Managing Difficult Workplace Conversations

Abstract

Many conversations involve sending or receiving “bad news.” These conversations are often dreaded, poorly executed, or avoided altogether. Ways need to be found to make them less difficult and more productive. We explored these issues through three methodologically diverse studies. Study 1 comprised in-depth interviews with 24 nurse managers. Interviews shed light on the characteristics of difficult conversations and strategies for making them less awkward and more successful. Study 2 was a survey investigating relationships between six dimensions of supportive communication and participant satisfaction with a difficult superior-subordinate conversation. Study 3 experimentally manipulated two supportive communication behaviors, plus a third variable, face-work. Together, these studies show that successful outcomes from difficult workplace conversations require the parties to balance task and relational goals, with the latter particularly dependent on acts of empathy and face-giving.

Keywords

difficult conversations supportive communication defensiveness face-work management communication

Managing Difficult Workplace Conversations

Interacting with other people is not always easy. Over their working lives, most individuals participate in many difficult conversations. Such conversations involve topics that are “sensitive,” or hard to talk about, often because they involve the transmission or receipt of “bad news.” They occur in many contexts including, perhaps most obviously, as part of performance appraisal interviews. But they also occur in less preplanned ways, and in organizational directions other than downwards. Examples include supervisors telling employees they are to be transferred or their job is to be restructured; subordinates telling their managers that they want a raise, or that they feel undervalued or bullied; one coworker telling another about the latter’s irritating personal habits; and long-term customers telling suppliers that they are taking their business elsewhere. Participants in these conversations often anticipate feeling uncomfortable and, as a consequence, the conversations are frequently delayed, or avoided altogether, a phenomenon known as the “mum” effect (Tesser & Rosen, 1975). In each case, when the conversation does eventually take place, poorly executed communications can have a range of adverse worker and organizational outcomes (Geddes & Baron, 1997; Kluger & DeNisi, 1996).

Despite their ubiquity, difficult conversations have been the subject of relatively little empirical research. In this article, we report three studies of difficult conversations: an interview-based study involving 24 nurse managers, a survey of 137 individuals’ experiences, and an experiment that used written scenarios to manipulate key conversation variables. Our focus was restricted to conversations between fellow employees rather than with customers and clients. The three studies share a common quest to discover why these conversations are difficult and what can be done to make them less so. Our research provides the basis for theoretical and practical advances in the understanding and management of difficult conversations.

Difficult Conversations

Over the past decade, several popular books have been published dealing with the topic of difficult conversations (e.g., Cloud & Townsend, 2005; Dickson, 2006; Kosmoski & Pollack, 2005; MacDonald, 2004). Our own Internet search revealed dozens of websites referring to training materials, media packages, and/or motivational speakers that offer information and advice for managing difficult workplace conversations. Typically, these books and sites offer testimonial or case study information in support of their recommendations, but none appears to be grounded in theory or based on sound research evidence.

In contrast to this abundance of advice, the academic literature that refers explicitly to “difficult conversations” is limited. However, a search using alternative terms and conducted across multiple disciplines reveals much within the fields of human resources management, employment relations, organizational behavior, social psychology, applied linguistics, and communication science that has clear implications for the conduct of difficult conversations. For example, the literature pertaining to performance appraisal, performance management, workplace negotiation, and workplace conflict bears on some aspects of the topic (e.g., Brown, Hyatt, & Benson, 2010; Clifton, 2012; De Dreu & Gelfand, 2008; Deutsch, 1994), as does research into adverse interpersonal treatment in the workplace, including bullying, incivility, and aggression (Baron, 1988; Geddes & Baron, 1997; Madlock & Kennedy-Lightsey, 2010). Taken together, this literature makes clear that, while communication is a key determinant of organizational effectiveness, many workplace conversations are both difficult and poorly handled.

A widely studied business context requiring, but often not delivering, high-quality communication involves the provision of negative feedback from superior to subordinate. Whether as part of a formal appraisal interview or a more ad hoc exchange, this task can be viewed as a difficult conversation, and has been shown to lead to several adverse consequences including negative employee cognitions and affective reactions, damaged relationships, reduced work performance, and employee resistance and retaliation (Baron, 1988; Geddes & Baron, 1997; Ilgen & Davis, 2000). The potential for supervisor feedback to adversely affect employee behavior and organizational outcomes in these ways is thus a key managerial concern. Limited theory (e.g., Brown & Levinson, 1987; Kluger & DeNisi, 1996) and organizationally based research (e.g., Baron, 1990; Lizzio, Wilson, & MacKay, 2008; Tata, 2002) has provided clues as to specific processes and message features that may be critical to communication success. This research shows that adverse impacts can be attenuated by, for example, presenting feedback that is specific, presenting it in a tentative manner, offering recipients an opportunity to voice any mitigating circumstances, and actively listening to their responses.

