Extreme Voices:

Interest Groups and the Misrepresentation of Issue Publics

Ryan L. Claassen

Department of Political Science

Kent State University

rclaasse@kent.edu

(corresponding author)

Stephen P. Nicholson

Program in Political Science

University of California, Merced

snicholson@merced.edu

Accepted for publication in Public Opinion Quarterly. An earlier version of this paper was

presented at the Annual Meeting of the Western Political Science Association, April 1-3, 2010,

San Francisco. We wish to thank Tom Hansford, Ben Highton, Suzanne Mettler, Chris Parker,

and Alexander Theodoridis for helpful comments and suggestions.

Abstract

Studies of issue publics suggest that widespread political ignorance does not matter

because those affected by specific issues are involved, well informed, and can meaningfully

shape policy in their policy area. However, research on civic participation raises important

questions about whether the opinions of the active are representative of the less active. To

examine whether meaningful differences in policy attitudes exist between the politically active

and inactive within issue publics, we compare the policy attitudes of interest group members to

nonmembers. Across ten interest groups we find uniformly consistent evidence of policy

distinctiveness among group members and show that party identification and ideology largely

account for the difference. We also find that the policy differences between members and

nonmembers vary according to the primary incentive offered by an interest group. Groups

primarily offering expressive benefits exhibit the greatest opinion differences within an issue

public whereas opinion differences are muted for groups emphasizing solidary or material

incentives. Lastly, we find evidence of attitude extremism among group members. Taken

together, our study suggests that the voices of non-active citizens are not well represented within

issue publics.

2

Converse’s (1964) landmark essay on mass belief systems is best remembered for

showing that public opinion is characterized by non-attitudes. In an otherwise pessimistic

account, however, he finds unusually coherent opinion within issue publics, “groups” of citizens

that pay special attention to an issue. Since Converse (1964), scholars have largely investigated

whether issue publics compensate for general political ignorance. Over time, evidence has

accumulated that mass political ignorance may not be as bad as it seems because those with the

most intense interest in an issue are well informed (e.g., Burns, Schlozman, and Verba 2001;

Holbrook et al. 2005; Hutchings 2001, 2003; Iyengar et al. 2008; Kim 2009; Krosnick 1990) and

politically involved (e.g. Anand and Kronsnick 2003; Campbell 2003; Highton 2004; Krosnick

and Telhami 1995; Price et. al 2006).

The optimistic conclusions about the role issue publics play in democratic society,

however, neglect the question of political voice inequalities (Schattschneider 1960; Schlozman,

Verba, and Brady 2012; Verba, Schlozman, and Brady 1995). According to Schlozman, Verba,

and Brady (2012), questions of political voice are of huge consequence for democratic equality if

the active and inactive hold distinct political preferences. As their research program

demonstrates, such differences do exist and they can be substantial. Voice inequalities matter for

democracy, “Because citizen activity functions not only to communicate information to public

officials about citizens’ interests, preferences, and needs but also to generate pressure on them to

respond…(Verba, Brady, and Schlozman 1995, 464).” Furthermore, since public officials often

look to interest groups to gauge opinion on issues (Herbst 1998), voice inequalities within an

issue public may distort the policy making process.

We advance research on civic participation and public opinion by investigating whether

opinion differences exist between interest group members and nonmembers, the active and

3

inactive denizens of issue publics. To explore whether voice inequalities exist within issue

publics, we identify groups of people that are affected by a given policy using a key

demographic characteristic and distinguish those that formally belong to interest groups from

those that do not. Doing so enables us to compare the opinions of individuals who are active in

interest groups to those of individuals in the relevant latent issue publics. We note that this

research design is unique. While it is not unusual to use a demographic characteristic to identify

the contours of an issue public, we could find no prior study that explicitly investigates the role

of interest group membership within an issue public.

Using surveys with large national samples, we find consistent evidence of policy

distinctiveness among interest group members across ten policy areas. We explored the origins

of the difference between group and non-group members and found that party identification and

ideology explain much of the difference. Compared to the inactive members of their issue

publics, the opinions of interest group members are more polarized because groups attract

likeminded partisans and ideologues. We also found systematic differences in opinion

depending on whether the group primarily offers expressive, solidary, or material incentives.

Although we do not claim to have a representative sample of groups offering the different types

of incentives, we found that opinion differences between group members and their issue public

are largest for expressive groups. Lastly, members of expressive groups tend to express uniquely

extreme views relative to both their issue public and to members of other types of groups—even

after controlling for partisan and ideological self-selection. We conclude by assessing what these

results mean for policy makers.

4

The Role of Issue Publics In Democratic Society

Although Converse’s (1964) famous essay on mass belief systems is best known for

demonstrating that the public was largely innocent of ideology, he found that many people

belonged to “issue publics,” groupings of individuals who held crystallized attitudes on issues.

According to Converse (1964, 245), issue publics are the only means by which the public has

anything meaningful to say about an issue. “Members” of issue publics are typically better

informed about their issue (see Delli Carpini and Keeter 1995; Krosnick 1990; Hutchings 2001,

2003; Holbrook et al. 2005; Price et al. 2006; Iyengar et al. 2008; Kim 2009) and are more likely

to vote for candidates (Anand and Kronsnick 2003; Highton 2004; Hutchings 2003; Krosnick

1990) and partake in other forms of participation on behalf of their issue (Campbell 2003;

Krosnick and Telhami 1995; Price et al. 2006). Absent from research on issue publics, however,

is the question of whether active “members,” those who belong to a relevant interest group, hold

the same views as inactive “members,” those who do not belong to a relevant interest group.

Motivating our inquiry into voice inequalities within issue publics is research on civic

participation showing that politically active citizens hold different attitudes than the politically

inactive (Verba, Schlozman, and Brady 1995; Schlozman, Verba, and Brady 2012). Although

they do not compare interest group members to nonmembers as we do, Schlozman, Verba, and

Brady (2012, 266) suggest that expressions of preference through organized interests are the

most likely to promote political voice inequalities. Research on interest groups also suggests that

the preferences and concerns of group members will be different, and likely more extreme, than

nonmembers (e.g., Schattschneider 1960). Sabatier and McLaughlin’s (1990) commitment

theory proposes that more politically active individuals are likely to have more extreme

ideologies. Although their analysis focuses on differences between group leaders and group

5

members, the logic of commitment theory applies equally well to differences between members

and potential members: group members will be more extreme than potential members. Taken

together, we expect that group members, those who are politically active, will express distinct,

and more extreme, attitudes from potential members.

It is also likely that meaningful opinion differences between group members and

nonmembers within an issue public will vary according to the primary incentive groups use to

induce people to join. Whereas Olson identified material selective incentives as the key to

inducing individuals to join a group pursuing a collective goal, subsequent scholars have found

that expressive or solidary incentives can also play an important role in the membership calculus

of individuals (Berry 1977; Clark and Wilson 1961; King and Walker 1992; Moe 1980;

Salisbury 1969). Building on this research, we expect that one of the keys to understanding

whether interest group members will be distinctive in their policy opinions is the primacy of

expressive benefits. For material or solidary incentives, we expect to find minimal opinion

differences between members and potential members. In contrast, for groups in which

expressive benefits are central, we expect members to be distinctive and more extreme.

Why does it matter whether interest group members hold distinct attitudes from non-

members, the inactive denizens of issue publics? As Schlozman, Verba, and Brady (2012, 6)

remark, “government policy is likely to reflect more fully the preferences and needs of the active

part of the public.” These differences are crucially important when considering political voice

inequalities because interest groups represent the voices of issues publics before public officials.

Interest groups not only provide members with opportunities to participate directly (e.g. letter

writing campaigns) they also pool resources in order to provide members with indirect voice

(e.g. professional lobbying efforts). For many public officials, interest groups are de facto

6

official spokespersons for issues. Herbst’s (1998) research, for example, shows that legislators

look to interest groups as an indicator of opinion on an issue of importance to relevant segments

of society. Thus, if interest group members hold opinions different from issue publics, interest

groups introduce bias into the policy process by misinforming public officials about the nature of

opinion among an issue public.

Data and Methods

Our study takes advantage of two large samples—samples that are large enough to obtain

sizeable numbers of members of interest groups—to compare the issue preferences of members

of 10 interest groups to those of the associated issue publics. The bulk of our analyses involve

the 2006 Cooperative Congressional Election Survey (CCES), a nationally representative sample

of 36,500 individuals.1 For one of our groups, the NRA, we turned to the 2004 National

Annenberg Election Survey (NAES) because the CCES did not include a question about a gun

related issue. 2 1 The sample sizes in our analyses are somewhat smaller due to missing data and some items that

were only asked of split samples.

2 The CCES sample was constructed using sample matching, a technique wherein a list of all

U.S. consumers is used to generate a set of demographic characteristics that are matched from a

pool of opt-in respondents to closely resemble the consumer data. The data were collected over

a three-month period from September to November 2006. The response rate was 93.1 percent

(Vavreck and Rivers 2008). The National Annenberg Election Survey (NAES) consists of phone

interviews of U.S. residents using RDD samples (Romer et al. 2006). The data are weighted to

adjust for the number of adults and phone lines in the household and to match census

distributions of race, ethnicity, age, sex, and education. The overall cooperation rate was 54%

7

The CCES is ideal for our purposes because it queried respondents about memberships in

a variety of interest groups, which is uncommon in surveys. We selected 10 of 12 possible

groups for our study. The key in selecting groups was the availability of a policy question

important to each group. For example, we selected the AARP because of the availability of a

companion question about privatizing social security (a topic of national policy debate on which

the AARP took a public position—opposition). In addition to the AARP, the other nine groups

and issues include: 1) union membership and the minimum wage; 2) the Sierra Club and global

warming; 3) NARAL and abortion; 4) National Right to Life and abortion; 5) the Christian

Coalition and gay marriage; 6) the VFW and the Iraq war; 7) the American Legion and the Iraq

war; 8) the Parent Teacher Association/Organization and a question about the importance of

education and; 9) the NRA and gun control.3

and the overall response rate was 22-25%. The response rate is “given as a range to account for

the different formulas that can be used to calculate the total number of households eligible for a

survey” (2006, 21). They define response rate as “the percentage of telephone numbers released

into the field that resulted in a completed interview” (2006, 21). They define cooperation rate as

“the percentage of households that agreed to participated in the survey once they had been

successfully reached by an interviewer” (2006, 21).

