Taylor (1992); Herron and Jenkins-Smith (2006a, 2006b); Hurwitz and Peffley (1987); Hurwitz, Peffley, and Seligson (1993); Jenkins-Smith and Herron (2007); Jenkins-Smith, Mitchell, and Herron (2004); Peffley and Hurwitz (1985); and Sabatier and Jenkins-Smith (1993, 1999). We premise the causal models on three levels of hierarchically structured beliefs, assuming that demographic attributes and more general beliefs causally precede more specific beliefs. (16) Core beliefs consist of fundamental underlying normative dispositions that transcend specific policy issue areas. We employ four core belief measures: political ideology and the indices of hierarchism, egalitarianism and individualism previously described.

Domain beliefs reflect fundamental orientations and strategies that apply across a given policy domain. For our models, we employ domain specific measures provided by the following five indices previously described: (1) security index; (2) trust index; (3) freedom of speech index; (4) due process index; and (5) privacy index.

Context beliefs are measures reflecting policy contexts that may be relevant to how beliefs shape preferences in specific policy environments. In this model, the difference in preferred (normative) emphasis on liberty minus the perceived emphasis current policies place on liberty represents a predictor specific to the existing policy context representing satisfaction/dissatisfaction with balancing policies in mid-2007.

Policy preferences are beliefs that represent specific choices associated with a given issue area. For policy preferences we model responses to Q59 from Table 3 that force trade-offs between freedom and security. (17) Figure 1 shows results for all respondents. After describing key relationships in the overall model, we examine how causal connections change among three groups with different political orientations. In each of our causal models, we calculate sequential multivariate regressions and specify standardized coefficients among only those relationships that are statistically significant at the 95-percent confidence level. Relationships having a p-value greater than 0.05 are not shown. In the first stage, we employ previously described measures of core beliefs consisting of self-rated ideology plus three political culture indices (hierarchism, egalitarianism, and individualism) as independent variables in multiple regressions to explain variations in each of five measures of domain level beliefs, also previously described. Next we combine core and domain belief measures to predict our policy context belief represented by the difference between preferred normative emphasis on liberty and perceptions of the current emphasis being placed on liberty. The final stage of regression represents the full model in which core, domain, and context beliefs all are used to predict responses to the following statement (Q59): "In the future, I would like for the U.S. government to place more emphasis on protecting my civil liberties, even if doing so means I am slightly less secure from attacks by terrorists." Responses are recorded on a scale from 1 to 7 where 1 means strongly disagree and 7 means strongly agree. The direction and size of the standardized regression coefficients are interpreted as follows: a change of one standard deviation in the independent variable produces the fractional change of a standard deviation in the dependent variable represented by the standardized coefficient. For example, a standardized coefficient of 0.25 means that a change of one standard deviation in the independent variable results in a change of 0.25 standard deviations in the dependent variable. Because the coefficients are all standardized, they can be compared. Explanatory powers are shown as adjusted [R.sup.2] values. Solid lines represent first-order relationships between independent and dependent variables in adjacent echelons of the model. Dashed lines and dotted lines depict relationships extending beyond adjacent echelons.18

[FIGURE 1 OMITTED]

Notice that our first model, which includes all respondents, is consistent with our expectation that beliefs associated with relating liberty and security are systematically structured hierarchically, with core beliefs influencing or constraining domain beliefs, which in turn are causally linked to our policy context belief, and together, the three levels of beliefs explain approximately 47 percent of variation in our dependent variable. The causal paths in the model suggest the following.

Explaining Domain Beliefs (All Respondents)

Ideology: As ideological beliefs increase one standard deviation (more conservative), the security index decreases -0.15 standard deviations, trust in government increases 0.15 standard deviations, and support for free speech (-0.28), due process (-0.23) and privacy (-0.16) all decline.

Hierarchism. Similarly, an increase of one standard deviation in the hierarchism index causes a decrease of -0.25 standard deviations in perceptions of security, an increase of 0.20 in trust, and reductions in support of speech (-0.36), due process (-0.29), and privacy (-0.36).

Egalitarianism. As the egalitarianism index increases one standard deviation, support for speech (0.04), due process (0.14) and privacy (0.07) all increase, but egalitarianism is not systematically related to either perceptions of security or trust in government.

Individualism. Increasing individualism by one standard deviation causes trust to decline by -0.06 and support for free speech to increase by 0.13 standard deviations.

Explaining the Policy Context Belief (All Respondents)

Variation in the difference between preferred and perceived current emphasis on liberty (Q48 minus Q49), which can be conceived as latent discontent with the way liberty and security currently are being balanced, is explained as follows. As ideology, hierarchism, and trust in government increase, the difference between preferred and perceived emphasis on liberty decreases. As egalitarianism, individualism, assessments of security, and support for speech, due process, and privacy increase, the difference in preferred and perceived emphasis on liberty increases. Together, our core and domain beliefs explain about 30 percent of the variation in the difference between Q48 and Q49, which we consider a measure of latent policy discontent.

The Full Model (All Respondents)

In the final stage of this model, preference for more governmental emphasis on security from terrorism, even if it comes at some cost in liberties, increases systematically with increases in hierarchism and trust, and preference for that policy emphasis decreases systematically with individualism, feelings of increased security, support for speech, due process, and privacy, and decreasing perceptions of the difference between normative and preferred levels of emphasis on liberty (latent discontent). The full model explains about 47 percent of variation in the dependent variable. We conclude from this exploration that, among the full sample, the relationship between security and liberty is partially but importantly a function of hierarchically structured beliefs.

Effects of Differing Political Beliefs

Our final three models are designed to explore how these belief structures vary among subsamples holding different political beliefs. In Figure 2, we model the same dependent variable among participants identified as liberal Democrats (Table 1, matrix 1).

