Deal Search: an Approach For Computer-Controlled Information Processing Experiments Involving Bargainable Attributes

Paul H. Schurr, State University of New York - Albany
Merrie Brucks, University of Arizona
ABSTRACT - Deal Search is a microcomputer program that can facilitate the study of information processing issues concerning bargainable and non-bargainable attributes. This article (I) discusses an information processing approach to bargaining, (2) provides details on how Deal Search may be used in computer-controlled experiments in this area, and (3) proposes some starting points for future research.
[ to cite ]:
Paul H. Schurr and Merrie Brucks (1991) ,"Deal Search: an Approach For Computer-Controlled Information Processing Experiments Involving Bargainable Attributes", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 591-596.

Advances in Consumer Research Volume 18, 1991      Pages 591-596


Paul H. Schurr, State University of New York - Albany

Merrie Brucks, University of Arizona

[The software discussed in his paper is available from either author. The copyright to Deal Search ((C) 1986) is owned by Paul H. Schurr and Merrie Brucks.]


Deal Search is a microcomputer program that can facilitate the study of information processing issues concerning bargainable and non-bargainable attributes. This article (I) discusses an information processing approach to bargaining, (2) provides details on how Deal Search may be used in computer-controlled experiments in this area, and (3) proposes some starting points for future research.


In a recent article by Brucks and Schurr (1990) the bargaining purchase process was examined as a-multi-attribute, multi-alternative choice task in which the attribute values are subject to change. Knowledge of attribute value ranges was manipulated to examine its effects on bargaining and non-bargaining purchase tasks. They found that buyers reduce information search when they have the option of bargaining. Also, knowledge or attribute value ranges appears to increase the degree to which buyers replace information search with bargaining. This study demonstrates the use of a computer-controlled experiment to gain understanding about how consumers search for information and make choices when product attributes are bargainable.

Their work utilized Deal Search, a menu-driven user interface that combined a number or novel features: (1) comparability between purchase tasks that do and do not include a bargaining component, (2) multiple brands or dealers, representing alternative bargaining partners, and (3) an offer-sensitive algorithm for generating responses to subjects who elected to bargain.

The Brucks and Schurr (199()) article provides sufficient description Of their methods for the purpose of reporting their experiment. However, their article did not attempt to discuss l)cal Search in such a way that would allow other researchers to easily envision how this same approach might be used to address other research issues in bargaining and information processing. Similar to Brucks' (1988) article on Search Monitor, the purpose of this paper is to suggest research problems for which Deal Search is appropriate and to briefly describe its features and options.

This paper is divided into four sections. First background is provided on an information processing approach to bargaining. The next two sections give the essential characteristics of Deal Search and summarize how the program operates. The fourth section discusses research issues that might be addressed using Deal Search or similar software.


An Information Approach to Bargaining

When attributes are bargainable, additional tasks confront the consumer: the consumer must interact with the seller and mutually define the attribute values for the product. This means buyers in a bargaining situation evaluate the implications of sellers' offers relative to potential final agreement levels. And they must also formulate messages and counter-offers that will induce the sellers to make favorable concessions.

The increased cognitive load associated with bargaining changes information search and decision-making processes in purchase situations. For example, when bargaining is required, a buyer may curtail information search in order to devote more cognitive effort to the bargaining task. The Brucks and Schurr (1990) article proposes that such changes in information search and decision-making processes have not been examined in the bargaining literature. Even in marketing, this important aspect of buying behavior -- relevant to house and ear buying, for example -- has received little attention. One reason for this lack of attention, we believe, is that until recently adequate methods were not available to study bargaining from an information processing approach.

Adding Bargaining to the Information Search Paradigm

When product attributes are not bargainable, decision making requires finding out the value of attributes for the various alternatives. It also requires evaluating those values relative to some absolute criteria or the other decision alternatives. Thus, the consumer information processing literature (cf. Bettman 1979) studies the multi-attribute brand choice problem as follows. Subjects are presented with a choice of several brands, each characterized by values on several attributes. Data are collected on the information search and/or evaluation process by monitoring subjects' acquisition of attribute information and/or thoughts produced during choice.

