CONTRACT FRAMING IN INTER-ORGANIZATIONAL COOPERATIONS
A SOCIAL IDENTITY APPROACH
Research Master Thesis
MSc. Research Master: Business Research University of Groningen
Faculty of Economics and Business
Martin Kleis Pit
S2171740
2 ABSTRACT
Despite extensive research, scholars lack consensus as to how contracts impact trust dynamics in inter-firm relationships. Whereas some scholars argue that contracts inhibit trust, others propose that contracts facilitate trust among alliance partners. Drawing from social identity theory, we propose that contracts can have different psychological effects induced by means of contract framing. By altering the phrasing of a contract, economically equivalent contracts can induce either collective or separate identity perceptions. These identity perceptions in turn allowing alliance partners to form clear expectations regarding the cooperation, facilitating trust in the alliance context. In a series of three experimental studies, we found evidence to suggest that setting clear expectations among alliance partners facilitates trust dynamics. However, our results do not provide conclusive evidence to suggest that contract framing in itself plays a significant role in generating clearer expectations. Furthermore, in contrast to the common notions of social identity theory, we found no differences in trust dynamics based on whether collective or separate identity perceptions were induced. Together, this study provides an initial test of concept for contractual identity framing, opening a horizon for future research into the topic.
3 TABLE OF CONTENTS
INTRODUCTION... 4
THEORETICAL FRAMEWORK ... 7
Contract usage in inter-firm relationships ... 7
Inducing social identity perceptions using contract frames ... 8
Hypotheses ... 10
Contract framing and trust dynamics ... 10
Expectations under contract framing ... 12
Expectations and trust among SA partners ... 13
Trust under collective and separate identity frames ... 14
OVERVIEW OF THE PRESENT RESEARCH ... 16
STUDY 1A: IDENTITY FRAMING AND TRUST DYNAMICS ... 16
Participants and design ... 16
Procedure ... 17
Manipulation ... 17
Measures ... 18
Results ... 19
Discussion ... 21
STUDY 1B: EXPECTATIONS UNDER IDENTITY FRAMES ... 22
Participants and design ... 22
Procedure ... 23
Measures ... 23
Results ... 24
Discussion ... 25
STUDY 2: IDENTITY FRAMING UNDER REPEATED INTERACTION ... 26
Negative behavior under identity frames ... 26
Participants and design ... 28
Procedure ... 28 Manipulations ... 29 Measures ... 30 Results ... 31 Discussion ... 33 GENERAL DISCUSSION ... 34 Theoretical implications ... 35
Limitations and future research directions ... 38
4 INTRODUCTION
Cooperative inter-firm relationships such as strategic alliances (henceforth: SAs) have become an increasingly important source for organizations in creating a sustainable competitive advantage (Cao & Lumineau, 2015; Faems, Janssens, Madhok, & van Looy, 2008; Parmigiani & Rivera-Santos, 2011). Despite their potential advantages, many SAs have failed to successfully build a cooperative environment in these relationships, instigating numerous scholars to investigate the mechanisms of successful SA governance (Faems et al., 2008; Gill & Butler, 2003). Governance mechanisms represent the means used to coordinate SA resources and to lessen the risk of opportunistic behavior, and consist of two main types: contractual and relational mechanisms (Cao & Lumineau, 2015; Faems et al., 2008; Malhotsra & Lumineau, 2011). On the one hand, contractual governance mechanisms relate to the role of the formal contract as a safeguard against opportunism and conflict between parties. On the other hand, relational governance focuses on the role of trust (i.e. belief in the reliability or ability of another party) in helping both parties to effectively govern SAs (Cao & Lumineau, 2015; Faems et al., 2008).
5
elements in a contract are used, contracts can have both a controlling and a coordinating function and as such a contract could act both as a safeguard for opportunism and a means with which to facilitate trust and cooperation (e.g. Faems et al., 2008; Lumineau, 2014; Malhotra & Lumineau, 2011; Mellewigt, Madhok, & Weibel, 2007).
To enhance our understanding of the use of contractual and relational governance mechanisms in SAs, in this study we will examine how contractual framing can impact the relational quality of SAs. Weber and Mayer (2011) and Weber & Bauman (2017), coined the idea that the phrasing used in a contract can foster trusting relationships, separate from the economic consequences conveyed through the contract. The research by Weber & Mayer (2011) draws on regulatory focus theory and shows how contractual safeguards can be framed as either penalties or bonuses to induce different behaviors among SA partners. Following their reasoning, we argue that contractual framing can also be used to induce psychological effects that provide us with useful insights on the workings of contractual governance on trust dynamics in SAs (Weber & Mayer, 2011; Weber et al., 2017). However, in doing so in this study we propose a promising yet underexplored facet of inter-firm relationship, that of social group membership. Building on social identity theory (SIT; (Albert & Whetten, 1985; Ashforth & Mael, 1989; Tajfel & Turner, 1986; Whetten, 2006), we will examine whether contract framing can induce either collective or separate identity perceptions among SA partners, and whether these identity frames can foster trust in the SA context. Social identity theory (SIT; Tajfel & Turner, 1986) proposes that group memberships are central to people’s self-views and self-esteem. Therefore, individuals tend to display more concern for the welfare of fellow “in-group” members than for outsiders (Haslam, Reicher, & Platow, 2011; Kane, Argote, & Levine, 2005; Riek, Mania, Gaertner, Mcdonald, & Lamoreaux, 2010). Creating collective group identity perceptions between two entities (whether these are smaller groups or larger organizations) can therefore facilitate the development of interpersonal trust, whereas diverging group identity perceptions among entities foster distance and even competition (Ashforth & Mael, 1989; Cartwright & Schoenberg, 2006; Haslam et al., 2011).
6
behavior (Kane et al., 2005; Riek et al., 2010; Sethi, 2000). In the field of mergers and acquisitions (M&As), for example, it has already been found that organizational identity perceptions represent an important predictor of successful inter-firm cooperation (Cartwright & Schoenberg, 2006). Research suggests that contractual framing can lead to specific expectations and emotional reactions that further influence a cooperative relationship once confirmed or violated (Weber & Mayer, 2011; Weber et al., 2017). By combining our understanding of governance mechanisms with SIT, we aim build a more nuanced look at how contracts influence transactions and exchange relationships, addressing the debate about the effect of contracts on relationships and cooperative performance. This thesis sets out to answer the following research question:
RQ: Does the inclusion of (different kinds of) social identity framing in inter-organizational contracts influence the level of trust in strategic alliances?
