The SAGE Handbook of Persuasion

by James Price Dillard

  We argue in this chapter that attitudes do all that, and that the interplay of attitudes and behavior gives fascinating insight into how people make sense of, and behave sensibly in, this chaotic world.

  Behavior as an Outcome of Persuasion

  * * *

  Behavior as an outcome of persuasion is the gold standard because it is often the main point of a persuasive attempt. However, it can be very difficult to measure actual behavior, because researchers are not always present when the behavior of interest is performed (e.g., voting behavior). Even when it is possible to measure actual behavior, it can be very costly in terms of time and financial resources. Additionally, it is sometimes unethical to measure actual behavior (e.g., drug use, underage drinking, or smoking).

  Rather than examining actual behavior, researchers rely on participants’ self-reports of behavior in the vast majority of studies in persuasion. However, relying on self-reports of behavior is problematic because we can never be sure that participants’ reports are accurate. Participants’ memories could be faulty, they may want to portray themselves in a positive light, and the nature of the persuasion study may telegraph the “correct” answer. For all of these reasons, self-reports of behavior may not be reliable and research should be conducted validating the self-report measures that are used. The difficulties in observing actual behavior and in measuring reported behavior reliably has had a number of consequences. Researchers have begun to rely on a number of proxies, or stand-ins, for actual behavior measurement. Some of these are described in the following.

  Proxies for Behavior

  Aggregated Behavior

  An intriguing line of research by Fishbein and Ajzen (1974) focused on the specificity with which we measure both attitudes and behavior (see also Ajzen & Fishbein, 2005). Their contention was that attitudes are measured in very general terms in traditional work on attitudes and persuasion. For example, if the study is concerned with a government policy, the policy is explained and then participants indicate whether they are in favor of the policy or not. In political persuasion, attitudes toward candidates are measured by participants indicating how favorably they feel toward the candidate. In marketing research, the participants’ overall judgment of a product may be measured. Yet, the behavioral measure tends to be something very specific, such as whether the participant voted for the candidate in the election or whether the respondent purchased the product in question at his or her last trip to the store.

  Ajzen and Fishbein noted that when people make aggregated judgments of an attitude object, they are taking into account the range of beliefs they have about that object in different contexts. Confining behavioral measures to a single place and time disadvantages the attempt to correlate attitudes and behaviors because there may be vagaries in the specific situation that affect the behavior. However, by examining the pattern of behavior over time, one would obtain a more reliable measure of the individual’s propensity to engage in the behavior. For example, in the case of whether attitudes direct TV watching behavior, a person with a favorable attitude toward one of the late-night comic hosts might or might not have watched the show in the previous evening for a number of reasons—they may have been out with friends, they may have had to catch up on work, they may have fallen asleep early. However, by measuring viewing behavior over, say, a one-month period, it is likely that those having a positive attitude toward the program will tune in more frequently than those having a negative attitude.

  Behavioral Intention and Behavioral Willingness

  Another way in which Fishbein and Ajzen (1975) dealt with the lack of predictive validity of attitudes was to espouse the use of measures of behavioral intention (see also Ajzen & Fishbein, 2005). In their model, behavioral intention is a strong predictor of actual behavior. The key for Fishbein and Ajzen is that the measure of behavioral intention must be anchored to a specific time and place. Thus, asking someone if they intend to watch Conan O’Brian on a specific night will give a closer approximation to the actual behavior that night than if they were asked if they intend to watch Conan O’Brian in some vaguely defined future.

