Media priming effects: accessibility, association, and activation.
Drawing from both of these research domains, we argue that media coverage of political issues has the potential to activate particular cognitions, which then may be used in evaluating other elements of the political environment. Specifically, media framing of issue information--by highlighting some dimensions of issues while excluding other aspects--seems likely to foster priming effects. In this article, we posit that framing issues in a manner that focuses on their moral or ethical dimensions can prime voters to (1) make attributions about candidate character, particularly integrity, and/or (2) evaluate other political issues in ethical terms. However, these relationships should be mediated by the specific features of any issue, the particular values emphasized in media coverage, and individuals' cognitive networks, thus fostering different priming effects from different issues for different people. To examine these ideas, this research systematically altered a single issue--which varied in the types of values in conflict--across four otherwise constant political environments to examine how voters process, interpret, and use issue information while `voting' in a simulated election.
Two sub-populations expected to differ in their core values and the interconnection of those values with political attitudes were selected for study: evangelical Christians and university undergraduate students. These groups were chosen for three reasons: (1) evangelical Christians have demonstrated increasing prominence and power in recent U.S. elections; (2) there has been little systematic study of evangelical Christians' political cognitions, although many have speculated about their values and behavior in electoral contexts; (3) most important, evangelical Christians, as a group, seem much more likely than undergraduate students to have well-integrated mental frameworks in which ethical or moral values--a key component of the theory advanced in this article--are closely linked with political attitudes. Thus, utilizing the same experimental design with these two populations provides a more thorough examination of our theory and allows consideration of differences in priming.
Insight into priming effects may be gained from research on information processing, particularly scholarship on construct accessibility, the spread of activation, and personal values.
Much political information processing research reflects a grounding in schema theory, which posits that people organize their perceptions of the environment into cognitive knowledge structures--i.e., mental clusters of information. Once activated, schema facilitate and shape the processing of information, thereby providing the raw materials upon which individuals form evaluations and come to understand their social milieu (Graber 1988, Fiske and Taylor 1991). According to this perspective, individuals do not draw upon all applicable cognitive constructs to guide information processing; rather, people tend to over-sample `accessible' schema (Higgins and King 1981, Higgins and Bargh 1987). For familiar objects, the relevant schema are highly accessible; however, when objects are more obscure, research suggests that contextual cues--such as media framing of issues--may activate relevant cognitive structures to guide information processing and the construction of attitudes (Tourangeau and Rasinski 1988, Lau 1989, Zaller 1992, Zaller and Feldman 1992).
According to Wyer and Srull (1989), cognitive structures recently or frequently activated by contextual cues remain on top of the mental bin, making them highly accessible for at least a period of time (see also Srull and Wyer 1989). These `available' schema, if judged to be applicable, may alter the basis for evaluating seemingly unrelated objects because judgments often `depend less on the entire repertoire of people's knowledge and more on which aspects of their knowledge happen to come to mind' (Iyengar and Kinder 1987, p. 64). For example, a growing body of political communication research indicates that mass media emphasis on particular political issues increases the accessibility of certain ideas for individuals, which then shape the factors considered in evaluations of political candidates (Iyengar et al. 1982, Iyengar and Simon 1994, Krosnick and Kinder 1990 Johnston et al. 1992, Krosnick and Brannon 1993, Mendelsohn 1996).
While considerable research asserts that the frequency and recency of information use influence its accessibility, there is less agreement on how these bits of information are arrayed in the mind. Fazio (1986, 1989), focusing on intra-attitudinal structure, views an attitude as an association in memory between the attitude object and the evaluation of the object. The strength of the link between object and evaluation is thought to influence the accessibility of the attitude when faced with the object. Scholars concerned with inter-attitudinal structures make a similar assertion, arguing for a conception of memory as a network of connected cognitive structures (Anderson and Bower 1973, Collins and Loftus 1975, Anderson 1985). These scholars maintain that any one concept is associated with other constructs when encoded in memory, and that the linkages between constructs are strengthened each time they are activated in tandem. Further, as the number of separate linkages to any particular construct increases, so does the likelihood that it will be activated indirectly due to an `implicational relation' (Judd and Krosnick 1989).
As a result, Berkowitz and Rogers (1986, pp. 58-9) argue, `When a thought element is activated or brought into focal awareness, the activation radiates out from this particular node along the associative pathways to other nodes', thereby increasing the probability that related constructs will come to mind. Drawing from these perspectives, theories of `spreading activation' contend that the stronger or more numerous the mental pathways between constructs, the greater the chance that thoughts activated to process information about one construct will cascade through memory to other constructs, influencing subsequent evaluations and the formation of impressions (Lodge and Stroh 1993).
Research on accessibility and the spread of activation, then, suggests that priming effects may depend on a combination of factors: the frequency or recency of activation and the inter-connectedness between distinct cognitions. Only a few scholars, however, have examined how cognitions activated by media coverage may spread to other constructs via mental pathways. For example, in research from an agenda-setting perspective, Schleuder, McCombs, and Wanta (1991) examined how linkages between mental constructs influence the retrieval of information about candidates from memory; their conclusions suggest that examination of spreading activation should be expanded to study other media related effects.
As many scholars argue, media do not merely serve an agenda-setting role in public discourse but, through selection and `framing' of news and opinion, are crucial in establishing the range of criteria for constructing, debating, and resolving social issues (Gitlin 1980, Graber 1988, Gamson 1992, Neuman et al. 1992, Shah et al. 1996). In particular, Ball-Rokeach and colleagues assert that emphasis on particular values in media coverage shapes policy debates about social issues as political groups attempt to articulate and legitimate why their choice is more appropriate or `moral' than their opponents' (Ball-Rokeach and Loges 1997, Ball-Rokeach et al. 1990). In this process, media constructions often probably serve as a heuristic for citizens, whose understanding of issues is `powerfully shaped' by the values involved (Tetlock 1989, p. 130, Rokeach 1973, Johnson and Eagly 1989, Boninger et al. 1995). For example, research indicates that voters form different psychological linkages with issues based upon the interaction of their core values with media coverage; in turn, these issue interpretations' shape information processing and judgment (Domke and Shah 1995, Shah et al. 1997).
