Implications of a Recognitional Decision Model For Consumer Behavior

Citation:

Caroline E. Zsambok (1993) ,"Implications of a Recognitional Decision Model For Consumer Behavior", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 239-244.

Advances in Consumer Research Volume 20, 1993      Pages 239-244

IMPLICATIONS OF A RECOGNITIONAL DECISION MODEL FOR CONSUMER BEHAVIOR

Caroline E. Zsambok, Klein Associates Inc.

OVERVIEW

This paper presents a different approach to understanding decision making compared to the traditional decision research paradigm within the field of psychology. First, in my introductory comments, I will briefly describe those differences that have relevance for this audience. Second, I will describe Recognition-Primed Decision Making (RPD), which models how experienced people often make decisions in their operational settings. Then, I will discuss the methodology that we used to study experienced decision makers. Last, I will conclude with implications of the model and methodology for studying and influencing consumer behavior.

INTRODUCTION

Traditionally, decision researchers have used a laboratory research paradigm to study the varieties of option selection strategies that people (usually college students) use to select a single option from many. Most of these strategies require comparing options to each other in order to choose an acceptable (or the best) one. Further, while the options in the choice set often concern a domain that is familiar to the subjects (e.g., selecting a car; choosing a job), subjects are not experts with that task and cannot draw upon an experience base of decision making within that context (Zsambok, Beach, & Klein, 1992).

In contrast, we at Klein Associates Inc. have worked in operational settings with experienced people like urban firefighters, military battle managers, neo-natal intensive care unit nurses, data programmers, paramedics, and design engineers. From these studies, we have modeled a decision-making process that these experienced people use for many of the decisions they make on the job. Called the Recognition-Primed Decision Making (RPD) model (Klein, 1989), it shifts the emphasis from option selection to situation assessment. It also describes ways that decision makers evaluate options without comparing them to each other.

In settings like these, we found that the lion's share of the task involves sizing up the situation, or achieving a situation assessment. In this assessment, decision makers pay attention to critical cues from the environment, they watch for violations to their expectancies about the situation, they become aware of goals that are plausible for this situation, and they consider typical actions they have taken in the past to cope with similar situations. Therefore, the understanding of the situation suggests a reasonable option. Generally, to assess the situation is to make the option apparent.

One of the most interesting findings from our studies is that much of the time, once an option is apparent, expert decision makers do not attempt to generate more options in order to compare them and select the best one. Rather, they use mental simulation to evaluate whether the option they are currently considering will work. If they believe it will not, and if they cannot imagine how they would modify it to make it workable, they reject it and generate another one. And, after they have generated the next option, they do not compare it to the previous one to determine which is better. Thus, Recognition-Primed Decision Making involves a non-comparative option adoption process.

Why is this model of interest to people who study consumer behavior? First, if targeted consumers are not using an option-comparison strategy in order to make their purchase decisions, then marketing and advertising that intends to affect option comparison is not relevant. Rather, marketing and advertising messages should address the non-comparative option evaluation processes that we found people using, like mental simulation. Or, if the consumer base uses both comparative and non-comparative strategies, then messages should be developed that are compatible with both of them.

A second reason why this model could be of interest to the study of consumer behavior concerns the methods we used to develop the model. As I will describe below, these methods allow an interviewer to uncover the mental model that a person has of a situation or decision event. Knowing about the information contained in mental models of relevant consumers can be useful in developing marketing and advertising messages, both to retain existing customers and to attract new ones.

RECOGNITION-PRIMED DECISION MAKING

Klein (1989; Klein, Calderwood, & Clinton-Cirocco, 1986) has developed a model of Recognition-Primed Decision Making that describes how experienced people commonly make decisions in their operational settings. Based on observations from six field studies in different domains such as firefighting and tank platoon maneuvers, we found that decision makers:

1. focused more on situation assessment than on option selection

2. did not generate a number of options and then compare them to each other

3. did not look for the optimal options, but for a satisfactory one

4. evaluated options through mental simulation and selected the first satisfactory one.

As described in Klein and Klinger (1991), our initial work that led us to these conclusions began by observing fireground commanders (FGCs) and obtaining incident accounts from them. These FGCs were in charge of allocating resources and directing personnel. We studied their decisions in handling non-routine incidents during emergency events. Some examples of these types of decisions include whether to initiate search and rescue, whether to initiate an offensive attack or concentrate on defensive precautions, and where to allocate resources.

