Abstract
A meta-analysis assessed whether exposure to information is guided by defense or accuracy motives. The studies examined information preferences in relation to attitudes, beliefs, and behaviors in situations that provided choices between congenial information, which supported participants' pre-existing attitudes, beliefs, or behaviors, and uncongenial information, which challenged these tendencies. Analyses indicated a moderate preference for congenial over uncongenial information (d. = 0.36). As predicted, this congeniality bias was moderated by variables that affect the strength of participants' defense motivation and accuracy motivation. In support of the importance of defense motivation, the congeniality bias was weaker when participants' attitudes, beliefs, or behaviors were supported prior to information selection, when participants' attitudes, beliefs, or behaviors were not relevant to their values or not held with conviction, when the available information was low in quality, when participants' closed-mindedness was low, and when their confidence in the attitude, belief, or behavior was high. In support of the importance of accuracy motivation, an uncongeniality bias emerged when uncongenial information was relevant to accomplishing a current goal.
The availability of diverse information in an environment does not guarantee that a person's views will be equally diverse. Former United States vice-president Dick Cheney, for example, reportedly requires the television set be tuned into a conservative news channel before he enters a hotel room (The Smoking Gun, 2006). Individuals strongly committed to certain religions often avoid contact with information or people that can tempt them away from their doctrine. For example, science teachers at a public school in Arkansas were prevented from discussing evolution after complaints from religious parents, teachers, and faculty (Wiles, 2006). But what is the extent of people's inclination to receive congenial information? Is there a predominance of exposure to information that confirms pre-existing views? And, if there is such a bias, is it mitigated by factors that highlight the benefits of reaching accurate conclusions? Research on information exposure, which is synthesized in this paper, can answer these questions.
Although recent research has carefully analyzed the role of motivated reasoning in creating positive illusions (e.g., Kunda, 1990; Molden & Higgins, 2005), processes that allow access to the truth are just as important. Receiving information that supports one's position on an issue allows people to conclude that their views are correct but may often obscure reality. In contrast, receiving information that contradicts one's view on an issue can make people feel misled or ignorant but may allow access to a valid representation of reality. Therefore, understanding how people strive to feel validated versus to be correct is critical to explicate how they select information about an issue when several alternatives are present. A meta-analysis of field and laboratory studies on information exposure was conducted to shed light on these issues.
The classic assumption in selective exposure research is that people are motivated to defend their attitudes, beliefs, and behaviors from challenges (e.g., Festinger, 1957; Olson & Stone, 2005). In attitude theory (e.g., Albarracín, Johnson, & Zanna, 2005; Eagly & Chaiken, 1993; Zanna & Rempel, 1988), attitude is defined as the individual's evaluation of an entity (an issue, person, event, object, or behavior; e.g., President Obama); belief as an association between an entity and an attribute or outcome (e.g., President Obama is honest); and behavior as an overt action performed in relation to an entity (e.g., voting for President Obama). Selective exposure enables people to defend their attitudes, beliefs, and behaviors by avoiding information likely to challenge them and seeking information likely to support them. Selectivity of this type has often been called a congeniality bias (e.g., Eagly & Chaiken, 1993, 1998, 2005), but has also been called a confirmation bias (e.g., Jonas, Schulz-Hardt, Frey, & Thelen, 2001). In this paper, we will use the term congeniality bias.
Although the idea that selective exposure typically takes the form of a congeniality bias has a history extending back to William James (1890) and even to Francis Bacon (1620/1960), the topic first attained prominence among social psychologists in the context of Festinger's (1957, 1964) theory of cognitive dissonance. According to dissonance theory, after people commit to an attitude, belief, or decision, they gather supportive information and neglect unsupportive information to avoid or eliminate the unpleasant state of post decisional conflict known as cognitive dissonance. Typically researchers have tested this congeniality principle in a laboratory paradigm in which participants select information from alternatives. Prior to this selection, participants make a decision (e.g., about the guilt of a defendant in a mock trial), form an attitude (e.g., toward a work of art), report an existing attitude (e.g., on abortion), or report a prior behavior (e.g., whether they have smoked). Then participants are given an opportunity to receive information about the same issue (e.g., abortion, smoking) from a list of options usually presented as titles or abstracts of available articles. Typically half of these options support the participant's attitude, belief, or behavior, and the other half contradict it. The researcher records the numbers of chosen articles that agree or disagree with each participant's attitude, belief, or behavior. Selection of more articles that agree and fewer that disagree indicates a congeniality bias. Selection of more articles that disagree and fewer that agree indicates an uncongeniality bias.
In one of the initial studies testing selective exposure (Adams, 1961), mothers reported their belief that child development was predominantly influenced by genetic or environmental factors and then could choose to hear a speech that advocated either position. Consistent with the congeniality principle, mothers overwhelmingly chose the speech that favored their view on the issue. More recent investigations have used more complex designs to identify the moderators of the congeniality principle. For example, in a study showing that people select more uncongenial information when it is viewed as easy to refute, participants were offered congenial and uncongenial information attributed to either expert or novice sources (Lowin, 1969). Moreover, many studies have included manipulations to study the effects of perceiving that a previously reported decision could be altered (Frey & Rosch, 1984; Lowe & Steiner, 1968) and of challenging initially-reported attitudes (Brodbeck, 1956; Frey, 1981b).
As the intensive study of moderators might suggest, Festinger's (1957) assumptions about selective exposure did not receive universal support. In fact, Freedman and Sears' (1965) narrative review revealed that selective exposure appears to be strong when people are exposed to information in natural settings because congenial information predominates in their environment (de facto selective exposure). In contrast, this review indicated that laboratory experiments in which people were free to choose the information were as likely to disconfirm as confirm the congeniality principle. However, in the mid-1980s, reviewers who took a fresh look at the available research concluded that considerable evidence supported Festinger's theory (Cotton, 1985; Frey, 1986). Specifically, these reviewers argued that selectivity in favor of attitudes, beliefs, and behaviors occurs more under some conditions than others, such as when people possess high (vs. low) commitment to their attitudes. Like Festinger (1957, 1964), they also maintained that a congeniality bias is not the only psychological principle regulating information selection. These additional principles, which need to be controlled in testing selective exposure, include preferences for information that is unfamiliar (e.g., Sears, 1965) and information that is useful for making decisions or performing upcoming tasks (e.g., Lowe & Steiner, 1968; for a discussion of these principles, see Wicklund & Brehm, 1976).
To date, only qualitative reviews have examined selective exposure research. Importantly, however, a meta-analysis is the best way to examine whether a congeniality bias exists, as well as its precise size and variability. Our meta-analysis corrects this omission and provides the most inclusive literature coverage to date. In the first available review, Freedman and Sears (1965) analyzed 14 research reports and found little support for the congeniality principle. In subsequent reviews, Cotton (1985) and Frey (1986) examined 29 and 34 research reports, respectively, and concluded that congeniality exists under a variety of circumstances consistent with dissonance theory. Although these past reviews were comprehensive, our meta-analysis includes 21 new research reports that have emerged since 1986. Given the additional research on this topic, it is important to re-examine the issue of selective exposure in light of the most recent evidence. Moreover, re-examining past conclusions is critical because many of the recent studies have assessed selective exposure using novel methods (e.g., Jonas, Greenberg, & Frey, 2003; Lundgren & Prislin, 1998). In conducting this reanalysis, we were also able to examine new moderators and estimate the contribution of motivational factors not examined in earlier reviews.
Given the acknowledged complexities of the determinants of selective exposure, we present a general framework, displayed in Figure 1, of the motivational forces that shape exposure decisions. These motivational forces and their empirical instantiations organize our meta-analysis of the direction, size, and variability of exposure biases. In this framework, information choices are meant to fulfill goals to defend attitudes, beliefs, and behaviors and to accurately appraise and represent reality (Chaiken, Liberman, & Eagly, 1989). By extending our analysis beyond the defense motivation principle central to cognitive dissonance theory (Cotton, 1985; Frey, 1986), we present a framework for understanding selective exposure that is broad enough to encompass most empirical findings. In addition to investigating whether defense and accuracy motivations guide selective exposure, our review furthers understanding by examining the relative strength of these motivations.
Figure 1.
The opposing motivations and their concrete instantiations influence exposure to congenial over uncongenial information (congeniality bias).
Defense and accuracy motives have proven to be popular in analyses of how people process attitude-relevant information (Chaiken, Wood, & Eagly, 1996; Eagly, Chen, Chaiken, & Shaw-Barnes, 1999; Johnson & Eagly, 1989; Prislin & Wood, 2005; Wyer & Albarracín, 2005). In one of the most prominent discussions of motivated information processing, Chaiken et al. (1989) distinguished between defense and accuracy motivation. Defense motivation is the desire to defend one's existing attitudes, beliefs, and behaviors; accuracy motivation is the desire to form accurate appraisals of stimuli. Although these theorists also proposed a third motive, impression motivation, the desire to form and maintain positive interpersonal relations, the research on this aspect of selective exposure does not offer sufficient evidence for a meta-analysis. Even though past research has varied the anonymity of attitudes and selection decisions, such manipulations are uninformative because the effect of anonymity on selective exposure should depend on characteristics of the audience that one intends to impress (Schlenker, 1980; e.g., the congeniality of the audience). In the absence of appropriate manipulations, our meta-analysis focused only on defense and accuracy motivations.
Defense Motivation
In dissonance theory, selective exposure to congenial information is a strategy to relieve or avoid cognitive dissonance, which is the discomfort arising from the heightened presence of dissonant cognitions (Festinger, 1957). This discomfort can arise from the mere presence of cognitive conflict (Beauvois & Joule, 1996; Harmon-Jones, 2000; Harmon-Jones et al., 1996) or from a self-threat, such as the perception one is poorly informed (Aronson, 1968; Greenwald & Ronis, 1978; Schlenker, 1980, 2003; Steele, 1988). Presumably, experiencing or anticipating cognitive dissonance motivates people to defend themselves by seeking more congenial than uncongenial information. Hence, factors that enhance the experience or anticipation of cognitive dissonance should strengthen defense motivation and in turn accentuate the congeniality bias.
Defense motivation should be stronger when people who just reported an attitude or belief, or engaged in a behavior, receive challenging (vs. supporting) information prior to information selection (Frey, 1986). If people encounter a challenge to recently expressed attitudes, beliefs, or behaviors, their effort to reduce the cognitive conflict may enhance the congeniality bias (Beauvois & Joule, 1996; Festinger, 1964). In one study (Frey, 1981b), participants made a decision about whether to extend the contract of a store manager. Afterwards, participants were asked to read congenial information, uncongenial information, both congenial and uncongenial information, or no information prior to selecting additional reading material. Results revealed that participants manifested an enhanced congeniality bias when they were asked to read uncongenial rather than congenial information prior to this selection.
Another consideration pertains to the quality of the information available for selection. Whereas the selection of high-quality uncongenial information has the potential to threaten individuals, the selection of low-quality uncongenial information does not. Hence, to the degree that defense motivation guides exposure decisions, the presence of apparently high-quality uncongenial information for selection may enhance the congeniality bias (i.e., people will be more likely to avoid such information). Correspondingly, whereas high-quality congenial information can potentially bolster one's pre-existing position, low-quality congenial information may threaten one's position. Hence, expectations of high-quality congenial information for selection may enhance selection of congenial information as a way of defending a prior view (Festinger, 1964). As a result, regardless of whether information supports or refutes one's own position, expecting high-quality information should enhance the congeniality bias, and expecting low-quality information should lessen it (Frey, 1986; Lowin, 1969).
Defense motivation is presumably also strengthened by individuals' commitment to the pre-existing attitude, belief, or behavior and by high relevance of the issue to enduring values. Personal commitment to an attitude, belief, or behavior is presumed to increase defense motivation because of the greater discomfort produced by holding an incorrect view on an important issue (Brehm & Cohen, 1962; Kiesler, 1971). Personal commitment is often conceptualized as feeling highly attached to a view (Kiesler, 1971) or contributing to feeling ownership for a view (i.e., belief possession; see Abelson, 1988). Several factors have been identified that might lead to commitment, such as sacrificing for the view (e.g., dedicating much time or effort to make a decision), freely choosing the view (e.g., forming an attitude without coercion) and explaining the view publicly or privately (e.g., defending a belief in a written essay; for reviews, see Olson & Stone, 2005; Harmon-Jones & Harmon-Jones, 2008). Accordingly, commitment has sometimes been assessed directly by having participants self-report their attachment or loyalty to a view (e.g., Jonas & Frey, 2003a). Moreover, commitment has also been manipulated by leading participants to (a) engage in a behavior under high or low choice conditions (e.g., Frey & Wicklund, 1978), (b) dedicate more or less time or effort to attitude-relevant behavior (e.g., Betsch, Haberstroh, Glöckner, Haar, & Fiedler, 2001), or (c) justify (e.g., Schwarz, Frey, & Kumpf, 1980) or anticipate having to justify their opinions to an audience (e.g., Canon, 1964; Lowin, 1969; Sears & Freedman, 1965).
Another factor that may affect the strength of defense motivation is the ability to reverse a prior attitude, belief, or behavior (reversibility). On the one hand, reversibility may reduce defense motivation by, for example, reducing attachment to a prior view that is seen as tentative due to its reversibility (Abelson, 1988; Kiesler, 1971). On the other hand, reversibility may increase defense motivation by, for example, increasing thoughts about reasons to change the view and thus increasing the number of dissonant cognitions. As a result, reversibility may either attenuate or accentuate the congeniality bias.
Similarly, defense motivation should be strengthened when attitudes, beliefs, or behaviors are linked to individuals' enduring values (e.g., on the issues of euthanasia or abortion) and therefore promote value-relevant involvement with the issue (Johnson & Eagly, 1989). Value-relevant involvement with an issue often produces resistance to persuasion and, more generally, defensive processing of issue-relevant information (Chaiken et al., 1996). Hence, tendencies to prefer congenial over uncongenial information should be amplified when issues are high (vs. low) in value relevance (e.g., Festinger, 1964; Johnson & Eagly, 1989).
Finally, personality differences may affect the extent to which people are motivated to defend their views and behaviors. Closed-minded individuals may view challenging information as threatening, whereas open-minded people may view it as interesting (Adorno, Frenkel-Brunswick, Levinson, & Sanford, 1950; Altemeyer, 1981; 1998). Consequently, individuals with trait closed-mindedness (i.e., high scores on measures of dogmatism or authoritarianism, and high scores on the repression end of the repression-sensitization scale; Byrne, 1964) should manifest a stronger congeniality bias. Furthermore, people who view themselves as incapable of refuting challenging information may be more motivated to proactively guard against such threats (e.g., Albarracín & Mitchell, 2004). If so, the congeniality bias should be more pronounced for individuals with lesser confidence in their attitude, belief, or behavior. Researchers have operationalized confidence by providing bogus positive (vs. negative) feedback about participants' ability to form accurate attitudes, beliefs, or decisions (e.g., Micucci, 1972; Thayer, 1969) or by assessing participants' (a) confidence in their attitude, belief, or behavior (e.g., Berkowitz, 1965; Brechan, 2002; Brodbeck, 1956), (b) chronic anxiety (Frey, Stahlberg & Fries, 1986), or (c) consistency (vs. inconsistency) among behaviors and beliefs (Feather, 1962).1
Accuracy Motivation
Accuracy motivation should promote tendencies to process information in an objective, open-minded fashion that fosters uncovering the truth (Chaiken et al., 1989; Kunda, 1990). One motivational variable linked to accuracy motivation is outcome-relevant involvement (Johnson & Eagly, 1989), which refers to attitudes, beliefs, and decisions linked to an important outcome. For example, in one study (Jonas & Frey, 2003b), participants made a decision assuming that they would (high outcomes relevance) or would not (low outcome relevance) receive a prize for a correct choice. Unlike value-relevant involvement, which heightens defense motivation, outcome-relevant involvement has been shown to foster accuracy concerns and objective processing of available evidence (Albarracín, 2002; Chaiken et al., 1996; Johnson, 1994; Johnson & Eagly, 1989; Petty & Wegener, 1998). Therefore, the congeniality bias may be weaker for information about issues with important personal outcomes (high outcome relevance) than issues without such outcomes (low outcome relevance).
Another factor linked to accuracy motivation is information utility, defined as the extent to which information can be used to facilitate good decisions. Accuracy motivation should direct individuals to information of the highest utility regardless of its congeniality and may therefore weaken the congeniality bias. Researchers have manipulated information utility by assigning participants either to debate an issue or to write an essay in support of their attitudes, beliefs, or behaviors (e.g., Canon, 1964; Freedman, 1965b). The expectation of participating in a debate enhances the selection of uncongenial information because accurate knowledge of the opposition's arguments is useful for planning a rebuttal (i.e., uncongenial information is higher in utility than congenial information; Canon, 1964). In contrast, the expectation of writing a supporting essay enhances the selection of congenial information because this information is useful for preparing an intelligent defense of a current view (i.e., congenial information is higher in utility than uncongenial information; Canon, 1964). Also, accuracy motivation, unlike defense motivation, should direct individuals to information that is of high quality regardless of its congeniality. Therefore, unlike defense motivation, accuracy motivation should reduce the congeniality bias when the uncongenial information is high (vs. low) in quality. But, similar to defense motivation, accuracy motivation should accentuate the congeniality bias when the congenial information is high (vs. low) in quality.
The Present Meta-analysis
Our focus is on the analysis of whether people prefer information that supports pre-existing attitudes, beliefs, and behaviors more than information that challenges pre-existing attitudes, beliefs, and behaviors. Hence, we included studies that measured information selection on the basis of a pre-existing attitude, belief, or behavior. Our search produced 67 eligible reports of selective exposure, which contained 91 studies incorporating 300 statistically-independent groups with a total of just under 8,000 participants. Our synthesis of the selective exposure research has two primary objectives. The first objective is to assess the average magnitude, direction, and variability of selection biases. The second objective is to examine whether moderators related to defense and accuracy motivation (see Figure 1) account for variability in information selection. In general, attempts to defend attitudes, beliefs, or behaviors from attack should accentuate the congeniality bias, whereas attempts to reach accurate conclusions might often attenuate this bias. Other variables were analyzed in an exploratory fashion, including year of publication, source of report, study country, and amount of congenial and uncongenial information available for selection.
Method
Sample of Studies
To locate studies, we first conducted a computerized search of PsycINFO, Medline, Educational Resources Information Center, Dissertation Abstracts International, Social Science Citation Index, the conference proceedings of the Association for Consumer Research, ComAbstracts (http://www.cios.org), the Foreign Doctoral Dissertations Database of the Center for Research Libraries (http://www.crl.edu), and the databases of the Institute of Psychology Information for the German-Speaking Countries (http://www.zpid.de). The keywords were selective exposure, confirmation bias, congeniality bias, information seeking, information avoidance, information preference, attitude selectivity, selective processing, post decision changes, exposure to information, post decision exposure, selectivity, and information seeking. Additional keywords were cognitive dissonance, cognitive consistency, consonant information, dissonant information, supportive information, nonsupportive information, supporting information, consistent information, inconsistent information, decision reversibility, and decision irreversibility.