Further, how people interpret issues often conceived in `social-moral' terms--such as abortion, gay rights, school prayer--may influence their evaluations of candidate character (see Page and Clelland 1978, Luker 1984, Moen 1984). Such moral Issues may be particularly symbolic for evangelical Christians (Lorentzen 1980, Wuthnow 1989). Many issues in the present political environment, however, are not as `easy' to evaluate because they are more complex and `means-oriented' (Carmines and Stimson 1980). For example, policy issues such as health care, the environment, and welfare are probably more open to differing interpretations because they have been discussed in both moral and economic terms (see Cook et al. 1994, Brown and Carmines 1995). For such issues, the particular values emphasized in media coverage are likely to play a critical role in voters' evalutions and information processing.
In this study, we posit that media presentation of political issues, by selecting and emphasizing certain values while excluding others, is likely to influence which cognitions are activated as voters evaluate a political environment. In this process, the particular features of any issue and the specific values emphasized in media coverage should both interact with individuals' cognitions, resulting in clear differences in priming effects.
We contend that media coverage of `social-moral' issues such as abortion or euthanasia may prime voters' evaluations of candidate character, in particular integrity, because positions on these issues are commonly considered symbolic of a candidate's deeper concerns and convictions. As a result, media coverage emphasizing such issues, which are primarily conceived in terms of morals and rights, is likely among certain voters to activate thoughts which then may be applied in evaluating components of a candidate's integrity such as morality, decency, and compassion. In comparison, attributions about candidate integrity should be less likely to occur when citizens encounter media coverage of issues regularly discussed in economic terms, since such issues (even if framed by media in ethical terms) are not considered by most voters to be symbolic of a candidate's core priorities and concerns.
We theorize that different priming effects may occur, however, when media coverage emphasizes issues that are more ambiguous-that is, they are not clearly moral or economic in nature. Such issues are discussed convincingly in either symbolic or instrumental terms by both politicians and media on a regular basis. Drawing upon conceptual distinctions developed by Shah, Domke, and Wackman (1996), we argue that, for many political issues, media framing focuses on either ethical or material concerns. Framing that focuses on ethical concerns emphasizes the dimensions of human rights, civil rights, religious morals, or personal principles present in the value conflict of an issue. Framing that focuses on material concerns emphasizes the tangible considerations grounded in economics, expedience, practicality, or personal self-interest present in the value conflict of an issue. In turn, research suggests that media framing of issues interacts with voters' values and predispositions to guide individual issue interpretations.(1)
We argue, further, that cognitions activated by media coverage of one issue may subsequently be used by voters to process information about additional issues. Such a `carryover' effect seems most likely to occur when individuals evaluate issues that share similar dimensions and, as a result, may be considered appropriate targets for the activated constructs. Thus, value-relevant thoughts brought to mind by ethical framing of one ambiguous issue, such as health care or welfare, would seem likely to foster ethical interpretations of other issues that contain a similar mix of ethical and material dimensions. In contrast, cognitions brought to mind by media coverage of social-moral issues should be less likely to influence voters' interpretations of more materially based issues, since ideas about these differing types of issues are not likely to be considered applicable to one another. Accordingly, we state research hypotheses 1 and 2:
H1: Individuals evaluating a political environment which includes a socialmoral issue (e.g., abortion or euthanasia) will be more likely to make attributions about candidate integrity than individuals evaluating a political environment which does not include a social-moral Issue.
H2: Individuals evaluating a political environment which includes an ambiguous issue framed in ethical terms (e.g., health care) will be more likely to form ethical interpretations of material issues than individuals evaluating a political environment which does not include an ambiguous issue framed in ethical terms.
This research also explores whether the complexity and inter-connectedness of individuals' mental structures mediates how the process of priming occurs. In considering the populations utilized in this study and the topics under consideration, evangelical Christians, as a group, seem much more likely than undergraduate students to have well-integrated mental frameworks for two reasons: (1) the evangelical Christians are, on average, considerably older; and (2) moral or ethical values--a key component of our theory--are more likely to be closely associated with certain political attitudes among evangelical Christians. Because of these differences, we expect to find differing patterns of priming between these two populations.
We posit that `spreading activation' priming effects should occur primarily among individuals who possess intersecting and highly integrated cognitive frameworks about the objects under consideration. Such tightly connected mental frameworks increase the likelihood that thoughts activated to process information about one construct will cascade through memory to other associated constructs. Among these individuals, then, media priming effects seem likely to occur via the following process: media coverage activates particular thoughts or ideas, which guide the evaluation of, first, the issue emphasized in media coverage, and then other constructs related in memory. Such an `indirect' process of spreading activation, in which a person's initial evaluation triggers similar evaluations along a cognitive pathway, should be reflected in an identifiable pattern or `trail' of evaluations in information processing.
In contrast, priming effects due primarily to 'construct accessibility' should occur among individuals whose mental frameworks are not as highly integrated. Without inter-woven pathways to direct active thoughts and ideas about objects under consideration, cognitions recently or frequently activated remain on top of one's mental bin, making them highly accessible for at least a period of time. Among these individuals, then, media priming effects seem likely to occur via the following process: media coverage activates particular thoughts or ideas, which then are used to evaluate the issue emphasized in media coverage and/or other constructs deemed applicable, though there may not be a correspondence between these two evaluations. In such `direct' priming, a person's initial evaluation does not determine subsequent evaluations, since both are based simply on which ideas come to mind; as a result, there should not be an identifiable pattern of evaluations in information processing. Accordingly, the final research hypothesis may now be stated:
H3: Individuals with highly integrated mental frameworks about the objects under consideration will be more likely to exhibit spreading activation priming than individuals with less-integrated mental frameworks about the objects under consideration.
