What is the theory—the logical explanation of how and why the concepts are related (3 points)

Mass Media and the Depoliticization of Personal Experience Author(s): Diana C. Mutz Source: American Journal of Political Science, Vol. 36, No. 2 (May, 1992), pp. 483-508 Published by: Midwest Political Science Association Stable URL: http://www.jstor.org/stable/2111487 Accessed: 22-11-2016 19:15 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted

digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about

JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

http://about.jstor.org/terms

Wiley, Midwest Political Science Association are collaborating with JSTOR to digitize, preserve and extend access to American Journal of Political Science

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

Mass Media and the Depoliticization of Personal Experience*

Diana C. Mutz, Department of Political Science and School of Journalism and Mass Communication, University of Wisconsin-Madison

This study combines contemporary research on the effects of mass communication with findings on sociotropic voting to build a general model that explains the origins and effects of economic per- ceptions. This model is then tested in the context of retrospective personal and social concerns about unemployment.

Survey evidence suggests that retrospective assessments of unemployment result primarily from mediated information rather than from direct experiences. Mass media are found to have an "imper- sonal impact," influencing social, but not personal perceptions of the issue, while personal experi- ences with unemployment influence exclusively personal-level judgments.

Mass media also influence the weighting of pocketbook as opposed to sociotropic concerns by means of a "sociotropic priming effect." Rather than priming all considerations that surround eco- nomic issues, high levels of media exposure to economic news prime the importance of collective perceptions to political evaluations and decrease the importance of personal concerns.

Early studies of economic influences on voting simply assumed that people voted their pocketbooks: when national economic conditions worsened, more citizens experienced economic problems in their own lives, and these people logically voted against the incumbent party. When empirical findings at the indi- vidual level failed to support this explanation, research shifted from a focus on

personal economic experiences to an emphasis on "sociotropic" judgments; that is, individuals' retrospective assessments of economic change at the collective level (see, e.g., Kinder and Kiewiet 1979, 1981; Schlozman and Verba 1979; Kinder 1981; Kiewiet 1983).

Perceptions of collective economic change at the national level have proved to be fairly consistent predictors of individual political preferences in U.S. congressional and presidential elections. However, studies have revealed little about the origins of these perceptions. By examining multiple sources of infor- mation, this study demonstrates the contribution of personal experiences and me- diated information sources to perceptions of economic change at state and na- tional levels.

Past neglect of mediated information sources as influences on economic per- ceptions is altogether understandable; the relative importance of other sources of information seemed superfluous compared to the immediate, tangible impacts that economic change had on people's everyday lives. But the general failure of

*I would like to thank Dick Merelman, Chuck Jones, and Jack Dennis for their helpful com- ments on an earlier draft of this paper. Special thanks are due to Steve Chaffee for his advice and encouragement.

American Journal of Political Science, Vol. 36, No. 2, May 1992, Pp. 483-508 0 1992 by the University of Texas Press, P. 0. Box 7819, Austin, TX 78713

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

484 Diana C. Mutz

the pocketbook voting hypothesis brought the understanding of communication

influences to the forefront of research on economic influences on voting (see,

e.g., Conover, Feldman, and Knight 1986; Kinder and Kiewiet 1981; Weather-

ford 1983a). Kiewiet (1983), for example, suggested that the information about

economic conditions that informs sociotropic judgments must come from televi-

sion or newspapers, or indirectly from mass media through a "two-step flow" of

other people's reports of what they have heard from mass media. In short, the

difference between the pocketbook and sociotropic hypotheses became construed

as a controversy over the kind of economic information posited to have the most influence on economic judgments and ultimately, on political behavior as well (Kiewiet 1983).

Despite assertions about the importance of mass media in informing soci-

otropic perceptions, a substantial body of empirical findings suggests this is un- likely. Past research indicates that economic issues are precisely the type least

likely to be influenced by mass media, presumably because of the immediacy

and accessibility of personal experiences with economic problems as rival sources of information. In studies of agenda setting, for example, findings are

often substantial when dealing with nondomestic issues, since most people must

depend exclusively on news media for their information about international af- fairs. But such effects are weak to nonexistent for obtrusive issues that enter into

people's personal life-space on a regular basis (e.g., McCombs and Shaw 1972; Zucker 1978). Personal experiences in particular are considered "superior" infor-

mation sources that will override any mass or interpersonal communication influ- ences. (For a review of this literature, see Chaffee and Mutz 1988.)

Moreover, the influence process suggested by Kiewiet and others would have to be a directional effect of positive or negative news presentations. There has been little research on effects from directional media content, perhaps owing to early disappointments (e.g., Lazarsfeld, Berelson, and Gaudet 1944). Instead, most research that documents media influence has focused on effects that result

from the mere presence or absence of issue coverage, regardless of direction (e.g., Iyengar and Kinder 1987; MacKuen and Coombs 1981; Weaver et al. 1981). Studies of effects stemming from the nature of the content itself have been far less successful. The well-worn dictum that media tell people what to think about, rather than what to think, has remained intact with few exceptions. '

This study has two major goals. First, it aims to resolve the apparent contra- diction between assumptions about the importance of media in forming soci- otropic perceptions, and empirical evidence that suggests such influence is un-

'A few studies have found evidence of directional effects (e.g., Haight and Brody 1977; Page, Shapiro, and Dempsey 1987), but this evidence remains primarily (if not entirely) at the aggregate level. The hazards of relying on such evidence for inferring individual-level effects are obvious.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 485

likely at best. Understanding the sources of these collective judgments is

important because the accountability of political leaders centers on the accuracy

of these perceptions; unlike personal experiences, collective economic percep-

tions may vary independently of the real world effects of economic policies. Sec-

ond, this study furthers research on the politicization of personal experiences by demonstrating for whom and under what conditions personal experiences are politicized. Mass media are found to play an important role in determining the extent to which pocketbook or sociotropic concerns shape political evaluations.

This study also furthers an understanding of the processes that underlie the

formation of retrospective judgments by demonstrating how the process of im- pression formation may be fundamentally different when evaluating state as op- posed to national economic conditions. Findings suggest that the case against personal experiences as predictors of political preferences may be overstated as a result of reliance on national-level data.

Study Design: The Origins and Effects of Retrospective Economic Perceptions

To illustrate the theoretical development that underlies this study, Figures 1 through 4 take the reader through a progression of stages in the study of eco- nomic influences on voting, leading up to a general model that incorporates the information sources that influence economic perceptions. Survey evidence then documents these information sources' effects on people's perceptions of their own employment-related concerns and on their perceptions of state and national unemployment conditions.

The Pocketbook and Sociotropic Models

Figure 1 shows the relationship that so many scholars have looked for but

seldom found, the influence of personal experiences with economic problems on evaluations of an incumbent. The broken line in Figure 1 suggests a weak rela- tionship, if any, in keeping with the many studies that have reached weak or null conclusions.

In response to the failure of the pocketbook model, the sociotropic model provided an individual-level explanation for the aggregate-level findings. Thus, Figure 2 shifts the emphasis from personal experiences to collective judgments about economic conditions at the social level.