3 Each opinion item is coded so that high values are the liberal response (e.g. increase gun

control) and standardized to range from 0 to 1. For each item, “Don’t know,” “other,” or “not

sure” responses were placed in the middle category. We also note that the results are nearly

identical if “Don’t know,” “other,” or “not sure” responses are coded missing (see Table 2 of the

online appendix).

8

We also identified the relevant issue public associated with each interest group. For

several groups, the requirements are obvious: the NRA and gun ownership, the AARP and age

over 50, the PTA and having children, union membership and employment, the VFW/American

Legion and military service (in all cases we restrict membership and key requirements to those of

the respondent, not the household). For the remaining groups, we identified a characteristic that

was both theoretically and empirically related to membership. For example, ideology is strongly

correlated with membership in the Sierra Club, NARAL, National Right to Life and the Christian

Coalition. Fewer than 6% of Sierra Club members and less than 1% of NARAL members are

conservatives (the most conservative two categories on a five point scale). Likewise, fewer than

6% of National Right to Life members and about 3% of Christian Coalition members identify as

liberal. Ideology performs reasonably well as a sufficient, but not a necessary condition for

membership in these groups (also see Hutchings 2001). However, we also identify potential

members of National Right to Life and the Christian Coalition without using ideology (we were

unable to find alternatives for the Sierra Club and NARAL). For National Right to Life, we use

“born again Christian” or Catholic identity to identify potential members and for the Christian

Coalition we use frequent attendance at religious services to identify potential members.4 In the

absence of better alternatives for NARAL and the Sierra Club, the fact that the results for 4 Shared characteristics are coded as dummy variables. In most cases the coding is described in

the second column of Table 1. However, several require additional comment. Liberals, for

example, include only those in the most liberal two categories from a five category variable.

Conservatives include only those in the most conservative two categories from the same item.

Frequent attenders include those in the most frequent two categories and those who identify as

“born again Christians” and Catholic are coded 1 in the first National Right to Life row.

9

National Right to Life and the Christian Coalition are nearly identical whether ideology is used

or not increases our confidence that ideology is a reasonable proxy for defining some issue

publics.

Finally, we categorized groups according to the primary benefit each provides. Although

groups provide a variety of benefits, in most cases a scholarly consensus exists regarding the

primary benefit. Though unions provide primarily material benefits, they are also unique in their

ability to coerce potential members into joining. So called “union shop” rules enable unions to

force all members of a collective bargaining unit to join once 50% plus 1 have voted to join by

secret ballot (Wagner Act of 1935 as amended by the Taft-Hartley Act of 1947).5 Therefore we

treat unions as a special type of material benefit group—one with a coercive “incentive.”

Material benefits are the primary incentive offered to potential AARP members (Dipak,

Hofstetterb, Buss 1997). The single issue groups, NARAL and National Right to Life, are

primarily expressive (Skinner 2007, 27). In a similar vein, Layman notes that groups such as the

Christian Coalition “exist for clearly political purposes” (2001, 275). The Sierra Club is often

mentioned as a group that attracts members by offering Ansel Adams prints, but recent

scholarship describes membership as an “expressive statement without incurring other

organizational obligations” (Bosso 2003, 409). Although earlier accounts of the NRA

categorized it as primarily emphasizing material benefits (Weil and Hemenway 1993),

contemporary accounts have detailed its shift to emphasizing expressive incentives (Patterson

and Singer 2002; Skinner 2007). Lastly, Mettler (2011) describes the VFW and the American

Legion as primarily solidary groups and Clark and Wilson (1961) also classify the PTA as a

5 These rules have been modified in “right to work” states.

10

solidary group. For the six groups we study in common (AARP, Unions, NARAL, National

Right to Life, NRA, and Sierra Club), our categorizations are identical to Skinner (2007, 27).

In sum, we restrict our comparisons to individuals with a characteristic that is relevant to

the interest group’s policy area—that is we attempt to compare members and non-group

members within each issue public. The defining characteristic we select not only links the

individual to an issue public, it links the individual to the side of the issue public represented by

the interest group in question. In so doing, our research design attempts to identify the portion of

the issue public represented by each group so that we can draw inferences about how well each

interest group represents the wider issue public.

Results

We rescaled each variable to range from zero to one to provide a graphic representation

of the data. Figure 1 depicts the opinion of group members and potential group members, our

comparison of active to inactive issue public “members,” respectively. The bars on the left side

represent each group whereas the middle bars represent the latent group or non-active issue

public “members.” In the bars on the right side, we include public opinion (excluding the issue

public and interest group respondents) on each issue to provide a sense of whether, and how,

interest groups and issue publics differ from the general public. On most issues, interest group

members appear to take distinctive positions in comparison to non-active “members” of the issue

public. Not surprisingly, interest groups members also appear to hold distinctive attitudes

compared to the public.

<Figure 1 About Here>

Figure 2 depicts the same data from Figure 1 reconfigured to examine whether interest

group opinion introduces distortion into the policy process. As mentioned, Herbst’s (1998)

11

research suggests that legislators look to interest groups as an indicator of opinion on an issue of

importance to an issue public, so examining whether groups have the same or different opinions

as their respective issue publics helps us understand whether groups distort the policy process.

Since the policy opinion questions are standardized to range from 0 to 1, we subtract .5 from

each variable to provide a neutral midpoint.6 In so doing, Figure 2 illustrates whether an interest

group opinion is on the same or opposite side of its issue public and how close or far apart group

opinion is from the opinion of the relevant issue public. As seen in Figure 2, interest groups and

issue publics are on opposite sides for six of the ten issues. On the four issues for which opinion

is on the same side, the opinions of group members are always more extreme. Although we do

not provide statistical controls in Figure 2 (or Figure 1 for that matter), they are irrelevant here

since a hypothetical legislator is unlikely to apply controls when listening to group members.

<Figure 2 About Here>

In Table 1, we test whether differences between the opinions of group members and

potential members within the issue public are statistically significant with, and without, statistical

controls. The estimates in the “Restricted Model” column are regression coefficients for

membership dummy variables from bivariate regressions of opinion on group membership in

samples limited to issue public members only. For example, in the NRA rows, the samples

include only gun owners and the coefficients reported in the “Restricted Model” column capture

opinion differences between NRA members and non-members within the issue public (e.g. gun

owners). In this way the estimates assess whether the differences between interest group

members and their respective issue publics shown in Figure 1 and Figure 2 are statistically 6 The PTA is excluded from Figure 2 because the middle of a “most important issue” item is not

a neutral point (.5 would indicate half of the sample feel education is the most important issue).

12

significant. The data speak loudly in the affirmative. The coefficients reported in the

“Restricted Model” column of Table 1 show that the opinions of interest group members are

significantly different than the opinions of non-members within each of the thirteen comparisons

we make for ten interest groups. Are these interest groups representative of their respective issue

publics? The first column of results in Table 1 raises a red flag for those concerned with equality

and voice within issue publics. Within each of the issue publics we investigate, those with

interest group representation express significantly different views than those without.

<Table 1 About Here>

In the next two columns of Table 1 we identify that portion of member distinctiveness

that is an artifact of other factors associated with both the opinions individuals express and

whether individuals join groups. Multiple regression will “purge” the effect of membership by

controlling for covariation in opinion that is shared by membership and other factors associated

with joining, such as SES.7 For example, if a portion of the effect of membership is actually

because members tend to be wealthy, and wealth shapes opinion, then controlling for income 7 We use the “KHB” command in Stata (see Kohler, Karlson, and Holm 2011) to make the

multiple regression estimates comparable to the bivariate estimates. For non-linear estimators

(such as the ordered-logit estimates we report here), part of the difference in the effect of a

variable evaluated in two nested models is due to the confounding effects of the other variables

and part of the difference is a methodological artifact (Karlson, Holm, and Breen 2012; Karlson

and Holm 2011; and Kohler, Karlson, and Holm 2011). A method devised by Karlson, Holm,

and Breen (the KHB method) eliminates the methodological artifact and ensures that the

difference in nested model estimates is due entirely to the confounding effect of the other

variables.

13

will eliminate the portion of the effect of membership that is actually due to wealth (proxied by

income). What remains can be thought of as the independent effect of membership.

We have attempted to model a wide range of potential individual differences to ensure

that the members and non-members we compare are statistically similar on many dimensions.

Toward that end we control for political awareness (0-1),8 party identification (-3 to +3),

ideology (0-4), age (in years), income (CCES 0-14, Annenberg 0-8), education (CCES 0-5,

Annenberg 0-8), sex, marital status, occupation (dummy variables for retired, homemaker,

unemployed), region (dummy variables for south, east, west), race, Hispanic ethnicity,

attendance at religious services (CCES 0-3, Annenberg 0-4), religious identity (dummy variables

for Catholic, Protestant, Jewish), and military service.9

We glean several noteworthy patterns from the last two columns of Table 1. First, in

absolute terms, the independent effects of membership are uniformly smaller in the “Full Model”

column compared to the “Restricted Model” column. In fact, the differences in the effects of

membership reported in the last column are uniformly statistically significant. As expected, a

significant portion of member distinctiveness can be attributed to opinion variation caused by

members’ other distinctive characteristics. Also noteworthy in the “Restricted Model” column is

8 See the online Question and Coding Appendix for a detailed description of scale construction.

9 We used Stata’s multivariate imputation command, “impute,” to compute observations for

missing cases for each control variable using the information available from all other control

variables. The results are nearly identical when other methods of dealing with missing data, such

as mode/mean substitution or listwise deletion, are used.