Our model of liberal Democrats exhibits directionally consistent relationships, but some independent variables drop out of the model. By restricting this group to those who identify strongly or somewhat with the Democratic Party, ideology is systematically predictive of only two of our domain beliefs (the speech and privacy indices). Individualism drops out of the model entirely. All the domain beliefs are predictive of the difference in our two marbles questions, and together with core beliefs, they explain almost 45 percent of variation in the policy context belief. The full model explains about 43 percent of the variation in preferences for greater emphasis on security among liberal Democrats. Next we turn to our group at the opposite end of the political spectrum by modeling conservative Republicans in Figure 3.

Among conservative Republicans, variation in ideology is relatively more influential than among liberal Democrats, exerting predictable causal effects on perceptions of security, trust in government, and support for due process. Note also that individualism does not drop out of the model, as was the case among those on the political left. The explanatory power for the difference between our two marbles questions (context belief) is substantially lower at only about 17 percent, but overall, the model explains about 41 percent of variation in the dependent variable, which is comparable to that among liberal Democrats. Our final model examines the same structures among political moderates who are not strongly aligned with either major political party (Figure 4).

[FIGURE 2 OMITTED]

Note that among those who are politically moderate and do not exhibit strong affinity for any political party (Table 1, matrix 5), the model is somewhat more elaborate than the two previous models depicting belief structures among liberal Democrats or conservative Republicans. This model more closely resembles our introductory model that includes all respondents (Figure 1). All of the significant causal relationships shown are directionally consistent with those in Figure 1, and the model explains about 25 percent of variation in differences between our two marbles questions (context belief), and about 43 percent of variation in the dependent variable. These results support findings by Goren (2004) that political sophistication (represented here by the strength and integration of political beliefs) does not substantially affect the absorption of domain-specific principles nor strengthen the impact those principles have on our measured policy preference. Said differently, those respondents who do not exhibit tightly structured political beliefs in which the cognitive and affective belief structures that are as elaborate and reliably predictive of policy preferences as those found among our groups on the political left and right with more highly integrated political beliefs.

[FIGURE 3 OMITTED]

Conclusions

We set out to explore with cross-sectional data whether political beliefs are causally linked to understandings of the relationship between freedom and security among ordinary members of the U.S. public. Our findings show that preferences for the preferred balance and assessments of the actual current balance between liberty and security are systematically influenced by political beliefs and supported by elaborate belief structures. Of those we surveyed, liberal Democrats, on average, prefer the normative balance to be weighted toward liberty over security, and they perceive the current balance to be weighted toward security over liberty. Conservative Republicans, on average, prefer the normative balance to be weighted toward security over liberty, and they perceive the current balance to be distributed roughly equally between liberty and security. These findings support those of Davis (2007), who reported similar effects of ideology and partisanship in trade-offs between civil liberties and security over the first three years following 9/11. With causal modeling, we show that structures of beliefs underlying normative preferences and perceptions of the existing balance are similarly coherent and robust among liberal Democrats, nonaligned moderates, and conservative Republicans who participated in our survey. These findings support Goren's (2004) domain specific theory of mass policy reasoning.

[FIGURE 4 OMITTED]

The influence of political values is not only apparent in evaluative beliefs about how freedom and security should be balanced, they also are systematically related to cognitive judgments about security from terrorism and the value of key civil liberties relative to the threat/security context. Thus, political beliefs are not only shaping perceptions of and preferences for the freedom-security relationship; they also are influencing cognition of reality. Assessments of security from terrorism vary systematically with political perspective. Liberal Democrats perceive security from terrorism to be greater than do conservative Republicans. Politically moderate and nonaligned respondents assess the reality of security to be between the two more politically integrated groups. This raises the issue for future research of better understanding how values may influence cognition.

The issue of trust in government and our policy context measures are of particular interest for future research as they relate to the hypothesized dynamic by which perceptions of risk may shift from the primarily external threats of terrorism to the internal threats from intrusive government. Differences between beliefs about a normative preferred balance of U.S. liberty and security and beliefs about the actual current balance represent latent public discontent, the effect of which in our models is directionally consistent with trust in government across varying political orientations. We expect that, together, these contextual factors provide early indications of increasing tension or potential political energy for resisting policies intended to prevent terrorism. In the aftermath of terrorist events, public sentiment may temporarily dampen normal ideological and partisan influences, create a rallying effect for government, and shift preferences toward the security side of the balance (Davis 2007). But as time passes without such attacks, latent discontent with increased emphasis on security grows (Matthew and Shambaugh 2005). If accompanied by a decline in trust in government, pressures increase for normative readjustment, and we expect emphasis would shift toward the liberty side of the balance. If such an equilibrating dynamic exists among the public for influencing the freedom-security balance, its functions will depend on structured beliefs about the relative values of liberty and security and cognition of their current and preferred relationships. Fully exploring these dynamics awaits the availability of longitudinal data, but our cross-sectional data help illuminate systematic belief structures that are needed for such equilibrating dynamics to function as we hypothesize.

As the struggle against terrorism proceeds, policy choices about how liberty and security are balanced should take into account substantial differences in both normative preferences for that balance and perceptions of how security policies are affecting that balance. Ordinary citizens of widely varying political orientations hold structured beliefs about liberty and security that militate toward a normative relationship, even though that ideal varies substantially across political orientations. When the perceived balance varies substantively from normative expectations, history suggests that public opinion may tolerate intrusions into civil liberties for purposes of enhanced security for temporary periods (when the perceptions of threat are most alarming), but a pendulum effect is likely to return in the direction of normative expectations when the threat recedes (such as the end of a war) or when the intrusions become sufficiently egregious and intolerable (such as with McCarthyism). The challenge for future research is to better understand the equilibrating dynamics underlying that pendulum effect.