Following Kinder and Kiewiet's work, economic variables generally have been designated as either personal, "pocketbook" variables or collective, "soci- otropic" variables referring to collective assessments of national economic con- ditions (Eulau and Lewis-Beck 1985). But structuring this theoretical puzzle as a choice among dichotomous alternatives ignores a wide range of intermediate referents between self and nation. One such option, group-level economic inter-

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

486 Diana C. Mutz

Figure 1. Pocketbook Model of Relationship between Economic Conditions and

Evaluations of Incumbents

Personal

Experiences Elt

~~ _ _ Evaluation of

Incumbent

ests, has received limited attention thus far (see, e.g., Conover 1985; Kinder,

Rosenstone, and Hansen 1983). Other possibilities, such as the state or local community context, also hold the potential for influence but have generated little research to date. Although some studies have addressed questions of economic influence in elections other than presidential and congressional ones, the mea-

sures of economic conditions have almost always been at the national level, with a few exceptions (see, e.g., Piereson 1977; cf. Klorman 1978; Pollard 1978). There is an unfortunate lack of research that matches local or state election out- comes with the economic context of their corresponding geopolitical units.

Nonetheless, Chubb (1988) suggests that the greater economic responsibil- ity being taken on by states in recent years makes the relationship between voting and economic change especially likely to hold at the state level. Economic activ- ities aside, general parallels between governors and presidents also recommend the likelihood of this relationship. As Schlesinger (1966) notes, executive, as opposed to legislative, incumbents are held most accountable for government activity during their term of office. Issues in general are also more important in high-information executive contests than in low-information legislative elec- tions, for which voters fall back on such cues as party identification (Hinckley, Hofstetter, and Kessel 1974). On the other hand, the relatively low salience of state, as opposed to local or national politics, may prevent citizens from holding governors accountable for state economic conditions. As Jennings and Zeigler (1970, 524) note, "In a sense the states are caught between the immediacy of the local [political] system and the glamour and importance of the national and inter- national systems."

In the one study that directly examines state-level voting and state-level eco- nomic conditions, Chubb (1988, 149) finds that "state economic conditions and the assumption of gubernatorial responsibility for them have a significant impact on gubernatorial election outcomes" but concludes that the influence of state con-

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 487

Figure 2. Sociotropic Model of Relationship between Economic Conditions and Evaluations of Incumbents

Objective Level

Subjective Level Personal

Experiences _

n eonom Evaluation

Incumbent

Perception of/ Unemployment/Economy/

as Social Problem

ditions is small in comparison with the effects of national economic conditions. Chubb interprets these results as providing little support for the notion that state politicians are held accountable for a state's economic performance (see also

Peltzman 1987).

As with the early studies of national economic performance, Chubb's evi-

dence was based on pooled time series data using states, not individuals, as the unit of analysis. It remains to be seen whether subjective assessments of state economic conditions have effects similar to those found at the national level. The individual-level mechanism through which national economic change renders na- tional political consequences remains controversial, and at the state level it is

altogether unexplored territory. Although the model in Figure 2 is stated in gen- eral terms, it is likely that people make the attribution link between political

causes and personal consequences more easily at state or local levels; it is prob- ably less of a mental leap to hold state or local officials responsible for losing

one's job than to hold a president responsible. Still, in interpreting the relative strengths of the pocketbook and sociotropic

paths, it is important to note that these two measures, which have often been pitted against one another empirically, are on fundamentally different levels. Per- sonal experiences are exogenous, objective things that happen to a person, while

perceptions of economic change are subjective assessments of social conditions. Since subjective reality almost always predicts attitudes better than objective re- ality, the winner in these contests should be no surprise; it is indicated by the solid line in Figure 2.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

488 Diana C. Mutz

Figure 3. Elaborated Sociotropic Model

Subjective Perceptions

Personal (Retrospective) Experiences

Perception of Political Unemployment/Economy Evaluations

as Personal Problem

Evaluation

of

Incumbent

Perception of Unemployment/Economy

as Social Problem

Subjective Problem Perceptions

Personal experiences with economic problems are neither necessary nor suf- ficient conditions for viewing these issues as pressing personal problems (Kie- wiet 1983); people who are currently employed or doing well financially may still be worried about these issues because they anticipate layoffs or salary cut- backs. With this problem in mind, I have incorporated in Figure 3 a personal-

level concept comparable to the social perception measure: perceptions of economic issues as problems in people's own lives. In previous studies, the in- vestigator has usually defined some condition as a problem for the respondent using personal experience as an indicator (e.g., unemployment); here, by con- trast, respondents are allowed to assess their own levels of personal economic concern.2 I include prospective personal concerns at this position in the model, since they are the result of informational influences and thus endogenous to the model; anticipated or feared economic changes come about as a result of infor- mation gleaned from past personal experiences or from outside sources of infor- mation (see Fiorina 1981 for evidence on the effects of prospective evaluations).

Since this concept incorporates future as well as past concerns as subjec- tively defined by the respondent, it is more likely to have a direct influence on incumbent evaluations than is an objective measure of personal experience, al- though the relationship between the subjective personal-level measure and in-

2Although Kiewiet (1983) tried to circumvent this problem by comparing open-ended responses

to personal and collective "most important problem" questions, these questions clearly do not measure

the same thing as personal and social-level retrospective economic evaluations.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 489

cumbent evaluations is still not expected to exceed the strength of the sociotropic

relationship. As Figure 3 illustrates, I hypothesize no direct relationship between

personal experiences and incumbent evaluations, but personal experiences may have an indirect effect through their strong influence on perceptions of issues as

personal problems.

The final, and most important step in building this model is to factor in the

sources of economic perceptions at the collective level. What are the possible

sources of information that lead to perceptions that economic conditions have

gotten better or worse? To some extent, social perceptions may be generalized

from perceptions of these problems at the personal level, as the central, down-

ward arrow in Figure 3 indicates. Beyond generalizations from personal-level

judgments, the two other possible sources of information on social conditions are mass media and other people.

Mass media. The key factor that weakens the case for mass media effects

with respect to economic issues in particular is that economic issues have "real

world" consequences that researchers have long assumed to override potential media effects (e.g., Erbring, Goldenberg, and Miller 1980; Graber 1984; Weaver et al. 1981). Typically, "media dependency" theories downplay the potential for media effects in situations in which alternative information sources are available (see Ball-Rokeach and DeFleur 1976; Eyal, Winter, and DeGeorge 1981; Mc- Combs and Shaw 1972; Palmgreen and Clarke 1977; Tipton, Haney, and Base- heart 1975; Zucker 1978). Inflation and unemployment in particular affect many

people's lives on an everyday basis. Under these conditions, the media depen- dency literature suggests weak, if any, media effects on people's perceptions of

economic issues, since there are highly credible, more easily accessible sources of information available in the form of people's own personal experiences (see,

e.g., Ball-Rokeach and DeFleur 1976; Graber 1984). But to the extent that polit- ical behavior is based on perceptions of collective economic conditions, mass media may play an important role in this process.

Impersonal influence. Recent evidence suggests that mass media's primary impact may be on social-level perceptions. Studies of agenda setting, for ex- ample, can be differentiated based on whether personal- or social-level defini- tions of audience agendas are used. McLeod, Becker, and Byrnes (1974) point out that mass media are typically found to have a greater impact on people's perceptions of the collective salience of issues than on the salience of the issues to individuals themselves (see also, Becker, McCombs, and McLeod 1975). In other words, a person may not perceive some problem as more important to him

or her personally just because he or she has seen much news concerning it, but that person would be very likely to think that it is an important issue to other people and, thus, an important social problem (see also Hawkins and Pingree 1982).

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

490 Diana C. Mutz

Additional studies have demonstrated that mass-mediated information has

an "impersonal impact" on perceptions of the frequency or severity of a problem, but not on personal judgments of concern or importance (Tyler and Cook 1984;

Tyler 1980, 1984). Tyler (1980), for example, found that mass media coverage

of crime influenced perceptions of crime as a social problem, but not as a per-

sonal one (see also Gordon and Heath 1981; Skogan and Maxfield 1981).