For the second and third of the current studies, we drew not only on this human resources management and organizational behavior literatures, but also on those pertaining to assertive communication, attribution biases and processes, organizational justice, persuasion, reactance, self-presentation, social interaction processes, and many more (see, e.g., Brehm & Brehm, 1981; Brown & Levinson, 1987; Cialdini, 2001; Gibb, 1961; Goffman, 1967; Weiner, 1985). But, given the dearth of theory and empirical research directly investigating the topic of work-based difficult conversations, we began with an exploratory study. This study adopted an inductive or open-ended approach, rather than being guided by, or limited to, a single theoretical focus or set of hypotheses.

Study 1: Exploring the Difficult Conversations Experienced by Nurse Managers

Study 1 investigated difficult conversations from the perspective of a sample of middle and senior-ranked nurses. We sought answers to questions regarding when and why these conversations occur, who is usually involved, what issues are commonly at stake, how the conversations typically unfold, and how they can best be managed.

Study 1 Method

Participants

The study involved anonymous, individual interviews with a volunteer sample of 24 experienced nurses. While the topic of difficult workplace conversations is relevant to most occupational groups, the sample was drawn from the nursing profession because nurses typically work within complex, multilayered bureaucracies, under conditions of considerable time, load, and emotional pressure, where difficult conversations are likely to occur frequently and with potentially grave consequences (Meyer et al., 2009). All but one of the interviewees was female. All were experienced workers (aged 35–60+ years), and most had substantial supervisory responsibilities. Position titles included Director of Nursing, Assistant Director of Nursing, Nurse Unit Manager, Nurse Educator, and Clinical Nurse Consultant.

Procedure

Interviews were semistructured and lasted between 30 and 50 minutes. An information sheet was provided to all respondents, and signed consent was obtained. Interviews were digitally recorded and transcribed. Participants were given a copy of their interview transcript and were invited to correct or retract any part of it. None elected to do so.

Analysis of interview content

We examined the interviews using a content analysis approach (Neuman, 2011). We sought to produce high-quality coding without resort to preconceived ideas as to the themes likely to be extracted. To begin, a single coder read all transcripts and grouped the content into one of six categories, namely, content pertaining to (a) personal details of the interviewees; (b) features of the difficult conversations experienced; (c) anticipated or actual consequences of the conversations; (d) strategies (not) used and/or recommended for handling difficult conversations; (e) external/contextual factors affecting the causes of, processes in, and outcomes from, difficult conversations; and (f) other.

Using eight initially coded interviews, a second coder constructed an embryonic coding structure within each of the six initial, broad categories. This coder then read all remaining transcripts and re-played audio-recordings of key interviews, progressively coding content into emerging themes and refining the scheme as coding progressed. Text word searches were also conducted to check on coding. When all interviews were coded, the first coder checked the integrity of the coding. And the two coders discussed any discrepant cases against the final coding structure. Following Miles and Huberman (1994), reliability was defined as the number of agreements divided by the total number of instances to be coded, and discussion continued until a reliability criterion of 95% was achieved. Through these processes, the coders identified more than 100 specific content categories that were subsumed within the six categories used by the first coder. For reporting purposes, we have organized these content categories into two parts: those pertaining to the problems caused by difficult conversations, and those pertaining to possible solutions to these problems.

Study 1 Findings

The nature of the “difficult conversations problem.”

Content for this section came from interviewee responses to such questions as:

How often, if at all, do you have difficult conversations at work? With whom?

What kinds of issues or topics have been under discussion in such conversations?

Why are they so difficult? What has been the outcome of these conversations?

All interviewees reported experiences of difficult workplace conversations. The frequency of such conversations ranged from rarely to daily, with one interviewee opining that nearly all conversations are difficult—it just depends on the degree. Frequency tended to increase with workloads and employee stress levels.

It depends on how busy and sort of frantic and intense the mood is. . . . More often than not, you get those clusters. . . . You have them at busy times, they go cyclic.

Most of the reported conversations were initiated by the interviewee, and most were part of a scheduled meeting with a single member of their nursing staff. More rarely reported were difficult conversations (a) initiated by another person, (b) with superiors, (c) with members of other professions, and/or (d) with groups rather than a single individual. The topic discussed was most often a subordinate’s (unsatisfactory) performance, including non-adherence to clinical and hospital procedures, inappropriate interpersonal manner, and poor work attendance. Many involved delivery of negative feedback or unwanted directions; some were triggered by a complaint about an employee from patients or their families.

All interviewees, even the most experienced, expressed a dislike of these conversations. They described a range of responses: uneasiness, anger, frustration, churning in the stomach, feeling drained and stressed, lost sleep, and reduced self-confidence. Specific comments included:

I would come out shaking.

It’s very draining emotionally.

I just felt I am useless.