14

the fact that, even after eliminating the significant confounding effects of a host of demographic

and attitudinal factors, there remain substantively and statistically significant differences

between members and non-members for most issue publics. Only members of the VFW and

American Legion (two of our three Solidary groups), express statistically identical opinions

when compared to otherwise similar individuals who have also served in the military.

Table 2 presents changes in predicted probabilities associated with group membership

from the “Full Model” depicted in Table 1 with control variables set to their means. For liberal

groups we assess the change in the likelihood of expressing the most liberal opinion associated

with group membership and for conservative groups we do the same but instead examine

conservative opinion. As expected, the changes in predicted probabilities associated with

membership in solidary groups are smaller than most of the others. The probability changes

associated with membership in unions and the AARP are also small. Again, it makes sense that

a coercive group would be less distinctive since “closed shop” rules circumvent self-selection

processes (also note that ideological and partisan self-selection are already controlled

statistically). Likewise, material groups appear to be less distinctive than expressive groups.

Once again, it makes sense that near universal desire for material benefits within certain issue

publics would mitigate self-selection processes that lead to distinctive opinions.

<Table 2 About Here>

Why Are Group Members Different?

Having noted that a variety of characteristics and attitudes account for many opinion

differences between interest group members and non-members—and given the potential

distortion these differences may introduce in the policy process—we investigate the confounding

effects of these variables to understand why group members and nonmembers from the same

15

issue public are different. To help understand what is causing what, on substantive grounds we

can probably rule out that group membership caused any of our control variables. For example,

joining a group clearly does not cause one’s age. If the reverse were true (if membership caused

the control variables), then the addition of control variables would reveal the indirect effects of

group membership through the control variables. Theoretically the difference is important

because indirect effects would be effects properly attributed to membership, notwithstanding

manifestation through a mediating variable. Confounds, on the other hand, show the effect of

membership to be spurious. Specifically controlling for age might confound the effect of

membership if age is associated with both membership and opinion. Thus one of the reasons

group members might appear to be distinctive is because they tend to be composed of older

individuals who tend to express different opinions than younger individuals. If both membership

and opinion were function of age, then a portion of the difference in the opinions of members

and non-members would be due to age differences, not membership per se. In general then, the

reductions in the effects of membership associated with the statistical controls in the “Full

Models” reported in Table 1 are instances of spurious distinctiveness.

We begin in Table 3 by assessing whether group membership is a function of each of six

confounding variables.10 Table 3 provides the results of regressing group membership on each

of the six column variables. Clearly most groups are stratified on most of these six

characteristics and attitudes. However, where an attitude or a characteristic is not associated

with group membership, it will not be a significant confound. Here we highlight that the two 10 There are 23 control variables and space does not permit a detailed assessment of each. The

six variables selected for detailed analysis were selected because, in nearly every case, they were

the largest and most robust of the 23 confounding effects.

16

variables that affect group membership most consistently are ideology and party identification—

we do so because these will also prove to be the two attitudes that matter most when it comes to

understanding why group members express distinctive opinion.

<Table 3 About Here>

Having demonstrated the confounding potential of six characteristics and attitudes, we

obtain the actual confounding effect of each.11 Each estimate in Table 4 describes the

independent effect of the column variable on the “Difference” estimate reported in the last

column of Table 1. For example, in the first row of Table 4, controlling for ideology was a

significant part of the difference between the effect of union membership in the restricted model

compared to the full model. Though we report only the six most important, the total of all 23 of

the decomposed effects, for a given group, would be equal to the difference reported in Table 1.

So for union membership, .12 of the total .27 difference reported in Table 1 is due to ideology.12

Put another way, the effect of union membership in the restricted model in Table 1 would be .12

smaller (e.g. union members would appear less distinctive) if a control for ideology were added.

<Table 4 About Here>

Taken together, the results from Table 3 and Table 4 indicate that ideology and party

identification are two consistent and important confounds associated with interest group

distinctiveness. All of the groups are stratified by ideology. Even when the issue public is

restricted to liberals, as in the case of the Sierra Club and NARAL, how liberal one is still 11 We again use the KHB method.

12 Note, however, that controls can have opposite confounding effects and so it is also possible

for a single variable to have a larger confounding effect, in absolute terms, than the total

difference reported in Table 1.

17

predicts significant variation in the likelihood of membership. Likewise every group, except for

the Sierra Club is stratified by party identification. Stratification on these two attitudinal

dimensions produces significant confounding effects for each group except for the PTA and, of

course, party identification does not produce a significant confound for the Sierra Club (since it

is not associated with membership). Naturally, the size of the coefficients are sensitive to the

range of the variables, but only age covers a substantially different range than the other control

variables, and so the large and robust coefficients in the ideology and the party identification

columns are suggestive of their leading roles in producing distinctive opinion among group

members.

Also, the confounding effects in the ideology and party identification columns uniformly

reinforce the effects of group membership for each group. For conservative groups, more

conservative and more Republican members make their opinions appear more conservative on

the group issue than they would otherwise. Likewise for liberal groups. This is in contrast to the

mixed confounding effects of several of the other control variables. Income, for example,

actually makes union members and AARP members appear more conservative than they would

otherwise—less distinctive. Income also makes NRA members appear more liberal than they

would otherwise—again, less distinctive. Although a portion of interest group members’

opinions is different because of their higher SES (in most cases), they would actually be more,

rather than less, distinctive if SES were not correlated with joining. Certainly this is a different

pattern of income related distortion than the type identified by Verba, Brady and Schlozman

(1995) and Schlozman, Verba and Brady (2012). Yet, on balance the confounds contribute to

how distinctive members are—which is again suggestive of the larger role played by variables

that reinforce opinion distortion, such as ideology and party identification.

18

To be clear members of every group we study are distinctive (first column, Table1). But,

many of the groups we study would appear much less distinctive if not for ideological and

partisan self-selection into likeminded interest groups. In fact, if one adds the ideology and Party

ID effects in Table 4, their sum is very similar to the total coefficient reduction reported in the

last column of Table 1 for Union, AARP, VFW, American Legion, National Right to Life, and

Christian Coalition members. Clearly party identification and ideology have emerged as the

major sources of spurious distinctiveness. More importantly, ideology and party identification

contribute to a specific type of opinion distinctiveness—extremism. Self-selection into interest

groups by likeminded partisans and ideologues pushes members’ opinions farther toward an

ideological pole than would otherwise be the case (the reinforcing effect discussed above).

Previous scholars have noted that interest group members tend to express extreme attitudes

(Schattschneider 1960; Sabatier and McLaughlin’s 1990), here we identify an important

mechanism behind the extreme attitudes of interest group members.

Extremism and Group Membership

As a consequence of partisan and ideological self-selection, group members express more

extreme opinions that would otherwise be the case. For most coercive, material, and solidary

groups there is very little distinctiveness left once partisan and ideological self-selection are

accounted for. Otherwise similar members of the VFW and the American Legion are not

statistically distinctive in the second column of Table 1. And substantively, the non-spurious

distinctiveness of Union members, AARP members, and PTA members appears to be mostly

trivial in Table 2—even if technically robust in Table 1. The distinctiveness that remains once

we add our statistical controls appears to be concentrated in the expressive groups. More

importantly, the type of distinctiveness in the expressive groups is of a kind with that of the other

19

groups—extremism—even if it is less a function of the control variables. A glance at Figure 2

reveals that group members express opinions that are farther from the neutral point in every case

except for the NRA. And even the NRA could be characterized as extremely conservative when

compared to non-member, gun owners. Clearly partisan and ideological self-selection explain a

good deal of extremism for members of coercive, material, and solidary groups. Could it be that

the primary incentive associated with expressive groups—expression of political views—

produces extremism beyond that produced by partisan and ideological self-selection in the other

groups?

In order to conduct a formal comparison of opinion extremism across group types we

constructed a general attitudinal extremism scale using respondents’ views on all seven of the

policy questions in the CCES.13 Because liberal positions are associated with high values and

each opinion item is scaled to range from 0 to 1, the range of the sum is 0 to 7—with 0 being the

most extremely liberal overall position and 7 being the most extremely conservative overall

position. In order to create a measure of distance from the midpoint, we zero-center the sum by

subtracting 3.5 and then take the absolute value of the zero-centered sum. Thus the final

measure ranges from 0 to 3.5—with 0 being the most moderate overall position and 3.5 being the

most extreme overall position. We regress this measure of general opinion extremism on a set of

dummy variables measuring the type of incentive offered by the group and our set of control

variables. For this regression we limited the sample to members of only a single group so that

the dummy variables represent a set of mutually exclusive and exhaustive group types. Solidary

13 We use the six group questions and a question about affirmative action. Whether education is

the most important issue is excluded because there is no neutral point at the center of the scale.

20

is the omitted category in the regression so the coefficients test for significant differences

between members of the indicated type of group and members of solidary groups.

The coefficient for expressive groups in Table 5 indicates that members are significantly

more extreme than otherwise similar members of solidary groups. Also, the size of the

coefficient for expressive groups compared to the others indicates that members of these groups

are more extreme than members of coercive or material groups. Finally, we cannot reject the

null hypothesis that members of coercive and material groups are no more extreme than

members of solidary groups. Although the results in Table 5 suggest that expressive groups are

more extreme, we acknowledge that our analysis only includes nine interest groups and we do

not make any claims about how representative these groups are of the interest group universe.