  Measures of behavioral intention are prevalent in health communication research. In part, this is because it is unethical to present opportunities for research participants to engage in health risk behaviors in the lab. Thus, studies in which anti-smoking messages are presented will ask about specific intentions to smoke in the future instead of offering participants a cigarette to determine if the anti-smoking message was successful. For example, participants might be asked questions such as “Do you think you will smoke a cigarette sometime today?” or “Do you think you will smoke a cigarette sometime in the coming week?” Measures such as these have been used as proxies for behavior with good success. In studies where follow-up surveys have been conducted, the ability of the behavior intention measure to predict actual behavior is reasonably strong, for example, resulting in an average correlation of .5 in one meta-analysis (Albarracín, Johnson, Fishbein, & Mullerleile, 2001). To be sure, behavioral intention falls short of being a perfect measure of behavior, but it does provide an index of propensity to behave.

  One of the criticisms of measuring behavioral intentions is that it assumes that a person’s behavior is the result of deliberative decision-making (Fazio, 1990; Roskos-Ewoldsen, & Fazio, 1997). However, there are many varieties of behavior and it is important to understand how different behaviors influence the relationship between attitudes and behavior (Jaccard & Blanton, 2005). Certainly, people often deliberate carefully about a decision before forming an intention to engage in some behavior. For example, people may carefully deliberate about important decisions, such as buying a car or deciding what college to attend. However, much of people’s everyday social behavior is more spontaneous in nature (Fazio, 1990; Gerrard, Gibbons, Vande Lune, Pexa, & Gano, 2002; Gerrard, Gibbons, Houlihan, Stock, & Pomery, 2008). For example, people rarely carefully consider all of the available information about all the different choices in a vending machine before making a selection.

  Fazio’s Motivation and Opportunities as DEterminants (MODE) model attempts to capture this difference between deliberative and more spontaneous behavior (Fazio, 1990; Olson & Fazio, 2009). According to the model, there are two factors that can influence whether a judgment or behavior is spontaneous or deliberative in nature: motivation and opportunity. Motivation can influence the judgment when individuals have a reason to more carefully scrutinize the attitude object. Some motivations that might lead to more careful scrutiny of an attitude object include desire to make accurate judgments, the need to belong, and motivation to avoid appearing prejudiced (Olson & Fazio, 2009). Opportunity refers to an individual’s actual ability to make a deliberative judgment at a given time. Factors such as time and available cognitive resources can influence a person’s opportunity to make deliberative judgments (Olson & Fazio, 2009). According to the MODE model, when people are highly motivated and they have sufficient opportunity to carefully consider the available information, they are more likely to make decisions in a deliberative fashion. However, when motivation or opportunity is low, people are more likely to engage in a spontaneous manner.

  Spontaneous decisions are characterized by the consideration of information that is accessible, that is, quickly activated from memory (Fazio, 1990; Gerrard et al., 2008; Roskos-Ewoldsen, 1997; Olson & Fazio, 2009). Consider the adolescent who, in surveys, reports that peer pressure has no influence on his or her behavior. Yet when asked by a best friend at a party if he or she wants to try a cigarette, this same adolescent takes a few puffs. Behavioral willingness refers to a person’s willingness to engage in risky behavior in circumstances that promote that behavior such as the friend asking if you want a cigarette. Consequently, behavioral willingness is an outcome measure that represents a teen’s likelihood of spontaneously engaging in smoking behavior if an opportunity presents itself (Gerrard et al., 2002; Gerrard et al., 2008; Rhodes & Ewoldsen, 2009) and is an indicator of spont
aneous decision making.

  Self-Reports of Behavior

  Although we said earlier that it is difficult to trust respondents’ reports of their own behavior, self-reports are often used as outcome variables in persuasion research. The use of such self-reports is subject to criticism because of the issues previously described, but it is also true that sometimes respondents’ own reports are the best information it is possible to obtain. To address this problem, work in various behavioral domains has focused on constructing reliable indices for behavior. For example, cigarette smoking behavior is frequently measured with a set of items developed for the behavioral risk factor surveillance system, which has been shown to have good reliability and validity.

  To some extent, the necessity for accurate measures of actual behavior depends on the goals of the research being conducted. From the perspective of researchers interested in how individuals process and react to persuasive appeals, showing that message characteristics have predicted effects on these types of indices is usually sufficient for the purposes of testing their hypotheses. For researchers in the field, however, more precise measures are often required to evaluate the effectiveness of a health-related intervention.