The FGCs' accounts of their decision making did not fit into a decision-tree framework, nor did it resemble an option-comparison strategy. The FGCs argued that they were not "making choices," "considering alternatives," or "assessing probabilities." They saw themselves as acting and reacting on the basis of prior experience; they were generating, monitoring, and modifying plans to meet the needs of the situations. We found no evidence for extensive option generation. Rarely were even two options concurrently evaluated. We did not see them trying to make optimal choices. The FGCs were more interested in finding an action that was "workable," "timely," and "cost effective."

Based on these and other protocol data (Klein et al., 1986), we attempted to model the way that experienced people were arriving at a workable plan, or option (see Figure 1). We found that much of their work in the decision event concerned situation assessment.

As an example, consider an experienced fire chief arriving at the scene of an unoccupied house that is on fire. He (in our studies, all chiefs were males) perceives a number of cues (features) from the environment. He sees smoke coming from under the eaves of a pitched roof, a red flame shooting out the attic window, a yellowish flame forming at an adjacent second-story window. The model postulates that if the situation is familiar to the chiefCif he has seen fires like this beforeCthese features will activate his memory for other situations he has seen, or for a prototypical instance which is the amalgamation of many such situations he has seen. [The model does not specify the nature of stored information--examples or prototypes or both. The descriptive capability of the model remains functionally the same regardless.] This process corresponds to "framing" the situation, as described by Image Theory (Beach, 1990, 1992).

FIGURE 1

RECOGNITION-PRIMED DECISION MODEL

The initial features of the situation cue the chief's memory for past situations of this type. As a result, the chief becomes aware of additional critical cues to look for (wind strength and direction) besides the ones that initially drew his attention; about feasible goals (saving the property is not feasible, but saving the adjacent house is); about typical actions (three separate streams and two converging streams of water will be necessary to save the adjacent property and contain the primary fire), and about expectancies (you should be able to bring the red flame under control within five minutes, given these actions).

Before implementing a course of action, the chief may choose to evaluate it by mental simulation: mentally enacting or envisioning a course of action. (This process is akin to the "walk-through" process described by a number of researchers such as Gettys, 1983. [For a review of literature about mental simulation, see Klein and Crandall, in press]) That is, the chief imagines training the hoses on the designated areas using a particular angle and water pressure, for the expected amount of time. If he "sees" that the plan won't work, (he can't get the angle he needs, the plan is too vulnerable to wind shifts, the hotter of the two flames might require more of the water resource than he had planned) he modifies it, then implements. If he can't make it workable during his mental simulation and mental modification, he rejects it and retrieves other actions from memory to generate a new course of action. But, he does not compare the rejected course of action to the current one as a way to determine which is preferred.

We are not arguing that the mark of experienced people is that they use only a recognitional decision strategy like RPD in their operational settings. Rather, we are arguing that recognitional decisions do occur often, and that they constitute another type of decision making besides the option selection strategies that have been reported and studied under normative decision theory. And, our studies do show that the proportion of recognitional decisions increases as a function of level of expertise.

TABLE 1

PROPORTION OF RECOGNITIONAL DECISION POINTS IN DECISION EVENTS FROM FIVE DIFFERENT OPERATIONAL SETTINGS

Table 1 depicts the proportion of recognitional decisions that were made by experienced people in several domains we have studied. These data were gathered in both verbal incident accounts we collected from research participants and also in observations we recorded (and later analyzed) as they were performing their jobs. First, we identified decision points within each incident. Then, we evaluated each of these decision points to determine what type of strategy was used. We looked for evidence of option comparison versus no comparison of options. To be classified as a recognitional decision, an option or course of action had to be adopted without comparing it to other options. Inter-rater reliability for each of these studies ranged from 87% to 94% (Taynor, Crandall, & Wiggins, 1987).