This study, part of a broader research program, examines how citizens in distinctly different populations process media messages and evaluate issues and candidates in making voting decisions. In this study, 172 members of five evangelical Christian churches and 201 undergraduate students in a large U.S. midwestern city were presented copies of newspaper articles and a questionnaire.(2) Most respondents took 35-45 minutes to complete the materials.
The core of this research strategy is the controlled presentation of political environments. Each environment contained newspaper articles, written by a former professional journalist, that presented the contrasting views of three candidates on four issues (for examples of the articles and a copy of the questionnaire, see Shah et al. 1996).(3) Subjects were randomly assigned to one of four environments. All subjects received articles on three issues framed to emphasize material dimensions (education, economy, and government cuts), hereafter referred to as the `material' or `other' issues. Each political environment also included an article on a single manipulated issue, varied across the environment conditions.
The first environment contained the issue of abortion, framed in an ethical manner by pitting the sanctity of an unborn child's life against the personal liberty of a woman to control her body and destiny. The second environment contained the issue of euthanasia, framed in an ethical manner by pitting the sanctity of all life against individual decisions about the quality of one's own existence. The third environment contained the issue of health care, framed in an ethical manner by pitting personal responsibility to provide for one's self against equality and compassion in access to medical treatment. Varying these issues across conditions created two environments (abortion, euthanasia) with issues similarly 'moral' in nature but differing in their duration in U.S. politics, while the third environment (health care) contained an ethically framed version of a long-standing issue often discussed in a mix of material and ethical terms. The fourth environment, which serves as a baseline in this study, contained a different version of health care, framed to emphasize material dimensions by pitting the merits of the free market against the need for government intervention to control costs.
Several steps were taken to ensure that (1) priming effects across research conditions were due to media coverage of different issues, and (2) priming effects within research conditions were due to subjects' processing of issue information. Across political environments, all information was held constant except for varying the issues of abortion, euthanasia, health care (ethical frame), and health care (material frame). Within environments, subjects were provided with only four items of (identical or nearly identical) personal information about the candidates: gender, age, educational background, and occupational background. Candidates and issues were then rotated in each packet of articles to avoid order effects.
QUESTIONNAIRE AND CODING OF DEPENDENT VARIABLES
After reading the articles, subjects completed a questionnaire, which began by asking them to make a candidate choice, followed by a series of open-ended questions probing the factors considered in the decision. These questions were designed to assess specific decision-making strategies used by subjects in making a candidate choice. However, responses to these questions also were examined for the presence of attributions by subjects about candidate integrity. Although none of the questions focused specifically on evaluations of candidates, nevertheless a number of subjects formed impressions or drew conclusions about the candidates' integrity, particularly such factors as morality, convictions, and beliefs. Because the questionnaire did not explicitly examine media priming to candidate integrity, then, the amount of attributions actually made by subjects is likely to be under-estimated. These measurement limitations, however, should not interfere with any tests of the hypotheses.
Guided by research on voters' evaluations of candidate character (Kinder 1985, Iyengar and Kinder 1987, Graber 1993), responses were content analyzed for attributions about candidate integrity.(4) For evangelical Christians, two coders agreed on 166 of 172 subjects as having made an attribution about candidate integrity or not, yielding an inter-coder reliability coefficient of .96, which was determined to be 89 percent greater than by chance (see Scott 1955). For undergraduate students, two coders agreed on 181 of 201 subjects as having made an attribution about candidate integrity or not, yielding an inter-coder reliability coefficient of go, which was 64 percent greater than by chance. All remaining responses were discussed and then classified.
Individual interpretations of issues were measured by three open-ended questions that engaged subjects in a thought-listing procedure to tap how the issues related to their personal values, concerns about society, and personal life-situations. Each issue was coded as having received an ethical interpretation, material interpretation, a combination of these two, or as not mentioned. Issues were coded as receiving an ethical interpretation if the individual discussed the issue within the framework of human rights, civil rights, religious morals or personal principles. Issues were coded as receiving a material interpretation if the individual discussed the issue in terms of economics, expedience, practicality, or personal self-interest. Individuals could receive a coding of either ethical or material regardless of whether they discussed the issues in personal terms or in relation to the broader society. Guided by our theory and in an effort to maintain reasonable cell sizes for hypothesis testing, the small number of issue responses coded as having combinations of ethical and material interpretations were collapsed into the ethical category.
For evangelical Christians, two coders agreed on 618 of 688 individual-issue codings, producing an inter-coder reliability coefficient of .90, which was 80 percent greater than by chance. For undergraduate students, two coders agreed on 723 of 804 individual-issue codings, producing an inter-coder reliability coefficient of .90, which was 80 percent greater than by chance. All remaining individual-issue codings were discussed and then classified.
For analysis, subjects who did not mention an issue were combined with those who formed a material interpretation, thereby creating the categories of ethical interpretation or non-ethical interpretation for each issue. A manipulation check was run to examine if, as expected, the combination of issues and media emphasis upon ethical dimensions produced differences in subjects' interpretations of the manipulated issues. As expected, individuals in both populations were more likely to ascribe an ethical interpretation to the issues framed in ethical terms than to the issue framed in material terms (baseline environment). A one-way ANOVA using the Least Significant Differences (LSD) post-hoc test confirmed statistically significant differences between the first three environments and the baseline environment in both sub-populations (see Table 1).
