Is Public Opinion Stable? Resolving the …faculty.wcas.northwestern.edu/~jnd260/lab/Druckman...

JAMES N. DRUCKMAN, elected aFellow of the American Academyin 2012, is the Payson S. Wild Pro-fessor of Political Science and Fac-ulty Fellow in the Institute for Pol-icy Research at Northwestern Uni-versity.

THOMAS J. LEEPER is a Ph.D. can-didate in the Department of Polit-ical Science and a Graduate Fellowin the Institute for Policy Researchat Northwestern University.

(*See endnotes for complete contributorbiographies.)

© 2012 by the American Academy of Arts & Sciences

50

Is Public Opinion Stable? Resolving theMicro/Macro Disconnect in Studies ofPublic Opinion

James N. Druckman & Thomas J. Leeper

Public opinion matters. In theory, it serves as thefoundation on which democratic governmental ac-tion is based.1 In practice, elected of½cials tend torespond to public opinion2; moreover, politiciansinvest massive resources in an effort to track andinfluence opinions that will affect election out-comes. Scholars have been interested in the originsand nature of public opinion since the emergenceof the modern social sciences. Yet a number of ques-tions remain, in particular: is public opinion stable?Stability in public preferences suggests that senti-ments expressed at one point in time will largelysustain and thus may reflect clearly held beliefs. Onthe other hand, instability could suggest that lessstock should be placed in the meaning of publicpreferences at a given point in time.3

Whether one concludes stability or instability de-pends to a signi½cant extent on whether one looksto macro trends in aggregated opinions (for exam-ple, the percentage of the public that supports in-

Abstract: Public opinion matters, both as a central element of democratic theory and as a substantivefoundation for political representation. The origins and nature of public opinion have long attracted theattention of social scientists. Yet a number of questions remain; among the more perplexing is whether

–and under what conditions–public opinion is stable. The answer depends in large part on whether onelooks at aggregations of individual opinions (macro public opinion) or at the individual opinions them-selves (micro public opinion). In this essay, we explore the macro/micro divide and offer a framework todetermine when opinions are likely to be stable or volatile. This framework reflects both the content of thepolitical environment and the nature of individuals’ opinions. Using public opinion dynamics surroundingthe Patriot Act as a primary example, we discuss the role of opinion stability in interpreting public opinionand in understanding the normative implications of public preferences.

51

James N.Druckman& Thomas J.Leeper

141 (4) Fall 2012

creased defense spending) or micro-levelindividual opinions (for example, an indi-vidual’s preference for defense spend-ing). Consider the conclusions from twohighly influential books published in 1992,the ½rst focused on macro opinion and thesecond on micro:

• “Our data reveal a remarkable degree ofstability in America’s collective policypreferences.”4

• “Opinion statements vary randomlyacross repeated interviews of the same people; entirely trivial changes in ques-tionnaire constructions . . . can easily pro-duce [large] shifts in aggregate opinion.”5

These conclusions are not time-bound,as similar conclusions can be found inrecent research on macro trends6 andmicro-level opinions.7

We explore the sources of the micro-instability and macro-stability divide. Webegin with a general discussion of microversus macro studies via an extended ex-ample of public opinion surrounding thePatriot Act. We then offer a frameworkfor understanding when opinions shouldbe stable or volatile. Next, we identify threesources of the micro/macro disconnectthat we believe explain why the type ofdata employed yields such distinct con-clusions. We end by discussing the impli-cations of our argument for both under-standing public opinion and interpretingwhat (in)stability implies from a norma-tive perspective. We consider why thismatters for those who report on and readabout public opinion in the news. Amongother ideas, we conclude that stability,often presumed to indicate “higher qual-ity” opinions, may bring with it some un-desirable features.

The divide between micro and macroperspectives in the social sciences is wellestablished, studied by such prominent

scholars as William James, Harold Lass-well, Kurt Lewin, and Thomas Schelling.8In his aptly titled autobiography, Micro-Macro Dilemmas in Political Science, HeinzEulau explains, “The fancy terms ‘micro’and ‘macro’ have come to mean large andsmall or individual and aggregate or partand whole. . . . Once micro and macro hadbeen attached to persons or groups . . . [i]twas only a small step to insist on ‘bridg-ing’ the micro-macro gap.”9 This gap per-vades a range of topics, but we focus hereon how it manifests in relation to publicopinion and communication.

We should be clear in what we mean bymicro and macro public opinion data. Formicro data, the unit of analysis is an indi-vidual (for example, a survey respondent).Typically, the researcher is interested inknowing what opinion(s) that personholds, why, and with what effects. For ex-ample, one may be interested in knowingwhether an individual respondent oppos-es or supports the Patriot Act, which is apiece of legislation enacted by the U.S.Congress and signed by President GeorgeW. Bush shortly after the September 11,2001, terrorist attacks. It increases thepowers that law enforcement agencieshave to monitor communications, records,and ½nancial transactions in an effort toidentify terror threats.10 With microdata, it is instructive to understand whythe individual holds an opinion–does it reflect deeply held values, knowledgeabout an issue, social experiences, and/ormedia coverage?–and whether the opin-ion shapes subsequent behavior: for ex-ample, is the individual willing to sign apetition in support of that issue? Muchof this work employs surveys that mea-sure an individual’s support for an issue,asking, for example:

• The Patriot Act was enacted in the weeksafter September 11, 2001, to strengthenlaw enforcement powers and technol-

r

g

52

Is PublicOpinionStable?

Dædalus, the Journal of the American Academy of Arts & Sciences

ogy. What do you think–do you opposeor support the Patriot Act?

Researchers then correlate answers tothis opinion measure (typically measuredon a seven-point scale ranging from 1, op-pose strongly, to 7, support strongly) withother variables such as demographic fea-tures (gender or income, for example), par-tisan attributes, experiences (media ex-posure, for example), values (such as im-portance of law and order), and so on.Some of the ½rst survey research reportedresponsive instability, meaning individu-als’ opinions measured at one point in timechanged at a later point in time.11

More recent work has built on this½nding by employing experiments thatrandomly assign respondents to differenttypes of questions.12 For example, somerespondents randomly receive the follow-ing (civil liberties) version of the PatriotAct question:

• The Patriot Act was enacted in theweeks after September 11, 2001, tostrengthen law enforcement powers and technology. Under the Patriot Act,the government has access to citizens’con½dential information from tele-phone and e-mail communications. As a result, it has sparked numerous con-troversies and been criticized for weak-ening the protection of citizens’ civil liberties. What do you think–do youoppose or support the Patriot Act?

Others receive a distinct (terrorism) ver-sion that asks:

• The Patriot Act was enacted in theweeks after September 11, 2001, to strengthen law enforcement powers and technology. Under the Patriot Act, the government has more resources forcounterterrorism, surveillance, borderprotection, and other security policies.As a result, it enables security to identi-fy terrorist plots on American soil and

to prevent attacks before they occur.What do you think–do you oppose orsupport the Patriot Act?

