Abstract
Much attention has been devoted to the “first offer” in negotiation research. Rightly so, as strong empirical evidence shows that the first offer has a significant impact on the negotiated outcome and, therefore, is a highly relevant topic for negotiation scholars and practitioners. Scholars typically recommend making the first offer. However, in the field, we have observed an alternative opening tactic—asking for the best price that the counterpart is willing to accept. This question represents a real alternative to making the first offer by initiating the discussion of specific settlement proposals, provided the counterpart answers the query. Does it, however, lead the other side to make a better offer? How does the question impact the economic and relational outcomes of the negotiation? Is it advisable to use this tactic in negotiations? We investigated these questions based on a controlled laboratory experiment, in which 227 dyads of cellphone buyers and sellers negotiated synchronously via a text chat. We found that the best‐price question has an impact on not only the first offer but also the negotiation outcome. When the buyers in our experiment asked the question, the results were not significantly different than those from negotiations in which they made the first offer. This effect was driven by the first offer in response to the question. Additionally, we found that the best‐price question did not negatively impact the relational outcome. Moreover, the effect was reduced when list price information was available. These findings suggest rethinking the traditional view of the offer‐counteroffer sequence and provide an alternative opening tactic to making the first offer in the context of high information asymmetry.
Introduction
The first offer has received much attention in research on negotiations (Ochs and Roth 1989; Kristensen and Gärling 2000; Galinsky and Mussweiler 2001). Lipp, Smolinski, and Kesting (2023: 2) defined the first offer as “the first settlement proposal put forward by either negotiating party”; in sales negotiations, for example, this is typically a price. The first offer is a critical moment in negotiations, and we refer to this moment as the “opening,” not in the sense of the actual beginning of negotiations, but the beginning of the exchange and discussion of concrete proposals for an agreement. In the phase model of Lewicki, Saunders, and Barry (2015), this phase would correspond to the “bidding” phase that follows the phases of preparation, relationship building, information gathering, and information using.
Researchers have identified a strong correlation between the first offer and the outcome of a negotiation. In a meta‐study of 16 research papers, Orr and Guthrie (2006) reported a correlation of nearly 0.5 between the first offer and the negotiation outcome. The positive correlation between the first offer and a negotiation's economic outcomes is often explained by the anchoring effect (Tversky and Kahneman 1974). The anchoring effect describes the relationship between a mentioned numeric value and a subsequent decision; Tversky and Kahneman (1974) demonstrated this effect when participants spun a wheel of fortune and then answered estimation questions. The estimation question was influenced, or anchored, by the random number shown on the wheel of fortune. This anchoring effect has prompted researchers to suggest that it can be leveraged by negotiators. However, research also has shown that first offers can be counterproductive if the “right” amount is not proposed (Wang, Zhang, and Han 2008). First offers can damage relationships or lead to an impasse if they are too extreme; if not ambitious enough, they can lead to an inferior economic outcome. Additionally, they can provide unwanted information disclosure (Loschelder et al. 2014). These findings led to Thompson (2001: 68) pie‐slicing strategies for first offers, which emerged as a standard in negotiation research and teaching:
“Make the first offer (if you are prepared).”
“Immediately re‐anchor if the other party opens first.”
These strategies reflect the general recommendation to make a first offer. The counter‐anchor is intended to mitigate the effect of the first offer, if the first offer cannot be made. If there is a risk of making a wrong first offer, it could be advantageous to wait for the other party to begin, then immediately follow up with a counteroffer. However, there are few negotiation experts who advise waiting for the first offer rather than making an offer first (e.g., Voss and Raz 2016).
There is, however, an alternative opening tactic that has received almost no attention in the literature: posing a question about the best price of the counterpart (best‐price question). Asking this question can be understood as an opening because it initiates the exchange and discussion of concrete settlement proposals. However, this question is not about just asking for an offer; the responder is asked for the best offer (price) they can make. What effect does this question have on the first offer and, thus, on this critical point of negotiations? Do negotiators truly disclose their best price or, at least, better prices than they would give as their first offers if they were not being asked this question? How does this affect the course of a negotiation and its outcome? The tactic of asking for the best price can also be considered quite offensive, as it asks for the seller's reservation point—sensitive information that is not advisable for a seller to disclose. So, how does this question affect trust and the relational outcome of negotiations? As far as we know, research has not tackled these questions yet.
We consider examining the best‐price question to be relevant for at least three reasons. First, the tactic represents a logical alternative to the offer‐counteroffer sequence and should be considered as such, especially at such a critical point of negotiations. Second, this question is frequently used in the field. A search on Google delivers 6,510,000,000 results for the query “what is your best price,” and there are multiple posts and forums on this question on the Internet (e.g., Butel 2018; Powell 2019). However, these contributions and forums are based less on systematic research and more on practical experience. In addition, we conducted a pre‐study,1 finding that 149 (81.4 percent) of 183 adults in the United States already have encountered the best‐price question in a second‐hand setting (e.g., Craigslist.com). Against this background, to say that this question is a relevant element of negotiations is not an exaggeration. Third, we know the fundamental importance of questions in and beyond negotiations. For example, Minson et al. (2018) showed that assumptions conveyed through questions could lead to greater disclosure of information from the counterpart than otherwise possible. Their findings and others suggest that the best‐price question could impact the course of negotiations. These are reasons enough to consider taking a closer look worthwhile.
The subject of this empirical study is to examine how the best‐price question affects the first offer and the outcome (economic and relational) of a negotiation. The opening tactic analyzed in this article, in its generic form, asks about the best offer of the responder and can be used in various types of negotiations. In our experimental setup, we chose to investigate the effects of the opening in the context of a distributive negotiation, in which the tactic takes on a specific form of a question about the best price. Distributive negotiation is also often referred to as a zero‐sum situation in which parties attempt to get a larger share of a fixed resource (Lewicki, Saunders, and Barry 2015). We conducted a controlled laboratory experiment in which we manipulated the negotiation opening (the buyer's first offer, seller's first offer, and seller's first offer in response to the “best price” question) and the amount of additional information available (only market price range, additional advertised price). The scenario consisted of a buyer and a seller negotiating the price of a mobile phone being sold on a classified ads website. The negotiation was conducted online via chat by 227 dyads.
First Offers and Questions in Negotiations
Role of the First Offer
The first offer in a negotiation has been extensively studied in negotiation research with a broad consensus that the size of the first offer influences the outcome of negotiations. This first‐offer effect has been shown in multiple studies in different analytical, experimental, and field settings (Sadanand 1996; Kristensen and Gärling 2000; Chi, Friedman, and Shih 2013; Yao, Ma, and Zhang 2018; Davis and Hyndman 2019).