TABLE 1 Subjects' interpretations of the manipulated issue in different political environments (percentages)
Political environment Abortion Euthanasia Health care (ethical frame) Issue interpretation Population 1: Evangelical Christians Non-ethical 28.3 41.9 51.2 Ethical 71.7 58.1 48.8 Totals 100 100 100 (n = 46) (n = 43) (n = 43) [chi square] = 32.3, d.f. = 3, p < .001 Population 2: Undergraduate students Non-ethical 32.0 52.0 63.5 Ethical 68.0 48.0 36.5 Totals 100 100 100 (n = 50) (n = 50) (n = 52) [chi square] = 33.2, d.f. = 3, p < .001 Health care (material Issue interpretation frame) Population 1: Evangelical Christians Non-ethical 87.5 Ethical 12.5 Totals 100 (n = 40) [chi square] = 32.3, d.f. = 3, p < .001 Population 2: Undergraduate students Non-ethical 87.8 Ethical 12.2 Totals 100 (n = 49) [chi square] = 33.2, d.f. = 3, p < .001
Using individuals' interpretations of education, economy, and government cuts, the variable `interpretation of the material issues' was constructed. To determine the overall amount of ethical interpretations of any of these three issues, which have often been discussed by media and politicians in material terms and were framed similarly in this study, subjects were coded as having an ethical interpretation if they interpreted as ethical any of the issues.
Hypothesis 1 predicted that individuals evaluating a political environment which includes a social-moral issue will be more likely to make attributions about candidate integrity than individuals evaluating a political environment which does not include a social-moral issue. To test the hypothesis, crosstabs were run in both research populations between the experimental conditions and whether or not subjects made attributions about candidate integrity. The results provide support for the hypothesis (see Table 2).
TABLE 2 Subjects' attributions about candidate integrity in different political environments (percentages)
Political environment Abortion Euthanasia Health care (ethical frame) Candidate integrity Population 1: Evangelical Christians No attributions 73.9 65.1 88.4 Attributions 26.1 34.9 11.6 Totals 100 100 100 (n = 46) (n = 43) (n = 43) [chi square] = 7.5, d.f. = 3, p = .056 Population 2: Undergraduate students No attributions 70.0 72.0 96.2 Attributions 30.0 28.0 3.8 Totals 100 100 100 (n = 50) (n = 50) (n = 52) [chi square] = 20.8, d.f. = 3, p < .001 Health care (material Candidate integrity frame) Population 1: Evangelical Christians No attributions 82.5 Attributions 17.5 Totals 100 (n = 40) [chi square] = 7.5, d.f. = 3, p = .056 Population 2: Undergraduate students No attributions 93.9 Attributions 6.1 Totals 100 (n = 49) [chi square] = 20.8, d.f. = 3, p < .001
In both populations, roughly 30 percent of subjects evaluating an environment containing a social-moral issue (abortion or euthanasia) made attributions about candidate integrity. When the social-moral issues were replaced in the environment by health care (regardless of the framing), the percent of subjects making attributions about candidate integrity was substantially lower--between 4 and 17 percent. A one-way ANOVA using the LSD post-hoc test showed significant differences among undergraduate students between the abortion and euthanasia environments and both health care environments; among evangelical Christians, statistical significance was achieved only between subjects receiving the euthanasia environment and subjects receiving the ethically framed health care environment.
It is noteworthy that attributions about candidate integrity--in both research populations--were highest among subjects receiving either abortion or euthanasia and lowest among subjects receiving the ethically framed version of health care. These results indicate that media emphasis upon the ethical dimensions of issues, alone, was not sufficient to prime subjects to make attributions about candidate integrity. Rather, a solid minority of subjects formed evaluations of candidate integrity only when media coverage included the social-moral issues abortion and euthanasia, which are commonly considered symbolic of a candidate's deeper concerns and convictions. These findings, then, provide support for hypothesis 1.
Hypothesis 2 predicted that individuals evaluating a political environment which includes an ambiguous issue framed in ethical terms (e.g., health care) will be more likely to form ethical interpretations of material issues than individuals evaluating a political environment which does not include an ambiguous issue framed in ethical terms. To test the hypothesis, crosstabs were run in both research populations between the experimental conditions and subjects' interpretations of the three unmanipulated `material' issues (see Table 3).
TABLE 3 Subjects' interpretations of the other issues in different political environments (percentages)
Political environment Abortion Euthanasia Health care (ethical Interpretation of other issues frame) Population 1: Evangelical Christians Non-ethical 84.8 86.0 67.4 Ethical 15.2 14.0 32.6 Totals 100 100 100 (n = 46) (n = 43) (n = 43) [chi square] = TS, d-f. = 3, p < .058 Population 2: Undergraduate students Non-ethical 92.0 86.0 82.7 Ethical 8.0 14.0 17.3 Totals 100 100 100 (n = 50) (n = 50) (n = 52) [chi square] = 5.5, d.f. = 3, n.s. Health care (material Interpretation of other issues frame) Population 1: Evangelical Christians Non-ethical 87.5 Ethical 12.5 Totals 100 (n = 40) [chi square] = TS, d-f. = 3, p < .058 Population 2: Undergraduate students Non-ethical 95.9 Ethical 4.1 Totals 100 (n = 49) [chi square] = 5.5, d.f. = 3, n.s.
As predicted, subjects evaluating the environment with the ethical frame of health care were the most likely to interpret at least one of the issues of economy, education, and government cuts in ethical terms. Among evangelical Christians, a one-way ANOVA using the LSD post-hoc test showed significant differences between the ethically framed health care environment, in which 33 percent of subjects formed an ethical interpretation of the material issues, and the other three environments, in which no more than 15 percent of subjects interpreted the other issues in ethical terms. Among students, the highest percentage of `carryover' also occurred in the health care ethically framed environment, but the only statistically significant difference found was between the health care environments: 17 percent of subjects receiving the ethical frame interpreted the other issues in ethical terms, compared to 4 percent of subjects receiving the material frame (p [is less than] .05).