Deal Search overlays the possibility of bargaining on this basic paradigm. For example, as in the purchase of a house or a computer for a small business, a subject might begin by gathering attribute information about different alternatives. At some point, however. the subject may choose to bargain with the seller to obtain desired values for particular attributes. In Deal Search, information search when attributes are bargainable is analogous to information search in non-bargaining tasks, except that the attribute values that are acquired are subject to change through an exchange of offers and counter-offers regarding specific attributes.


Deal Search itself has roots in a prior information search study (Brucks 1985) and a multiple-opponent bargaining study (Schurr and Ozanne 1985). These prior studies provide the rationale for using a computer-controlled task in consumer behavior studies (see also Brucks 1990). The following discussion focuses on the specific characteristics of Deal Search.

Degree of Customization

Deal Search can be modified to adapt a number of design characteristics to the needs of the research: the number and type of alternatives (i.e., dealers, stores, or brands), the number and type of attributes, the bargainability of the attributes (bargainable or non-bargainable), the values of attributes, and the pattern of counter-offers in the bargaining mode. (In addition, the researcher can design both pre-test and post-test questionnaires that are administered by the computer.) This adaptability applies to both the warm-up familiarization task and the main experimental task. As operationalized in the Brucks and Schurr (1990) article, subjects purchased a business computer from one of 6 dealers. Twelve product attributes described the computers. The following discussion refers to this version of Deal Search.

Sequential Decisions

There are two fundamental decisions relating to alternatives and attributes. Essentially, a subject first must choose which store to visit. Then the subject must decide which product attribute(s) to ask (or bargain) about while at the store. The stores are the alternatives, and could be relabeled as brands if the researcher so desired. The following menus indicate the options that are related to the basic store-choice and attribute-choice decisions.

The menu for choosing a store to visit. The focal buying task requires a series of sequential decisions, and these decisions vary according to where the subject is in the search process. First, the subject chooses which store to visit. Then the computer provides a waiting-time graphic to denote travel time to the store.

The menu for making an initial inquiry at a store. On the first visit to a store the choices include: asking about attributes, purchasing the product without further ado, or backtracking so that a different store can be visited. This latter option is necessary to allow for times when a subject goes to a store and then discovers that s/he intended to visit a different store.

Choosing an attribute to inquire about. If the subject opts for asking about attributes, the subject must type in the name of the attribute. In the Brucks and Schurr (1990) study the subjects knew about the attribute names by reading about them in a pre- task instruction set.

The menu accompanying attribute value information. Upon receiving attribute information, the choices become: ask about a different attribute, go to a different store, or purchase the product. The latter option recognizes that a consumer may elect to make a purchase at any point in the search process.

In the bargaining mode this menu also includes the option to bargain about the attribute. The option to bargain also appears in the preceding menu, but only if a subject has already asked about the store's offering (i.e., attribute values). (Generally speaking it would be unnatural for a person to walk into a store and launch into bargaining without first asking some questions aimed at revealing a seller's initial offer. However, in non-retail settings the seller may or may not make the first offer; Deal Search would require modification in this ease.)

The bargaining option. If the subject elects to bargain, the following screen requests the subject's counter-offer in a fill-in-the-blank format. Accompanying the seller's response are the same options as in the preceding menu plus the option to continue bargaining about the same attribute.