Adverse effects frequently extended to the participants’ work team and/or other stakeholders. The quality of patient care was reported to be sometimes compromised.

When asked why the conversations were so difficult, many responses related to the risk that the communication would (a) hurt the other person, (b) result in resistance and/or retaliatory action from that person, (c) have adverse consequences for the initiator, (d) damage the relationship between the parties, and/or (e) adversely affect other interests and stakeholders. Illustrative responses regarding the sources of conversation difficulty were:

Because I have to tell them something negative. I have to point out to them that their performance is not quite as shiny as they think it is, and that can be hard to hear.

Sometimes it’s hard to know if people are mentally well enough to have these conversations. I would be loathe to be the person that tipped someone over the edge, and they did some self-harm or something like that. So that’s a tricky situation.

They make it personal. . . . They know they’re sinking, and they’re scrambling. So they are attacking on the way down. I understand that, but I find it difficult to deal with.

Some of the conversations had been held many times before, even over a period of years, without a satisfactory resolution. In these cases, the other person typically refused to cooperate or change their behavior. The interviewees’ prior experience of these conversations suggested that any bad news they sought to impart would be greeted with denial or defensiveness, and their directions for behavior change would be met with dissent and resistance.

Often when people get very prickly about it, and resistant, it is difficult to get co-operation to move people in the direction that I think they need to move in [so as] to become competent and efficient and safe. . . . Generally their attitude to me becomes quite resistant and they can become quite subversive.

Perceptions of having limited power, fears of aggressive counterattack, and concerns about being unjustly accused of bullying, harassment, or victimization were widely held and strongly voiced. For example:

These people who have these challenging behaviors mostly go into denial about them and then the first thing is, “You’re bullying me,” or “You’re harassing me.”

In sum, the interviewees described the task of having difficult conversations as daunting and demanding. Reasons included the uncertain, but likely adverse, outcomes at stake, the need for well-developed and carefully enacted communication skills, doubts as to their capacity to execute these skills, and the operation of a myriad of personal, organizational, and industrial demands and constraints upon their actions.

Solutions to the “difficult conversations” problem

A central focus of the study was on the processes that occur within difficult workplace conversations, and particularly on the strategies used, and recommended for use, to achieve desired outcomes. To these ends, the interviews included questions such as:

How do you manage difficult conversations? Why do you do that?

How well do you handle these conversations? What do you do or say that works well? What do you do or say that does not work so well?

What can individuals do to improve the process and facilitate a better outcome for those involved? How can difficult conversations be made “easy”?

We grouped the responses to these and similar questions into a set of 24 recommendations regarding how best to manage difficult workplace conversations. They are presented in Figure 1 under three broad headings reflecting a temporal sequence from preparing for the conversation, through conducting the conversation, to concluding the conversation.

Figure 1.

The 24 recommended strategies for managing difficult conversations.

We further divided the second category of strategies into two groups:

those that are primarily concerned with minimizing the negative consequences of a difficult conversation, that is, “damage control” (i.e., strategies 7–11), and

those that are primarily concerned with maximizing the positive outcomes from conversations, including enhancing the likelihood of sustained attitudinal and/or behavioral change (i.e., strategies 12–20).

Space limitations prevent elaboration of the full set of recommended strategies, but we have provided quotations to illustrate the kind of interview content used as a foundation for three strategies. For example, the following is one of many comments that led to the formulation of recommendation 3 (Decide on goals and prepare a strategy):

So it’s about having to really be prepared before going in. I remember I actually wrote that conversation out, and politely asked that person to allow me to finish before responding and not to cut me off. I structured what I needed to do and practised it a couple of times. That person’s body language was very intimidating … which was really off-putting, and I was very happy that I’d written it all out.

The following is indicative of recommendation 10 (Word concerns and criticisms carefully):

It’s a matter of trying to get the message across but without becoming threatening or being perceived as being bullying or aggressive. . . . So getting that message across . . . picking the words, it’s an art that you have to learn. You want a win/win situation.

Similarly, we formulated recommendation 12 (Give the other person voice) on the basis of comments such as the following:

You have to be open. It has to be a conversation. . . . They’ve got to think that they have input, because if they don’t, that’s the end of the conversation. . . . They’ll tell you what you want to hear, and they won’t be interested in taking anything in.

Study 1 Conclusions

Study 1 provided a profile of difficult workplace conversations, several reasons for their difficulty, and 24 strategies for helping to achieve successful outcomes. Despite proposing plausible strategies, and despite the interviewees’ considerable job experience and possible well-honed communication skills, the interviewees made clear that their conversations seldom were entirely successful. Difficult conversations more often led to responses of denial, disagreement, and resistance than to contrition and cooperativeness. Some conversations ended in complete stalemate, and others led to angry and aggressive acts of retaliation and/or accusations of bullying. Some were expected to be repeated many times because requested changes in work behaviors were never made.