The mechanism behind the first‐offer effect was initially attributed to the anchoring‐and‐adjustment heuristic (Tversky and Kahneman 1974), which stated that negotiators do not sufficiently adjust from the initially mentioned value (anchor). However, this explanation was later extended using the selective accessibility perspective, which argued that both the accessibility and applicability of a reference value (anchor) influence the decision‐making process (Strack and Mussweiler 1997; Chapman and Johnson 1999; Mussweiler and Strack 1999, 2001). Furnham and Boo (2011) posited that the selective accessibility theory is the currently dominant explanation for the anchoring effect of first offers.
Additionally, the first offer can encompass more than a mere value. Studies also have demonstrated that there are different types of first offers and that the surrounding circumstances also influence behaviors and outcomes. For example, precise first offers function as stronger anchors than rounded offers (Mason et al. 2013; Loschelder, Friese, and Trötschel 2017). Furthermore, the impression of the counterpart can change the size of the first offer. For example, negotiators that are perceived to be wealthy tend to receive more unfavorable first offers (Maaravi and Hameiri 2019). Finally, a strategic reaction to a first offer (e.g., a flinch) can yield better results than no reaction (Fassina and Whyte 2014).
Role of Asking Questions
Our study investigates the impact of a specific question—the best‐price question. However, to better understand the context and derive our hypotheses, we start our analysis with selected aspects of asking questions in general and in negotiations specifically.
A question can be defined as “a sentence, phrase, or word that asks for information or is used to test someone's knowledge” (Merriam Webster 2011). As negotiations are typically characterized by high information asymmetry, asking questions can play a critical role in negotiations. Accordingly, the importance of questions for negotiations has already been addressed by research, for example, in connection with not only (1) active listening and open‐ended questions (Royce 2005) and (2) interests and priorities (Thompson 1991; Neale and Bazerman 1992) but also (3) the critical questioning of claims (Fisher, Ury, and Patton 2011). However, there are still important research gaps in the field (Miles 2013), and the best‐price question is one that requires more investigation.
According to Athanasiadou (1991), questions can serve four functions: request information (e.g., What do you know about him?), provide information (via rhetorical questions, e.g., Do you know it is 12:30?), examine the knowledge of others (e.g., How many syllables does the word “dog” contain?), and request activity indirectly (e.g., Would you mind not making so much noise?). The best‐price question can be considered a request for information.
In addition, questions can influence behavior inadvertently (i.e., even if the enquirer did not intend that effect). For example, Williams, Block, and Fitzsimons (2006) found that asking people about their likelihood to exercise or do illegal drugs in the next two months increases the likeliness of these behaviors happening, even if they are non‐normative (i.e., drug use). Minson et al. (2018) summarized two behavioral effects of questions. On the one hand, the authors describe how the conveyance of information affects responses. An interesting example is the study of Loftus and Palmer (1974), who found that when people were asked, “About how fast were the cars going when they smashed into each other?,” their estimates were higher than when verbs like “collided, bumped, contacted, or hit” were used instead. On the other hand, Minson et al. (2018) demonstrate that a question can communicate information about the appropriateness of behaviors. These behavioral effects suggest that the best‐price question could have effects beyond the mere information request.
Asking Sensitive Questions
The best‐price question stands out from other questions because it is a request for sensitive information. Hart, VanEpps, and Schweitzer (2021: 1) define sensitive questions as “questions about topics that are uncomfortable to discuss, inappropriate for the social context, or about information individuals would rather keep private.”
The best‐price question refers to the reservation price (also known as the resistance point), that is, the highest or lowest price the negotiator is willing to accept. Knowing the reservation price of our negotiation partner is sensitive because it can be used strategically to maximize the opponent's share of the zone of possible agreement. Therefore, a negotiator typically wants to hide their reservation point (Lewicki and Stark 1996).
The best‐price question presents the addressee with the dilemma of either having to refuse to answer, lie, or give out sensitive information that the other side should not have. All these alternatives are problematic in different ways and can lead to tensions or even stress for the party being asked. Refusing to answer is impolite, lying unethical, and disclosing the reservation price leads to a strategic disadvantage.
Minson et al. (2018) investigated a comparable dilemma and demonstrated that a question that carries a negative assumption (vs. a positive or neutral assumption) leads to greater disclosure of undesirable information. The authors also found support for two mechanisms of this effect; both the inferred assertiveness and knowledgeability of the enquirer led to increased disclosure. This disclosure effect has also been demonstrated in earnings conference calls (Haag et al. 2019).
Even though the best‐price question can be considered neutral (as per Minson et al. 2018), one could also argue that it carries an assumption that there is a price better than what would otherwise be communicated. Otherwise, the negotiator could simply ask for the price only. Hence, the best‐price question could also be interpreted as a request to make a better price or an initial concession.
Notably, people generally avoid asking sensitive questions as they fear the assumed negative effects, even though these effects often do not occur (Hart, VanEpps, and Schweitzer 2021). Thus, using sensitive questions in negotiations could be a powerful—but underused—strategy.
Responses to Sensitive Questions
The immediate reaction to a sensitive question is a response. Bitterly and Schweitzer (2020) described six types of responses to direct, difficult questions: (1) deflection, (2) dodging, (3) paltering, (4) declining to respond, (5) lie of commission, and (6) honest disclosure. Deflection is defined as “responding to a direct question with a new question that shifts the conversation to another person” (Bitterly and Schweitzer 2020: 945). “Dodging” is evading a question by answering a different question (Rogers and Norton 2011). Finally, paltering is the “active use of truthful statements to create a false impression” (Rogers et al. 2017: 456).
In addition to the above responses, a question can be answered with either a truthful statement, a lie by commission (an outright false statement), or a “lie by omission,” which is defined in the literature on deception as “withholding information with an intention to deceive” (Mahon 2008). However, if a direct question is asked, no lie of omission is possible; thus, if a responder lies, the lie would be one of commission.
Very few studies have investigated (1) what one could expect if a direct question is asked and (2) how often the response types described above are used to answer sensitive questions.
Ethics of Asking Questions
Our study also has an ethical dimension. Asking a negotiation partner for information that they might not want to disclose could be considered ethically problematic. Lewicki and Robinson (1998) researched what is considered ethical in negotiations. They found that a misrepresentation of one's position and bluffing are considered ethical, whereas deception and falsification are considered unethical. Thus, one could argue that asking for information that a responder does not want to reveal invites deception (as the negotiator does not want to reveal their position) and is thus unethical. Furthermore, experienced negotiators could capitalize on this tactic to elicit concessions from inexperienced negotiators who tend to answer questions honestly.
Strudler (1995) used the specific example of lying about one's bottom line and evaluated the ethicality of the lie from both the “right to self‐defense” and the “mutual advantage” perspectives. These two perspectives led to differing conclusions. From the self‐defense perspective, lying about the bottom line would be allowed as a means of self‐protection. However, from the mutual advantage view, which requires a tactic that produces an advantage for each party, this deception would cause a one‐sided advantage. Strudler advocates the mutual advantage perspective. In this perspective, lying about the bottom line would not be ethical.