Notice that the lowest proportion of RPD decisions was found for three groups. First, 46% of decisions made by the novice fireground commanders (they were not inexperienced with firefightingCthey were inexperienced as commanders) were recognitional, compared to 58% for experienced commanders. Second, for the tank platoon leaders, who had only five weeks experience on the job and were not considered expert, a proportion of 42% recognitional decisions was found. Third, the organizational decisions made by wildland commanders consisted of 39% recognitional decisions. These organizational decisions were made by teams, in which it was common for different members to throw out different ideas about courses of action that the team then discussed. Yet, even here 39% of the decisions were made without comparing the options. This proportion can be compared to the expected larger proportion of functional decisions that commanders made individually (56%), and with which they had considerable experience and expertise.

We believe that the reason less-experienced decision makers use an option comparison strategy more frequently than do their experienced counterparts is that they cannot assess the situation as well. Empirical support for this conclusion rests on our findings that non-experts spend less time assessing the situation than do experts, and more time comparing possible courses of action than experts do. Again, if the situation is understood, a good option often becomes apparent.

More recently, we analyzed decisions made by 31 experienced Naval officers in the Command Information Centers of AEGIS cruisers while they were at sea (Kaempf, Wolf, Thordsen, & Klein, 1992). These incidents were about situations that needed to be diagnosed as either hostile or non-hostile air threats to individual ships or whole fleets. Many of these incidents were marked with harassing acts from enemy aircraftCothers contained perplexing activity from unknown aircraft. The task of the officers was to make decisions that would maintain the safety of their ship and fleet without shooting at innocent or merely harassing aircraft. Again, using protocol data from incident accounts, and applying very stringent criteria to what constituted a decision point and a non-comparative decision strategy versus any other type of decision strategy, we found that approximately 95% of the actions taken by the decision makers were based on recognitional (non-comparative) decision processes. Inter-rater reliability was 96%. The decision makers generated multiple options for purposes of comparison for only 4% of the cases. In most cases, decision makers knew what to do once they understood the situation.

The high proportion of recognitional decisions in this study and also from the battle managers mentioned in Table 1 is very likely due to the domain: many of their decisions required that they follow doctrine or standard operating procedure. In both cases, situation assessment was key. Once they understood the situation, they knew what they had to do.

There are other models of decision making that, like the RPD model, emphasize the role of situation assessment and allow for non-comparative option adoption strategies. Examples include the story model of Pennington and Hastie, Beach's Image Theory, and Rasmussen's skill/rule/knowledge-based model of cognitive control, but space does not permit a discussion of these here. For a recent review of this literature, see Klein, Orasanu, Calderwood, and Zsambok, in press; and Zsambok, Beach, and Klein, 1992.

METHODOLOGY AND THE RPD MODEL

The methods we used to develop the RPD model include an interview process called the Critical Decision method (CDM) as well as concept mapping. As described in Klein, Calderwood, and McGregor (1989), CDM is a knowledge elicitation strategy based on Flanagan's (1954) critical incident technique. Using recollection of a specific incident as its starting point, CDM employs a semi-structured interview with specific, focused probes designed to elicit particular types of information from the interviewee. Solicited information includes goals that were considered during the incident, cues that the person attended to, expectancies held about what would happen as the decision event unfolded, violations to those expectancies that they noticed, and actions they took during the course of the event.

FIGURE 2

CONCEPT MAP OF DRIVING A CAR

Researchers at Klein Associates developed CDM to elicit the decision strategies used by experienced people like fireground commanders and emergency rescue personnel at the scene of a fire or emergency. We found that many of these decisions relied on subtle perceptual cues and assessments of changing events that were not easily articulated by the experts. Thus, probes had to be developed that would allow experts to focus on and describe aspects of their task that are normally only tacitly understood. CDM has been demonstrated to yield information richer in variety, specificity, and quantity than is typically available in experts' verbal reports (Crandall, 1989), and we have used it successfully in over a dozen studies and in domains as varied as fireground command, battle planning, critical care nursing, corporate information management, and commercial and helicopter piloting. For a more detailed description of the CDM interview process and the work surrounding it, see Klein (1989) and Klein, Calderwood, and MacGregor (1989).