Studies of the "third-person effect" also suggest that people perceive others as influenced by mass media more than they themselves are (Davison 1983).

Several examinations of this hypothesis have demonstrated that one major effect

of exposure to mass media is to influence people's perceptions of what others

think regardless of whether it influences individuals' personal perceptions (see

also Cohen et al. 1988; Mutz 1989). This collection of evidence suggests that despite the obtrusive nature of eco-

nomic issues, mass media may still have some effect on perceptions of economic

trends at the collective level. Very little research bears directly on questions about the effects of mass media reporting of economic news on perceptions of eco- nomic conditions. A few studies, however, indirectly address these questions and are well worth mentioning in this regard.

In an effort to understand the origins of perceptions of economic trends, Weatherford (1983a) compared people who relied heavily on newspapers for po-

litical information with those who did not and found that personal experiences weighed more heavily in voting considerations among the information-poor

subgroup, while the information-rich relied on perceptions of collective eco- nomic conditions. These findings mesh well with Adoni and Cohen's (1978) sug- gestion that mass media use contributes to a subjective sense of economic knowl- edge; heavy media users in Weatherford's study relied on their subjective perceptions whether they were correct or not, while low media users relied on an information source of which they could be certain-their own personal experi-

ences. The findings of Conover and colleagues (1986) loosely paralleled these results: those having inaccurate information about unemployment and inflation

relied on personal experiences to form collective economic perceptions, while respondents with accurate unemployment and inflation knowledge did not. The extent to which people generalize collective or sociotropic evaluations from per- sonal experiences seemingly depends on access to and assimilation of informa- tion about experiences beyond one's own; that is, in all likelihood, information from mass media.

Figure 4 assigns mass-mediated information a role consistent with these findings. As suggested by the literature on agenda setting, the impersonal impact

hypothesis, and related studies, I expect mass-mediated information to have its greatest effects on perceptions of economic issues at the collective level. In the absence of mass-mediated information, personal and collective concerns would become less distinct, and people would default to personal experiences and more parochial sources as the basis of their perceptions of social change.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 49I

Figure 4. General Model of Relationships between Information Sources, Subjective Perceptions, and Evaluations of Incumbents

Sources of Information

Subjective Perceptions

Personal (Retrospective)

Experiences \

Perception of Political

Unemployment/Economy Evaluations as Personal Problem

Interpersonally J l Evaluation

Mediated of Information ofumen

Perception of

Unemployment/Economy

as Social Problem

Mass-Mediated

Information

Legend:

Strong relationship - - - Weak relationship

Interpersonal sources. Although mass media influence has captured the bulk of researchers' attention as a potential source of economic information, some studies have also explored the flow of economic information in an interper- sonal context.3 Few, if any, studies have measured directly interpersonal discus- sion with respect to economic problems, but studies of "contextual" effects and group economic interests often suggest interpersonal communication as the underlying cause. As Weatherford (1983b, 870) noted, "The evaluation of eco- nomic conditions is a natural situation for contextual effects to operate through interpersonal contact; individuals are readily aware of co-workers and acquaint- ances who are unemployed, and shoppers in markets as diverse as food and real estate commonly compare their experiences with inflation" (see also Kinder, Ro- senstone, and Hansen 1983). Results of the few studies that examine contextual

effects and group-level economic interests do not argue for the superiority of these influences, but they suggest that interpersonally transmitted economic in-

3Subsequent research on the two-step flow idea proposed by Katz and Lazarsfeld (1955) sug- gested that it was more the exception than the rule; most mass-mediated information reaches people

directly rather than through an interpersonal network (see Chaffee 1972, 1982, for a review of this

literature).

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

492 Diana C. Mutz

formation may have an independent effect on political evaluations beyond the

effects of personal experiences (see Conover 1985). Studies of issues outside the

economic realm also have demonstrated that interpersonal discussion can influ-

ence both personal and social levels of judgment (Tyler 1980). On the basis of these limited findings, the model assigns interpersonal com-

munication the potential to influence both personal and social perceptions in Fig-

ure 4. This means that hearing about other people's economic problems can make

people more aware that the problem is shared by others and thus an important

social problem, and it also may make them more worried about their own per- sonal situations. Nonetheless, the broken lines indicate that personal experience

should be the strongest influence on perceptions of unemployment as a personal problem, and mass media should be the strongest influence on perceptions of

unemployment as a social problem. In Figure 4, the one remaining link that deserves explanation is the central

arrow pointing upward from social- to personal-level perceptions. As previously mentioned, people may generalize from personal- to social-level concerns, but the recognition of a social problem may also trigger the reverse process; for ex-

ample, national economic decline may lead people to believe that their own per- sonal finances are in peril. Although one would expect the generalization from

personal to social levels to be the dominant influence in the economic context, both directions are theoretically possible, and I test both possibilities in the con- text of the full model.

Methods

Unemployment was chosen as a narrowly defined economic issue appro- priate for an initial examination of these ideas.4 The data used to test the theoret- ical model illustrated in Figure 4 come from two surveys designed specifically for this purpose. For each wave of data, interviewers contacted a random sample of 300 Indiana residents by telephone using random digit dialing. This procedure resulted in two successive cross-sectional samples spaced at six-month intervals, in June and December 1987.5

Two of the three information source variables at the far left side of the model

were assessed through survey questions. Personal experience with joblessness required that the respondent or someone in his or her immediate family had been

laid off or had had trouble finding a job within the past year.6 Instead of asking

people how often they talked to others about some topic-the traditional measure

4Methodological considerations also suggest that unemployment would provide a cleaner and

more exacting test of the model than, say, changes in personal financial status (see Kramer 1983; Rosenstone, Hansen, and Kinder 1986).

'For details on sampling and data collection procedures, consult Appendix A.

6This question incorporated many more respondents into the personal experience category than would a measure of respondents' personal experiences alone. For example, although only 7% to 9%

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 493

of interpersonal communication-in this case they were asked how frequently

others they knew told them about having trouble finding or keeping a job.' Con-

structing a measure of mass media exposure to information about unemployment involved combining survey responses on general media use with data derived from content analyses of respondents' newspapers.8

The questionnaire measured support for incumbents and retrospective per-

ceptions of unemployment conditions at both state and national levels.9 It also

assessed subjective perceptions of unemployment change using parallel ques-

tions that addressed personal and collective levels of judgment. The personal-

level measure of unemployment concern was designed to include those worried about job security and future layoffs as well as those with family members having actual experiences with these problems. 10

At an operational level, care has been taken to ensure that the unidirectional

arrows in Figure 4 are truly recursive relationships, with the exception of the one

of respondents reported having been personally unemployed during the past year, more than 20%

reported either themselves or someone in their immediate family had been personally affected within

the past year.

7Again, this indicator should be distinguished from the interpersonal communication measures

traditionally included in election research in which interpersonal communication is conceived as a

force in competition with mass-mediated information. Here, by contrast, it is seen as yet another

channel through which information about unemployment (contrasting or, more likely, congruent in-

formation) may flow.

'Although newspapers obviously do not cover the entire spectrum of mass media available to

people, newspapers were chosen as the indicator of mass media content because some evidence has

suggested that their effects are more reliable and enduring than effects of television news broadcasts.

Furthermore, broadcast agendas tend to imitate newspaper agendas with a small time lag. (For more

on television-newspaper comparisons, see Benton and Frazier 1976; Erbring 1980; Eyal 1981; Mc-

Clure and Patterson 1976; Tipton, Haney, and Baseheart 1975.)