When discussing the ethics of asking questions about the bottom line, a legal perspective is also of interest. Shell (1991) makes clear that bottom‐line values or other “estimates of price or value” are not material facts, and only a misrepresentation of the material facts is illegal. However, Shell also argues that there is a thin line. Making up other offers to make someone pay more already has been ruled as fraud.
Expected Effects of the “Best‐Price” Question
As shown above, questions can have powerful effects and could even lead to the disclosure of unfavorable information. Therefore, we expected that opening a negotiation with a best‐price question could influence the course of the negotiation. This section describes the study's theoretical background and our hypotheses.
In the standard paradigm, a negotiation can be opened in two ways: either (1) the buyer makes a first offer, or (2) the seller makes a first offer. If we add a tactic of asking for the best price, two additional openings need to be considered; either (3) the buyer makes a first offer after being asked for the “best price,” or (4) the seller makes a first offer after being asked for the “best price.”
In our study, we focused on the effects that the “best‐price” question has on the seller (4). We made this decision because it is typically the buyer who asks about the seller's best (lowest) price. We compared this opening with a “normal” buyer and a seller's first offer (1 and 2). To reduce the complexity of this first study, we decided not to include the alternative (3), for which the opposite direction of the question would be required (asking the buyer for the maximum price that the buyer is willing to accept).
We theorized that one key aspect of the best‐price question is its impact on the resulting first offer from the counterpart. Thus, we started the investigation with the first offer and then proceeded to investigate the outcomes.
From our point of view, there are four main ways to answer the best‐price question asked at the beginning of a negotiation—and these answers imply a lower first offer. First, although answering the best‐price question truthfully is disadvantageous, some individuals may still do so because of a general aversion toward deception. Lying is not the default mode for answering questions, and negotiators generally try to avoid it (Minson et al. 2018).
Additionally, lying leads to distress, and people tend to use minimization strategies to reduce distress, such as denying that they have lied (Aquino and Becker 2005). This explanation is in line with the truth default perspective of Verschuere et al. (2011), who state that lying involves an intentional suppression of the predominant truth response. Gneezy (2005) empirically confirmed that there is an aversion to lying in a decision‐making context. In Gneezy's experiment, a player 1 had to choose whether she communicated to a player 2 the truth regarding the payoffs of two options of the game. Subsequently, player 2 had to decide which option to take. A truthful communication would put player 1 at a disadvantage. The deception percentage range of player 1 ranged from 17 percent to 52 percent, and the other players told the truth. The consequences for both players impacted the lying behavior in that if the stakes were higher for oneself, their lying increased. However, the impact on the other party was also relevant; lying increased if the consequences to the counterpart were less severe.
Gneezy's study relates well to our study: After the best‐price question is asked, the receiver must decide whether to answer it truthfully, and they have a certain expectation regarding the benefits. The aversion toward deception has been confirmed by Lundquist et al. (2009), who extended the findings of Gneezy (2005) by showing that aversion to lying increases with the size of the lie.
Second, the answer could be a lie. Since it is strategically disadvantageous to answer the best‐price question truthfully, it can be expected that many negotiators would answer it with an offer that is not their true reservation price. This assertion has some empirical backing. Lying seems to be a ubiquitous phenomenon in daily life; DePaulo et al. (1996) reported that about 15 percent of social interactions involve lies. Schweitzer and Croson (1999) studied the impact of direct questions and demonstrated that direct questions reduced the number of lies by omission but increased the number of lies by commission. However, they also demonstrated that 61 percent of responders revealed unfavorable information about the negotiation issue after being directly asked about any problems.
Furthermore, when confronted with the best‐price question, negotiators might initially think about their true reservation point, but once they realize that revealing it might not be in their best interest, they try to move away from it in their response. However, their adjustment is influenced by the value of their own reservation point. This mechanism would be based on the anchoring effect and causes the seller's initial offer to be lower than it would be if the question had not been asked. We refer to this mechanism as an “anchored lie.” Schaerer, Loschelder, and Swaab (2016) found that having multiple lower alternatives led to lower first offers as it increased the salience of the reference value and, accordingly, being nudged by the question could anchor the negotiator to a lower value than initially planned.
Third, the answering party could dodge yet answer the question with a better (but not the best) offer. The dodging answer is interesting because it helps the responder avoid the dilemma that must be faced when the question is interpreted literally. By interpreting the question as a request for a good/better price, negotiators can proceed without compromising their integrity or disclosing sensitive information.
Finally, there is a probability that the best‐price question will not be answered at all or answered with information that does not relate to the value of the negotiated object or issue. In such cases, it is unlikely that the question will impact the value of the first offer.
Thus, we predicted that the answer to the best‐price question would be significantly lower than a voluntary first offer by the sellers. Although we expected a material adjustment, the resulting first offer should still be above the first offer an average buyer would make. If a typical buyer sets their first offer below the seller's reservation price, the first offer in reply to the question should be higher than the buyer's first offer.
These theoretical considerations led us to the first hypothesis:
Hypothesis 1 (H1): If a buyer asks about the best (lowest) price, the resulting first offer from the seller is lower than the voluntary first offer of the seller and higher than the first offer from the buyer.
In line with the literature related to the anchoring effect, the elicited first offer is expected to induce the anchoring effect. Regardless of how the first offer is triggered, the first value provided should cause the same effects as that caused by the selective accessibility model (Strack and Mussweiler 1997) and predict the overall economic outcome (agreement price) of the negotiation. Therefore, we posit the following hypothesis:
Hypothesis 2 (H2): The first offer in reply to the “best‐price question” functions as an anchor and mediates the relation between the opening and the economic outcome of a negotiation. The outcome in the question treatment is thus lower than in the “seller first offer” and higher than in the “buyer first offer” treatment.
Moreover, we expected that the number of reference points would moderate the impact of the negotiation opening on the first offer. As H1 suggests that the question about the lowest price influences the first offer, we were interested in situations in which the question about the lowest price does not yield a different first offer. According to the anchoring perspective, the availability of additional reference points could be such a situation. Galinsky and Mussweiler (2001) showed that the anchoring effect can be reduced when the negotiator focuses on information that is inconsistent with the anchor. If the seller has included a list price in the product information, this list price could work as an inconsistent reference point that would reduce the anchoring effect. Moreover, Liebert et al. (1968) showed that incomplete information strengthened the first‐offer effect. Multiple other authors showed the effect of information availability on the first‐offer effect (Orr and Guthrie 2006; Maaravi and Levy 2017; Jiang and Ma 2019). Thus, reduced information asymmetry and information that conflicts with the other available information are expected to moderate the effect of the negotiation opening on the first offer. This led us to the third hypothesis.
Hypothesis 3 (H3): The presence of a reference point (list price) acts as a moderator and reduces the effect of the negotiation opening on the first offer.