To supplement these database searches, we examined the reference lists of numerous review articles, chapters, and books discussing selective exposure. Also, we examined the abstracts of all of the publications by authors of multiple articles on selective exposure. Finally, we contacted researchers to request unpublished data and sent requests to the email lists of the Society for Personality and Social Psychology and the Association for Consumer Research. Our search extended through February 2008.
Selection Criteria
Five criteria determined the selection of studies. These criteria yielded a relatively large set of studies that used a similar methodology.
Studies were included if they assessed selective exposure on the basis of prior attitudes, beliefs, and behaviors (including decisions). Studies assessed attitudes and beliefs using self-report rating scales (e.g., agree vs. disagree). Behavior was usually operationalized by (a) a choice made in the session (e.g., choosing to extend a manager's contract; e.g., Frey, 1981b), (b) a self-report of past behavior (e.g., smoking; e.g., Feather, 1962), or (c) a behavior carried out in the experimental session (e.g., playing a computer game; e.g., Betsch et al., 2001). We excluded studies of exposure as a function of mood (e.g., studies of whether people who chronically suffer from a negative mood watch televised-news programs less than those who do not suffer from a negative mood; e.g., Anderson, Collins, Schmitt, & Jacobvitz, 1996), psychological disorders (e.g., studies of whether depressed vs. non-depressed people vary in exposure to comedy programs; e.g., Hammen, 1977; Potts & Sanchez, 1994; Raghunathan & Pham, 1999), biological factors (e.g., preferences for different television programs as a function of time of the menstrual cycle; e.g., Meadowcroft & Zillman, 1987; Potts, Dedman & Halford, 1996), demographic variables (e.g., gender differences in reading about achievement related topics; e.g., Dillman, Knobloch, & Zillman, 2003; Knobloch-Westerwick & Hastall, 2006) or personality (e.g., preferences for different types of music as a function of rebelliousness; e.g., Carpentier, Knobloch & Zillman, 2003).
Studies were included if they assessed information selection or preference and excluded if they pertained to selective interpretation (e.g., Robinson, Keltner, Ward, & Ross, 1995), memory (e.g., Levine & Murphy, 1943), or liking of already viewed material (e.g., Boden & Baumeister, 1997). Typical assessments of selective exposure compared counts of participants' choices from a list of congenial and uncongenial alternatives (e.g., Fischer, Jonas, Frey, & Schulz-Hardt, 2005; Jonas, Graupmann, & Frey, 2006). In some studies, information selection was assessed by participants' ratings or rankings of their preferences for congenial and uncongenial information (e.g., Brannon, Tagler, & Eagly, 2007; Feather, 1963). Finally, selective exposure was sometimes assessed by the amount of time participants devoted to viewing congenial versus uncongenial information (e.g., Brock & Balloun, 1967; Olson & Zanna, 1979).
Studies were included if they arranged choices between congenial and uncongenial information and excluded if they presented only one-sided information or only neutral information (fifteen articles; e.g., Behling, 1971; Edeani, 1979; Frey, 1981c; Otis, 1979; Sweeney & Gruber, 1984; Wellins & McGinnies, 1977). Note that a bias in information selection can only be diagnosed when choices are provided between consonant and dissonant information (Freedman & Sears, 1965). For example, finding that voters who supported Nixon (vs. did not) paid less attention to anti-Nixon information does not necessarily imply a congeniality bias if these same voters also pay less attention to the news in general (Sweeney & Gruber, 1984). Based on this criterion, we also excluded studies on positive hypothesis testing, which examine whether individuals tend to select more questions that are consistent than inconsistent with a prior belief (e.g., Johnston, 1996). For example, research in this tradition might ask participants to test whether someone is an extravert by selecting questions to ask to this person. Some of these questions might confirm the hypothesis (Do you enjoy parties?), whereas others might disconfirm it (Do you enjoy spending time alone?). Selecting more confirming than disconfirming questions has been termed positive hypothesis testing and is distinguished from the congeniality bias examined in research on selective exposure. Specifically, questions testing a hypothesis can sometimes provide disconfirming answers, thus departing from a direct choice of congenial or uncongenial information (Klayman & Ha, 1987).
Studies were included if they focused on an individual's information seeking and excluded if they focused on a group's information seeking (Shulz-Hardt, Frey, Luthgens, & Moscovici, 2000).
Finally, studies were excluded if they lacked adequate statistics (e.g., F-ratios, frequencies, and p-values) for calculating an effect size representing the difference in exposure to congenial and uncongenial information (seven articles; e.g., Donohew, Parker, & McDermott, 1972).
Partitioning of Studies, Calculation of Effect Sizes, and Analytical Considerations
Results were often partitioned into experimental conditions or samples of participants. Whenever possible, effect sizes were computed according to the conceptually-important moderators discussed by the researcher even when this partitioning did not reflect our hypothesized moderators (e.g., unlimited vs. limited choices of information to receive; Fischer et al., 2005). This procedure allowed us to analyze the overall sample of effect sizes without assuming equality in effect sizes across the levels of moderators that were of interest to the researcher (see Table 1).2
Table 1.
All Included Studies, Effect Sizes, and Moderator Values (Levels)
| Short reference for report and condition | d | Challenge or support |
Quality congenial |
Quality uncongenial |
Commitment | Reversibility | Value relevance |
Closed- mindedness |
Confidence | Outcome relevance |
Utility Congenial |
Utility uncongenial |
Relative utility |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Adams (1961) | |||||||||||||
| Heard congenial speech | 0.71 | Support | H | H | H | Reversible | H | M | H | H | No goal | No goal | Equal |
| Heard uncongenial speech | 0.55 | Challenge | H | H | H | Reversible | H | M | H | H | No goal | No goal | Equal |
| Berkowitz (1965) | |||||||||||||
| Support | −1.50 | Support | L | L | M | Irreversible | H | M | H | L | No goal | No goal | Equal |
| Moderate dissonance | −0.85 | Challenge | L | L | M | Irreversible | H | M | H | L | No goal | No goal | Equal |
| Strong dissonance | 1.04 | Challenge | L | L | M | Irreversible | H | M | L | L | No goal | No goal | Equal |
| Betsch et al. (2001) | |||||||||||||
| Strong routine, familiar task | 1.04 | No | M | M | H | Reversible | L | M | H | L | H | H | Equal |
| Weak routine, familiar task | 0.61 | No | M | M | M | Reversible | L | M | H | L | H | H | Equal |
| Strong routine, new task | −0.28 | No | M | M | H | Reversible | L | M | M | L | H | H | Equal |
| Weak routine, new task | 0.46 | No | M | M | M | Reversible | L | M | M | L | H | H | Equal |
| Bosotti (1984) | |||||||||||||
| Study 1 | |||||||||||||
| Low quality information | 0.40 | No | L | L | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| High quality information | 0.38 | No | H | H | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Study 2 | |||||||||||||
| Low quality information | −0.26 | No | L | L | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| High quality information | 0.70 | No | H | H | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Brannon et al. (2007) | |||||||||||||
| Study 1a | 0.72 | No | H | H | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Study 1b | 0.73 | No | H | H | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Study 2 | 0.49 | No | H | H | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Brechan (2002) | |||||||||||||
| Study 1 | |||||||||||||
| High confidence | 1.72 | No | M | M | M | Irreversible | H | M | H | L | No goal | No goal | Equal |
| Low confidence | 1.07 | No | M | M | M | Irreversible | H | M | L | L | No goal | No goal | Equal |
| Study 2 | |||||||||||||
| High confidence | 1.79 | No | M | M | M | Irreversible | H | M | H | L | No goal | No goal | Equal |
| Low confidence | 0.99 | No | M | M | M | Irreversible | H | M | L | L | No goal | No goal | Equal |
| Brock (1965) | |||||||||||||
| Low commitment, smoker | 0.53 | No | M | M | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Low commitment, non-smoker | 0.01 | No | M | M | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| High commitment, smoker | 3.32 | No | M | M | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| High commitment, non-smoker | 2.24 | No | M | M | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Brock et al. (1970) | |||||||||||||
| New information, high commitment | 0.81 | No | H | H | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| New information, less commitment | −0.09 | No | H | H | M | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Old information, high commitment | −0.17 | No | H | H | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Old information, less commitment | 0.30 | No | H | H | M | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Brock & Balloun (1967) | |||||||||||||
| Study 1 | |||||||||||||
| Smoker | 0.74 | No | H | H | H | Irreversible | L | M | M | H | H | H | Equal |
| Non-smoker | 0.18 | No | H | H | M | Irreversible | L | M | M | L | H | H | Equal |
| Study 2 | |||||||||||||
| Smoker | 0.86 | No | H | H | H | Irreversible | L | M | M | H | H | H | Equal |
| Non-smoker | 0.11 | No | H | H | M | Irreversible | L | M | M | L | H | H | Equal |
| Study 3 | |||||||||||||
| Smoker | 0.99 | No | H | H | H | Irreversible | L | M | M | H | H | H | Equal |
| Non-smoker | −0.06 | No | H | H | M | Irreversible | L | M | M | L | H | H | Equal |
| Study 4 | |||||||||||||
| Smoker | 1.21 | No | H | H | H | Irreversible | L | M | M | H | H | H | Equal |
| Non-smoker | 0.42 | No | H | H | M | Irreversible | L | M | M | L | H | H | Equal |
| Brodbeck (1956) | |||||||||||||
| Challenge, decrease in confidence | −0.37 | Challenge | L | L | M | Irreversible | H | M | L | H | No goal | No goal | Equal |
| Challenge, no decrease in confidence | −0.70 | Challenge | L | L | M | Irreversible | H | M | H | H | No goal | No goal | Equal |
| Support | −0.97 | Support | L | L | M | Irreversible | H | M | M | H | No goal | No goal | Equal |
| Canon (1964) | |||||||||||||
| Debate goal, high confidence | −0.55 | No | M | M | H | Irreversible | L | M | H | L | H | H | Uncongenial |
| Debate goal, low confidence | 0.23 | No | M | M | H | Irreversible | L | M | L | L | H | H | Uncongenial |
| Expression goal, high confidence | 0.32 | No | M | M | H | Irreversible | L | M | H | L | H | H | Congenial. |
| Expression goal, low confidence | 1.14 | No | M | M | H | Irreversible | L | M | L | L | H | H | Congenial. |
| Canon & Matthews (1972) | |||||||||||||
| Non-smoker, low concern for health (CH) | 0.16 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Non-smoker, high CH | 0.52 | No | H | H | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Smoker, low CH | −0.06 | No | H | H | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Smoker, high CH | 0.89 | No | H | H | H | Irreversible | H | M | M | H | No goal | No goal | Equal |
| Clarke & James (1967) | |||||||||||||
| Expect debate | 0.47 | No | M | M | M | Irreversible | H | M | M | L | H | H | Uncongenial |
| Expect discussion | 0.31 | No | M | M | M | Irreversible | H | M | M | L | H | H | Equal |
| Cotton & Hieser (1980) | |||||||||||||
| Low choice | 1.04 | No | M | M | M | Reversible | H | M | H | L | No goal | No goal | Equal |
| High choice | 0.17 | No | M | M | L | Reversible | H | M | H | L | No goal | No goal | Equal |
| Ehrlich et al. (1957) | |||||||||||||
| New car owners | 0.95 | No | M | M | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Old car owners | 0.63 | No | M | M | H | Irreversible | L | M | H | H | No goal | No goal | Equal |
| Feather (1962) | |||||||||||||
| Smoker, congenial beliefs | −0.78 | No | M | M | H | Irreversible | L | M | H | H | No goal | No goal | Equal |
| Smoker, uncongenial beliefs | 0.39 | No | M | M | H | Irreversible | L | M | L | H | No goal | No goal | Equal |
| Non-smoker, uncongenial beliefs | −0.39 | No | M | M | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Non-smoker, congenial beliefs | 0.26 | No | M | M | M | Irreversible | L | M | H | L | No goal | No goal | Equal |
| Feather (1963) | |||||||||||||
| Smoker | 0.06 | No | M | M | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Non-smoker | −0.12 | No | M | M | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Feather (1969) | |||||||||||||
| High dogmatism, old information | 1.22 | No | H | H | M | Irreversible | H | H | M | L | No goal | No goal | Equal |
| High dogmatism, new information | 1.40 | No | H | H | M | Irreversible | H | H | M | L | No goal | No goal | Equal |
| Low dogmatism, old information | 0.41 | No | H | H | M | Irreversible | H | L | M | L | No goal | No goal | Equal |
| Low dogmatism, new information | 0.18 | No | H | H | M | Irreversible | H | L | M | L | No goal | No goal | Equal |
| Fischer et al. (2005) | |||||||||||||
| Study 1 | |||||||||||||
| No restrictions | −0.20 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Lower limit restrictions | −0.07 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Upper-limit restrictions | 1.00 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Specific restrictions | 1.65 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Study 2 | |||||||||||||
| Restricted, no scarcity cue | 2.43 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Restricted, scarcity cue | 1.61 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Unrestricted, no scarcity cue | 0.66 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Unrestricted, scarcity cue | 1.77 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Study 3 | |||||||||||||
| Restricted, load | −0.34 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Restricted, no load | 2.22 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Unrestricted, no load | −0.01 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Unrestricted, load | 0.08 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Study 4 | |||||||||||||
| Restricted, before | 0.55 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Restricted, after | 0.71 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Unrestricted, before | −0.07 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Unrestricted, after | 0.08 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Fischer et al. (2008) | |||||||||||||
| Study 1 | |||||||||||||
| 2 pieces | −0.39 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| 10 pieces | 0.46 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Study 2 | |||||||||||||
| 2 pieces | −0.29 | No | M | M | H | Reversible | H | M | M | H | H | H | Equal |
| 4 pieces | 0.65 | No | M | M | H | Reversible | H | M | M | H | H | H | Equal |
| 10 pieces | 0.85 | No | M | M | H | Reversible | H | M | M | H | H | H | Equal |
| Study 3 | |||||||||||||
| 2 pieces with content cues | −0.41 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| 10 pieces with content cues | 1.06 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| 2 pieces no content cues | −0.31 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| 10 pieces no content cues | −0.01 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Study 4 | |||||||||||||
| 2 no focus | −0.44 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| 10 no focus | 0.78 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| 2 quality focus | 0.48 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| 10 quality focus | 0.68 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| 2 direction focus | −1.29 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| 10 direction focus | −0.20 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Freedman (1965a) | |||||||||||||
| Positive interview | −1.49 | No | H | H | M | Irreversible | L | M | H | L | No goal | No goal | Equal |
| Negative interview | −1.44 | No | H | H | M | Irreversible | L | M | H | L | No goal | No goal | Equal |
| Freedman (1965b) | |||||||||||||
| Low confidence, expression goal | 0.35 | No | M | M | H | Irreversible | L | M | L | L | H | H | Congenial |
| Low confidence, debate goal | −0.29 | No | M | M | H | Irreversible | L | M | L | L | H | H | Uncongenial |
| Low confidence, no goal | 0.04 | No | M | M | M | Irreversible | L | M | L | L | No goal | No goal | Equal |
| High confidence, expression goal | 0.34 | No | M | M | H | Irreversible | L | M | H | L | H | H | Congenial |
| High confidence, debate goal | −0.27 | No | M | M | H | Irreversible | L | M | H | L | H | H | Uncongenial |
| High confidence, no goal | 0.04 | No | M | M | M | Irreversible | L | M | H | L | No goal | No goal | Equal |
| Frey (1981a) | |||||||||||||
| Information costs, 7 | 0.14 | No | M | M | M | Reversible | H | M | L | L | No goal | No goal | Equal |
| Information costs, 15 | 0.72 | No | M | M | M | Reversible | H | M | L | L | No goal | No goal | Equal |
| Information costs, 25 | 0.64 | No | M | M | H | Reversible | H | M | L | L | No goal | No goal | Equal |
| Information costs, 33 | 0.62 | No | M | M | H | Reversible | H | M | L | L | No goal | No goal | Equal |
| Information free, 7 | 0.22 | No | M | M | M | Reversible | H | M | L | L | No goal | No goal | Equal |
| Information free, 15 | 0.78 | No | M | M | M | Reversible | H | M | L | L | No goal | No goal | Equal |
| Information free, 25 | −0.25 | No | M | M | H | Reversible | H | M | L | L | No goal | No goal | Equal |
| Information free, 33 | −0.18 | No | M | M | H | Reversible | H | M | L | L | No goal | No goal | Equal |
| Frey (1981b) | |||||||||||||
| Study 1 | |||||||||||||
| High quality congenial, high quality Uncongenial | 0.50 | No | H | H | H | Irreversible | L | M | M | L | H | L | Congenial |
| Low quality congenial, high quality Uncongenial | −0.07 | No | L | H | H | Irreversible | L | M | M | L | L | L | Equal |
| High quality congenial, low quality Uncongenial | 0.23 | No | H | L | H | Irreversible | L | M | M | L | H | L | Congenial |
| Low quality congenial, low quality Uncongenial | 0.07 | No | L | L | H | Irreversible | L | M | M | L | L | L | Equal |
| Study 2 | |||||||||||||
| High quality congenial, high quality Uncongenial | 0.58 | No | H | H | M | Irreversible | H | M | M | H | H | H | Equal |
| Low quality congenial, high quality Uncongenial | −0.01 | No | L | H | M | Irreversible | H | M | M | H | H | H | Equal |
| High quality congenial, low quality Uncongenial | 0.69 | No | H | L | M | Irreversible | H | M | M | H | H | H | Equal |
| Low quality congenial, low quality Uncongenial | 0.24 | No | L | L | M | Irreversible | H | M | M | H | H | H | Equal |
| Study 3 | |||||||||||||
| Unlimited, no information | 0.40 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Unlimited, uncongenial information | 0.54 | Challenge | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Unlimited, congenial information | −0.23 | Support | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Unlimited, both | 0.21 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Limited, no information | 0.91 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Limited, uncongenial information | 0.44 | Challenge | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Limited, congenial information | −0.04 | Support | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Limited, both | 0.36 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Frey (1982) | |||||||||||||
| High gain | 0.74 | Support | M | M | H | Reversible | L | M | H | H | H | H | Equal |