For each newspaper named by a respondent, a content analysis was done of 10 front pages

appearing during two reconstructed weeks (weekday editions only) selected from the two months

preceding each survey. Two coders measured the number of column inches for each identified unem-

ployment-related article and noted the directional thrust of the article along a five-point good news/

bad news scale. Amount-by-direction measures were summed, creating a single score for each news-

paper for each wave. The intercoder reliability coefficient was .76. Intracoder reliability was .95 for

the first coder, and .87 for the second coder. Discrepancies were resolved in favor of the more reliable

coder. The final measure of Exposure to Newspaper Coverage of Employment Trends was created by

standardizing the variable described above, and then multiplying it by a dummy variable representing

the respondent's frequency of newspaper exposure (0 for nonreaders, 1 for occasional readers, and 2

for regular readers).

9Both state and national assessments are parallel in the amount of time preceding an election.

Further, both the gubernatorial and presidential incumbents were two-term Republicans who were not

running for reelection.

'?Details on the operationalization of concepts in Figure 4 are described in Appendix B. To avoid problems resulting from priming respondents to think about economic issues in the context of

the survey interview, the subjective unemployment-related questions were not asked in close proxim-

ity to questions dealing with evaluations of incumbents (see Sears and Lau 1983).

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

494 Diana C. Mutz

acknowledged reciprocal link;1I I include control variables to partial out preexist- ing differences as much as, if not more than, is appropriate. 12 In addition to tra-

ditional controls, the analyses of both direct and indirect effects included one

additional variable, the community unemployment context. This variable was operationalized using county-level unemployment statistics for the month pre-

ceding each wave of data collection. The design includes this variable to control for the possibility that people were responding to events in their community that

were observed but not directly experienced, nor learned about through mediated

sources; for example, when one observes long lines at the unemployment of-

fice. 13 In this way, the test for media influence controls for "real world cues" at

both aggregate and individual levels. If the hypothesized relationships hold up

under these stringent conditions with the small sample sizes used in this study, one can be assured that the findings represent relatively robust effects.

Findings

The presentation of results proceeds in four parts. First, the direct, additive effects of perceptions of unemployment on evaluations of incumbents are calcu- lated using ordinary least squares regression. Next is an analysis of the origins of personal and collective retrospective perceptions of unemployment using two- stage least squares to examine influences on personal and social perceptions si- multaneously. The third step evaluates the political importance of personal and collective perceptions with respect to the amount of unemployment coverage

"To avoid simultaneity problems, operationalizations of concepts purposely erred on the con- servative side. For example, the coding procedure used in this study results in a very general measure of exposure to media coverage of unemployment news. Questions about attention to job-related news are not included as part of the measure. If questions were asked in more specific terms, information seeking that might result from social or personal perceptions of the issue-as-problem would make it necessary to draw causal arrows from subjective perceptions to attention to unemployment news as well as the other way around.

2Controlling for party identification is one means of establishing a baseline that explains as much of the preexisting variation as possible; however, the usual party identification scale is likely to pick up short-term as well as long-term influences, so a model estimated with these controls would

likely overcontrol for preexisting differences and underestimate the true effects of perceptions of un-

employment trends (see, e.g., Brody 1977; Converse and Markus 1979; Fiorina 1981; Meier 1975; Page and Jones 1979; Shively 1977). Following Kiewiet (1983), this dilemma was resolved by using two multiple regression models, one using party identification and the other demographic variables, as upper and lower boundaries of the effects to be estimated. I have presented the more conservative estimates of these relationships in Table 1. The alternative models have greater explanatory power overall, but do not change the relationships between variables significantly. Both party identification

and demographic controls are used in the remainder of the analyses in Tables 2, 3, and 4. "3These variables are likely to "overcontrol" for the influence of mass and interpersonally me-

diated messages. Areas with high unemployment rates, for example, are likely to have high levels of interpersonal discussion and high media coverage of this issue. Although much of this information will be conveyed to people through mediated channels, the community-level indicators will account for this same variance to the extent they mirror real world conditions.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 495

over time and exposure levels within subgroups of the population. Finally, a col- lapsed version of the model is considered to investigate further the potential for politicization of personal experiences.

At each stage in this progression, results are presented for the spring and

fall waves of data at the state level and the fall wave at the national level. Testing three replications of the model allows over-time comparisons at the state level, and comparisons between state- and national-level unemployment effects at the same point in time.

Direct Effects on Incumbent Evaluations

As the unstandardized beta coefficients in Table 1 demonstrate, perceptions of unemployment at the social level have a consistently significant effect on ap-

Table 1. Predictors of Support for Incumbents

Govemor Govemor President

(Spring '87) (Fall '87) (Fall '87)

Subjective evaluations

of unemployment:

Social problem (better)" .32 (.14)* .41 (.13)** .43 (.19)* Personal problem (better) .26 (.15) .08 (.13) .34 (.18) Sources of information:

Personal experience -.47 (.33) .22 (.33) .23 (.45) Interpersonally mediated

infonnation - .18 (.12) .02 (.13) - .31 (.17) Exposure to newspaper

coverage of

employment trends

(bad news) -.01 (.09) .06 (.13) -.03 (.03) Community context: .08 (.09) -.00 (.08) .18 (.12) Party identification:

Republican .41 (.30) .64 (.27)* 2.14 (.41)*** Democrat -.78 (.32)* -.74 (.29)** -.61 (.40)

R2 .20 .16 .26

n (204) (228) (265)

Note: Entries are unstandardized betas with standard errors in parentheses. The dependent variable is evaluation of incumbent performance, where a high score indicates approval. This model repre- sents conservative estimates of these relationships (see note 13); including both party and demo- graphic variables produces considerably higher R2 values, but does not change significantly the pat- tern of relationships.

aDirection of higher values indicated in parentheses.

p < .05; **p < .01; ***p < .001, two-tailed tests.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

496 Diana C. Mutz

proval of incumbents; this remains true across both waves for perceptions of state

unemployment trends in relation to the incumbent gubernatorial administration and for perceptions of national unemployment trends in relation to the incumbent

president.

As hypothesized in Figure 4, the sociotropic effect is the strongest direct

influence on approval of incumbents outside the direct effects of party identifica-

tion. Personal experiences with unemployment do not have direct effects, just as previous research would indicate. However, the measures of subjective percep-

tions of unemployment as a personal problem approach statistical significance in some cases, just as the weaker arrow in Figure 4 would suggest.

Origins of Retrospective Evaluations

Given the political importance of perceived unemployment trends, it be- comes important to know on what kind of information these perceptions are based. What sources of information contribute to the social- and personal-level

perceptions that ultimately influence political attitudes? To answer these questions, I used two-stage least squares to resolve the

complications posed by the potential reciprocal relationship between perceptions of unemployment as a personal and social problem. Preliminary tests of the gen- eral model using ordinary least squares confirmed the two key instrumental vari- ables indicated in the theoretical model and used in these analyses; personal ex-

perience uniquely identifies perceptions of unemployment as a personal problem across all three tests of the model, and exposure to newspaper coverage of un- employment trends uniquely identifies perceptions of unemployment as a social problem. 14

The unstandardized betas in Table 2 illustrate the results of these efforts. At the state level, the pattern is clear; in both waves, personal perceptions signifi- cantly influence social-level perceptions, but not the other way around. At the national level, personal- and social-level concerns about unemployment remain virtually independent: personal concerns do not translate to social ones, and per- ceptions of unemployment as a national problem do not trigger concern at the personal level.

Table 2 also illustrates the consistent influence of personal experiences on

personal concern, and mass media on social concern, just as the impersonal im- pact hypothesis would predict. Interpersonally mediated unemployment infor- mation represents the one inconsistent finding, demonstrating influence in two out of three cases.