<Table 5>

Of course general policy extremism need not mean members of expressive groups stake

out more extreme positions on their group’s issue. To test whether members of expressive

groups are more extreme on their issue, we created another indicator of extremism by folding the

opinion scales associated with each group.14 In this way we measure each respondent’s distance

from each scale’s neutral point. We then regressed these quantities on membership and repeated

the process again with the control variables. Once again, we restrict the samples to the issue

publics and use the KHB method to compare the effect of membership with and without controls.

Table 6 presents the results of the analysis of attitudinal extremism on the relevant issue. With

the exception of the NRA (as expected given Figure 2), membership in the other expressive

groups is consistently associated with extremism. Moreover, for these groups, the robust

association between membership and extremism remains even after controlling for a wide range 14 The assault weapon ban question has too few response categories for folding.

21

of other attitudes and characteristics. Finally, we confirm that ideology and party identification

are, in each case, significant predictors of a portion of membership related extremism—though

again significant extremism remains unexplained for these groups.15 Though we have not

identified the specific mechanism, we note that extremism beyond that produced by ideological

and partisan self-selection is concentrated in expressive groups. To be clear, we are not claiming

that extremism is the mechanism that causes extremism—that would be tautological. Some

process that has not yet been identified leads individuals both to stake out more extreme

positions on political issues and to join expressive groups.

<Table 6 About Here>

In any case, we conclude with the following observations. Interest groups distort the

opinions of their respective issue publics in a variety of ways, but generally interest groups tend

to stake-out more extreme positions and this tendency is especially apparent in expressive

groups.

Conclusion

The literature on issue publics has optimistically concluded that widespread political

ignorance is not a problem for democracy because those affected by specific issues are well

informed, involved, and represented. Yet, guided by insights from research on civic

participation, we explored whether active members of issue publics, those in formal membership

groups, held the same opinions as those “members” of an issue public that were not interest

group members. On the whole, they are not. In our examination of ten membership groups we

find consistent evidence that group members hold policy attitudes that are distinct from their

counterparts in a broader issue public. In fact, Figure 2 demonstrates that a policy-maker guided

15 We also used the KHB method for this analysis.

22

by interest group representation, rather than a more comprehensive survey of issue public

opinion, might actually come down on the wrong side of an issue in most cases. Clearly opinion

distortion could be problematic.

We also found that interest group members are more extreme than nonmembers of the

same issue public. In contrast to research on civic participation (Schlozman, Verba and Brady

2012; Verba, Schlozman, and Brady 1995), however, the roots of the difference are not socio-

economic. Instead, we found that the extreme attitudes of group members can be attributed to

ideology and party identification. Because party identification is a relatively stable disposition

(Campbell et al. 1960), we surmise that the distinct and often extreme attitudes of group

members are more likely to be a product of selection. Furthermore, in the online appendix we

provide an analysis of non-group-related issues to examine further the question of selection or

persuasion. For most of the groups we examined, we found that the opinions of group members

were significantly different on non group-related issues suggesting that people with distinct

preferences join groups rather than groups creating distinct preferences. Taken together, these

results suggest that the policy distortion produced by interest groups may ultimately stem from

those who are different, and more extreme in their opinions, self-selecting into groups. For

coercive, material, and solidary groups, policy distortion appears to be primarily a function of

ideological and partisan self-selection into likeminded groups. This is also the case for

expressive groups, but, in addition, we found policy distortion beyond that produced by

ideological and partisan self-selection in expressive groups. Yet, we wish to stress that we

cannot rule out alternative mechanisms including persuasion or recruitment without additional

research. Identifying the mechanisms that make interest group members distinctive from

23

nonmembers will no doubt be an interesting avenue for future work investigating interest group

distinctiveness.

Thus, opinion distortion wrought by interest group representation is likely to contribute to

political polarization more generally. When policy-makers rely on interest groups to

communicate the positions of issue publics, they perceive greater polarization than they would if

they had a more accurate measure of issue public opinion. Fiorina (2005) highlights a disconnect

between elite-level polarization and a comparatively moderate public—this study suggests that

elites may also be misinterpreting the positions of issue publics because of interest group

representation.

The implications of these differences are potentially important for understanding the

policy process. Although interest groups do not always influence policy (see Schlozman, Verba

and Brady 2012 for a review), they sometimes do. In those cases where interest groups matter,

especially on issues that involve mass membership groups such as those featured in our research,

legislators who attend to interest group members are listening to unrepresentative voices. In this

way, interest groups distort the views of the broader, attentive public on a particular issue. We

agree with Converse that aggregate opinion on any given issue is more meaningful when shaped

by those for whom the issue matters most—the issue public (1964, 246). However, our results

suggest that issue publics fall prey to some of the same problems they are credited with solving.

A uniformly active issue public would ensure that the voices of those for whom the issue matters

most are heard. But issue publics are not uniformly active. More problematic those active in

interest groups hold positions that are more extreme than, and often at odds with, the positions of

less active members within the issue public.

24

Appendix A: Descriptive Statistics

Type of Shared Group Range Gen. Shared Characteristic Group Characteristic Issue (1=Liberal) Public (Non-Members) Members Coercive Union Employed Minimum 0-1 .79 .73 .80 Wage (12,625) (15,523) (1,973) Material AARP Over 50 Privatize 0-1 .43 .47 .62 Soc. Sec. (14,479) (9,601) (5,573) Solidary VFW Military Iraq War 0-1 .60 .51 .45 Service (23,214) (5,258) (990) American Military Iraq War 0-1 .60 .52 .42 Legion Service (23,130) (5,121) (1,127) PTA Children Ed. Most Imp. 0-1 .02 .02 .04 Issue (24,828) (3,192) (663) Expressive Sierra Club Liberals Global 0-1 .49 .88 .97 Warming (10,265) (3,002) (343) NRA^ Gun Gun 0-1 .86 .68 .48 Owners Control (21,280) (10,889) (2,695) NRA^ Gun Assault 0-1 .80 .67 .43 Owners W. Ban (24,133) (12,682) (3,100) NARAL Liberals Abortion 0-1 .56 .87 .96 (21,466) (5,980) (419) National Right Born Again & Abortion 0-1 .78 .52 .18 to Life Catholics (14,918) (13,802) (548) National Right Conservatives Abortion 0-1 .79 .40 .17 to Life (17,562) (9,944) (516) Christian Conservatives Gay Marriage 0-1 .79 .16 .02 Coalition (8,399) (5,278) (196) Christian Frequent Gay Marriage 0-1 .66 .35 .03 Coalition Attendees (8,855) (4,842) (179) Notes: Cells in the last three columns are means; number of observations in parentheses. ^ The NRA models come from the 2004 Annenberg survey because the CCES did not ascertain views on gun related issues. All others come from the 2006 CCES survey. CCES weight is V1001.

25

References

Anand, Sowmya and Jon A. Krosnick. 2003. “The Impact of Attitudes Toward Foreign Policy

Goals on Public Preferences Among Presidential Candidates: A Study of Issue Publics

and the Attentive Public in the 2000 U.S. Presidential Election.” Presidential Studies

Quarterly. 33: 31-71.

Ansolabehere, Stephen. 2008. Guide to the 2006 Cooperative Congressional Election Survey.

Data Release No. 2.

Berry, Jeffrey. 1977. Lobbying for the People. Princeton: Princeton University Press.

Bosso, Christopher J. .2003. “Rethinking the Concept of Membership in Nature Advocacy

Organizations.” The Policy Studies Journal. 31(3): 397-411.

Burns, Nancy, Kay Lehman Schlozman, and Sidney Verba. 2001. Private Roots of Public

Action: Gender, Equality, and Political Participation. Cambridge, MA: Harvard

University Press.

Campbell, Andrea Louise. 2002. “Self-Interest, Social Security, and the Distinctive Participation

Patterns of Senior Citizens.” American Political Science Review 96: 565-74.

Campbell, Andrea Louise. 2003. “Participatory Reactions to Policy Threats: Senior Citizens and

the Defense of Social Security and Medicare.” Political Behavior 25: 29-49.

Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald Stokes. 1960. The

American Voter. New York: John Wiley and Sons.

26

Converse, Philip. 1964. “The Nature of Belief Systems in Mass Public.” In Ideology and

Discontent. David E. Apter. ed. New York: Free Press.

Clark, Peter B. and James Q. Wilson. 1961. “Incentive Systems: A Theory of Organizations.”

Administrative Science Quarterly 6 (2):129-166.

Delli Carpini, Michael X., and Scott Keeter. 1996. What Americans Know about Politics and

Why It Matters. New Haven: Yale University Press.

Dipak K. Gupta, C. Richard Hofstetterb, and Terry F. Buss. 1997. “Group utility in the micro

motivation of collective action: The case of membership in the AARP.” Journal of

Economic Behavior & Organization 32: 301-320.

Fiorina, Morris. P. 2005. Culture War? The Myth of a Polarized America. New York: Pearson

Longman.

Herbst, Susan. 1998. Reading Public Opinion: How Political Actors View the Democratic

Process. Chicago: University of Chicago Press.

Highton, Benjamin. 2004. “Policy Voting in Senate Elections: The Case of Abortion.”

Political Behavior. 26: 181-200.

Hutchings, Vincent L. 2001. “Political Context, Issue Salience, and Selective Attentiveness:

Constituent Knowledge of the Clarence Thomas Confirmation Vote.” Journal of

Politics. 63: 848-868.

Hutchings, Vincent L. 2003. Public Opinion and Democratic Accountability: How Citizens

Learn about Politics. Princeton: Princeton University Press.

