1 Introduction

Personality traits have been found to predict a wide range of important outcomes such as job performance (Barrick and Mount 1991), career success (Judge et al. 1999), academic success (Hirsh and Peterson 2008), and life expectancy (Roberts et al. 2007). It was therefore only natural for negotiation researchers and practitioners to investigate the personality traits of negotiators, not only to make predictions about individual potential for negotiation, but also to advance negotiation theory by describing general constellations of negotiator behaviors and mental patterns present in negotiation situations. However, despite decades of effort, scholars have struggled to isolate broad personality traits that consistently predict outcomes in negotiation tasks (Sharma et al. 2013). This is surprising because research has found effects of personality in tasks that have parallels with negotiation tasks – that is, goal directed, collective and time-limited tasks – such as collective brainstorming tasks (Bolin and Neuman 2006) and group decision making tasks (Hakim et al. 2021).

To explain the limited predictive power of personality traits for negotiation outcomes, some scholars have suggested that many negotiation situations (especially those in research settings) may be too scripted, resulting in too few individual differences in the psychological meaning attached to the negotiation situation. Knowledge of individuals’ relative positions on trait dimensions would then be of minimal benefit to predicting negotiators’ behavior and subsequent performance (Thompson 1990). Others have argued the opposite and stated that negotiation situations are in fact so open that no personality trait in isolation can be systematically associated to negotiation outcomes, but only in interaction with situational factors (Wilson et al. 2016; Elfenbein et al. 2018).

In the current paper we make a new case for personality research in the field of negotiation. Building on recent theorizing and research we present an alternative conceptualization and assessment of personality that builds on a more dynamic approach to personality traits. While in classic approaches to personality researchers describe the behaviors and mental patterns of individuals as average tendencies across situations, recent approaches contend that average trait scores are only one way of quantifying individuals’ behavioral and mental tendencies. Another relevant quantification of individuals’ behavioral and mental tendencies is to map individual differences in within-person variability in behaviors and mental states, that is, the variability trait (Fleeson and Gallagher 2009; Lievens et al. 2018; Lang et al. 2021). Until now negotiation research has focused on average traits and has shown that they do not help much in predicting negotiation outcomes, but what about a negotiator’s variability trait? In the present paper we propose an explanation of why average traits have limited predictive power for the outcomes of a negotiation task and why the variability trait should be a better predictor of negotiation outcomes. In two empirical studies, we investigate the predictive value of the variability trait for negotiation outcomes.

1.1 Personality Traits and Negotiations Outcomes

Identifying the stable characteristics of efficient negotiators has long been a goal of negotiation research for both theoretical and practical reasons (Sharma et al. 2013). On a theoretical level, it provides insights into the psychosocial processes at work in negotiations, whereas on a practical level, it allows the identification of individuals with high potential in negotiations. In this line of research, personality is probably one of the factors that has been the most scrutinized (Sharma et al. 2013; Elfenbein 2015). Yet, to date, the empirical investigation of the links between personality traits and negotiation outcomes has not been very fruitful. A meta-analysis revealed that negotiators’ personality traits only contribute to negotiators’ sense of satisfaction at the end of a negotiation task, but that they have close to no predictive value for economic outcomes (Sharma et al. 2013). This is surprising because personality traits predict important outcomes in other types of dyadic and group interactions – such as collective brainstorming tasks (Bolin and Neuman 2006) or group decision making tasks (Hakim et al. 2021). In what follows, we argue that in most prototypical negotiation situations a wide range of goals are activated in a relatively short time span, and hence a variety of mental states and behaviors are required to perform well. This could explain why classic and broad personality traits – which describe average behavioral tendencies across situations (Fleeson 2004; Fleeson and Jayawickreme 2015) – have limited predictive value for negotiation outcomes (Sharma et al. 2013).

Prototypical negotiations as investigated by negotiation researchers are social situations with an important distinctive feature: the presence of mixed motives in both negotiation partners (Walton and McKersie 1966). Negotiations are induced by incompatible positions that are difficult to reconcile without a careful balancing act of cooperative and competitive behaviors (Walton and McKersie 1966), which sets negotiations apart from collaborative problem-solving tasks. Negotiations lead to what is generally known as the negotiator’s dilemma, that is, the choice that negotiators have to make between competing to claim value and cooperating to create value (Lax and Sebenius 1987; Lewicki and Hiam 2007). To be successful in negotiations, negotiators need to dynamically adjust their mental states and behaviors, in short lapses of time, to be more or less competitive or cooperative depending on the issue at stake and/or the strategy of the negotiating partner.

More specifically, many negotiations include multiple issues that are of varying importance to the negotiator (Pruitt 1981; Weingart et al. 1990). Research suggests that negotiators should yield on issues that are not important for them, and compete on issues that are important for them (Pruitt 1981; Weingart et al. 1990). This means that negotiators should be able to modify their mental states and behaviors in sometimes very short lapses of time, being assertive or even disagreeable at one moment, and compliant in the following moment, depending on which issue is discussed. Classic and broad personality traits are not suited to capture such dynamic changes in mental states and behaviors, which could be one of the reasons why they have limited predictive value for the outcome of a negotiation task.