Much of the work that takes this (experi-mental) approach ½nds that respondents’opinions, on average, differ widely depend-ing on which version of the question theyreceive. To many researchers, this ½ndingsuggests that opinions are not groundedand are malleable based on whatever rhet-oric is most recently heard by respon-dents.13 In many ways, these conclusionsoffer an explanation for responsible insta-bility by showing that instability stems,at least in part, from alternative rhetoricfound in discourse or in survey questions.

Other relevant work has tracked indi-viduals’ opinions over time by asking thesame respondents the same question sev-eral weeks apart. The modal ½nding hereis that opinions change and any effects(for example, from a certain type of ques-tion at one point in time) quickly decay.14

For instance, when individuals receive theterrorism version of the Patriot Act ques-tion, they likely become more supportiveof the Act. Yet for the modal individual,that support quickly dissipates and, in fact,may flip if the individual later receives thecivil liberties frame. According to a 2010study by Dennis Chong and James Druck-man, “[W]hen competing messages areseparated by days or weeks, most individ-uals give disproportionate weight to themost recent communication because pre-vious effects decay over time.”15

Whether this instability suggests thatcitizens’ opinions are baseless and of littlevalue is a topic of debate; reasonable move-ments, rather than ineptitude, could ex-plain an individual’s change in opinion.16

Still, when studied at the micro-level, indi-viduals’ political attitudes appear unsta-ble on many issues.17 Such dynamics ledSamuel Best and Monika McDermott toconclude that “reported opinions on . . .

53


141 (4) Fall 2012

the USA Patriot Act . . . vary greatly due to simple variations in question wording,content, and response options.”18

This view of public opinion as ½ckle issomewhat puzzling because it appears tocontradict macro-level studies. For macrostudies, the unit of analysis is not the in-dividual per se, but rather a given issue ora given point in time. The focus is oftenon the overall percentage of individualswho support or oppose a perspective,such as the percentage that support thePatriot Act or the frequency of each re-sponse at a given point in time.19 Muchmacro-level work studies whether govern-ment policies respond to aggregate trendsin opinions (does the government in-crease Patriot Act spending when supportincreases over time?), and conversely,whether public opinion reacts to govern-mental actions (does support wane oncespending increases?) or other events (forexample, the effect that a terrorist threathas on support).20 Studies of macro opin-ions toward the Patriot Act report tre-mendous stability, contradicting the micro½ndings: a 2011 report from the PewResearch Center states, “Public views ofthe Patriot Act, whose renewal is beingdebated by Congress, have changed littlesince the Bush administration.”21 This as-sertion means that the level of support forthe Act at one point in time, for example,is near equivalent to support at a later time.These ½ndings of micro instability andmacro stability are not unique to the Pa-triot Act; rather, they extend across count-less issues and times.22 Peter Mortensenexplains, “Studies convincingly demon-strate that aggregated voter opinions arerather sticky . . . [yet there are] randomfluctuations at the individual level.”23

This contradiction emerges even thoughmacro opinion is the aggregation of microattitudes: macro support for the PatriotAct comes from simply counting thenumber of individual respondents who

expressed support. What explains thisstriking micro/macro instability/stabilityinconsistency?24 Unraveling the osten-sible micro/macro inconsistency is morethan a pedantic exercise.25 Politiciansoften turn to aggregate opinion for guid-ance,26 and media outlets typically reporton aggregate trends.27 To interpret thesetrends and to understand how one mightgo about altering them, we must contem-plate their micro-foundations. Do thesetrends reflect reasoned judgments, or istheir meaning less substantive?

What generates stability? Opinions arestable if they sustain or do not change whenmeasured at two or more points in time.Two factors are critical for creating unstableopinions. The ½rst is a weak attitude. Atti-tudes can range from nonexistent (a non-attitude) to weak to extremely strong.28

For example, an individual may be askedfor her opinion on a policy that she hasnever heard of (regulation of vendingmachines, say) or an issue on which she is highly committed to a position (abor-tion, for instance). As attitudes becomestronger, they also exhibit greater stability;indeed, by some de½nitions, a strong atti-tude is (tautologically) one that persistsand resists change.29 Thus, change occursmostly when attitudes are weak.

Attitude strength is a multidimensionalconcept. The strength of a given attitudedepends on the nature of the attitude (forexample, more extreme opinions tend tobe stronger), the attitude’s structure (moreaccessible attitudes tend to be stronger),and the process by which one forms atti-tudes (those based on elaborative think-ing tend to be stronger, as are attitudesformed in an “online” fashion30). Atti-tudes also tend to be stronger when theyare deemed personally important or areviewed as more certain.31 Finally, atti-tude strength grows when individualsthink about their attitudes or have atti-

54



tude-relevant experiences,32 includingbeing exposed repeatedly to the same in-formation (as from continuous mediacoverage).33 Attitude strength lies on acontinuum from weak to strong; however,we focus here on either strong or weakattitudes.

The second factor that contributes toinstability in opinions is the presence of astimulus. For an attitude to change, theretypically must be a stimulus that inducesthe change; such stimuli might include anostensibly persuasive argument (even onenot consciously processed), a world event,a novel experience, and/or rethinking aviewpoint. That said, most micro studiesattend to stimuli contained in communi-cations, as in the case of the experimentalexample described above. These studiesare meant to mimic the types of rhetoricfound outside the study context (com-munications that may influence macrotrends). Macro movements, and hence in-stability, could be driven by other factorssuch as world events and experiences.Because we seek to explain micro insta-bility and macro stability (rather than viceversa), we limit our following discussionto communications. We also attend tostimuli that are potentially persuasive: thatis, information that has suf½cient cred-ibility to induce change under at leastsome conditions.

Our attitude strength x stimulus frameworkmaps into four model scenarios (Table 1).All else constant, we expect stability atboth the micro and macro level to occurin three of the four situations. In the ½rsttwo cases, in which there are no stimuli,we expect stability because there are noexperiences that would stimulate recon-sideration of an attitude, such as encoun-tering new information. We expect insta-bility when opinions are weak and thereis a stimulus (assuming the stimulus issuf½ciently credible to induce change).As explained, weak attitudes are relatively

open to change, and thus a stimulus mayinduce such modi½cations (assuming thestimulus pushes the opinion in a direc-tion counter to the prior stance).