Our findings using cross-sectional data point to potentially important causal linkages, but understanding those dynamics will require further research. Of particular interest is predicting when public concerns about external threats become subordinate to concerns about risks from intrusive government. Are there criteria by which policy makers can judge the acceptability of policy alternatives intended to reduce risks from terrorism, and are there reliable indications when such policies may no longer be acceptable? Given the events of 9/11, the contemporary struggle against nonstate radical jihadism, and the potential for future acts of terrorism in which weapons of mass destruction produce unacceptable losses, has the normative balance between liberty and security already been permanently altered? Further research into the dynamics between freedom and security may help provide important insights for both values and how policies affect their interaction.

Appendix

Sampling: The sample frame was provided by Survey Sampling International (SSI), which maintains an Internet panel of U.S. residents interested in participating in online research. The panel, titled SurveySpot, consists of volunteer members from many sources, including several thousand web properties, multiple online recruitment methods, and random digit dialing telephone recruitment. SurveySpot members are recruited exclusively using permission-based techniques. Unsolicited email is not employed. SurveySpot consists of approximately 2 million U.S. households with about 5 million household members. Only one member in each household can participate in the panel. SSI maintains a subpanel of approximately 400,000 members whose demographics are roughly proportional to U.S. national census characteristics. Our sample was randomly drawn from the 400,000-strong census-balanced subpanel.

Terrorism Index: We will begin by asking a series of questions about your perceptions of the threat of terrorism. For each, please consider both the likelihood of terrorism and its potential consequences. Each is answered on a scale from 0 to 10, where 0 means no threat and 10 means extreme threat.

Q4: Remembering to consider both the likelihood and potential consequences, how do you rate the overall threat of terrorism of all types throughout the world today.

Q5: Focusing more specifically on our own country, and considering both foreign and domestic sources of terrorism, how do you rate the threat of all kinds of terrorism in the United States today?

Q6: Narrowing our focus to the threat of nuclear terrorism, how do you rate the threat of terrorists creating a nuclear explosion in the United States today?

Q7: So-called dirty bombs are devices that use conventional explosives to scatter radioactive materials. How do you rate the threat of terrorists using a dirty bomb in the United States today?

Q8: Biological devices are used to spread biological agents, such as germs and viruses. How do you rate the threat of terrorists using a biological device in the United States today?

Q9: Chemical terrorism could result from terrorist attacks on U.S. chemical installations or by terrorists purposely dispensing dangerous chemical agents. How do you rate the threat of chemical terrorism in the United States today?

Q10: How do you rate the threat of suicide bombings by terrorists in the United States today?

Q11: Turning now to the future, how do you rate the overall threat of terrorism to the United States in the next ten years?

Freedom of Speech Index: Next we will present a series of statements regarding various forms of speech in the United States. Please respond to each statement on a scale from 1 to 7 where 1 means you strongly disagree and 7 means you strongly agree. (random order)

Q12: The right to free speech should be granted to everyone, even if they are unwilling to recognize the right to free speech by others.

Q13: Groups that express extreme political views, including hatred and violence, should be allowed to buy access to television, radio, and newspapers just like anyone else.

Q14: Even if they pay their own way, foreigners who dislike our government and criticize the American way of life should not be allowed to study in U.S. universities. (responses reversed)

Q15: On issues of religion, morals, and politics, teachers have the right to express their opinions in a public high school class, even if they go against the community's most precious values and beliefs.

Q16: When a media outlet reports classified information illegally obtained from the U.S. government, it is just doing its job of informing the American public.

Q17: The United States should block access to information in books and on the Internet that explains how to construct nuclear or biological devices that could be used by terrorists. (responses reversed)

Q18: A person who spits on or burns the American flag in the United States is behaving badly, but should not be punished for doing so by law.

Q19: If members of an extremist group in the United States who advocate terrorism pay for the use of a civic auditorium, they have every right to hold open meetings to advocate such activities in that public space.

Q20: Members of the Islamic community in the United States who want to march and demonstrate against U.S. actions in the Middle East should be allowed to conduct such activities if they acquire all necessary permits from local authorities.

Q21: The freedom of nonbelievers to criticize God and the Christian religion in the United States should be legally protected regardless of who might be offended.

Q22: The freedom of nonbelievers to criticize Allah and the Islamic religion in the United States should be legally protected regardless of who might be offended.

Q23: If someone advocates terrorism, but they do not actively participate in terrorist acts, they should be arrested and tried in a court of law, even if they are a religious leader or teacher. (responses reversed)

Q24: Government law enforcement agencies should never infiltrate or spy on religious groups, even if they are suspected of advocating or supporting terrorism.

Q25: If a particular religious sect or group is found to be advocating or promoting terrorism, that organization should be shut down by the government. (responses reversed)

Due Process Index: Our next series of statements relates to due process under the law. Again, please respond to each on a scale from 1 to 7, where 1 means strongly disagree and 7 means strongly agree. (random order)

Q26: When people who are not American citizens are captured outside the United States for suspected terrorist activities, they should be afforded all the legal rights provided by the U.S. Constitution and the American judicial system for U.S. citizens.

Q27: When people who are not American citizens are captured inside the United States for suspected terrorist activities, they should be afforded all the legal rights provided by the U.S. Constitution and the American judicial system for U.S. citizens.

Q28: When U.S. citizens are captured outside the United States for suspected terrorist activities, they should be afforded all the legal rights provided by the U.S. Constitution and the American judicial system.

Q29: When U.S. citizens are captured inside the United States for suspected terrorist activities, they should be afforded all the legal rights provided by the U.S. Constitution and the American judicial system.