In addition, variations in the characteristics of negotiation partners should be taken into account by negotiators. Negotiators are indeed typically confronted with various kinds of negotiation partners (Wilson et al. 2016). The strategy and/or the personality of the negotiation partner can require for example overall a more aggressive and extroverted response style or a more agreeable approach. This view is in line with the interactionist approach, which states that the effectiveness of a negotiator’s personality depends on the characteristics of the negotiation partners (Wilson et al. 2016; Elfenbein et al. 2018). Taking into account partner characteristics in balancing competition with cooperation requires a high level flexibility from the negotiator. This could explain why neither trait extroversion nor trait agreeableness predict negotiation outcomes.

Finally, moving beyond the interactionist approach and holding partner characteristics constant, every negotiation confronts negotiators with multiple goals that negotiators should correctly recognize, and flexibly respond to. Indeed negotiation situations typically consist of distinct sub-processes or phases – such as the preparation phase, the bargaining phase, and the post-negotiation phase – which all contribute to the outcome of a negotiation (Jang et al. 2018). A recent general theory of negotiation offers a rich description of the variety of goals that are pursued by negotiators in a focal negotiation (Jang et al. 2018). For example, during the planning phase of a negotiation, the main goal of negotiators should be to systematically gather information about one’s own and potential partners’ preferences (Jang et al. 2018). During the bargaining phase of a negotiation, on the other hand, the most important goals of negotiators are to form a psychological contract, make offers, and react to counter offers (Jang et al. 2018). This suggests that to be successful in a negotiation task, negotiators need to dynamically adjust their mental states and behaviors depending on the requirements of the phase in which they are. For example, being relatively more conscientious and open during the preparation phase should be particularly adaptive as this phase requires to process complex information in a thorough way (Jang et al. 2018). Conversely, being more extroverted and agreeable during the bargaining phase should be particularly adaptive as this phase requires to socialize and to be persuasive (Oreg and Sverdlik 2014; McCabe and Fleeson 2016).

Taken together, the fact that negotiation tasks require a wide variety of mental states and behaviors could explain why classic and broad personality traits, which capture average behavioral tendencies across situations (Fleeson 2004; Fleeson and Jayawickreme 2015), have a low predictive value for negotiation outcomes (Sharma et al. 2013). If negotiations require various mental states and behaviors, the best negotiators should not be those who are most of the time more conscientious, open, extroverted, or agreeable than others, but those who are most able to modulate their mental states and behaviors depending on the requirements and constraints of the particular negotiation issue, negotiation partner or negotiation phase. This is the reason why a trait like extroversion could lead to positive outcomes in some but not all negotiation situations.

1.2 The Variability Trait: Another Dimension of Personality

Due to the complex nature of negotiation tasks, identifying a trait that could predict negotiation outcomes requires to focus on an another aspect of personality. In recent personality theories, average tendencies across situations are only one way to look at individual differences in personality (Fleeson and Gallagher 2009; Fleeson and Jayawickreme 2015; Lang et al. 2021). These theories posit that individual differences in within-person variability in mental states and behaviors described by personality traits – that is, the variability trait (Fleeson and Gallagher 2009; Lievens et al. 2018; Lang et al. 2021) – should be an integral part of descriptive models of personality next to average traits (Fleeson and Gallagher 2009; Lang et al. 2021).

To better understand the nature of the variability trait, it is important to recall that personality traits describe clusters of mental states and behaviors that help achieve personal goals (Fleeson and Jayawickreme 2015). For example, behaving in an extroverted manner (i.e., going to parties, being talkative) allows socializing and connecting, while behaving conscientiously allows getting things done (McCabe and Fleeson 2016). When an individual is in a given personality state in a given situation, this is due in part to stable characteristics of the individual such as their values or beliefs that cause them to implement this personality state, but also to characteristics of the situation that make this personality state functional and desirable (Fleeson and Jayawickreme 2015). The influence of situational characteristics on personality states is evidenced by experience sampling studies – that is, studies in which participants are asked to report personality states at different times of the day over several days – which show that, although average personality states across situations are relatively stable within an individual from one period of time to another, there are also large variations in personality states within the same individual from one moment to another (Fleeson 2001; Fleeson and Gallagher 2009; Judge et al. 2014). Within-person variations in personality states are explained by variations in the goals activated by different situations (McCabe and Fleeson 2016). In other words, it appears that individuals are able to modify their personality states in order to achieve the goals they set for themselves in a given situation.

