Advances in Consumer Research Volume 18, 1991 Pages 225-231
EXPERT-NOVICE DIFFERENCES IN COMPLAINT SCRIPTS
Ingrid Martin, University of Southern California
[The author gratefully acknowledges the guidance and helpful comments of Henrianne Sanft as well as three referees.]
Script theory is the theoretical framework used to investigate expert/novice differences in a complaint domain. This exploratory study investigated the differences in content knowledge between experts and novices as well as the differences in the ability to generalize to novel complaint situations. Three hypotheses were examined. H1 found that there is a significant difference between experts and novices based on prior knowledge, certainty and effort expended on the complaint process. H2 and H3 found that experts are better able to generalize their knowledge to a novel problem.
We all know what it is like to discover that your luggage hasn't arrived or that your flight will be delayed for several hours. But, do we all know how to resolve these problems? Typically, these problems are resolved by complaining. However, not all people know how to complain, and those that do, may not know how to complain effectively. For example in a recent article in the Wall Street Journal, it was found that "... many fliers don't know that most airlines will put them on another carrier - if asked to - when a flight is canceled". It was concluded that "... you've got to know to ask" and many consumers just don't have the appropriate knowledge base to deal effectively with complaint situations.
When representative groups of consumers were examined in various studies, it was found that a large number of consumers do not complain or complain ineffectively when confronted with problems concerning products, services or issues (e.g. Best & Andreasen 1977). Some of the factors that have been consistently found to influence propensity to complain include the economic and psychological costs and benefits of seeking redress as well as various demographic characteristics. The critical issue that has not been addressed in the complaint literature is whether this large group of consumers has the awareness and knowledge required to effectively complain.
It is important to understand why some people are more effective at complaining than others. This requires determining the content of the knowledge base for complaining and understanding how an effective complainer differs from an ineffective complainer or a noncomplainer in a complaint domain.
This exploratory study categorizes effective complainers as experts and ineffective complainers as novices. This differentiation of complainers is based on the premise that experts have more complete representations of a task domain (Chi, et al. 1982). The appropriate organization of knowledge should allow experts to perform the complaint task more effectively than novices.
Using scripts as the theoretical framework to investigate consumers' knowledge in the complaint domain addresses the call for theoretical constructs to better understand consumer complaint behavior (Fornell & Didow 1980). This type of knowledge structure stores a set of standard event sequences that guide behavior in a routine situation. The objective of this study is to investigate the content of scripts for experts and novices in a specific complaint domain. For example, expert complainers should have elaborate scripts which contribute to more effective complaint outcomes, whereas novices in a similar complaint situation, will have a much simpler script resulting in less effective complaining.
What follows is an exploratory study designed to address the differences in content knowledge of experts and novices in a complaint domain. Previous research has established the need for further study regarding the complaint process to better understand the phenomenon (e.g. Day, et al., 1978). The critical differentiating factors in the content of scripts for expert and novice complainers are the amount of prior knowledge in the complaint process, the certainty they have in the complaint outcome, and the overall effort expended on complaining.
A script is defined as a "... coherent sequence of events expected by the individual, involving him either as a participant or as an observer" (Abelson 1976, 1981). It consists of knowledge stored in long-term memory as sets of well structured cognitions that have been learned over time as experience accumulates. Specifically, these mental representations contain a series of actions causally linked in hierarchical order involving props and participants in common activities.
The expectations that an individual has in a particular situation act as a cue to automatically trigger the appropriate script. For example, if you order a hamburger and the wait-person brings you a ham and cheese sandwich instead, the expectation that you were to get a hamburger acts as a cue to trigger the complaint script. At this point, you will tell the waitperson a mistake has been made and to bring the correct order. The usefulness of a script is related to its ability to conserve limited processing capacity. Since a script retains the action sequences of prior situations, the individual doesn't have to relearn the steps every time this sequence of actions is needed. She can use that processing capacity to better develop her script to include new steps needed to more effectively complain.