Classic contributions in the study of negotiation (e.g., Pruitt, 1983) and conflict management (e.g., Deutsch, 1994) have observed that social interactants are often faced with multiple, partially incompatible goals. Consistent with this and past research in related fields (e.g., Bradley, Sparks, Zapf, McColl-Kennedy, & Jimmieson, 2013), we discerned three competing goals in the Study 1 interview content. First, there was a task goal. Often, the goal pertained to changing another person’s attitudes or behaviors, and was pursued through such strategies as 2 (Gather evidence), 17 (Persuasion), and 21 (Take responsibility). A second goal was to care for the other person’s well-being and maintain a positive relationship with that person. This goal is reflected in such strategies as 4 (Forewarn), 12 (Give voice), and 15 (Be empathic). The third goal relates to protecting and enhancing oneself and one’s position within the organization, as indicated by strategies 6 (Enlist support), 11 (Adhere to policies), and 24 (Keep records). In general, the conversations can be viewed as mixed-motive encounters, with trade-offs required between these different goals. Research in related fields (e.g., Bradley et al., 2013;; De Dreu & Gelfand, 2008; Deutsch, 1994) suggests that progress toward a prioritized goal is often made more difficult by a failure to consider other goals. In particular, task-focused behaviors that ignore the needs of the other person may lead not to the desired behavior changes but to resentment and resistance. Put another way, conversations that are effective (in meeting task goals) may not be appropriate (in maintaining cordial relations) (Bisel, Messersmith, & Kelley, 2012).

One of many themes we observed in the interviews was a tendency for recipients of bad news to become “defensive,” defiant, and even subversive. Interviewees frequently attributed these reactions to recipient feelings of being attacked rather than supported, with criticisms leveled by the interviewee in the heat of a difficult conversation being interpreted by the recipient as a personal attack, and a source of embarrassment and loss of “face.” Studies 2 and 3 continued to pursue our general aim of identifying reasons for conversation difficulty and ways of reducing these difficulties. The studies particularly focused on the interrelated themes of communication defensiveness, supportiveness, and loss of face.

Difficult Conversations and Defensive-Supportive Communication Climates

Defensive behaviors are those that involve attempts to deny, conceal, justify, disguise, or otherwise minimize aspects of the self that are subjected to criticism. According to Gibb (1961), Stamp, Vangelisti, and Daly (1972) and others, defensive reactions are triggered by communications that threaten, attack, or find flaws in another person. On the basis of eight years of research observing group discussions, Gibb proposed six dimensions of behavior, the positive poles of which create a sense of supportiveness while the negative poles produce a climate of defensiveness. We have listed these six dimensions of defensive and supportive communicative behaviors in Table 1. Gibb observed that defensiveness tends to be reciprocated and self-perpetuating, with destructively critical language often leading to defensive listening and closed-mindedness. In contrast, when supportive communication is used, accusations and blame are reduced and message transmission and acceptance are facilitated.

Table 1.

Gibb’s (1961) Model of Defensive-Supportive Communication.

Behaviors that promote a defensive climate	Behaviors that promote a supportive climate
Evaluation (assessing, judging, criticizing)	Description (accepting, nonjudgmental understanding)
Control (forcing, dominating, persuading)	Problem Orientation (permissiveness, focusing on task, collaborating)
Strategy (scheming, deceiving, manipulating)	Spontaneity (honesty, genuineness, straightforwardness)
(Emotional) Neutrality (indifference, lacking in care and warmth)	Empathy (perspective-taking, sharing feelings, validating)
Superiority (dismissing, dominating, denigrating)	Equality (participating, disclosing, respecting)
Certainty (dogmatic, foreclosed, single-minded)	Provisionalism (tentative, open, tolerant)

Gibb’s (1961) ideas about defensive and supportive communication are presented in a range of contemporary communication textbooks (e.g., Adler & Rodman, 2009). Forward, Czech, and Lee (2011, p. 2) claim that Gibb’s model has now achieved “iconic status.” The model has, however, received limited empirical scrutiny. An exception is a recent study by Forward et al. (2011) which found that more effective chairs of academic departments were perceived to use all six of Gibb’s supportiveness dimensions more frequently, and five of the six defensiveness behaviors less frequently, than did less effective chairs. Perceived effectiveness was best predicted by the supportiveness dimensions of problem orientation and description, while faculty’s satisfaction with their relationship with their chair was best predicted by the dimensions of empathy and spontaneity. In other studies, Alexander (1979) found that communication characterized by a problem orientation helped reduce conflict, and Winer and Majors (1981) observed that negative feedback was perceived to be more supportive if it contained provisional verbs (e.g., “seem,” “appear”) rather than certainty verbs (e.g., “be,” “know”).