Importantly it appears that people systematically differ with regard to the size of the variability in their personality states (Fleeson and Gallagher 2009; Lang et al. 2021), and therefore systematically differ in their ability to adapt their personality states to the situation. This trait has been referred to as the variability trait (Beckmann et al. 2021). In the literature two methods have been proposed to measure individual differences in the variability trait. The first method is based on experience sampling studies, in which the variability trait is derived from variations in individuals self-reported mental states and behaviors at different times of the day over several days (Beckmann et al. 2021). Capturing individual differences in the variability trait with this method is challenging because the situations to which individuals are exposed are not standardized (Lievens et al. 2020). A second method is based on standardized self-report instruments like Situational Judgment Tests (SJT) (Lievens et al. 2020). These are questionnaires commonly used in recruitment, consisting of short scenarios of various common situations, along with a list of possible behavioral reactions corresponding to different personality states in one trait (Weekley and Ployhart 2006; Lievens et al. 2020). Respondents indicate for each situation the likelihood of each one of the suggested behavioral reactions. Because in such tests the situations to which individuals are exposed are all the same, SJT guarantee standardization and allow measuring pure individual differences in average traits and in the variability trait (Lievens et al. 2018; Lievens et al. 2020). Prior research suggests that the variability trait measured with standardized instruments captures the ability to adapt personality states to the requirements of the situation as it was found to be positively associated with functional flexibility (Lievens et al. 2018), job performance (Lievens et al. 2018) and career adaptability (Storme et al. 2020).

1.3 Overview of Studies and Hypotheses

In negotiation tasks, the variability trait should be a positive predictor of performance. Indeed, negotiators who score high on the variability trait can be expected to display personality states that are adapted to the specific requirements of the negotiation task, leading ultimately to higher performance and satisfaction. In contrast, average traits should have limited predictive power regarding the outcomes of a negotiation task. Indeed, since negotiations require very different personality states, a general behavioral tendency should lead to mental states and behaviors that are sometimes functional and sometimes dysfunctional depending on the situation. In sum, we hypothesize that only the variability trait is a positive predictor of performance in negotiations, and more specifically of individual gains (H1), joint gains (H2), and post-negotiation satisfaction (H3).

An important step in testing our hypotheses is to choose a method to measure individual differences in the variability trait. As outlined in the previous section, it was suggested that while experience sampling studies demonstrate the existence of within person variability in personality, they are less powerful in capturing individual differences in this variability due to their unstandardized nature (Lievens et al. 2020). In our empirical studies, we rely on Situational Judgment Tests (SJT) because they provide a purer measure of individual differences in average traits and the variability trait (Lievens et al. 2018; Lievens et al. 2020).

We test in a first study the link between the outcomes of a dyadic three-issue integrative negotiation and the variability trait assessed in extroversion, agreeableness, conscientiousness, and openness, measured using an adaptation of a validated personality SJT (Mussel et al. 2018). In a second study, we use a dyadic eight-issue negotiation with more integrative potential than the negotiation in Study 1 in order to gain more insights about the relationship between the variability trait and joint gains. In Study 2, we also test whether negotiation outcomes can be predicted from the variability trait estimated on a single personality trait (agreeableness).

2 Study 1

The aim of Study 1 was to investigate the link between negotiators’ variability trait and negotiators’ outcomes in a dyadic three-issue integrative negotiation. We estimated the variability trait in extroversion, agreeableness, conscientiousness, and openness using a validated personality SJT (Mussel et al. 2018). Note that previous research suggests that the variability trait is not specific to a personality dimension (Baird et al. 2006; Lievens et al. 2018; Lang et al. 2019; Storme et al. 2020). In other words, an individual with a high level of within-person variability in extroversion is expected to show a high level of within-person variability in all other traits as well. In line with previous research, we consider the variability trait as unidimensional. We hypothesized positive relationships between the variability trait and (1) individual gains, (2) joint gains, and 3) post-negotiation satisfaction.

2.1 Method

2.1.1 Sample and Procedure

Participants were 116 third year French business administration undergraduate students (52 men, 64 women) at a French business school enrolled in a negotiation course. Participants were compensated with extra course credits for their participation.

Before starting the data collection, we had determined using APIMPowerR [34] that we needed to recruit a minimum of 65 dyads in order to have a statistical power greater than .80 to detect an effect of medium size of the variability trait on each negotiation outcome. We expected a medium effect size because prior studies on the predictive power of the variability trait reported medium to large effects on outcomes such as job performance (Lievens et al. 2018) or career adaptability (Storme et al. 2020). The data collection stopped when we were not able to recruit more participants in the available pool of students who were available to take part in studies during the semester. We were only able to recruit 58 dyads. We ran a sensitivity power analysis using APIMPowerR and found that the power that we reached to detect a medium effect size with 58 dyads was .75, which suggests that our study is slightly underpowered.