In conclusion, this study contributes to the existing literature on SA outcomes (e.g., Cao & Lumineau, 2015; Faems et al., 2008; Weber & Mayer, 2011; Weber & Bauman., 2017) by examining how contractual framing sets clear expectations that affect trust perceptions in SAs. In doing so, this study integrates psychological insights into the strategy literature. Furthermore, we will examine whether the different identity frames trigger distinct emotional and behavioral responses once a contract breach occurs. While we hypothesize initial positive effects of contract frames in facilitating expectations and trust, the question remains as to how sustainable these effects are over time. To test this, we will theorize and empirically examine how firms respond to a breach in trust under different contract frames. The following section will present a literature review on (a) the impact of contracts on trust dynamics in SAs, and (b) contractual framing and SIT, which will lead into the hypotheses derived from the review. The hypotheses will be tested empirically using a series of three experiments (Study 1a, Study 1b and Study 2). The relationships proposed in this paper are presented schematically in the following conceptual model (see Figure 1). The individual relationships will be discussed in the following section.
7 THEORETICAL FRAMEWORK
Contract usage in inter-firm relationships
Following Doucette and Wiederholt (1997), effective SA cooperation is the process in which organizations pursue their own goals and keep their autonomy, while orienting their actions towards joint outcomes. As scholars generally agree that inter-firm cooperation is a key predictor of SA success (e.g. Dyer & Singh, 1998; Hagedoorn & Duysters, 2002; Lambooij & Koster, 2016), there is a large body of literature studying which governance mechanisms will ensure effective SA cooperation. Previous literature has acknowledged trust as a crucial mechanism governing many social exchange relationships that are characterized by uncertainty, vulnerability and dependence (Pavlou & Anderson, 2007; Riegelsberger, Sasse, & McCarthy, 2005). In the context of inter-firm cooperation, literature too has regarded trust as one of the key mechanisms that drives cooperative behavior and performance (Inkpen & Currall, 2004; Dyer & Singh, 1998; Gulati, 1995; Palmatier et al., 2007; Weber & Bauman, 2017; Weber & Mayer, 2011). Hence, to better understand the effects of contract frames on cooperation in SAs, it is important to acknowledge trust as one of the key mechanisms at play. Whereas trust has been defined in literature in multiple different ways, most literature defines two core concepts constituting inter-firm trust: (1) reliance and (2) risk (Currall & Inkpen, 2002; Inkpen & Currall, 2004; R. C. Mayer, Davis, & Schoorman, 1995; Rotter, 1980). The first dimension, reliance, constitutes to the action through which one party (individual, group, organization) allows its fate to be determined by another party (Currall & Inkpen, 2002; Inkpen & Currall, 2004). The dimension of risk relates to the fact that the trusting party is open to negative outcomes if the other party takes opportunistic action (Inkpen & Currall, 2004; Lane, Salk, & Lyles, 2001). More specifically, trust can thus be defined as follows: “the willingness of
a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party” (Mayer, Davis & Schoorman, 1995: 712). (Doucette & Wie derholt, 199 7)
8
tool to monitor the behavior of both parties. Monitoring in this case provides a degree of transparency, which inhibits opportunistic behavior and signals goodwill between parties, thus reducing relational hazards. Accordingly, contracts are assumed to facilitate trust between SA parties (Cao & Lumineau, 2015; Zhang & Zhou, 2013). The other perspective sees contracts as substitutes to relational characteristics; arguing that contracts inhibit the development of inter-firm trust and cooperation (e.g. Ghoshal & Moran, 1996; Gulati & Nickerson, 2008; Malhotra, 2009). Specifically, scholars from this stream argue that the strengths contracts bring in terms of monitoring and safeguarding weaken or replace the development of relational mechanisms (Cao & Lumineau, 2015; Poppo & Zenger, 2002). The rules and regulations specified in the contracts will, from this perspective, signal a feeling of distrust towards the other party in the SA (Cao & Lumineau, 2015; Lee & Cavusgil, 2006). By relying too heavily on contract usage to reduce opportunistic behavior, parties signal to one another that they are unwilling to rely on trust in the SA (Ghoshal & Moran, 1996; Gulati & Nickerson, 2008; Malhotra, 2009). In this case, contracts create a situation where SA parties treat one another with wariness and distance, making them less likely to engage in cooperative behavior that benefits the SA (Wijk, Jansen, & Lyles, 2008; Zahra & George, 2002). From this perspective, contracts make it less likely that firms in a SA will cooperate to reach a competitive advantage (Lichtenthaler, 2009; Wijk et al., 2008; Zahra & George, 2002). Drawing from social identity theory (SIT; (Tajfel & Turner, 1986), we propose that contract framing (with the use of identity phrasing) is a fruitful contingency in explaining trust development in SAs.
Inducing social identity perceptions using contract frames
9
likely to positively evaluate fellow colleagues (Ashforth & Mael, 1989; Chakravarty, Fonseca, Ghosh, & Marjit, 2016; Kane et al., 2005). These identity-based perceptions, in turn, create positive expectations regarding cooperation with those from the own group versus other groups (Ashforth & Mael, 1989; Kane et al., 2005; Riek et al., 2010). In contrast, employees who identify highly with their own organization tend to foster distrust and competition towards other firms, which hinders cooperation with those perceived as out-group members (Ashforth & Mael, 1989; Chakravarty, Fonseca, Ghosh, & Marjit, 2016; Gaertner & Dovidio, 2000 A. Kane et al., 2005; Tajfel & Turner, 1986). (Albert & Whetten, 1985)
Previous research into social identity has shown that not only can group membership be conveyed by means of interpersonal interaction, but group membership can also be easily communicated by cues in the environment (e.g. Rogers & Lea, 2005; Spears & Lea, 1994; Spears & Postmes, 2015). Situational features highlighting group membership (e.g. a collective name, symbol, purpose or fate) can increase the salience of group membership and thus affect perceptions of who belongs to the in-group (has a shared identity) and out-group (has a divergent identity) (e.g. Ellemers, Kortekaas, & Ouwerkerk, 1999; Gaertner & Dovidio, 2000; Kane, 2010; Kane et al., 2005). In the context of inter-firm relationships, Weber and Bauman (2017) showed that contract framing can function as environmental cues to induce different regulatory foci among SA partners (Weber & Mayer, 2011; Weber & Bauman, 2017). Hence, following this reasoning we expect that contractual identity framing could function as a fruitful tool with which to induce social identity perceptions among SA partners.
10
environmental cues on which the firms involved in SAs shape their perceptions of each other and their cooperation. Contracts can phrase regulations and deadlines in such a way that they evoke a sense of shared group identity (e.g. ‘shared regulations’ and ‘common procedures’), or contracts can be phrased in such a way that they highlight the fact that both organizations represent different social groups, creating out-group perceptions. Thus, by including identity frames in contracts, organizations are capable to foster additional psychological effects among SA partners, that go beyond the effects of a standard contract.