These results indicate that value-relevant thoughts activated by media coverage of one ambiguous issue primed subjects to form value-based evaluations of other issues sharing similar dimensions. Individuals evaluating the ethically framed health care environment were the most likely to form ethical interpretations of economy, education, or government cuts. Differences across political environments were much stronger for evangelical Christians than undergraduate students. Although the results are far from conclusive, a `carryover' of ethically based cognitions to material issues may be less likely when subjects evaluate social-moral issues, perhaps because, as we theorized, ideas about these differing types of issues are not considered applicable to one another. The pattern of findings, then, provides some support for hypothesis 2.
The consistent pattern of results across the research populations for hypotheses 1 and 2 indicates that media coverage, at a minimum, activated certain value-relevant thoughts which were then applied by subjects to attitude objects deemed appropriate. Within-groups analysis is necessary, however, to gain insight into whether these priming effects were due to spreading activation along mental pathways or merely to the increased accessibility of certain ideas at the top of mental bins.
Hypothesis 3 predicted that spreading activation priming will be more likely to occur among individuals with highly integrated mental frameworks (i.e., evangelical Christians given the focus of this study) than among individuals with less-integrated mental frameworks (i.e., undergraduate students). When focusing `within the cognitive system of the individual' (Lavine et al. 1996, p. 298), one possible method to investigate the process of priming, which almost certainly occurs outside the conscious awareness of individuals, is to examine whether patterns are apparent in subjects' information processing and evaluations.
Our theory suggests that a key variable in any spreading activation is individuals' interpretations of the manipulated issues (abortion, euthanasia, or either version of health care). Specifically, we posit that, if there is spreading activation, thoughts activated in the formation of an ethical issue interpretation will trigger a pattern of ethically based evaluations that reflect linkages between mental constructs. Thus, evangelical Christians who form an ethical interpretation of the manipulated issue should become more likely to exhibit priming to either integrity attributions or ethical interpretations of the other issues, depending on the respective political environment. In contrast, there should not be a distinguishable pattern of evaluations if priming is due primarily to construct accessibility, since such effects, which we expect among undergraduate students, should occur based simply on which ideas come to mind, regardless of whether subjects initially form an ethical interpretation of the manipulated issue.
To explore these potential differences, we ran ANOVAs and crosstabs to examine the contribution to each priming effect of (1) the experimental manipulation, by isolating the political environment conditions in which each effect was maximized,(5) and (2) an ethical interpretation of the manipulated issues. The results of these analyses, arranged in Figures 1 to 4, clearly show differences between the populations in the process of priming.
Results in Figures 1 and 2 suggest that, among undergraduate students, the type of political environment encountered was the primary influence on both attributions about candidate integrity (F= 18.0, p [is less than] .001) and ethical interpretations of the other issues (F = 2.4, [is less than] .13). The lines in each figure indicate that an ethical interpretation of the manipulated issue was unrelated--either as a main effect or as part of an interaction-to both priming effects. These results, then, suggest that priming among these subjects was probably simply due to temporary construct accessibility.
[Figures 1-2 ILLUSTRATION OMITTED]
In contrast, the pattern of results in Figures 3 and 4 suggest that, among evangelical Christians, both interpretation of the manipulated issue and the type of political environment contributed to the priming effects, particularly the `carryover' to subjects' interpretations of the material issues. Three points deserve further interpretation.
[Figures 3-4 ILLUSTRATION OMITTED]
First, the distances between the lines indicate that, among evangelical Christians, interpretation of the manipulated issue had the strongest influence on both attributions about candidate integrity (F = 4.7, p [is less than] .05) and ethical interpretations of the other issues (F= 16.0, p [is less than] .001). Subjects with an ethical issue interpretation were consistently more likely to exhibit the priming effects than subjects without an ethical issue interpretation: this gap ranged from 6 percent to 38 percent, depending on the type of political environment. Second, the upward slope of the lines-particularly among subjects with an ethical issue interpretation-suggests that manipulation of the political environment produced priming effects. That is, subjects evaluating social-moral environments were the most likely to make attributions about candidate integrity (F=2.9, p [is less than] .10), while subjects exposed to An ethical framing of health care were the most likely to form ethical interpretations of the materially based issues (F= 8.5, p [is less than] .01). Finally, that the lines are not parallel indicates the priming effects occurred primarily when subjects (1) evaluated a particular political environment, emphasizing either social-moral or more ambiguous issues and (2) formed an ethical interpretation of the manipulated issue (the interaction was statistically significant only for priming to other issues, F = 3.9, P [is less than] .05). These results, then, strongly suggest that priming among evangelical Christians in this study was characterized by spreading activation that radiated from one cognition to other constructs via mental pathways.
These findings across two different sub-populations and four distinct political environments offer substantial insight into priming effects. Two consistent patterns emerged in analysis: (1) experimental tests indicated a clear pattern of different issues producing different effects, with some issues (i.e., social-moral) priming perceptions of candidate integrity and others (i.e., a more ambiguous issue with an ethical frame) priming ethical interpretations of more material issues; (2) within-groups analysis revealed differences between sub-populations in the process of priming, with only evangelical Christians exhibiting a pattern of linkages between ethical interpretations of the manipulated issues and other ethically based evaluations.