Participants completed the study over two time points. They were initially invited to complete individually a Situational Judgment Test (SJT) measuring extroversion, agreeableness, conscientiousness, and openness (Mussel et al. 2018). Two weeks later, they participated in a face-to-face negotiation exercise in the classroom. We used a business contract negotiation exercise between a buyer and a seller. Participants were randomly assigned into one of the two roles and were given 20 minutes to read independently their role instructions and prepare for the negotiation. Participants were told that the buyer and the seller needed to negotiate three unresolved issues: (1) Price, (2) Percentage of royalties, and (3) Duration of license. Each issue involved nine possible options. We provided participants with details of the payoff for their role. The structure of the payoff schedule left room for integrative potential: price and percentage of royalties were logrolling issues, whereas the duration of license was a distributive issue.

After preparing, participants were given 25 minutes to negotiate and reach an agreement on all three issues. Participants were instructed not to share payoff tables but were otherwise free to negotiate as they pleased. Agreements were finalized by signing off each of the issues on an sale contract sheet. All dyads reached an agreement.

After the sale contract was signed, participants were asked to fill in a post-negotiation survey assessing their post-negotiation satisfaction. To control for the presence of unmeasurable effects from pre-existing relationships, we also asked participants to assess their level of acquaintance with their partner prior to the negotiation simulation. Finally, participants were debriefed about the purpose of the study.

2.1.2 Measures

Average traits and the variability trait We adapted four scales of a validated Situational Judgment Test (Mussel et al. 2018) to measure average trait levels and the variability trait in extroversion, agreeableness, conscientiousness, and openness. Each item consisted of a short description of a specific situation (e.g., for agreeableness: "You want to go on holiday with your partner, but you disagree about the destination") followed by the question "What would you do?". Each description was followed by four different behavioral responses. Behavioral responses reflected higher (e.g., "I agree with my partner’s proposal and hope to be able to determine the destination of our next holiday") versus lower levels of agreeableness (e.g., "I stand my ground and try to convince my partner"). Participants were asked to indicate how well each behavioral response described the way they would behave in the particular situation using a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The test was filled in individually two weeks before the negotiation task. We used four items for each dimension (i.e., extroversion, agreeableness, conscientiousness, and openness) and each participant was thus asked to indicate what would be his/her behavior for 16 situations in total. The SJT, like all other questionnaires, was administered in English to be consistent with the language used in the whole cursus. Although English is not the native language of most participants, all participants have, per recruitment, an English level equivalent to a TOEIC score of at least 850, corresponding to the level expected for professional buyers (Oliveri and Tannenbaum 2017). Cronbach’s \(\alpha \) ranged between .64 and .76 for average trait levels, which is satisfactory for SJTs (Catano et al. 2012; Campion et al. 2014). More information about the scoring and reliability of this scale is provided in the Data analysis subsection.

Individual and joint gains Individual gain was operationalized as the sum of the points earned by each negotiator during the negotiation exercise. Joint gain was operationalized as the sum of individual gains within the same dyad.

Satisfaction We used the 16-item Subjective Value Inventory (Curhan et al. 2006) to measure satisfaction with the negotiation. This scale measures the extent to which a negotiator is satisfied with (1) the instrumental outcome (4 items, “I am satisfied with my own outcome”), (2) the self (4 items, “This negotiation made me feel more competent as a negotiator”), (3) the process (4 items, “I would characterize the negotiation process as fair”), and (4) the relationship (4 items, “I am satisfied with my relationship with my counterpart as a result of this negotiation”) with regard to a specific negotiation. Participants rated their level of satisfaction on a 7-point Likert scale. The internal consistency of the scale was satisfactory (Cronbach’s \(\alpha \) = .75).

Acquaintance We used two items to measure the degree of acquaintance between participants prior to negotiation (e.g., “How close were you with this person?”) (Fischer and Roseman 2007). Responses were given on a 5-point Likert scale (from “not close at all” to “very close”). The two items were highly consistent and averaged per participant (Cronbach’s \(\alpha \) = .82). Acquaintance scores from recruiters were highly consistent with acquaintance scores from candidates (Cronbach’s \(\alpha \) = .74), we therefore estimated an overall dyad acquaintance score by averaging the two scores that we used as a control variable in our main analyses.

2.1.3 Data Analysis

Estimating average trait levels and the variability trait We relied on a recently developed IRT model – the Trait Variability Tree Model (TVTM) (Lang et al. 2019) – to estimate simultaneously average trait levels and the variability trait in extroversion, agreeableness, conscientiousness, and openness from the SJT. This IRT model has several advantages over simpler scoring approaches, such as using the within-person average and the within-person standard deviation. A first advantage of TVTM is that the level of the trait is functionally independent from the variability on the trait and from the item difficulties (Lang et al. 2019). Second, TVTM circumvents the maximum score problem by modeling variability at the level of the individual item. Indeed, when using within-person standard deviation to model within-person variability, individuals with high trait levels receive automatically low within-person variability scores, which represents a methodological limitation. Third, TVTM disentangles meaningful within-person variability and variability due to measurement error (Lang et al. 2019).