27

Iyengar, Shanto, Kyu S. Hahn, Jon A. Krosnick, and John Walker. 2008. “Selective Exposure to

Campaign Communication: The Role of Anticipated Agreement and Issue Public

Membership.” Journal of Politics. 70: 186-200.

Karlson, K. B. , Holm, A. & Breen, R. 2012. “Comparing Regression Coefficients Between

Same-Sample Nested Models using Logit and Probit: A New Method.” Sociological

Methodology. 42.

Karlson, K. B. & Holm, A. Jun 2011. “Decomposing Primary and Secondary Effects: A New

Decomposition Method.” Research in Social Stratification and Mobility. 29: 221-237.

Kim, Young Mie. 2009. “Issue Publics in the New Information Environment: Selectivity,

Domain Specificity, and Extremity.” Communication Research. 36: 254-284.

Kohler, U. , Karlson, K. B. & Holm, A. 2011 “Comparing Coefficients of Nested Nonlinear

Probability Models.” Stata Journal. 11: 420-438.

King, David C. and Jack L. Walker. 1992. “The Provision of Benefits by Interest Groups in the

United States.” Journal of Politics 54:394-426.

Krosnick, Jon. 1990. “Government Policy and Citizen Passion: A Study of Issue Publics in

Contemporary America.” Political Behavior 12: 59-92.

Krosnick, Jon, and Shibley Telhami. 1995. “Public Attitudes Toward Israel: A Study of the

Attentive and Issue Publics.” International Studies Quarterly 39: 535-554.

Layman, Geoffrey. 2001. The Great Divide: Religious and Cultural Conflict in American Party

Politics. Columbia University Press.

28

Mettler, Suzanne. 2011. Personal communication with the author, February 7.

Moe, Terry M. 1980. The Organization of Interests. Chicago: University of Chicago Press.

Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of Groups.

Cambridge: Harvard University Press.

Patterson, Kelly D. and Matthew Singer. 2002. “The National Rifle Association in the Face of

the Clinton Challenge.” In Interest Group Politics, 6th edition. Allan J. Cigler and Burdett

Loomis. Washington, D.C.: Congressional Quarterly Press.

Price, Vincent, Clarissa David, Brian Goldthorpe, Marci McCoy Roth, and Joseph N. Cappella.

2006. “Locating the Issue Public: The Multi-Dimensional Nature of Engagement with

Health Care Reform.” Political Behavior 28:33-63.

Romer, Daniel, Kate Kenski, Kenneth Winneg, Christopher Adasiewicz, and Kathleen Hall

Jamieson. 2006. Capturing Campaign Dynamics, 2000 & 2004: The National

Annenberg Election Survey. Philadelphia: University of Pennsylvania Press.

Sabatier, Paul A. and Susan M. McLaughlin. 1990. “Belief Congruence between Interest-Group

Leaders and Members: An Empirical Analysis of Three Theories and a Suggested

Synthesis.” Journal of Politics 52:914-35.

Salisbury, Robert H. 1969. “An Exchange Theory of Interest Groups.” Midwest Journal of

Political Science 13:1-32.

29

Schattschneider, Elmer E. 1960. The Semi-Sovereign People: A Realists View of Democracy in

America. New York: Holt, Rinehart, and Winston.

Schlozman, Kay L., Sidney Verba, and Henry E. Brady. 2012. The Unheavenly Chorus:

Unequal Political Voice and the Broken Promise of American Democracy. Princeton:

Princeton University Press.

Skinner, Richard M. 2007. More than Money: Interest Group Action in Congressional Elections.

Lanham, MD: Rowman & Littlefield.

Verba, Sidney, Kay L. Scholzman, and Henry E. Brady. 1995. Voice and Equality: Civic

Voluntarism in American Politics. Cambridge: Harvard University Press.

Weil, Douglas S. and David Hemenway. 1993. “I Am the NRA: An Analysis of a National

Random Sample of Gun Owners.” Violence and Victims 8(4):353-365.

30

Table 1: Comparing Members to Potential Members

Type of Issue Public Dependent Restricted Model Full Model Difference in Group Characteristic Variable (no controls) (w/ controls) Effect of Membership Coercive Union Employed Minimum .49*** .22** .27*** Wage (.08) (.07) (.04) Material AARP Over 50 Privatize .89*** .38*** .51*** Soc. Sec. (.04) (.04) (.03) Solidary VFW Military Iraq War -.40*** -.19 -.21** Service (.10) (.10) (.08) American Military Iraq War -.56*** -.19 -.37*** Legion Service (.10) (.10) (.08) PTA Children Ed. Most Imp. .97*** .81** .16* Problem (.26) (.25) (.08) Expressive Sierra Club Liberals Global 1.58*** 1.29*** .29*** Warming (.22) (.22) (.05) NRA Gun Gun -1.00*** -.85*** -.15*** Owners Control (.04) (.04) (.02) NRA Gun Assault -1.07*** -.93*** -.14*** Owners W. Ban (.04) (.04) (.01) NARAL Liberals Abortion 1.30*** .69** .61*** (.21) (.22) (.06) National Right Born Again & Abortion -2.28*** -.96*** -1.32*** to Life Catholics (.11) (.10) (.06) National Right Conservatives Abortion -1.69*** -.75*** -.94*** to Life (.11) (.11) (.05) Christian Conservatives Gay Marriage -2.57*** -1.51*** -1.06*** Coalition (.42) (.42) (.10) Christian Frequent Gay Marriage -3.67*** -1.58** -2.09*** Coalition Attenders (.46) (.42) (.13) * p<.05; ** p<.01; *** p<.001. “KHB” command in Stata used to obtain restricted and full model estimates and to compute the difference in the effect of group membership once statistical controls are accounted for (restricted model estimate adjusted for scaling of control variables). Ordered logit estimators. Samples restricted to those in the issue public. Estimates for the coefficient for the membership dummy variable reported only. Also included, but not reported, are the control variables mentioned in the main text. Standard errors in parentheses. High values coded in liberal direction. CCES weight is V1001.

31

Table 2: Substantive Effects of Group Membership on Policy Attitudes

Type of Issue Public Dependent Chg. in Likelihood Assoc. Group Characteristic Variable w/ IG Membership: Coercive Union Employed Minimum 3% More Likely Lib. Wage Material AARP Over 50 Privatize 8% More Likely Lib. Soc. Sec. Solidary VFW Military Iraq War 4% More Likely Con. Service American Military Iraq War 5% More Likely Con. Legion Service PTA Children Ed. Most Imp. 1% More Likely Lib. Problem Expressive Sierra Club Liberals Global 19% More Likely Lib. Warming NRA Gun Gun 13% More Likely Con. Owners Control NRA Gun Assault 22% More Likely Con. Owners W. Ban NARAL Liberals Abortion 9% More Likely Lib. National Right Born Again & Abortion 13% More Likely Con. to Life Catholics National Right Conservatives Abortion 13% More Likely Con. to Life Christian Conservatives Gay Marriage 17% More Likely Con. Coalition Christian Frequent Gay Marriage 31% More Likely Con. Coalition Attenders Notes: Changes in predicted probabilities (converted to percentages) computed using SPOST in Stata. Underlying model is the “Full” model from Table 1. All control variables were set to their means. For liberal groups (positive coefficients in Table 1) we assess the change in the probability of the most liberal response. For conservative groups (negative coefficients in Table 1) we assess the change in the probability of the most conservative response.

32

Table 3: Bivariate Estimates of Group Stratification Issue Public Political Group Characteristic Awareness Ideology Party ID Age Income Education Coercive Union Employed .53*** .15*** .14*** .022*** .02** -.02 (.08) (.02) (.01) (.002) (.01) (.02) Material AARP Over 50 .27** .24*** .13*** .047*** .02** .10*** (.08) (.02) (.01) (.003) (.01) (.01) Solidary VFW Military -.06 -.12** -.06** .005 .01 .11*** Service (.15) (.04) (.02) (.003) (.01) (.03) American Military .29 -.19*** -.10*** .025*** .003 .004 Legion Service (.15) (.04) (.02) (.003) (.011) (.028) PTA Children .27 .21*** .07** .007 .07*** .22*** (.16) (.04) (.02) (.005) (.01) (.03) Expressive Sierra Club Liberals 1.22** .36** .04 .017*** .04** .19*** (.29) (.13) (.06) (.004) (.02) (.04) NRA Gun .05 -.23*** -.14*** -.012*** .04*** -.04*** Owners (.07) (.02) (.01) (.001) (.01) (.01) NARAL Liberals 1.71*** .71*** .21*** .008* .04** .08* (.21) (.10) (.05) (.004) (.01) (.03) National Right Born Again & 1.14*** -1.06*** -.40*** .016*** -.01 -.08* to Life Catholics (.17) (.06) (.03) (.003) (.01) (.03) Christian Frequent .05 -1.33*** -.46*** .024*** -.01 -.22*** Coalition Attenders (.34) (.13) (.08) (.006) (.02) (.06) * p<.05; ** p<.01; *** p<.001. Bivariate logit models of group membership. Samples restricted to those in the issue public and those not missing on the issue opinion. Standard errors in parentheses. High values coded in liberal direction. CCES weight is V1001.