Hypotheses
In the previous section we have laid out the foundation for this thesis; contracts (through framing) can function as environmental cues with which to induce identity perceptions among SA partners. In the next section, we will develop the hypotheses as to how contract framing influences trust dynamics both directly (by fostering positive emotions regarding the other party’s intention and goodwill) and indirectly by setting clear expectations for the SA partners (H1-H4). Furthermore, we propose that there is a distinct difference between collective and separate identity framing, that could directly affect trust dynamics in the SA context.
Contract framing and trust dynamics
11
Hence, we expect that the effect of contract framing on trust dynamics is also partially mediated as contractual identity framing helps organizations foster clear expectations regarding the other party and the SA cooperation. As stated, we will first develop the direct effect of contract framing on trust dynamics, where we argue that contractual identity framing facilitates trust development as it functions as a sign of goodwill surrounding behavior and intention of the SA partner: building confidence that the own organization will not fall prey to opportunism. Trust is built up based on judgements regarding the other parties perceived ability, integrity and benevolence (Kramer, 1999; Mayer & Davis, 1999; Pavlou & Anderson, 2007). Judgements related to ability are based on whether the party has the right set of skills, competencies and characteristics to behave in a fruitful manner (Mayer & Davis, 1999). Benevolence relates to the degree to which the party is perceived to want to do good towards the trustor; whether the party cares about the best interest of the other (Mayer and Davis, 1999, p. 124). Finally, integrity relates to the perception that the other party adheres to a set of rules and principles that are acceptable to the trustor; and that the party acts in congruence with these principles (Mayer & Davis, 1999, p. 124).
12
positive judgements with regards to the benevolence and integrity of the other organization. Those organizations that make identity salient through contract framing show that they are willing to discuss with their partner and are perceived as fairer and more altruistic than those that remain closed. Therefore, we expect that there is a positive relationship between contract framing and trust. This relationship is portrayed in hypothesis 1. (Rin k & Ellem ers, 2007)
H1: Contract framing (regardless of which social identity perception is induced) is positively related to trust between SA partners.
Expectations under contract framing
13
information based on which they can form clear expectations regarding the behavior which their SA partner is likely to show. In this sense, contractual framing means that safeguards and regulations not only help align incentives for both organizations, but they also aid in aligning expectations by facilitating clarity regarding future SA cooperation. Therefore, we expect a positive effect of contract framing on clarity of expectations, described in hypothesis 2.
H2: : Contract framing (regardless of which social identity perception is induced) fosters clarity of expectations regarding the cooperation among SA partners.
Expectations and trust among SA partners
14
expectations (i.e. expectation alignment) builds trust as it reduces uncertainty among alliance partners. Under these conditions, organizations are more willing to be vulnerable to opportunism, as they have greater confidence that their partner is able to cooperate in a manner fruitful to SA outcomes (Beccerra & Gupta, 1999; Miles, Snow & Miles, 2000; Inkpen & Currall, 2004; Mayer et al., 1995) fitting the definition of trust as per Mayer and Davis (1999). Concluding, we predict a positive direct effect of clarity of expectations on trust, which is portrayed in hypothesis 3. (Mulder, van D ij k, De Crem er, & Wilke, 2006)
H3: Clarity of expectations has a positive effect on trust between SA partners.
Combining the previous hypotheses, we can derive that the effect of contract framing on trust dynamics in SAs is twofold. First, contract framing fosters positive emotions in terms of goodwill and intent among the SA partners, which directly affects trust dynamics. Furthermore, contracts provide clarity among SA partner, which can help reduce uncertainty as the cooperation continues (reducing the likelihood of unwanted surprises). We therefore derive that clarity of expectations is an important mechanism in explaining the effects of contract frames on trust development, next to the direct effect of contract framing. When including contractual identity frames, organizations can help form clear expectations regarding future interaction, which in turn facilitates trust development among parties. The indirect effect of contract framing on trust, as mediated through clarity of expectations is hypothesized as follows:
H4: The positive relationship between contract framing and trust between SA partners is partially mediated by clarity of expectations.
Trust under collective and separate identity frames
15
with more positive emotions and judgements, and are regarded as being more likeable, honest and cooperative (Gaertner & Dovidio, 2000; Kane et al., 2005). Thus, when organizational members cooperate with members of another organization whom they perceive as in-group members, they are more likely judge those organizational members as trustworthy; they are more willing to open themselves up to vulnerability with the thought that the other party will not act opportunistically (Kane et al., 2005; Riek et al., 2010). As such, contracts that include collective social identity frames are more likely to facilitate trust between SA partners.
However, research into social identity has also shown that people who belong to social groups are more likely to show favoritism towards the in-groups, which in turn fosters fewer positive attitudes and judgements towards the out-group members (Ashforth & Mael, 1989; Chakravarty et al., 2016; Kane et al., 2005; Tajfel & Turner, 1986). Whereas people are more positive and accepting towards in-group members, they are more likely to be wary and distant towards group members. As such, when a SA partner is perceived as belonging to the out-group, organizational members are less inclined to form positive judgements regarding the interaction, being less likely to perceive the intent of the other organization to be in the best interest of the SA (Ashforth & Mael, 1989; Chakravarty et al., 2016; Mayer & Davis, 1999). Hence, contracts that include separate identity frames facilitate a degree of wariness among SA parties. All in all, based on the above reasoning, we predict that whereas contract framing (regardless of identity perception) fosters more trust than a baseline condition, there are differences between collective and separate identity perceptions, where contracts with a collective identity frame will foster more trust than contracts with separate identity frames. This relationship is portrayed in the following hypothesis (H5).
16 OVERVIEW OF THE PRESENT RESEARCH
The hypotheses presented in this paper were tested using two experimental designs (study 1a and 1b and study 2). Study 1a and 1b follow a cross-sectional design, serving as an initial test of the effects proposed in hypotheses 1 through 5. Specifically, Study 1a serves to examine the initial effects of identity framing on trust, as well as the indirect effect mediated by clarity of expectations. This study does so by means of a hypothetical scenario-based experiment, set out among participants via the Prolific Academic platform. Study 1b is a continuation of Study 1a, which serves to address the generalizability of the results by testing the hypothesized effects among alliance managers. The purpose of the second study design (study 2) is twofold, first Study 2 is meant to provide a replication of the effects tested in Study 1a and 1b, and to extend upon these studies with a behavioral design to measure trust. Additionally, because Study 2 uses a behavioral measure of trust, it is also meant to test how sustainable the effects of contract framing are over repeated interaction. To do so, study 2 specifically addresses the issue of how contract framing affect trust development under breaches in trust. In this way, Study 2 is meant to address the limitations provided by the design of Studies 1a and 1b. The second study was conducted in the FEB Research Lab, set out among students at the University of Groningen.
STUDY 1A: IDENTITY FRAMING AND TRUST DYNAMICS Participants and design
17
conditions1 , making sure that participants were distributed among conditions evenly: (1) collective
identity frame, (2) separate identity frame and (3) a neutral baseline condition.