Perhaps most interesting is when anindividual possesses a strong opinion andencounters a potentially persuasive stimu-lus (countering one’s present opinion, suchas a terrorism argument presented to anindividual who opposes the Patriot Act).When this occurs, we expect that, all elseconstant, the individual will reject thestimulus and cling to the extant opinion.This happens because individuals withstrong attitudes tend to engage in moti-vated reasoning, whereby they seek outinformation that con½rms priors (con-½rmation bias), view evidence consistentwith prior opinions as stronger (prior-attitude effect), and spend more timecounterarguing and dismissing evidenceinconsistent with prior opinions, regard-less of objective accuracy (discon½rma-tion bias).34

Strong attitudes are likely to “comeinescapably to mind, whether conscious-ly recognized or not, and for better orworse these feelings guide subsequentthought.”35 When people receive newinformation about George W. Bush, forexample, those with strong feelings inter-pret that information in light of their ex-isting opinions about Bush. Thus, a pro-Bush voter might interpret informationsuggesting that Bush misled voters aboutthe Iraq War either as false or as evidenceof strong leadership in a time of crisis,rather than as an indication of incompe-tence or deception. Such voters maintaintheir support of Bush and may evenbecome more supportive. An individualstrongly opposed to the Patriot Act, bycontrast, will reject arguments about itsutility for combating terrorism, even if theargument is otherwise objectively sound.Ironically, those with less developed,weaker attitudes “are processing infor-

55


141 (4) Fall 2012

mation more ‘objectively’ than thosewith [stronger] attitudes.”36

As explained, micro-level public opin-ion work tends to suggest instabilitywhile macro-level work suggests stability.With our strength x stimulus framework inmind, we can now turn to three possiblesources that may explain the inconsistentmicro/macro ½ndings.

Measurement Error. Measurement errorcan generate instability on individual sur-vey responses that, when randomly dis-tributed in the sample, cancel out at themacro level. Measurement error occurswhen a survey response departs from its“true value”; for example, on the seven-point scale measuring support for the Pa-triot Act, ranging from strongly opposedto strongly support, a respondent’s trueattitude could be around 5.5. If the surveywere to be administered twice, the respon-dent might report a 5 in one instance anda 6 in another.

Measurement error can stem from char-acteristics of the respondent (for exam-ple, he or she is not paying attention or didnot understand the question), the inter-viewer (misreading the question, includ-ing the response options), the question-naire (the order in which questions areasked), or other factors such as the con-text of data collection. At the micro level,measurement features can cause a re-spondent to offer different answers at

distinct points in time, leading to insta-bility over time. Yet at the macro level,random measurement error cancels outbecause roughly the same number ofrespondents who move in one direction(for example, 40 percent offer a lower levelof support at the time a question is ½rstanswered than at the second time) willmove in the other (40 percent offer ahigher level of support the ½rst time thanthe second).37 Thus, stability exists in ag-gregation (even though 40 percent of re-spondents increased their support betweenthe ½rst and second instances, and 40 per-cent decreased their support, the averagesat each time are the same).38

Measurement error can generate microinstability in any of the four scenarios pre-sented in Table 1. Such error would appearto be less likely among individuals withstrong opinions, because they tend tocling to those attitudes. Yet measurementerror is not about substantive changes,and therefore susceptibility is not contin-gent on attitude strength. Jon Krosnickand Robert Abelson posit that the “rela-tively simple hypothesis that these effects[that is, responsive instability] are greaterin the case of weaker attitudes has clearlybeen discon½rmed.”39

Stephen Ansolabehere and colleaguesoffer compelling evidence that once cor-rections for measurement error are put inplace (for example, using multiple mea-sures and taking averages), the result is

Table 1Conditions for Opinion Stability

Attitude Strength Stimuli Stability?

Weak No Stimuli YesStrong No Stimuli YesWeak Stimuli NoStrong Stimuli Yes

Source: Table created by authors.

56



micro-level (and macro-level) stability.“[T]he low correlations of individuals’issue preferences over time,” they explain,“are easily reconciled with a model inwhich there is a high degree of measure-ment error and a high degree of stabilityin preferences.”40

Sample Inconsistencies. Most discussionsabout a survey sample focus on the selec-tion of respondents: for example, are therespondents representative of the targetpopulation? Yet sampling also includes theselection of issues and times. Researchersaim to draw inferences about opinions onthe universe of issues across time; how-ever, they have no choice but to focus onselect issues at particular times. We suspectthat at least some of the micro/macrodiscrepancy can be traced to distinct foci inthe issues examined and the timing of thestudies. Aggregate studies almost alwaysrely on publicly available survey data fromcredible polling organizations (Gallup,American National Election Study, andso on); consequently, these studies focuson public opinion toward the issues thatwere asked about in these surveys. Thisselection turns out to be a very small andlikely nonrandom sample of the possibleuniverse of issues (for instance, all issuesthe government addresses over a term).Paul Burstein explains that “the entire setof issues studied may be so small that it isunrepresentative of the set of all issuesand an inadequate basis for generaliza-tion. . . . [W]hat should be emphasized ishow our capacity to generalize is limitedby the narrowness of the range of issuesstudied.” He also states that “it’s no secretthat public opinion data don’t exist formost policies legislatures consider.”41

Importantly, the issues that tend to beincluded in public surveys are those thatare more salient, and it makes sense thatsurvey organizations would prefer to gaugeissues salient to the public. James Druck-man and Lawrence Jacobs explain that

there is pressure “to collect policy opiniondata on issues seen as important by thepublic.”42 In his survey of extant work,Burstein shows that these issues includesocial welfare, taxes, and defense: issuesthat have the potential to affect citizensdirectly.43

Thus, macro studies may be biasedtoward issues on which citizens possessstronger opinions because the issues aremore likely to be of personal importance,a key dimension of attitude strength.44

Also, these issues are more likely to becovered in the media, thereby providingcitizens with repeated exposure, which, asmentioned, enhances attitude strength.45

In Figure 1, we chart the number of ques-tions asked regarding the Patriot Act (byall survey organizations contained in theiPoll database) along with media cover-age of the Patriot Act (as captured by non-editorial mentions of “Patriot Act” inSection A of The New York Times). Thenumber of survey questions in the ½eld(gray bars) peaks when media coverageincreases. Survey questions are not askedconsistently across the period; none werein the ½eld during initial authorization inOctober 2001, and few were asked betweenthe July 2005 and May 2011 reauthoriza-tions. Effectively, polls that are respon-sive to media coverage select upon opin-ions that are strong and salient; this non-random selection of times for assessingpublic opinion problematizes the assess-ment of stability. Indeed, as mentionedabove, access to information tends to gen-erate stronger attitudes, which in turn leadto stability.