As the preceding details indicate, the menu options are tailored to a particular step in the subjects' information search task and to whether the attributes are bargainable or not. Because of this sequential structure, of course, the key-stroke-and-time sequential trace of a subject's decisions provides appropriate data for information search analyses

Methods for Requesting Information and Making Counter-Proposals

Requesting attribute information. In Deal Search subjects make inquiries about a particular attribute by typing in the name of that attribute. This approach has both advantages and disadvantages (also see Brucks 1988). One advantage of this approach is that people normally have to bring to mind which attributes they want to ask about and then communicate their choices. Another advantage of this approach is that it offers the researcher flexibility in the way attribute information is presented to subjects. For example, in Brucks and Schurr (1990) the knowledge manipulation was incorporated in the hard-copy attribute descriptions Another consideration is that the format of hard-copy attribute information can be varied, recognizing the possibility that information presentation may affect consumer search. To avoid order effects in attribute listings, order may be varied in the different versions of the instruction sets.

The primary disadvantage of forcing a subject to type in an attribute name is that it imposes a cost or search that may vary according to the typing skills of the subject. This problem is minimized by keeping keywords short While Deal Search can handle some variations of the attribute keywords, typing errors evoke a message indicating that the input is not recognized.

Making counter-proposals. Subjects make counter-proposals to a seller's offers by stating an attribute value for the attribute selected counter-proposals can be made for only one attribute at a time. On one hand, this one-at-a-lime feature preserves the sequential nature of a subject's decisions; this facilitates the monitoring of information search (e.g, Brucks and Schurr 1990) On the other hand, it does remove the possibility of log-rolling through the simultaneous presentation of counter-proposals on, say, three attributes at one time. Log-rolling refers to giving concessions on one attribute so that gains can be obtained on an alternative attribute (Pruitt 1981; for a paradigm that incorporales computer-controlled log-rolling see Schurr and Ozanne 1985)

Allowing subjects to enter any attribute value they choose for their counter-proposals creates some challenges for the algorithm that determines the computerized responses. However, this characteristic is in some respects a significant advance over the widely used Kelly (1966) paradigm (e.g., Graham et at 1988; Schurr 1987) because the Kelly paradigm reveals the feasible range of attribute value offers; Deal Search does not require that subjects know the researcher's pre-conceived range of bargainable attribute values Thus, a primary advantage of this approach is that the range Of possible responses is not communicated to subjects unless the researcher so chooses This means researchers can study the effect that a subject's attribute-value knowledge has on counter-proposals that are made. Another possible advantage not exploited in the Brucks and Schurr (1(990) study is that more latitude is provided for the study of how consumers formulate a series of counter-proposals in a bargaining situation

Search Costs

As discussed in Brucks (1988), the major advantage of building wailing time into a computer-controlled experiment is that it parallels the cost of search in actual purchase situations In Deal Search, waiting time can be adjusted for any of the menus or computerized responses One or the more important search cost variables is the time it lakes to go from store to store. When the cost of traveling to alternative stores increases, one might expect consumers to increasingly focus on bargaining rather than search since bargaining may achieve more favorable outcomes without traveling to another store.

Bargaining Interaction

An important characteristic that distinguishes Deal Search from other computer-controlled approaches to the study of bargainable attributes (e.g., Clopton 1984; Schurr and Ozanne 1985) is the strategy employed for developing computerized responses to a subject's counter-proposals Each rule governing the bargaining process was designed to be consistent with certain assumptions about a realistic bargaining process The following discussion presents these rules and explains the underlying assumptions and rationale for the rules.

1. In non-bargaining and bargaining conditions dealers make the initial offers that are predetermined by the researcher. In the bargaining condition, concessions are then determined in response to a subject's counter-offer.

This rule fits selling situations, particularly retail situations, whore the product has an "asking" price. An advantage of this rule is that it keeps non-bargaining and bargaining tasks identical until the subject in the bargainable attribute condition chooses to bargain. However, in bargaining situations opening offers commonly are different from agreement points. Yet Deal Search equates a seller's opening offer in a bargaining situation with a seller's best and only offer in a non-bargaining situation This may mean that an attribute value that a subject perceives as a reasonable opening offer in a bargaining situation (say, a list price of a house in a buyer's market) will be perceived as less than reasonable in a non-bargaining situation (see Brucks and Schurr 1990 for a discussion of this point)

2. If a subject's counter-offer is poorer than (less attractive to the seller) or the same as that subject's most recent offer, then the dealer will not make a concession. A dealer will never retract an offer once it is made.