Concept mapping is a method that produces a schematic representation of the meaningful relationships among units of information, like events, objects, and states. For example, Figure 2 is a concept map about driving a car. Non-relational concepts are depicted in ellipses which in this example consist of either objects (steering wheel), actions (stop), or events (driving a car). Relational concepts are depicted by arrows and consist of linkages like "by," "need to," and "has." The most significant feature about concept maps is that the information is formatted in a non-linear fashion, and is thought by most psychologists who study memory storage and retrieval to reflect more closely the organizational structure of information in memory than a linear format would. [See semantic network and spreading activation research as originated by Collins and Loftus, 1975; Collins and Quillian, 1982.] Concept maps allow you to quickly sense the amount of inter-relationships contained in the network (an index of its complexity), and the specific nature of the system. Originally devised as an instructional and evaluation tool for use in academic settings (e.g., Gowin & Novak, 1984), concept mapping has been used more recently in applied settings. For example, Air Force operations researchers used concept mapping to identify needs of users of a decision support system, and to develop work station designs (McFarren, 1987; McNeese, Zaff, Peio, Snyder, Duncan, & McFarren, 1990).

Both CDM interview data and concept mapping provide information about a person's mental model of different types of knowledge. As described by Donald Norman (1983), a mental model is a conceptual model of a target system. A target system is the system that the person is learning or using. It could include an event (the course of a particular type of house fire, and how to put it out), or an object that is a system (a nuclear reactor). A conceptual model is invented to provide an appropriate representation of the target system, appropriate in the sense of being accurate, consistent, and complete.

According to Norman, mental models are naturally evolving models. Through interaction with a target system, people formulate mental models of that system. These models need not be technically accurate (and, he says, usually they are not), but they must be functional. A person, through interaction with the system, will continue to modify the mental model in order to get to a workable result. Mental models will be constrained by such things as the user's technical background and previous experiences with a similar system.

The important point about mental models for this discussion is that they contain information about people's understanding of how things work (or how things happen). And, we believe that knowledge of how things work can be described in terms of the information considered during situation assessment: critical cues, plausible goals, expectancies, and actions. In fact, in research projects where we have been asked to discover what experts know about their operational domain that can be taught to intermediate-level personnel, the information that we have documented for training purposes concerns just those things: cues, goals, expectancies' and actions (Calderwood & Crandall, 1989; Crandall, Kyne, Militello, & Klein, 1992; Weitzenfeld, Klein, Riedl, Freeman, & Musa, 1991).

For example, in a study involving a neo-natal intensive care unit (NICU) at a hospital, we used the CDM method to discover what experienced nurses knew that nurses new to the unit did not know about one segment of their patient population. This segment consisted of very premature babies, called Extremely Low Birth Weight, or ELBW, babies who are born at 30 weeks gestational age and weigh less than three pounds. The purpose of the study was to discover how certain nurses could tell that an infant was in the beginning stages of what often becomes a fatal, systemic infection, called sepsis. This condition is common in ELBW infants because of their compromised immune system. Sometimes, if NICU staff waits until results from "hard" diagnostic tools like blood tests come back to confirm the onset of sepsis, the infant can be beyond help. The infection could have engulfed the system. But, we found that some nurses could tell when the baby was just beginning to get sick, and could alert the health care team so that early intervention could save the baby's life.

What we found (Calderwood & Crandall, 1989; Crandall & Gamblian, 1991) was that these nurses had a very different mental model, compared to new NICU nurses, of what cues to pay attention to (skin tone, alertness, muscle tone); what expectancies they should have and what the violations to these might mean (reaction to medical procedures and routine handling by nurses); what the plausible goals were ("we can't wait much longer to confirm diagnosis through testsCit's worth the risk to commence drug intervention, even with its potentially serious side effects"); and what actions they needed to take ("I'm going to stick really close to this babyCat the first sign of one more problem, I'm going to wake up the doctor to get an order for an antibiotic series").

We discovered that these nurses had learned what had not been taught to them in school. In Norman's terms, by interacting with the target system (the sick ELBW infant), they had developed a mental model that gave them a workable result: the ability to recognize and cope with the early onset of a deadly infection. In terms of the RPD model, they had refined the cue constellation that they attended to (compared to what they had been taught in school about general signs of infection), they had learned a different set of expectancies concerning the course of the illness, they had developed a set of different actions to take, and they had included new goals as plausible.