In addition to the different first offers that depend on the negotiation opening, we also expected changes in seller behavior to be induced by the question.
As discussed above, a seller would start with a lower offer as either an honest disclosure of the reservation price, a dodge, or an (anchored) lie. Lewicki, Saunders, and Barry (2015) proposed that negotiators typically factor in some bargaining range, as they expect a give‐and‐take process in the negotiations that follow.
A low offer for any of the above reasons will already have reduced the bargaining range and the room for further concessions. First, an honest disclosure eliminates any room for further concessions and effectively ends the negotiation process. In reaction to an honest disclosure, buyers can only accept the first offer or walk away, both of which end the negotiation process. Second, a dodging response is based on an interpretation of the best‐price question as a request for a concession, and once an accommodating first offer has been made, it is perceived by the responder (seller) as her first concession. The concession reduces the distance that must be bridged to complete the negotiation process. Third, in the case of the (anchored) lie, responders (sellers) are subconsciously influenced by an implied reference to their reservation point (best price), which results in adjusted (lower) first offers. Although in such cases responders might not perceive their first offers as deliberate concessions, their effect is comparable to the effect of a dodging response. They shorten the distance between the parties and decrease the necessary magnitude of concessions in the remaining part of the negotiation process.
Therefore, sellers who answer the best‐price question will likely start with lower offers. Hence, the overall effect of the best‐price question will likely decrease the distance between the seller's first offer and the buyer's aspiration point, as well as the total magnitude of possible and required concessions. Thus, the next hypothesis is proposed.
Hypothesis 4 (H4): If asked for the “best price,” sellers will make a smaller total magnitude of concessions compared to the other treatments.
Negotiators typically want to hide their reservation point or best alternative to a negotiated agreement (BATNA) (Lewicki and Stark 1996). Thus, being asked for the best price creates a dilemma for a negotiator. On the one hand, negotiators do not want to disclose their best price; on the other hand, they tend to tell the truth (Verschuere and Shalvi 2014) and need to engage in neutralization strategies when lying (Aquino and Becker 2005). Combined with the potential of reduced trust, the negotiation could be perceived as less pleasant. Therefore, we expected that the negotiator receiving the best‐price question would have reduced levels of satisfaction regarding the negotiation. This led us to the following hypothesis.
Hypothesis 5 (H5): Answering the “best‐price” question leads to lower satisfaction with the negotiation by the seller when compared to a negotiation that proceeds with a “normal” first offer.
In addition to the above hypotheses, we expected to generate relevant descriptive data, such as the number of participants answering the question, the number of lies received in response to the question, and the extent of those lies.
Laboratory Experiment: The Sale of a Smartphone
We carried out a controlled study in the behavioral economics laboratory of a large university. In this experiment, a buyer and a seller of a smartphone negotiated its price via an online text chat.
Participants and Design
The experiment was designed as a 3 (negotiation opening: the buyer makes the first offer, the seller makes the first offer, and the seller makes the first offer after being asked a best‐price question) × 2 (presence of a list price: yes, no) factorial design. We employed self‐selection for the “opening offer” condition and did not manipulate who made the first offer. We only manipulated the treatment wherein the buyer asked for the lowest price. We decided to employ self‐selection to achieve a “natural” first offer rather than a required first offer, which also allowed us to observe the participants' most natural behavior.
Our hypotheses aimed to assess whether the best‐price question has an impact on a responding negotiator. Therefore, we could only investigate one member of a negotiation dyad; thus, the unit of investigation was the dyad. We estimated a priori required sample size for a two‐way ANOVA using the G*Power software (effect size 0.25, alpha error probability 0.05, power 0.80, 6 groups), resulting in a required sample size of 244. This number was also in line with a more general “rule of thumb” of 30 observations per cell (Van Voorhis and Morgan 2007); to account for potential dropouts and data quality issues, we aimed to recruit 40 dyads per experiment cell of our factorial design, for a total goal of 240 dyads (six experiment cells) or 480 participants. A total of 553 participants (289 female, 260 male, 3 diverse, and 1 other) participated in the study. The mean age was 24.3 years (SD = 4.0).
Task
We chose a common everyday item (a used smartphone) for the negotiation task so that the participants could relate to the situation. Our negotiation task was constructed similarly to that used in Minson et al. (2018, study 3). However, the information was reduced to a minimum. We also provided a market price for the used smartphone. According to the task description, the participants had to sell (seller) or buy (buyer) a used smartphone; to do so, they were required to negotiate its price with their counterpart. In the “best‐price question” treatment, we instructed the buyers to ask for the best (lowest) price at the beginning of the negotiation. The sellers were instructed to hold off on the price discussion until the buyers brought up the topic. In the low information asymmetry treatment, an additional list price was shown in the instructions. Detailed instructions are provided in Appendix One.
We incentivized the participants toward two goals without letting them know the value of the incentive and compensation. The first incentive was given for reaching a deal, while the second incentive was given based on the deal price (the lower/higher the price, the better, depending on the role). The base compensation for participation in the 11‐minute experiment was 3 EUR, and the incentive was up to 1 EUR.
Procedure and Materials
The experiment was conducted in English. We used oTree (Chen, Schonger, and Wickens 2016) to develop the experiment and ORSEE (Greiner 2015) to manage invitations and administration.
Before the start of the experiment, participants were invited to a short briefing session via videoconference to note their attendance and provide technical information regarding the payment for participation. Afterward, the participants started the experiment in their browsers and provided their informed consent.
If the participants agreed, they were taken to the experiment content, which began with the displayed instructions page (see Appendix One), followed by the actual negotiation page. The key information regarding the participant's role was shown above the chat window. In the chat window, the participants could exchange messages. Below the chat window was a text box where the participants could enter the deal value or indicate that they had not reached a deal. They could only proceed if their deal values (or “no deal” indications) matched. The negotiation was limited to 11 minutes (1 minute for orientation +10 minutes for negotiation).
After completing the negotiation, the participants answered the subjective value inventory (SVI) (Curhan, Elfenbein, and Xu 2006) and demographic data questionnaires. The SVI was used to obtain satisfaction with the negotiation as per H5 and can be considered the standard for evaluating a negotiator's subjective value/satisfaction across four dimensions (satisfaction with the process, the outcomes, the relationship to the counterpart, and satisfaction with oneself). Finally, the payment data were displayed for the participants.
Variables
The following variables were recorded for every participant. Measures like outcome and negotiation time can be considered measurements recorded at the dyad level, as they are the same for the buyer and the seller.
Reservation Price. The participants entered the reservation price (the minimum or maximum price depending on the role) on the instructions page before the start of negotiations.
Chat Transcript. The chat record of each participant was saved. We also used the chat record to obtain additional variables (see the next paragraph).
Subjective Value. The SVI was used to obtain the subjective value of the negotiation with its four subscales (Curhan, Elfenbein, and Xu 2006).