  Attitudes as Outcomes

  * * *

  One approach to the problem of measuring behavior reliably is to ignore the problems associated with measuring behavior and simply study the effects of persuasive messages on attitudes. Researchers in this stream have largely been content to leave the question of behavior to other scholars and focus on the processes through with attitude change occurs. Huge advances in theory about the effects of persuasive messages have occurred, including Petty and Cacioppo’s (1986) elaboration likelihood model and Chaiken’s heuristic-systematic model (Chaiken, Liberman, & Eagly, 1989), which have advanced thinking about how dual processes of decision making relate to persuasive attempts.

  A great deal of research along these lines has been conducted to test predictions of dual process models. Consistent with these models, studies such as these have generally found that strong arguments are persuasive when participants are motivated and able to carefully think about the argument claims. Attitude formation or change that is based on a careful consideration of the arguments tends to be more durable and more predictive of behavior than attitudes that are formed or changed on the basis of persuasive elements that are peripheral to the message claims, such as the attractiveness of the spokesperson (Andrews & Shimp, 1990). There are numerous sources that address these approaches to the study of attitudes and persuasion (Albarracín, Johnson, & Zanna, 2005; Crano & Prislin, 2008), including other chapters in this volume. We focus the bulk of our review of the recent work on accessible and implicit attitudes because this is a burgeoning area of research with important theoretical and methodological contributions.

  Accessible or Implicit Attitudes

  Recent work has focused on spontaneously or implicitly activated attitudes. As discussed earlier, work on accessible and implicit attitudes is an outgrowth of the cognitive revolution in social science research and particularly in social psychology that began in the 1970s. This perspective is consistent with dual process approaches, and focuses directly on the role of spontaneous or automatic processing of attitudinally relevant constructs. Accessible attitudes are those attitudes that are readily available in memory. An example of an accessible attitude is one’s attitude toward a cockroach. Most people do not need to search their memory long to find their attitude toward this object when they see it crawling on their kitchen floor: the evaluation is retrieved extremely quickly, and appropriate action is initiated. What is important is that these implicit and accessible attitudes are activated automatically and without controlled thought, and thus they can affect behavior at times when motivation to process information carefully is low.

  Attitude Accessibility

  Attitude accessibility involves the ease with which an attitude is activated from memory (Fazio, 1986; Roskos-Ewoldsen, 1997). Within the attitude accessibility framework, attitudes are defined as associations between objects and evaluations of those objects stored in memory (Fazio, 1986, 1989). Based on the idea of memory as a semantic network, the mental representation of the attitude object is stored as a node in this network. Similarly, the evaluation of this object is also represented as a node in the network. To the extent that an evaluation is strongly associated with the object, the evaluation will be highly accessible: that is, when the node for the attitude object is activated, the strength of the association will ensure that the node containing the evaluation of the object is also activated. In this way, judgments can be made rapidly and without extensive reflection (Fazio, 1986).

  In contrast, for attitudes that are not accessible, the associations between the object and the evaluation of that object are not as strong, or perhaps the object has no evaluation associated with it. In this case, the activation of the object does not spontaneously activate the evaluation of the object, and, it may take more time to activate the judgment. Many attitudes lie somewhere between these two extremes. Most people, when asked, can report their evaluation of a wide range of attitude objects. For any given person, some of these attitudes will be strong and very quickly retrieved, others may be relatively weak and slowly accessed.

  Accessible attitudes are important predictors of various phenomenon of interest to communication scholars and psychologists (Dillard, 1993; Roskos-Ewoldsen, 1997). We continue Fazio’s (1989) practice of discussing the functions that accessible attitudes serve for the individual because we feel it is important to focus on what the attitude does for the individual. To use Allport’s words (1935; previously quoted), how does the accessible attitude influence how a person finds their way in an ambiguous universe?