In short, the strong opinions on issuesthat are polled during times of increasedmedia activity lead to stability. This factsharply contrasts with the foci of manymicro-level studies that typically chooseissues for the exact opposite reason. Thesestudies search for issues on which prioropinions are weak, since that may allow

57


141 (4) Fall 2012

for change (the focus of many of thesestudies), and/or issues that have been ab-sent from recent media coverage. DennisChong and James Druckman echo manyother micro studies in stating that theyselected issues for their 2010 study because“opinions on these issues are liable tochange, which allows us to test hypoth-eses [about opinion change].”46 Studiesalso opt to select issues “that receive scantattention outside of the experiment it-self.”47 Examples from the micro-studiesthat demonstrate volatility include atti-tudes about a particular ballot propo-sition,48 an election involving a new can-didate about whom individuals havescant prior opinions,49 regulation of hogfarms,50 urban sprawl in situations whererespondents are not directly affected,51

or abstract and impersonal subjects, suchas people’s trust in institutions.52

In sum, varying measures of stability instudies of macro- and micro-level opin-

ion may stem in part from differences inthe issues explored and the timing of thatexploration.53 The disconnect originatesin samples of issues and times that are in-comparable.

To see how opinion strength can gener-ate distinct patterns of stability, considerChong and Druckman’s survey experi-ment.54 Their December 2009 study in-volved a nationally representative sampleof about 1,300 individuals and focused onopinions about the Patriot Act. Theirspeci½c dependent measure was the sameas that presented above, where respon-dents reported their support for the PatriotAct on a seven-point scale, with higherscores indicating increased support. Theymeasured opinions at two points in time(t1 and t2), separated by about ten days.

There are two critical features of thisstudy. First, it employed versions of theaforementioned terrorism (“pro”) andcivil liberties (“con”) frames. (The frames,

Figure 1Patriot Act Survey Questions and Mentions in The New York Times

The histogram depicts the number of survey questions ½elded each month (grouped into three-month intervalsfor clarity of presentation). The dark black line is a kernel-smoothed density plot of noneditorial mentions of“Patriot Act” in Section A of The New York Times (NYT) over the same period. Source: Figure created by authors.

58



however, were presented as a series ofstatements rather than in the wording ofthe question, as in the example above.)Respondents received different mixes ofthese frames at t1 and t2. Second, Chongand Druckman randomly assigned respon-dents to conditions that induced them toform strong opinions at t1, or induced themto form weaker opinions at t2.55 We willnot go into the details of the speci½copinion-strength manipulations, but suf-½ce it to say that Chong and Druckmanoffer evidence that their inducements(which are commonly used in psychology)did in fact generate stronger or weaker t1opinions about the Patriot Act.56

Figure 2 reports the average opinions att1 and t2 for the weak-attitude condi-tions, for various frame combinations.57

Figure 2a shows conditions that did notinclude a frame at t2, while Figure 2b isfrom conditions with a t2 frame. Sub-stantial over-time volatility is evident inthe ½gure, with opinions at t1 reflectingthe direction of whatever frame the re-spondents received, but then either mov-ing toward the control group at t2 (that is,the t1 “No,” t2 “No” condition) when not2 frame is offered or flipping to reflect thedirection of the t2 frame when a t2 frameis offered. There is no stability whatsoever.

Figure 3, which contains analogous re-sults but in this case for those induced toform strong opinions, presents an entire-ly different portrait. Here we see tremen-dous stability when no t2 frame is offered(Figure 3a). Moreover, Figure 3b showssimilar stability even in the presence of a contrary t2 frame; individuals withstrong attitudes reject it and cling to theirt1 opinion (which was affected by the t1 frame). This latter dynamic reflectsmotivated reasoning, whereby respon-dents counterargue and reject contraryevidence.58

These results have been replicated withvarious issues, including attitudes about

urban sprawl, a state-funded casino, newscienti½c technologies, and health care.59

The implication is that if macro studiesfocus on issues at times when individualsdevelop strong attitudes, then stability isto be expected; however, instability wouldbe the norm for micro studies to the extentthat they focus on less-developed issues.

While Chong and Druckman’s exper-iment reveals a source of the macro/micro disconnect, it cannot explain thediscrepancy in the case of the Patriot Act,given that it focuses on one issue duringone time period. Moreover, there are un-doubtedly issues on which most possessweak opinions that nonetheless lead todiffering macro and micro dynamics(putting measurement error aside). Wesuspect that these issues as well as theaforementioned Patriot Act inconsistencystem from a third possible cause of in-consistency.

Ecological Validity of the Rhetorical Envi-ronment. One possible reason why microinstability on a given issue at a certaintime would exhibit macro stability is thatthe instability cancels out. Consider theweak-attitude conditions in the Patriot Actexperiment. In that case, proportionalnumbers of individuals were exposed tothe pro and con frames at each point intime. There was considerable movement;but because the numbers were largelyequivalent (due to the assignment to con-ditions), the consequence was a cancel-ing out. Indeed, if we merge all the weak-attitude scenarios, it would appear as ifthere was aggregate macro stability, as theoverall t1 mean is 4.40 (standard devia-tion = 1.79; N = 575) and the t2 mean is4.38 (standard deviation = 1.70; N = 575).

This ½nding suggests one possibility:that stability stems from a macro envi-ronment that includes a broad array ofcontrasting information. Such environ-ments would differ from micro studiesthat often expose individuals to informa-

59

James N.Druckman& ThomasJ. Leeper

141 (4) Fall 2012

Figure 2aWeak Opinions/No t2 Frame

Figure 2bWeak Opinions/t2 Frame

*p < 0.01 for one-tailed tests (for changes between t1 and t2). Source: Dennis Chong and James N. Druckman,“Dynamic Public Opinion: Communication Effects over Time,” American Political Science Review 104 (4) (2010);used here with permission of Chong and Druckman.

60



Figure 3aStrong Opinions/No t2 Frame

Figure 3bStrong Opinions/t2 Frame

*p < 0.10; **p < 0.05 for one-tailed tests (for changes between t1 and t2). Source: Dennis Chong and James N.Druckman, “Dynamic Public Opinion: Communication Effects over Time,” American Political Science Review 104(4) (2010); used here with permission of Chong and Druckman.

61


141 (4) Fall 2012

tion pushing them in a single direction.In terms of the Patriot Act, Chong andDruckman report that the civil libertiesand terrorism frames appeared with nearlyidentical frequency in The New York Timesfrom 2001 through 2005. When this oc-curs, the competing frames often cancelout, leaving opinions unaffected.60 Theresults in Figures 2 and 3 support this con-tention. Notice that for both the strong-and weak-attitude conditions, when in-dividuals receive pro and con frames at t1,their opinions are unmoved relative to thecontrol and consequently sustain until t2.61

Micro work may be lacking in ecologi-cal validity, that is, the extent to whichstudies approximate “real life” situa-tions. If most stimuli in the world (andthus in macro studies) involve competinginformation streams, but micro studiesexplore asymmetric information, the dis-connect may simply reflect a lack of eco-logical validity in micro studies (particu-larly those experimental studies on whichwe have focused). Chong and Druckmanmake this exact point upon discoveringthat across many issues, media coverageincorporates competing information:“Because news stories typically containmore than one or two effective frames,readers rarely encounter a scenario–common in experimental studies–inwhich they are restricted to a single mono-lithic frame of the issue. Thus, framingeffects that occur outside of controlledexperimental settings are not well under-stood.”62