IMPLICATIONS OF THE MODEL AND THE METHODS IN CONSUMER BEHAVIOR

We have just begun using concepts derived from the RPD model and the methods we used to develop the model in order to study consumer behavior. Very recently, we completed market research projects for Procter & Gamble and Johnson & Johnson. Our approach was first, to assume that consumers might engage in situation assessment as part of their decision-making task, as opposed to generating numerous options that they would compare by weighing the pros and cons of the options' features. This meant we needed to use knowledge elicitation techniques like the CDM interview and concept mapping to uncover the mental models of consumers concerning the situation related to product use, and the manner in which the product could affect the situation.

While we cannot disclose the content of our findings, we can say that they have implications for the marketing and advertising messages that would be directed at the portion of the consumer base which uses or could be influenced to use recognitional decision making about these products. Further, we were interested in consumers who are not "expert" users of the product or diagnosers of the problem situation that the product was designed to address. These are people whose mental models do not include the most critical cues to look for in the problem situation; do not contain the most salient expectancies; do not reflect that higher goals (which the product can address) are plausible; or do not contain a full set of actions that are necessary on their part in order for the product to work as expected.

In general, the implications of our model and methods to the study of consumer behavior follow from this line of thinking:

1. The data from our studies of decision makers show that most of the work is in situation assessment (not option evaluation) and that when options are evaluated, frequently the method is mental simulation rather than comparing multiple options to each other.

2. The methods used to develop the RPD model include the CDM interview and concept mapping.

3. These methods expose knowledge about cues, goals, expectancies, and actions.

4. This information can be translated into a mental model held by an individual.

5. These methods can be useful in research about consumer behavior to

- uncover what people believe to be true of situations that are relevant to using (buying) a particular product or service

- uncover what people believe about the way a product or service works in those situations and what they notice in order to form these beliefs

- develop marketing and advertising messages that reinforce, deepen, or correct these mental models as they relate to the product or service.

6. In developing these messages, our model and methods would emphasize

- depicting how the product or service achieves its intent in ways that match consumers' mental models (for example, show how anti-lock brakes function on a car, using a visual and simplified representation to emphasize critical functions)

- raising consumer expectations about what this type of product or service should do (and how your own product does not violate those expectations), and teaching them what cues to look for, in order to enable them to notice violations of expectancies in competitors' products or services (for example, showing how smoothly the car should stop, and what counts as less-than-perfect smoothness)

- changing consumers' understanding of what constitutes a feasible goal (for example, slippery pavement does not mean that cars have to skid)

7. From our perspective, it would be important to know these things:

- How much of a role does situation assessment play in the use of a particular product? (Products or services that can be cast as solving a problem, such as how to relieve a stomach ache or how to protect your home, would seem to be strong candidates for this.)

- Are consumers likely to use an option-comparison strategy or a non-comparative strategy in selecting the product or service? (We would expect that buying items like cars and houses, which are expensive and for which consumers don't have a lot of purchasing expertise, would pull for comparative strategies. Less expensive items and those for which people have some expertise would pull for non-comparative option decisions.)

In conclusion, what we are suggesting is that marketing efforts and advertising messages can go beyond these three common current strategies and their hybrid combinations: 1) Comparing your product or service to other leading competitors in terms of how the options' features stack up to one another; 2) associating your product or service to an emotionally appealing life-style; and 3) price-point advertising. In those cases where situation assessment is likely, an additional strategy is highlighting information about the problem (situation) addressed by your product or service. This is done in such a manner that potential consumers come to refine and correct their mental model of the situation itself, as well as the effects of your product/service on the situation. Using methods like the CDM interview and concept mapping, you can pinpoint precisely which information is important to highlight.

REFERENCES

Beach, L. R. (1990). Image theory: Decision making in personal and organizational contexts. West Sussex, England: John Wiley & Sons Ltd.

Beach, L. R. (1992). Image theory: An alternative to normative decision theory. Proceedings of the Association for Consumer Research Conference. October, Vancouver, British Columbia.

Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82, 407-428.

Collins, A. M., & Quillian, M. R. (1972). Experiments on semantic memory and language comprehension. In L. W. Greg (Ed.), Cognition in learning and memory. NY: Wiley.

Crandall, B. (1989). A comparative study of think-aloud and critical decision knowledge elicitation. ACM SIGART, 108, April.

