Measuring Functions
Several scholars (Clary & Snyder, 1992; Clary, Snyder, Ridge, Copeland, Stukas, Haugen, & Miene, 1998; Clary, Snyder, Ridge, Miene, & Haugen, 1994) have tested the functional matching hypothesis using Herek’s (1987) method of measuring attitude functions by self-report. Clary and Snyder’s (1992) work provides an example. Their volunteer functions inventory (VFI) was designed to assess five different motivations for engaging in volunteerism based on the five previously described functions. They tested the functional matching effect with five different videotaped messages promoting volunteerism that were each designed to target one of the five functions. They used the subjects’ scores on the VFI to assign a message to them that either matched the function with which their attitude was most associated or the function with which their attitude was least associated. If the message matched the function identified by one’s highest score on the VFI, then it was more persuasive than if it matched their lowest score. Subsequently, they replicated this effect for brochures. Clary et al. (1994) conducted a similar study using the VFI to measure the primary function served by the subjects’ attitudes. They found that functional matching was also associated with higher ratings of trustworthiness toward the source of the message and a stronger belief that the message addressed their individual goals. These studies are unique in that they created messages to target all five of the previously elucidated functions rather than targeting a subset of the traditional five functions (e.g., using the self-monitoring scale to target either the social-adjustive functions or utilitarian function, Snyder & Debono, 1985). Although the VFI is a promising tool for targeting messages supporting volunteering, in order to use the functional measurement approach to tailor messages in other contexts, a more general instrument is needed, one that is not specific to a particular attitude object.
Several attempts have been made to produce such a measure. Herek’s (1987) items are topic specific, Herek making the argument that although his measure can be adapted to other domains, the wording may need to be changed from topic to topic to reflect the specifics of that domain. Alternatively, Locander and Spivey (1978) produced evidence consistent with the validity of a general measure of Katz’s (1960) four functions, but their measure was developed assuming that the only source of an attitude function is personality, this instrument being an attempt to measure features of personality thought to be linked with particular functions. Thus, the instrument neglects the influence of the attitude object and the situational features that also impact attitude function. Franc and Brkljačic´ (2005) also present a general measure, but it is limited in that it assesses only the utilitarian, social-expressive, and value-expressive functions. Consequently, it provides a foundation on which to build indicators of the knowledge function and the ego-defensive function. If attitude functions can be measured effectively, the influence of personality, object, and situation can be estimated and theory developed to explain their varying effects.
Explanations for the Functional Matching Hypothesis
The majority of studies examining the functional matching hypothesis have found evidence consistent with its predictions regardless of the source of the attitude function. Although the effect appears robust, explanations for it remain uncertain. Lavine and Snyder (1996) argued that the effect arises because of biased processing. Specifically, when the audience receives a functionally matched message, they are predicted to perceive the message as of higher quality and to produce more favorable message-relevant thoughts than when they are exposed to a mismatched message. Furthermore, Lavine and Snyder predicted that perceptions of message quality would mediate the relationship between functional matching and attitudes, and they found evidence consistent with these hypotheses in two studies examining attitudes toward voting.
The elaboration likelihood model (ELM; Petty & Cacioppo, 1986) has also been used to explain the functional matching hypothesis results. There is inconsistency, however, in deriving predictions from the ELM. Snyder and Debono (1985) propose that functional matching operates as a peripheral cue such that those who are not processing centrally employ the heuristic cue that matched arguments are better regardless of quality. If a message produces attitude change via means other than message content, it follows from the ELM that peripheral processes are operative. Debono (1987, study 2) found that evidence consistent with this reasoning. Specifically, functional matching increased the persuasiveness of a message even when the speaker produced no actual arguments.
Alternatively, Debono and Harnish (1988) argued that functional matching increases depth of processing so that mismatched messages are processed peripherally and matched messages are processed centrally. They found, consistent with ELM predictions, that when a message provided a functional match, argument strength had a stronger impact on the persuasiveness of the message than when the message did not provide a functional match. Furthermore, consistent with the ELM-based hypothesis of increased processing, they found that the ratio of positive to negative thoughts generated by their subjects was associated strongly with post-message attitudes when the message matched functions but not when the message failed to match functions. This latter effect was not replicated, however, by Debono (1987).
In their second study, Petty and Wegener (1998) found evidence consistent with the ELM explanation as the effects of argument strength and functional matching were limited to those low in need for cognition. They argued that if the audience is motivated strongly to process arguments carefully and thoroughly (in this case, those who chronically process centrally due to their high need for cognition), functional matching will not affect the depth of processing the message. Functional matching is anticipated to increase the effect of argument strength for only those low in need for cognition (i.e., for those for whom the default is limited processing). On the other hand, in their first study in this article, Petty and Wegener (1998) found that functional matching did increase the persuasiveness of strong arguments and decrease the persuasiveness of weak arguments. Yet, there was also a strong main effect for functional matching such that matched arguments were less persuasive than mismatched arguments. Thus, this experiment failed to replicate functional matching hypothesis predictions that matching is expected to increase the persuasiveness of the arguments. Given the inconsistency of these results, the ELM may not be adequate as an explanation of this phenomenon (cf., Stiff & Boster, 1987).