Tests of hypotheses 1 and 2 suggest that media coverage of political issues activates certain cognitions, which are then more broadly accessible for use by individuals in evaluating other applicable elements of the political environment. Media frames that focus on moral or ethical dimensions of issues appear to prime perceptions of candidate integrity or ethical interpretations of other, materially based issues, depending on the nature of the issue presented. These results, then, suggest that media priming effects may be contingent upon the specific features of an issue and the manner in which media frame issue coverage. Moreover, analysis of hypothesis 3 suggests that priming effects that appear similar on their surface may be substantially different when researchers look within the `black box' of human cognition. While experimental tests of hypotheses 1 and 2 reveal similar results among both evangelical Christians and undergraduate students, very different patterns emerged during within-groups analysis.
Data arranged in Figures 1-4 show that, when considering the contribution to priming of both the political environments encountered and individuals' interpretations of the manipulated issues, only evangelical Christians exhibited a string of ethically based evaluations. Indeed, among subjects in this population, the strongest effects were consistently provided by issue interpretation, indicating that this variable played a critical intervening role in initiating `indirect' priming. These results suggest that the priming effects among evangelical Christians were characterized by the spread of activation, triggered when individual cognitions activated by media content radiated outward to activate related nodes, thereby creating a `trail' or `string' of similar value-based evaluations along cognitive pathways. In considering the broader implications of this finding, individuals who are highly cognitively integrated seem likely to display clearly patterned and predictable priming effects, in the process making implicit connections triggered initially by an ethical issue framing.
In contrast, a similar trail of ethical evaluations was not found among undergraduate students. Rather, the results suggest that among these subjects the priming effects, due almost solely to the experimental manipulation, were characterized by the temporary accessibility of particular cognitions. Indeed, priming effects among individuals with less cognitively integrated mental frameworks may depend primarily upon contextual cues such as media coverage. As a result, what issues news media emphasize, and how those issues are covered, may substantially influence which thoughts or ideas come to mind for voters as they evaluate even seemingly unrelated elements of a political environment. Perhaps even more importantly, in priming effects based on construct accessibility the activation (or not) by media content of particular values does not appear to generate a patterned or predictable string of linked evaluations. That is, there is no correspondence between an individual's. interpretation of the media content and the priming effects, which appear to occur simply due to which ideas happen to be at the top of one's mental bin. These findings, then, suggest that priming effects that appear similar or even identical at the aggregate level may actually be a result of different mental processes.
In considering more broadly the linkages between news media and public opinion, these findings may offer some additional insights. In particular, this research questions the long-standing dichotomy in popular understanding of the roles of political issues and candidate images in voting behavior. Findings presented in this article indicate that research should consider how issues, particularly the manner in which they are framed by both politicians and news media, influence voters' perceptions of candidate image characteristics. Indeed, it is quite likely that image-based voting represents the outcome, in part, of perceptions encouraged by individuals' issue preferences and interpretations. Further, media coverage of issues seems likely to influence how other issues are interpreted in cross-issue priming effects. Both of these conclusions have implications for models of candidate choice and more broadly for democratic theory. Issues may not only influence voting behavior directly, due to voters' acceptance or rejection of candidates based on issue stands, but also indirectly, due to priming effects that begin with voters' thoughts about issues and then `carryover' to shape evaluations of candidate images and other issues in the electoral environment. What is clearly needed in political communication research, then, is a more nuanced understanding of the role of issues, and media framing, in individuals' political judgments.
While this study offers insights into priming effects, we recognize that it has some clear weaknesses. In particular, we theorize that how the process of priming occurs depends on the degree of inter-connectedness of an individual's mental system; with this perspective, we used two populations expected to differ in cognitive integration, but we are unable to offer concrete evidence of differences in mental structures, partly due to the difficulty in getting inside the cognitive `black box'. Future research should examine differences in priming effects by using measurement strategies that explore in greater depth (1) the role of values for individuals, particularly the linkages between values and political attitudes, and (2) the potential explanatory power of variables such as degree of political sophistication, need for cognition, amount of political knowledge, level of integrative complexity, and extent of public affairs media use. While this research has focused on the centrality and integration of values, the observed relationships might also be explained by personality, cognitive or behavioral variables. Future research that examines these factors--with, for example, other politically relevant sub-populations such as environmentalist groups or women's rights activists--would be an important step toward understanding the contributions to priming of both the accessibility of cognitions in short-term memory and the pathways among information in long-term memory.
This research was supported by a Kriss Fund grant from the School of journalism and Mass Communication at the University of Minnesota. The authors thank Sooyeon Yoo for research assistance, and William J. Gonzenbach and three anonymous reviewers for thoughtful comments and suggestions on an earlier version of this article.
Correspondence should be addressed to David Domke, University of Minnesota, School of journalism and Mass Communication, 111 Murphy Hall, 206 Church Street S.E., Minneapolis, MN 55455, USA. Fax: (612) 620-8251.
(1) Some scholars have similarly grouped together concerns about religious morals, personal principles, and basic rights into a broader conception of symbolic or ethical concerns as contrasted against concerns grounded in economics or individual self-interest. While drawing upon this previous work, our distinction between ethical and material frames (and individual issue interpretations) substantially differs from these earlier conceptualizations. Specifically, Sears and Funk (1991) distinguish between 'self-interest' (material- or economic-oriented) and `symbolic' (value-oriented) interpretations of issues, but they offer an affective approach to this distinction and generally do not examine socio-tropic concerns. Abramson and Inglehart (1995) distinguish between 'materialist' (economic- or fiscal-oriented) and `postmaterialist' (freedom-, quality of life-, and self-expression-oriented) values, but these are conceived as traits tied to individuals or nations rather than contextualized ways in which issues might be discussed or conceived.
(2) Pastors or ministers self-identified each church as `evangelical--that is, with a Biblical foundation in faith and practice. While it is obvious that these two sub-populations are not mutually exclusive, analysis of demographic and orientational variables--such as age, income, religiosity, and political party affiliationindicate that they differ substantially. For example, among evangelical Christians ages ranged relatively evenly between 20 and 76, while 83 percent of students were between the ages of 18 and 25. Throughout this article, we use the terms `evangelical Christians' and `undergraduate students' to distinguish the two sub-populations.