The TVTM approach consists in using IRT tree models – and more specifically the three-process model (Bockenholt 2012) – to quantify systematic individual differences in response patterns. TVTM assumes that the choice of an answer option on a 5-point Likert scale can be decomposed into three subdecisions represented by three pseudoitems in the statistical model. First, the participant decides whether he or she chooses the middle point of the Likert scale (“neither agree, nor disagree”) or not. This first decision reflects the tendency to be indifferent and is captured by Pseudoitem I in TVTM. Then, if the participant decides not to choose the middle point, he or she has to choose the direction of the response he or she will give – either disagreeing or agreeing. This decision reflects the average trait level – the parameter is indeed highly correlated with traditional scale scores (Zettler et al. 2016) – and is captured by Pseudoitem II in TVTM. Finally, the participant has to decide on the intensity of the disagreement (or the agreement). This decision is captured by Pseudoitem III. Pseudoitems I and III capture response variability, they are typically strongly correlated with one another, and they are less strongly correlated with Pseudoitem II (Lang et al. 2019).

TVTM were fitted with the R library lme4 (Bates et al. 2015). Consistent with previous empirical studies (Zettler et al. 2016; Lievens et al. 2018; Lang et al. 2019), we found a strong correlation at the item level between pseudoitem I and pseudoitem III (r = -.74). We therefore estimated variability trait scores based on the two pseudoitems (Lievens et al. 2018; Lang et al. 2019).

To assess the internal consistency of variability trait scores derived from TVTM, it is not possible to rely on the traditional Cronbach’s \(\alpha \) (Lang et al. 2019). Therefore, we relied on the split-half approach to reliability (Lang et al. 2019). We found that the variability trait scores estimated on the odd-numbered items of the SJT were strongly positively correlated with variability trait scores estimated on the even-numbered items of the SJT (r = .81), indicating satisfactory internal consistency (Lang et al. 2019).

Actor-Partner Interdependence Model To account for the dyadic nature of our data, we relied on the Actor-Partner Interdepence Model (APIM) (Kenny et al. 2006) when examining our main hypotheses. APIM allows estimating simultaneously and independently the effect of individuals’ personality characteristics on their own negotiation outcomes (actor effect) and on their partners’ negotiation outcomes (partner effect). In addition, APIM accounts for the degree of interdependence between negotiation partners from the same dyad in the predictor and outcome variables. In all APIM, we controlled for the level of acquaintance between partners. Individual and joint gains were standardized.

Because dyads were distinguishable (buyer vs. seller), we relied on the distinguishable approach to APIM using Structural Equation Modeling (Kenny and Ledermann 2010). For each negotiation outcome, we first estimated a model in which effects were freely estimated for buyers and sellers. We then estimated a second model in which effects were constrained to be equal for buyers and sellers. For each negotiation outcome, the two models were compared using the minimal Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) procedure (Vrieze 2012). AIC and BIC are relative indicators of “parsimony”, which can be defined as the extent to which a model captures the true relationships between the variables of interest while preventing from over-fitting the data. The best model is the one with the lowest AIC (or BIC) value (Vrieze 2012). We considered a difference between the AIC (or BIC) of two competing models greater than 2, as evidence for the superiority of the model with the lowest AIC (or BIC) (Vrieze 2012).

All models were estimated with the R package lavaan (Rosseel 2012). We relied on full-information maximum likelihood estimation. Regarding absolute model fit, we considered the four following statistical indices: \(\chi ^{2}/df\) ratio of a good model should be less than 3, Comparative Fit Index (CFI) of a good model should be more than .90, Standardized Root Mean Square Residual (SRMR) of a good model should be less than .08, and Root Mean Square Error of Approximation (RMSEA) of a good model should be less than .09 (Schumacker and Lomax 2004).

2.2 Results

For each negotiation outcome, we first investigated whether effects were equal across roles or not. We found that models assuming equal effects across roles had better relative fit than models assuming different effects for individual gains (\(AIC_{equal} = 966\) vs. \(AIC_{different} = 977\), \(BIC_{equal} = 1140\) vs. \(BIC_{different} = 1171\)), joint gains (\(AIC_{equal} = 900\) vs. \(AIC_{different} = 906\), \(BIC_{equal} = 1055\) vs. \(BIC_{different} = 1070\)), and satisfaction (\(AIC_{equal} = 1051\) vs. \(AIC_{different} = 1058\), \(BIC_{equal} = 1224\) vs. \(BIC_{different} = 1251\)). The absolute fit of the model assuming equal effects across roles was acceptable for individual gains (\(\chi ^{2}/df\) = 1.39, CFI = .95, SRMR = .07, RMSEA = .08), joint gains (\(\chi ^{2}/df\) = 1.50, CFI = .95, SRMR = .07, RMSEA = .09), and satisfaction (\(\chi ^{2}/df\) = 1.51, CFI = .94, SRMR = .07, RMSEA = .09). Consequently, in what follows, we report only the estimates of APIM assuming equal effects across roles (see Table 1 for standardized estimates of all three models).