Q30: Regardless of their nationality or where they were captured, it is wrong for the United States to hold suspected terrorists for as long as a year without formally charging them with their crimes.

Q31: Regardless of their nationality or where they were captured, suspected terrorists deserve the right to a fair trial.

Q32: If authorized by Congress and the president, military courts should be used to prosecute suspected terrorists who are not U.S. citizens. (responses reversed)

Q33: All systems of justice can sometimes make mistakes, but it is far worse to convict an innocent person that it is to let a guilty person go free.

Q35: Some terrorists are just too dangerous to be allowed to go free, even if not convicted of a specific crime. (responses reversed)

Q36: If the U.S. government decides that torture is the only way to prevent a terrorist group from using nuclear weapons, torture would be acceptable to me, because the potential good far outweighs the potential bad. (responses reversed)

Privacy Index: Using a scale from 1 to 7 where 1 means strongly oppose and 7 means strongly support, how would you feel about the following measures for preventing terrorism in the United States? (random order)

Q38: Requiring national identification cards for all U.S. citizens.

Q39: Restricting immigration into the United States to prevent terrorism.

Q40: Collecting personal information about you, such as your name, address, phone number, income, and social security number.

Q41: Collecting information about your behavior, such as where you shop, what you buy, what organizations, you belong to, and where you travel.

Q42: Conducting pat-down searches of your clothing and inspections of your belongings.

Q43: Taking photographic images of you without your knowledge.

Q44: Taking harmless electronic scans of your hands and face.

Q45: Taking a sample of your DNA.

Q46: Secretly monitoring phone conversations between people in the United States and people suspected of terrorist connections in other countries.

Q47: Secretly monitoring transfers of large sums of money between people in the United States and people suspected of terrorist connections in other countries.

Trust Index: Q63: On a scale from 0 to 10, where 0 means none of the time and 10 means all of the time, how much of the time do you trust the government in Washington to do what is right for the American people?

Lead-in: Please respond to each of the following questions on a scale from 0 to 10 where 0 means no trust and 10 means complete trust. (random order within each of three sets)

Q64: How much do you trust the current U.S. president to meet the threat of terrorism today?

Q65: How much do you trust the current U.S. Congress to meet the threat of terrorism today?

Q66: How much do you trust the current courts to meet the threat of terrorism today?

Q67: How much do you trust the current U.S. president to protect our civil liberties today?

Q68: How much do you trust the current U.S. Congress to protect our civil liberties today?

Q69: How much do you trust the current courts to protect our civil liberties today?

Q70: How much do you trust the current U.S. president to properly balance our security and our liberties today?

Q71: How much do you trust the current U.S. Congress to properly balance our security and our liberties today?

Q72: How much do you trust the current courts to properly balance our security and our liberties today?

Political Culture Indices: Please respond to each of the following statements using a scale from 1 to 7, where 1 means strongly disagree and 7 means strongly agree. (random order)

Hierarchism Index: (1 = strongly disagree--7 = strongly agree)

Q83: The best way to get ahead in life is to work hard and do what you are told to do.

Q86: Society is in trouble because people do not obey those in authority.

Q89: Society would be much better off if we imposed strict and swift punishment on those who break the rules.

Egalitarianism Index: (1 = strongly disagree--7 = strongly agree)

Q81: What society needs is a fairness revolution to make the distribution of goods more equal.

Q84: Society works best if power is shared equally.

Q87: It is our responsibility to reduce differences in income between the rich and the poor.

Individualism Index: (1 = strongly disagree--7 = strongly agree)

Q82: Even if some people are at a disadvantage, it is best for society to let people succeed or fail on their own.

Q85: Even the disadvantaged should have to make their own way in the world.

Q88: We are all better off when we compete as individuals.

References

ALMOND, GABRIEL A. 1956. "Public Opinion and National Security Policy." Public Opinion Quarterly 20 (2): 371-378.

ARIAN, ASHER, and MICHAL SHAMIR. 1983. "The Primarily Political Functions of the Left-Right Continuum." Comparative Politics 15 (2): 139-158.

BERRENS, ROBERT P., ALOK K. BOHARA, HANK JENKINS-SMITH, CAROL SILVA, and DAVID L. WEIMER. 2003. "The Advent of Internet Surveys for Political Research: A Comparison of Telephone and Internet Samples." Political Analysis 11 (1): 1-22.

BEST, SAMUEL J., BRIAN KRUEGER, CLARK HUBBARD, and ANDREW SMITH. 2001. "An Assessment of the Generalizability of Internet Surveys." Social Science Computer Review 19 (2): 131-145.

BIMBER, Bruce, 1998. "The Internet and Political Mobilization: Research Note on the 1996 Election Season." Social Science Computer Review 16 (4): 391-401.

BLALOCK, HUBERT M., Jr 1985. "Simple Recursive Models and Path Analysis." In Causal Models in the Social Sciences. 2d edition, edited by H. M. Blalock. New York: Aldine de Gruyter. 1-6.

BRINKLEY, ALAN. 2003. "A Familiar Story: Lessons from Past Assaults on Freedoms." In The War on Our Freedoms: Civil Liberties in an Age of Terrorism, edited by Richard Leone and Greg Anrig, Jr. New York: Public Affairs. 23-46.

BUZAN, BARRY. 1991. People States and Fear. 2d edition. Boulder, CO: Lynne Rienner Publishers.

CAMPBELL, ANGUS, PHILIP E. CONVERSE, WARREN E. MILLER, and DONALD E. STOKES. 1960. The American Voter. New York: Wiley.

CHUBB, JOHN E., MICHAEL G. HAGEN, and PAUL M. SNIDERMAN. 1991. "Ideological Reasoning." In Reasoning and Choice: Explorations in Political Psychology, edited by Paul M. Sniderman, Richard A. Brody, and Philip E. Tetlock. New York: Cambridge University Press. 141-163.