Demographic Variables. The demographic variables (age, gender, highest education, work experience, and occupation) were requested at the end of the experiment.
Further, we obtained several variables from the chat transcripts:
- 5.
First offer maker. As we employed self‐selection for the start of the negotiation, we manually coded who made the first offer into the cells “buyer first offer” and “seller first offer” based on the chat transcript.
- 6.
First offer. The first offer was the first number proposed by either party.
- 7.
Number of offers. The number of offers was coded as a distinct count of the numeric offers made; therefore, an offer was not counted as an offer when it was reiterated.
- 8.
Type of reply to the best‐price question. We coded the reply to the question on the best price with answers of five different categories: answer (if a numeric answer was given; in the case of a range, we took the average), answer_other (if another question was answered, e.g., the question “What is the list price?”), boomerang (if the seller asked for the highest price of the buyer as a response), deflect (if the seller tried to shift the conversation), and ignore (if the seller just ignored the question).
- 9.
Agreement (only for verification). This was coded to verify the agreement from the chat against the agreement the pair of participants entered in the system, which enabled us to include the dyads that came to an agreement shortly before the time‐out.
- 10.
Timing of the question about the lowest price. Some buyers did not ask the best‐price question at the beginning of the negotiation. This action was coded so these cases could be removed from the data later.
In addition, we calculated variables from the data described above:
- 11.
Total concessions. This variable was calculated as the difference between the first offer and the final agreement and thus reflected the total concessions made.
- 12.
First offer to reservation price. The distance between the first offer and the reservation price was computed to measure how much negotiation room the negotiators implemented.
We did not obtain any additional discourse data from the chat transcripts, as this was not required to test our hypotheses. The data are available in the Negotiation Data Repository and the OSF repository (see author note for the repository URLs).
Results
A total of 553 participants took part in the experiment. We removed 53 datasets, as these dyads did not reach an agreement (n = 46), did not have a counterpart (n = 2), or did not carry out a negotiation but quickly agreed to meet in the middle (n = 5). Next, we reduced the dataset to include the seller data only, as the seller received the question, and we are investigating dyadic data. Finally, we removed another 23 datasets from the resulting n = 250 dyadic data points, as these buyers did not comply with the instructions to ask the lowest price question in the beginning; instead, they asked the lowest price question when other values were already put forward. Hence, we were left with n = 227 data points. A detailed overview of the sample flow is provided in Table One.
Step . | Participants . | Dyads . | Comment . |
---|---|---|---|
All participants | 553 | ||
Removed participants | −53 | No agreement, collusion, no partner | |
Sub‐result | 500 | 250 | |
Removed dyads | −23 | First‐offer question asked too late | |
Final sample | 227 |
Step . | Participants . | Dyads . | Comment . |
---|---|---|---|
All participants | 553 | ||
Removed participants | −53 | No agreement, collusion, no partner | |
Sub‐result | 500 | 250 | |
Removed dyads | −23 | First‐offer question asked too late | |
Final sample | 227 |
Note: All participants included one case where no partner was found, and no negotiation happened, which explains the uneven number in the “All participants” row. The dataset was subsequently removed as part of the ‐53 datasets.
Even though our hypotheses were agnostic of the answer to the best‐price question, we started by examining the statistics based on the answer type. Most sellers (82.1 percent) who received and answered the best‐price question from the buyers mentioned a number in response to the question. Of those who answered, 15.6 percent responded with their true reservation price. The second most common strategy (11.5 percent) was the boomerang—reversing the question with questions like “what is your best price?” A smaller share of respondents used other tactics, including ignoring the question or answering another question altogether. The details can be found in Table Two.
Answer Type . | Frequency . | % . |
---|---|---|
Answer | 64 | 82.1% |
Thereof true answer = reservation price | 10 | 15.6% |
Boomerang | 9 | 11.5% |
Ignore | 3 | 3.8% |
Answer other question | 2 | 2.6% |
Answer Type . | Frequency . | % . |
---|---|---|
Answer | 64 | 82.1% |
Thereof true answer = reservation price | 10 | 15.6% |
Boomerang | 9 | 11.5% |
Ignore | 3 | 3.8% |
Answer other question | 2 | 2.6% |
Note: This table includes the cases in which the question was asked too late (n = 250; see Table One for the detailed participant flow). Although the question was asked too late, the reaction could be obtained.
For the following analyses, we only used those datasets wherein the best‐price question was asked at the beginning of the negotiation, in addition to the buyer and seller making first offers (n = 227; see Table One).
We first compared whether the variable means were significantly different across the different treatment groups. Table Three summarizes the results for each treatment group. The “best‐price” group's sample size was lower than that of the other treatments, as we had to remove dyads where the buyer asked for the best price after other values were put forward (see also Table One). However, this imbalance did not impede our analysis strategy, as we used tests that did not require equal sample sizes (regression, Kruskal–Wallis test) or observed equal variances (ANOVA).
Variable . | No List Price . | List Price . | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Buyer First . | Seller First . | Question Seller First . | Buyer First . | Seller First . | Question Seller First . | ||||||
. | n = 41 . | n = 45 . | n = 28 . | n = 42 . | n = 47 . | n = 24 . | ||||||
. | M . | SD . | M . | SD . | M . | SD . | M . | SD . | M . | SD . | M . | SD . |
First offer | 390.2 | 85.2 | 631.3 | 85.0 | 544.1 | 62.4 | 404.9 | 69.9 | 588.4 | 41.6 | 548.5 | 40.2 |
Agreement | 493.7 | 44.3 | 508.5 | 40.2 | 480.5 | 44.4 | 497.8 | 30.4 | 507.7 | 32.3 | 489.5 | 24.5 |
Total concession | 106.6 | 75.2 | 122.8 | 65.6 | 63.6 | 62.1 | 92.9 | 52.9 | 80.7 | 51.9 | 59.0 | 44.8 |
SVI seller | 3.8 | .7 | 3.4 | .7 | 3.6 | .6 | 3.7 | .7 | 3.7 | .5 | 3.8 | .8 |
First offer to reservation price | 121.3 | 64.0 | 182.0 | 106.8 | 102.0 | 71.6 | 116.0 | 72.9 | 132.6 | 67.6 | 102.7 | 57.1 |
Variable . | No List Price . | List Price . | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Buyer First . | Seller First . | Question Seller First . | Buyer First . | Seller First . | Question Seller First . | ||||||
. | n = 41 . | n = 45 . | n = 28 . | n = 42 . | n = 47 . | n = 24 . | ||||||
. | M . | SD . | M . | SD . | M . | SD . | M . | SD . | M . | SD . | M . | SD . |
First offer | 390.2 | 85.2 | 631.3 | 85.0 | 544.1 | 62.4 | 404.9 | 69.9 | 588.4 | 41.6 | 548.5 | 40.2 |
Agreement | 493.7 | 44.3 | 508.5 | 40.2 | 480.5 | 44.4 | 497.8 | 30.4 | 507.7 | 32.3 | 489.5 | 24.5 |
Total concession | 106.6 | 75.2 | 122.8 | 65.6 | 63.6 | 62.1 | 92.9 | 52.9 | 80.7 | 51.9 | 59.0 | 44.8 |
SVI seller | 3.8 | .7 | 3.4 | .7 | 3.6 | .6 | 3.7 | .7 | 3.7 | .5 | 3.8 | .8 |
First offer to reservation price | 121.3 | 64.0 | 182.0 | 106.8 | 102.0 | 71.6 | 116.0 | 72.9 | 132.6 | 67.6 | 102.7 | 57.1 |
Note: No list price: no price has been listed; M and SD stand for mean and standard deviation, respectively. The statistical tests for group differences are reported below in the hypothesis tests.