Gibb’s model is not without shortcomings. For example, Gibb did not specify the verbal behaviors indicative of the poles on each dimension, and this has hampered the development of valid tools to measure his constructs. Moreover, no empirical evidence has been published as to either the mutual exclusivity or the exhaustiveness of his six dimensions, few researchers other than that of Forward et al. have assessed all of Gibb’s dimensions in a single study, and studies that report supportive findings also report findings that run counter to predictions.

Study 2: Correlates of Defensive-Supportive Communication

Research by Gibb (1961) and others suggests that particular conversation acts induce a supportive interpersonal climate, and thereby work against participant defensiveness, resistance, and/or withdrawal. Following previous research (e.g., Carson & Cupach, 2000; Tata, 2002), in Study 2 we used a critical incidents approach in which participants recalled a recent workplace difficult conversation, evaluated the supportiveness of the other party, and indicated their satisfaction with the conversation. We predicted that each of Gibb’s six supportive behaviors would be associated with high levels of participant satisfaction.

Study 2 Method

Participants

Following a small pilot study (N = 13 undergraduates), 137 people (80% female, aged 17–59 years, M = 23.4 years) recruited on a convenience basis participated in Study 2. The single eligibility criterion was recent (preceding 12 months) participation as the junior person within a superior-subordinate workplace difficult conversation.

Materials

We developed an e-questionnaire that included items pertaining to the content and context of a difficult communication and level of satisfaction with this conversation.

Difficult conversation recall items

Participants first completed several open-ended items that required the recall of a recent conversation in which they had received negative feedback from a superior relating to their work performance and/or behavior. Participants reported details of this conversation (e.g., the manner in which they were approached and spoken to, the nature of the issue under discussion, the negativity of any feedback received, the duration of the conversation, the presence of others), as well as information about previous experiences of difficult conversations, and their relationship with their superior.

Defensive-supportive communication behaviors

We sought participants’ views as to the defensiveness-supportiveness of their superior’s communication behavior using items from the Communication Climate Inventory (CCI; Costigan & Schmeidler, 1984). This scale contains subscales assessing each of Gibb’s (1961) six dimensions. When used by Forward et al. (2011), the subscales had alpha reliability coefficients of between .65 and .93, and acceptable concurrent validity, although discriminant validity was limited. We used the three items from this scale measuring each of Gibb’s six supportive communication categories. Consistent with Forward et al., item wording reflected the workplace superior-subordinate feedback context. Response ranged from 1 = strongly disagree to 7 = strongly agree.

Dependent variable (DV)

Participants reported their level of satisfaction with the conversation by responding to eight items (e.g., “In general, how satisfied were you with the conversation?”) taken from previous research. Following reverse scoring of two items, we formed a composite scale by averaging all responses, with higher scores indicating greater levels of satisfaction.

Procedure

Participants responded to an advertisement to take part in an anonymous, 20-minute study of difficult conversations in the workplace. Participants read and retained the study information sheet, reported details of their difficult conversation, and then completed the questionnaire items with this conversation in mind.

Study 2 Results

Two of the supportiveness items (one intended to measure each of provisionalism and empathy) were highly correlated with other scales, and one of the description scale items was weakly associated with the other two description items (rs = .14 and .16). Therefore, we used composites for these three scales comprising just two items, and only 15 of the original 18 items to measure the six supportiveness behaviors. Despite this, correlations between the six communication subscales remained very high, ranging from .56 to .87

We present descriptive statistics and scale reliabilities in Table 2. As shown, all supportiveness features were highly correlated with satisfaction, with the dimensions of empathy, equality, and description more closely related to this outcome than were the remaining three dimensions.

Table 2.

Study 2 Descriptive Statistics, Correlations, and Regression Model Predicting Satisfaction With the Conversation.

					Standard regression predicting satisfaction
Variable	Mean	SD	Cronbach’s alpha	Correlation with satisfaction	B	SE B	β
Predictors: Supportive Behaviors
Spontaneity	4.57	1.46	.77	.67	.15	.06	.14*
Provisionalism	3.55	1.59	.86	.80	.06	.09	.06
Description	3.55	1.78	.86	.81	.17	.07	.19*
Empathy	3.74	1.81	.92	.86	.28	.09	.32**
Equality	3.68	1.57	.75	.82	.26	.07	.26***
Problem-orientation	4.12	1.47	.76	.73	.01	.08	.01
					R² = .80
Criterion
Satisfaction	3.01	1.55	.94

p < .05. ** p < .01. *** p < .001.

Predicting the DVs from the six communication variables

We examined relationships between the six supportive communication behaviors and the criterion, satisfaction with the conversation. Because Gibb’s theory does not differentiate between the importance of the supportiveness variables, all six were entered in a single step as part of a standard regression analysis. Assumptions underlying regression were met, although variance inflation factors were as high as 6.5. Overall, the six predictors explained 80% of the variance in satisfaction, F(6, 108) = 83.75, p < .001 (see Table 2). Four predictors made unique contributions, with communication characterized by empathy and equality explaining the most variance. In line with past research (e.g., Tata, 2002), we repeated the analysis with message negativity included as a control at the first step. Results remained substantially unaltered.