Others have offered a provocative alternative derived from the unimodel (Thompson, Kruglanski, & Spiegel, 2000). The unimodel posits that the functional matching effect results from functionally matched messages providing information that is relevant to the audience. According to Thompson et al., attitude functions serve to determine what kind of information is relevant to the audience’s goals. Information that is irrelevant is ignored just as easily as information from bad arguments. No studies have been published reporting a critical test of this uni-model prediction as compared to the various ELM predictions, but Hullett (2002, 2004) has found consistently that the functional matching effect is mediated by the perception that the message is relevant to one’s values.
Structure of Attitude Functions
* * *
Katz (1960) and Smith et al. (1956) produced similar lists of functions. Herek (1986) divided them up somewhat differently but other than expanding the types of utilitarian functions and dropping the knowledge function, he produced a similar list. More recent scholarship has begun to question the adequacy of that list (Hullett & Boster, 2001; Maio & Olson, 2000; Watt, Maio, Haddock, & Johnson, 2008).
Maio and Olson (2000) argued that utilitarian attitudes are not conceptually distinct from value-expressive attitudes because attitudes that are based on maximizing utility require some value to determine what is good and bad for the individual. They found that their subjects’ attitudes toward attending a proposed music festival immediately before finals were substantially correlated with utilitarian values related to enjoyment and negatively related to utilitarian values re
lated to achievement. They found that just as responses to appeals based on other values more commonly associated with value-expressive attitudes were substantially related to the specific values that were targeted, targeting utilitarian values operated the same way. It is also worth noting that these investigators were following up on an important point they had made previously (Maio & Olson, 1994); namely, that knowing that an attitude is value-expressive does not allow functional matching in and of itself. Rather, one must know the specific targeted value. One could determine that the target of a message has a value-expressive function for that attitude, but if the message was not targeted at the correct value, it would fail to persuade.
Although Maio and Olson (2000) argued that utilitarian attitudes may simply be value-expressive attitudes with a particular utilitarian value, Hullett and Boster (2001) make a similar argument such that social-adjustive attitudes are a particular type of value-expressive attitudes. They argued that people with social-adjustive attitudes are simply pursuing an other-directed value (the extent to which the individual’s values conform to the wishes of others). They found that when the message was based on supporting conformity, the degree to which the target had other-directed values was related directly to agreement with the message.
The bulk of the research that has demonstrated a successful functional matching effect in the second wave of functional attitude research has been conducted on utilitarian, social-adjustive, or value-expressive attitudes (see Clary & Snyder, 1992 and Julka & Marsh, 2006 for exceptions). If utilitarian, social-adjustive, and value-expressive attitudes are all simply value-expressive attitudes, all this line of research has demonstrated is that it is more effective to target the value that the individual associates with the attitude than to target a value the audience either does not support or does not associate with the attitude being targeted. In order to determine if any given proposed function is actually a separate function, it would be helpful to return to Katz’s (1960) original purpose of delineating functions; namely, to determine if the processes by which attitudes change differ based on the function. In an early statement of functional attitude theory, Katz and Stotland (1959) discussed the possibility of attitude functions varying based on the intervening variables between the function and the attitude. The third wave of functional attitude research examines the different types of cognitive processes that occur when different attitude functions predominate or are salient (Hullett, 2002; 2004; 2006, Hullett & Boster 2001; La France & Boster, 2001; Lapinski & Boster, 2001). These studies can be used to demonstrate which functions produce the same cognitive processes, and thus could be profitably combined, and which produce different cognitive processes and should therefore be distinguished.
Although the ego-defensive function produced the most research in the initial phase of functional attitude research (e.g. Katz, McClintock, & Sarnoff, 1957; Katz, Sarnoff, & McClintock, 1956), it is the most understudied function in the second wave of functional attitude theory research, a wave ushered in by the extensive use of the self-monitoring scale (Snyder & Debono, 1985). Lapinski and Boster (2001) sought to address this lacuna by employing causal modeling to understand the cognitive processes that occur when an ego-defensive attitude is challenged. Rather than exploring how to target ego-defensive attitudes, they modeled the process by which ego-defensive attitudes become more resistant to change, focusing on how people defend their ego from a message that attacks a pivotal aspect of their self-image. In addition to the inherent benefit of studying resistance to persuasion, their work suggests future work focusing on reducing the processes that cause resistance. They found evidence consistent with a causal chain in which the degree to which a message is ego-threatening produces predominantly negative thoughts in the audience. The production of negative thoughts leads to message discounting and then source derogation. Finally, increasing source derogation resulted in an increasingly negative attitude. This model describes a serial process that people experience in order to protect their ego from a message that suggests a discrepancy between their behavior and their self-image. This process is different from the processes identified by Hullett (Hullett, 2002, 2004; Hullett & Boster, 2001) concerning value-expressive, social-adjustive, and utilitarian functions.