These overall patterns suggest that personal experiences may indeed have

'4Additional exogenous variables in the equations included interpersonally mediated informa-

tion, demographic variables, party identification, retrospective assessments of family financial status,

the unemployment rate in the respondent's county of residence, and indicators of change in unemploy-

ment within that county (see Table 2 for details).

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

* * I-~~~~~~4-u

00 00 -w~

00 oE .

0"~~~~~~~~~~~~~~~~~~~~1

00 r

0 r~~~~~~~~~~ - ~ ~

N '.0~~~~~~0

n 0 0 0 W I ' (A

131 I~~~~~~~~~~~~~~( 00 I I

00~~~~~~

IN~~~~~

I.-~~~~~~~~~~~~~~~~~~~Z

C * 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~-

*1 * 4 . o ~~~~~~~~~~~~~' -'.~~~~~~~~~~~~~~~~~i UU ~~~~~~~~~~~~~r

*0 0 '.0 00 ~ U0

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

498 Diana C. Mutz

an indirect influence on political evaluations at the state level by means of their influence on personal and ultimately social concerns. At the national level, how-

ever, the path that translates these concems to political significance is incom- plete.

The lack of connection between personal and social concern jibes well with

many previous studies that find personal experiences and collective perceptions weakly, if at all, related. At the national level, the level of most research on sociotropic and pocketbook voting, there is not even an indirect path from per- sonal experiences to political evaluations. This may overstate the case somewhat, since the subjective personal problem measures do approach significance in some

analyses. Still, the overall pattern is clear: the experiences of others that reach one through mass and interpersonally mediated communication are weighted much more heavily in forming retrospective evaluations than are one's own per-

sonal experiences. It is mediated information, rather than information drawn from personal experiences, that is ultimately more important in bringing about

the electoral effects of changes in unemployment at the national level. At the state level, personal experiences may play a more important role because social per- ceptions are to some extent generalized from personal-level perceptions.

Weighting of Personal and Collective Perceptions

The pattern of results in the fall and spring waves squares well with the changes in unemployment coverage that occurred during the study. Mass media coverage of unemployment went up considerably between waves 1 and 2, most likely due to the general media emphasis on economic issues brought on by "Black Monday," 19 October 1987. Subjective measures of the emphasis given to employment-related news increased by 70%; the number of column inches nearly doubled from an average of nine inches per sample of front page news in the spring to 17 inches in the fall.

Accordingly, the generalization of personal concern to state-level percep- tions is much stronger in the first wave than the second wave, just as the theoret- ical model would lead one to expect; the closeness of the relationship between personal and social levels of judgment fluctuates according to the amount of out- side information that reaches people through news coverage; mass mediated in- formation is, after all, the only variable in the model capable of distinguishing social from personal judgments.

As personal and social levels of judgment became increasingly distinct, the weighting of these two factors in influencing attitudes toward incumbents also shifted. As a reexamination of Table 1 indicates, the weighting of perceptions of unemployment as a personal problem in predicting incumbent evaluations de- creased considerably from wave 1 to wave 2 while the importance of social per- ceptions in evaluating incumbents increased. Although personal experiences did

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 499

not have significant direct effects on incumbent evaluations in either wave, they came closest to having an impact during the first wave, again the time when media coverage was least.

The sociotropic priming effect. To examine further the role of mass media in

influencing the weighting of personal and social judgments, Table 3 shows these patterns separately for regular newspaper readers as opposed to occasional or nonreaders. Here the relationships become even more pronounced. As the top half of the table demonstrates, collective perceptions at either state or national levels clearly overwhelm the influences of personal concerns. On the other hand, among occasional and nonreaders, personal concerns, if anything, influence in- cumbent evaluations. It should be noted that this pattem is not simply a function of greater personal experiences with joblessness among the low readership group. These analyses included personal experiences, along with community

employment context and a host of other controls.

Table 3. Weighting of Personal and Collective Perceptions in Evaluating

Incumbent Performance by Newspaper Readership

Spring 1987 Fall 1987. Fall 1987

Governor Governor President

Regular readers:"

Personal problem .17 (.19) .03 (.23) - .05 (.28) State/national

problem .30** (.16) .49* (.23) .63* (.27)

R 2 .20 .30 .43

Occasional and

nonreaders:"

Personal problem .28* (.12) .08 (.17) .50* (.21)

State/national

problem .06 (.12) .18 (.19) .34 (.26)

RI2 .14 .25 .33

Note: Table 3 represents the results of six individual OLS regressions. In addition to the indicators of personal- and social-level concern shown here, each equation included demographic variables,

party identification, community context, and all of the information sources appearing in the general

model. Entries are unstandardized beta coefficients with standard errors in parentheses.

"Regular readers were defined as those who read a newspaper seven days a week. This group com-

prised 53% of the sample in the spring 1987 wave and 44% of the sample in the fall 1987 wave. The

remainder were occasional and nonreaders.

*p < .05; **p < .10, two-tailed tests.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

500 Diana C. Mutz

A comparison of coefficients in the top and bottom of Table 3 further illus-

trates that the weights attached to personal unemployment concems are consist- ently greater among occasional and nonreaders while precisely the opposite is

true for collective unemployment concerns; here, coefficients are consistently larger among regular readers.

The explanatory power of these models is also worth noting. Across all

three columns, the regular readers register an r-square value five to 10 percentage

points higher than the low readership group. Thus, the model accounts for in- cumbent evaluations better among regular readers than for others.

Clearly, media exposure heightens the salience of social-level judgments as well as informing collective evaluations. The former effect may be seen as a

variant of Iyengar and Kinder's (1987) "priming" effect; by focusing on some issues more than others, media increase the importance attached to these issues

in evaluating political figures. In this case, however, the effect is not simply the priming of all considerations surrounding an issue but rather "sociotropic prim-

ing": the priming of collective perceptions and a deemphasis on personal con- cerns. By heightening the importance of social concems to political evaluations and decreasing the importance of personal concerns, mass media contribute to the depoliticization of personal experience.

The collapsed model. In the face of uncertainty or little knowledge of state or national conditions, this model should collapse upon itself so that collective perceptions would become largely projections of personal experience in the ab- sence of other information sources. It was possible to test this idea with the sec-

ond wave of data by utilizing questions that asked respondents to estimate the current state and national unemployment rates (see Appendix B for wording). I

used these questions to identify the subjectively unknowledgeable; that is, those who knew they had no sense of collective conditions, and who demonstrated this

through their unwillingness even to venture a guess as to the state of unemploy- ment conditions after several probes. A full 37% of the sample fell into this category.

With a dummy variable to represent subjective unemployment knowledge, the interaction between personal experience and subjective unemployment knowledge in Table 4 represents the effects of personal experiences on percep- tions of national unemployment conditions specifically for those who do not have any idea as to current unemployment conditions and who are aware that they are unknowledgeable. While, in the aggregate, personal experiences do not enter into perceptions of overall employment conditions, the significant interaction term in Table 4 indicates that personal experiences with unemployment do influ- ence perceptions of national unemployment trends, specifically among those who have no definite ideas about national unemployment conditions. The general

pattem for national assessments in Tables 1 and 2 may be true of the aggregate, but individual differences in the awareness of collective economic conditions

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 501

Table 4. Subjective Unemployment Knowledge: Effects of Personal Experiences on

Perceptions of Unemployment as a National Problem

Standard

Variables beta Error

Personal problem

(better) .14 (.06)* Information sources:

Subjective knowledge

Personal experience .12 (.16)

Personal experience by

subjective unemployment

knowledgea -.20 (09)*

Interpersonally mediated

information about

unemployment -.14 (.06)* Exposure to newspaper

coverage of employment

trends (bad news) -.02 (.01)*

Community context: .06 (.04) Party identification:

Republican .29 (.14)*

Democrat -.19 (.13) Demographic controls:

Age -.02 (.04)

Gender (female) - .20 (.11) Race (nonwhite) .03 (.25)

Income .01 (.05)

Education .05 (.05)

R2 .23

Note: Entries are unstandardized betas with high score indicating conditions are perceived to be improving.