Procedure
Study 1a presented participants with a hypothetical scenario, in which they were asked to imagine themselves working as an alliance manager for the company Print, a fictional company in the printer industry. The goal of the study was to immerse participants in a scenario where we could capture their trust and expectations under identity frame manipulations. Participants were informed that recently, Print had engaged in a strategic alliance with Head to develop a new product line. It would be the responsibility of the participant to represent the interest of Print in the SA. The scenario was formed in such a way that it would not be too difficult for participants to assess, but we also did not make the scenario too straightforward, our intention was to leave some room for wariness among participants. For a full description of the scenario, see Appendix A
Participants went through the following procedure: in the first part of the study, participants received the scenario described above for evaluation. The second part contained several questions regarding personality traits and characteristics. Once participants had been informed of their role in the scenario, they were presented with one of three contract conditions (see Appendix B ). Once participants had read the contract, they were asked to answer questions regarding their expectations of the alliance and to what degree they could identify themselves with their partner ‘Head’. All participants were then presented with the rest of the scenario, followed by several items to measure their trust in Head as a cooperation partner. Finally, participants were presented with the several descriptive items and supplementary variables, after which the participants were debriefed and thanked for their participation.
Manipulation
Social identity perceptions. Social identity perceptions were manipulated in two ways, by
means of the contract governing the SA as presented to the participants (see Appendix B for the contract manipulations) and indirectly through text in the scenarios. The first condition, collective
1 To test the contract manipulations, we ran a pre-test among students at the University of Groningen (N=67). This study included two manipulations of social identity
18
identity frame induced the identity perceptions in a direct manner, by including an identity clause to the contract that specifically stated both parties would function under a newly developed collective group identity. The second condition, separate identity frame functioned similarly to the first condition, however the identity clause specifically stated that both parties would retain their original, separate organizational identities. The third condition consisted of a neutral baseline contract without any identity manipulations, to check whether the responses that do include an identity manipulation yield significantly different results. In addition to the contract manipulations, we designed logos for each company to further induce the identity perceptions (see Appendix B). In the separate identity frame condition, participants were shown the separate logos for both Print and Head. The collective identity condition however included a common logo, that combined the old logos of Print and Head as one. These logos were included in the contract, the scenario descriptions as well as in the blocks of questions linked to the business case. The baseline condition did not include logos anywhere in the experiment. To test whether identity manipulations worked successfully among the participants, we included two robustness checks, rated on a 7-point Likert scale that ranged from strongly disagree (1) to strongly agree (7). Participants were presented with the following two items: (1) “In the alliance cooperation... each party held their own, unique work
values and norms” and (2) “In the alliance cooperation... both parties held similar work values and norms”. For item 1, results of a one-way ANOVA show that there are statistically significant
differences among the manipulation checks between collective and separate identity group means (F(2, 134) = 6.86, p = 0.00) however, post-hoc results also indicate that for both questions the
common identity condition does not differ significantly from the baseline condition (1: Mseparate =
5.48 (SD = 1.27), Mbaseline = 4.87 (SD = 1.29), Mcollective = 4.97 (SD = 1.31). Furthermore, item 2
shows no significant differences between conditions (F(2, 134) = 1.22, p = 0.29).
Measures
19
Trust. To measure our dependent variable trust, we used a 10-item measurement scale
adapted from Mayer and Davis (1999), capturing trust along the dimensions of ability, integrity and benevolence. An example of a statement related to ability-based trust: “I believe that head still
realizes their promises”. An item used to measure perceived benevolence-based trust: “I believe that HEAD will help PRINT cope with these schedule changes”. Finally, an example item used for
integrity: “I believe that HEAD will consider the effects of their decision on PRINT” (Cronbach’s Alpha = 0.92).
Clarity of expectations. Clarity of expectations was measured using a 3-item scale adapted
from Lee, Stajkovic and Bongsoon. (2011) and Swaab, Postmes, van Beest and Spears (2007). Items used for this variable are: “In the contract... - the rules regarding cooperation were clearly
formulated”, “In the contract... - the timeline for the project was clearly formulated” and “In the contract... - costs were clearly formulated” (Cronbach’s Alpha = 0.85).
Control variables. For the control variables, besides age and gender, we included two
individual level traits that have shown to impact the way in which people develop expectations and trust towards those around them: need for affiliation and disposition to trust. Need for
affiliation was measured using 5 items, adapted from Hill (1987) and Wiesenfeld, Raghuram &
Garud (2001). Participants were, among other items, asked to respond to: “I think being close to
others, listening to them, and relating to them is one of my favorite and most satisfying pastimes”
(Cronbach’s Alpha = 0.88). Finally, disposition to trust was measured using 6 items adapted from Wang, Ngamsiriudom and Hsieh. (2015). An example statement used: “I tend to count upon other
people” (Cronbach’s Alpha = 0.90). Correlation results showed that whereas none of the control
variables are correlated with trust, disposition to trust is positively correlated with clarity of expectations (B = 0.30, SE = 0.07, p = 0.00). These results indicate that people that have an inherent tendency to trust those around them experience clearer expectations when interacting with others. For the regression models, only disposition to trust was found to be positively related to clarity of expectations (B = 0.30, SE = 0.07, p = 0.00).
Results
Descriptive statistics and correlations. Table C1 (see Appendix C) reports the descriptive
20
Hypothesis testing. For the initial effect of contract framing, we proposed that contract
framing has a positive direct effect on trust (H1). Results of a one-way ANOVA of contract framing on trust shows that there are no statistically significant differences between contract framing and baseline condition group means (F(1, 135) = 1.18, p 0.28), meaning that trust under
contract framing (Mframing = 4.14, SD = 1.11), is not statistically different from that in baseline
condition (Mbaseline = 3.91, SD = 1.23). Additionally, a regression analysis using a dummy variable
for contract framing (with the baseline condition being the reference category) shows no significant evidence to suggest a direct effect of contract framing on trust (B = 0.21, SE = 0.21, p = 0.33). As such, we reject H1. H2 proposes that contract framing has a positive direct effect on clarity of expectations. One-way ANOVA results of contract framing on clarity of expectations show that there are no statistically significant differences between contract framing and baseline condition group means (F(1,135) = 0.01, p = 0.95). Clarity of expectations under contract framing
(Mframing = 5.44, SD = 1.09) does not differ statistically from the baseline condition (Mbaseline = 5.43,
SD = 1.25). Regression results also indicate no direct effect of contract framing on clarity of
expectations (B = 0.05, SE = 0.21, p 0.80). Hence, we reject H2. For hypothesis 3, we predicted that clarity of expectations facilitates trust, expecting a positive direct effect. Regression results however show no evidence to suggest a significant positive effect on trust (B = 0.10, SE = 0.09, p = 0.27). Therefore, we reject H3. The mediation effect (H4) predicts that the effect of contract framing on trust is mediated through clarity of expectations. Given the previous results of hypotheses 1 through 3, we did not find significant evidence to suggest a mediation effect to take place as both the relationship between IV and mediator as well as the mediator and DV need to be significant (Frazier, Tix, & Barron, 2004). These results are supported using the PROCESS macro for SPSS (Hayes, 2013), which show that there is no significant indirect effect of contract framing on trust through clarity of expectations (B = 0.01, SE = 0.03, p = 0.19, CI = (-0.06, 0.06)). Based on these results, H4 is rejected.