(3) The candidates' positions were composites or variations of actual political stances, taken from several campaigns. The fictitious candidates were competing for the Democratic nomination in a Congressional District primary in a nearby state, a political sphere likely to be unknown among the subjects. All articles were given female bylines.
(4) Factor analysis has demonstrated that voters focus on two correlated but empirically distinct dimensions of candidate character--integrity and competence (see Iyengar and Kinder 1987). This research, due to its focus on the priming effect of media coverage emphasizing ethical and moral values, examined attributions about candidate integrity. Specifically, subjects were coded as making integrity attributions if they discussed the following traits of at least one of the candidates: trustworthiness, honesty, morality/ethnicality, compassion or decency, courage/convictions, degree of power-hungriness.
(5) For Figures 1 and 3, which present the priming effect of attributions about candidate integrity, the political environments were split so that individuals evaluating the social-moral environments (abortion or euthanasia) were separated from individuals evaluating the health care environments (regardless of framing). For Figures 2 and 4, which present the priming effect of ethical interpretations of `material' issues, the political environments were split so that individuals evaluating the health care ethical framing environment were separated from individuals evaluating the other three environments. The political environments were coded in this fashion to isolate the experimental conditions in which each priming effect was maximized. In addition, subjects in the two political environments containing a social-moral issue were collapsed together because the results of hypotheses 1 and 2 revealed nearly parallel priming effects in these two political environments, in both research populations.
Abramson, P. and Inglehart, R. (1995): Value Change in Global Perspective, Ann Arbor, University of Michigan Press.
Anderson, J. R. (1983): The Architecture of Cognition, Cambridge, MA, Harvard University Press.
Anderson, J. R. (1985): Cognitive Psychology and its Implications, New York, Freeman.
Anderson, J. R. and Bower, G. H. (1973): Human Associative Memory, Washington, DC, Winston.
Ball-Rokeach, S. J. and Loges, W. E. (1997): `Making choices: Media roles in the construction of value-choices'. In Seligman, C., Olson J., and Zanna, M. (eds.), The Psychology of Values: The Ontario Symposium, 8, Mahwah, NJ, Erlbaum, 277-98.
Ball-Rokeach, S. J., Power, G. J., Guthrie, K. K. and Waring, H. R. (1990): `Value-framing abortion in the United States: An application of media system dependency theory', International Journal of Public Opinion Research, 2, 249-73.
Berkowitz, L. and Rogers, K. H. (1986): `A priming effect analysis of media influences'. In Bryant,. J. and Zillman, D. (eds.), Perspectives on Media Effects, Hillsdale, NJ, Erlbaum, 57-81.
Boninger, D. S., Krosnick, J. A. and Berent, M. K. (1995): `Origins of attitude importance: Self-interest, social identification, and value relevance', Journal of Personality and Social Psychology, 68, 61-80.
Brown, R. D. and Carmines, E. G. (1995): Materialists, postmaterialists, and the criteria for political choice in U.S. presidential elections', Journal Of Politics, 57, 483-94.
Carmines, E. G. and Stimson, J. A. (1980): `The two faces of issue voting', American Political Science Review, 74, 78-91.
Collins, A. M. and Loftus, E. F. (1975): `A spreading activation theory of semantic processing', Psychological Review, 82, 407-28.
Cook, E. A., Jelen, T. G. and Wilcox, C. (1994): `Issue voting in gubernatorial elections: Abortion and post-Webster politics', Journal of Politics, 56, 187-99.
Domke, D. and Shah, D. V. (1995): `Interpretation of issues and voter decision-making strategies: A new perspective on "issue-oriented" election coverage', journalism and Mass Communication Quarterly, 72, 45-71.
Fazio, R. H. (1986): `How do attitudes guide behavior?' In Sorrentino, R. M. and Higgins, E. T. (eds.), Handbook of Motivation and Cognition: Foundations of Social Behavior, New York, Guilford, 204-43.
Fazio, R. H. (1989): `On the power and functionality of attitudes: The role of attitude accessibility'. In Pratkanis, A. R., Breckler, S. J. and Greenwald, A. G. (eds.), Attitude Structure and Function, Hillsdale, NJ, Erlbaum, 153-79.
Fiske, S. and Taylor, S. (1991): Social Cognition, New York, McGraw-Hill.
Gamson, W. A. (1992): Talking Politics, Cambridge, Cambridge University Press.
Gitlin, T. (1980): The Whole World is Watching: Mass Media in the Making and Unmaking of the New Left, Berkeley, University of California Press.
Graber, D. A. (1988): Processing the News: How People Tame the Information Tide, New York, Longman, 2nd edn.
Graber, D. A. (1993): Mass Media and American Politics, Washington, DC, Congressional Quarterly Press, 4th edn.
Higgins, E. T. and Bargh, J. A. (1987): `Social cognition and social perception', Annual Review Of Psychology, 38, 369-425.
Higgins, E. T. and King, G. (1981): `Accessibility of social constructs: Information-processing consequences of individual and contextual variability'. In Cantor, N. and Kihlstrom, J. (eds.), Personality, Cognition, and Social Interaction, Hillsdale, NJ, Erlbaum, 69-121.
Iyengar, S. and Kinder, D. (1987): News that Matters: Television and American Opinion, Chicago, University of Chicago Press.
Iyengar, S. and Simon, A. (1994): `News coverage of the Gulf crisis and public opinion: A study of agenda-setting, priming, and framing'. In Bennett, W. L. and Paletz, D. (eds.), Taken by Storm, Chicago, University of Chicago Press, 167-85.