This implies that once a person has a well developed script for some sequence of actions such as eating in a restaurant or complaining about poor service on an airline flight, the expert's script will result in routine response behavior. The novice, in contrast, will have to seek out information and put more effort into learning the appropriate complaint process.
Experts and Novices
The literature on expert/novice differences addresses several important factors that are relevant to this study. First, novices and experts have consistently been found to differ on the complexity and size of their knowledge base. Experts have larger and more complex knowledge bases from which to draw on when "... mapping new problems in terms of familiar patterns" (Charness, 1983). This is extended to the problem solving situation that experts and novices face when a problem arises concerning consumer products or services. The expert's complaint script should have the knowledge to allow her to generalize from one familiar complaint situation to a new complaint situation in a given domain. For example, an expert complainer in poor service at a restaurant, when faced with a new problem of a broken toaster, will have the knowledge to effectively complain.
Following from above, the literature shows that experts in domains such as chess, and other problem solving situations, have been found to produce more accurate solutions than novices (e.g. Larkin, et al. 1980). This would imply that expert complainers are also more effective complainers than novice complainers. Chase and Simon (1973) found that skill level in chess players varies by the number of move generations. In other words, less skilled players (novices) might run out of potential moves before the more skilled players (experts). The solution to these problem solving situations (i.e. chess) is mediated by the level of knowledge, potential moves, and other variables encountered by the player. This is similar to the mediating factors (e.g. who is receiving the complaint, what type of day they are having) encountered by the complainant. This implies that novice complainers either give up or resort to "sabotage" behavior unlike an expert complainer.
Charness (1981) conducted a study on age and skill differences in chess playing and found that expert chess players search more extensively and deeply for a solution. This can be extended to the complaint arena where an expert complainer would expend more overall effort to pursue a problem than would a novice. This increase in effort to solve the problem is a result of a knowledge structure with a larger, more complex content.
Underlying Dimensions Of Scripts
The study of expert-novice differences in schematic and category structures has resulted in several common underlying dimensions (e.g. Galambos & Rips, 1982). The underlying factors used to differentiate the content of the knowledge base of the expert and novice include centrality, contingency and hypotheticality measures. These three dimensions play a critical role in the different complaining results found between experts and novices. [A fourth dimension, distinctiveness, has been identified as a factor in cognitive scripts. This dimension refers to the extent to which an action is unique to a particular track of a script. Subjects were asked for a specific complaint situation rather than a general complaint script. There were little, if any, listings of distinctive actions in the written protocols of either experts or novices.] Centrality is a "... measure of how important an action is to its activity" (Galambos, 1986). These actions are the major goals or subgoals of the activity. For example, in the restaurant script the central actions would be "look at a menu, order meal, eat food, pay bill, leave" (Bower, et al., 1979). An individual who has little experience in a given domain will have the major goals and subgoals relevant to a particular domain related task. Galambos (1986) found that subjects' centrality information tended to be coordinated with subgoal information in activities. The above implies that as prior experience, certainty in the outcome and overall complaining effort increase, based on many single experiences in a particular domain, so will the number of possible central actions within the complaint script.
Contingency is the hierarchical structure of well developed scripts among experienced subjects which includes conditional statements to cover more variations. These conditional statements serve as decision trees containing alternative sets of subgoals when obstacles are encountered in a complaint situation. For example, if a store clerk is unable to help with a defective toaster, then a expert complainer would ask to speak to the store manager. This would continue until a satisfactory outcome was obtained much the same as the expert bridge player who revises his bid given added information from his partner. This also shows that subjects with less expertise are more constrained in how they are able to deal with variations in the script. For example, a novice bridge player may not realize the need to revise her bid.
The last dimension, hypotheticality, refers to the level of abstraction of the knowledge structure. Greater expertise allows individuals to generalize from previous experiences to a new experience. Individuals with less expertise have scripts with more concrete representations and display less successful behavior when confronted with a novel situation. For example, an expert should be able to effectively complain in a new complaint situation in the same domain. This does not hold for a novice who has less general knowledge about complaining due to less prior experience, less certainty in the complaint outcome, and less effort expended on the complaining process.