| Moderate gain | 0.76 | Support | M | M | H | Reversible | L | M | H | H | H | H | Equal |
| Low gain | 1.26 | Support | M | M | H | Reversible | L | M | H | H | H | H | Equal |
| Low loss | 0.72 | Challenge | M | M | H | Reversible | L | M | L | H | H | H | Equal |
| Moderate loss | −0.18 | Challenge | M | M | H | Reversible | L | M | L | H | H | H | Equal |
| High loss | −0.72 | Challenge | M | M | H | Reversible | L | M | L | H | H | H | Equal |
| Frey & Rosch (1984) | |||||||||||||
| Reversible, old information | 0.72 | No | H | H | L | Reversible | L | M | M | L | L | L | Equal |
| Reversible, new information | 0.15 | No | H | H | L | Reversible | L | M | M | L | H | H | Equal |
| Irreversible, old information | 0.73 | No | H | H | H | Irreversible | L | M | M | L | L | L | Equal |
| Irreversible, new information | 1.19 | No | H | H | H | Irreversible | L | M | M | L | H | H | Equal |
| Frey & Stahlberg (1986; Study 1) | |||||||||||||
| Congenial information | 0.18 | Support | H | H | H | Irreversible | H | M | H | L | No goal | No goal | Equal |
| No information | 0.65 | No | H | H | H | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Frey et al. (1986) | |||||||||||||
| High anxiety, low score | 1.32 | No | H | H | H | Reversible | H | H | L | L | No goal | No goal | Equal |
| High anxiety, high score | −0.07 | No | H | H | H | Reversible | H | H | L | L | No goal | No goal | Equal |
| Low anxiety, low score | 0.50 | No | H | H | H | Reversible | H | H | H | L | No goal | No goal | Equal |
| Low anxiety, high score | 0.31 | No | H | H | H | Reversible | H | H | H | L | No goal | No goal | Equal |
| Frey & Wicklund (1978) | |||||||||||||
| No choice, restricted search | −0.07 | No | H | H | L | Irreversible | L | M | L | L | No goal | No goal | Equal |
| No choice, restricted search | 0.47 | No | H | H | L | Irreversible | L | M | L | L | No goal | No goal | Equal |
| No choice, unrestricted search | 0.14 | No | H | H | L | Irreversible | L | M | L | L | No goal | No goal | Equal |
| No choice, restricted search | −0.40 | No | H | H | L | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Choice, restricted search | 0.72 | No | H | H | H | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Choice, restricted search | 1.53 | No | H | H | H | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Choice, unrestricted search | 0.50 | No | H | H | H | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Choice, restricted search | 0.60 | No | H | H | H | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Hillis & Crano (1973) | |||||||||||||
| Strong pro-choice attitude, pro-choice talk | 1.12 | No | H | H | H | Irreversible | H | M | H | L | H | L | Congenial |
| Pro-choice attitude, pro-choice talk | 0.83 | No | H | H | M | Irreversible | H | M | M | L | H | L | Congenial |
| Strong pro-life attitude, pro-choice talk | −0.51 | No | H | H | H | Irreversible | H | M | H | L | L | H | Uncongenial |
| Pro-life attitude, pro-choice talk | 0.00 | No | H | H | M | Irreversible | H | M | M | L | L | H | Uncongenial |
| Strong pro-life attitude, pro-life talk | 0.73 | No | H | H | H | Irreversible | H | M | H | L | H | L | Congenial |
| Pro-life attitude, pro-life talk | 0.73 | No | H | H | M | Irreversible | H | M | M | L | H | L | Congenial |
| Strong pro-choice attitude, pro-life talk | −1.26 | No | H | H | H | Irreversible | H | M | H | L | L | H | Uncongenial |
| Pro-choice attitude, pro-life talk | −1.26 | No | H | H | M | Irreversible | H | M | M | L | L | H | Uncongenial |
| Holton & Pyszczynski (1989) | |||||||||||||
| Study 1 | 1.30 | No | H | H | H | Reversible | H | M | M | L | No goal | No goal | Equal |
| Janis & Rausch (1970) | |||||||||||||
| Refused to sign quickly | −0.70 | No | M | M | M | Irreversible | H | M | H | H | No goal | No goal | Equal |
| Refused to sign | −0.74 | No | M | M | M | Irreversible | H | M | M | H | No goal | No goal | Equal |
| Might sign | 0.30 | No | M | M | M | Irreversible | H | M | L | H | No goal | No goal | Equal |
| Have signed | −0.14 | No | M | M | H | Irreversible | H | M | H | H | No goal | No goal | Equal |
| Jecker (1964) | |||||||||||||
| Moderate commitment | 0.54 | No | H | H | M | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Low commitment | 0.10 | No | H | H | L | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Jonas & Frey (2003a) | |||||||||||||
| Dm | −0.25 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Euro | 0.35 | No | H | H | L | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Jonas & Frey (2003b) | |||||||||||||
| Study 1 | |||||||||||||
| Personal, friendly atmosphere | 0.24 | No | H | H | H | Reversible | L | M | M | L | H | H | Equal |
| Personal, business atmosphere | 0.63 | No | H | H | H | Reversible | L | M | M | L | H | H | Equal |
| Advisor, friendly atmosphere | −0.08 | No | H | H | H | Reversible | L | M | M | H | H | H | Equal |
| Advisor, business atmosphere | 0.10 | No | H | H | H | Reversible | L | M | M | H | H | H | Equal |
| Study 2 | |||||||||||||
| Personal | 0.36 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Advisor | −0.13 | No | H | H | M | Reversible | L | M | M | H | H | H | Equal |
| Jonas, Frey et al. (2001) | |||||||||||||
| Support, verbal justification | 0.35 | Support | H | H | H | Reversible | L | M | H | L | H | H | Equal |
| Challenge, verbal justification | 0.11 | Challenge | H | H | H | Reversible | L | M | L | L | H | H | Equal |
| Support and challenge, verbal justification | 0.40 | No | H | H | H | Reversible | L | M | M | L | H | H | Equal |
| Support, no verbal justification | −0.19 | Support | H | H | M | Reversible | L | M | H | L | H | H | Equal |
| Challenge, no verbal justification | 0.32 | Challenge | H | H | M | Reversible | L | M | L | L | H | H | Equal |
| Support and challenge, no verbal justification | −1.16 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Jonas, Graupmann et al. (2003) | |||||||||||||
| Low party awareness, low relevance | 0.57 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Low party awareness, high relevance | 2.18 | No | H | H | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| High party awareness, low relevance | 0.50 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| High party awareness, high relevance | 0.82 | No | H | H | H | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Jonas et al. (2006) | |||||||||||||
| Study 1 | |||||||||||||
| Positive mood | −0.01 | No | H | H | H | Reversible | L | M | H | L | H | H | Equal |
| Negative mood | 0.85 | No | H | H | H | Reversible | L | M | L | L | H | H | Equal |
| Study 3 | |||||||||||||
| Positive mood | 0.26 | No | H | H | H | Reversible | L | M | H | L | H | H | Equal |
| Neutral mood | 0.64 | No | H | H | H | Reversible | L | M | M | L | H | H | Equal |
| Negative mood | 0.89 | No | H | H | H | Reversible | L | M | L | L | H | H | Equal |
| Jonas, Greenberg et al. (2003) | |||||||||||||
| Mortality salience, worldview issue | 1.50 | No | H | H | H | Reversible | H | M | M | H | H | H | Equal |
| Control, worldview issue | 0.73 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Mortality salience, fictitious issue | −0.02 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Control, fictitious issue | 0.28 | No | H | H | M | Reversible | L | M | M | L | H | H | Equal |
| Jonas, Schulz-Hardt & Frey (2001) | |||||||||||||
| Sequential | 2.56 | No | H | H | H | Reversible | H | M | M | H | No goal | No goal | Equal |
| Simultaneous | 1.28 | No | H | H | M | Reversible | H | M | M | H | No goal | No goal | Equal |
| Jonas et al. (2005) | |||||||||||||
| Study 1 | |||||||||||||
| Decision-maker for self | 2.24 | No | M | M | M | Reversible | L | M | M | H | H | H | Equal |
| Advisor as recommender (non-binding) | 0.76 | No | M | M | M | Reversible | L | M | M | H | H | H | Equal |
| Advisor as decision-maker (binding) | 2.76 | No | M | M | H | Reversible | L | M | M | H | H | H | Equal |
| Study 2 | |||||||||||||
| No meeting, recommendation | 0.78 | No | M | M | M | Reversible | L | M | M | H | H | H | Equal |
| Meeting, recommendation | 0.84 | No | M | M | H | Reversible | L | M | M | H | H | H | Equal |
| No meeting, decision-maker | −0.81 | No | M | M | M | Reversible | L | M | M | H | H | H | Equal |
| Meeting, decision-maker | 1.14 | No | M | M | H | Reversible | L | M | M | H | H | H | Equal |
| Jonas, Schulz-Hardt, Frey, & Thelen (2001) | |||||||||||||
| Study 1 | |||||||||||||
| Simultaneous search | 0.40 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Sequential search | 1.25 | No | H | H | H | Reversible | H | M | M | H | H | H | Equal |
| Study 2 | |||||||||||||
| Simultaneous-simultaneous focus | 0.56 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Simultaneous-sequential focus | 0.40 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Sequential-simultaneous focus | 0.90 | No | H | H | H | Reversible | H | M | M | H | H | H | Equal |
| Sequential-sequential focus | 1.84 | No | H | H | H | Reversible | H | M | M | H | H | H | Equal |
| Study 3 | |||||||||||||
| Sequential search | 0.91 | No | H | H | H | Reversible | H | M | M | H | H | H | Equal |
| Study 4 | |||||||||||||
| Sequential-control focus | 1.40 | No | H | H | H | Reversible | H | M | M | H | H | H | Equal |
| Simultaneous-control focus | 0.43 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Sequential-information focus | 0.46 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Simultaneous-information focus | 0.50 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| Kleck & Wheaton (1967) | |||||||||||||
| Study 1 | 0.47 | No | M | M | M | Irreversible | H | M | M | H | No goal | No goal | Equal |
| Lavine et al. (2005) | |||||||||||||
| Control, high authoritarianism | 0.15 | No | M | M | M | Irreversible | H | H | M | L | No goal | No goal | Equal |
| Control, low authoritarianism | 0.06 | No | M | M | M | Irreversible | H | L | M | L | No goal | No goal | Equal |
| Mortality salience, high authoritarianism | 0.74 | No | M | M | H | Irreversible | H | H | M | L | No goal | No goal | Equal |
| Mortality salience, low authoritarianism | 0.00 | No | M | M | H | Irreversible | H | L | M | L | No goal | No goal | Equal |
| Lavoie & Thompson (1972) | |||||||||||||
| Study 1 | 0.20 | No | M | M | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Lowe & Steiner (1968) | |||||||||||||
| Reversible, consequences | −0.44 | No | M | M | L | Reversible | L | M | M | H | H | H | Equal |
| Reversible, no consequences | −0.25 | No | M | M | L | Reversible | L | M | M | L | H | H | Equal |
| Irreversible, consequences | −0.53 | No | M | M | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Irreversible, no consequences | 0.28 | No | M | M | H | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Lowin (1969) | |||||||||||||
| High quality information, high confidence | 0.52 | No | H | H | H | Irreversible | L | M | H | L | No goal | No goal | Equal |
| High quality information, low confidence | 0.64 | No | H | H | H | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Low quality information, high confidence | −0.50 | No | L | L | H | Irreversible | L | M | H | L | No goal | No goal | Equal |
| Low quality information, low confidence | −0.18 | No | L | L | H | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Lundgren & Prislin (1998) | |||||||||||||
| Study 1 | |||||||||||||
| Accuracy motive | −0.18 | No | L | L | M | Reversible | L | M | M | L | H | H | Equal |
| Impression motive | −0.51 | No | L | L | M | Reversible | L | M | M | L | L | H | Uncongenial |
| Defense motive | 0.42 | No | L | L | M | Reversible | L | M | M | H | H | L | Congenial |
| Control | −0.26 | No | L | L | M | Reversible | L | M | M | L | No goal | No goal | Equal |
| Study 2 | |||||||||||||
| Impression and accuracy motive | 0.16 | No | L | L | M | Reversible | L | M | M | H | H | H | Uncongenial |
| Defense and accuracy motive | 1.24 | No | L | L | M | Reversible | L | M | M | H | H | H | Congenial |
| Defense and impression motive | 0.26 | No | L | L | M | Reversible | L | M | M | H | H | H | Equal |
| Maccoby et al. (1961) | |||||||||||||
| Support | 0.83 | Support | L | L | M | Reversible | H | M | M | L | H | H | Equal |
| Challenge | 0.75 | Challenge | L | L | M | Reversible | H | M | M | L | H | H | Equal |
| McFarland & Warren (1992) | |||||||||||||
| Fundamentalist Christians | 1.54 | No | H | H | H | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Micucci (1972) | |||||||||||||
| Low self-esteem | 1.04 | No | M | M | M | Irreversible | H | M | L | L | No goal | No goal | Equal |
| Moderate self-esteem | −0.29 | No | M | M | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| High self-esteem | −0.30 | No | M | M | M | Irreversible | H | M | H | L | No goal | No goal | Equal |
| Miller (1977) | |||||||||||||
| Immediate | 0.17 | No | M | M | M | Irreversible | H | M | L | H | No goal | No goal | Equal |
| 4 minutes | −0.57 | No | M | M | M | Irreversible | H | M | L | H | No goal | No goal | Equal |
| 12 minutes | 0.82 | No | M | M | M | Irreversible | H | M | L | H | No goal | No goal | Equal |
| Nemeth & Rogers (1996) | |||||||||||||
| High relevance, majority dissent | 0.56 | No | H | H | M | Reversible | H | M | M | H | H | H | Equal |
| High relevance, minority dissent | 0.00 | Support | M | M | M | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Low relevance, majority dissent | 0.77 | Challenge | M | M | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Low relevance, minority dissent | 0.30 | Support | M | M | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Olson & Zanna (1979) | |||||||||||||
| Repressors, choice | 0.83 | No | N/A | N/A | M | Irreversible | L | H | M | L | No goal | No goal | Equal |
| Repressor, liking | 0.05 | No | N/A | N/A | M | Irreversible | L | H | M | L | No goal | No goal | Equal |
| Sensitizers, choice | −0.21 | No | N/A | N/A | M | Irreversible | L | L | M | L | No goal | No goal | Equal |
| Sensitizers, liking | 0.16 | No | N/A | N/A | M | Irreversible | L | L | M | L | No goal | No goal | Equal |
| Pyszczynski et al. (1985) | |||||||||||||
| Study 1 | 1.06 | No | H | H | H | Reversible | H | M | M | L | No goal | No goal | Equal |
| Study 2 | 1.20 | No | H | H | H | Reversible | H | M | M | L | No goal | No goal | Equal |
| Rhine (1967) | |||||||||||||
| 0–1 contradiction | −0.61 | No | H | H | H | Irreversible | H | M | H | H | No goal | No goal | Equal |
| 2 contradictions | 0.29 | Challenge | H | H | H | Irreversible | H | M | M | H | No goal | No goal | Equal |
| 3 contradictions | 0.37 | Challenge | H | H | H | Irreversible | H | M | M | H | No goal | No goal | Equal |
| 4 contradictions | 0.36 | Challenge | H | H | H | Irreversible | H | M | L | H | No goal | No goal | Equal |
| 5–6 contradictions | 0.86 | Challenge | H | H | H | Irreversible | H | M | L | H | No goal | No goal | Equal |
| No contradictions | 0.35 | No | H | H | H | Irreversible | H | M | H | H | No goal | No goal | Equal |
| Rosen (1961) | |||||||||||||
| Objective, high relevance | 0.65 | No | H | H | M | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Objective, low relevance | 0.42 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Essay, high relevance | 0.87 | No | H | H | M | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Essay, low relevance | 0.28 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Rosenbaum & McGinnies (1973) | |||||||||||||
| Study 1 | 0.50 | No | H | H | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Schulman (1971) | |||||||||||||
| High primary support, low secondary support | 0.37 | No | M | M | M | Irreversible | H | M | M | H | No goal | No goal | Equal |
| High primary support, high secondary | 0.41 | No | M | M | M | Irreversible | H | M | H | H | No goal | No goal | Equal |
| Moderate primary support, low secondary support | 0.41 | No | M | M | M | Irreversible | H | M | M | H | No goal | No goal | Equal |
| Moderate primary support, high secondary support | 0.72 | No | M | M | M | Irreversible | H | M | M | H | No goal | No goal | Equal |
| Low primary support, low secondary support | 0.81 | No | M | M | M | Irreversible | H | M | L | H | No goal | No goal | Equal |
| Low primary support, high secondary support | 0.70 | No | M | M | M | Irreversible | H | M | M | H | No goal | No goal | Equal |
| Schulz-Hardt et al. (2000) | |||||||||||||
| Study 1 | 1.01 | No | H | H | H | Irreversible | L | M | M | L | H | H | Equal |
| Schwarz et al. (1980) | |||||||||||||
| Supportive essay, one-sided support | 0.25 | Support | H | H | H | Irreversible | L | M | H | H | No goal | No goal | Equal |
| Supportive essay, two-sided support | −0.52 | Support | H | H | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Supportive essay, one-sided challenge | 1.58 | Challenge | H | H | H | Irreversible | L | M | L | H | No goal | No goal | Equal |
| Supportive essay, two-sided challenge | 0.79 | Challenge | H | H | H | Irreversible | L | M | M | H | No goal | No goal | Equal |
| No essay, one-sided support | 0.64 | Support | H | H | M | Irreversible | L | M | H | H | No goal | No goal | Equal |
| No essay, two-sided support | −0.38 | Support | H | H | M | Irreversible | L | M | M | H | No goal | No goal | Equal |
| No essay, one-sided challenge | 0.20 | Challenge | H | H | M | Irreversible | L | M | L | H | No goal | No goal | Equal |
| No essay, two-sided challenge | −0.08 | Challenge | H | H | M | Irreversible | L | M | M | H | No goal | No goal | Equal |
| Sears (1965) | |||||||||||||
| Old information | −0.64 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| New information | −0.50 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Sears (1966) | |||||||||||||
| No summation | −0.18 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Agrees with summation | −0.40 | Support | H | H | M | Irreversible | L | M | H | L | No goal | No goal | Equal |
| Disagrees with summation | 0.53 | Challenge | H | H | M | Irreversible | L | M | L | L | No goal | No goal | Equal |
| Two opposed summations | −0.03 | No | H | H | M | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Sears & Freedman (1963) | |||||||||||||
| Low commitment, expression goal | 0.35 | No | H | H | L | Reversible | L | M | M | L | H | H | Congenial |
| Low commitment, no goal | 0.35 | No | H | H | L | Reversible | L | M | M | L | No goal | No goal | Equal |
| High commitment, expression goal | 0.11 | No | H | H | H | Irreversible | L | M | M | L | H | H | Congenial |
| High commitment, no goal | 0.17 | No | H | H | H | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Sears & Freedman (1965) | |||||||||||||
| Convict, new information | −1.00 | No | H | H | H | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Acquit, new information | 0.06 | No | H | H | H | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Convict old information | −0.31 | No | H | H | H | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Acquit, old information | −0.02 | No | H | H | H | Irreversible | L | M | M | L | No goal | No goal | Equal |
| Smith et al. (2007) | |||||||||||||
| Study 1 | |||||||||||||
| Expression goal | 0.87 | No | M | M | H | Irreversible | H | M | M | L | H | H | Congenial |
| No goal | 0.21 | No | M | M | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Study 2 | |||||||||||||
| Expression goal, time pressure | 0.99 | No | M | M | H | Irreversible | H | M | M | L | H | H | Congenial |
| Expression goal, no time pressure | 0.26 | No | M | M | H | Irreversible | H | M | M | L | H | H | Congenial |
| No goal, time pressure | 0.33 | No | M | M | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| No goal, no time pressure | 0.15 | No | M | M | M | Irreversible | H | M | M | L | No goal | No goal | Equal |
| Thayer (1969) | |||||||||||||
| High confidence | 0.07 | No | M | M | M | Irreversible | L | M | H | L | No goal | No goal | Equal |
| Low confidence | 0.36 | No | M | M | M | Irreversible | L | M | L | L | No goal | No goal | Equal |
Note. The following abbreviations were used for the columns: H = High; L = Low; M = Moderate; No = No challenge or support.