Study 2 Conclusions

Study 2 results confirmed predictions based on Gibb’s work that supportive communication is positively associated with favorable participant evaluations. Although the scale used to assess the unique contributions of the six aspects of supportiveness did not provide adequate separation between these dimensions, some evidence was obtained for the relevance of all six characteristics. These findings extend past work, most of which has investigated only a subset of Gibb’s dimensions, has examined only bivariate relationships, and has done so within ongoing relationships, rather than in relation to one-off interactions, as was the case here.

The obtained links between communication behaviors and favorable conversation outcomes among employees align only partially with the recommendations for handling difficult conversations proposed by the Study 1 nurse managers. In that study, while several interviewees emphasized the positive role played by displays of empathy, many of their other comments suggested that they favored acts of superiority, evaluation, and control over their more supportive opposites. For example, rather than advocating a (collaborative) problem orientation, a preference for its polar opposite, control, may be inferred from remarks such as “You have to be the one running the conversation,” “You’ve just got to not let them talk over you,” and “. . . being in control is what I need.” Comments such as these suggest that a higher priority was placed on achieving task- and self-focused goals than on achieving relatedness goals. In other words, effectiveness was favored over appropriateness. As Gibb’s model predicts, communication acts that reflect such priorities create a climate in which defensiveness and resistance-to-change are likely to prevail.

While the Study 1 interviewees expressed dismay over the frequency with which their difficult conversations resulted in subordinate negativity and defensiveness, many seemed to acknowledge that these behaviors were displayed in response to a lack of supportiveness on their part, and, more particularly, to a threatened or actual loss of face. Strategies recommended by the Study 1 interviewees to avoid face loss include insuring that difficult conversations are conducted in private, criticisms are worded tactfully, and subordinates are given an opportunity to voice their version of events. The extent to which the nurse managers actually implemented these behaviors is however moot, especially in light of arguments put forth by Bisel et al. (2012) that, within the command structure of most organizations, supervisors are not highly motivated to protect the face of their subordinates. Given this background, we explored the notions of face, loss of face, and “face-work,” as well as continuing to examine some of Gibb’s (1961) ideas about supportiveness, in the third of our studies.

Study 3: Experimental Evidence of Communicative Acts During Difficult Conversations

According to Goffman (1967), “face” refers to the positive social value that a person claims for himself/herself and presents in interaction with others. “Face-work” refers to communicative acts used either to defend, maintain, or restore one’s preferred social identity or to protect that of an interaction partner. Applying these concepts, it can be seen that much of the information conveyed in difficult conversations, especially that pertaining to another person’s mistakes, flaws, performance lapses, or similar failings, is potentially “face-threatening.” When criticized, blamed, shamed, or similarly face-threatened, participants are likely to engage in acts of “face-saving” so as to (re-)establish their status. Goffman argued that face threats are mitigated by acts such as denying, disowning, redefining, or compensating for the offending act. Excuses and self-justificatory remarks may be offered, and, if not fully successful, these may be followed by counter-attack or avoidance behaviors. Conversations that degenerate into face-threatening (offensive) and face-saving (defensive) cycles are unlikely to be productive, because resources are diverted from the substantive topic to more personal issues (e.g., reputation, pride).

Conversations can be made less threatening, and hence less difficult, if the parties communicate in ways that “protect” and “give” face to each other. According to Goffman, these acts may involve downplaying or not drawing attention to the other person’s failings, giving the other an opportunity to rationalize or justify their actions, and supporting and enhancing the other’s status through compliments, affirmations, and indications of group inclusion. Such acts of positive face show respect for the other person and help that person feel competent, liked, and valued. In support, research (e.g., Carson & Cupach, 2000; Clifton, 2012; Kerssen-Griep, Trees, & Hess, 2008; Tata, 2002; Trees, Kerssen-Griep, & Hess, 2009; Tynan, 2005) shows that face-saving acts can reduce defensiveness and enhance positive conversational outcomes.

No study could be located that explored the relationships between Gibb’s (1961) dimensions of supportive communication and Goffman’s (1967) notions of face and face-work. Yet there are several parallels. Many difficult conversations involve attempts by one person to evaluate, control, and /or “pull rank” on another. These acts are both face-threatening (Goffman, 1967) and defensiveness-inducing (Gibb, 1961) and are likely to trigger adverse reactions, including counterattack and withdrawal (Baron, 1988). In contrast, acts that are face-giving and supportive are likely to have opposite effects, including attenuation of conversation difficulty and enhancement of the self, the relationship, and work performance.