Determinants of Script Dimensions
The three independent variables believed to determine the level of centrality, contingency, and hypotheticality of the knowledge structure are prior knowledge in the specific domain of interest, certainty in the complaint outcome, and the overall effort that subjects will devote to the complaint process. This is based on the concept that learning involves any process that "... modifies a system so as to improve, more or less irreversibly, its subsequent performance of the same task or of tasks drawn from the same population" (Langley & Simon, 1981). This implies that as expertise increases subjects should be able to go from using a situation specific knowledge structure to a more general knowledge structure.
The independent variable, prior knowledge, is based on the direct and vicarious experience that a person has with the complaint process. Alba and Hutchinson (1987) found that the ability to solve a problem is partially determined by one's prior experience and knowledge in that domain. In familiar complaint situations, prior experience may lead to the direct retrieval of a prior solution outlined in an elaborate, routinized script. Another aspect of prior knowledge is the simple repetition of the task performance resulting in a more elaborate knowledge base.
The next independent variable is certainty in the successful complaint outcome. In other words, the amount of confidence that a person has in the resulting outcome of a problem has a positive effect on knowledge (Fischoff & MacGregor, 1982). This implies that lack of expertise in an area tends to increase perceived risk- and decrease certainty in the knowledge a person has about an issue. Hence, a person is more likely to cease complaining when confronted with an unresponsive complainee. The resulting uncertainty tends to invoke heuristics (Folkes, 1986). For example, if the ticket agent doesn't help you with your problem then you conclude that airlines are unresponsive to customer needs and you cease to complain. This means that certainty in a successful complaint outcome can lead to an increase in expertise. Therefore, experts are believed to have a higher certainty measure than novices when involved in a complaint situation.
The independent variable, overall effort, is based on the attribution theory of motivation. This theory suggests that when people make behavioral attributions concerning a problem they are motivated to act (Weiner 1980). People want to maximize the probability of a successful complaint, therefore, putting more effort into complaining can increase the certainty that they will be successful. The increased effort involved in ensuring a successful outcome furthers the mastery of the complaint process. Thus, to ensure a successful outcome, an expert will expend more effort to pursue all the required subgoals. This is in contrast to the novice, who lacking the elaborate script structure, will display minimal effort in complaining. Instead the novice may expend more effort in such action as negative word of mouth, boycotting stores and switching brands. This is be discussed further in the supplemental analysis section.
The content of the knowledge base that expert and novice complainers have is organized in terms of central actions, contingency sub-branches, and hypotheticality measures. The central actions are the goals and subgoals of the script whereas the contingency statements lead the subjects to the relevant subgoals. The hypotheticality information determines the level of abstraction that the knowledge is organized in memory. This content base is shaped by the different levels of prior experience, certainty, and effort that experts and novices attach to the complaint process. The research hypotheses of interest are:
H1: As prior knowledge, certainty, and overall effort increase, subjects' scripts will provide mote central actions, contingency statements, and hypotheticality measures.
H2: An expert's script will not differ significantly across familiar and unfamiliar complaint situations for centrality, contingency and hypotheticality measures.
H3: A novice's scripts will be significantly different across familiar and unfamiliar complaint situations for centrality, contingency, and hypotheticality measures.
The first hypothesis tests the content of the knowledge between experts and novices whereas the remaining two hypotheses investigate the differences in the ability of experts and novices to generalize to domain tasks in which they have no experience. It is expected that novices should not be able to generalize across new complaint tasks whereas experts should have no problem generalizing to new complaint situation.
The dependent variables in this study are centrality, contingency, and hypotheticality. Centrality is operationalized as the total number of central or basic actions reported in the written protocols. The centrality measure was determined by comparing both scripts for common statements. These basic actions were determined to be critical to all complaint situations. An example would be "... complain about problem to appropriate person".