This rule assumes that in general bargainers do not backtrack on their previous offers, and a seller will not reward such backtracking with a concession. Exceptions to this norm are possible in real life. of course

3. However, if a subject persists in making offers that do not represent a concession, the dealer will occasionally respond with a small concession.

Osgood (1959) introduced a widely discussed notion of initiating reciprocity by making a small concession in a stand-off situation. Deal Search casts the seller in the position of trying to stimulate a give-and-take exchange. However, in a laboratory setting subjects will try to "game" the task if they discover that the simple rule "be persistent" yields consistent rewards in bargaining. Therefore, Deal Search is designed such that persistence by itself is an inefficient strategy for reaching favorable agreements In fact, the researcher can control when and how the programmed seller responds to a subject's persistence -- even retracting a previous concession, if this makes sense in the research context.

4. If a subject's current offer represents a concession when compared to the subject's most recent offer, then the dealer will make a concession. The computer determines an appropriate concession by first matching the subject's offer to one of a number of predetermined offer ranks for a particular dealer. An offer rank is a subset of attribute values defined by an upper and lower bound between which a buyer's offer to a seller may fall. Each offer rank is associated with a specific concession size for the dealer.

For example, suppose a subject asks a dealer about the attribute "free software," and the dealer replies by offering free software valued at 5100 Then suppose the subject counters with a higher figure: free software valued at $1200 The Deal Search algorithm first examines the attribute's offer ranks in order to classify the subject's proposal for the value of free software:


Deal Search determines that the subject's $1200 proposal falls within offer Rank 6 (bounded by $1150 and $1299) and determines that a concession of $200 is called for. Consequently the dealer would add $200 worth of free software to the initial offer of $100, resulting in a counter-proposal of free software valued at $300.

Rule 4 utilizes Osgood's 1959 notion of graduated reciprocity, which says that an opponent's concessions should be reciprocated in order to facilitate agreement To implement this reciprocity, the Deal Search algorithm requires that the researcher predetermine dealer responses by creating offer rank ranges and corresponding dealer concessions Each dealer can have different initial offers (c g, sec Table 1 in Brucks and Schurr 1990). Because a dealer's concessions are added to a dealer's initial offer, each dealer's attribute value offers varies accordingly.

5. Concessions for some attributes (i c, where feasible) are smaller as a dealer approaches the preset settlement point for that dealer.

This rule assumes Siegal and Fouraker's (1960) notion that bargainers signal that their settlement point is approaching by making smaller concessions. By incorporating this rule into Deal Search we approach the goal of making a dealer's pattern of offers informative to the alert subject The implementation of this rule can be observed by inspecting the Dealer's Concession column in the preceding example Note that as a subject's offers fall into progressively lower offer ranks, the dealer's concessions decrease.

Also, if a subject makes a second offer that falls within the same offer rank, the dealer responds with the smallest concession the subject could make -- $50 in the case of free software. Subsequent concessions are treated in the same way as repetitive offers (Rule 3) Note that a subject whose initial counter-proposal corresponds to a lower offer rank may actually obtain a less favorable final settlement point with a dealer than will a subject whose initial offer falls into a higher offer rank. This aspect of Deal Search bargaining agrees with observed patterns of bargaining outcomes (Siegal and Fouraker 1960)

6. Agreement is reached when a subject agrees to a dealer's standing offer (either the initial offer or an offer that has resulted from one or more concessions) or when a dealer makes a concession that results in agreement with a subject's most recent offer.

This rule assures that the dealer will stop making concessions when a point of agreement has been reached.