Table 1 Study 1 - Standardized estimates of the Actor Partner Interdependence Models (APIM)
Full size table

We found support for our first hypothesis (H1). Indeed, there was a positive actor effect of the variability trait on individual gains (\(B = 0.40, p < .05\)). In addition, we found a negative partner effect of the variability trait on individual gains (\(B = -0.40, p < .05\)). None of the average traits predicted individual gains. Our second hypothesis (H2) was not supported as there was no effect of the variability trait on joint gains. Note that none of the average traits predicted joint gains either. Finally, our third hypothesis (H3) was supported as we found a positive actor effect of the variability trait on satisfaction (\(B = 0.77, p < .01\)). None of the other traits was found to predict satisfaction neither in the actor nor in the partner. Note that running the analyses without acquaintance as a covariate did not change the findings.

2.3 Discussion

Study 1 showed that the variability trait predicts individual gains and post-negotiation satisfaction, but not joint gains. One explanation is that while the negotiation used in Study 1 is integrative by design, its integrative potential is limited because it consists of only two exchange issues. A small variance in joint gains could explain why no relationship was found between the variability trait and joint gains. We should thus replicate Study 1 with a negotiation exercise that has more integrative potential.

3 Study 2

Study 2 aimed at replicating the findings of Study 1 but also at extending them in two ways. First, we increased the integrative potential of the negotiation as compared to Study 1 by using a dyadic eight-issue integrative negotiation in order to improve our chances to detect a relationship between the variability trait and joint gains. Second, we estimated the variability trait based on a SJT measuring only agreeableness (Mussel et al. 2018). Since prior research suggests that the variability trait is a single factor encompassing equivalent levels of variability in all personality traits (Baird et al. 2006; Lang et al. 2019; Storme et al. 2020), even if derived from agreeableness scores only, it should be as predictive of negotiation outcomes as if derived from other traits as well.

3.1 Method

3.1.1 Sample and Procedure

Participants were 136 third year French business administration undergraduate students (61 men, 75 women) at a French business school enrolled in a negotiation course. Participants were compensated with extra course credits for their participation.

Before starting the data collection, we had determined using APIMPowerR (Ackerman et al. yyy) that we needed to recruit a minimum of 65 dyads in order to have a statistical power greater than .80 to detect an effect of medium size of the variability trait on each negotiation outcome. The data collection stopped when we were not able to recruit more participants in the available pool of students who were available to take part in studies during the semester. We were able to recruit 68 dyads. We ran a sensitivity power analysis using APIMPowerR and found that the power that we reached to detect a medium effect size with 68 dyads was .82, which suggests that our study is adequately powered.

The design of Study 2 was very similar to the design of Study 1. Participants completed the study over two time points. First, they completed individually a SJT measuring agreeableness (Mussel et al. 2018). Two weeks later, they participated in a face-to-face negotiation exercise in the classroom. We used a job contract negotiation exercise between a job candidate and a recruiter (Neale 1997). Participants were randomly assigned into one of the two roles and were given 20 minutes to read independently their role instructions and prepare for the negotiation. Participants were told that the recruiter and the candidate were seriously interested in working together but needed to negotiate eight unresolved issues: (1) Bonus, (2) Job assignment, (3) Vacation time, (4) Starting date, (5) Moving expense covered, (6) Insurance covered, (7) Salary, and (8) Location. Each issue involved five possible options. We provided participants with details of the payoff for their role. The structure of the payoff schedule left room for integrative potential: Holidays and bonus as well as moving expense covered and insurance covered were logrolling issues, job assignment and location were compatible issues, whereas salary and starting date were distributive issues.

After preparing, participants were given 25 minutes to negotiate and reach an agreement on all eight issues. In our sample, all dyads reached an agreement. After the contract was signed, participants were asked to fill in the same post-negotiation survey as in Study 1 and were debriefed about the purpose of the study. Like in Study 1, all questionnaires were administered in English.

3.1.2 Measures

Average agreeableness and the variability trait We adapted the 22-item agreeableness subscale of the Situational Judgment Test used in Study 1 (Mussel et al. 2018) to measure both average agreeableness levels and the variability trait in agreeableness. The observed Cronbach’s \(\alpha \) for agreeableness was .84, which is satisfactory for SJTs (Catano et al. 2012; Campion et al. 2014). More information about the scoring and reliability of this scale is provided in the Data analysis subsection.

Negotiation outcomes and control variable All negotiation outcomes – that is, individual gains, joint gains, and post-negotiation satisfaction (Cronbach’s \(\alpha \) = .89) – were measured the same way as in Study 1. We used the same measure as in Study 1 to control for acquaintance (Cronbach’s \(\alpha \) = .84).