Contingency was measured by counting the number of conditional clauses in the elicited scripts. The methodology developed by Martin, et al. (1980) was used to code statements in each protocol that included such clauses as "either, or; neither, nor; if-then; depending on; unless; etc".
Hypotheticality was determined also using the Martin, et al. (1980) methodology of summary totals of the number of general statements made, roles mentioned, and frequency indicators. The approach was based on the premise that "... increased ability for abstraction requires more general statements, frequency indicators, and roles to summarize multiple incidents whereas describing isolated episodes does not" (Leong, et al. 1989). [The preceding measures were evaluated by two coders who were blind to the objectives of the study. The judges were required to code separately each sentence of the written protocol. For each sentence the judges coded the centrality, contingency and hypotheticality measures. The results of each protocol were compared and discrepancies were discussed between the judges until an agreement was made as to the appropriate code.]
The hypotheticality measure, was coded for four different categories of statements. The first was the number of times that frequency words were used such as "frequently, some, many, and other synonyms. The second coded group was the number of general statements that subjects made about the complaint process. This included statements such as "the waitress will usually..." or other statements which refer to all events. The third group of codes related to the number of job or role titles that were mentioned in each protocol. For example, "... I -reported the problem to the clerk ... then I asked to speak to the manager...". The final code was tallied for the number of examples or incidental events included in each protocol.
The independent variables for this experiment were prior knowledge, certainty, and effort. Prior knowledge was a composite measure of frequency of complaint experiences and familiarity with the complaint process. Certainty and overall effort were measured using likert scales. The certainty measure looked at the likelihood that experts would have a higher level of confidence in the complaint outcome. Overall effort measured the likelihood of pursuing all possible complaint options.
The students in this study were 35 undergraduate students in an advertising course in a major university. The dependent constructs for this study were operationalized in the form of two script elicitations. The subjects were asked to report the types of complaint situations they had been involved in from a comprehensive list of possible complaint situations. This list included such complaint situations as poor service in a restaurant, defective merchandise, lost baggage, and so on. They indicated the approximate number of times they had complained about a problem over the last year. Next, half of the subjects were asked to select a complaint situation in which they had no experience from this list and to write down all the possible steps they would take to complain until all avenues were exhausted. The other half of the group was asked to do the same task for the complaint situation in which they had the most experience. Then the next task was reversed for both groups so that all subjects provided two script elicitations, one for an experienced complaint situation and one for a novel complaint situation. The students were given as much time as they wished to finish the task. After completing the questionnaire, the students were debriefed concerning the study.
Due to the exploratory nature of this study, expertise was measured, not manipulated, resulting in the need to conduct a median split on the three independent variables. This approach was used so that subjects could be categorized into either expert or novice groups. The-median value of the prior knowledge was five complaint experiences in one year reported by the subject. The second independent variable was a five point certainty scale with a median value of three. This was midway between being very certain and very uncertain of a successful resolution to the complaint. The last independent variable was the likelihood that a person would make the effort to complain about the problem. A median split resulted in a value of three, midway between very unlikely and very likely that someone would make this effort. The next step categorized subjects into expert and novice groups if they scored either above or below the median value on all the independent variables. The result was a sample size of 10 novices and 9 experts.
The first hypothesis (H1) was concerned with the difference between experts and novices in measures of centrality, contingency, and hypotheticality. Three separate repeated measures ANOVAs were run with one within-subjects factor (familiar and novel experience) and one between-subjects factor (expert-novice). The result of the analysis found that expert complainers had a significantly higher number of central actions, contingency statements and hypotheticality measures. (See Table One.) This implies that experts may have a more abstract and elaborate script content. The interaction effects were not significant for centrality and contingency measures. There was a significant interaction effect between expertise and familiar and novel experience situation in the hypotheticality measure. This significant interaction points to the possibility that an increase in hypotheticality measures (i.e. roles, general statements, incidental events) may be critical to making a novice an expert. That is, increasing people's knowledge about various roles when complaining may point to the alternatives available to the complainer. In the airline complaint script, learning that other people (i.e. customer relations personnel, ticket agents, pilots, travel agents) are possible outlets for complaining is critical to a successful complaint outcome. This means they are not confined to stop the process at the first person with whom they complain. The knowledge that other functional roles are available to move their complaint forward will increase expertise in the process.