"This variable represents the effects of personal experience with unemployment for those unwilling

to estimate national unemployment conditions; that is, the effects of personal experience for those subjectively unknowledgable about national employment conditions.

*p < .05; **p < .01; ***p < .001, two-tailed tests.

suggest that personal experience may still play an important role among signifi- cant subgroups of the population. Although people may not readily translate per- sonal experiences to national concerns, in the absence of other sources of infor-

mation, people will default to this readily available source of knowledge. In one sense, this finding replicates the analyses of Conover and colleagues (1986), who suggest that the weighting of personal experiences in forming social perceptions

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

502 Diana C. Mutz

depends upon knowledge or, more appropriately, a subjective sense of knowledge of social conditions.

Discussion

Reading in a newspaper that the unemployment rate has increased by a few

percentage points seems a fundamentally different experience from receiving a pink slip that indicates that one has lost a job. Losing a job has an inescapable

impact on an individual's everyday life; on the other hand, this study suggests

that reading about unemployment in a newspaper may have greater consequences

for U.S. political life.

Although they may be limited by time and locale,'5 these findings have im- plications for the study of economic influences on voting, and for analyses of

mass media's role in the political process. First, these findings suggest that the

answer to the question of what kind of information sustains the political impact of unemployment trends depends on the level of office being considered and on characteristics of the individual's information environment. At the presidential level, the general answer to this question clearly is mediated information. At the

state level, personal perceptions influence social perceptions, and thus personal experiences indirectly enter into the process as well. Even at the national level, personal experiences may enter into these evaluations in the absence of other information. Just as Weatherford (1983a, 162) is correct in suggesting that "the

dilemma of choosing between personal and national referents is more apparent than actual," the dilemma of choosing between personal experiences and me-

diated sources of information as a basis for political decisions is also contrived.

Contrary to the claims of Kramer (1983), these results demonstrate that individual-level variations in perceptions of unemployment conditions are, in fact, meaningful; they primarily demonstrate differences in the amount and type of information available to people. Those with much information from mass me- dia sources will have perceptions of unemployment consistent with media pre- sentations, while those relying on more parochial sources will have perceptions that reflect their immediate environments. These results further confirm the hy-

'5For most purposes, a statewide sample does not pose generalizability problems for this kind

of study, since hypotheses are about the processing of political information that is unlikely to differ

systematically from state to state. But using a statewide sample does mean that the conclusions may

be bound to the particular employment conditions found in the state at that particular time. By most

standards, unemployment was low during the period of this study. Indiana's rate of unemployment

had closely paralleled the national average for several years, including the period of this study. More-

over, the advantages of a statewide sample clearly outweigh the problems when the purpose is to look

at variation in people's information environments. Studies using national samples typically have been

forced to aggregate media content measures over time from a single, metropolitan source, such as the

New York Times (see, e.g., Behr and Iyengar 1985; Funkhouser 1973a, 1973b; MacKuen and Coombs

1981). Such aggregation risks the same ecological fallacy that has slowed understanding of economic

influences on voting.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 503

potheses of Weatherford (1983a) and Conover and colleagues (1986) that per-

sonal experiences are influential information sources among unknowledgeable

people, but these findings also extend this general argument by demonstrating the sources of social perceptions among those who are knowledgeable.

The case against personal experiences in the literature on economic influ- ences on voting has probably been overstated. Personal experiences do influence assessments of social conditions among the substantial proportion of the popula- tion that lacks altemative sources of information. Furthermore, when personal and social judgments move closer together in the absence of outside information, it becomes increasingly meaningless to distinguish between personal and social

impacts, since one is simply an extension of the other. When a link does exist

between personal and social levels of judgment as in the state-level model, per- sonal experiences may influence social perceptions and thus, indirectly, political evaluations as well. Finally, even personal perceptions themselves, when mea- sured subjectively, appear to have the potential for relatively weak, but signifi- cant direct effects on evaluations of incumbents.

On the whole, however, social perceptions and mediated information matter more. This is especially true when one takes into consideration the fact that the

highly involved people who follow political news and form attitudes on a soci- otropic basis are also most likely to be the ones voting. At the national level in

particular, reading about unemployment in the newspaper or hearing about it from friends with employment problems may indeed have greater political con- sequences than personally losing a job. The closer to home the political contest, however, the stronger one would expect the personal-social link to become.

Perhaps the most surprising finding from this study in light of the conven- tional wisdom on mass media effects is that mediated information has any influ-

ence at all with an issue that so clearly affects individuals' everyday lives as unemployment does. The findings of this study suggest the exact inverse of what media dependency theories would suggest. Instead of arguing that mediated in- formation sources will have effects in the absence of more personal information, this study suggests that for lack of more representative, social-level information from mediated sources, personal experiences will inform these judgments.

In addition to serving to inform social-level judgments, mass media also serve as an obstacle to the politicization of personal experience. Mass media coverage primes the sociotropic side of the general model and decreases the im- portance of the pocketbook path. This consistent pattern appears when we look at changes over time as media coverage increased and when we examine high- and low-exposure subgroups of these samples.

Understanding the normative implications of these findings is complex be- cause mass media are functioning in two different capacities with respect to un- employment, each with different implications for the economic accountability of incumbent leaders. On the one hand, mass media coverage of unemployment

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

504 Diana C. Mutz

trends encourages accurate perceptions of social conditions. 16 This is a valuable function for mass media to serve, especially when citizens are relying primarily

on social-level perceptions as a basis for political decisions. At the same time, mass media are an obstacle to the politicization of personal experiences because

they provide a steady flow of information that establishes a social world beyond

one's personal experiences and interpersonal contacts. Widening the gap be-

tween personal and social judgments may be dysfunctional to the extent that people's social perceptions become independent of their aggregated personal ex- periences, and democratic accountability breaks down. Those who are not ex- posed enough to mass media to have accurate perceptions of state or national conditions may punish incumbents for economic declines that have not truly oc-

curred, or reward incumbents for economic improvements that have no basis in collective individual realities.

Manuscript submitted 31 August 1989

Final manuscript received 3 April 1991

APPENDIX A

Sampling and Data Collection Procedures

Data collection was carried out by professional staff interviewers of Walker Research, Inc., as

part of a biannual statewide poll. Telephone surveys were conducted between 1 and 5 June for the first

wave, and between 7 and 11 December for the second wave. A random digit computer-generated

sample was used with the number of telephone exchanges listed proportionate to the state population.

An evening and weekend calling pattern was utilized with each number receiving three attempts. All

respondents were at least 18 years old. The number of interviews conducted in each county was

roughly proportionate to the county's percentage of the state population.

APPENDIX B

Wording of Survey Questions

Personal experience: "In the past year have you or anyone in your family been laid off or had

trouble finding a job?" (No; Yes, respondent; Yes, family member). Coded (0) if no one in the imme- diate family has been unemployed, (1) if respondent or someone in family has been unemployed.

Interpersonally mediated unemployment experiences: "How often do other people talk to you about their employment problems, that is having trouble finding or keeping a job? Would you say they

talk to you about job security or unemployment problems everyday, 3 or 4 times a week, once or twice a week, or less often than that?" Coded as everyday (4), 3 or 4 times a week (3), once or twice a week (2), or less often than that (1).