3.1.3 Data Analysis

The data analysis for Study 2 was conducted the same way as for Study 1. More specifically, we used the Trait Variability Tree Model approach (Lang et al. 2019) to estimate simulateneously average agreeableness levels and the variability trait in agreeableness. Like in Study 1, we found that the split-half reliability estimate was satisfactory for the variability trait (r = .89). Like in Study 1, we used the Actor-Partner Interpendence Model framework (Kenny et al. 2006) to model the relationships between personality characteristics and negotiation outcomes.

3.2 Results

For each negotiation outcome, we first investigated whether effects were equal across roles or not. We found that models assuming equal effects across roles had better relative fit than models assuming different effects for individual gains (\(AIC_{equal} = 736\) vs. \(AIC_{different} = 741\), \(BIC_{equal} = 787\) vs. \(BIC_{different} = 801\)), joint gains (\(AIC_{equal} = 555\) vs. \(AIC_{different} = 559\), \(BIC_{equal} = 593\) vs. \(BIC_{different} = 601\)), and satisfaction (\(AIC_{equal} = 711\) vs. \(AIC_{different} = 716\), \(BIC_{equal} = 762\) vs. \(BIC_{different} = 776\)). The absolute fit of the model assuming equal effects across roles was satisfactory for individual gains (\(\chi ^{2}/df\) = 0.97, CFI = 1.00, SRMR = .07, RMSEA = .00), joint gains (\(\chi ^{2}/df\) = 0.95, CFI = 1.00, SRMR = .07, RMSEA = .00), and satisfaction (\(\chi ^{2}/df\) = 0.98, CFI = 1.00, SRMR = .07, RMSEA = .00). Consequently, in what follows, we report only the estimates of APIM assuming equal effects across roles (see Table 2 for standardized estimates of all three models).

Table 2 Study 2 - Standardized estimates of the Actor Partner Interdependence Models (APIM)
Full size table

Our first hypothesis (H1) was supported: We found a positive actor effect (\(B = .47, p < .05\)) of the variability trait on individual gains. Note that we also found a negative actor effect (\(B = -.68\), \(p < .05\)) and a positive partner effect (\(B = .59, p < .05\)) of average agreeableness levels on individual gains. We found support for our second hypothesis (H2) as there was a positive effect of the variability trait on joint gains (\(B = .39, p < .05\)). For joint gains, we did not find any effect of average agreeableness levels. Finally, our third hypothesis (H3) was supported. We found a positive actor effect of the variability trait on satisfaction (\(B = .80, p < .01\)). None of the other effects were significant. Note that running the analyses without acquaintance as a covariate did not change the findings.

3.3 Discussion

Like in Study 1, we found that the variability trait was positively associated with individual gains and post-negotiation satisfaction. We also found that the variability trait was positively linked to joint gains. Importantly, Study 2 demonstrates that the variability trait, even when estimated on a single trait (here agreeableness), has predictive value for negotiation outcomes.

4 General Discussion

Our aim was to provide preliminary evidence of the value of the variability trait for predicting negotiation outcomes. Consistent with our hypotheses, we found in both studies that the variability trait positively predicts individual gains and post-negotiation satisfaction over and beyond average traits. In Study 2, we found that the variability trait predicts positively joint gains as well. Our findings have several theoretical and practical implications that we discuss below.

4.1 Theoretical and Practical Implications

We believe our paper sheds new light on the mystery of the weak links between negotiators’ personalities and their negotiation outcomes (Sharma et al. 2013; Elfenbein 2015). Negotiations are situations in which having one rigid attitude or another is not adaptive, which could explain why average traits have been found to have little predictive power so far (Sharma et al. 2013) or only in interaction with situational characteristics (Wilson et al. 2016; Elfenbein et al. 2018), in spite of the existence of important individual differences in the ability to negotiate (Elfenbein et al. 2008). The fact that individual differences in the variability trait predict negotiation outcomes could tempt some readers to infer from our findings that the personality of negotiators – in the classic sense of stable traits – is irrelevant in predicting success in a negotiation. However, the variability trait should be regarded as a stable trait (Lievens et al. 2018; Lang et al. 2021). Our findings thus demonstrate that there is at least one trait of negotiators that predicts negotiation outcomes, although it is a trait capturing a form of flexibility.

In Study 2, we found a negative actor effect and a positive partner effect of average agreeableness levels on individual gains. These findings are not in line our general idea that average personality traits play a minimal role in predicting negotiation success. Note that we did not find the same pattern of findings in Study 1, which might suggest that the results of Study 2 are coincidental. But prior research on the role of the personality dimension unmitigated communion may also shed light on the link we have highlighted between agreeableness and individual gains (Amanatullah et al. 2008). Unmitigated communion can be viewed as an anxious and defensive facet of agreeableness defined by a tendency to display high relational concern without balancing self-concern. In negotiations, it has been shown that unmitigated communion leads negotiators to make concessions in order to avoid jeopardizing the relationship, which leads, in turn, to lower outcomes (Amanatullah et al. 2008). It may be that the SJT we used in Study 2 captures a form of unmitigated communion, which could explain the links we observed between agreeableness and individual gains. Another possible explanation is that something in the context of Study 2 made agreeableness a relevant predictor of individual gains. For example, it is possible that our participants engaged in the negotiation simulation predominantly with a competitive mindset. This could explain why individuals high in agreeableness – who yielded too much – achieved lower outcomes in our sample than individuals low in agreeableness. Replicating our work with other populations would allow a better understanding of the role of average agreeableness levels in negotiations and would be a valuable test of the person-situation interactionist approach (Wilson et al. 2016; Elfenbein et al. 2018).