RESULTS FROM REPEATED MEASURES ANALYSIS OF VARIANCE FOR EXPERTISE AND FREQUENCY
MEANS AND STANDARD DEVIATIONS FOR DEPENDENT MEASURES
The second hypothesis predicted that experts would have no problem generalizing their script knowledge to a new complaint situation. This was tested in a difference of means test for the three dependent measures. The results show that experts were not significantly different in centrality, contingency and hypotheticality measures when asked to relate a complaint script for a complaint task in which they had no experience (t=1.47, p=.09; t=.92, p=.19; t=.288, p=.71, respectively). The results indicate that experts have the knowledge base needed to generalize their complaint script to a novel situation. This implies that an expert should be able to successfully complain about a new problem. This is similar to the expert chess player, who encounters a new chess move made by an opponent and is able to counter that move due to his ability to generalize from past experiences.
The third hypothesis (H3) stated that novices would not be able to generalize their script knowledge to a new complaint situation. The difference of means test for centrality and hypotheticality were highly significant (t=2.72, p=.008; t=2.30, p=.0175, respectively) and the contingency measure was marginally significant (t=1.54, p=.065). This means that a novice complainer, however capable of handling a specific problem, will not be able to generalize that knowledge to a new situation. Much the same occurs with the novice chess player who must concede the game when a new move is made by her opponent.
Earlier it was argued that novices would put less overall effort into the complaint process and instead, turn to such actions as boycotting the store, switching brands, warning friends and family to avoid the firm and so on. A directional difference of means test was conducted on the number of "sabotage" statements made by both experts and novices. The results show that novices did tend to include more "sabotage" statements in their scripts than experts (t=2.02, p=. )3).
DISCUSSION AND CONCLUSIONS
This exploratory study was designed to address two specific questions about the content of the knowledge base in a given domain. This was accomplished through the comparison of the knowledge base of an expert with that of a novice. The first critical finding is associated with the content of the knowledge base in memory between these two groups. Experts have significantly more central actions, contingent actions, and hypothetical information stored in their scripts. This allows them the flexibility to deal with many different obstacles that could be encountered in the complaint process.
The second critical finding is the difference between the ability of an expert to abstract to new complaint situations whereas a novice is more restricted in her/his ability to generalize. This restriction in the novice's knowledge is what a columnist for the Wall Street Journal referred to when he said that "... many travelers say complaints don't bring the results they used to, but speaking up still can have rewards with airlines". It goes back to the basics; you have to know how to complain effectively.
By understanding the content of the knowledge structure one uses to guide behavior, we are one step closer to understanding the process by which a novice becomes an expert. An important theoretical implication from this study is that this lays the foundation for further research in understanding how novices become experts. Understanding the differences between the two knowledge bases can provide some insight into how knowledge for routine event sequences develop in adults.
Understanding the complaint process that consumers use is not only critical to a firm's customer relations but intrinsic to intraorganizational relationships as well. If a firm has a good grasp of what consumers' perceive as the sequential set of complaint actions, then a firm can develop a responsive set of policies that will portray them as customer-oriented. An example of this is Nordstrom's liberal policy of allowing their customers to exchange any product purchased at their stores no matter how worn or old it may be. The issue is that firms should use knowledge of consumer complaining as an insight into customer needs.