33

Table 4: Decomposing the Confounding Effects of Political and Demographic Variables Type of Issue Public Dependent Chronic Political Group Characteristic Variable Awareness Ideology Party ID Age Income Education Coercive Union Employed Minimum -.04*** .12*** .23*** .02* -.006* .005 Wage (.01) (.02) (.02) (.01) (.003) (.005) Material AARP Over 50 Privatize .002 .15*** .28*** .08*** -.009** -.003 Soc. Sec. (.002) (.01) (.02) (.01) (.003) (.003) Solidary VFW Military Iraq War -.002 -.10*** -.15** .01 .001 .02* Service (.005) (.03) (.05) (.01) (.001) (.01) American Military Iraq War .01 -.16*** -.25*** .07*** .0003 .001 Legion Service (.01) (.03) (.05) (.02) (.0010) (.006) PTA Children Ed. Most Imp. .005 .04 -.01 -.02 -.02 .13** Prob. (.010) (.04) (.03) (.02) (.03) (.05) Expressive Sierra Club Liberals Global .05* .06** .01 .02 .01 .05* Warming (.02) (.02) (.01) (.02) (.01) (.02) NARAL Liberals Abortion .17*** .07*** .05*** .02* .03** .03 (.03) (.02) (.01) (.01) (.01) (.02) NRA Gun Gun .0001 -.04*** -.08*** .002 .004* -.002 Owners Control (.0003) (.01) (.01) (.004) (.002) (.002) National Right Born Again & Abortion .02* -.58*** -.34*** .02* -.01 -.02* to Life Catholics (.01) (.04) (.03) (.01) (.01) (.01) Christian Frequent Gay Marriage .003 -1.13*** -.58*** -.06** -.002 -.09** Coalition Attenders (.023) (.09) (.06) (.02) (.005) (.03) * p<.05; ** p<.01; *** p<.001. “KHB” command in Stata used to obtain estimates of the confounding effect of each control variable. Samples restricted to those in the issue public. Also included, but not reported, are the control variables mentioned in the main text. Standard errors in parentheses. High values coded in liberal direction. CCES weight is V1001.

34

Table 5: Extreme Policy Attitudes and Type of Group Incentive Type of Group Coercive .13 (.07) Material .05 (.06) Expressive .39** (.09) Clusters 9 N 4,135 * p<.05; ** p<.01; *** p<.001. OLS estimators. Clustered standard errors in parentheses adjusted for membership in each of the nine groups (PTA excluded). CCES weight is V1001. Sample restricted to members of only one group. Solidary is the reference group for the set of “group type” dummy variables. Also included, but not reported, are the control variables mentioned in the main text.

35

Table 6: Extremism on Group Related Issues

Issue Public Folded Restricted Model Full Model Difference in Group Characteristic Scales (no controls) (w/ controls) Effect of Membership NRA Gun Gun -.28*** -.16*** -.12*** Owners Control (.04) (.04) (.01) Sierra Club Liberals Global 1.52*** 1.25*** .27*** Warming (.22) (.22) (.05) NARAL Liberals Abortion 1.15*** .67** .48*** (.22) (.22) (.05) National Right Born Again & Abortion .27* .57*** -.30*** to Life Catholics (.11) (.11) (.03) Christian Frequent Gay Marriage 1.95*** 1.59** .36*** Coalition Attenders (.51) (.51) (.05) * p<.05; ** p<.01; *** p<.001. “KHB” command in Stata used to obtain restricted and full model estimates and to compute the difference in the effect of group membership once statistical controls are accounted for (restricted model estimate adjusted for scaling of control variables). Ordered logit estimators. Samples restricted to those in the issue public. Estimates for the coefficient for the membership dummy variable reported only. Also included, but not reported, are the control variables mentioned in the main text. Standard errors in parentheses. High values coded in liberal direction. CCES weight is V1001

36

!"#$%&'()'*&+,'-./"01'2$33.%4'51'6,4&%&74'8%.$3'*&95&%7:"3;'-.4&,4"+/'8%.$3'*&95&%7:"3;'+,<'4:&'-$5/"0

!"

!#$"

!#%"

!#&"

!#'"

!#("

!#)"

!#*"

!#+"

!#,"

$"

2.0"+/'2&0$%"41'

6%+='>+%' 8+1'*+%%"+#&'

?5.%@.,' ?5.%@.,' 8$,'A.,4%./'

?77+$/4'>&+3.,7'

8/.5+/'>+%9",#'

*","9$9'>+#&'

6%+='>+%' B<$0+@.,'

8%.$3'*&95&%' -.4&,@+/'*&95&%' -$5/"0'

??C-'?9&%"0+,''D&#".,'

A:%"7@+,'A.+/"@.,'

E?C?D'

E+@.,+/'C"#:4'4.'D"F&'

EC?'EC?'

2"&%%+'A/$5'

G,".,'

H!>'

-I?'

37

"!"#$%&'J)'I:&'K"%&04".,'+,<'BL4%&9"41'.F'-./"01'M3",".,'F.%'6,4&%&74'8%.$37'+,<'677$&'-$5/"07''

2.0"+/'2&0$%"41'N??C-O' 6%+='>+%'

N?9&%"0+,'D&#".,O'

6%+='P+%'NH!>O'

8+1'9+%%"+#&'NA:%"7@+,'A.+/"@.,O'

?5.%@.,'NE?C?DO'

?5.%@.,'NE+4/Q'C"#:4'4.'D"F&O'

8$,'0.,4%./''NEC?O'

?77+$/4'P&+3.,7'NEC?O'

8/.5+/'P+%9",#'

N2"&%%+'A/$5O'

*","9$9'P+#&'

NG,".,7O'

-!#("

-!#'"

-!#&"

-!#%"

-!#$"

!"

!#$"

!#%"

!#&"

!#'"

!#("M3",".,

'R"+7'

6,4&%&74'8%.$3' 677$&'-$5/"0'

Online Appendix I. Question and Coding Questions from the 2006 Cooperative Congressional Election Study !"#$%&'()*+,-./0+.1)"'/)2.1$,'1.)2"3.14))!"#$%%&'$()*+,#$-)($./0(123.1#4$3(506$()*+,#$*)1."5067$)$1#."0583#$-"#2#50$)$,5(1$/9$),,$:;';$./0(3*#2($5($3(#4$1/$6#0#2)1#$)$(#1$/9$4#*/62)+"5.$.")2).1#25(15.($1")1$)2#$*)1."#4$92/*$)$+//,$/9$/+1<50$2#(+/04#01($1/$.,/(#,=$2#(#*>,#$1"#$./0(3*#2$4)1);$$!"#$4)1)$-#2#$./,,#.1#4$/?#2$)$1"2##<*/01"$+#25/4$92/*$'#+1#*>#2$1/$@/?#*>#2$ABBC;$$!"#$2#(+/0(#$2)1#$-)($DE;F$+#2.#01$GH)?2#.I$)04$J5?#2($ABBKL; Generally speaking, do you think of yourself as a ...? Democrat Republican Independent Other (please specify): Not sure If R answered Democrat or Republican… Would you call yourself a strong [Democrat/Republican] or a not very strong [Democrat/Republican]? Strong Democrat Not very strong Democrat Strong Republican Not very strong Republican If R answered Independent… Do you think of yourself as closer to the Democratic or the Republican Party? Democratic Party Republican Party Neither Not sure Thinking about politics these days, how would you describe your own political viewpoint?

Very Liberal Liberal Moderate Conservative

Very Conservative Not sure

What is your gender?

Male Female

In what year were you born? ____ What racial or ethnic group best describes you?

White Black or African-American

Hispanic or Latino Asian or Asian-American Native American Mixed Race Other (please specify): Middle Eastern What is the highest level of education you have completed? Did not graduate from high school High school graduate Some college, but no degree (yet) 2-year college degree

4-year college degree Postgraduate degree (MA, MBA, MD, JD, PhD, etc.)

What is your marital status? Married, living with spouse Separated Divorced Living as married Widowed Single, never married Domestic partnership Don’t know What is your religious preference?$$ Protestant (denomination (optional)): Catholic Jewish

Muslim None Some other religion (please specify) Another type of Christian (denomination (optional) Do you consider yourself to be "born again"? Yes No How often do you attend formal religious services? Once a week or more A few times a month Less than once a month Almost never or never Not sure Thinking back over the last year, what was your family's annual income? less than $10,000 $10,000 - $14,999 $15,000 - $19,999 $20,000 - $24,999 $25,000 - $29,999 $30,000 - $39,999 $40,000 - $49,999 $50,000 - $59,999 $60,000 - $69,999 $70,000 - $79,999 $80,000 - $99,999 $100,000 - $119,999 $120,000 - $149,999 $150,000 or more Prefer not to say Do you or have you ever belonged to a labor union? Current member (please specify which union): Past member Never have belonged to a labor union We'd like to know whether you or someone in your immediate family is currently serving or has ever served in the U.S. military. Immediate family is defined as your parents, siblings, spouse, and children. Please check all boxes that apply. I am currently serving in the U.S. military

I have immediate family members currently serving in the U.S. military I previously served in the U.S. military but I am no longer active

Members of my immediate family have served in the U.S. military but are no longer active Neither myself nor any members of my immediate family have ever served in the What is your current employment status? Working full time now Working part time now Temporarily laid off Unemployed Retired Permanently disabled Homemaker Student Other (please specify) Do you happen to remember the party affiliation of the Governor of your state? Do you happen to remember the party affiliation of Senator 1? Do you happen to remember the party affiliation of Senator 2? Do you happen to remember the party affiliation of your Representative in the House of Representatives? One way that people talk about politics in the United States is in terms of left, right, and center, or liberal, conservative, and moderate. We would like to know how you view the parties and candidates using these terms. The scale below represents the ideological spectrum from very liberal (0) to very conservative (100). The most centrist American is exactly at the middle (50). Ideology Placement-Democratic Party Ideology Placement-Republican Party As you may know, the federal minimum wage is currently $5.15 an hour. Do you favor or oppose raising the minimum wage to $7.25 an hour over the next two years, or not? Favor Oppose Do you think it was a mistake to invade Iraq? Yes No Not sure

From what you know about global climate change or global warming, which one of the following statements comes closest to your opinion?