Iyengar, S., Peters, M. D. and Kinder, D. R. (1982): `Experimental demonstrations of the not-so-minimal political consequences of mass media', American Political Science Review, 76, 848-858.
Johnson, B. T. and Eagly, A. H. (1[989): `Effects of involvement on persuasion: A meta-analysis', Psychological Bulletin, 106, 290-314.
Johnston, R., Blais, A., Brady, H. E. and Crete, J. (1992): Letting the People Decide: Dynamics of a Canadian Election, Stanford, Stanford University Press.
Judd, C. M. and Krosnick, J. A. (1989): `The structural bases of consistency among political attitudes: Effects of political expertise and attitude importance'. In Pratkanis, A., Beckler, S., and Greenwald, A. (eds.), Attitude Structure and Function, Hillsdale, NJ, Erlbaum.
Kinder, D. (1985): `Presidential character revisited'. In Lau, R. and, Sears, D. (eds.), Cognition and Political Behavior, Hillsdale, NJ, Erlbaum.
Krosnick, J. A. and Brannon, L. (1993): `The media and the foundations of presidential support: George Bush and the Persian Gulf conflict', Journal of Social Issues, 49, 167-82.
Krosnick, J. A. and Kinder, D. R. (1990): `Altering the foundations of support for the president through priming', American Political Science Review, 84, 497-512.
Lau, R. (1989): `Construct accessibility and electoral choice', Political Behavior, 11, 5-32.
Lavine, H., Sullivan, J. L., Borgida, E. and Thomsen, C. J. (1996): `The relationship of national and personal issue salience to attitude accessibility on foreign and domestic policy issues', Political Psychology, 17, 293-316.
Lodge, M. and Stroh, P. (1993): `Inside the mental voting booth: An impression-driven process model of candidate evaluation'. In Iyengar, S. and McGuire, W. (eds.), Explorations in Political Psychology, London: Duke University Press, 225-63.
Lorentzen, L. J. (1980): `Evangelical life style concerns expressed in political action', Sociological Analysis, 41, 144-54.
Luker, K. (1984): Abortion and the Politics of Motherhood, Berkeley: University of California Press.
Mendelsohn, M. (1996). `The media and interpersonal communications: The priming of issues, leaders, and party identification', Journal Of Politics, 58, 112-25.
Moen, M. (1984): `School prayer and the politics of life style concern', Social Science Quarterly, 65, 1065-71.
Neuman, W. R., just, M. R. and Crigler, A. N. (1992): Common Knowledge: News and the Construction of Political Meaning, Chicago, University of Chicago Press.
Page, A. L. and Clelland, D. A. (1978): `The Kanawha county textbook controversy: A study of the politics of life style concern', Social Forces, 57, 265-81.
Rokeach, M. (1973): The Nature of Human Values, New York, The Free Press.
Schleuder, J., McCombs, M. and Wanta, W. (1991): `Inside the agenda-setting process: How political advertising and TV news prime viewers to think about issues and candidates'. In Biocca, F. (ed.), Television and Political Advertising, Vol. 1, Hillsdale, NJ, Erlbaum, 265-309.
Scott, W. A. (1955): `Reliability of content analysis: The case of nominal scale coding', Public Opinion Quarterly, 19, 321-5.
Sears, D. O. and Funk, C. L. (1991): `The role of self-interest in social and political attitudes', Advances in Experimental Social Psychology, 24, 1-91.
Shah, D. V., Domke, D. and Wackman, D. (1996): "`To thine own self be true": Values, framing, and voter decision-making strategies', Communication Research, 23, 509-60.
Shah, D. V., Domke, D. and Wackman, D. (1997): `Values and the vote: Linking issue interpretations to the process of candidate choice', journalism and Mass Communication Quarterly, 74, 357-387.
Srull, T. and Wyer, R. (1989): `Person memory and judgment', Psychological Review, 96, 58-83.
Tetlock, P. E. (1989): `Structure and function in political belief systems'. In Pratkanis, A. R., Breckler, S. J., and Greenwald, A. G. (eds.), Attitude Structure and Function, Hillsdale, NJ, Erlbaum, 129-151.
Tourangeau, R. and Rasinski, K. (1988): `Cognitive processes underlying context effects in attitude measurement', Psychological Bulletin, 1103, 299-314.
Wuthnow, R. (1989): The Struggle for America I Soul: Evangelicals, Liberals, and Secularism, New York, W. B. Eerdmans.
Wyer, R. and Srull, T. (1989): Memory and Cognition in its Social Context, Hillsdale: Erlbaum.
Zaller, J. (1992): The Nature and Origins of Mass Opinion, Cambridge, Cambridge University Press.
Zaller, J. and Feldman, S. (1992): `A simple theory of the survey response: Answering questions versus revealing preferences', American Journal of Political Science, 36(3), 579-616.
David Domke is a visiting faculty member in the School of Journalism and Mass Communication at the University of Minnesota. His research focuses on the relationships among individual values and attitudes, media framing and priming, and social change.
Dhavan V. Shah is a doctoral candidate in the School of Journalism and Mass Communication and a graduate student affiliate of the Center for the Study of Political Psychology, University of Minnesota. His research focuses on cognitive and motivational approaches to the study of political communication, with particular emphasis on media framing and priming.
Daniel B. Wickman is a professor in the School of Journalism and Mass Communication and a faculty affiliate of the Center for the Study of Political Psychology, University of Minnesota. His research focuses on people's information processing, with particular emphasis on their use of information in making decisions.
|Printer friendly Cite/link Email Feedback|
|Author:||Domke, David; Shah, Dhavan V.; Wackman, Daniel B.|
|Publication:||International Journal of Public Opinion Research|
|Date:||Mar 22, 1998|
|Previous Article:||Medium or message? Predicting dimensions of political sophistication.|
|Next Article:||Ethical considerations when establishing survey standards.|