Hullett and Boster (2001) examined messages that targeted the social-adjustive and value-expressive functions. When the message advocated conformity, the relationship between the extent to which the audience valued conformity and their agreement with the message was mediated by perceptions of message quality (although the extent to which they valued other-directedness did have a direct effect on message acceptance). When they used a message that advocated self-direction (values associated with independence and following one’s own desires), the relationship between that value and message acceptance was again mediated by perceptions of message quality. Lavine and Snyder (1996) found a similar effect for social-adjustive and value-expressive functions such that perceptions of message quality mediated the relationship between the functional match of the message and message acceptance.
Hullett (2002) expanded this process-oriented approach by measuring the extent to which targets of a value-expressive message actually held that value and the extent to which the message was relevant to the basis of their attitude. Hullett found that the extent to which the audience held a value influenced the extent to which they viewed the message targeting that value as relevant to their attitudes, which then increased their perceptions of message quality. His analysis suggests a process in which the audience initially determines if a message is relevant to the values associated with their attitudes, and then the extent to which the message is relevant determines the audience’s subjective determination of the quality of the message. As before, message quality perceptions directly influence message acceptance. Hullett (2004) went on to demonstrate that this process occurs for other-directed values (similar to the basis of the social-adjustive function) and self-centered values (similar to the basis of the utilitarian function), suggesting that regardless of the value that is targeted, the same process occurs. Hullett also found that regardless of which value the persuasive message targeted, if the message was perceived to be relevant to one’s reasons for holding the attitude, it was associated with guilt arousal, which increased message acceptance independent of the impact of message relevance on perceptions of message quality.
These studies on the cognitive processes underlying different functions are consistent with the perspective that value-expressive, utilitarian, and social-adjustive attitudes all seem to be changed via the same processes (Hullett, 2002, 2004; Hullett & Boster, 2001) and thus may not represent different functions. On the other hand, the research by Lapinski and Boster (2001) on the unique processes associated with ego-defensive attitudes, suggests that this function may be a theoretically distinct attitude function. The remaining function of the original five, the knowledge function (Katz, 1960), has only been the focus of a small number of studies (Clary et al., 1998; Clary et al., 1994; Clary & Snyder, 1992; Julka & Marsh, 2006). The studies by Clary and colleagues have all been basic functional matching studies that did not investigate whether or not the knowledge function was associated with a different process. Julka and Marsh’s (2006) study did not investigate the process by which attitudes with a knowledge function change, but they did find that the knowledge function could be induced similarly to how the value-expressive function could be induced (e.g., by causing the audience to feel that they had failed to reach the goal associated with each function). It remains to be seen whether the processes associated with the knowledge function simply represent another value and could be subsumed as another type of value-expressive attitudes or if there are unique cognitive processes associated with the knowledge function that would require a unique message strategy as the ego-defensive function seems to require.
Challenges
* * *
Theoretical Challenges
Hullett (2002, 2004) found evidence consistent with the pe
rspective that value-relevant, social-adjustive, and utilitarian functions all engage similar, serial cognitive processes. Critical tests are needed to compare some of the ELM predictions with the more complex causal processes Hullett examined. In general, not enough attention has been paid to the cognitive processes that generate the functional matching effect. Unless and until functional attitude scholars advance from focusing on mere functional matching effects to trying to understand the causal processes associated with the different functions as initiated by the third wave of functional attitude research, progress will not be made toward a comprehensive functional theory of attitudes.
As with much persuasion research, the studies that emerge from viewing attitude functions as a process are cross-sectional, and have the corresponding limitations in the strength of the causal inferences that can be drawn from this method of data collection. And, although future research would profit from longitudinal data collection, it would profit even more if longitudinal data sets emerged from designs derived from dynamic theories. Meeting this condition would require that the structural equations implicit in Hullett and Boster’s (2001) and Lapinski and Boster’s (2001) papers be replaced by change equations or differential equations detailing the processes by which each variable in these cross-sectional models change. Theory of this level of sophistication would then dictate many features of design and data collection, as well as provide straightforward tests of the model (cf., Boster, Mayer, Hunter, & Hale, 1980; Boster, Hunter, Mongeau, & Fryrear, 1982).
The SAGE Handbook of Persuasion Page 23