Global climate change has been established as a serious problem, and immediate action is necessary. There is enough evidence that climate change is taking place and some action should be taken. We don't know enough about global climate change, and more research is necessary before we take any actions.

Concern about global climate change is unwarranted. President Bush recently spoke out in favor of a Constitutional Amendment defining marriage as strictly between a man and a woman. Do you support or oppose a Constitutional amendment banning gay marriage? Strongly support Somewhat support Somewhat oppose Strongly oppose Don't know There has been some discussion about abortion during recent years. Which one of the opinions on this page best agrees with your view on this issue? By law, abortion should never be permitted.

The law should permit abortion only in case of rape, incest or when the woman's life is in danger. The law should permit abortion for reasons other than rape, incest, or danger to the woman's life, but only after the need for the abortion has been clearly established. By law, a woman should always be able to obtain an abortion as a matter of personal choice.

Other (please specify): [text] Now, we'd like to ask you about Social Security. A proposal has been made that would allow people to put a portion of their Social Security payroll taxes into personal retirement accounts that would be invested in private stocks and bonds. Do you favor or oppose this idea? Strongly favor Somewhat favor Neither favor nor oppose it Somewhat oppose Strongly oppose What is the most important problem facing the country?

War in Iraq Terrorism Education Health care and health costs Corruption in government Energy supply/gas and oil prices Economy and jobs Rising prices Poverty Housing Immigration Crime Drug abuse Taxes/deficit Social security and pensions Abortion Gay marriage Pollution and the environment Other: [text]

Are you a member of any of the following organizations? Please check all that apply: American Association of Retired Persons Veterans of Foreign Wars America Legion Sierra Club NARAL National Right to Life Christian Coalition of America PTA/PTO None of the above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

)J#(125.1506$1"#$I504($/9$630($1")1$+#/+,#$.)0$>3=V("/3,4$1"#$9#4#2),$6/?#20*#01$4/$*/2#$)>/31$517$4/$1"#$()*#$)($0/-7$4/$,#(($)>/31$517$/2$4/$0/1"506$)1$),,W$$X/2#$')*#$Y#(($@/1"506$N/0Z1$I0/-$$!"#$.322#01$9#4#2),$,)-$>)00506$)(()3,1$-#)+/0($5($)>/31$1/$#P+52#;$N/$=/3$1"50I$1"#$:;';$%/062#(($("/3,4$+)(($1"5($,)-$)6)507$/2$0/1W$$$ [#($$ @/$$ N/0Z1$I0/-$$Extending the federal law banning assault weapons—do you favor or oppose this? Favor Oppose Don’t know Generally speaking, would you describe your political views as very conservative, conservative, moderate, liberal or very liberal? Very conservative Conservative Moderate Liberal Very liberal Don’t know Generally speaking, do you usually think of yourself as a Republican, a Democrat, an Independent, or something else? Republican Democrat Independent Something else Don’t know If R identifies as Republican, Democrat, or Independent… Do you consider yourself a strong or not a very strong [Republican/Democrat/Independent]?

Strong Not strong Don’t know If not Republican or Democrat… Do you think of yourself as closer to the Republican or Democratic Party? Republican Democrat Neither Republican nor Democrat Don’t know What is your age? Number 18-97 Last year, what was the total income before taxes of all the people living your house or apartment? Just stop me when I get to the right category—less than $10,000; $10,000 to less than $15,000; $15,000 to less than $25,000; $25,000 to less than $35,0000; $35,000 to less than $50,0000; $50,000 to less than $75,000; $75,000 to less than $100,000; $100,000 to less than $150,000; $150,000 or more. Less than $10,000 $10,000 to less than $15,000 $15,000 to less than $25,000 $25,000 to less than $35,0000 $35,000 to less than $50,0000 $50,000 to less than $75,000 $75,000 to less than $100,000 $100,000 to less than $150,000 More than $150,000 Don’t know What is the last grade or class you completed in school? Grade 8 or lower Some high school, no diploma High school diploma or equivalent Technical or vocational school after high school Some college, no degree Associate’s or two-year college degree Four-year college degree Graduate or professional school, no degree Graduate or professional degree Don’t know

The interviewer observed the respondent’s sex from his or her voice. If the voice was ambiguous, the interviewer asked for clarification. Male Female Are you married, living as married, widowed, divorced, separated, or have you never been married? Married Living as married Widowed Divorced Separated Never married Don’t know Are you or any other member of your household, that is, any other adult living in your home or apartment, a member of a labor union? If yes: Is that person you or someone else? Yes, respondent Other household member Both respondent and other household member No one is union member Don’t know Are you working full time or part time? Working full time Part time Temporarily laid off or unemployed Retired Permanently disabled Homemaker Student Other Don’t know Are you of Hispanic or Latino origin or descent Yes No Don’t know

What is your race—white, black or African-American, Asian, Native American or American Indian, or some other race? If Hispanic: Are you white Hispanic, black Hispanic, or some other race? White Black Asian American Indian Other Don’t know How often do you attend religious services, apart from special events like weddings and funerals—more than once a week, once a week, once or twice a month, a few times a year, or never? More than once a week Once a week Once or twice a month Few times a year Never Don’t know Do you consider yourself an evangelical or born-again Christian? Yes No Don’t know Do you mostly attend a place of worship that is Protestant, Roman Catholic, Jewish, Mormon, an Orthodox Church, Muslim, or some other religion?

Protestant (including Baptist, Christian, Episcopal, Jehovah’s Witness, Lutheran, Methodist, Presbyterian, and non-denominational Christian)

Catholic Jewish Mormon Orthodox (including Greek, Russian, and Eastern) Muslim Other Don’t know Including yourself, has anyone living in your house or apartment ever served in the U.S. military? This includes the National Guard and the Reserves. Yes, respondent Other household member

Both respondents and other household member No one served in military Don’t know Do you or does anyone living in your house or apartment own a handgun, rifle, or other type of firearm? Yes No Don’t know If yes: Would that be you or someone else in your household or both? Yes, respondent Other household member Both respondents and other household member No one served in military Don’t know Is the gun owner living in your house or apartment a member of the National Rifle Association, also known as the NRA? Yes No Don’t know (Are you/Is the other person living in your house or apartment/ Are you or the other person) a member of the National Rifle Association, also known as the NRA? Yes No Don’t know At the conclusion of the interview, the interviewer assigned a grade for how knowledgeable the respondent was about politics. The grade is the interviewer’s personal opinion. A, excellent B, good C, average D, poor F, very poor Which one of the parties would you say is more conservative on the national level? Democratic Republican

Don’t know Do you happen to know which party has the most members in the United States House of Representatives? Democratic Republican Don’t know How much of a majority is required for the U.S. Senate and House to override a presidential veto? Two-thirds Other answer Who has the final responsibility to determine if a law is constitutional or not? Is it the president, the Congress, or the Supreme Court? President Congress Supreme Court Don’t know Do you happen to know which job or political office is now held by Dick Cheney? Vice president Other answer Don’t know How many days in the past week did you discuss politics with your family or friends? Number 0-7 Don’t know Some people seem to follow what is going on in government and public affairs most of the time, whether is an election or not. Others are not that interested, or are interested in other things. Would you say you follow what is going on in government and public affairs most of the time, some of the time, only now and then, or hardly at all? Most of the time, Some of the time Only now and then Hardly at all Don’t know During the past week, how much attention did you pay to newspaper articles about the campaign

for president—a great deal of attention, some, not too much, or no attention at all? Great deal Some Not much None Don’t know During the past week, how much attention did you pay to stories about the campaign for president on national network or cable TV news—a great deal of attention, some, not too much, or no attention at all? Great deal Some Not much None Don’t know Political Awareness Scales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`50),,=7$(3*($)2#$./0?#21#4$1/$+#2.#015,#$(./2#($)04$2#(.),#4$1/$2)06#$92/*$B$1/$F;

==>)?0+3@.+)='A.13&("3&'()3@.)B&13&'-3&A.'.11),C)='3.+.13)D+,0$)E.#:.+1)

!"#$45(150.15?#$/+505/0($/9$62/3+$*#*>#2(7$#(+#.5),,=$50$#P+2#((5?#$62/3+(7$)++#)2$

1/$/25650)1#$92/*$2#,)15?#,=$(1)>,#$+2#45(+/(515/0($(3."$)($+)215()0("5+$G%)*+>#,,$#1$),;$

FDCBL$(366#(1506$1")1$#P12#*#$+#/+,#$(##I$/31$62/3+(;$$M0/1"#2$+/((5>5,51=$5($1")1$

*#*>#2($/9$#P+2#((5?#$62/3+($)2#$0/1$#P12#*#$-"#0$1"#=$O/50$1"#$62/3+(7$>31$50(1#)4$

>#./*#$(/$)($1"#=$)11#04$#?#01(7$+#23(#$0#-(,#11#2(7$)04$(/$/0$G)04$1")1$1"5($")++#0($,#(($

50$/1"#2$1=+#($/9$62/3+(L;$$\0$1#2*($/9$!"#$%&#!$45(150.15?#0#((7$-#$9504$54#/,/65.),$)04$

+)215()0$(#,9<(#,#.15/0$1/$>#$5*+/21)01;$$a"#0$51$./*#($1/$1"#$'&'(!"#$%&#!$45(150.15?#0#(($