7. A dealer never agrees to a final settlement less favorable to the dealer than a preset best offer

This rule prevents persistent subjects from obtaining unrealistically favorable deals. Also, note that by manipulating the best offer that is made by a dealer, the researcher can create a task in which some alternatives (i e, dealers) dominate other alternatives By observing the efficiency and effectiveness that subjects exhibit in terms of identifying the dominant alternatives, a researcher can draw some conclusions about the effectiveness of alternative patterns of bargaining or information search.


From the subject's standpoint, Deal Search is easy to use Deal search tells subjects when to read instructions contained on separate pages, solicits responses to pretest and post-test questionnaires, and manages the menu-driven search task as well as keyword and attribute value input Creation of the keystroke trace, which is the researcher's raw data, is invisible to subjects

From the researcher's standpoint, Deal Search is not especially user friendly While Deal Search incorporales debugging features, structured data entry for many parameters, and special operator-only screens, understanding of the source program is necessary in order to use Deal Search. Adapting Deal Search to the research project at hand and transforming data traces to data bases suitable for statistical analysis requires knowledge of BASIC and competent programming skills.

Input Required

Deal Search requires information for a variety of parameters, including: store names, attribute keywords, units that describe how the attribute values are measured, the number of offer ranks for each attribute, concession range information for each rank, concessions for each offer rank, and additional parameters that control certain aspects of bargaining. A variety of waiting time parameters can also be adjusted.

Program Requirements and Documentation

Deal Search currently utilizes Zenith's Microsoft GW-BASIC (Version 2) Currently it operates on computers with 512K of memory and one disk drive Deal Search contains documentation in the programming statements. Some additional documentation is available from the authors.


We believe that there is considerable merit in approaching bargaining as an information search and evaluation task Such an approach reveals new issues for study that have received very little attention. Because this approach to bargaining is so new, the value of Deal Search probably lies more with the issues a researcher might address, rather than with the program itself Three general research streams appear appropriate (l) comparing bargainable to non-bargainable purchases, (2) examining the factors that affect the trade-off between bargaining and information search, and (3) examining the factors that produce various bargaining strategies. In this concluding section we enumerate a few starting points for- such research

1. Bargaining and Decision Frames

Starting with the notion that a major task of bargainers is to process the great variety of information pertaining to a negotiation, it is interesting to consider the effects of different frames of reference that might influence how a bargainer interprets information For example, it was found in one negotiation study concerned with risky decisions (Schurr 1987) that the very same pay-offs stated in terms Of "the chances of obtaining ne profits" (i.e., potential gains, a positive frame) instead of "the chances of incurring expenses" (i e, potential reduced losses, a negative frame) caused negotiators to make less risky bargaining agreements. Besides affecting outcomes, risk aversion may also affect (1) the trade-off between bargaining and information search and (2) bargaining strategy.

2. The Perceived Cost of Information Search and Bargaining

Search costs have received almost no attention in the bargaining literature More research is needed on how time pressure and time-costs of information affect consumer behavior in connection with both bargainable and non-bargainable attributes. It has become conventional wisdom that when one bargainer is subject 10 time pressure and his or her opponent is not, the opponent has an advantage Yet we know little about patterns of information search and bargaining or about the effectiveness of these efforts in connection with time costs.

3. Perceived Benefits of Information Search and Bargaining

In the study reported in Brucks and Schurr (1990) subjects varied considerably in the degree to which they engaged in bargaining to achieve a favorable outcome. While much research has been carried out regarding individual differences in amount Of information search (see Moore and Lehmann 1980), little has been published in the consumer literature on individual differences in the amount of bargaining, Motivational factors, such as involvement and past enjoyment, should be examined.

4. Attribute Level Knowledge

Attribute level knowledge emerged as an important factor in the Brucks and Schurr (1990) study, which suggested that bargainers tended to discount the value of initial offer information and utilize bargaining influence to improve their outcomes. Future research can explore the effects of knowledge that gives greater meaning to initial offers For example, industry norms or other sources of expectations would influence the ability or a consumer to utilize initial offers to screen dealers by means of initial offers. Alternatively, bargainers may be most interested in knowing the feasible zone of agreement and indifference points to use in their bargaining efforts.