CONOVER, PAMELA JOHNSTON, and STANLEY FELDMAN. 1981. "The Origins and Meaning of Liberal/Conservative Self-Identifications." American Journal of Political Science 25 (4): 617-645.

CONVERSE, PHILIP E. 1964. "The Nature of Belief Systems in Mass Publics." In Ideology and Discontent, edited by David E. Apter. New York: Free Press. 206-261.

--. 1970. "Attitudes and Non-Attitudes: Continuation of a Dialogue." In The Quantitative Analysis of Social Problems, edited by Edward R. Tufte. Reading, MA: Addison-Wesley. 168-190.

--. 1987. "Changing Conceptions of Public Opinion in the Political Process." Public Opinion Quarterly 51 (Part 2: Supplement): S12-S24.

DAHL, ROBERT. 1972. Polyarchy: Participation and Opposition. New Haven, CT: Yale University Press.

DAVIS, DARREN W. 2007. Negative Liberty: Public Opinion and the Terrorist Attacks on America. New York: Russell Sage Foundation.

DELLI CARPINI, MICHAEL X., and SCOTT KEETER. 1996. What Americans Know About Politics and Why It Matters. New Haven, CT: Yale University Press.

DOUGLAS, MARY. 1970. Natural Symbols: Explorations in Cosmology. London: Barrie and Rockliff.

Douglas, Mary, and AaronWildavsky. 1982. Risk and Culture: An Essay on the Selection of Technical and Environmental Dangers. Berkeley: University of California Press.

ELLIS, RICHARD J., and DENNIS J. COYLE. 1994. "Introduction." In Politics, Policy and Culture, edited by Dennis J. Coyle and Richard J. Ellis. Boulder, CO: Westview Press. 1-14.

FELDMAN, STANLEY. 1988. "Structure and Consistency in Public Opinion: The Role of Core Beliefs and Values." American Journal of Political Science 32 (2): 416-440.

FISK, SUSAN T., and SHELLY E. TAYLOR. 1992. Social Cognition. New York: McGraw-Hill.

FREEDMAN, LAWRENCE. 1992. "The Concept of Security." In The Encyclopedia of Government and Politics, Vol. 2, edited by Mary E. Hawkesworth and Maurice Kogan. London: Routledge. 730-741.

GALLIE, WALTER B. 1962. "Essentially Contested Concepts." In The Importance of Language, edited by Max Black. Englewood Cliffs, NJ: Prentice Hall. 121-146.

GOREN, PAUL. 2004. "Political Sophistication and Policy Reasoning: A Reconsideration." American Journal of Political Science 48 (3): 462-478.

HAND, LEARNED. 1952. The Spirit of Liberty. New York: Alfred A. Knopf.

HERRON, KERRY G., and HANK C. JENKINS-SMITH. 2002. "U.S. Perceptions of Security in the Wake of the Cold War: Comparing Public and Elite Belief Systems." International Studies Quarterly 46 (4): 450-479.

--. 2006a. Critical Masses and Critical Choices: Evolving Public Opinion on Nuclear Weapons, Terrorism, and Security. Pittsburgh, PA: University of Pittsburgh Press.

--. 2006b. American Views on Nuclear Security and Terrorism: Comparing Phone and Internet Surveys: 2005. Sandia Report: SAND2006-0753P. Albuquerque, NM: Sandia National Laboratories.

--. 2007. Public Views on Multiple Dimensions of Security: Nuclear Weapons, Terrorism, Energy and the Environment: 2007. Sandia Report: SAND2007-7733. Albuquerque, NM: Sandia National Laboratories.

HUDDY, LEONIE, NADIA KHATIB, and TERESA CAPELOS. 2002. "The Polls--Trends: Reactions to the Terrorist Attacks of September 11, 2001." Public Opinion Quarterly 66 (3): 418-450.

HURWITZ, JON, and MARK PEFFLEY. 1987. "How Are Foreign Policy Attitudes Structured? A Hierarchical Model." American Political Science Review 81 (4): 1099-1120.

HURWITZ, JON, MARK PEFFLEY, and MITCHELL A. SELIGSON. 1993. "Foreign Policy Belief Systems in Comparative Perspective: The United States and Costa Rica." International Studies Quarterly 37 (3): 245-270.

JENKINS-SMITH, HANK C., and KERRY G. HERRON. 2005. "United States Public Response to Terrorism: Fault Lines or Bedrock?" Review of Policy Research 22 (5): 599-623.

--. 2007. American Views on Energy and Environmental Security: Comparing Phone and Internet Surveys: 2006. Sandia Report: SAND2006-7236P. Albuquerque, NM: Sandia National Laboratories.

JENKINS-SMITH, HANK C., NEIL J. MITCHELL, and KERRY G. HERRON. 2004. "Foreign and Domestic Policy Belief Structures in the U.S. and British Publics." Journal of Conflict Resolution 48 (3): 287-309.

JENKINS-SMITH, HANK C., and WALTER K. SMITH. 1994. "Ideology, Culture, and Risk Perception." In Politics Policy, and Culture, edited by Dennis J. Coyle and Richard J. Ellis. Boulder CO: Westview Press. 17-32.

LEVITIN, TERESA E., and WARREN E. MILLER. 1979. "Ideological Interpretations of Presidential Elections." American Political Science Review 73 (3): 751-771.

LEWIS, ANTHONY. 2003. "Security and Liberty: Preserving the Values of Freedom." In The War on Our Freedoms: Civil Liberties in An Age of Terrorism, edited by Richard C. Leone and Greg Anrig, Jr. New York: PublicAffairs. 47-73.