We tested Hypotheses 1 and 2 with a mediation model using the PROCESS macro (Hayes 2018). This was useful, as we could investigate both the direct correlations and the mediation of the first offer to the negotiation outcome. The resulting mediation model is depicted in Figure One. We dummy‐coded the group variable with the baseline level when the buyer made the first offer (Buyer_nolist).
H1 predicted that the best‐price question leads to a seller offer that is lower than a normal “seller first offer” but higher than a “buyer first offer.” The process model analysis found that group membership is a significant predictor of the first offer (Figure Three, path a). Furthermore, the coefficients are in the predicted order (“buyer first offer” > “seller first offer” after question > “seller first offer”). Only the “Buyer_list” path was not statistically significant, as the difference to the baseline level (Buyer_nolist) was too marginal.
For additional robustness, we also conducted a group comparison test of means in addition to the process model analysis. The group comparison confirms the regression results of the process model and finds significant differences between the treatments. The results of the ANOVA test can be found in Appendix Two. Therefore, the analysis supports our H1 in that the first offer in response to the question was in between the other treatments. We can thus accept H1.
In H2, we predicted that the first offer mediates the relation between group membership and that the economic outcome in the question treatment is lower than in the “seller first offer” and higher than in the “buyer first offer” treatment. Our mediation analysis partly supports this hypothesis. The model shows that the group has a significant relative indirect effect on the negotiation result via the first offer, as indicated by the confidence interval that does not include zero. Compared to the baseline treatment (the buyer makes the first offer), the first offer was significantly higher if (a) the seller made it and (b) it was in between the other two treatments if the seller made it in response to the question. The “buyer_list” treatment was not significantly different due to its closeness to the baseline term. As expected, the negotiation opening predicted the outcome via the first offer, in that the question led to a coefficient between a “seller first‐offer” and a “buyer first‐offer.” We can, therefore, accept the mediation part of H2.
However, the group membership and the above‐confirmed mediation effect only partly explain the negotiation outcome, as the group comparison only partly shows the same differences. We conducted a group comparison using a Kruskal–Wallis rank sum test to compare the different negotiation outcomes. There was a significant difference between the experimental treatments [H(5) = 12.17, p = 0.03]. A pairwise comparison with a Wilcoxon rank sum test showed that the agreement was lower when the buyer asked the question than when the seller made the first offer in both list price conditions (p = 0.02 in no list price [i.e., no price was listed] treatment and p = 0.02 in the list price treatment). There was no significant difference in the other comparisons. The mean differences are also reported in Table Three. Thus, the second part of H2—the difference in outcomes—needs to be rejected, as we did not observe three levels of outcomes in the three treatments. The difference in outcome can also be observed in Figure Two.
The interaction hypothesis between the opening and the presence of a list price (H3) was tested with a linear regression model using these two factors with the Buyer Opening as the reference category. The model was significant [F(5,22) = 88.67, p < 0.001] with both openings as significant predictors of the First Offer [opening with question at b = 143.66, t(220) = 8.23, p < 0.001 and Opening by Seller at b = 183.52, t(220) = 12.67, p < 0.001]. In the case of the absence of a list price, the first offer was only higher if the seller made the first offer [b = 59.52, t(220) = 2.87, p = 0.01] and not if the negotiation was opened with the question by the buyer [b = 12.20, t(220) = 0.50, p = 0.62]. The regression model is depicted in Table Four.
Dependent Variable . | Dependent Variable . |
---|---|
. | First Offer . |
No list price | −16.63 (15.07) |
Opening with question | 143.66*** (17.45) |
Opening by seller | 183.52*** (14.48) |
No list price × opening with question | 12.20 (24.23) |
No list price × opening by seller | 59.52*** (20.72) |
Constant | 404.88*** (10.53) |
Observations | 226 |
R2 | 0.67 |
Adjusted R2 | 0.66 |
Residual Std. Error | 68.21 (df = 220) |
F Statistic | 88.67*** (df = 5; 220) |
Dependent Variable . | Dependent Variable . |
---|---|
. | First Offer . |
No list price | −16.63 (15.07) |
Opening with question | 143.66*** (17.45) |
Opening by seller | 183.52*** (14.48) |
No list price × opening with question | 12.20 (24.23) |
No list price × opening by seller | 59.52*** (20.72) |
Constant | 404.88*** (10.53) |
Observations | 226 |
R2 | 0.67 |
Adjusted R2 | 0.66 |
Residual Std. Error | 68.21 (df = 220) |
F Statistic | 88.67*** (df = 5; 220) |
Note: No list price = no price was listed.
*p < 0.1; **p < 0.05; ***p < 0.01.
Thus, we can confirm the interaction effect as per H3 because the presence of an additional list price reduced the first offer in the “seller” condition. This interaction effect is also depicted in Figure Three.
H4 proposed that if sellers received the question concerning the best price, they would make concessions of a smaller total magnitude than those in other conditions. Due to the non‐normality of the concession data, we used a Kruskal–Wallis test to test for group differences. The test confirmed significant differences between the groups, H(5) = 31.02, p < 0.001. In addition, we used a pairwise Wilcoxon test to understand the differences between the groups. In the absence of a list price, the total magnitude of concessions was significantly lower in the case that the question was asked (question_nolist – seller_nolist p < 0.001; question_nolist – buyer_nolist p = 0.006). For the condition containing the list price, the question led to lower total concessions only when compared with the “buyer‐first” condition (question_list – buyer_list p = 0.007). There was one significant interaction effect of the list price: in the case of the seller making the first offer, the total concessions were significantly higher (seller_nolist – seller_list p = 0.001). Therefore, H4 can be only partly confirmed as the concessions in the “list price” condition did not differ between “seller first offer” and “seller first offer after questions.” The results can also be observed in Figure Four.