The implication is that stability is thenorm, due to competing communications,and that micro studies overstate instabil-ity due to scant attention to competition.This raises the question of how these com-peting communications work. On the onehand, Paul Sniderman and Sean Theriaultsuggest that “political debate, being ex-posed to opposing sides, tightens the link-ages of mass belief systems and increases

the constraint between basic principlesand speci½c issue choices.”63 In otherwords, individuals exposed to competingmessages largely ignore them and fall backon their well-formed values. On the otherhand, John Zaller suggests that “the massmedia routinely carry competing politi-cal messages [and] each message . . . hasits effects, but the effects tend to be mu-tually canceling in ways that produce theillusion of modest impact.” That is, citi-zens do not rely on well-formed, reasonedvalues, but rather move back and forth inresponse to the messages.64

We began by asking whether publicopinion is stable. Our answer may be lessthan satisfying: it depends. More impor-tant, however, is our identi½cation of whenwe can expect stability. We predict thatopinions will be stable on issues and attimes when individuals possess strongopinions or, putting measurement erroraside, when there is a lack of persuasivestimuli in the environment. We arguedthat micro-level studies signi½cantly over-state the malleability of the mass publicby focusing on issues on which individu-als possess weak attitudes. On the flip side,macro studies likely overstate the extentof stability by relying on publicly avail-able data that overrepresent issues thatreceive substantial media coverage and onwhich individuals possess strong opinions.

We offer a fairly clear blueprint for stepsthat can be taken to vitiate the micro/macro gap:• All possible efforts should be put forth

to reduce measurement error in surveys.While some approaches to doing so–

such as using multiple items, as StephenAnsolabehere and colleagues suggest–come with costs (for example, the cost ofsurvey time or demand effects), there arealso more straightforward steps that canbe taken to minimize error.65

62



• Studies should consciously assess the representativeness of the issues and thetimes on which they focus.A ½rst step is to carry out a more sys-

tematic appraisal of the exact issues andtimes that have been the focus in microand macro study. Then, going forward,studies should attempt to incorporate mul-tiple issues (ones that are likely to havevarying distributions of attitude strength),or at least recognize the consequences ofnot doing so. While most instruments in-clude a vast array of questions about re-spondent demographics, they rarely incor-porate attitude-strength questions thatcould provide insight into expected sta-bility or instability.66 Twenty years ago,Jon Krosnick and Robert Abelson made aplea for the regular inclusion of attitude-strength measures in public opinion sur-veys, but thus far, it has gone largelyunheeded.67

• More attention should be paid to issuesof ecological validity.The intellectual evolution of many po-

litical communication studies led to anoveremphasis on documenting the possi-bility of effects.68 This is no longer a crit-ical goal, and scholars should invest moretime in identifying the nature of the rhe-torical environment that surrounds anissue. They should seek to theorize andemulate the effects of that environment.We recognize that this task brings with ita host of challenges: it requires more in-tensive content analyses, and it introducesthe likelihood of fewer statistically signi½-cant ½ndings, which then face a publicationbias. This raises a larger concern about thepublication process and the biases that re-sult from a narrow focus on p-values.69

We urge caution to anyone inferringmuch at all from survey evidence that sug-gests mass opinions either have changedor remained stable on a given issue. Pol-iticians frequently legitimize their stances

by referring to public opinion, particularlywhen majorities are on their side or opin-ions seem to be shifting their way. Whethera bare majority or a few-percentage-pointshift is meaningful requires an understand-ing of survey practice–question wording,sampling, and so on–but also some senseof why opinions might behave the waythey do. To comprehend the latter, reportsof mass opinion need to be contextual-ized with information about the environ-ment in which opinions were measuredand some sense of the strength of thoseopinions. Unfortunately, present report-ing rarely mentions either. Observers andreporters should aim to present richer nar-ratives to make sense of public opinion.

A ½nal point concerns the normative im-plications of our argument. Strong opin-ions and stability are often seen as signsof an engaged and thoughtful citizenry–coveted attributes. Attitude strength pro-motes constraint70 and engagement.71 Yetstrong attitudes also lead to motivated rea-soning that can cause individuals to resistconsideration of relevant alternative per-spectives. At the extreme, such individu-als can be close-mindedly dogmatic, whichmight be as problematic as extremely labilepreferences. In terms of opinion “qual-ity,” theorists should not presume that thequality of well-developed and thought-out opinions always trumps that of fleet-ing opinions.72

Micro/macro gaps pervade the socialsciences, and we have focused on just oneexample. In so doing, however, we af½rmHeinz Eulau’s hope for the ½eld of commu-nication and public opinion. He believedthat it had the potential to bridge themicro/macro gap: “the new ‘discipline’of Communication represents the ful½ll-ment of the dream for . . . [i]nterdiscipli-nary behavioral science that can addressthe gap.”73

63


141 (4) Fall 2012

endnotes* Contributor Biographies: JAMES N. DRUCKMAN, elected a Fellow of the American Academy

in 2012, is the Payson S. Wild Professor of Political Science and Faculty Fellow in the Insti-tute for Policy Research at Northwestern University. He edited the Cambridge Handbook ofExperimental Political Science (with Donald P. Greene, James H. Kuklinski, and Arthur Lupia,2011). He is currently Editor of the Public Opinion Quarterly as well as the American Politicsseries from the University of Chicago Press.

THOMAS J. LEEPER is a Ph.D. candidate in the Department of Political Science and a Grad-uate Fellow in the Institute for Policy Research at Northwestern University. His researchfocuses on the role of information in politics, especially in the U.S. context. Polarized, a bloghe maintains for Psychology Today, examines psychological perspectives on public opinion.

1 See, for example, Robert A. Dahl, Polyarchy: Participation and Opposition (New Haven, Conn.:Yale University Press, 1971).

2 Robert Y. Shapiro, “Public Opinion and American Democracy,” Public Opinion Quarterly 75(5) (2011).

3 We emphasize that it could suggest that less meaning should be attached to public opinion;another possibility is that instability stems from systematic and thoughtful opinion changesin response to meaningful events.

4 Benjamin I. Page and Robert Y. Shapiro, The Rational Public: Fifty Years of Trends in Americans’Policy Preferences (Chicago: University of Chicago Press, 1992), 45.