LIPPMANN, WALTER. 1922. Public Opinion. New York: Macmillan.

--. 1925. The Phantom Public. New York: Harcourt Brace.

--. 1955. Essays in Public Philosophy. Boston: Little, Brown.

LUPIA, ARTHUR, and MATTHEW D. MCCUBBINS. 1998. The Democratic Dilemma: Can Citizens Learn What They Need to Know? Cambridge: Cambridge University Press.

LUSKIN, ROBERT C. 1987. "Measuring Political Sophistication." American Journal of Political Science 31 (4): 856-899.

--. 1990. "Explaining Political Sophistication." Political Behavior 12 (4): 331361.

MARCUS, GEORGE E. 2002. The Sentimental Citizen: Emotion in Democratic Politics. University Park, PA: Pennsylvania State University Press.

MATTHEW, RICHARD, and GEORGE SHAMBAUGH. 2005. "The Pendulum Effect: Explaining Shifts in the Democratic Response to Terrorism." Analyses of Social Issues and Public Policy 5 (1): 223-233.

MUELLER, JOHN. 2006. Overblown: How Politicians and the Terrorism Industry Inflate National Security Threats, and Why We Believe Them. New York: Free Press.

NEIER, ARYEH. 2004. "Lost Liberties: Ashcroft and the Assault on Personal Freedom." In Civil Liberties vs. National Security in a Post-9/11 World, edited by Katherine B. Darmer, Robert M. Baird, and Stuart E. Rosenbaum. Amherst, NY: Prometheus Books. 31-41.

PAGE, BENJAMIN I., and ROBERT Y. SHAPIRO. 1992. The Rational Public: Fifty Years of Trends in Americans' Policy Preferences. Chicago: University of Chicago Press.

PEFFLEY, MARK A., and JON HURWITZ. 1985. "A Hierarchical Model of Attitude Constraint." American Journal of Political Science 29 (4): 871-890.

REHNQUIST, WILLIAM H. 1998. All the Laws But One: Civil Liberties in Wartime. New York: Random House.

ROTHSCHILD, EMMA. 1995. "What Is Security?" Daedalus 124 (3): 53-98.

SABATIER, PAUL A., and HANK C. JENKINS-SMITH. 1993. Policy Change and Learning: An Advocacy Coalition Approach. Boulder, CO: Westview Press.

--. 1999. "The Advocacy Coalition Framework: An Assessment." In Theories of the Policy Process, edited by Paul A. Sabatier. Boulder, CO: Westview Press. 117-166.

SNIDERMAN, PAUL M., RICHARD A. BRODY, and PHILIP E. TETLOCK. 1991. Reasoning and Choice: Explorations in Political Psychology. New York: Cambridge University Press.

SNIDERMAN, PAUL M., and PHILIP E. TETLOCK. 1986. "Interrelationships of Political Ideology and Public Opinion." In Political Psychology, edited by Margaret E. Hermann. San Francisco: Jossey-Bass. 62-96.

STONE, GEOFFREY R. 2003. "Civil Liberties in Wartime." Journal of Supreme Court History 28 (3): 215-251.

--. 2004. Perilous Times: Free Speech in Wartime from the Sedition Act of 1798 to the War on Terrorism. New York: W. W. Norton.

THOMPSON, MICHAEL. 1997. "Cultural Theory and Integrated Assessment." Environment Modeling and Assessment 2 (3): 139-150.

WITTKOPF, EUGENE R. 1983. "The Two Faces of Internationalism: Public Attitudes Toward American Foreign Policy in the 1970s and Beyond." Social Science Quarterly 64: 288-304.

--. 1987. "Elites and Masses: Another Look at Attitudes Toward America's World Role." International Studies Quarterly 31 (2): 131-159.

--. 1990. Faces of Internationalism: Public Opinion and American Foreign Policy. Durham, NC: Duke University Press.

--. 1994. "Faces of Internationalism in a Transitional Environment." Journal of Conflict Resolution 38 (3): 376-401.

WOLFERS, ARNOLD. 1952. "'National Security' as an Ambiguous Symbol." Political Science Quarterly 67 (4): 481-502.

ZALLER, JOHN R. 1992. The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press.

HANK C. JENKINS-SMITH KERRY G. HERRON

Center for Applied Social Research University of Oklahoma

(1) On the subject of judicial restraint in times of war, see Rehnquist (1998) and Lewis (2003, 68) who quotes Chief Justice Rehnquist from a speech in 2000 as follows: "While we would not want to subscribe to the full sweep of the Latin maxim inter arms silent leges--'in time of war the laws are silent'--perhaps we can accept the proposition that though the laws are not silent in wartime, they speak with a muted voice."

(2) For a cogent discussion of how risks of low probability and high consequence can be used to exaggerate threats, see Mueller (2006).

(3) For a more detailed contrast of traditional and revisionist assumptions about public capacities, see Herron and Jenkins-Smith (2006a, 1-19).

(4) How political sophistication is characterized and measured varies. Some emphasize the role of factual knowledge; others stress knowledge about political processes; others look to the consistency and connectedness with which views are held. One approximation of political sophistication is the degree of integration demonstrated between political ideology, party identification, and partisanship.

(5) For an in-depth discussion of the role of emotion in political reasoning, see Marcus (2002).

(6) The sample frame was provided by Survey Sampling International (SSI) which maintains an Internet panel of U.S. residents interested in participating in online research. The panel, titled SurveySpot, consists of volunteer members from many sources, including several thousand web properties, multiple online recruitment methods, and random digit dialing telephone recruitment. SurveySpot members are recruited exclusively using permission-based techniques. Unsolicited email is not employed. SurveySpot consists of approximately 2 million U.S. households with about 5 million household members. Only one member in each household can participate in the panel. SSI maintains a subpanel of approximately 400,000 members whose demographics are roughly proportional to U.S. national census characteristics. Our sample was randomly drawn from the 400,000 census balanced subpanel.