In Study 3, we explored the main and interactive effects of two of Gibb’s supportive communication dimensions, and Goffman’s notion of positive face, on participant reactions to a difficult conversation. Given the argument of Bisel et al. (2012) that supervisors are disinclined to give face to their subordinates, we were particularly interested in whether communicative behavior that is in other ways supportive would compensate for a lack of supervisor face-work. Rather than selecting supportiveness dimensions that have well-established effects and are well supported by past research, we chose to study the dimensions of strategy versus spontaneity, and certainty versus provisionalism. The effects of these two dimensions on conversation outcomes are not so intuitively obvious as are some others, yet there is some evidence that each of them (spontaneity: Forward et al., 2011; and provisionalism: Winer & Majors, 1981) can have powerful effects on conversation outcomes.

While the method used in Study 2 (i.e., asking participants to report their own experiences of difficult conversations) ensured high levels of realism, it relied on participant recall accuracy, and it did not permit causal conclusions to be drawn. In contrast, in Study 3 we sought evidence of causality through the use of a 2 (face-giving: present vs. absent) × 2 (provisionalism vs. certainty) × 2 (spontaneity vs. strategy) between-subjects factorial experiment. Following prior research (e.g., Lizzio et al., 2008), the independent variables (IVs) were manipulated via written scenarios. We hypothesized that, within the context of a difficult workplace conversation, a superior’s display of (1) face-giving, (2) spontaneity (rather than strategy), and (3) provisionalism (rather than certainty) would result in high levels of subordinate satisfaction with the conversation and low levels of subordinate intended withdrawal and retaliatory behaviors. In addition to these three hypothesized main effects, interaction effects (where, e.g., the joint effects of spontaneity and provisionalism may buffer the adverse impact of a lack of face-giving) were explored.

Study 3 Method

Participants

Following a pilot study (N = 28), 204 undergraduate students (71% female, aged 17–54 years, mean = 22.0 years) participated in Study 3.

Materials

We provided study details and consent procedures in an information sheet, and developed a paper questionnaire that included a version of a difficult conversation scenario, and measures of three dependent variables: conversation satisfaction, likely withdrawal behaviors, and likely retaliatory behaviors.

Scenarios

We developed eight versions of a written scenario to operationalize the three communication behaviors. The scenario was set in a retail employment context and entailed a superior delivering negative work performance feedback to a subordinate. We manipulated the IVs in ways that closely reflect seminal theory (Gibb, 1961; Goffman, 1967) and past research (e.g., Winer & Majors, 1981). Thus, face-giving was manipulated by the superior showing concern for the subordinate’s self- and social esteem (Face Present) or by showing no such concern (Face Absent). Similarly, the superior either used straightforward and honest language when addressing the subordinate (Spontaneity) or used language that reflected hidden and multiple motives (Strategy), and used either tentative and open-ended language (Provisionalism) or definite and dogmatic language (Certainty). The eight versions of the scenarios included all possible combinations of the three communication behaviors. In all other respects including statements by the subordinate, they were identical.

Dependent variables

We measured satisfaction as in Study 2. In addition, participants completed scales indicating the extent to which the conversation would likely entice them to engage in increased amounts of four psychological withdrawal behaviors (e.g., “Thought about quitting my job”), and six retaliatory behaviors (e.g., “Broke rules or policies”). We adapted these items from Lehman and Simpson (1992), and required responses from 1 = strongly disagree to 7 = strongly agree, with higher scores on the two scales indicating greater conversation-induced increases in withdrawal and retaliation behaviors.

Manipulation and credibility checks

Items assessing the manipulation of the IVs were based on existing instruments (e.g., Costigan & Schmeidler, 1984; Kerssen-Griep et al., 2008). Four items assessed each of face-giving (e.g., “I think the superior was genuinely concerned for my self-esteem and confidence”), spontaneity versus strategy (e.g., “I think the superior was very honest and straightforward”), and provisionalism versus certainty (e.g., “I think the superior was very definite as to what change should be made to overcome the problem with my work”). All required a response ranging from 1 = strongly disagree to 7 = strongly agree. We also used four items from Willson and McNamara (1982) to measure the credibility of the scenarios (e.g., “I think there are superiors who behave like this in real life”).

Procedure

We randomly assigned participants to one of eight versions of the conversation scenario. Participants read the scenario and, imagining themselves as the subordinate in the conversation, completed the questionnaire.

Study 3 Results

Preliminary analyses

We first conducted three 2 (presence vs. absence of face) × 2 (spontaneity vs. strategy) × 2 (provisionalism vs. certainty) ANOVAs with either perceived face-giving, perceived provisionalism, or perceived spontaneity as the DV. These ANOVAs confirmed that all variables were manipulated as intended. We averaged the responses to the four scenario credibility items. One-sample t-tests revealed that the mean credibility score was significantly (p < .001) greater than the neutral scale point (i.e., 4.0) on the 7-point response scale for all eight conditions, and for the sample as a whole, indicating that the scenarios were perceived as realistic and believable.