Finally, H5 predicts that there is a distinction in trust developed under collective and separate identity frames, stating that collective identity frames foster greater trust than separate identity frames. Results of a one-way ANOVA of identity framing (collective vs separate vs baseline) on trust show that there are no significant differences in trust between collective and separate identity group means (F(2,134) = 0.64, p = 0.53), meaning that trust under collective
21
identity framing (Mseparate = 4.10, SD = 1.27). Furthermore, regression results show no direct effect
of social identity perceptions on trust (B = 0.07, SE = 0.24, p = 0.76). Given these results, we reject
H5. For an overview of the regression results, see Table C2 (Appendix C).
Supplementary analyses. The results of the previous analyses provide no evidence to
suggest contract framing has a direct effect on trust, nor does it provide an indirect effect through clarity of expectations. However, upon further examination of our measurement items for clarity of expectation, we theorized that the lack of an indirect effect could be explained because not all three measurement items included adequately reflect clarity of expectations (i.e. identity perceptions does not influence all items, causing absence of significant effects). Therefore, for a supplementary analysis we singled out the first item of clarity of expectations “In the contract... -
the rules regarding cooperation were clearly formulated”, which most adequately captures our
theoretical proposition, and performed the mediation analysis accordingly. Following H2, results indicate a positive direct effect of contract framing on clarity of expectations, however this effect was marginally significant (B = 0.45, SE = 0.26, p = 0.09). Additionally, results show that clarity of expectations has a positive direct effect on trust, although also only marginally significant (B = 0.12, SE = 0.07, p = 0.08). Finally, using the PROCESS macro (Hayes, 2013), we tested for mediation as per H4, however these results failed to find a significant indirect effect of contract framing on trust through clarity of expectations (B = 0.05, SE = 0.05, p = 0.26).
Discussion
22
limited in scope, given that it consisted of only 3 measurement items. Therefore, we theorized that our lack of significant results for the indirect effect could be attributed to the fact that we did not fully capture our theoretical concept. To test this, we performed a supplementary analysis, where we singled out the measurement item best capturing clarity of expectations, which did provide initial (albeit marginal) support for a direct effect of contract framing on clarity of expectations, as well as a direct effect of clarity of expectations on trust; however, it failed to find an indirect effect. Hence, we decided that if we were to draw decisive conclusions, the next step would be to further expand our measurement of clarity of expectations, to better capture our proposed effect. Additionally, given the fact that the identity manipulations did not work as intended, for study 1b we decided to make adjustments to the scenario and the contracts to make the identity manipulations stand out stronger.
STUDY 1B: EXPECTATIONS UNDER IDENTITY FRAMES
The purpose of study 1b was to further examine the relationship between contract framing and trust development, as described in the discussion for Study 1a. In Study 1a, we did not find significant evidence to suggest that contract framing directly affect trust development, however we theorized that the clarity of expectations formed under social identity frames could provide useful insights into the mechanisms that underly trust development. Additionally, study 1b was conducted among managers familiar with alliances, to allow for generalizability of the results.
Participants and design
23 Procedure
The procedure for Study 1b was performed similarly to that of Study 1a: participants were asked to imagine themselves as a manager for Print in a SA with Head. Participants reviewed a hypothetical scenario with a contract, where we were specifically interested in their expectations and trust towards their SA partner under contractual identity framing. We manipulated identity perceptions by means of contract framing and company logos, as per Study 1a . All items used in this study were measured using a 7-point Likert scale that ranged from strongly disagree (1) to strongly agree (7). To test whether identity manipulations worked successfully among the participants, we included two robustness checks (as per Study 1a), rated on a 7-point Likert scale that ranged from strongly disagree (1) to strongly agree (7). Participants were presented with the following two items: (1) “In the alliance cooperation... each party held their own, unique work
values and norms” and (2) “In the alliance cooperation... both parties held similar work values and norms”. Results for both items however show that there are no significant statistical
differences between the collective and separate identity group means, nor the baseline condition, indicating that there are no significant differences in identity perceptions between conditions (1: (F(2, 54) = 1.29, p = 0.28), 2: (F(2,54) = 0.35, p = 0.71).
Measures
Trust. To measure trust, we used the 10-item measurement scale as used in Study 1a,
adapted from Mayer and Davis (1999), capturing the dimensions of ability, integrity and
benevolence. An example of a statement related to ability-based trust: “I believe that head still realizes their promises.” An item used to measure perceived benevolence-based trust: “I believe that HEAD will help PRINT cope with these schedule changes.” Finally, an example item used for
integrity: “I believe that HEAD will consider the effects of their decision on PRINT” (Cronbach’s Alpha = 0.82).
Clarity of expectations. Clarity of expectations was measured using an 8-item scale
adapted from Lee, Stajkovic and Bongsoon. (2011) and Swaab, Postmes, van Beest and Spears (2007). Example items used for this variable are: “In the contract... - the rules regarding
24 you feel about the collaboration with HEAD - I understand how to approach HEAD during the collaboration” (Cronbach’s Alpha = 0.87).
Control variables. The same control variables as in Study 1a were included (i.e. age,
gender, need for affiliation and disposition to trust). Need for affiliation followed the measurement scale from Study 1a, using 5 items adapted from Hill (1987) and Wiesenfeld , Raghuram & Garud (2001). Participants were, among other items, asked to respond to: “I think
being close to others, listening to them, and relating to them is one of my favorite and most satisfying pastimes” (Cronbach’s Alpha = 0.78). Finally, disposition to trust was measured using
5 items adapted from Wang, Ngamsiriudom and Hsieh. (2015). An example statement used: “I
tend to count upon other people” (Cronbach’s Alpha = 0.86). None of the control variables were
found to be significantly correlated with clarity of expectations or trust, however gender was found to be positively correlated with clarity of expectations (r = 0.26, p = 0.03), indicating that women in general were more likely to indicate having formed clear expectations regarding the SA.
Results
Descriptive statistics and correlations. Table D1 (see Appendix D) reports the descriptive
statistics (means, standard deviations) and zero-order correlations among the variables included in the study. The coefficient alphas are presented on the diagonal in parentheses.