5. Reputational Knowledge

Reputational knowledge is another factor that affects a consumer's ability to form evaluations about dealers For example, dealers in the Brucks and Schurr (1990) study had no established reputation, which may be realistic in some, but not all, settings Reputations take different forms. Reputations for quality or service reliability would affect expectations for product attributes levels. In contrast, reputation for fairness and honesty is more likely to influence the process of bargaining, rather than information search (cf., Schurr and Ozanne l 985).

In addition to the effects of reputational knowledge, interactions between reputation and product attribute information deserves consideration. For example, an interesting research issue concerns the affects of disconfirming information on search patterns and choice behavior. Disconfirming information may cause consumers to discount reputational knowledge, leading to increased information search

6. Bargaining Strategies

Bargaining processes reflect a bargaining strategy or an absence of one While in the Brucks and Schurr (1990) study a bargaining strategy was suggested to subjects in order to reduce variance from different strategies, bargaining strategy itself warrants attention as a research variable. One might examine the buyer's bargaining strategy as a function of the different bargaining strategies Of the different sellers.


Our purpose in this paper has been to (I) suggest an information processing approach to bargaining, (2) provide details on how Deal Search could be used to conduct a computer-controlled experiment in this area, and (3) propose some starting points for future research the Deal Search software is available from either author.


Bettman, James R. (1979), An Information Processing Theory of Consumer Choice, Reading, MA: Addison-Wesley

Brucks, Merrie (1985), "The Effects Of Product Class Knowledge on Information Search Behavior," Journal of Consumer Research, 12 (June), 1-16.

Brucks, Merrie (1988), "Search Monitor: An Approach for Computer-Controlled Experiments Involving Consumer Information Search, Journal of Consumer Research, 15 (June), 117-121.

Brucks, Merrie (1990), "Computer-Controlled Experimentation in Consumer Decision Making and Judgment," in Marvin E. Goldberg, Gerald J Gorn, and Richard Pollay, eds Advances in Consumer Research, Vol. 17, in press

Brucks, Merrie and Paul H Schurr (1990), "The Effects of Bargainable Attributes and Attribute Range Knowledge on Consumer Choice Processes," Journal of Consumer Research, 16 (March), 409-419.

Clopton, Stephen W. (1984), "Seller and Buying Firm Factors Affecting Industrial buyers' Negotiation behavior and Outcomes," Journal of Marketing Research, 21 (February), 39-53

Graham, John L., Dong Ki Kim, Chi-Yuan Lin, and Michael Robinson (1988), "Buyer-Seller Negotiations Around the Pacific Rim: Differences in Fundamental Exchange Processes," Journal of Consumer Research, 15 (June), 48-54

Kelley, Harold H (1966), "A Classroom Study of the Dilemmas in Interpersonal Negotiations in Strategic Interaction and Conflict, ed. Kathleen Archibald, Berkeley, CA Institute Of International Studies, University Of California 49-73.

Moore, William L. and Donald K Lehmann (1980), "Individual Difference in Search Behavior for a Nondurable," Journal of Consumer Research, 7 (December), 296-307.

Osgood, Charles E. (1959), "Suggestions for Winning the Real War with Communism," Journal of Conflict Resolution, 3 (4), 295-325

Pruitt, Dean G. (1981), Negotiation Behavior, New York: Academic Press

Schurr, Paul H. (1987), 'The Effects Of Gain and Loss Decision Frames on Risky Purchase Negotiations," Journal of Applied Psychology, 72 (August), 351-358

Schurr, Paul H. and Julie L. Ozanne (1985), "Influences on Exchange Processes:Buyers' Preconceptions of a Seller's Trustworthiness and Bargaining Toughness," Journal of Consumer Research, 11 (March), 939-53.

Siegel, Sidney and Lawrence E. Fouraker (1960), Bargaining and Croup Decision Making, New York: McGraw-Hill.