LIMITATIONS AND FUTURE RESEARCH
The methodology used to elicit the written protocols was established by Bower, et al. (1979). It was found that using this method of script elicitation, distinctive actions are not part of the subjects' explicit knowledge structure. Had the verbal protocol method been used, through direct probing by interviewers, the subject could also provide more of the implicit knowledge in memory related to distinctive actions. This is based on the argument that Ericsson and Simon (1980) make for the use of verbal reports, which if "... elicited with care and interpreted with full understanding of the circumstances under which they were obtained, are a valuable and thoroughly.reliable source of information about cognitive processes". This limitation can also be viewed as an contribution to the methodological aspect of script elicitation.
A second limitation was the need to measure rather than manipulate the independent variables resulting in a loss of half the subjects. Since this was an exploratory study, however, it allowed the determination of which independent variables may be critical for complaint expertise.
The next step in this research should be to investigate the same variables in an experimental setting using the verbal protocol methodology while manipulating the independent variables to determine how a novice becomes an expert. Based on the work in the learning literature on analogical learning, learning through doing, learning through reminding, etc. a plausible approach would be to develop three or four learning contexts. Then randomly assign subjects to one of the learning contexts and manipulate expertise based on the three independent measures used in this exploratory study. This should lead to interesting and relevant findings based on this exploratory study.
Abelson, R. P., (1976), "Script Processing in Attitude Formation and Decision Making", in Cognition and Social Behavior, John S. Carroll & John W. Payne (eds.), LEA, Hillsdale, NJ, 3345.
Abelson, R. P., (1981), "Psychological Status of the Script Concept", American Psychologist, 37(7), 715-29.
Alba, J. W. & W. J. Hutchinson (1987), Journal of Consumer Research, 13(1), 411 -54.
Best, A. & A. Andreasen (1977), "Consumer Response to Unsatisfactory Purchases: A Survey of Perceiving Defects, Voice Complaints, and Obtaining Redress", Law and Society, 701-37.
Bower, G. H., et al. (1979), "Scripts in Memory for Text", Cognitive Psychology, 11, 177-220.
Charness, N. (1983), "Age, Skill, and Bridge Bidding: A Chronometric Analysis", Journal of Verbal Learning and Verbal Behavior, 406-16.
Charness, N. (1981), "Search in Chess: Age and Skill Differences", Journal of Experimental Psychology: Human Perception and Performance, 7(2), 467-76.
Chi, M.T.H., et al. (1980), "Representation of Physics Knowledge by Experts and Novices", Technical Report No. 2, Learning Research and Development Center, University of Pittsburgh.
Day, R. et al. (1981), "The Hidden Agenda of Consumer Complaining", Journal of Retailing, 57(3), 86-105.
Fischoff, B. & D. MacGregor (1982), "Subjective Confidence in Forecasts", Journal of Forecasting, 1, 155-72.
Folkes, V. (1986), 'The Availability Heuristic and Perceived Risk", Journal of Consumer Research, 15(June), 13-23.
Galambos, J.A., (1986), "Knowledge Structures for Common Activities", Knowledge Structures, James A. Galambos, et al. (eds), LEA, Hillsdale, NJ.
Galambos, J.A. & Lance J.R. (1982), "Memory for Routines"> Journal of Verbal Learning and Verbal Behavior, 260-81.
Langley, P. & H.A. Simon (1981), "The Central Role of Learning in Cognition", Cognitive Skills and Their Acquisition, J.R. Anderson (ed), LEA, Hillsdale, NJ, 361-81.
Larkin, J.H., et al. (1980), "Expert and Novice Performance in Solving Physics Problems", Science, June, 1335-42.
Leong, M., et al. "Knowledge Bases and Salesperson Effectiveness: A Script-Theoretic Analysis", Journal of Marketing Research, 26(2), 164-78.
Martin, J. et al. (1980), "The Development of Knowledge Structures", Research Paper No. 557, Stanford University.
Schank, R. (1982), Dynamic Memory, Cambridge University Press, Cambridge, MA.
Weiner, B. (1980), Human Motivation, Holt, Rinehart, and Winston.