5 John Zaller, The Nature and Origins of Mass Opinion (New York: Cambridge University Press,1992), 28.

6 For example, Robert S. Erikson, Michael B. MacKuen, and James A. Stimson, The Macro Polity(New York: Cambridge University Press, 2002), 235; Bryan D. Jones and Frank R. Baum-gartner, “A Model of Choice for Public Policy,” Journal of Public Administration Research andTheory 15 (2005): 325–351; B. Dan Wood and Arnold Vedlitz, “De½nition, Information Pro-cessing, and the Politics of Global Warming,” American Journal of Political Science 51 (3) (2007):553; Frank R. Baumgartner, Jeffrey M. Berry, Marie Hojnacki, David C. Kimball, and Beth L.Leech, Lobbying and Policy Change: Who Wins, Who Loses, and Why (Chicago: University ofChicago Press, 2009), 175–178; Stuart N. Soroka and Christopher Wlezien, Degrees of De-mocracy (New York: Cambridge University Press, 2009), 69.

7 For example, Christopher H. Achen and Larry M. Bartels, “Musical Chairs: Pocketbook Vot-ing and the Limits of Democratic Accountability,” paper presented at the Annual Meetingof the American Political Science Association, Chicago, Illinois, 2004; Douglas A. Hibbs, Jr.,“Implications of the ‘Bread and Peace’ Model for the 2008 Presidential Election,” PublicChoice 137 (2008): 1–10; Dennis Chong and James N. Druckman, “Dynamic Public Opinion:Communication Effects over Time,” American Political Science Review 104 (4) (2010): 665.

8 For a historical discussion, see Heinz Eulau, Micro-Macro Dilemmas in Political Science: Per-sonal Pathways Through Complexity (Norman: University of Oklahoma Press, 1996); see alsoThomas C. Schelling, Micromotives and Macrobehavior (New York: Norton, 1978).

9 Eulau, Micro-Macro Dilemmas in Political Science, 37–38.10 The Act contains a number of other elements, such as rede½ning terrorism to include domes-

tic incidents. The actual name of the Act is the USA PATRIOT Act, which stands for Unitingand Strengthening America by Providing Appropriate Tools Required to Intercept and Ob-struct Terrorism. The Patriot Act is a good issue to focus on insofar as it resembles manyother issues by being periodically salient and touching on both economic and social dimen-sions.

11 For example, Philip E. Converse, “The Nature of Belief Systems in Mass Publics,” in Ideologyand Discontent, ed. David Apter (New York: Free Press, 1964).

64



12 See James N. Druckman, Jordan Fein, and Thomas J. Leeper, “Framing and Biased Informa-tion Search,” paper presented at the Annual Meeting of the American Political Science Asso-ciation, Seattle, Washington, 2011.

13 For instance, Zaller, The Nature and Origins of Mass Opinion; and Larry M. Bartels, “Democracywith Attitudes,” in Electoral Democracy, ed. Michael Bruce MacKuen and George Rabinowitz(Ann Arbor: University of Michigan Press, 2003).

14 Chong and Druckman, “Dynamic Public Opinion.”15 Ibid., 663.16 See James N. Druckman, “On the Limits of Framing Effects: Who Can Frame?” Journal of

Politics 63 (2001): 1041–1066; Paul M. Sniderman and Sean M. Theriault, “The Structure ofPolitical Argument and the Logic of Issue Framing,” in Studies in Public Opinion: Attitudes,Nonattitudes, Measurement Error, and Change, ed. William E. Saris and Paul M. Sniderman(Princeton, N.J.: Princeton University Press, 2004).

17 As is typical, we treat the terms opinion and attitude as interchangeable.18 Samuel J. Best and Monika L. McDermott, “Measuring Opinions vs. Non-Opinions–The

Case of the USA Patriot Act,” The Forum 5 (2007): 1.19 The measure could be the same as the previously presented individual-level measure, where

support is construed as any score above 4. Alternatively, a measure could report percentagesfor each of the seven response options (for example, percentage of respondents who regis-tered a 6) or use a distinct set of response options (for example, “support,” “not sure,”“oppose”).

20 See Page and Shapiro, The Rational Public; Erikson et al., The Macro Polity; and Soroka andWlezien, Degrees of Democracy. There is a related debate on whether political actors respondto issue-speci½c opinions (for example, speci½c trends regarding the Patriot Act) or moregeneralized ideological trends (for example, liberalism versus conservatism); see James N.Druckman and Lawrence R. Jacobs, “Lumpers and Splitters: The Public Opinion Informa-tion that Politicians Collect and Use,” Public Opinion Quarterly 70 (4) (2006): 453–476.

21 Pew Research Center, “Public Remains Divided Over the Patriot Act,” 2011, http://pewresearch.org/pubs/1893/poll-patriot-act-renewal.

22 For example, see Bartels, “Democracy with Attitudes,” in Electoral Democracy, ed. MacKuenand Rabinowitz.

23 Peter B. Mortensen, “The Impact of Public Opinion on Public Policy,” Ph.D. dissertation,Aarhus University, Denmark, 2006, 18.

24 The micro/macro inconsistency is a prototypical example of an ecological inference prob-lem, whereby stable trends at the macro level belie the underlying volatility at the micro level.

25 John E. Jackson and Ken Kollman, “Connecting Micro- and Macropartisanship,” PoliticalAnalysis 19 (2011): 507.

26 Erikson et al., The Macro Polity.27 Michael W. Traugott and Paul J. Lavrakas, The Voter’s Guide to Election Polls (Lanham, Md.:

Rowman & Little½eld, 2008). 28 Russell H. Fazio, “Attitudes as Object-Evaluation Associations of Varying Strength,” Social

Cognition 25 (5) (2007): 603–637.29 Jon A. Krosnick and Wendy A. Smith, “Attitude Strength,” in Encyclopedia of Human Behav-

ior, ed. V. S. Ramachandran (San Diego: Academic Press, 1994); Joanne M. Miller and DavidA.M. Peterson, “Theoretical and Empirical Implications of Attitude Strength,” The Journalof Politics 66 (3) (2004): 847–867; Penny S. Visser, George Y. Bizer, and Jon A. Krosnick,“Exploring the Latent Structure of Strength-Related Attitude Attributes,” Advances in Exper-imental Social Psychology 38 (6) (2006): 1–67.

65


141 (4) Fall 2012

30 Richard E. Petty and Jon A. Krosnick, eds., Attitude Strength: Antecedents and Consequences(New York: Psychology Press, 1995); George Y. Bizer, Jon A. Krosnick, Allyson L. Holbrook,S. Christian Wheeler, Derek D. Rucker, and Richard E. Petty, “The Impact of Personality onCognitive, Behavioral, and Affective Political Processes,” Journal of Personality 72 (2004):995–1027.

31 Jon A. Krosnick, “Attitude Importance and Attitude Change,” Journal of Experimental SocialPsychology 24 (1988): 240–255; Visser et al., “Exploring the Latent Structure of Strength-Related Attitude Attributes.”