'6Using the questions about estimated state and national unemployment rates as dependent var- iables, probit analysis indicated that daily newspaper reading did, in fact, encourage more accurate perceptions of unemployment conditions at both state and national levels, even beyond the effects of education and other demographic indicators. This probit regression included the three information source variables, party identification, race, gender, age, education, income, and the county unem- ployment rate in the respondent's community. Gender and newspaper exposure accounted for the bulk of variance in unemployment knowledge.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 505

Unemployment as a statelnational problem: State level: "Let's talk about the employment situ- ation here in Indiana. Would you say that over the past year people across our state have had a harder

time finding enough work, an easier time, or have things stayed about the same? Is that a little harder/

easier or a lot harder/easier?" Coded on a 5-point scale from a lot harder (1) to a lot easier (5). "Not

sure" and "Don't know" recoded as (3). National level: "And what about the United States as a whole.

Would you say that over the past year people across the United States have had a harder time finding

enough work, an easier time, or have things stayed about the same? Is that a little harder/easier or a lot harder/easier?" (Wave 2 only). Coded on a 5-point scale from a lot harder (1) to a lot easier (5). "Not sure" and "Don't know" recoded as (3).

Unemployment as a personal problem: "How about you, or people in your own household. Over the past year, have you been more worried about finding a job or keeping the one you have, less

worried about these things, or have things stayed about the same? Is that a little more/less worried or

a lot more/less worried?" Coded on a 5-point scale from a lot more worried (1) to a lot less worried

(5). "Not sure" and "Don't know" coded as (3).

Approval of incumbent administrations: "Overall do you approve or disapprove of the way

(Name of governor/lt. governor) is doing his job as ? Is that strongly approve/disapprove or

somewhat approve/disapprove?" Additive index comprised of two 5-point approve-disapprove scales.

Same questions repeated for presidential incumbents to form comparable scales corresponding to

approval of state/national incumbent administrations.

Exposure to mass media coverage of employment trends: (A) Newspaper identification: "What

one newspaper do you read most often?" Interviewers coded name and city of publication of news- paper. For each newspaper named by a respondent(s), a systematic content analysis was done of front pages appearing during two reconstructed weeks from the two months preceding each survey. (B)

Frequency of exposure: "How many days in the past week did you read the front page news in a

newspaper?" 0 to 7 raw scale recoded to 3-point scale representing nonreaders (0), occasional readers

(1), and regular, 7-day a week readers (2). Party identification: "Generally speaking, do you think of yourself as a Republican, a Demo-

crat, an Independent or of some other political party?" Two party identification measures created representing whether respondent was Republican (1) or not (0), and Democrat (1) or not (0).

Unemployment knowledge: "Do you happen to know what the current unemployment rate is in the state of Indiana/the United States?" (If at first don't know, probe) "Could you take a guess?"

Answers rounded to nearest whole number. Answers were counted as correct (1) if the respondent

named a number that had been the actual state or national unemployment rate within the previous year

(from 5% to 8%, when rounded), and incorrect (0) if the respondent was outside of this range or refused to guess.

Age: "Now I'm going to read some'age groups. Please tell me into which group your age falls. Would that be 18-24 (1); 25-34 (2); 35-44 (3); 45-54 (4); 55-64 (5); 65 and over (6)."

Education: "What was the last grade of school that you, yourself, completed?" Grade school or

less-grades 1-8 (1); some high school-grades 9-11 (2); graduated high school (3); some college/ technical school (4); graduated college (5); postgraduate work (6).

Income: "Which of the following best represents your total annual household income? Under

$10,000 (1); $10,000 to $20,000 (2); $20,000 to $30,000 (3); $30,000 to $40,000 (4); $40,000 or more (5)."

Race: "In order to ensure that all groups are properly represented in this survey are you white,

black, Hispanic American, American Indian, or some other nationality?" Coded as (0) for white, (1) for nonwhite.

Gender: Recorded by interviewer as male (1) or female (2).

REFERENCES

Adoni, Hannah, and Akiba A. Cohen. 1978. "Television Economic News and the Social Construction

of Economic Reality." Journal of Communication 28:61-70.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

5o6 Diana C. Mutz

Ball-Rokeach, Sandra J., and Melvin L. DeFleur. 1976. "A Dependency Model of Mass Media Ef-

fects." Communication Research 1:3-21.

Becker, Lee B., Maxwell E. McCombs, and Jack M. McLeod. 1975. "The Development of Political

Cognitions." In Political Communication: Issues and Strategies for Research, ed. Steven H.

Chaffee. Beverly Hills: Sage.

Behr, Roy L., and Shanto Iyengar. 1985. "Television News, Real World Cues, and Changes in the

Public Agenda." Public Opinion Quarterly 49:38-57.

Benton, Marc, and P. Jean Frazier. 1976. "The Agenda-Setting Function of the Mass Media at Three

Levels of Information Holding." Communication Research 3:261-74.

Brody, Richard A. 1977. "Stability and Change in Party Identification: Presidential to Off-Years."

Presented at the annual meeting of the American Political Science Association, Washington, DC.

Chaffee, Steven H. 1972. "The Interpersonal Context of Mass Communication." In Current Perspec-

tives in Mass Communication Research, ed. F. Gerald Kline and Phillip Tichenor. Beverly Hills:

Sage.

. 1982. "Mass Media and Interpersonal Channels: Competitive, Convergent, or Complemen-

tary?" In InterlMedia: Interpersonal Communication in a Media World, ed. Gary Gumpert and

Robert Cathcart. New York: Oxford University Press.

Chaffee, Steven H., and Diana C. Mutz. 1988. "Comparing Mediated and Interpersonal Communi- cation Data." In Advancing Communication Science: Merging Mass and Interpersonal, ed. Rob- ert P. Hawkins, John M. Wiemann, and Suzanne Pingree. Newbury Park, CA: Sage.

Chubb, John E. 1988. "Institutions, the Economy, and the Dynamics of State Elections." American Political Science Review 82:133-54.

Cohen, Jeremy, Diana C. Mutz, Vincent Price, and Albert C. Gunther. 1988. "Perceived Impact of

Defamation: An Experiment on Third Person Effects." Public Opinion Quarterly 52:161-73. Conover, Pamela J. 1985. "The Impact of Group Economic Interests on Political Evaluations." Amer-

ican Politics Quarterly 13:139-66.

Conover, Pamela J., Stanley Feldman, and Kathleen Knight. 1986. "Judging Inflation and Unemploy- ment: The Origins of Retrospective Evaluations." Journal of Politics 48:565-88.

Converse, Philip E., and Gregory B. Markus. 1979. "Plus ga Change ... : The New CPS Election Study Panel." American Political Science Review 73:32-49.

Davison, W Phillips. 1983. "The Third Person Effect in Communication." Public Opinion Quarterly

47:1-15.

Erbring, Lutz. 1980. "Media Monitoring and Public Opinion Change." Presented at the annual meet-

ing of the American Political Science Association, Washington, DC. Erbring, Lutz, Edie N. Goldenberg, and Arthur H. Miller. 1980. "Front-Page News and Real World

Cues: A New Look at Agenda-Setting by the Media." American Journal of Political Science 24:16-47.

Eulau, Heinz, and Michael S. Lewis-Beck. 1985. Economic Conditions and Electoral Outcomes: The United States and Western Europe. New York: Agathon Press.

Eyal, Chaim H. 1981. "The Roles of Newspapers and Television in Agenda-Setting." Mass Commu- nication Review, Yearbook 2. Beverly Hills: Sage.