(7) Though this study is not based on a probabilistic sample from which extrapolations about the distributions of opinions can be made to the U.S. public at large, our emphasis in this analysis is on relationships among variables, and true probability samples are not always necessary to make valid inferences about relationships, especially when variables are based on "treatments" randomly applied to respondents (Berrens et al. 2003). As reported by other researchers, mechanisms underlying decision-making processes (as opposed to distributions of opinions) do not differ importantly between Internet users and the general population (Berrens et al. 2003; Best et al. 2001; Bimber 1998). In extensive paired phone/Internet surveys on nuclear security, terrorism, energy, and the environment, Herron and Jenkins-Smith (2006b), Jenkins-Smith and Herron (2007), and Herron and Jenkins-Smith (2007) find phone and Internet policy preferences to be directionally consistent, although distributional patterns and central tendencies can vary across survey modes. More importantly, this research supports earlier findings about relationships among key independent and dependent variables holding across survey modes.

(8) Each of the component questions is provided in the Appendix.

(9) Cronbach's alpha is a measure of internal reliability or consistency of items in a scale or index. It ranges from 0 (not at all reliable) to 1.0 (perfectly reliable) and indicates the degree to which items in an index are measuring the same thing.

(10) Each of the component statements is provided in the Appendix.

(11) Ibid.

(12) Ibid.

(13) Ibid.

(14) Ibid.

(15) A fourth cultural type--fatalists--are omitted because by definition, they believe they have little control over their lives. The balance of freedom and security is out of their hands.

(16) The assumption of causal direction is necessary for causal modeling, as described in Blalock (1985). These models test for causal linkages from demographic attributes, to general beliefs, to more specific policy beliefs, assuming that the beliefs are structured hierarchically. For purposes of this analysis, however, our interest is in the comparison of the model structures across the three groups, rather than the quantitative estimates of the path coefficients themselves. For that reason, the causal models--while overly simple in some respects--are appropriate for the analysis we pursue.

(17) Space does not allow modeling multiple dependent variables, but similar relationships (with differing explanatory powers) are found when other policy preferences shown in Table 3 are modeled.

(18) To simplify the model graphic, we do not show implications of demographic predispositions. In multiple regressions of all respondents in which age, gender, education, race/ethnicity, and household income are included as independent variables, demographics exert indirect influence through core, domain, and context belief measures. For example, increasing education is causally linked to more liberal ideology, perceptions of greater security from terrorism, and greater support for civil liberties. But only age is systematically and directly predictive of the dependent variable. As age increases one standard deviation, preference for more emphasis on security, even if it requires slightly restricting some civil liberties, increases 0.07 standard deviations (p <.0001).