32 Ibid.; Krosnick and Smith, “Attitude Strength,” in Encyclopedia of Human Behavior, ed. Rama-chandran; Laura R. Glasman and Dolores Albarracín, “Forming Attitudes that PredictFuture Behavior: A Meta-Analysis of the Attitude-Behavior Relation,” Psychological Bulletin132 (2006): 778–822.

33 Dennis Chong and James N. Druckman, “Public-Elite Interactions,” in The Oxford Handbookof American Public Opinion and the Media, ed. Robert Y. Shapiro and Larry R. Jacobs (New York:Oxford University Press, 2011).

34 See Milton Lodge and Charles S. Taber, “Three Steps Toward a Theory of Motivated Polit-ical Reasoning,” in Elements of Reason: Cognition, Choice, and the Bounds of Rationality, ed.Arthur Lupia, Mathew D. McCubbins, and Samuel L. Popkin (New York: Cambridge Uni-versity Press, 2000); Ziva Kunda, Social Cognition: Making Sense of People (Cambridge, Mass.:mit Press, 1999); Larry M. Bartels, “Beyond the Running Tally: Partisan Bias in PoliticalPerceptions,” Political Behavior 24 (2) (2002): 117–150; David P. Redlawsk, “Hot Cognitionor Cool Consideration? Testing the Effects of Motivated Reasoning on Political DecisionMaking,” The Journal of Politics 64 (4) (2002): 1021–1044; Thomas J. Rudolph, “TriangulatingPolitical Responsibility: The Motivated Formation of Responsibility Judgments,” PoliticalPsychology 27 (1) (2006): 99–122; Alan S. Gerber and Gregory A. Huber, “Partisanship andEconomic Behavior: Do Partisan Differences in Economic Forecasts Predict Real EconomicBehavior?” American Political Science Review 103 (2009): 407–426; Alan S. Gerber and Greg-ory A. Huber, “Partisanship, Political Control, and Economic Assessments,” American Jour-nal of Political Science 54 (2010): 153–173; Paul Goren, Christopher M. Federico, and MikiCaul Kittilson, “Source Cues, Partisan Identities, and Political Value Expression,” AmericanJournal of Political Science 53 (4) (2009): 805–820; James N. Druckman, Cari Lynn Hennessy,Kristi St. Charles, and Jonathan Weber, “Competing Rhetoric Over Time,” The Journal ofPolitics 72 (2010): 136–148; James N. Druckman and Toby Bolsen, “Framing, Motivated Rea-soning, and Opinions About Emergent Technologies,” Journal of Communication 61 (4) (2011):659–688; Howard Lavine, Christopher Johnston, and Marco Steenbergen, The AmbivalentPartisan: How Critical Loyalty Promotes Democracy (forthcoming, Oxford University Press).

35 Milton Lodge and Charles S. Taber, “The Rationalizing Voter: Unconscious Thought in Po-litical Information Processing,” unpublished manuscript, Stony Brook University, 2008, 33.

36 David A. Houston and Russell H. Fazio, “Biased Processing as a Function of Attitude Acces-sibility: Making Objective Judgments Subjectively,” Social Cognition 7 (1) (1989): 64. Moti-vated reasoning is less likely to occur if the individuals are motivated to process informationin an accurate way. Taber and Lodge, however, argue that in most political contexts, this isunlikely; see Charles S. Taber and Milton Lodge, “Motivated Skepticism in the Evaluationof Political Beliefs,” American Journal of Political Science 50 (3) (2006): 755–769 (although seeJames N. Druckman, “The Politics of Motivation,” forthcoming in Critical Review).

37 There is also the possibility of nonrandom measurement error, such as a question written ina biased fashion (“Most support the Patriot Act since it helps prevent terrorism. What doyou think?”). This type of measurement error would generate micro instability only if it werecorrected or changed in over-time surveys.

38 Page and Shapiro, The Rational Public; Mortensen, “The Impact of Public Opinion on PublicPolicy,” 18. See also Larry M. Bartels, “Uninformed Votes: Information Effects in Presiden-tial Elections,” American Journal of Political Science 40 (1) (1996): 194–230.

y

ff

dd

w

d

66



39 Jon A. Krosnick and Robert P. Abelson, “The Case for Measuring Attitude Strength in Sur-veys,” in Questions About Questions: Inquiries into the Cognitive Bases of Surveys, ed. Judith M.Tanur (New York: Russell Sage, 1992), 193. For a more recent discussion, see Visser et al.,“Exploring the Latent Structure of Strength-Related Attitude Attributes.”

40 Stephen D. Ansolabehere, Jonathan Rodden, and James M. Snyder, Jr., “The Strength of Is-sues: Using Multiple Measures to Gauge Preference Stability, Ideological Constraint, andIssue Voting,” American Political Science Review 102 (2) (2008): 216.

41 Paul Burstein, “The Impact of Public Opinion on Public Policy: A Review and an Agenda,”Political Research Quarterly 56 (2003): 31, 36, 68.

42 Druckman and Jacobs, “Lumpers and Splitters,” 470.43 Burstein, “The Impact of Public Opinion on Public Policy.” Another point is that publicly

available surveys tend to focus on policy generalities rather than on speci½c topics: for in-stance, questions about support for welfare in general rather than for a particular welfareprovision. It may be that attitudes toward such general areas are more stable.

44 On attitude strength, see Visser et al., “Exploring the Latent Structure of Strength-RelatedAttitude Attributes.” While we focus on the issue and time, we also note that another pos-sible reason for the micro/macro gap is the nature of the samples used (for example, morestudent samples in many micro studies). However, recent work that relies on representativesample survey experiments leads us to put less emphasis on this possibility.

45 Chong and Druckman, “Public-Elite Interactions,” in The Oxford Handbook of American PublicOpinion and the Media, ed. Shapiro and Jacobs.

46 Chong and Druckman, “Dynamic Public Opinion,” 667. 47 Claes H. de Vreese, “The Effects of Strategic News on Political Cynicism, Issue Evaluations,

and Policy Support: A Two-Wave Experiment,” Mass Communication and Society 7 (2) (2004):191–214; Chong and Druckman, “Dynamic Public Opinion”; James N. Druckman andThomas J. Leeper, “Learning More from Political Communication Experiments: The Im-portance of Pretreatment Effects,” American Journal of Political Science (forthcoming): 3.

48 Bethany L. Albertson and Adria Lawrence, “After the Credits Roll: The Long-Term Effectsof Educational Television on Public Knowledge and Attitudes,” American Politics Research 37(2) (2009): 275–300.

49 For example, Alan S. Gerber, James G. Gimpel, Donald P. Green, and Daron R. Shaw, “HowLarge and Long-Lasting are the Persuasive Effects of Televised Campaign Ads? Results froma Large-Scale Randomized Experiment,” American Political Science Review 105 (1) (2011):135–150.

50 David Tewksbury, Jennifer Jones, Matthew W. Peske, Ashlea Raymond, and William Vig,“The Interaction of News and Advocate Frames,” Journalism and Mass Communication Quar-terly 77 (4) (2000): 804–829.