Hypothesis testing. First, we tested the mediation effect as proposed in hypotheses 1
through 4. A one-way ANOVA of contract framing on trust shows no significant differences
between the contract framing and baseline group means (F(1, 55) = 2.29, p = 0.14) meaning that
trust under contract framing is not significantly different than under the baseline condition (Mframing
= 2.79 (SD = 0.68), Mbaseline = 3.13 (SD = 0.99)). A regression analysis using a dummy variable
for contract framing (with the baseline condition being the reference category) further indicates no direct effect of contract framing on trust (B = -0.35, SE = 0.24, p = 0.13). Hence, H1 is rejected. The second model, using clarity of expectations as the outcome variable shows with a one-way ANOVA that there are differences in clarity of expectations between the contract framing and baseline condition (F(1, 55) = 4.93, p = 0.03). Clarity of expectations under contract framing
differs marginally from that under the baseline condition (Mframing = 3.42 (SD = 1.07), Mbaseline =
25
Clarity of expectations was found to be positively related to trust, however this effect was marginally significant (B = 0.19, SE = 0.11, p = 0.09), showing support for H3. Testing for mediation (H4) we used the PROCESS macro for SPSS (Hayes, 2013). However, these results shows no significant indirect effect of contract framing on trust through clarity of expectations (B = 0.16, SE = 0.10, p = 0.11, CI(0.03, 0.40)). Again, these results are not surprising given that we did not find a significant direct effect of the IV to the mediator, a prerequisite for mediation to take place (Frazier, Tix & Barron, 2004). Based on these results, H4 is rejected. Lastly, H5 predicts that a collective identity frame fosters greater trust than a separate identity frame. However, results of a one-way ANOVA show that there are no significant differences in trust between collective and
separate identity frames (F(1, 36) = 0.01 p = 0.95; Mcollective = 3.48 (SD = 1.12), Mseparate = 3.38
(SD = 1.05)). Furthermore, regression results also indicate no direct effect of identity perceptions on trust (B = -0.12, SE = 0.30, p = 0.70). Therefore, H5 is rejected. For an overview of the regression results, see Table D2 (Appendix D).
Discussion
26 STUDY 2: IDENTITY FRAMING UNDER REPEATED INTERACTION
Study 1a and 1b failed to find conclusive support to suggest that contract framing can build trust by setting clearer expectations regarding the cooperative behavior of the other party. However, these results are limited due to restrictions in the design of both studies which could explain a lack of significant effects. Study 2 sets out to address the limitations of study 1a and 1b, by replicating the conceptual model using a behavioral measure of trust. Given that studies 1a and 1b make use of self-report measures, this leaves open room for retrospective bias among respondents, and answers can be affected by more recent events that are recalled during the study (Nisbett & Wilson, 1977; Searle, 2015). In addition, whereas trust is conceptualized as a “willingness to be vulnerable to actions based on another party” (Mayer & Davis, 1999), self-report measures do not always directly capture the perceived vulnerability, as there is no direct threat to the position of the respondent. Hence, literature argues that self-report and behavioral measures of trust do not always correlate (Tanis & Postmes, 2005; Weber & Bauman, 2017). Thus, whereas studies 1a and 1b provide an initial examination of the hypothesized effects, they can be further complemented by an alternate experimental design that specifically addresses these weaknesses (to be discussed further below). Finally, whereas the hypotheses predict initial positive effects of contract framing in building trust and cooperative behavior, the question remains as to how sustainable these effects are over time. To test this, in Study 2 we will also theorize and empirically examine how firms respond to a negative event (i.e. a breach in trust) under different social identity frames.
Negative behavior under identity frames
27
likeable and cooperative, and will set their expectations regarding the relationship accordingly (e.g. Burgoon & Burgoon, 2001; Burgoon & Walther, 1990; Tajfel & Turner, 1989). However, under a separate identity frame firms are more likely to view one another with distance and wariness. Since the other party is seen as an outsider, organizational members will adjust their expectations for the cooperation accordingly, which from SIT we have seen is more distant than under a collective frame (Ashforth & Mael, 1989; Kane et al., 2005; Riek et al., 2010).
28 H6: When experiencing a negative event in a cooperative relationship, parties under collective identity frames will show a greater fall in trust than parties that experience separate identity frames.
Participants and design
Study 2 was set out in the Research Lab, among students at the Faculty of Economics and Business at the University of Groningen. The study provided a total sample of 273 respondents, which upon reviewing attention checks and missing data left us with a final sample of 261 (121 male, 140 female). The average age among the respondents is 21 years, with a minimum age of 18, a maximum age of 35 and a SD of 3 years. Most of the respondents stemmed from the Netherlands (99), Germany (25) and Indonesia (26). The study was constructed using the Qualtrics survey platform and was distributed in English. Participants were informed that with this experiment we were interested in cooperation behavior, and that for this experiment they would be performing a cooperation task. Participants were randomly assigned to one of three conditions, making sure that participants were distributed evenly: (1) shared identity frame, (2) separate identity frame and (3) a neutral baseline condition.
Procedure
29
Once participants entered the cubicle, they were presented with the background information on Print, Head and the alliance as well as the rules of the trust game. In addition, participants were told that they would be randomly paired with another participant, who would make decisions for the other company, however their identity would remain anonymous, even after completion of the experiment. In reality, the responses of the other participant were pre-programmed. Given that in this experiment we are interested in trust development, all participants were assigned as the manager for Print, where the responses for Head were pre-programmed. Having read the scenario and the contract, we measured participants’ expectations regarding the cooperation, after which they engaged in the trust game. Once all five rounds of the trust game were played, participants were asked to fill in an additional self-report measure of trust, as well as several items measuring individual characteristics. Finally, participants were asked several open questions with suggestions to further improve the set-up for this study and were subsequently debriefed and thanked for their participation. Participants were rewarded either money or research points for their participation.
Manipulations
Social identity perceptions. We manipulated identity perceptions by means of contract
framing and company logos, as per Study 1a and 1b. Participants were shown a version of the contract right before starting round 1 of the trust game, and once after having played two rounds as a reminder, to re-prime the identity frame. To test whether identity manipulations worked successfully among the participants, we included two robustness checks (as per Study 1a and 1b), rated on a 7-point Likert scale that ranged from strongly disagree (1) to strongly agree (7). Participants were presented with the following two items: (1) “In the alliance cooperation... each
party held their own, unique work values and norms” and (2) “In the alliance cooperation... both parties held similar work values and norms”. For item 1, results of a one-way ANOVA show that
there are statistically significant differences among the manipulation checks between collective and separate identity group means (F(2, 258) = 7.25, p = 0.00) however, post-hoc results also indicate that for this item the collective identity condition does not differ significantly from the
baseline condition (Mseparate = 5.38 (SD = 1.17), Mbaseline = 4.76 (SD = 1.30), Mcollective = 4.72 (SD
30
Breach in trust. To measure hypothesis 6, we induced a negative event during the trust
game as follows. As stated, participants took the role of Print and were tasked to decide upon the investment in the strategic alliance each round. Participants were informed that the manager of Head would decide upon the distribution of the payout after each round. We programmed the responses of Head and used these responses to induce a breach in trust at the end of the third round. For the first two rounds, participants received an equal distribution (i.e. resources were split 50/50 between Print and Head). However, after participants had invested their resources in the third round, they were informed that the manager of Head decided to take most of the resources for themselves (i.e. participants only received back their initial investment). Overall, participants show a significant drop in investments between round 3 (M = 7.68) and 4 (M = 5.10), meaning that we did successfully induce a breach in trust (t(260) = 12.74, p = 0.00). For rounds 4 and 5, participants were presented with equal payouts, to examine how trust develops after a negative event.