Crandall, B., & Calderwood, R. (1989). Clinical assessment skills of experienced neonatal intensive care nurses. Yellow Springs, OH: Klein Associates Inc. Final Report prepared for the National Center for Nursing Research, National Institutes for Health under Contract No. 1 R43 NR01911 01.

Crandall, B., & Gamblian, V. (1991). Guide to early sepsis assessment in the NICU. Fairborn, OH: Klein Associates Inc. Instruction manual prepared for the Ohio Department of Development under the Ohio SBIR Bridge Grant program.

Crandall, B. W., Kyne, M., Militello, L., & Klein, G. A. (1991). Describing expertise in one-on-one instruction. Final Report. Prepared under contract MDA903-91-C-0058 for the U.S. Army Research Institute, Alexandria, VA.

Flanagan, J. C. (1954). The critical incident technique. Psychological Bulletin, 51, 327-358.

Gettys, C.F. (1983). Research and theory on predecision processes (TR 11-30-83). Norman, OK: University of Oklahoma, Decision Processes Laboratory.

Gowin, D. B., & Novak, J. D. (1984). Learning to learn. NY: Cambridge University Press.

Kaempf, G. L., Wolf, S., Thordsen, M. L., & Klein, G. (1992). Decision making in the AEGIS combat information center. Fairborn, OH: Klein Associates Inc. Prepared under contract N66001-90-C-6023 for the Naval Ocean Command, Control and Surveillance Center, San Diego, CA.

Klein, G. A. (1989). Recognition-primed decisions. In W.B. Rouse (Ed.), Advances in man-machine system research Vol. 5, 47-92. Greenwich, CT: JAI Press, Inc.

Klein, G. A., Calderwood, R., & Clinton-Cirocco, A. (1986). Rapid decision making on the fire ground, Proceedings of the 30th Annual Human Factors Society, 1, 576-580. Dayton, OH: Human Factors Society.

Klein, G. A., Calderwood, R., & MacGregor, D. (1989). Critical decision method for eliciting knowledge. IEEE Transactions on Systems, Man, and Cybernetics, 19(3), 462-472.

Klein, G. A., & Crandall, B. W. (In press). The role of mental simulation in naturalistic decision making. In J. Flach, P. Hancock, J. Caird, and K. Vicente (Eds.), The ecology of human-machine systems. Hillsdale, NJ: Lawrence Erlbaum Associates.

Klein, G. A., & Klinger, D. W. (1991). Naturalistic decision making. CSERIAC Gateway, II(1), 1-4.

Klein, G. A., Orasanu, J., Calderwood, R., & Zsambok, C. E. (In press). Decision making in action: Models and methods. Norwood, NJ: Ablex Publishing Corporation.

McFarren, M. R. (1987). Using concept mapping to define problems and identify key kernels during the development of a decision support system. Master's thesis, School of Engineering, Air Force Institute of Technology, Wright-Patterson AFB, OH.

McNeese, M. D., Zapf, B. S., Peio, K. J., Snyder, D. E., Duncan, J. C., & McFarren, M. R. (1990). An advanced knowledge and design acquisition methodology: Application for the pilot's associate. Final Report AAMRL-TR-90-060. WPAFB, OH: Armstrong Aerospace Medical Research Laboratory.

Norman, D. (1983). Some observations on mental models. In D. Gentner and A. Stevens (Eds.), Mental models. Hillsdale, NJ: Erlbaum Associates.

Taynor, J., Crandall, B., & Wiggins, S. (1987). The reliability of the critical decision method (KATR-863(B)-87-07F). Yellow Springs, OH: Klein Associates Inc. Prepared under contract MDA903-86-C-0170 for the U.S. Army Research Institute Field Unit, Alexandria, VA.

Weitzenfeld, J. S., Klein, G. A., Riedl, T. R., Freeman, J. T., & Musa, J. (1991). Knowledge elicitation for software engineering expertise. Proceedings of the Fifth Software Engineering Institute (SEI) Conference on Software Engineering Education. NY: Springer-Verlag.

Zsambok, C. E., Beach, L. R., & Klein, G. (1992). A literature review of analytical and naturalistic decisiommaking. Fairborn, OH: Klein Associates Inc. Prepared under contract N66001-90-C-6023 for the Naval Command, Control and Ocean Surveillance Center, San Diego, CA.

----------------------------------------