Finally, H5 suggested that the best‐price question is perceived negatively by the responder (seller) and will thus have a negative impact on the subjective value of the negotiation. To test this hypothesis, we conducted a one‐way ANOVA comparing the SVI score across experimental conditions. The test did not confirm a difference between SVI scores in the treatment groups [F(5, 221) = 2.142, p = 0.06]. Therefore, H5 was rejected, as we found no difference in satisfaction in different conditions.
As an additional analysis, we investigated the first offers in relation to the reported reservation price. This distance can also be considered a measure of the intensity of deception because a true answer to the best‐price question would equal the reservation point, resulting in a distance of “zero.” For the other conditions, this measure captures the aggressiveness of the first offer. The data are represented in Figure Five.
Here too, we used a Kruskal–Wallis test to test for group differences. The test confirmed significant differences between the groups, H(5) = 17.92, p = 0.003. A pairwise Wilcoxon test showed only one significantly different treatment condition. In the “no‐list‐price” condition, sellers had a higher distance or made more aggressive first offers than in the presence of the list price. Further, the “asking the price” treatment led the sellers to reduce their deception, but only in the “no‐list price” condition.
Discussion
As discussed above, previous research has focused on the first offers of the buyer or the seller. This study extends the focus to the broader theme of negotiation openings and adds a new opening tactic—asking the seller about the best price. Our study demonstrated that the buyer's question concerning the best price does change negotiation behaviors and outcomes when compared to a normal first offer made by either a buyer or a seller. The best‐price question leads to a seller's first offer that is lower than the first offer that the seller would normally make. Additionally, it is also above the first offer that the buyer would normally make. We found that additional reference points weaken this effect.
Furthermore, the first offer after the question also predicts the final negotiation outcome. There is no difference in the predictive quality of the first offer regardless of whether it comes from the buyer or the seller, even if the seller is responding to the “best‐price” question; in all cases, the anchoring effect is equally strong. This finding replicates the vast findings related to the anchoring effect for elicited first offers. We also confirmed the indirect effect of the negotiation opening on the outcomes and showed that the outcome for the opening with the best‐price question did result in the same outcomes as buyer first offer.
Additionally, we confirmed the impact of the question on negotiation behavior. Having been asked the best‐price question, sellers made smaller total concessions as compared to the buyer‐first or seller‐first treatment. The additional list price information also moderates the effect comparable to the “higher‐asymmetry” condition.
Finally, we did not confirm a negative effect of the best‐price question on the negotiator's subjective value. This is a counterintuitive finding, as we expected this tough question to have detrimental effects, as per H5. However, there is also evidence that people overestimate the impact of sensitive questions (Hart, VanEpps, and Schweitzer 2021), and this could be a potential explanation for this counterintuitive finding.
Theoretical Contribution
With the question about the best offer, we introduced a new theoretically and practically relevant element in this study that has so far been largely ignored in research. This element is at the intersection of the fields of asking questions, deception, anchoring, and negotiation. However, our results stem from a narrow experimental setting, in which the best‐price question was asked at the beginning and only by buyers. Thus, a generalization of the findings requires more research, and it should be handled with care. Against this background, our contributions to the literature are as follows.
First, we showed that asking even a very sensitive question in negotiation leads to significant response rates. These rates were higher than what we expected but did not include the negotiators' true reservation points, which indicates that deception seems to be the standard in a negotiation setting. However, even though the answer to the best‐price question is typically not true, asking about the best price changes the negotiation process. Both the first offer and concession‐making can be altered by the best‐price question, and these effects translate to changed negotiation outcomes. This is in line with the previous literature on the anchoring effect.
Second, we demonstrated that asking sensitive questions in negotiations had no downsides in terms of subjective value. As described above, negotiators may be hesitant to ask questions, as they assume downsides; however, this assumption is not supported by the results of our study.
Third, we substantiated the literature on information availability and found that additional reference points reduce the effect of asking for the best price.
Further, we uncovered additional data points of interest for negotiation research. We showed how negotiators answer the best‐price question. We also provided descriptive data on where the anchors are positioned in relation to the reported reservation price. Even though this might not be generalizable beyond this experimental context, an improved understanding of first offer sizes could help to understand negotiation openings better.
Practical Implications
We challenged the previous notion of the offer‐counteroffer sequence. The results of this study suggest modifying the pie‐slicing strategies in Thompson (2001: 68) as follows:
According to our data and in our setting, it is not even certain whether it is advisable to make the first offer or whether it is always better to open with the best‐price question. We found that there was at least no disadvantage in making the first offer.
The findings of this study do not change the guidance for the counteroffer as an effective way of offsetting the negative effect of the negotiation partner's first offer—even though we are not aware of any research that investigated this recommendation.
The gist of the practical implications is that the negotiators should be aware of the best offer question. As it might influence their behavior, they should actively ask it to influence the behavior of their negotiation partners, but also be prepared to answer it themselves.
Limitations and Future Research Avenues
In this study, we showed the fundamental importance of the question of the best price. We think that our evidence is sufficient to arouse interest and take a closer look. But then, many questions arise.
First, the scope of this study is limited. At first, we only investigated what happens when the buyer asks for the best price. We cannot conclude that it would be the same for a seller asking the question. Theoretically, however, we would expect similar effects, but this would have to be verified in future research. Moreover, the setting involving a buyer–seller dyad in an online context is still very narrow. The channel itself, for example, could have an impact on the process. Digital channels have a lower information richness (Daft and Lengel 1984), and this lower richness could inhibit the transmission of cues. However, this argument might even strengthen our analysis. As we were already able to detect the effects in a low‐richness channel, the effects should be even stronger in a rich channel, as it conveys more information. For example, asking the question in a face‐to‐face (F2F) setting might make lying even more challenging for a responder. Conversely, Hancock, Thom‐Santelli, and Ritchie (2004) found that email has lower deception rates than F2F communication.
In addition, it would be valuable to learn more about contextual factors, such as how experience affects the effectiveness of the best offer tactic. Our participants were students with relatively little experience, and this could influence our observations. How robust are the effects of asking for the best price? Our data allow only a few insights. Therefore, this tactic should be treated with caution for the moment.
Second, it would be interesting to learn more about how best to respond to the best offer question. There are a few options for strategic responses here. One would be to ignore the question and make the same offer as if the question had not been asked. Does that give a strategic advantage because the other side assumes the answer is a good price? A second strategic response would be to name a good price but not deviate from it, arguing that it is the best price and, therefore, no further concessions are possible. A third strategic response is to ask the boomerang question. It needs to be examined how effective these strategic responses are and what other options there are.
Third, the lack of a significant difference in outcome between the first offer and the best‐price question is very counterintuitive and should be further substantiated. How robust is this connection? What contextual factors affect it? These aspects of our findings need to be critically questioned by future research.
Fourth, our study is limited to distributive negotiation. Mixed‐motive or integrative negotiations could be influenced differently by the question of the first offer. There is also a remarkable gap in the literature, as only a few studies have investigated the first‐offer effect (Lipp, Smolinski, and Kesting 2023). Theoretically, we would expect a lower effect of the first offer and, consequently, a lower effect of the questions in integrative negotiations as more issues and reference points are present, and these reference points reduce the anchoring effect (see our above mediation effect).