After completing the coding, we calculated effect sizes (g) representing selective exposure from means and standard deviations, proportions or frequencies, F-ratios, t-tests, and correlations. When a report included means (e.g., ratings of interest in the information), we calculated g by subtracting the mean ratings of the uncongenial information from the mean ratings of the congenial information and dividing by the pooled standard deviation. From other documents, g was estimated from t-tests or F-ratios. For proportions, an odds or an odds ratio was calculated. When there was a mutually exclusive choice between congenial and uncongenial information (i.e., selecting a congenial article meant not selecting an uncongenial article), the odds of selecting congenial information was calculated by dividing the proportion of participants choosing congenial information by the proportion choosing uncongenial information. When there were independent choices of congenial and uncongenial information, we calculated separate odds and then an odds ratio by dividing the odds for congenial information by the odds for uncongenial information. To produce g, the log of the odds or the odds ratio was divided by 1.81 (Haddock, Rindskopf, & Shadish, 1998; Hasselblad & Hedges, 1995). All gs were converted to ds to correct for sample size bias (Hedges & Olkin, 1985). Positive ds indicate greater selection of congenial information, negative ds indicate greater selection of uncongenial information, and zero indicates the absence of bias.
We used Hedges and Olkin's (1985) procedures to calculate weighted mean effect sizes, effect sizes (d) and to estimate a homogeneity statistic (Q). Q has a distribution similar to a chi-square with k −1 degrees of freedom, where k is the number of effect sizes, and indicates whether the variance in effect sizes is no greater than sampling error. When a d implied a within-subjects comparison (e.g., between mean ratings of congenial and uncongenial information), the correlation between the two measures can be used to calculate the between-subjects variance in the statistic (Morris, 2000). We estimated this correlation (r = .27) using procedures suggested by Seignourel and Albarracín (2002) and then calculated the variance of the effect sizes using this imputed correlation.3 When d implied a between-subjects comparison, we used Hedges and Olkin's (1985) procedures to calculate the between-subjects variance in the statistic.
In the absence of homogeneity, we examined whether our moderators, entered alone and jointly with other moderators, accounted for variability among effect sizes using both fixed-effects and random-effects models.4, 5 In addition, we examined whether the effects of the moderators replicated using only effect sizes that derived from studies that measured or manipulated the moderator variable of interest. Because these analyses relied on a smaller number of cases, only univariate analyses using fixed- and random-effects models are presented. These analyses ensure that the effects of moderators are not due to uncontrolled differences across studies. We analyzed the effects of the moderators on selective exposure using analysis of variance. In this type of analysis, the inverse of the variance of the effect size being predicted is used as a weight and the significance of the moderators of interest are determined by examining the significance of the QB, which is a sums of squares comparable to an F-ratio but distributed similar to a chi-square with l −1 degrees of freedom, where l is the number of levels of the moderator. QBs were obtained to test for the main and simple effects of the moderator variable on selective exposure.
Moderators
Potential moderators were independently coded by two of the authors with adequate agreement (average kappa = .79; all kappas > .70). Disagreements were resolved by discussion with a third author.
For descriptive purposes, we recorded (a) year of publication; (b) publication form (journal article, unpublished dissertation or thesis, or other unpublished document); (c) participant population (university students, high school students, other, or mixed); (d) country where the study was conducted (United States and Canada, Germany, Australia, or Italy); (e) research setting (lab or field); (f) type of issue used in the study (e.g., politics, religion and morality, game play, betting and buying behavior, or personal health and development); (g) artificiality of issue (artificial, e.g., a hypothetical hiring decision, or real, e.g., abortion); (h) breadth of issue (broad, e.g., euthanasia, or narrow, e.g., decision about the guilt of a particular defendant); (i) exposure measure (choice of information to receive, rating of information preference, or ranking of information preference); (j) amount of congenial and uncongenial information offered for selection (number of congenial choices and number of uncongenial choices in the selection array); (k) psychological predictor used in the research (attitude, belief, or behavior); (l) the anonymity of the attitude, belief, and choice (anonymous or not anonymous); and (m) the novelty of the congenial and uncongenial information offered for selection (familiar or novel).
Coding of Potential Motivational Moderators
To examine the motivational determinants of selective exposure, we coded several variables with potential motivational properties (see Figure 1).
Defense motivation
In some studies participants' pre-existing attitudes, beliefs, and behaviors were challenged or supported prior to the information selection by learning that their decision was poor (vs. smart; e.g., Frey, 1982), hearing that their attitude was a minority (vs. majority) position (e.g., Nemeth & Rogers, 1996), and receiving more or less challenging (vs. supporting) information (e.g., Berkowitz, 1965). We coded challenge or support received prior to information selection as challenge (i.e., more uncongenial than congenial information received), no challenge or support (i.e., neither congenial nor uncongenial information received or equal amounts of congenial and uncongenial information received), or support (i.e., more congenial than uncongenial information received).
Also, we coded the quality of the available information presented for selection as high when the presumed source of the information was an expert on the topic (e.g., a scientist) and low when the presumed source was a novice or a peer (e.g., in a financial decision, high for an economics professor and low for a fifteen-year-old student or a passerby on the street; Frey, 1981b). When the source was neither clearly high nor low in expertise (e.g., a newspaper columnist or magazine writer), quality was coded as moderate.
We coded participants' commitment to their pre-existing attitude, belief, or behavior as high, moderate, or low. Commitment was high if the participants (a) justified (e.g., Jonas & Frey, 2003b; Schwarz et al., 1980) or anticipated having to justify (Canon, 1964; Janis & Rausch, 1970; Lowin, 1969) an attitude, belief, or behavior to an audience; (b) freely spent a relatively large amount of time or effort on a given behavior (e.g., playing a game; Betsch et al., 2001; smoking; Brock, 1965; writing random numbers, Frey & Wicklund, 1978); (c) engaged in sequential information searches (Jonas, Graupmann, & Fischer, 2003), which are known to enhance commitment to the decision (Jonas et al., 2001); (d) thought about their own death (Jonas, Greenberg, & Frey, 2003; Lavine, Lodge, & Freitas, 2005), which is known to enhance commitment to attitudes, beliefs, and behaviors that are tied to world views (review by Pyszczynski, Greenberg, Solomon, Arndt, & Schimel, 2004); or (f) reported that they held their attitude or belief with high commitment (Jonas & Frey, 2003a; Rhine, 1967) or viewed the belief as relevant to their self-worth (e.g., intelligence; Frey & Stahlberg, 1986; sociability, Holton & Pysczynski, 1989). Commitment to a pre-existing attitude, belief, or behavior was low when the participants freely engaged in attitude-inconsistent behavior (Cotton & Hieser, 1980), did not freely choose their behavior, attitude, or beliefs (e.g., behavior was assigned; Frey & Wicklund, 1978), or indicated a low amount of commitment to the choice (Jonas & Frey, 2003a). When commitment was not clearly high or low, it was coded as moderate. In addition, we coded for the reversibility of participants' reported attitudes, beliefs, or behaviors by noting whether, at the time of information selection, participants believed that they could (reversible) or could not (irreversible) change their attitudes, beliefs, or behaviors at a later time in the experiment (e.g., Frey & Rosch, 1984).
We also coded the value relevance of the issue. Value relevance was high if the issue was judged to be connected to the participants' enduring values (e.g., abortion, euthanasia, how to raise children); otherwise value relevance was low (e.g., a specific hiring decision, choosing among gifts). We also coded, whenever possible, participants' closed-mindedness as high or low as assessed by Rokeach's (1960) Dogmatism Scale, Altemeyer's (1996) Right Wing Authoritarianism Scale, and the Repression-Sensitization scale (Byrne, 1964). If the sample was not partitioned on closed-mindedness, this variable was coded as moderate.6 Participants' confidence in their attitude, belief, or behavior was registered as high, moderate, or low. Confidence was high (low) if participants reported high (low) confidence in their attitude, belief, or behavior (e.g., Adams, 1961; Brechan, 2002; Berkowitz, 1965; Brodbeck, 1956), reported beliefs that were consistent (inconsistent) with their behavior (Feather, 1962), received bogus positive (negative) feedback about their ability to form accurate attitudes, beliefs, or decisions (e.g., Thayer, 1969), were placed in a positive (negative) mood state after forming a decision (Jonas et al., 2006),7 were provided positive (negative) self-relevant feedback (Micucci, 1972), or possessed low (high) dispositional levels of anxiety (Frey et al., 1986). Without a confidence manipulation or partitioning of the sample, confidence was coded as moderate.
Accuracy motivation
We coded outcome relevance of the topic as high if the issue could have foreseeable effects on participants' outcomes in the near future (e.g., a choice of a gift, use of a type of exam, career choice) or distant future (e.g., developing cancer from smoking); otherwise, outcome relevance was coded as low. For example, manipulations of outcome relevance had participants select potential dates assuming that they would (high outcomes relevance) or would not (low outcome relevance) actually date the person (Lowe & Steiner, 1968).
We coded the utility of the available information presented for selection as high or low for fulfilling an experimental goal. Utility was high if the available information was high or moderate in quality and novel, and could facilitate accomplishing an immediate goal in the session (e.g., deciding whether to extend the contract of a manager, or writing an essay to justify their beliefs, attitudes or behaviors) or low if it was low quality and familiar, and could not facilitate accomplishing an immediate goal. When no such goal was present, utility was coded as no goal. We also coded the relative utility of the available congenial and uncongenial information presented for selection (congenial more useful; equally useful; uncongenial more useful). Conditions were coded as equally useful when there was no immediate goal in the session or the congenial and uncongenial information were judged equally likely to facilitate or hinder goal attainment. For example, the congenial and uncongenial information would be equally useful for preparing to select among gifts (Jonas et al., 2005). However, uncongenial information would be more useful for preparing to debate (e.g., Canon, 1964) or to write an uncongenial essay (Hillis & Crano, 1973). Congenial information would be more useful for planning to discuss one's opinion (Canon, 1964; Smith, Fabrigar, Powell, & Estrada, 2007) or to defend one's attitudes, beliefs, or behaviors (Frey, 1981b; Lundgren, & Prislin, 1998).
Results
Distribution of Effect Sizes
Our effect sizes are displayed in the stem-and-leaf plot in Figure 2. We first analyzed the distribution of effect sizes to check for potential biases in the study retrieval or publication. To estimate potential study retrieval and publication biases, we examined the funnel plot of effect sizes (see Figure 3) and the normality of the distribution under examination (see Figure 4). For Figure 3, if no bias is present, the plot should take the form of a funnel centered on the mean effect size, with smaller variability as the sample size increases. Instead, in the presence of publication bias, there is a distortion in the shape of the funnel. If the true effect size is zero and there is bias, the plot has a hollow middle. If the true effect size is not zero, the plot tends to be asymmetrical, having a large and empty section where the estimates from studies with small sample sizes and small effect sizes would be located in the absence of bias. Following these guidelines, an examination of the plot in Figure 3 suggests no retrieval or publication bias.
Figure 2.
Stem-and-leaf plot of effect sizes (ds).
Figure 3.
This funnel plot presents mean effect sizes on the Y-axis and sample sizes on the X-axis; a symmetric and inverted funnel shape suggests no publication bias
Figure 4.
Normal quantile plot. The line on the diagonal indicates normality; the lines around the diagonal represent the 95% confidence interval around the normality line.
In addition to examining the funnel plot, we used the normal-quantile plot method to uncover evidence of bias (Wang & Bushman, 1999). In a normal-quantile plot, the observed values of a variable are plotted against the expected values given normality. If the sample of effect sizes is from a normal distribution, data points cluster around the diagonal; if the sample of effect sizes is biased by publication practices or eligibility criteria, data points deviate from the diagonal (Wang & Bushman, 1999). As can be seen from Figure 4, the standardized effect sizes followed a straight line and generally fell within the 95% confidence intervals of the normality line.
Study Characteristics
Prior to testing our hypotheses, we examined some descriptive characteristics of the samples in our meta-analysis. As shown in Table 2, samples generally (a) were published in earlier decades, (b) appeared in journals, (c) included college students as participants, (d) took place in The United States and Canada, and (d) with the exception of a minority of field studies, took place in the laboratory. In terms of the issues, conditions generally used issues that were (a) political (e.g., scandals, campaign issues, war); (b) real (e.g., abortion) rather than artificial (e.g., a bogus hiring decision); and (c) specific in scope (e.g., extending the contract of a particular manager) rather than general (e.g., euthanasia). Choices of information to receive were most frequently assessed and most often made between two pieces of congenial information and two pieces of uncongenial information. Information choices were most often predicted from measures of prior behaviors and measures that were not anonymous in the experimental setting. The congenial and uncongenial information offered for selection was most often novel rather than familiar.
Table 2.
Distribution of Descriptive Moderators
| Variables | Value |
|---|---|
| Median publication year | 1981 |
| Publication form | |
| Journal article | 279 (93%) |
| Unpublished document | 8 (3%) |
| Dissertation or master's thesis | 7 (2%) |
| Book chapter | 6 (2%) |
| Participant population | |
| University students | 252 (84%) |
| High school students | 35 (12%) |
| Other or mixed | 13 (4%) |
| Country where study was conducted | |
| United States and Canada | 147 (49%) |
| Germany | 139 (46%) |
| Australia | 10 (3%) |
| Italy | 4 (1%) |
| Research setting | |
| Lab | 257 (86%) |
| Field | 43 (14%) |
| Issue type | |
| Politics | 72 (24%) |
| Organization and business administration | 70 (23%) |
| Personal development, personal health, self-related | 69 (23%) |
| Religion and values | 51 (17%) |
| Buying behavior, game play, or betting | 38 (13%) |
| Artificiality of issue | |
| Real | 219 (73%) |
| Artificial or bogus | 81 (27%) |
| Generality of issue | |
| Specific | 169 (56%) |
| General | 131 (44%) |
| Exposure measure | |
| Choice of information to receive | 197 (66%) |
| Rating of information preference | 85 (28%) |
| Ranking of information preference | 18 (6%) |
| Modal amount of congenial information offered | 2 |
| Modal amount of uncongenial information offered | 2 |
| Predictor | |
| Behavior | 194 (65%) |
| Attitude | 63 (21%) |
| Belief | 43 (14%) |
| Anonymity of attitude, belief, or behavior | |
| Not anonymous | 224 (75%) |
| Anonymous | 76 (25%) |
| Novelty of congenial and uncongenial Information | |
| Familiar | 13 (4%) |
| Novel | 287 (96%) |
Note. Unless otherwise specified, values are number of conditions or samples, with percents in parentheses.
The distributions of other important descriptive characteristics appear in the third column of Table 3. For moderators relevant to defense motivation, typically (a) challenge or support of the pre-existing attitudes, beliefs, or behaviors was absent; (b) quality of the available congenial and uncongenial information for selection was high (vs. moderate or low); (c) commitment to the pre-existing attitude, belief, or behavior was moderate (vs. high or low); (d) reversibility of the pre-existing attitude, belief, or behavior was absent (irreversible; vs. present, reversible); (e) value relevance of the issues was low (vs. high); (f) closed-mindedness was high or low in the samples in which it was assessed; and (g) confidence in the pre-existing attitudes, beliefs, or behaviors was moderate (vs. low or high). For moderators relevant to accuracy motivation, a majority of the conditions pertained to issues that (a) were not outcome relevant and (b) did not provide an immediate goal in the session. In the conditions that did provide a goal, the available information presented for selection was generally high (vs. low) in utility. The correlations between the defense-motivation and accuracy-motivation moderators appear in Table 4. As one might expect, the quality of the congenial and uncongenial information intercorrelated highly, and the utility of the congenial and uncongenial information also intercorrelated highly. Although many of the other correlations were weak or non-significant, we used multiple-regression procedures to determine the independent contribution of each moderator.
Table 3.
Moderator Analyses for All Included Studies
| Moderator and level | d. | k | Fixed-effect QB | Fixed-adjusted effect QB | Random-effect QB | Random-adjusted effect QB |
|---|---|---|---|---|---|---|
| Challenge or support | 14.72*** | 10.44** | 3.32 | 3.15 | ||
| Challenge | 0.27a | 24 | ||||
| No challenge or support | 0.38a | 257 | ||||
| Support | 0.16b | 19 | ||||
| Quality of available congenial information | 43.82*** | 38.32*** | 10.41** | 7.14* | ||
| High | 0.41a | 173 | ||||
| Moderate | 0.37a | 100 | ||||
| Low | 0.00b | 23 | ||||
| Quality of available uncongenial information | 26.23*** | ---- | 7.64* | ---- | ||
| High | 0.40a | 173 | ||||
| Moderate | 0.37a | 100 | ||||
| Low | 0.01b | 23 | ||||
| Commitment to the attitude, belief, or behavior | 19.99*** | 25.06*** | 6.66* | 6.87* | ||
| High | 0.42a | 117 | ||||
| Moderate | 0.35b | 170 | ||||
| Low | 0.13c | 13 | ||||
| Reversibility of the attitude, belief or behavior | 0.06 | 1.62 | 4.35* | 0.75 | ||
| Reversible | 0.37a | 114 | ||||
| Irreversible | 0.36a | 186 | ||||
| Value relevance | 84.15*** | 67.63*** | 11.87*** | 10.19** | ||
| High | 0.51a | 120 | ||||
| Low | 0.24b | 180 | ||||
| Closed-mindedness | 17.85*** | 17.94*** | 4.05 | 3.48 | ||
| High | 6 | 0.69a | ||||
| Moderate | 288 | 0.36b | ||||
| Low | 6 | 0.11c | ||||
| Confidence in attitude, belief, or behavior | 11.40** | 14.40*** | 4.73# | 4.54 | ||
| High | 0.23a | 36 | ||||
| Moderate | 0.37b | 224 | ||||
| Low | 0.45 b | 40 | ||||
| Outcome relevance | 1.65 | 1.19 | 4.52* | 0.94 | ||
| High | 0.39a | 102 | ||||
| Low | 0.35a | 198 | ||||
| Utility of congenial information | 31.59*** | 0.47 | 12.19** | 0.84 | ||
| High | 0.36a | 123 | ||||
| No goal | 0.39a | 168 | ||||
| Low | −0.16c | 9 | ||||
| Utility of uncongenial information | 8.87** | ---- | 1.26 | ---- | ||
| High | 0.31a | 121 | ||||
| No goal | 0.39b | 168 | ||||
| Low | 0.51b | 11 | ||||
| Relative utility | 100.83*** | 66.92*** | 26.74*** | 17.05*** | ||
| Congenial more useful | 0.54a | 17 | ||||
| Both equally useful | 0.38b | 272 | ||||
| Uncongenial more useful | −0.39c | 11 |
Note; d. = weighted mean effect size; k = number of cases. QB = Homogeneity statistic distributed as a χ2 with degrees of freedom equal to the number of moderator levels minus one. Effect sizes (d.) were estimated using a fixed- and random-effects model. For d., positive numbers indicate approach to congenial information, whereas negative numbers indicated approach to uncongenial information. The fixed and random-effect QB reflect the between group effect of the moderator when entered independently into the respective model. To determine the significance of simple effects, a one-tailed criterion was used when a directional hypothesis was assessed; otherwise, a two-tailed criterion was used, d.s within columns not sharing subscripts are significantly different from each other at p < .05 when entered independently into the fixed-effect model. The fixed- and random-adjusted effect QB was estimated using the respective model with all (non-redundant) moderators in this table entered simultaneously into a regression equation.
p < .10
p < .05
p < .01
p < .001
Table 4.