51 For example, Dennis Chong and James N. Druckman, “Framing Public Opinion in Compet-itive Democracies,” American Political Science Review 101 (4) (2007): 637–655.

52 For example, de Vreese, “The Effects of Strategic News on Political Cynicism, Issue Evalua-tions, and Policy Support”; Diana C. Mutz and Byron Reeves, “The New Videomalaise: Effectsof Televised Incivility on Political Trust,” American Political Science 99 (1) (2005): 1–15. Fordiscussion, see Kathleen M. McGraw and Thomas Dolan, “Personifying the State: Conse-quences for Attitude Formation,” Political Psychology 28 (2007): 299–328.

53 See also Wood and Vedlitz, “De½nition, Information Processing, and the Politics of GlobalWarming.”

54 Chong and Druckman, “Dynamic Public Opinion.”55 They also included conditions with no attitude-strength inducement, but we do not discuss

those results here.

67


141 (4) Fall 2012

56 They did this by inducing “online” or memory-based processing; for details, see Chong andDruckman, “Dynamic Public Opinion.”

57 We recognize that it may be ironic that we are using aggregated averages to make a caseabout micro opinion, which we have argued differs from macro opinion because it is notaggregated. Nonetheless, we take this approach for presentational clarity, noting that if weinstead looked at individual opinion change across periods, the story would be the same.What is critical here are the conditions that generate different types of micro-level opinions–and these conditions accentuate common differences in micro and macro samples givenwhen measurement typically takes place (as explained).

58 Direct evidence of this dynamic comes from a question that asked strong-attitude individu-als to rate the “effectiveness” of the t1 and t2 statements, in terms of “providing informa-tion and/or making an argument about the Patriot Act,” on a seven-point scale, with higherscores indicating increased effectiveness. Participants who received the con frame at t1reported an average effectiveness score of 4.84 (standard deviation = 1.65; N = 83) whereasthose who received the same frame at t2 (after having received the pro frame at t1) regis-tered a signi½cantly lower 4.40 (standard deviation = 1.44; N = 83) average (t164 = 1.83, p < 0.05for a one-tailed test). Similarly, the average t1 pro rating is 5.05 (standard deviation = 1.54;N = 83), but only 4.29 (standard deviation = 1.59; N = 83) at t2 (t164 = 3.13, p < 0.01 for a one-tailedtest). In short, those induced to form strong attitudes at t1 downgraded the t2 frames thatcontradicted their t1 priors.

59 For example, see James N. Druckman, “Competing Frames in a Political Campaign,” in Win-ning with Words, ed. Brian F. Schaffner and Patrick J. Sellers (New York: Routledge, 2010);Druckman and Bolsen, “Framing, Motivated Reasoning, and Opinions about Emergent Tech-nologies”; Druckman et al., “Framing and Biased Information Search”; and Druckman andLeeper, “Learning More from Political Communication Experiments.”

60Chong and Druckman, “Public-Elite Interactions,” in The Oxford Handbook of American Pub-lic Opinion and the Media, ed. Shapiro and Jacobs, 257. One challenge to this, however, is ifindividuals select information consistent with their prior opinions and thus segment them-selves such that those with pro (or con) prior opinions view only pro (or con) information.Indeed, this is a manifestation of motivated reasoning. In this scenario, we would ½nd microstability rather than instability; see Druckman et al., “Framing and Biased InformationSearch.”

61 Alternatively, stimuli in the world may be less likely to change in macro contexts: for example,see Baumgartner et al., Lobbying and Policy Change, 175–178.

62 Chong and Druckman, “Public-Elite Interactions,” in The Oxford Handbook of American PublicOpinion and the Media, ed. Shapiro and Jacobs, 255. See also Sniderman and Theriault, “TheStructure of Political Argument and the Logic of Issue Framing,” in Studies in Public Opinion,ed. Saris and Sniderman, 141; and Donald R. Kinder, “Curmudgeonly Advice,” Journal ofCommunication 57 (1) (2007): 155–162.

63 Sniderman and Theriault, “The Structure of Political Argument and the Logic of Issue Fram-ing,” in Studies in Public Opinion, ed. Saris and Sniderman, 158.

64 John Zaller, “The Myth of Massive Media Impact Revived: New Support for a DiscreditedIdea,” in Political Persuasion and Attitude Change, ed. Diana C. Mutz, Richard A. Brody, andPaul M. Sniderman (Ann Arbor: University of Michigan Press, 1996), 20. Based on an explicittesting of these alternative accounts, Chong and Druckman report, “The participants in ourexperiments were open to argumentation on both sides of the issue and did not merelyrevert to standing positions”; see Chong and Druckman, “Framing Public Opinion in Com-petitive Democracies,” 651.

65 Ansolabehere et al., “The Strength of Issues.” For a general discussion, see Jon A. Krosnickand Stanley Presser, “Question and Questionnaire Design,” in Handbook of Survey Research,ed. Peter V. Marsden and James D. Wright, 2nd ed. (Bingley, U.K.: Emerald, 2010).

68



66 Page and Shapiro explore potential volatility differences among various demographic sub-groups (based, for example, on age, income, education, region, and race). They concludethat “the bulk of the evidence indicates that different groups do not tend to change theirpreferences very often in very different ways”; see Page and Shapiro, The Rational Public, 285–320, italics in original. Our point, however, is that much more relevant subgroup differencesmay be found if one focuses on attitude strength instead of conventional demographics.

67 Krosnick and Abelson, “The Case for Measuring Attitude Strength in Surveys,” in Questionsabout Questions, ed. Tanur.

68 See James N. Druckman, James H. Kuklinski, and Lee Sigelman, “The Unmet Potential ofInterdisciplinary Research,” Political Behavior 31 (2009): 485–510.

69 See Alan S. Gerber, Neil Malhotra, Conor M. Dowling, and David Doherty, “Publication Biasin Two Political Behavior Literatures,” American Politics Research 38 (2010): 591–613.

70 Christopher M. Federico and M. C. Schneider, “Political Expertise and the Use of Ideology:Moderating Effects of Evaluative Motivation,” Public Opinion Quarterly 71 (2) (2007): 221–252.

71 Thomas J. Leeper, “Information Choice and Opinion Change Over Time,” paper presentedat the Annual Meeting of the Midwest Political Science Association, Chicago, Illinois, 2012.

72 See Druckman, “The Politics of Motivation.”73 Eulau, Micro-Macro Dilemmas in Political Science, 359.

Is Public Opinion Stable? Resolving the …faculty.wcas.northwestern.edu/~jnd260/lab/Druckman...

Documents

Transcript of Is Public Opinion Stable? Resolving the …faculty.wcas.northwestern.edu/~jnd260/lab/Druckman...