Our findings, although preliminary, can guide professional negotiators in their practice. A first area of practical application is the selection of negotiators. Our studies suggest that negotiators who score high on the variability trait should be favored in a selection process as they have the highest potential for negotiation. For recruiters interested in this application, the variability trait can be assessed by combining Situational Judgement Tests and Trait Variability Tree Models (Lievens et al. 2018; Lang et al. 2019). A second area of practical application is about the general attitude that negotiators should have during a negotiation. Our findings suggest that it is better to try to pay attention to characteristics of the negotiation situation and adopt personality states that seem natural in the situation, rather than trying to have a certain posture and to keep it no matter what. It is still too early to make recommendations as to the best personality states for each negotiation situation. Our research is still in a preliminary stage and more research on these issues will be necessary before concrete recommendations can be made.

In the educational context, one of the roles of instructors is to help students become aware of their behavioral tendencies in order to learn how to manage them. Here again, the combined use of Situational Judgment Tests and Trait Variability Tree Models could be a way to make students aware of their behavioral tendencies. With regard to personality traits, students could either try to change their behavioral tendencies through behavioral change exercises or discover strategies that allow their strengths to be exploited and/or weaknesses to be circumvented (Movius 2008; Elshenawy 2010). We describe in the next section how dynamic personality theories could provide valuable ideas as to how one could make individuals more flexible.

4.2 Limitations and Future Research

Our goal with this paper was to propose a new approach to the contribution of negotiators’ personality to negotiation outcomes. As with any pioneering work, our work has many limitations which are as many avenues to explore for future research. First, we did not test directly that the variability trait predicts the adoption of adaptive behaviors during the negotiation. It is especially important that this interesting possibility is further investigated, because the current design of our studies does not allow ruling out alternative explanations for our findings. It is possible, for example, that the variability trait and negotiation outcomes are correlated because they are both caused by a third variable. For instance, general mental ability could both help in displaying functional behavioral flexibility and achieving good negotiation outcomes via a process of understanding the economic dynamics underlying the economic game (Mischel 1999; Sharma et al. 2013). This could mean that the variability trait is positively associated with negotiation outcomes without the former having any direct effect on the latter.

In line with this limitation, more research will also be necessary to define what is an adaptive personality state, depending on the phases of a negotiation or the attitude of the partner. One way to do this would be to start by listing the most important goals for each micro-situation constituting a focal negotiation. Goals could be defined on the basis of a general situational taxonomy such as DIAMONDS (Rauthmann et al. 2014) or CAPTION (Parrigon et al. 2017). In a second step, the extent to which different personality states are predictive of negotiation outcomes depending on the goals pursued in the situation should be investigated. Personality states associated with success in a given situation could be considered as adaptive in that particular situation. The principle of this research would be in line with the recommendations of recent calls for more granularity in the approach to the negotiation process (Jang et al. 2018). Furthermore, research on this issue could lead to new insights on how to behave best in different phases of a negotiation, with potentially important theoretical and practical implications.

Another limitation is that we did not theorize finely about what causes individual differences in the variability trait, which could be important to understand what leads to the adoption of adaptive behaviors during a negotiation. A first possibility is that flexible negotiators are flexible because they identify better than others the most important goals in a given situation. For example, they might understand that during the preparation phase it is crucial to be conscientious and that during the bargaining phase it is crucial to be extroverted. A second possibility is that flexible negotiators are more knowledgeable as to which personality states are most effective to reach a given goal. For example, they might have a better understanding of how being conscientious can help gather information prior to a negotiation. A third possibility is that flexible negotiators are better able to implement the personality state that enables them to achieve the goal. For example, they might know better what it takes to be assertive in the bargaining phase. All these possibilities are of course not mutually exclusive. Better understanding the relative role of these different processes could have important practical implications. For example, in the context of training negotiators, one would know whether it is better to focus educational efforts on explaining the different goals in different phases, the function of different personality states, or how to implement a personality state to reach a specific goal.

4.3 Conclusion

Identifying personality traits that are predictive of negotiation outcomes is important for theoretical and practical reasons. Although preliminary, our findings suggest that it is especially the variability trait that predicts negotiation outcomes. Negotiators who tend to adapt their personality states to the requirements of a situation relatively more than others achieve higher individual and joint gains, and are more satisfied at the end of the negotiation. We hope that our work will encourage personality and negotiation researchers to further investigate an idea that has many potential theoretical and practical implications.