Table 1. Cognitive and Affective Dimensions of Political Beliefs                                            Slightly Democrat (3)Freedom vs. Security      Strongly          + Independent (4)Values Survey 2007        Democrat (1)      + None (0)Web                       + Somewhat        + Slightlyn = 2,924                 Democrat (2)      Republican (5)Strongly Liberal (1)      Liberal           Liberals  + Liberal (2)             Democrats         Nonaligned                            n = 343           n = 123                            (11.7%)           (4.2%)                            Matrix 1          Matrix 2Slightly Liberal (3)      Moderate          Moderates  + Middle of Road (4)      Democrats         Nonaligned  + Slightly                n = 626           n = 680    Conservative (5)        (21.4%)           (23.3%)                            Matrix 4          Matrix 5Conservative (6)          Conservative      Conservatives  + Strongly                Democrats         Nonaligned  Conservative (7)          n = 46            n = 134                            (1.6%)            (4.6%)                            Matrix 7          Matrix 8Freedom vs. Security      SomewhatValues Survey 2007        Repub. (6)Web                       + Stronglyn = 2,924                 Republican (7)Strongly Liberal (1)      Liberal  + Liberal (2)             Republicans                            n = 18                            (0.6%)                            Matrix 3Slightly Liberal (3)      Moderate  + Middle of Road (4)      Republicans  + Slightly                n = 432    Conservative (5)        (14.8%)                            Matrix 6Conservative (6)          Conservative  + Strongly                Republicans  Conservative (7)          n = 522                            (17.9%)                            Matrix 9Table 2. Mean Emphasis on Liberty by Political Belief GroupingsEmphasis                        #1             #2 + 4on Liberty                      Liberal        Liberal(White Marbles)    All          Democrats      LeaningPreferred          49.02        57.95          49.21Current            46.09        38.03          43.56Difference         2.93         19.92          5.65p-value            <.0001       <.0001         <.0001Emphasis           #5           #6 + 8         #9on Liberty         Nonaligned   Conservative   Conservative(White Marbles)    Moderates    Leaning        RepublicansPreferred          49.98        46.07          44.66Current            46.45        48.63          52.00Difference         3.53         -2.56          -7.34p-value            .0002        .0068          <.0001Notes: 1. Differences in means among all parings for preferredemphasis on liberty are statistically significant with twoexceptions: differences between liberal leaning vs.  nonalignedmoderates, and differences between conservative leaning vs.conservative Republicans.2. Differences in means among all parings for current emphasison liberty are statistically significant.3.  Mean differences between preferred and current emphasis onliberty are statistically significant for all parings exceptliberal leaning vs. nonaligned moderates.Table 3. Comparing Issue Means                                          Liberal    Liberal/Dem.Question/Index                          Democrats         LeaningTerrorism Security Index (0 = not            4.00            3.26  at all secure; 10 = extremely  secure)Speech Index (1 = least                      4.26            3.54  supportive; 7 = most supportive)Due Process Index (1 = least                 4.90            4.25  supportive; 7 = most supportive)Privacy Index (1 = least                     4.37            3.84  supportive; 7 = most supportive)Trust Index (0 = no trust; 10 =              3.76            4.13  complete trust)Q48: Marbles 1: Preferred emphasis          57.95           49.21  on liberties (0-100)Q49: Marbles 2: Gov't. emphasis             38.03           43.56  on liberties today (0-100)Q51: I am concerned the U.S.                 5.31            4.43  government is unnecessarily  threatening our civil liberties  to meet the threat of terrorism  today. (1 = strongly disagree;  7 = strongly agree)Q52: I believe too much power is             5.72            4.76  being exercised by the U.S.  president in meeting the threat  of terrorism today. (1 =  strongly disagree; 7 = strongly  agree)Q54: If suicide bombings occur in            4.63            5.61  the United States, I would  support imposing much greater  restrictions on immigration into  our country of all types,  whether legal or illegal.  (1 = strongly disagree; 7 =  strongly agree)Q55: If a nuclear weapon or dirty            3.84            4.49  bomb is used in United States,  we will need to forfeit more of  our freedoms in order to regain  adequate security. (1 = strongly  disagree; 7 = strongly agree)Q59: In the future, I would like             3.29            4.36  for the U.S. government to place  more emphasis on ensuring my  security from terrorism, even if  doing so requires slightly  restricting some of my civil  liberties. (1 = strongly  disagree; 7 = strongly agree)Q60: In the future, I would like             4.65            3.81  for the U.S. government to place  more emphasis on protecting my  civil liberties, even if doing  so means I am slightly less  secure from attacks by  terrorists. (1 = strongly  disagree; 7 = strongly agree)                                       Moderate &    Conserv./Rep.Question/Index                         Nonaligned         LeaningTerrorism Security Index (0 = not            3.26            2.97  at all secure; 10 = extremely  secure)Speech Index (1 = least                      3.46            3.18  supportive; 7 = most supportive)Due Process Index (1 = least                 4.02            3.77  supportive; 7 = most supportive)Privacy Index (1 = least                     3.74            3.36  supportive; 7 = most supportive)Trust Index (0 = no trust; 10 =              4.14            5.07  complete trust)Q48: Marbles 1: Preferred emphasis          50.00           46.07  on liberties (0-100)Q49: Marbles 2: Gov't. emphasis             46.45           48.63  on liberties today (0-100)Q51: I am concerned the U.S.                 4.19            3.44  government is unnecessarily  threatening our civil liberties  to meet the threat of terrorism  today. (1 = strongly disagree;  7 = strongly agree)Q52: I believe too much power is             4.26            3.18  being exercised by the U.S.  president in meeting the threat  of terrorism today. (1 =  strongly disagree; 7 = strongly  agree)Q54: If suicide bombings occur in            5.75            6.04  the United States, I would  support imposing much greater  restrictions on immigration into  our country of all types,  whether legal or illegal.  (1 = strongly disagree; 7 =  strongly agree)Q55: If a nuclear weapon or dirty            4.52            4.82  bomb is used in United States,  we will need to forfeit more of  our freedoms in order to regain  adequate security. (1 = strongly  disagree; 7 = strongly agree)Q59: In the future, I would like             4.31            4.79  for the U.S. government to place  more emphasis on ensuring my  security from terrorism, even if  doing so requires slightly  restricting some of my civil  liberties. (1 = strongly  disagree; 7 = strongly agree)Q60: In the future, I would like             3.76            3.20  for the U.S. government to place  more emphasis on protecting my  civil liberties, even if doing  so means I am slightly less  secure from attacks by  terrorists. (1 = strongly  disagree; 7 = strongly agree)                                         Conserv.Question/Index                        RepublicansTerrorism Security Index (0 = not            2.68  at all secure; 10 = extremely  secure)Speech Index (1 = least                      2.96  supportive; 7 = most supportive)Due Process Index (1 = least                 3.34  supportive; 7 = most supportive)Privacy Index (1 = least                     2.96  supportive; 7 = most supportive)Trust Index (0 = no trust; 10 =              5.23  complete trust)Q48: Marbles 1: Preferred emphasis          44.66  on liberties (0-100)Q49: Marbles 2: Gov't. emphasis             52.00  on liberties today (0-100)Q51: I am concerned the U.S.                 2.60  government is unnecessarily  threatening our civil liberties  to meet the threat of terrorism  today. (1 = strongly disagree;  7 = strongly agree)Q52: I believe too much power is             2.22  being exercised by the U.S.  president in meeting the threat  of terrorism today. (1 =  strongly disagree; 7 = strongly  agree)Q54: If suicide bombings occur in            6.33  the United States, I would  support imposing much greater  restrictions on immigration into  our country of all types,  whether legal or illegal.  (1 = strongly disagree; 7 =  strongly agree)Q55: If a nuclear weapon or dirty            5.03  bomb is used in United States,  we will need to forfeit more of  our freedoms in order to regain  adequate security. (1 = strongly  disagree; 7 = strongly agree)Q59: In the future, I would like             5.19  for the U.S. government to place  more emphasis on ensuring my  security from terrorism, even if  doing so requires slightly  restricting some of my civil  liberties. (1 = strongly  disagree; 7 = strongly agree)Q60: In the future, I would like             2.77  for the U.S. government to place  more emphasis on protecting my  civil liberties, even if doing  so means I am slightly less  secure from attacks by  terrorists. (1 = strongly  disagree; 7 = strongly agree)Note: Differences in means between liberal Democrats andnonaligned moderates, between nonaligned moderates and conservativeRepublicans, and between liberal Democrats and conservativeRepublicans for each issue are statistically significant (p < .001).