Inter-correlations between Moderators
| Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Challenge or support (1) | ---- | −0.01 | −0.01 | 0.02 | −0.04 | 0.05 | 0.00 | −0.02 | 0.06 | −0.03 | −0.03 | 0.00 |
| Quality congenial (2) | ---- | 0.96** | 0.03 | 0.08 | −0.16** | 0.00 | 0.01 | −0.08 | 0.09 | 0.07 | −0.01 | |
| Quality uncongenial (3) | ---- | 0.03 | 0.08 | −0.16** | 0.00 | 0.01 | −0.08 | 0.06 | 0.07 | 0.02 | ||
| Commitment (4) | ---- | 0.02 | 0.02 | 0.00 | 0.08 | 0.13* | 0.08 | 0.05 | −0.07 | |||
| Reversibility (5) | ---- | −0.04 | 0.00 | −0.01 | 0.09 | 0.57** | 0.58** | 0.03 | ||||
| Value relevance (6) | ---- | 0.00 | 0.01 | 0.16** | −0.19** | −0.17** | 0.01 | |||||
| Closed-mindedness (7) | ---- | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | ||||||
| Confidence (8) | ---- | 0.02 | 0.06 | 0.06 | 0.00 | |||||||
| Outcome relevance (9) | 0.12* | 0.12* | 0.03 | |||||||||
| Utility congenial (10) | 0.81** | −0.25** | ||||||||||
| Utility uncongenial (11) | 0.16** | |||||||||||
| Relative utility (12) | ---- |
Note. Entries are Spearman's correlation coefficients. Levels of the moderators were coded as follows: challenge or support (1= support; 2 = no challenge or support; 3 = challenge), quality congenial and uncongenial (1 = low; 2 = moderate; 3 = high), commitment (1 = low; 2 = moderate; 3 = high), reversibility (1 = irreversible; 2 = reversible), value relevance (1 = low; 2 = high), closed-mindedness (1 = low; 2 = moderate; 3 = high), confidence (1 = low; 2 = moderate; 3 = high), outcome relevance (1 = low; 2 = high), utility congenial (1 = low; 2 = moderate; 3 = high), utility uncongenial (1 = low; 2 = moderate; 3 = high), relative utility (1 = congenial more useful; 2 = both equally useful; 3 = uncongenial more useful).
p < .05
p < .01
Average Exposure Effect Size and Between-Effect Variability
We first obtained a weighted-mean average of information preferences and tested for variability among effect sizes. The average effect was d. = 0.36 (95% CI = 0.34, 0.39) according to fixed-effects analysis, indicating a moderate congeniality bias, and d. = 0.38 (95% CI = 0.32, 0.44) according to the random-effects analysis, indicating a moderate congeniality bias as well. Both of these average effects were statistically different from zero, Q (299) = 611.57, p < .001 for the fixed-effects analysis and Q (299) = 132.02, p < .001 for the random-effects analysis, and were heterogeneous, Q (299) = 1,354.55, p < .001 for the fixed-effects analysis and Q (299) = 372.45, p < .001 for the random-effects analysis. Notably, the mean unweighted effect size of 0.38 was similar to both of these estimates.
Moderator Analyses
Because there was a large amount of variability between effect sizes, we tested whether our moderators accounted for a significant amount of this variability. Generally, the results from fixed- and random-effects models converged. Thus, we focus on the fixed-effects models, which are more powerful, and are summarized in columns four and five of Table 3 (but see the sixth and seventh column of Table 3 for random-effects results). Table 3 presents analyses of all conditions, which provide the most complete description of our synthesis. Table 5 presents analyses using only the effect sizes for which the levels of the moderator varied within a study; these analyses protect against different levels of a moderator being spuriously confounded with study characteristics. Therefore, the Table 3 analyses included all samples, whereas the Table 5 analyses relied on studies with manipulations or partitioning based on a particular moderator. Importantly, the patterns of cell means were generally similar across these two types of analyses.
Table 5.
Moderator Analyses for Studies with Variability in the Levels of Moderators
| Moderator and level | d. | k | Fixed-effect QB | Random-effect QB |
|---|---|---|---|---|
| Challenge or support | 2.21 | 1.07 | ||
| Challenge | 0.27a | 24 | ||
| No challenge or support | 0.15a | 11 | ||
| Support | 0.16a | 19 | ||
| Quality of congenial information | 24.14*** | 4.49* | ||
| High | 0.55a | 8 | ||
| Low | −0.01b | 8 | ||
| Quality of uncongenial information | 4.53* | 1.33 | ||
| High | 0.38a | 8 | ||
| Low | 0.13b | 8 | ||
| Commitment to the attitude, belief, or behavior | 21.80*** | 7.97* | ||
| High | 0.43a | 54 | ||
| Moderate | 0.24b | 57 | ||
| Low | 0.15b | 9 | ||
| Reversibility of the attitude, belief or behavior | 0.41 | 0.24 | ||
| Reversible | 0.10a | 6 | ||
| Irreversible | 0.18a | 6 | ||
| Value relevance | 10.58** | 4.24* | ||
| High | 0.80a | 6 | ||
| Low | 0.33b | 6 | ||
| Closed-mindedness | 17.82*** | 4.02* | ||
| High | 0.69a | 6 | ||
| Low | 0.11b | 6 | ||
| Confidence in attitude, belief, or behavior | 18.87** | 5.82# | ||
| High | 0.19a | 30 | ||
| Moderate | 0.29a | 16 | ||
| Low | 0.53b | 20 | ||
| Outcome relevance | 0.67 | 1.58 | ||
| High | 0.22a | 25 | ||
| Low | 0.28a | 23 | ||
| Utility of congenial information | 21.82*** | 6.68* | ||
| High | 0.32a | 21 | ||
| No goal | 0.13b | 10 | ||
| Low | −0.17c | 9 | ||
| Utility of uncongenial information | 25.59*** | 9.25** | ||
| High | 0.13a | 19 | ||
| No goal | 0.13a | 10 | ||
| Low | 0.74b | 7 | ||
| Relative utility | 92.48*** | 23.48*** | ||
| Congenial more useful | 0.54a | 17 | ||
| Both equally useful | 0.19b | 13 | ||
| Uncongenial more useful | −0.39c | 11 |
Note. d. = weighted mean effect size; k = number of cases. QB = Homogeneity statistic distributed as a χ2 with degrees of freedom equal to moderator levels minus one. Effect sizes (d.) were estimated using a fixed- and random-effects model. For d., positive numbers indicate approach to congenial information, whereas negative numbers indicated approach to uncongenial information. The random effects QB reflect the between-group effect of the moderator when entered independently into the random-effects model. The fixed effects QB reflect the between-group effect of the moderator when entered independently into the fixed-effects model. To determine the significance of simple effects, a one-tailed criterion was used when a directional hypothesis was assessed; otherwise, a two-tailed criterion was used. d.s within columns not sharing subscripts are significantly different from each other at p < .05 according to the fixed-effect model.
p < .10
p < .05
p < .01
p < .001
Defense Motivation
Six of seven of our findings provided at least partial support for the hypothesis that defense motivation enhances the congeniality bias (see Figure 1). First, as anticipated, the congeniality bias was smaller when there was support rather than no challenge or support of the preexisting attitude, belief, or behavior prior to information selection. However, the congeniality bias was not larger when there was a challenge rather than no challenge or support prior to information selection. Second, as predicted, the congeniality bias was larger when the uncongenial or congenial information available for selection was high or moderate in quality (vs. low), although the high and moderate levels did not differ from one another. Third, as anticipated, the congeniality bias was larger for samples with high than moderate commitment to an attitude, belief, or behavior and smaller for samples with low than moderate commitment. Fourth, the congeniality bias was larger when the value relevance of the issue was high than low. Fifth, as expected, the congeniality bias was larger for samples high in closed-mindedness (vs. moderate) and smaller for samples low in closed-mindedness (vs. moderate). Sixth, the congeniality bias was smaller among samples with high (vs. moderate or low) confidence in the attitude, belief, or behavior. Although many of the findings supported the hypothesis that defense motivation enhanced the congeniality bias, one finding did not. Specifically, although the fixed-effects analysis showed that the congeniality bias was not influenced by whether the attitude, belief, or behavior was reversible, the random-effects analysis showed that the congeniality bias was larger when the attitude, belief, or behavior was reversible (vs. irreversible; d = 0.47 vs. 0.32).
Accuracy Motivation
Most of our major findings were consistent with the hypothesis that accuracy motivation can guide information selection (see Figure 1). First, as anticipated, the congeniality bias was larger when the congenial information was highly useful relative to when it was not useful or when there was no experimental goal. In fact, an uncongeniality bias appeared when the congenial information was not useful. Second, the congeniality bias was smaller when the uncongenial information was high than low in utility or when there was no goal. Third, as hypothesized, the congeniality bias was larger when the congenial information was more useful than the uncongenial information rather than when they were equally useful. In addition, the congeniality bias was smaller (and reversed) when the uncongenial information was more useful than the congenial information rather than when they were equally useful. Two findings were inconsistent with the hypothesis that accuracy motivation guides exposure decisions. First, although the fixed-effects analysis showed that the congeniality bias was not influenced by the outcome relevance of the issue, the random-effects analysis showed that the congeniality bias was larger when issues were high in outcome relevance (vs. low; d = 0.48 vs. 0.33). Second, the congeniality bias was larger when the uncongenial information was high or moderate in quality rather than low in quality. This latter finding, as mentioned earlier, supports defense motivation predictions more than accuracy motivation predictions.
Defense Motivation vs. Accuracy Motivation: Relative Contributions
To examine the relative influence of defense and accuracy motivations on the congeniality bias, we entered all seven non-redundant defense motivation moderators (i.e., challenge or support, quality of available congenial information, commitment, reversibility, value relevance, closed-mindedness, confidence) and the two accuracy motivation moderators (i.e., relative utility, outcome relevance) into a hierarchical regression analysis. Prior to entering these variables, they were dummy-coded with l - 1 dummy codes for each variable, where l represents the number of levels in the moderator. For example, challenge or support had two dummy codes. One dummy code represented a comparison between challenge and the other two groups (1 = challenge, 0 = support and no challenge or support), and the other dummy code represented a comparison between support and the other two groups (1 = support, 0 = challenge and no challenge or support). Note that when these two dummy codes are entered into a regression equation simultaneously, they completely account for the effect of the variable on congeniality (for more on dummy-coding see Keith, 2006).
The congeniality bias was predicted using a hierarchical-regression analysis with the defense-motivation moderators entered in the first step and the accuracy-motivation moderators entered in the second step. This analysis revealed that the defense-motivation moderators alone accounted for a significant amount of variance (13%; QR = 179.64, p < .001). Importantly, adding the accuracy-motivation moderators accounted for an additional 7% of the variance, which was significant (QR = 90.61, p < .001). Thus, it seems that both of these variables may contribute to selective exposure, but as the moderate-sized congeniality bias (d. = 0.36) would imply, defense motivation has a greater influence. Indeed, when we entered the accuracy-motivation moderators in the first step and defense-motivation moderators in the second step (i.e., reversed the order of entry), results were similar (accuracy accounted for 8% and defense accounted for 13% of the variance). Note that the individual effects of the moderators in this analysis are presented in the fifth column of Table 3.
Supplementary Analyses and Analyses of Descriptive Moderators
Comparing the analyses of the studies that varied the levels of the moderator (Table 5) with the analyses of all conditions (Table 3), we find a large amount of agreement. As can be gleaned from Table 5, the patterns of cell means were comparable for all nine of the moderator analyses that were significant according to both analyses. Challenge or support was the only moderator that failed to reach conventional levels of significance for this analytic approach but did for the analyses of all conditions.
Table 6 contains analyses for the descriptive moderators. Of the 16 descriptive moderators, 12 were significant predictors of information selection. The year the paper was published and the amount of congenial and uncongenial pieces of information in the selection array were each positively correlated with congeniality scores. In addition, congeniality biases were generally larger when reported in dissertations and theses, when the study concerned religion and values or politics, when the issues were real and general, when belief (vs. attitudes and behaviors) was the predictor, when participants ranked the information, and when the samples were not composed entirely of college and high school students. Possible interpretations of these findings appear in the general discussion.
Table 6.
Descriptive Moderator Analyses
| Moderator and level | B | k | Fixed-effect QB | Fixed-effect adjusted QB | Random-effect QB |
|---|---|---|---|---|---|
| Publication year (Median = 1981) | 0.20 | 300 | 53.24*** | 13.67** | 14 74*** |
| Amount of congenial information (Mode = 2) | 0.08 | 284 | 7.70** | 4.65* | 6.37* |
| Amount of uncongenial information (Mode = 2) | 0.08 | 284 | 7.96** | 5.25* | 6.49* |
| d. | |||||
| Publication form | 53.63*** | 47.66*** | 6.69 | ||
| Journal article | 0.35a | 279 | |||
| Book chapter | 0.25a | 6 | |||
| Dissertation or master's Thesis | 1.00b | 7 | |||
| Unpublished document | 0.28a | 8 | |||
| Country where study was Conducted | 0.51 | 2.09 | 4.59 | ||
| United States and Canada | 0.37a | 147 | |||
| Germany | 0.37a | 139 | |||
| Australia | 0.34a | 10 | |||
| Italy | 0.29a | 4 | |||
| Research setting | 0.01 | 0.33 | 0.06 | ||
| Laboratory | 0.36a | 257 | |||
| Field | 0.36a | 43 | |||
| Issue type | 50.38*** | 4.67 | 12.59** | ||
| Politics | 0.46a | 72 | |||
| Organization and business administration | 0.20b | 70 | |||
| Personal development, personal, health, self-related issues | 0.36c | 69 | |||
| Religion, and values | 0.48a | 51 | |||
| Buying behavior, game play, or betting | 0.27bc | 38 | |||
| Artificiality of issue | 8.61** | 0.41 | 0.12 | ||
| Real | 0.39a | 219 | |||
| Artificial or bogus | 0.28b | 81 | |||
| Generality of issue | 84.34*** | 10.96*** | 14.49*** | ||
| Specific | 0.23a | 169 | |||
| General | 0.50b | 131 | |||
| Predictor | 34.41*** | 11.43** | 0.39 | ||
| Behavior | 0.29a | 194 | |||
| Attitude | 0.42b | 63 | |||
| Belief | 0.53c | 43 | |||
| Exposure measure: Choice of information to receive | 12.32*** | 1.23 | 4.29* | ||
| Yes | 0.41a | 186 | |||
| No | 0.30b | 114 | |||
| Exposure measure: Rating of Preference | 25.00*** | 33.56*** | 12.24*** | ||
| Yes | 0.26a | 85 | |||
| No | 0.42b | 215 | |||
| Exposure measure: Ranking of Preferences | 6.10* | 15.54*** | 1.99 | ||
| Yes | 0.51a | 18 | |||
| No | 0.35b | 282 | |||
| Participant population | 21.83*** | 26.69*** | 5.81# | ||
| University students | 0.35a | 252 | |||
| High school students | 0.35a | 35 | |||
| Other or mixed | 0.66b | 13 | |||
| Anonymity of attitude, belief, or behavior | 16.08*** | 0.03 | 1.72 | ||
| Not anonymous | 0.32a | 224 | |||
| Anonymous | 0.45b | 76 | |||
| Novelty of congenial and uncongenial information | 0.54 | 0.07 | 0.44 | ||
| Familiar | 0.31 | 13 | |||
| Novel | 0.37 | 287 |
Note. B = slope; d. = weighted mean effect size; k = number of cases. QB = Homogeneity statistic distributed as a χ2 with degrees of freedom equal to one minus the levels of the moderator. Effect sizes (d) were estimated using a fixed- and random-effects model. For d, positive numbers indicate approach to congenial information, whereas negative numbers indicated approach to uncongenial information. The fixed effect (random effect) QB reflects the between group effect of the variable when entered independently into the fixed effects (random-effects) model. The fixed-adjusted effect QB was estimated using the respective model with all (non-redundant) defense and accuracy moderators entered simultaneously into a regression equation. To determine the significance of simple effects, a two-tailed criterion was used, d.s within columns not sharing subscripts are significantly different from each other at p < .05 according to the fixed-effect model.
p < .10,
p < .05,
p < .01,
p < .001
GENERAL DISCUSSION
People's attitudes and behaviors are often inappropriate and inaccurate, as is the case when investors make a poor investment decision, physicians misdiagnose patients, and children persist in their belief in Santa Claus. Although information relevant to these attitudes and behaviors can provide opportunities for change, our review demonstrates biases in what information is selected for reception. People are almost two times (OR = 1.92, based on d. = 0.36) more likely to select information congenial rather than uncongenial to their pre-existing attitudes, beliefs, and behaviors. The moderate size of the bias is perhaps not surprising given that selective exposure is responsive to motivations that can occasionally exert opposing influences on selection preferences. As our analyses have shown, variables associated with defense motivation (e.g., commitment, value relevance, confidence, and challenge or support) uniformly increased the selection of congenial information. In contrast, information utility, a moderator associated with the accuracy motivation, increased or decreased the preference for congenial information, depending on whether the congenial or uncongenial information possessed a utility advantage. Selecting congenial information can facilitate feeling validated about one's view or even maintaining stable views of the world but may reduce accuracy and flexibility. Hence, the occasionally opposing influences of defense and accuracy motivation create a balance between defending prior views and obtaining realistic views of an object or issue.
Motivational Factors
Several theorists have proposed that accuracy and defense motivations guide human behavior (e.g., Chaiken et al., 1996; Jonas et al., 2005; Katz, 1960; Smith, Bruner, & White, 1956; Wyer & Albarracín, 2005). People are presumed to want to believe in the accuracy of their views (a result of defense motivation) but also attain views that are rooted in external reality (a result of accuracy motivation; for broader theories, see Baumeister, 2005; Schlenker, 1980). Consistent with this notion of human motivation (see Figure 1), our meta-analysis confirmed that exposure decisions are guided by defense and accuracy motivation.