Finally, there is an ethical perspective around asking for the best price. Although the question can be considered an invitation to lie, most negotiators perceive lying about reservation prices or BATNAs to be perfectly ethical (Lewicki and Robinson 1998). Thus, an ethical evaluation of the strategy appears to be warranted. Disagreements in our internal discussions show that this is not a clear‐cut question.
Conclusion
We think that the topic of the first offer needs to be reconsidered in a new and more comprehensive way. The offer‐counteroffer sequence needs to be challenged. The best‐price question is certainly relevant in this regard. However, we must still learn a lot more about it to be able to really understand and use its full potential in different types of negotiations.
Acknowledgment
Open Access funding enabled and organized by Projekt DEAL.
Conflict of Interest Statement
We have no known conflicts of interest to disclose.
Data Availability Statement
All data have been made publicly available at the Negotiation Data Repository and can be accessed at https://dataverse.harvard.edu/dataverse/NegotiationDataRepository?q=&types=dataverses%3Adatasets&sort=dateSort&order=desc&page=2; and at the OSF Repository and can be accessed at https://osf.io/j3adr/?view_only=74c3d9b2682a42baa8b20aabd81acd98. This study was not preregistered.
NOTE
In a pre‐study, we asked the participants, “Have you ever encountered the question about the lowest price?” Even though this question was not directly related to the beginning of a negotiation, we believe this adds additional relevance for our research question at hand, as it shows the frequency of the “best‐price” question.
REFERENCES
Appendix One: Experiment Instructions
Buyer Instructions
Imagine you want to buy a new smartphone on a classifieds website (e.g., Craigslist, ebay Kleinanzeigen). After some research, you have a good sense for the market price, which is between 400 and 600 EUR. You recently found an interesting listing. The phone is listed for 600 EUR.1
In your preparation, you thought about your strategy and you recently learned that it is favorable to ask the seller for the best price. Thus, you decided to ask “What is the lowest price you would accept?” at the outset of the negotiation.2
You now used the chat function of the classifieds site to contact the seller. On the next page, you can chat with the seller. Please negotiate the price of the smartphone with the seller. You have 10 minutes time to negotiate. You will receive a bonus, if you reach a deal with the seller and you will receive an additional bonus, the lower the agreed price.
Seller Instructions
Imagine you are selling a used but fully functional smartphone via a classifieds website (e.g., Craigslist, ebay Kleinanzeigen). After some research, you have a good sense for the market price, which is between 400 and 600 EUR. As you want to realize a good deal and in order to leave room for negotiation, you listed the phone for 600 EUR on the classifieds website.3
You are now expecting inquiries regarding the phone and you prepared your strategy: You want to wait with the price discussion, until the buyer brings up the price topic.4
A potential buyer contacted you via the chat function of the classifieds website. In the window on the next page, you can chat with the buyer. Please negotiate a price for the smartphone with the buyer. You have 10 minutes negotiation time. You will receive a bonus, if you reach a deal with the buyer and you will receive another bonus, the higher the agreed price.
Only displayed in the “listprice” condition.
Only displayed in the “buyer asks the lowest price question” condition.
Only displayed in the “listprice” condition.
Only displayed in the “buyer asks the lowest price question” condition.
Appendix Two: ANOVA Results First Offer
We tested the “first offer” differences using a one‐way ANOVA. The results showed a significant difference in first offers among the various treatments, F(5,221) = 88.32, p < .001. A Tukey HSD post hoc test showed the following group differences:
Group 1 | Group 2 | Mean Difference | 95% CI Lower | 95% CI Upper | p |
Buyer_nolist | Buyer_list600 | −14.6 | −57.7 | 28.5 | 0.925 |
Question_list600 | Buyer_list600 | 143.7 | 93.4 | 193.9 | <.001 |
Question_nolist | Buyer_list600 | 139.2 | 91.3 | 187.1 | <.001 |
Seller_list600 | Buyer_list600 | 183.5 | 141.9 | 225.2 | <.001 |
Seller_nolist | Buyer_list600 | 226.4 | 184.3 | 268.5 | <.001 |
Question_list600 | Buyer_nolist | 158.3 | 107.9 | 208.7 | <.001 |
Question_nolist | Buyer_nolist | 153.9 | 105.7 | 202.0 | <.001 |
Seller_list600 | Buyer_nolist | 198.2 | 156.2 | 240.1 | <.001 |
Seller_nolist | Buyer_nolist | 241.0 | 198.7 | 283.4 | <.001 |
Question_nolist | Question_list600 | −4.4 | −59.0 | 50.2 | 0.999 |
Seller_list600 | Question_list600 | 39.9 | −9.4 | 89.1 | 0.189 |
Seller_nolist | Question_list600 | 82.7 | 33.1 | 132.4 | <.001 |
Seller_list600 | Question_nolist | 44.3 | −2.6 | 91.2 | 0.076 |
Seller_nolist | Question_nolist | 87.2 | 40.0 | 134.4 | <.001 |
Seller_nolist | Seller_list600 | 42.9 | 2.0 | 83.8 | 0.034 |
Group 1 | Group 2 | Mean Difference | 95% CI Lower | 95% CI Upper | p |
Buyer_nolist | Buyer_list600 | −14.6 | −57.7 | 28.5 | 0.925 |
Question_list600 | Buyer_list600 | 143.7 | 93.4 | 193.9 | <.001 |
Question_nolist | Buyer_list600 | 139.2 | 91.3 | 187.1 | <.001 |
Seller_list600 | Buyer_list600 | 183.5 | 141.9 | 225.2 | <.001 |
Seller_nolist | Buyer_list600 | 226.4 | 184.3 | 268.5 | <.001 |
Question_list600 | Buyer_nolist | 158.3 | 107.9 | 208.7 | <.001 |
Question_nolist | Buyer_nolist | 153.9 | 105.7 | 202.0 | <.001 |
Seller_list600 | Buyer_nolist | 198.2 | 156.2 | 240.1 | <.001 |
Seller_nolist | Buyer_nolist | 241.0 | 198.7 | 283.4 | <.001 |
Question_nolist | Question_list600 | −4.4 | −59.0 | 50.2 | 0.999 |
Seller_list600 | Question_list600 | 39.9 | −9.4 | 89.1 | 0.189 |
Seller_nolist | Question_list600 | 82.7 | 33.1 | 132.4 | <.001 |
Seller_list600 | Question_nolist | 44.3 | −2.6 | 91.2 | 0.076 |
Seller_nolist | Question_nolist | 87.2 | 40.0 | 134.4 | <.001 |
Seller_nolist | Seller_list600 | 42.9 | 2.0 | 83.8 | 0.034 |