Measures
The measures used in Study 2 were similar to those in Study 1a and 1b, the main difference being that trust was measured both using a self-report measure (similar to Study 1a/1b) as well as a behavioral measure. For the behavioral measure of trust, we examined the development of alliance investments in rounds 1 and 2 under different contract frames: higher investments in the alliance meaning a higher display of trust (Malhotra & Murninghan, 2002; Weber & Bauman, 2017). The impact of negative events under contractual identity frames was capture by examining the change in alliance investments between rounds 3 and 4. Finally, rounds 4 and 5 captured the development of trust after a negative event. All measures in this study were answered on a 7-point Likert scale that ranged from strongly disagree (1) to strongly agree (7).
Trust. In addition to the behavioral measure of trust, we also included a self-report
measure, the 10-item measurement scale as used in Study 1a (adapted from Mayer & Davis, 1999), including dimensions of ability, integrity and benevolence. An example of a statement related to ability-based trust: “I believe that head still realizes their promises”. An item used to measure benevolence-based trust: “I believe that HEAD will help PRINT cope with these schedule
31
Clarity of expectations. Clarity of expectations was measured using an 8-item scale
adapted from Lee, Stajkovic and Bongsoon. (2011) and Swaab, Postmes, van Beest and Spears (2007). Example items used for this variable are: “In the contract... - the rules regarding
cooperation were clearly formulated”, Please indicate how you feel about the collaboration with HEAD - I have a good idea of what to expect from the collaboration” and “Please indicate how you feel about the collaboration with HEAD - I understand how to approach Head during the collaboration” (Cronbach’s Alpha = 0.67).
Control variables. The same control variables as in Study 1a and 1b were included (i.e.
age, gender, need for affiliation and disposition to trust). Need for affiliation followed the measurement scale adapted from Hill (1987) and Wiesenfeld, Raghuram & Garud (2001). Participants were, among other items, asked to respond to: “I think being close to others, listening
to them, and relating to them is one of my favorite and most satisfying pastimes” (Cronbach’s
Alpha = 0.70). Finally, disposition to trust was measured using 5 items adapted from Wang, Ngamsiriudom and Hsieh. (2015). An example statement used: “I tend to count upon other people” (Cronbach’s Alpha = 0.87). With regards to the regression models, gender was found to be marginally correlated with trust, indicating that women overall are more likely to show trust in the experimental context (r = 0.12, p = 0.05). Furthermore, disposition to trust was found to be significantly correlated to clarity of expectations, suggesting that people with an inherent likelihood to trust those around them experience clearer expectations when cooperating with others (r = 0.22, p = 0.00).
Results
Descriptive statistics and correlation. Table E1 (see Appendix E) reports the descriptive
statistics (means, standard deviations) and zero-order correlations among the variables included in the study. The coefficient alphas are presented on the diagonal in parentheses.
Hypothesis testing (cross-sectional). First, the direct effect of contract framing on trust
32
contract frames do not perceive higher trust than those under the baseline condition (Mframing = 4.14
(SD = 0.96), Mbaseline = 4.04 (SD = 0.88)). Given these results, H1 is rejected. Additionally,
following H2 regression results show no direct effect of contract framing on clarity of expectations (B = 0.04, SE = 0.09, p = 0.68), providing no support for H2. Third, clarity of expectations shows a positive direct effect on trust (H3), however this effect was only marginally significant (B = 0.16,
SE = 0.09, p = 0.07). To test the mediation effect (H4), the PROCESS macro for SPSS was used
(Hayes, 2013), which showed no evidence to suggest an indirect effect of contract framing on trust (B = 0.02, SE = 0.02 , CI(-0.01, 0.07)). Finally, H5 predicts that collective identity framing fosters greater trust than separate identity framing. Results of a one-way ANOVA show that there are no significant differences in trust between collective and separate identity frames (F(1, 172) = 1.73 p
= 0.19; Mcollective = 4.24 (SD = 1.01), Mseparate = 4.04 (SD = 0.90)). Furthermore, regression results
also indicate no direct effect of social identity perceptions on trust (B = 0.17, SE = 0.0.15, p = 0.24). Therefore, H5 is rejected. For an overview of the regression results, refer to table E2 (see Appendix E).
Supplementary analysis (behavioral). Figure E1 shows SA investment per round as a
function of contract condition. As stated, Study 2 uses an experimental design with a behavioral measure of trust to test our conceptual model. To capture the effect of contract framing on trust, the levels of investment in round 1, as well as between round 1 and 2 (before the introduction of the breach in trust) are compared. We predicted that, compared to baseline conditions, contract framing fosters greater trust in the initial condition (round 1) and would show a stronger growth in investments between rounds 1 and 2 (H1). However, results of a one-way ANOVA show that there are no significant differences between contract framing and the baseline condition in round
1 investment (F(1, 259) = 0.53, p = 0.47; Mframing = 6.35 (SD = 1.63), Mbaseline = 6.20 (SD = 1.59)).
33
under separate identity frames. This effect was tested by examining whether there was a difference in investment between rounds 3 and 4 between collective and separate identity conditions. Overall, participants show a significant drop in investments between round 3 (M = 7.68) and 4 (M = 5.10),
t(260) = 12.74, p = 0.00. However, we do not find evidence to suggest that this difference is
dependent on collective or separate identity frames (F(1, 172) = 0.29, p = 0.59). Additionally, we examined the effects of contract framing after a negative event, by examining how trust develops between round 4 and 5. Results show that overall, participants show a significant increase in SA investment between round 4 (M = 5.10) and 5 (M = 6.09). However, there is no significant effect of identity framing on trust development between rounds 4 and 5 (B = 0.13, SE = 0.33, p = 0.69). Therefore, the results do not find significant evidence to support the effect of contract framing on breaches in trust as per H6.