Defense Motivation
The majority of our findings showed that a congeniality bias increases as a function of factors that presumably increase defense motivation. As expected, the congeniality bias was positively correlated with information quality, commitment, value-relevance, and closed-mindedness, but negatively correlated with confidence in or support given to one's pre-existing attitude, belief, or behavior. Although the majority of our findings suggested that defense motivation affects selective exposure, one finding did not. In particular, we predicted that irreversible decisions would promote a greater congeniality bias because people experience greater affective attachment to their irreversible decisions than their reversible ones (Kiesler, 1971; Schlenker, 1980). Although a fixed-effects analysis revealed that the ability to reverse a pre-existing attitude, belief or behavior had no effect on the congeniality bias, a random-effects analysis showed that the congeniality bias was larger when prior attitudes, beliefs, or behaviors could be reversed.
Another possible interpretation of the reversibility effect is that the ability to change one's position may enhance the experience of cognitive dissonance by prompting a consideration of reasons to change the position. For example, the possibility for change may automatically direct attention to why the unchosen position might be better than the chosen position. Consequently, dissonance arousal may be greater and congeniality more pronounced under reversible-decision conditions. Alternatively, the perceived ability to change one's position may enhance attempts to crystallize and defend this position (Dewey, 1938; Kruglanksi, 1990; Lewin, 1951; Pierce, 1877; Tajfel, 1969). Yet another possibility is that the perceived ability to change a decision enhances the congeniality bias by directly improving memory for the contents (e.g., beliefs) and decision-making strategies (e.g., congenial information searches) associated with that incomplete decision (Zeigarnik, 1927). Future work may disentangle these possibilities, perhaps as a function of individual differences in variables such as closed-mindedness (e.g., need for cognitive closure) and through assessments of memory. At present, the accumulated data are insufficient to explore these issues further.
Accuracy Motivation
Our meta-analysis revealed that participants selected information that best suited the goal they were pursuing in the session. Studies showed that selection favored congenial information when the congenial information was useful but favored uncongenial information when the uncongenial information was useful. Less supportive of the role of accuracy motivation in selective exposure were associations involving information quality and outcome relevance. The expected preference for high-quality congenial information was present, even though the expected preference for high-quality uncongenial information was absent. Importantly, this pattern was entirely consistent with the role of defense motivation but was only partially consistent with the role of accuracy motivation. Also, contrary to the possibility that outcome relevance negatively correlates with the congeniality bias, the random-effects analysis showed that the correlation was positive. However, closer inspection revealed that outcome relevance was correlated with value relevance (rs = .16, p = .005; see Table 4). In an analysis controlling for value relevance, outcome relevance no longer had a significant effect on congeniality (p > .10).
Effects of Descriptive Variables
Some of the effects of the descriptive variables on the congeniality bias (see Table 6) might reflect defense and accuracy motivation. For example, the findings that congeniality biases are enhanced for general issues, real issues, and belief-relevant topics may reflect enhanced defensiveness in these conditions. Real and belief-relevant issues are also more value relevant, and so value relevance should be responsible for these associations. Indeed, in analyses controlling for value relevance, neither variable significantly predicted the congeniality bias (ps > .10). Why general rather than specific issues (e.g., capital punishment vs. the guilt of a defendant) enhances the congeniality bias is less clear but might reflect the fact that general issues bring to mind many specific beliefs, attitudes, and behaviors. If so, disagreement on general issues may arouse more cognitive conflict than disagreement on specific issues.
Still other findings may support a cognitive mechanism affecting the congeniality bias. For example, the positive association between congeniality and the number of pieces of congenial and uncongenial information in the selection array might suggest that larger arrays make prior attitudes and behaviors more accessible as a basis for the selection. Alternatively, larger arrays may create a cognitive load and hence promote tendencies to rely on heuristics that promote congenial selections (e.g., “if it is (un)congenial, then it is probably (un)reliable”). We also found that congeniality biases were greater when information preferences were measured by rankings as opposed to ratings or yes/no selections. Perhaps ranking methods require more thought about the information and thereby aid retrieval of past views. Alternatively, ranking methods may force direct comparisons among the information in the array and therefore better highlight the congeniality or uncongeniality of each piece of information. Further, the finding that student samples exhibited a smaller congeniality bias than non-student or mixed samples may be due to more mature individuals' practice with selective exposure. Student samples are ordinarily younger than non-student samples and therefore have less experience with the selective exposure process and less developed views (Sears, 1986).
Additional findings may reflect publication practices or methodological changes over time. For example, the congeniality bias was larger in unpublished reports as opposed to published reports. Perhaps the controversial history of selective exposure led journal editors to publish various types of findings, including null ones (see review by Freedman & Sears, 1965). Also, the positive correlation between report year and congeniality may reflect improved methodologies through the years. Researchers now possess more refined experimental methods and a better grasp of the competing causes of information selection that must be controlled when studying this issue.
Our Review in the Context of the Past Reviews
More than two decades have passed since Frey's (1986) and Cotton's (1985) influential reviews of selective exposure. Guided at least in part by these reviews, many new research reports with innovative methods have emerged since 1986. This accumulation of new data created an ideal opportunity for a review that quantifies the congeniality bias and determines its variability. In doing so, this meta-analysis yielded some conclusions that support the earlier reviews and some that do not.
Our study strongly supported the earlier conclusion that defense motivation enhances the congeniality bias (Cotton, 1985; Frey, 1986). Some our findings, however, were not obtained in past reviews. For example, past reviews concluded that attitudinal confidence and congeniality are unrelated (Cotton, 1985; Freedman & Sears, 1965), but our results suggested that congeniality is weaker at high (vs. low or moderate) levels of confidence. Also, whereas Frey (1986) concluded that congeniality is stronger when decisions are irreversible than reversible (Frey, 1986), our results revealed that congeniality is stronger when decisions are reversible. Yet, Frey's conclusion was based on only two studies (Frey, 1981c; Frey & Rosch, 1984), the first of which presented only congenial information.
In addition to exploring defense motivation, which was the theoretical foundation for the reviews by Cotton (1985) and Frey (1986), our analysis highlighted the critical role of accuracy motivation. Our conclusions on accuracy motivation are reminiscent of Freedman and Sears' (1965) view that, although attitudinal selectivity can occur, utility may be a more important guide for information choices. Consistent with this notion, our study showed a moderate-size uncongeniality bias when the uncongenial information was clearly of higher utility than the congenial information. Our estimates suggested that both defense and accuracy motivations predict exposure decisions but, as the mean effect size signals a predominance of congeniality, defense is a stronger predictor.
Our review has greatly amplified understanding of the variability of selective exposure effects. Whereas past reviews have analyzed effects of moderators only within individual studies, our study examined their effects both between and within studies. Moreover, by coding all studies on all moderators, our conclusions regarding moderators are based on far more information than prior reviews. The new moderators we introduced also proved to be important. For example, we assessed the effect of value relevance on selective exposure and found greater congeniality for high (vs. low) value-relevant topics. All in all, our review advances the selective exposure literature well beyond past reviews.
Future Directions
Congeniality at Other Stages of Information Processing
Past research has examined whether congeniality biases exist at all stages of information processing—exposure, interpretation, and memory. To date, however, only congeniality biases at exposure and memory have been estimated meta-analytically. In this regard, the current review estimated the congeniality bias at exposure to be moderate in size (d. = 0.36) and influenced by accuracy and defense motivations. In contrast, the congeniality bias in memory was smaller (d. = 0.23, albeit artificially increased by methodological problems that were prevalent in earlier studies) and was also moderated by accuracy and defense motivations (Eagly et al., 1999). The variance of the overall size of congeniality bias across these two stages is interesting and might suggest that the strength of defense and accuracy motivation vary accordingly. Therefore, to get a clearer picture of congeniality biases, future research should explore the size and variability of the bias at information interpretation (Bargh, 1999; Bruner, 1957; Darley & Gross, 1983; Duncan, 1976; Hastorf & Cantril, 1954; Lord, Ross, & Lepper, 1979).
Cognitive Factors in Selective Exposure
Although motivational mechanisms appear to underlie selective exposure, cognitive mechanisms are also likely to be critical. For example, the congeniality bias might increase along with people's ability to retrieve past attitudes, beliefs, and behaviors (e.g., attitude accessibility). Attitudes, beliefs, and past behaviors may automatically influence information selection by making the selections consistent with the retrieved attitude, belief, or behavior (e.g., Chen & Bargh, 1999; Greenwald, McGhee, & Schwartz, 1998). In addition, the ability to retrieve these tendencies may make congenial information easier to process than uncongenial information and hence more attractive (Winkielman & Cacioppo, 2001). Such considerations were not amenable to testing within this meta-analysis, and they are prime candidates for future research. For example, studying the development of selective exposure may show that older children (who have greater resources to recall prior attitudes and behaviors) show an enhanced congeniality bias compared to younger children. In addition, examining factors that affect attitude retrieval may show that factors that impede retrieval of prior attitudes (e.g., distraction) decrease the congeniality bias.
Impression Motivation and Selective Exposure
The kind of information that people select can convey preferences and other personal attributes, leading them to attempt to strategically manage their selections to establish a desired identity. In our meta-analysis, a tendency towards trying to appear unbiased was revealed by a weaker congeniality bias when attitudes, behaviors, and beliefs were not anonymous relative to anonymous (see Table 6). Nonetheless, future research should investigate self-presentation issues in greater depth. For example, the presence of an audience may affect selective exposure by affecting the perceived desirability of appearing receptive versus resolute (Schlenker, 1980, 1985, 2003; see Jonas et al., 2005). In addition to manifesting strategic forms of impression management, people may select information to develop (or maintain) relationships and create a shared reality with likeable others (Higgins, 1992). For example, to maintain a relationship with an attractive group, an individual may select information consistent with its views (Lundgren & Prislin, 1998). In contrast, to cut ties from an unattractive group, an individual may select information inconsistent with its views.
Controlled and Automatic Processes Underlying Selective Exposure
A critical consideration for changing and alerting individuals about biases in information selection is whether the selective exposure process is controlled or automatic (e.g., Fiske & Taylor, 1991; Shiffrin & Schneider, 1977). Yet, little research has addressed this question to date. On the one hand, people may make a conscious decision to select congenial information. In this case, the process of selecting information is effortful and intentional, and it occurs with conscious awareness and may be intentionally interrupted. On the other hand, people may reduce dissonance without conscious awareness or intention. In a dramatic demonstration of this fact, patients who suffered from anterograde amnesia (i.e., a condition that prevents the formation of new memories) re-ranked a piece of artwork more positively when they had previously chosen it than when they had not (Lieberman, Ochsner, Gilbert, & Schacter, 2001, Study 1). By the same token, then, defense motivation (and possibly accuracy motivation) may be elicited automatically after retrieving an attitude or making a decision. In this situation, the effects of prior attitudes, beliefs, or behaviors on exposure could be effortless, unintentional, devoid of awareness, and uncontrollable.
To our knowledge, only Fischer et al. (2005, Study 3) have studied the automatic nature of selective exposure. In their study, participants were asked to decide whether to extend the contract of a fictitious manager and then were offered additional information about the manager in either distracting (cognitive load) or nondistracting (control) conditions. The congeniality bias was smaller (and nonsignificantly reversed) in the cognitive-load condition than in the control condition, suggesting a controlled process. Importantly, however, information selection may be more or less automatic depending on the nature of the pursued goals. Defense motivation may be easily satisfied by selecting congenial information, whereas accuracy motivation may require complex procedures that involve conscious monitoring. For example, satisfying defense motivation may require monitoring the direction and quality of the information in relation to a prior attitude, belief, or behavior. In contrast, satisfying accuracy motivation may require monitoring the direction and quality of the information as well as attending to and correcting for any systematic biases in exposure (e.g., Harkness, DeBono, & Borgida, 1985; McAllister, Mitchell, & Beach, 1979; Tetlock, 1983; Tetlock & Kim, 1987). Given these possibilities, future research might explore the automatic and controlled processes that influence information selection.
Increasing (or Decreasing) Exposure through Goal Accomplishment
The motivation to defend an attitude may lead to seeking more congenial than uncongenial information until defense motivation is satisfied, at which point this motive may become deactivated or inhibited (Zeigarnik, 1927). As a result, if defense or accuracy motivation is satisfied by means of behaviors other than selective exposure (e.g., self-affirmation; Steele, 1988), effects on exposure may be attenuated or possibly reversed. Performing mathematical calculations correctly, for example, may increase the congeniality bias if this behavior satisfies accuracy motivation. This prediction is counterintuitive because the calculations could potentially activate accuracy-related procedures, thus enhancing rather than reducing accuracy.
Another issue deserving of future research is whether satisfaction of defense or accuracy motivation in one information-search domain affects future information selection in other domains. For example, allowing an individual to satisfy defense motivation by selecting and reading congenial information on abortion may result in less defensiveness when selecting information on euthanasia. Such a possibility has important implications for daily life because people often search for information about more than one issue.
Practical Implications of Our Meta-Analysis
Although our study implemented a correlational method to assess the effects of various factors on the congeniality bias and hence possesses the weaknesses associated with this method, it is unlikely that unidentified differences across the studies and conditions could completely account for the effects of the moderators on the congeniality bias. For example, we found that the effects of the moderators generally replicated using only effect sizes from studies that measured or manipulated the moderator variable of interest. In addition, multiple regression analyses showed that the effects of the moderators generally remained significant even after controlling for the other moderators. Also, we employed various measures of the motivational processes that were of interest (see Figure 1), and the alternate measures generally had the same effect on the congeniality bias.
Health-Promotion Intervention Planning
Selective exposure can have implications for the health and well-being of a society. For example, a recent meta-analysis of intervention acceptance and attrition found that about a quarter of eligible participants turned down an opportunity to participate in an HIV-prevention program (Noguchi, Albarracin, Durantini, & Glasman, 2007). Even more unfortunate, people who rarely wear condoms and hence are most in need of prevention programs were more likely to turn down these programs than people who consistently wear condoms (Noguchi et al., 2007). Presumably, individuals in need of intervention programs are more likely to avoid them because they anticipate that the programs will challenge their behavior.
Despite this resistance, there may be several strategies for increasing participation among an unwilling audience. Individuals may be motivated to attend such programs when the intervention is perceived as facilitating the attainment of valued goals. For example, people may have an inherent need to help others, especially their children (Baumeister, 2005; Maslow, 1968) but not be aware of how this goal can be facilitated by taking part in an HIV intervention program. If a program is framed as facilitating the ability to provide important knowledge that can be transmitted to one's children and family, people may participate to that end. This approach seems plausible given our finding that people seek uncongenial information when the information facilitates achieving a current goal (i.e., helping others in this case). Furthermore, prevention programs may increase acceptance rates by minimizing cues that can trigger defense motivation. For example, people may be more willing to participate in a program called a “health discussion group” than an “HIV intervention group” or “HIV counseling group.” By implying an intention to produce change, such words as “intervention” and “counseling” may automatically strengthen defense motivation and increase tendencies to avoid the program (Albarracin et al, in press).
Democracy and Selective Exposure
Individual choice rather than governmental choice of information is characteristic of a democracy. Moreover, democracies rely on the ability of citizens to access a range of available information and make intelligent choices based on this information. Despite having relatively few governmental restrictions on information, citizens may select certain newspapers, televised-news programs, radio programs, and magazines that suit their political ideology. A 2004 survey by The Pew Research Center found that Republicans are about 1.5 times more likely to report watching Fox News regularly than Democrats (34% for Republicans and 20% of Democrats). In contrast, Democrats are 1.5 times more likely to report watching CNN regularly than Republicans (28% for Democrats vs. 19% of Republicans). Even more striking, Republicans are approximately five times more likely than Democrats to report watching “The O'Reilly Factor” regularly, and are seven times more likely to report watching “Rush Limbaugh” regularly.
Our review found a stronger congeniality bias for political issues than other issues (d = 0.46; see Table 6). Moreover, our review suggests strategies for increasing exposure to uncongenial political information among citizens. Individuals should be motivated to seek uncongenial political information when this information best suits their goals. For example, a strong motivation to debate an issue (vs. express one's view) may promote a search for uncongenial information with the objective of counterarguing it (e.g., Albarracín & Mitchell, 2004; Canon, 1964; see also Smith et al., 2007). In addition, citizens might be led to seek uncongenial information if political discussion is framed as an opportunity to build rapport (vs. establish interpersonal distance) with uncongenial audiences (see Lundgren & Prislin, 1998). These important issues deserve future research attention.
Closing Note
Although information selection could potentially proceed under the influence of the motivation to feel validated or the motivation to gain an accurate understanding of reality, our review suggests that both motivations are important. It seems likely that these often antagonistic tendencies may compensate for the potential dangers of seeking only self-validating or accurate information. Whereas defense motivation facilitates psychological stability and personal validation, accuracy motivation promotes accurate perceptions of reality. Given current evidence, however, it appears that tendencies toward congeniality prevail.
Acknowledgments
The research was funded by grants from the National Institutes of Health (K02-MH01861 and R01-NR08325). We thank the attitudes laboratories at the Psychology Departments of the University of Florida and the University of Illinois at Urbana-Champaign for discussion of the ideas reported in this paper. Also, we thank Shelly Chaiken, Blair Johnson, Tarcan Kumkale, and Moon-Ho Ringo Ho for their helpful comments on an earlier version of this paper and KyuHee Lee for her assistance in coding studies.
Footnotes
Although confidence and commitment should exert opposite effects on selective exposure, they may, in practice, go hand-in-hand. Therefore, our predicted effect of confidence assumes that commitment is controlled at a moderate level and our predicted effect of commitment assumes that confidence is controlled at a moderate level.
Partitioning studies in this way (versus only partitioning studies based on moderators of interest) allows a single study to contribute more than one effect size (e.g., each condition or sub-sample within a study contributes an effect size). Although such sub-samples within the studies of a meta-analysis are assumed to be statistically independent (e.g., Lipsey & Wislon, 2001; Raudenbush & Bryk, 2002), some researchers have suggested that sub-samples from the same study may share minor statistical dependencies even though the participants are different (see Wolf, 1990). For this reason, we re-analyzed our data after partitioning studies based on only the moderators of interest. Essentially, this procedure involved averaging effect sizes across moderators (not of interest) within a single study to reduce the number of effect sizes coming from that study (potential dependence). Of note, this change in partitioning procedure reduced the number of effect sizes to 211 (i.e., 70% of the original sample, 300). A majority of this decrease in the number of effect sizes (i.e., 89) can be attributed to only six papers (i.e., 40 effect sizes; or 45% of the decrease; Fischer et al., 2005, 2008; Frey, 1982; Frey, 1981a,b; Frey & Wicklund, 1978), in which moderators were not directly relevant to our theoretical framework (e.g., limited vs. unlimited searches), or had additional levels of one of our moderator of interest (e.g., high, moderate or low levels of challenge). This more conservative partitioning procedure did not alter the pattern of our reported results for the moderator analyses.