Eyal, Chaim H., James P. Winter, and William F. DeGeorge. 1981. "The Concept of Time Frame in Agenda-Setting." Mass Communication Review, Yearbook 2. Beverly Hills: Sage.

Fiorina, Morris P. 1981. Retrospective Voting in American National Elections. New Haven: Yale University Press.

Funkhouser, G. Ray. 1973a. "The Issues of the Sixties: An Exploratory Study in the Dynamics of Public Opinion." Public Opinion Quarterly 37:62-75.

. 1973b. "Trends in Media Coverage of the Issues of the '60s." Journalism Quarterly 50:533-38.

Gordon, Margaret T., and Linda Heath. 1981. "The News Business, Crime, and Fear." In Reactions to Crime: Individual and Institutional Responses, ed. Dan A. Lewis. Beverly Hills: Sage.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

MASS MEDIA AND DEPOLITICIZATION 507

Graber, Doris A. 1984. Processing the News. New York: Longman.

Haight, Timothy, and Richard A. Brody. 1977. "The Mass Media and Presidential Popularity." Com-

munication Research 4:41-60.

Hawkins, Robert P., and Suzanne Pingree. 1982. "Television's Influence on Social Reality." In Tele-

vision and Behavior, vol. 2, ed. David Pearl, Lorraine Bouthilet, and Joyce Lazar. Rockville,

MD: Government Printing Office.

Hinckley, Barbara, Richard Hofstetter, and John Kessel. 1974. "Information and the Vote: A Com-

parative Election Study." American Politics Quarterly 2:131-58.

Iyengar, Shanto, and Donald R. Kinder. 1987. News That Matters. Chicago: University of Chicago

Press.

Jennings, M. Kent, and Harmon Zeigler. 1970. "The Salience of American State Politics." American Political Science Review 64:523-35.

Katz, Elihu, and Paul F. Lazarsfeld. 1955. Personal Influence. Glencoe, IL: Free Press.

Kiewiet, D. Roderick. 1983. Macroeconomics and Micropolitics: The Electoral Effects of Economic

Issues. Chicago: University of Chicago Press.

Kinder, Donald R. 1981. "Presidents, Prosperity, and Public Opinion." Public Opinion Quarterly 45:1-21.

Kinder, Donald R., and D. Roderick Kiewiet. 1979. "Economic Discontent and Political Behavior:

The Role of Personal Grievances and Collective Economic Judgments in Congressional Voting."

American Journal of Political Science 23:495-527.

. 1981. "Sociotropic Politics: The American Case." British Journal of Political Science

11:129-61.

Kinder, Donald R., Steven J. Rosenstone, and J. Mark Hansen. 1983. "Group Economic Well-Being

and Political Choice." Pilot Study Report to the 1984 NES Planning Committee and NES Board.

Klorman, Ricardo. 1978. "Trend in Personal Finances and the Vote." Public Opinion Quarterly 42:31-48.

Kramer, Gerald H. 1983. "The Ecological Fallacy Revisited: Aggregate- versus Individual-level Find- ings on Economics and Elections, and Sociotropic Voting." American Political Science Review 77:92-111.

Lazarsfeld, Paul F., Bernard Berelson, and Hazel Gaudet. 1944. The People's Choice. New York: Columbia University Press.

McClure, Robert D., and Thomas E. Patterson. 1976. "Setting the Political Agenda: Print vs. Net-

work News." Journal of Communication 26:23-28.

McCombs, Maxwell E., and Donald L. Shaw. 1972. "The Agenda-Setting Function of the Mass

Media." Public Opinion Quarterly 36:176-87.

MacKuen, Michael B., and Stephen L. Coombs. 1981. More Than News: Media Power in Public

Affairs. Beverly Hills: Sage. McLeod, Jack M., Lee B. Becker, and James E. Byrnes. 1974. "Another Look at the Agenda-Setting

Function of the Press." Communication Research 1: 131-66.

Meier, Kenneth J. 1975. "Party Identification and Vote Choice: The Causal Relationship." Western Political Quarterly 28:496-505.

Mutz, Diana C. 1989. "The Influence of Perceptions of Media Influence." International Journal of Public Opinion Research 1:3-24.

Page, Benjamin I., and Calvin C. Jones. 1979. "Reciprocal Effects of Policy Preferences, Party Loy- alties, and the Vote." American Political Science Review 73:1071-89.

Page, Benjamin I., Robert Y. Shapiro, and Glenn R. Dempsey. 1987. "What Moves Public Opinion?" American Political Science Review 81:23-44.

Palmgreen, Phillip, and Peter Clarke. 1977. "Agenda Setting with Local and National Issues." Com-

munication Research 4:435-52.

Peltzman, Sam. 1987. "Economic Conditions and Gubernatorial Elections." American Economic Re- view 77:293-97.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms

5o8 Diana C. Mutz

Piereson, James E. 1977. "District Economic Conditions and Congressional Elections." Presented at

the annual meeting of the Midwest Political Science Association, Chicago.

Pollard, W. 1978. "Effects of Economic Conditions on Elections: A Study Controlling for Political

Variables." Presented at the annual meeting of the Public Choice Society, New Orleans.

Rosenstone, Steven J., J. Mark Hansen, and Donald R. Kinder. 1986. "Measuring Change in Personal

Economic Well-being." Public Opinion Quarterly 50:176-92.

Schlesinger, Joseph A. 1966. Ambition and Politics: Political Careers in the United States. Chicago:

Rand McNally.

Schlozman, Kay L., and Sidney Verba. 1979. Injury to Insult. Cambridge: Harvard University Press.

Sears, David O., and Richard R. Lau. 1983. "Inducing Apparently Self-Interested Political Prefer-

ences." American Journal of Political Science 27:223-52.

Shively, W Phillips. 1977. "Information Costs and the Partisan Life Cycle." Presented at the annual

meeting of the American Political Science Association, Washington, DC.

Skogan, Wesley G., and Michael G. Maxfield. 1981. Coping with Crime: Individual and Neighbor- hood Reactions. Beverly Hills: Sage.

lipton, Leonard, Roger Haney, and James Baseheart. 1975. "Media Agenda-Setting in City and State

Election Campaigns." Journalism Quarterly 52:15-22.

'ryler, Tom R. 1980. "Impact of Directly and Indirectly Experienced Events: The Origin of Crime- Related Judgments and Behaviors." Journal of Personality and Social Psychology 39:13-28.

. 1984. "Assessing the Risk of Crime Victimization: The Integration of Personal Victimization

Experience and Socially Transmitted Information." Journal of Social Issues 40:27-38.

Tyrler, Tom R., and Fay L. Cook. 1984. "The Mass Media and Judgments of Risk: Distinguishing Impact on Personal and Societal Level Judgments." Journal of Personality and Social Psychology

47:693-708.

Weatherford, M. Stephen. 1983a. "Economic Voting and the 'Symbolic Politics' Argument: A Rein-

terpretation and Synthesis." American Political Science Review 77:158-74.

. 1983b. "Evaluating Economic Policy: A Contextual Model of the Opinion Formation Pro-

cess." Journal of Politics 45:866-88.

Weaver, David H., Doris A. Graber, Maxwell E. McCombs, and Chaim H. Eyal. 1981. Media Agenda Setting in a Presidential Election. New York: Praeger.

Zucker, Harold G. 1978. "The Variable Nature of News Media Influence." In Communication Year- book 2, ed. Brent D. Ruben. New Brunswick, NJ: Transaction Books.

This content downloaded from 129.219.247.33 on Tue, 22 Nov 2016 19:15:23 UTC All use subject to http://about.jstor.org/terms