Discussion
34 GENERAL DISCUSSION
This study set out to further the understanding of the use of contractual and relational governance in SAs by specifically examining how contracts can impact the relational quality of SAs through contract framing. Extending the work by Weber and Mayer (2011) and Weber and Bauman (2017) we proposed that by phrasing elements of a contract in different ways, contracts can induce psychological effects that are separate from its economic consequences. Drawing form SIT, our conceptual model proposed that contract framing is related to trust among SA partners in that it helps foster clarity of expectations. Whereas we proposed that the overall effect of contract framing (regardless of social identity perception) is positive compared to a baseline contract, we did argue that in terms of trust development, the effect of contract framing can be further nuanced by looking at the distinction between collective and separate identity frames. Specifically, we predicted that collective identity frames foster more trust among SA partners than separate identity frames.
To test the hypotheses predicted in the conceptual model, a series of three experimental studies were conducted. Overall, these studies showed no evidence for the effects of contract framing on trust dynamics or on clarity of expectations. First, looking at the direct effect of contract framing on trust (H1) all three studies failed to show any support for the proposed effect. For the indirect effect (H2-H4) we first examined the effect of contract framing on clarity of expectations
(H2). Given our results, we did not find evidence to suggest that contract framing fosters clearer
35
a mediation effect to take place as predicted per H4. Finally, building on SIT our conceptual model predicted that there are differences in trust development based on whether organizational members perceive a collective or separate identity (H5). Contrary to our expectations, neither of the experiments found there to be differences in trust based on whether a collective or a separate identity frame was included. Whereas these results are surprising with relation to our own predictions as well as the general propositions of SIT, they can be attributed to the fact that overall our results showed that our operationalization of contract framing was not sufficiently strong to manipulate identity perceptions among participants. Altering the phrasing in the contract in combination with separate or collective logos did not prove to be sufficient stimuli to make social identity salient among participants.
To test our initial conceptual model proposed in the theoretical section, our experiments used a cross-sectional design. Whereas a cross-sectional design can function to capture the initial effects of contract framing, it does not fully replicate the effects of framing in terms of behavior (Tanis & Postmes, 2005; Weber & Bauman, 2017). For this reason, we decided to add a behavioral design in study 2, which allows to test the direct behavioral impact of contract framing. Besides this, study 2 also tested whether the effects of contract framing are sustainable on the long term, examining how contract faming affects people’s response to a breach in trust. Using a modified version of the ‘trust game’ (Malhotra & Murnighan, 2002), results however did not find any further evidence to support the hypotheses regarding the effect of contract framing on trust through clarity of expectations. Additionally, whereas the results did show a behavioral response to the breach in trust, there was no significant evidence to suggest that this response was affected by contract framing nor clarity of expectations. We will discuss our reasoning as to why we found these results under the limitations and future research section.
Theoretical implications
36
illustrate how experimental research designs could supplement existing research methods in the strategy field.
37
The second contribution of this study is that by taking a social identity approach to contract framing, we show how incorporating psychological processes in strategy literature can develop a more complete understanding of contract usage in SAs. Whereas we built on previous work by Weber and Bauman (2011), no other research has addressed the effects of framing in terms of contract usage. Therefore, examining the psychological consequences of contracts through language and phrasing could open fruitful new avenues of research. However, in contrast to the work of Weber and Bauman (2017), our work shows that the effects of phrasing and language, when taking a social identity perspective, are more limited than previously has been theorized. For example, in testing the initial manipulations for contractual identity framing, our pre-study showed that merely inducing social identity by changing how contractual elements are phrased (as proposed in Weber & Mayer, 2011) failed to induce social identity perceptions. Even when manipulating identity perceptions more directly (i.e. by specifically addressing them in contractual identity clauses, in addition to overall phrasing) our research at best found mixed results regarding the effects of contract framing in comparison to a baseline condition.
38
Our third main contribution to the field of strategy is that this research in its methodological design shows how different methodological approaches could be used to complement existing research in the field. Using an experimental design, we were able to draw conclusion regarding our conceptual model with a degree of accuracy that is difficult to achieve had we used more common methods in the field of strategy research (Croson, Anand & Agarwal, 2007; Weber & Mayer, 2017). For example, attempting to examine the effects of contractual identity framing as proposed in our conceptual model would have been difficult were we to use a large secondary database. Therefore, we attempt to show that by using both experimental and traditional research approaches we can further complement our understanding of the nature of contract usage in inter-firm relationships.
Limitations and future research directions
As with any study, the approach we took to test our hypotheses has certain limitations. First, our main conceptual model was tested using a cross-sectional design. One of the core limitations of using a cross-sectional design is that it only captures a single point in time. Whereas this design was used due to restrictions in scope of this study (i.e. it is difficult to capture longitudinal data on trust in inter-firm relationships), this does mean that it is difficult to capture the phenomenon, given that trust typically builds up over time; through repeated exchanges (e.g. Dyer & Singh, 1998; Mayer & Davis, 1999, Inkpen & Currall, 2002). Furthermore, by using self-report measures of trust and expectations, our study design left open room for retrospective bias among respondents; in that answers may be affected by more recent events recalled during the study (Nisbett & Wilson, 1977; Searle, 2015). Therefore, whereas the cross-sectional approach to test our conceptual model can provide initial results for certain effects, it does not fully capture how contractual frames would operate in a real-world scenario.
39
due to the absence of direct personal consequences, participants paid less attention to the scenario and the contract, which can explain why the identity manipulations did not work as intended. An additional difficulty here was that the majority of our participants (i.e. for Study 1a and 2) did not have a background specifically related to strategic alliances. Hence, not only did the participants have to acquaint themselves with the hypothetical scenario, but they also needed to familiarize themselves with the SA context. As we have concluded that the effect of contract framing in inducing social identity is fairly subtle, the additional complexity added for the participants could have suppressed our manipulation effects.
To address some of these limitations, we performed study 2, which included a behavioral measure of trust through an adapted version of the ‘trust game’ (Malhotra & Murnighan, 2002; Weber & Bauman, 2017). However, whereas this study was meant to address the limitations of Study 1a and 1b, the behavioral measure of trust did not provide any additional insights as to the workings of contract framing. A possible explanation as to why Study 2 did not provide the intended results lies within our adaptation and operationalization of the trust game. One of the core differences between the trust game as per Malhotra and Murnighan (2002) and our study is in the way that trust (i.e. a willingness to open up to vulnerability from another party) was induced. The original trust game by Malhotra and Murnighan (2002) had participants decide whether to invest their resources in a joint venture, or to keep their resources to themselves. Resources invested internally would grow at a steady, be it slow, rate: resources invested in the JV would grow at a substantially higher rate, however investing in the JV made participants vulnerable since the other player would get to decide how (and if) to distribute the resources (Malhotra & Murnighan, 2002; Weber & Bauman, 2017). Hence, by having participants decide either between a steady growth individually, or a risky joint investment, this trust game directly captures the dimension of vulnerability and loss.