To directly verify that our liberal partitioning strategy did not reduce the statistical independence of the effect sizes, we estimated the sampling error (see Lipsey & Wilson, 2001) for the effect sizes partitioned on only the moderators of interest (211 effect sizes; conservative partitioning strategy) and then for the effect sizes partitioned on the basis of the moderators used in the studies (300 effect sizes; liberal partitioning strategy). If the sampling error for the 211 effect sizes is larger than the sampling error for the 300 effect sizes, then the liberal (vs. conservative) partitioning procedure may have introduced dependencies in the data. Contrary to this possibility, however, the sampling error estimates were almost identical and thus suggested similar statistical independence. In fact, the sampling error for the sample of 211 (vs. 300) effect sizes was estimated to be slightly smaller (compare vθ = 0.23 vs. 0.24).
Due to a limited number of reports containing the statistics required to compute this correlation, we also calculated the variance of within-subject effect sizes using three different correlations between the preferences for congenial and uncongenial information to reflect extreme (r = .00 and r = .99) and moderate correlations (r = .50; see also Albarracín et al., 2003, 2005). Notably, the results were very similar across these various correlations, so we present only the ones with the imputed correlation (see also Albarracín et al., 2003, 2005).
Although fixed-effects models are “mixed” models (Raudenbush & Bryk, 2002), we chose to retain traditional meta-analytic terminology.
In the case of a fixed-effects model, one assumes a fixed population effect size and estimates its sampling variance, which is an inverse function of the sample size of each group. As a result, effects sizes generated from larger samples are considered to be more precise estimates of the fixed effect size and hence are weighted more heavily than effect sizes obtained from smaller samples. In contrast, random-effects models assume that effect sizes are sampled from a population of effect sizes. Hence, an effect size results from sampling an effect size at random (from a population of values) in addition to measurement error, which is an inverse function of the sample size. Because random-effects models account for these two sources of error in an effect size, they yield a larger error term and less statistical power than fixed-effects procedures. However, one of the benefits of the random-effects model (vs. the fixed-effects model) is the ability to generalize its results to a broader universe of studies.
In addition to comparing the congeniality bias across three groups of closed-mindedness, we compared only groups coded as high and low (see Table 5).
Past research indicates a fairly direct relation of confidence to positive and negative affect (see Erber, 1991; Forgas & Moylan, 1987; Johnson & Tversky, 1983; Salovey & Birnbaum, 1989). Hence, we coded positive mood as high confidence and negative mood as low confidence. Eliminating these conditions did not alter our results.
References
- Abelson RP. Conviction. American Psychologist. 1988;43:267–276. [Google Scholar]
- Adams JS. Reduction of cognitive dissonance by seeking consonant information. Journal of Abnormal and Social Psychology. 1961;62:74–78. doi: 10.1037/h0047029. [DOI] [PubMed] [Google Scholar]
- Adorno T, Frenkel-Brunswick E, Levinson D, Sanford N. The authoritarian personality. Harper & Row; New York: 1950. [Google Scholar]
- Albarracín D. Cognition and persuasion: An analysis of information processing in response to persuasive communications. In: Zanna MP, editor. Advances in experimental social psychology. Academic Press; San Diego, CA: 2002. pp. 61–130. [Google Scholar]
- Albarracín D, Durantini MR, Earl A, Gunnoe J, Leeper J. Beyond the most willing audiences: A meta-intervention to increase participation in HIV prevention intervention. Health Psychology. doi: 10.1037/0278-6133.27.5.638. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Albarracín D, Gillette J, Earl A, Glasman LR, Durantini MR, Ho MH. A test of major assumptions about behavior change: A comprehensive look at the effects of passive and active HIV-prevention interventions since the beginning of the epidemic. Psychological Bulletin. 2005;131:856–897. doi: 10.1037/0033-2909.131.6.856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Albarracín D, Johnson BT, Zanna MP. Handbook of attitudes. Lawrence Erlbaum; Hillsdale, New Jersey: 2005. [Google Scholar]
- Albarracín D, McNatt PS, Klein C, Ho R, Mitchell A, Kumkale GT. Persuasive communications to change actions: An analysis of behavioral and cognitive impact in HIV prevention. Health Psychology. 2003;22:166–17. doi: 10.1037//0278-6133.22.2.166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Albarracín D, Mitchell AL. The role of defensive confidence in preference for proatttitudinal information: How believing that one is strong can be a weakness. Personality and Social Psychology Bulletin. 2004;30:1565–1584. doi: 10.1177/0146167204271180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Altemeyer B. Right-wing authoritarianism. University Manitoba Press; Winnepeg: 1981. [Google Scholar]
- Altemeyer B. The authoritarian specter. Harvard University Press; Cambridge, MA: 1996. [Google Scholar]
- Altemeyer B. The other “authoritarian personality. In: Zanna MP, editor. Advances in experimental social psychology. Vol. 30. Academic Press; San Diego, CA: 1998. pp. 47–92. [Google Scholar]
- Anderson DR, Collins PA, Schmitt KL, Jacobvitz RS. Stressful life events and television viewing. Communication Research. 1996;23:243–260. [Google Scholar]
- Aronson E. Dissonance theory: Progress and problems. In: Abelson RP, Aronson E, McGuire WJ, Newcomb TM, Rosenberg MJ, Tannenbaum PH, editors. Theories of cognitive consistency: A sourcebook. Rand McNally; Chicago: 1968. pp. 5–27. [Google Scholar]
- Bacon F. New organon, and related writings. Liberal Arts Press; New York: 1620. 1960. [Google Scholar]
- Bargh JA. The cognitive monster: The case against controllability of automatic stereotype effects. In: Chaiken S, Trope Y, editors. Dual process theories in social psychology. Guilford; New York: 1999. pp. 361–382. [Google Scholar]
- Baumeister RF. The cultural animal: Human nature, meaning, and social life. Oxford University Press; New York: 2005. [Google Scholar]
- Beauvois JL, Joule RV. A radical dissonance theory. Taylor & Francis; Bristol, PA: 1996. [Google Scholar]
- Behling CF. Effects of commitment and certainty upon exposure to supportive and nonsupportive information. Journal of Personality and Social Psychology. 1971;19:152–159. [Google Scholar]
- Berkowitz L. Cognitive dissonance and communication preferences. Human Relations. 1965;18:361–372. [Google Scholar]
- Betsch T, Haberstroh S, Glöckner A, Haar T, Fiedler K. The effects of routine strength on adaptation and adaptation in recurrent decision-making. Organizational Behavior and Human Decision Processes. 2001;84:23–53. doi: 10.1006/obhd.2000.2916. [DOI] [PubMed] [Google Scholar]
- Boden JM, Baumeister RF. Repressive coping: Distraction using pleasant thoughts and memories. Journal of Personality and Social Psychology. 1997;73:45–62. doi: 10.1037//0022-3514.73.1.45. [DOI] [PubMed] [Google Scholar]
- Bosotti E. Cognitive elaboration and exposure to information: Structure of memory and reduction of dissonance. Giornale Italiano di Psicolgia. 1984;11:2. [Google Scholar]
- Brannon LA, Tagler MJ, Eagly AH. The moderating role of attitude strength in selective exposure to information. Journal of Experimental Social Psychology. 2007;43:611–617. [Google Scholar]
- Brechan I. Unpublished master's thesis. University of Florida, Gainesville; Florida, USA: 2002. Selective exposure and selective attention: The moderating effect of confidence in attitudes and the knowledge basis for these attitudes. [Google Scholar]
- Brehm JW, Cohen AR. Explorations in cognitive dissonance. Academic Press; San Diego, CA: 1962. [Google Scholar]
- Brock TC. Commitment to exposure as a determinant of information receptivity. Journal of Personality and Social Psychology. 1965;2:10–19. doi: 10.1037/h0022082. [DOI] [PubMed] [Google Scholar]
- Brock TC, Albert SM, Becker LA. Familiarity, utility and supportiveness as determinants of information receptivity. Journal of Personality and Social Psychology. 1970;4:292–301. [Google Scholar]
- Brock TC, Balloun JL. Behavioral receptivity to dissonant information. Journal of Personality and Social Psychology. 1967;6:413–428. doi: 10.1037/h0021225. [DOI] [PubMed] [Google Scholar]
- Brodbeck M. The role of small groups in mediating the effects of propaganda. Journal of Abnormal and Social Psychology. 1956;52:166–170. doi: 10.1037/h0042654. [DOI] [PubMed] [Google Scholar]
- Bruner JS. On perceptual readiness. Psychological Review. 1957;64:123–154. doi: 10.1037/h0043805. [DOI] [PubMed] [Google Scholar]
- Byrne D. Repression-sensitization. In: Maher BA, editor. Progress in experimental personality research. Academic Press; New York: 1964. pp. 169–220. [PubMed] [Google Scholar]
- Carpentier FD, Knobloch S, Zillman D. Rock, rap, and rebellion: comparisons of traits predicting selective exposure to defiant music. Personality and Individual Differences. 2003;35:1643–1655. [Google Scholar]
- Canon LK. Self-confidence and selective exposure to information. In: Festinger L, editor. Conflict, decision, and dissonance. Stanford University Press; Stanford, CA: 1964. pp. 83–96. [Google Scholar]
- Canon LK, Matthews KE. Concern over personal health and smoking-relevant beliefs and behavior. Proceedings of the Annual Convention of the American Psychological Association. 1972;7:271–272. [Google Scholar]
- Chaiken S, Liberman A, Eagly AH. Heuristic and systematic information processing within and beyond the persuasion context. In: Uleman JS, Bargh JA, editors. Unintended thought. Guilford Press; New York: 1989. pp. 212–252. [Google Scholar]
- Chaiken S, Wood W, Eagly AH. Principles of persuasion. In: Higgins ET, Kruglanski AW, editors. Social psychology: Handbook of basic principles. Guilford Press; New York: 1996. pp. 361–399. [Google Scholar]
- Chen M, Bargh JA. Consequences of automatic evaluation: Immediate behavioral predispositions to approach or avoid the stimulus. Personality and Social Psychology Bulletin. 1999;25:215–224. [Google Scholar]
- Clarke P, James J. The effects situation, attitude intensity and personality on information-seeking. Sociometry. 1967;30:235–245. [PubMed] [Google Scholar]
- Cotton JL. Cognitive dissonance in selective exposure. In: Zillmann D, Bryant J, editors. Selective exposure to communication. Erlbaum; Hillsdale, NJ: 1985. pp. 11–33. [Google Scholar]
- Cotton JL, Hieser RA. Selective Exposure to information and cognitive dissonance. Journal of Research in Personality. 1980;14:518–527. [Google Scholar]
- Darley JM, Gross PH. A hypothesis-confirming bias in labeling effects. Journal of Personality and Social Psychology. 1983;44:20–33. [Google Scholar]
- Dewey J. Logic: The theory of inquiry. Holt; New York: 1938. [Google Scholar]
- Dillman CF, Knoblach S, Zillman D. Rock, rap, and rebellion: Comparisons of traits predicting selective exposure to defiant music. Personality and Individual Differences. 2003;35:1643–1655. [Google Scholar]
- Donohew L, Parker JM, McDermott V. Psychophysiological measurement of information selection: Two studies. Journal of Communication. 1972;22:54–63. doi: 10.1111/j.1460-2466.1972.tb00131.x. [DOI] [PubMed] [Google Scholar]
- Duncan BL. Differential social perception and attribution of intergroup violence: Testing the lower limits of stereotyping of blacks. Journal of Personality and Social Psychology. 1976;34:590–598. doi: 10.1037//0022-3514.34.4.590. [DOI] [PubMed] [Google Scholar]
- Eagly AH, Chaiken S. The psychology of attitudes. Harcourt Brace Jovanovich; Fort Worth, TX: 1993. [Google Scholar]
- Eagly AH, Chen S, Chaiken S, Shaw-Barnes K. The impact of attitudes on memory: An affair to remember. Psychological Bulletin. 1999;125:64–89. doi: 10.1037/0033-2909.125.1.64. [DOI] [PubMed] [Google Scholar]
- Edeani DO. Field study of selective exposure and selective recall as functions of locus of control and information utility. Dissertation Abstracts International. 1979;40(6-A):2960. [Google Scholar]
- Ehrlich D, Guttman I, Schonbach P, Mills J. Postdecision exposure to relevant information. Journal of Abnormal and Social Psychology. 1957;54:98–102. doi: 10.1037/h0042740. [DOI] [PubMed] [Google Scholar]
- Erber R. Affective and semantic priming: Effect of mood on category accessibility and inference. Journal of Experimental Social Psychology. 1991;27:480–498. [Google Scholar]
- Feather NT. Cigarette smoking and lung cancer: A study of cognitive dissonance. Australian Journal of Psychology. 1962;14:55–64. [Google Scholar]
- Feather NT. Cognitive dissonance, sensitivity, and evaluation. Journal of Abnormal and Social Psychology. 1963;66:157–163. doi: 10.1037/h0049383. [DOI] [PubMed] [Google Scholar]
- Feather NT. Preference for information in relation to consistency, novelty, intolerance of ambiguity, and dogmatism. Australian Journal of Psychology. 1969;21:235–249. [Google Scholar]
- Festinger L. A theory of cognitive dissonance. Stanford University Press; Stanford, CA: 1957. [Google Scholar]
- Festinger L. Conflict, decision, and dissonance. Stanford University Press; Stanford, CA: 1964. [Google Scholar]
- Fischer P, Jonas E, Frey D, Schulz-Hardt S. Selective exposure to information: The impact of information limits. European Journal of Social Psychology. 2005;35:469–492. [Google Scholar]
- Fischer P, Schulz-Hardt S, Frey D. Selective exposure and information quantity: How different information quantities moderate decision makers' preferences for consistent and inconsistent information. Journal of Personality and Social Psychology. 2008;94:231–244. doi: 10.1037/0022-3514.94.2.94.2.231. [DOI] [PubMed] [Google Scholar]
- Fiske ST, Taylor SE. Social cognition. McGraw Hill; New York: 1991. [Google Scholar]
- Forgas JP, Moylan SJ. After the movies: the effects of transient mood states on social judgments. Personality and Social Psychology Bulletin. 1987;13:478–489. [Google Scholar]
- Freedman JL. Preference for dissonant information. Journal of Personality and Social Psychology. 1965a;2:287–289. doi: 10.1037/h0022415. [DOI] [PubMed] [Google Scholar]
- Freedman JL. Confidence, utility, and selective exposure: A partial replication. Journal of Personality and Social Psychology. 1965b;2:778–780. doi: 10.1037/h0022670. [DOI] [PubMed] [Google Scholar]
- Freedman JL, Sears DO. Selective Exposure. In: Berkowitz L, editor. Advances in Experimental Social Psychology. Academic Press; New York: 1965. pp. 58–98. [Google Scholar]
- Frey D. The effect of negative feedback about oneself and cost of information on preferences for information about the source of this feedback. Journal of Experimental Social Psychology. 1981a;17:42–50. [Google Scholar]
- Frey D. Postdecisional preference for decision-relevant information as a function of the competence of its source and the degree of familiarity with it the information. Journal of Experimental Social Psychology. 1981b;17:621–626. [Google Scholar]
- Frey D. Reversible and irreversible decisions: Preference for consonant information as a function of attractiveness of decision alternatives. Personality and Social Psychology Bulletin. 1981c;7:621–626. [Google Scholar]
- Frey D. Different levels of cognitive dissonance, information seeking and information avoidance. Journal of Personality and Social Psychology. 1982;43:1175–1183. [Google Scholar]
- Frey D. Recent research on selective exposure to information. In: Berkowitz L, editor. Advances in Experimental Social Psychology. Academic Press; New York: 1986. pp. 41–80. [Google Scholar]
- Frey D, Rosch M. Information seeking after decisions: The roles of novelty of information and decision reversibility. Personality and Social Psychology Bulletin. 1984;10:91–98. [Google Scholar]
- Frey D, Stahlberg D. Selection of information after receiving more or less reliable self-threatening information. Personality and Social Psychology Bulletin. 1986;12:434–441. [Google Scholar]
- Frey D, Stahlberg D, Fries A. Information seeking of high- and low-anxiety subjects after receiving positive and negative self-relevant feedback. Journal of Personality. 1986;54:694–703. doi: 10.1111/j.1467-6494.1986.tb00420.x. [DOI] [PubMed] [Google Scholar]
- Frey D, Wicklund R. A clarification of selective exposure: The impact of choice. Journal of Experimental Social Psychology. 1978;14:132–139. [Google Scholar]
- Greenwald AG, McGhee DE, Schwartz JLK. Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology. 1998;74:1464–1480. doi: 10.1037//0022-3514.74.6.1464. [DOI] [PubMed] [Google Scholar]
- Greenwald AG, Ronis DL. Twenty years of cognitive dissonance: Case study of the evolution of a theory. Psychological Review. 1978;85:53–57. [PubMed] [Google Scholar]
- Haddock CK, Rindskopf D, Shadish WR. Using odds ratios as effect sizes for meta-analysis of dichotomous data: A primer on methods and issues. Psychological Methods. 1998;3:339–353. [Google Scholar]
- Hammen CL. Effects of depression, feedback, and gender on selective exposure to information about the self. Psychological Reports. 1977;40:403–408. doi: 10.2466/pr0.1977.40.2.403. [DOI] [PubMed] [Google Scholar]
- Harkness AR, DeBono KG, Borgida E. Personal involvement and strategies for making contingency judgments: A stake in the dating game makes a difference. Journal of Personality and Social Psychology. 1985;49:22–32. [Google Scholar]
- Harmon-Jones E. Cognitive dissonance and experienced negative affect: Evidence that dissonance increases experienced negative affect even in the absence of aversive consequences. Personality and Social Psychology Bulletin. 2000;26:1490–1501. [Google Scholar]
- Harmon-Jones E, Brehm JW, Greenberg J, Simon L, Nelson DE. Evidence that the production of aversive consequences is not necessary to create cognitive dissonance. Journal of Personality and Social Psychology Bulletin. 1996;70:5–16. [Google Scholar]
- Harmon-Jones E, Harmon-Jones C. Cognitive dissonance theory: An update with a focus on the action-based model. In: Shah JY, Gardner WL, editors. Handbook of motivation science. The Guilford Press; New York: 2008. pp. 71–83. [Google Scholar]