  Initially, accessible attitudes influence what attracts our attention, literally determining what we notice in our environment. Roskos-Ewoldsen and Fazio (1992) found that research participants were more likely to orient their attention to objects in a complex visual field when they had more accessible attitudes toward those objects. Likewise, people are more likely to attend to brands within an advertisement when they already have an accessible attitude toward that brand (Goodall & Ewoldsen, 2011). Similarly, research participants who had more accessible attitudes toward particular consumer products were more likely to find those products on a shelf and choose them from a group of many other products toward which their attitudes were less accessible (Fazio, Powell, & Williams, 1989). The implications are clear in a world full of information and visual clutter. We cannot act positively (or negatively) toward objects we do not “see.” The orienting function of accessible attitudes will be the gateway to attitude-consistent behaviors in most cases.

  Second, accessible attitudes offer us an efficient, though not always desirable, means of processing myriad pieces of information we encounter each day by motivating us to attend to and carefully consider some messages and by facilitating avoidance or biased processing of other messages (Fazio, Roskos-Ewoldsen, & Powell, 1994; Fazio & Towles-Schwen, 1999; Roskos-Ewoldsen, 1997). Accessible attitudes often signal for us which topics are important, thereby encouraging us to attend to and elaborate on persuasive messages related to the important topic or delivered by a favored message source (Fabrigar, Priester, Petty, & Wegener, 1998; Roskos-Ewoldsen, Bichsel, & Hoffman, 2002). For example, Fabrigar et al. (1998) found that people with more accessible attitudes toward vegetarianism were more likely to centrally process a message about vegetarianism. Roskos-Ewoldsen et al. (2002) extended this finding by demonstrating that the activation of an attitude toward other components of the message such as the source of the message could also increase central processing. People with more accessible positive attitudes toward the source of the message are more likely to centrally process a message attributed to that source, and the attitude toward the source does not bias the processing of the message. Rather, the accessible attitude toward the source of the message acts as a piece of information indicating the importanc
e of the message, which motivates participants to more carefully process the message (see also Fabrigar et al., 1998; Roese & Olson, 1994).

  Accessible attitudes can also encourage us to cognitively reinforce our existing attitudes and behaviors by motivating us to process information in a biased manner. Early scholars recognized that attitudes can function as lenses through which we view and interpret our world (Allport, 1935; Katz, 1960; Smith, Bruner, & White, 1956). More recent research has shown that accessible attitudes often color our judgments of messages and attitude objects in a manner that is consistent with our attitudes (Fazio, 1990; Fazio et al., 1994). For example, research participants with more accessible, positive attitudes toward smoking were more likely to perceive antismoking PSAs as biased (Rhodes, Roskos-Ewoldsen, Edison, & Bradford, 2008; Shen, Monahan, Rhodes, & Roskos-Ewoldsen, 2009). An important caveat to these studies is that accessible attitudes are more likely to bias processing when the information or situation is relatively ambiguous (Roskos-Ewoldsen, 1997).

  Finally, perhaps the most important contribution of the conceptualization and study of attitude accessibility is the identification of consistently strong correlations between accessible attitudes and behavior (for general reviews, see Fazio, 1986, 1990; Fazio & Roskos-Ewoldsen, 2005) in the areas of health behaviors, voting behavior, consumer product choice, loyalty to retail stores, intended charitable contributions, exercise behaviors, choice of a game to play, and racist behaviors (Bassili, 1995; Kokkinaki, & Lunt, 1997; Posavac, Sanbonmatsu, & Fazio, 1997; Rhodes, Ewoldsen, Shen, Monahan, & Eno, 2011; Rhodes & Ewoldsen, 2009; Woodside & Trappey, 1996). It makes sense that attitudes are most likely to affect behavior when they are activated from memory at the moment the attitude object is initially observed.