/9$#P+2#((5?#$62/3+(7$)2#$*#*>#2($+#2(3)4#4$1/$1"#52$45(150.15?#$?5#-($)91#2$O/50506$/2$5($

(/*#7$(15,,$3054#01595#47$+2/.#(($/9$(#,9<(#,#.15/0$)1$-/2IW$$\4#),,=7$)0$#P+#25*#01$50$-"5."$

(/*#$*#*>#2($)2#$2)04/*,=$#P+/(#4$1/$+#2(3)(5?#$./**305.)15/0($92/*$1"#$62/3+$-"5,#$

/1"#2$)2#$0/1$-/3,4$),,/-$3($1/$*)I#$)$(12/06$.,)5*$.)3(),$509#2#0.#;F$$$

`),,506$("/21$/9$1"5($54#),7$-#$)2#$)>,#$1/$6)50$,#?#2)6#7$),>#51$50452#.17$/0$1"#$

(#,#.15/0$?#2(3($+#2(3)(5/0$83#(15/0$>=$#P+)04506$/32$50?#(156)15/0$9/2$#)."$62/3+$

>#=/04$5((3#($1")1$)2#$2#,#?)01$1/$1"#$62/3+;$$!"#$I#=$5($1"#$/>(#2?)15/0$1")1$501#2#(1$

62/3+($,5*51$1"#52$+#2(3)(5/0$#99/21($1/$5((3#($)>/31$-"5."$1"#$62/3+$.)2#(V5((3#($1")1$)2#$

45(.3((#4$)1$62/3+$(+/0(/2#4$9/23*($)04$-2511#0$)>/31$50$62/3+$(+/0(/2#4$

./**305.)15/0(;$$\9$-#$/>(#2?#$*#*>#2$45(150.15?#0#(($)*/06$/1"#2-5(#$(5*5,)2$

5045?543),($/0$62/3+$5((3#(7$>31$0/1$/0$0/0<62/3+$5((3#(7$1"#0$1"#$+#2(3)(5/0$"=+/1"#(5($5($

(3++/21#4;$$b31$59$-#$/>(#2?#$*#*>#2$45(150.15?#0#(($)*/06$/1"#2-5(#$(5*5,)2$5045?543),($

/0$>/1"$1=+#($/9$5((3#(7$1"#0$1"#$(#,#.15/0$"=+/1"#(5($5($(3++/21#4;$$!"#$1#(1$5($0/1$54#),$)04$

-#$*3(1$)4*51$1")1$+#2(3)(5/07$9/2$#P)*+,#7$./3,4$(15,,$>#$")++#0506$/0$62/3+<2#,)1#4$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$1 We are grateful to an anonymous reviewer for this idea.

5((3#($#?#0$59$*#*>#2($),(/$#P+2#(($45(150.15?#$/+505/0$/0$0/0<62/3+$5((3#(;$$&?#0$(/7$1"#$

2#(3,1($92/*$1"5($1#(1$.)0$+2/?54#$9321"#2$#?54#0.#7$#?#0$59$(366#(15?#7$/9$(#,#.15/0$/2$

+#2(3)(5/0;$

c!)>,#$FM>/31$]#2#d$

\0$!)>,#$F7$#)."$/9$1"#$/+505/0$51#*($+2#?5/3(,=$45(.3((#47$+,3($)0$/+505/0$51#*$

2#,)1#4$1/$)9952*)15?#$).15/07$")($>##0$2#62#((#4$/0$62/3+$*#*>#2("5+$)04$1"#$./*+,#1#$

(#1$/9$(1)15(15.),$./012/,(;$$a#$)44$)9952*)15?#$).15/0$50$/24#2$1/$/>1)50$2#(3,1($/0$)0$5((3#$

9/2$#)."$62/3+$1")1$-)($0/1$)$I#=$5((3#$9/2$)0=$/9$1"#$./0(54#2#4$62/3+(;$$]56"$?),3#($./4#4$

)($(3++/21$9/2$)9952*)15?#$).15/0$+/,5.5#(;$$^2/3+($92/*$1"#$%%&'$-#2#$0/1$1#(1#4$9/2$1"#$

^30$%/012/,$)04$1"#$M(()3,1$a#)+/0($b)0$51#*($>#.)3(#$FL$1"/(#$51#*($-#2#$0/1$)(I#4$50$

1"#$%%&'$)04$AL$1"#$62/3+$*#*>#2("5+$83#(15/0($-#2#$0/1$)(I#4$50$1"#$@M&';$$]/-#?#27$

>#.)3(#$@JM$*#*>#2("5+$)04$630$/-0#2("5+$509/2*)15/0$-#2#$/>1)50#4$50$1"#$%%&'$

(32?#=7$-#$-#2#$)>,#$1/$#(15*)1#$1"#$#99#.1($/9$@JM$*#*>#2("5+$9/2$#)."$/9$1"#$%%&'$

/+505/0$51#*(;$

!"#$+32+/(#$/9$1"#$)0),=(5($5($1/$)((#(($-"#1"#2$62/3+$45(150.15?#0#(($5($,5*51#4$1/$

62/3+$2#,)1#4$5((3#(;$$a#$50.,34#$50$1"#$1#(1$#)."$62/3+$-51"$2#*)50506$0/0<125?5),7$0/0<

(+325/3(7$45(150.15?#0#((;$$&)."$/9$1"#(#$62/3+($#P"5>51$45(150.15?#0#(($9/2$(#?#2),$0/0<

62/3+<2#,)1#4$5((3#(;$$'5#22)$%,3>$*#*>#2($/0,=$#P+2#(($30583#$?5#-($2#,)15?#$1/$/1"#2-5(#$

(5*5,)2$,5>#2),($/0$A$/31$/9$e$+/((5>,#$0/0<62/3+<2#,)1#4$5((3#($G1"/36"$1"2##$)44515/0),$

51#*($)2#$.,/(#$1/$(1)15(15.),$(560595.)0.#L;$$f0$1"#$/1"#2$")047$@JM$*#*>#2($#P+2#(($

(1)15(15.),,=7$(560595.)01,=7$4599#2#01$/+505/0($1")0$/1"#2-5(#$(5*5,)2$630$/-0#2($/0$C$/9$K$

+/((5>,#$0/0<62/3+<2#,)1#4$5((3#($)04$@)15/0),$J56"1$1/$Y59#$*#*>#2($#P+2#(($(1)15(15.),,=7$

(560595.)01,=7$4599#2#01$/+505/0($1")0$/1"#2-5(#$(5*5,)2$>/20$)6)50$%"25(15)0($)04$%)1"/,5.($

/0$C$/9$e$+/((5>,#$0/0<62/3+<2#,)1#4$5((3#(;$$@MJMY$)04$1"#$%"25(15)0$%/),515/0$9),,$

(/*#-"#2#$50$>#1-##0;$$$

a"5,#$-#$.)00/1$23,#$/31$+#2(3)(5/07$#?54#0.#$/9$45(150.15?#$/+505/0$/0$0/0<62/3+<

2#,)1#4$5((3#($(366#(1($1")1$(#,9<(#,#.15/0$+2/.#((#($,)26#,=$425?#0$>=$54#/,/6=$)04$

+)215()0("5+$)2#$)0$5*+/21)01$+)21$/9$45(150.15?#0#(($9/2$1"#(#$62/3+(;$$`321"#2*/2#7$

),1"/36"$-#$454$0/1$50.,34#$1"#*$50$!)>,#$Q7$-#$.)0$),(/$2#+/21$1")1$1"#$+)11#20$/9$

45(150.15?#0#(($/0$0/0<62/3+$2#,)1#4$5((3#($5($,#(($)++)2#01$9/2$1"#$(/,54)2=$62/3+(7$1"#$

_!M7$)04$305/0(;$$!"3($1"#$3054#01595#4$(#,9<(#,#.15/0$+2/.#(($1")1$,#)4($1/$45(150.15?#$

/+505/0$).2/(($)$2)06#$/9$5((3#$)++#)2($1/$/+#2)1#$*/2#$(12/06,=$-51"50$1"#$62/3+($-51"$

(560595.)01$0/0<(+325/3($45(150.15?#0#((7$1"#$#P+2#((5?#$62/3+(;$$

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`%MM%#_9RN`%MMM%#_9RRN9%%4?/7%>&C%>$#>$)$13)%?%>$8>$)).&1%B&2$-%&'%&#.1.&19%Q>2$>$2%-&8.3%$)3.B?3&>)%$a/$#3%'&>%37$%*1.&1%B&2$-%F-&8.3G9%%V$%7?\$%>$)3>./3$2%37$%)?B#-$)%3&%37&)$%.1%37$%.))*$%#*,-./9%%V$%?>$%>$#&>3.18%&1-"%37$%/&$''./.$13%'&>%37$%B$B,$>)7.#%2*BB"%\?>.?,-$9%%<-)&%.1/-*2$2b%,*3%1&3%>$#&>3$2b%?>$%37$%/&13>&-%\?>.?,-$)%B$13.&1$2%.1%37$%B?.1%3$a39%%[3?12?>2%$>>&>)%.1%#?>$137$)$)9%%c.87%\?-*$)%/&2$2%.1%-.,$>?-%2.>$/3.&19%%664[%C$.873%.)%@NRRN9%%%J#67?18$)%.1%#>$2./3$2%#>&,?,.-.3.$)%F/&1\$>3$2%3&%#$>/$13?8$)G%/&B#*3$2%*).18%[+Q[!9%%<--%&37$>%\?>.?,-$)%C$>$%)$3%3&%37$.>%B$?1)9%%X&>%-.,$>?-%8>&*#)%F#&).3.\$%/&$''./.$13)G%C$%?))$))%37$%/7?18$%.1%37$%#>&,?,.-.3"%&'%37$%B&)3%-.,$>?-%>$)#&1)$9%%X&>%/&1)$>\?3.\$%8>&*#)%F1$8?3.\$%/&$''./.$13)G%C$%?))$))%37$%/7?18$%.1%37$%#>&,?,.-.3"%&'%37$%B&)3%/&1)$>\?3.\$%>$)#&1)$9%