- Hasselblad V, Hedges LV. Meta-analysis of screening and diagnostic tests. Psychological Bulletin. 1995;17:167–178. doi: 10.1037/0033-2909.117.1.167. [DOI] [PubMed] [Google Scholar]
- Hastorf AH, Cantril H. They saw a game: A case study. Journal of Abnormal and Social Psychology. 1954;49:129–134. doi: 10.1037/h0057880. [DOI] [PubMed] [Google Scholar]
- Hedges LV, Olkin I. Statistical methods for meta-analysis. Academic Press; Orlando, FL: 1985. [Google Scholar]
- Higgins ET. Achieving “shared reality” in the communication game: A social action that creates meaning. Journal of Language & Social Psychology. 1992;11:107–131. [Google Scholar]
- Hillis JW, Crano WD. Additive effects of utility and attitude supportiveness in the selection of information. The Journal of Social Psychology. 1973;89:257–269. [Google Scholar]
- Holton B, Pyszczynski T. Biased information search in the interpersonal domain. Personality and Social Psychology Bulletin. 1989;14:42–51. [Google Scholar]
- James W. Principles of psychology. Holt; New York: 1890. [Google Scholar]
- Janis IL, Rausch CN. Selective interest in communications that could arouse decisional conflict: A field study of participants in the draft. Journal of Personality and Social Psychology. 1970;14:46–54. doi: 10.1037/h0028624. [DOI] [PubMed] [Google Scholar]
- Jecker JD. Selective exposure to new information. In: Festinger L, editor. Conflict, decision, and dissonance. Stanford University Press; Stanford, California: 1964. pp. 65–81. [Google Scholar]
- Johnson BT. Effects of outcome-relevant involvement and prior information on persuasion. Journal of Experimental Social Psychology. 1994;30:556–579. [Google Scholar]
- Johnson BT, Eagly AH. The effects of involvement on persuasion: A meta-analysis. Psychological Bulletin. 1989;106:375–384. [Google Scholar]
- Johnson E, Tversky A. Affect, generalization, and the perception of risk. Journal of Personality and Social Psychology. 1983;45:20–31. [Google Scholar]
- Johnston LC. Resisting change: Information seeking and stereotype change. European Journal of Social Psychology. 1996;26:799–825. [Google Scholar]
- Jonas E, Frey D. Searching for information about financial decisions in Euro versus DM. European Psychologist. 2003a;8:92–96. [Google Scholar]
- Jonas E, Frey D. Information search and presentation in advisor-client interactions. Organizational Behavior and Human Decision Processes. 2003b;91:154–168. [Google Scholar]
- Jonas E, Frey D, Henninger M, Pommer M, Haeften I. von, Schulz-Hardt S, Mandl H. Rechtfertigungsdruck und Expertenurteile als Einflussfaktoren auf Informationssuche und Informationsbewertung in einer sequentiellen Lernsituation. Zeitschrift für Entwicklungspsychologie und Pädagogische Psychologie. 2001;33:242–252. [Google Scholar]
- Jonas E, Graupmann V, Fischer P. Schwarze Kassen, weibe westen? Konfirmatorische informationssuche und-bewertung im knotext der parteispend. Zeirschrift Fur Sozialpsychologie. 2003;34:47–61. [Google Scholar]
- Jonas E, Graupmann V, Frey D. The influence of mood on the search for supporting vs. conflicting information: Dissonance reduction as a means of mood regulation? Personality and Social Psychology Bulletin. 2006;32:3–15. doi: 10.1177/0146167205276118. [DOI] [PubMed] [Google Scholar]
- Jonas E, Greenberg J, Frey D. Connecting terror management and dissonance theory: evidence that mortality salience increases the preference for supporting information after decisions. Personality and Social Psychology Bulletin. 2003;29:1181–1189. doi: 10.1177/0146167203254599. [DOI] [PubMed] [Google Scholar]
- Jonas E, Schulz-Hardt S, Frey D. Konfirmatorische informationssuche bei simultaner vs. sequentieller informationsvorgabe. Seitschrift fur Experimentelle Psychologie. 2001;48:239–247. [PubMed] [Google Scholar]
- Jonas E, Schulz-Hardt S, Frey D. Giving advice or making decisions in someone else's place: The influence of impression, defense and accuracy motivation on the search for new information. Personality and Social Psychology Bulletin. 2005;31:977–990. doi: 10.1177/0146167204274095. [DOI] [PubMed] [Google Scholar]
- Jonas E, Schulz-Hardt S, Frey D, Thelen N. Confirmation bias in sequential information search after preliminary decisions: An expansion of dissonance theoretical research on selective exposure to information. Journal of Personality and Social Psychology. 2001;80:557–571. doi: 10.1037//0022-3514.80.4.557. [DOI] [PubMed] [Google Scholar]
- Katz D. The functional approach to the study of attitudes. Public Opinion Quarterly. 1960;24:163–204. [Google Scholar]
- Kiesler CA. The psychology of commitment. Academic Press; New York: 1971. [Google Scholar]
- Keith TZ. Multiple regression and beyond. Allyn & Bacon; Boston: 2006. [Google Scholar]
- Klayman J, Ha YW. Confirmation, disconfirmation, and information in hypothesis testing. Psychological Review. 1987;94:211–228. [Google Scholar]
- Kleck RE, Wheaton J. Dogmatism and responses to opinion-consistent and opinion-inconsistent information. Journal of Personality and Social Psychology. 1967;5:249–253. doi: 10.1037/h0024197. [DOI] [PubMed] [Google Scholar]
- Knobloch-Westerwick S, Hastall M. Social comparisons with news personae. Communication Research. 2006;33:262–284. [Google Scholar]
- Kruglanski AW. Motivations for judging and knowing: Implications for causal attribution. In: Higgins ET, Sorrentino RM, editors. Handbook of motivation and cognition. Guilford; New York: 1990. pp. 333–367. [Google Scholar]
- Kunda Z. The case for motivated inference. Psychological Bulletin. 1990;108:480–498. doi: 10.1037/0033-2909.108.3.480. [DOI] [PubMed] [Google Scholar]
- Lavine H, Lodge M, Freitas K. Threat, authoritarianism and selective exposure to information. Political Psychology. 2005;26:219–244. [Google Scholar]
- LaVoie AL, Thompson SK. Selective exposure in a field setting. Psychological Reports. 1972;31:433–434. [Google Scholar]
- Levine JM, Murphy G. The learning and forgetting of controversial material. Journal of Abnormal and Social Psychology. 1943;38:507–517. [Google Scholar]
- Lewin K. Problems of research in social psychology. In: Cartwright D, editor. Field theory in social science. Harper & Row; New York: 1951. pp. 155–169. [Google Scholar]
- Lieberman MD, Ochsner KN, Gilbert DT, Schacter DL. Do amnesics exhibit cognitive dissonance reduction? The role of explicit memory and attention in attitude change. Psychological Science. 2001;12:135–140. doi: 10.1111/1467-9280.00323. [DOI] [PubMed] [Google Scholar]
- Lipsey MW, Wilson DB. Practical Meta-Analysis. Sage Publications, Inc; Thousand Oaks, CA: 2001. [Google Scholar]
- Lord CG, Ross L, Lepper MR. Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology. 1979;37:2098–2109. [Google Scholar]
- Lowe RH, Steiner ID. Some effects of the reversibility and consequences of decisions on postdecision information preferences. Journal of Personality and Social Psychology. 1968;88:172–179. doi: 10.1037/h0025301. [DOI] [PubMed] [Google Scholar]
- Lowin A. Further evidence for an approach-avoidance interpretation of selective exposure. Journal of Experimental Psychology. 1969;5:265–271. [Google Scholar]
- Lundgren SR, Prislin R. Motivated cognitive processing and attitude change. Personality and Social Psychology Bulletin. 1998;24:715–726. [Google Scholar]
- Maccoby EE, Maccoby N, Romney AK, Adams JS. Social reinforcement in attitude change. Journal of Abnormal and Social Psychology. 1961;63:109–115. doi: 10.1037/h0043154. [DOI] [PubMed] [Google Scholar]
- Maslow AH. Toward a psychology of being. Van Nostrand; New York: 1968. [Google Scholar]
- McAllister DW, Mitchell TR, Beach LR. The contingency model for the selection of decision strategies: An empirical test of the effects of significance, accountability, and reversibility. Organizational Behavior and Human Performance. 1979;24:228–244. [Google Scholar]
- McFarland SG, Warren JC. Religious orientations and selective exposure among fundamentalist Christians. Journal for the Scientific Study of Religion. 1992;31:163–174. [Google Scholar]
- Meadowcroft J, Zillmann D. Women's comedy preferences during the menstrual cycle. Communication Research. 1987;14:204–218. [Google Scholar]
- Micucci JA. Unpublished master's thesis. Cornell University; Ithaca, New York, USA: 1972. Self-esteem and preference for consonant information. [Google Scholar]
- Miller RL. The effects of postdecisional regret on selective exposure. European Journal of Social Psychology. 1977;7:121–127. [Google Scholar]
- Molden DC, Higgins ET. Motivated thinking. In: Holyoak KJ, Morrison RG, editors. The cambridge handbook of thinking and reasoning. Cambridge University Press; New York: 2005. pp. 295–320. [Google Scholar]
- Morris SB. Distribution of the standardized mean change effect size for meta-analysis on repeated measures. British Journal of Mathematical & Statistical Psychology. 2000;53:17–29. doi: 10.1348/000711000159150. [DOI] [PubMed] [Google Scholar]
- Nemeth C, Rogers J. Dissent and the search for information. British Journal of Social Psychology. 1996;35:67–76. [Google Scholar]
- Noguchi K, Albarracín D, Durantini MR, Glasman LR. Who participates in which health promotion programs? A meta-analysis of motivations underlying enrollment and retention in HIV-prevention interventions. Psychological Bulletin. 2007;133:955–975. doi: 10.1037/0033-2909.133.6.955. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Olson JM, Stone J. The influence of behavior on attitudes. In: Albarracín D, Johnson BT, Zanna MP, editors. The handbook of attitudes. Lawrence Erlbaum Associates; New Jersey: 2005. [Google Scholar]
- Olson JM, Zanna MP. A new look at selective exposure. Journal of Experimental Social Psychology. 1979;15:1–15. [Google Scholar]
- Otis L. Selective exposure to the film close encounters. Journal of Psychology: Interdisciplinary and Applied. 1979;101:293–295. [Google Scholar]
- Petty RE, Wegener DT. Attitude change: Multiple roles for persuasion variables. In: Gilbert D, Fiske S, Lindzey G, editors. The handbook of social psychology. McGraw-Hill; New York: 1998. pp. 323–390. [Google Scholar]
- Pierce C. Popular Science Monthly. 1877. The fixation of belief. [Google Scholar]
- Potts R, Dedmon A, Halford J. Sensation seeking, television viewing motives, and home television viewing patterns. Personality and Individual Differences. 1996;21:1081–1084. [Google Scholar]
- Potts R, Sanchez D. Television viewing and depression: No news is good news. Journal of Broadcasting & Electronic Media. 1994;38:79–90. [Google Scholar]
- Prislin R, Wood W. Social influence in attitudes and attitude change. In: Albarracín D, Johnson BT, Zanna MP, editors. The handbook of attitudes. Lawrence Erlbaum Associates; New Jersey: 2005. pp. 671–706. [Google Scholar]
- Pyszczynski TA, Greenberg J, Laprelle J. Biased search for social comparison information after success and failure. Journal of Experimental Psychology. 1985;21:195–211. [Google Scholar]
- Pyszczynski T, Greenberg J, Solomon S, Arndt J, Schimel J. Why do people need self-esteem? A theoretical and empirical review. Psychological Bulletin. 2004;130:435–468. doi: 10.1037/0033-2909.130.3.435. [DOI] [PubMed] [Google Scholar]
- Raghunathan R, Pham MT. All negative moods are not equal: Motivational influences of anxiety and sadness on decision making. Organizational Behavior and Human Decision Processes. 1999;79:56–77. doi: 10.1006/obhd.1999.2838. [DOI] [PubMed] [Google Scholar]
- Raudenbush SW, Bryk AS. Hierarchical linear models: Applications and data analysis methods. Sage Publications; 2002. [Google Scholar]
- Rhine RJ. The 1964 presidential election and curves of information seeking and avoiding. Journal of Personality and Social Psychology. 1967;5:416–423. doi: 10.1037/h0021211. [DOI] [PubMed] [Google Scholar]
- Robinson RJ, Keltner D, Ward A, Ross L. Actual versus assumed differences in construal: “Naïve realism” in intergroup perception and conflict. Journal of Personality and Social Psychology. 1995;68:404–417. [Google Scholar]
- Rokeach M. The open and closed mind. Basic Books; New York: 1960. [Google Scholar]
- Rosen S. Postdecision affinity for incompatible information. Journal of Abnormal and Social Psychology. 1961;63:188–190. doi: 10.1037/h0046311. [DOI] [PubMed] [Google Scholar]
- Rosenbaum LL, McGinnies E. Selective Exposure: An addendum. The Journal of Psychology. 1973;83:329–331. doi: 10.1080/00223980.1973.9915621. [DOI] [PubMed] [Google Scholar]
- Salovey P, Birnbaum D. Influence of mood on health-relevant cognitions. Journal of Personality and Social Psychology. 1989;57:539–551. doi: 10.1037//0022-3514.57.3.539. [DOI] [PubMed] [Google Scholar]
- Schlenker BR. Impression management: The self-concept, social identity, and interpersonal relations. Brooks/Cole; Monterey, CA: 1980. [Google Scholar]
- Schlenker BR. Identity and self-identification. In: Schlenker BR, editor. The self and social life. McGraw-Hill; New York: 1985. pp. 65–99. [Google Scholar]
- Schlenker BR. Self-presentation. In: Leary MR, Tangney JP, editors. Handbook of self and identity. Guilford; New York: 2003. pp. 492–518. [Google Scholar]
- Schulman GI. Who will listen to the other side: Primary and secondary group support and selective exposure. Social Problems. 1971;18:404–415. [Google Scholar]
- Schulz-Hardt S, Frey D, Luthgens C, Moscovici S. Biased information search in group decision making. Journal of Personality and Social Psychology. 2000;78:655–669. doi: 10.1037//0022-3514.78.4.655. [DOI] [PubMed] [Google Scholar]
- Schwarz N, Frey D, Kumpf M. Interactive effects of writing and reading a persuasive essay on attitude change and selective exposure. Journal of Experimental Social Psychology. 1980;16:1–17. [Google Scholar]
- Sears DO. Biased indoctrination and selectivity of exposure to new information. Sociometry. 1965;28:363–376. [Google Scholar]
- Sears DO. Opinion formation and information preferences in an adversary situation. Journal of Experimental Social Psychology. 1966;2:130–142. [Google Scholar]
- Sears DO. College sophomores in the laboratory: Influences of a narrow data base on social psychology's view of human nature. Journal of Personality and Social Psychology. 1986;51:515–530. [Google Scholar]
- Sears DO, Freedman JL. Commitment, Information Utility and Selective Exposure. Aug, 1963. USN technical report (ONR) nonr-233(54) NR 171–350, No. 12. [Google Scholar]
- Sears DO, Freedman JL. Effects of expected familiarity with arguments upon opinion change and selective exposure. Journal of Personality and Social Psychology. 1965;3:420–426. doi: 10.1037/h0022380. [DOI] [PubMed] [Google Scholar]
- Seignourel P, Albarracín D. Calculating effect sizes for designs with between-subjects and within-subjects factors: Methods for partially reported statistics in meta-analysis. Methodology of Behavioral Sciences. 2002;4:273–289. [Google Scholar]
- Shiffrin RM, Schneider W. Controlled and automatic human information processing: II. Perceptual learning, automatic attending, and a general theory. Psychological Review. 1977;84:127–190. [Google Scholar]
- Smith MB, Bruner JS, White RW. Opinions and personality. Wiley; Oxford, England: 1956. [Google Scholar]
- Smith SM, Fabrigar LR, Powell DM, Estrada M. The role of information processing capacity and goals in attitude-congruent selective exposure effects. Personality and Social Psychology Bulletin. 2007;33:948–960. doi: 10.1177/0146167207301012. [DOI] [PubMed] [Google Scholar]
- Steele CM. The psychology of self-affirmation: Sustaining the integrity of the self. In: Berkowitz L, editor. Advances in experimental social psychology. Vol. 21. Academic Press; New York: 1988. pp. 261–302. [Google Scholar]
- Sweeney PD, Gruber KL. Selective exposure: Voter information preferences and the Watergate affair. Journal of Personality and Social Psychology. 1984;46:1208–1221. [Google Scholar]
- Tajfel H. Cognitive aspects of prejudice. Journal of Social Issues. 1969;25:79–97. doi: 10.1017/s0021932000023336. [DOI] [PubMed] [Google Scholar]
- Tetlock PE. Accountability and the perseverance of first impressions. Social Psychology Quarterly. 1983;46:285–292. [Google Scholar]
- Tetlock PE, Kim J. Accountability and judgment in a personality prediction task. Journal of Personality and Social Psychology. 1987;52:700–709. doi: 10.1037//0022-3514.52.4.700. [DOI] [PubMed] [Google Scholar]
- Thayer S. Confidence and postjudgement exposure to consonant and dissonant information in a free choice situation. The Journal of Social Psychology. 1969;77:113–120. [Google Scholar]
- The Pew Research Center. 2004 Retrieved April 4, 2008, from http://people-press.org/reports/display.php3?PageID=1067.
- The Smoking Gun. 2006 Mar 23; Retrieved July 2, 2008, from http://www.thesmokinggun.com/archive/0322061cheney1.html.
- Wang MC, Bushman BJ. Integrating results through meta-analytic review using SAS software. SAS Institute Inc; Cary, NC: 1999. [Google Scholar]
- Wellins R, McGinnies E. Counterarguing and selective exposure to persuasion. Journal of Social Psychology. 1977;103:115–127. doi: 10.1080/00224545.1978.9924175. [DOI] [PubMed] [Google Scholar]
- Wicklund RA, Brehm JW. Perspectives on cognitive dissonance. Lawrence Erlbaum Associates; Hillsdale, NJ: 1976. [Google Scholar]
- Wiles JR. The missing link: Scientist discovers that evolution is missing from Arkansas classrooms. 2006 Mar 23; Retrieved July 2, 2008, from http://www.arktimes.com/Articles/ArticleViewer.aspx?ArticleID=e7a0f0e1-ecfd-4fc8-bca4-b9997c912a91.
- Winkielman P, Cacioppo JT. Mind at ease puts a smile on the face: Psychophysiological evidence that processing facilitation elicits positive affect. Journal of Personality and Social Psychology. 2001;81:989–1000. [PubMed] [Google Scholar]
- Wolf IM. Methodological observations on bias. In: Wachter KW, Straf ML, editors. The future of meta-analysis. Russel Sage Foundation; New York: 1990. pp. 139–151. [Google Scholar]
- Wyer RS, Albarracín D. The origins and structure of beliefs and goals. In: Albarracín D, Johnson BT, Zanna MP, editors. Handbook of attitudes. Lawrence Erlbaum; Hillsdale, New Jersey: 2005. pp. 273–322. [Google Scholar]
- Zanna MP, Rempel JK. Attitudes: A new look at an old concept. In: Bar-Tal D, Kruglanski AW, editors. The social psychology of knowledge. Cambridge University Press; Cambridge, UK: 1988. pp. 315–334. [Google Scholar]
- Zeigarnik B. The retention of completed and uncompleted actions. Psychologische Forschung. 1927;9:1–85. [Google Scholar]