We computed composite scores for each of the dependent variables: satisfaction (M = 3.57, SD = 0.75; α = .89), withdrawal behaviors (M = 2.34, SD = 1.04; α = .81), and retaliatory behaviors (M = 1.97, SD = 0.97; α = .89). Satisfaction correlated at r = –.30 with withdrawal and r = –.35 with retaliation, while the two measures of post-conversation behaviors were positively correlated, r = .65.

Main analyses

The first ANOVA, with satisfaction as the DV, revealed a significant main effect for face-giving, F(1, 195) = 26.27, p < .0005, η_p² = .12, and nonsignificant effects for spontaneity, provisionalism, and all two- and three-way interactions (all ps > .10). Second, with psychological withdrawal as the DV, a significant effect was revealed for face, F(1, 195) = 5.06, p = .026, η _p² = .03. While neither the spontaneity nor the provisionalism main effect was significant, their interaction was F(1, 195) = 4.59, p = .033, η_p² = .02. Contrary to expectations that withdrawal would be minimized under the joint presence of spontaneity and provisionalism, none of the paired contrasts was significant. The nonsignificant trend was for the spontaneity-certainty and strategy-provisionalism conditions to be associated with lower levels of withdrawal than were the other two conditions.

In the third ANOVA, with retaliation as the DV, face was again significant F(1, 196) = 7.30, p = .008, η_p² = .04, as was the Spontaneity × Provisionalism interaction, F(1, 196) = 6.21, p = .014, η_p² = .03. These effects were however qualified by a significant three-way interaction, F(1, 196) = 5.35, p = .022, η_p² = .03. As shown in Figure 2, when face-giving was present, certainty produced more retaliatory reactions than did provisionalism, but there was no difference between the spontaneity and strategy conditions and none of the pairs of cell means differed significantly (p > .05). In contrast, contrary to expectations, when face-giving was absent, communication characterized by both spontaneity and provisionalism generated the most retaliatory behaviors, with scores in this condition higher than in the spontaneity-certainty (p = .004) and the strategy-provisionalism (p = .013) conditions.

Figure 2.

Study 3 interaction of face, spontaneity, and provisionalism on retaliatory behaviors.

We conducted post hoc Pearson correlation analyses between scores on the three manipulation check scales and the three DVs. Perceived face-giving correlated at r = .59 with satisfaction, and r = –.31 with each of withdrawal and retaliation behaviors, whereas the corresponding correlations involving perceived spontaneity were r =.54, –.45, and –.34 and those involving perceived provisionalism were r = –.24, –.04, and –.04.

Study 3 Conclusions

Using a scenario-based experimental design, Study 3 demonstrated that, compared to a no face-giving alternative, displays of face-giving by a superior lead to high subordinate satisfaction with a difficult conversation and low likelihood of subordinate future engagement in withdrawal and retaliatory behaviors. Correlational analyses involving the measure of perceived face-giving supported these findings. The findings are consistent with past studies (e.g., Kerssen-Griep et al., 2008; Tata, 2002; Trees et al., 2009; Tynan, 2005) that have demonstrated positive effects of face-giving on conversation outcomes. The current findings of stronger effects of face on satisfaction than on intended behaviors is novel, but perhaps not surprising given that satisfaction is the more proximate outcome and therefore less likely influenced by other (personal and contextual) factors.

In contrast to past findings (e.g., Forward et al., 2011), spontaneity did not have a main effect on any of the DVs. However, correlations involving the measured version of spontaneity were in accord with prior findings, with Gibb’s (1961) model, and with the Study 1 results. Perceptions of communication spontaneity (r = –.45) were more closely associated with withdrawal behaviors than was perceived face-giving (r = –.31), an interesting finding possibly due to spontaneity being perceived as indicative of a more stable disposition than is face-giving, and hence as a more sound foundation on which to base future engagements.

Despite operationalizing it in ways that are both construct valid and consistent with past research (Winer & Majors, 1981), our manipulation of provisionalism revealed weak main effects. Correlational analyses yielded modest positive associations with satisfaction, and no relationships with either of the behavioral DVs. These findings are not in accord with predictions derived from Gibb’s theory or with the correlations obtained in Study 2.

The unexpectedly weak main effects of spontaneity and provisionalism were qualified by higher-order interactions. Contrary to predictions based on Gibb’s (1961) model, the combination of spontaneity and provisionalism did not have synergistically dampening effects on participant withdrawal behaviors. More remarkably, when face-giving was absent, the combination of these conditions yielded particularly high levels of retaliatory behavior. While replication using other methods and samples is needed, this finding suggests that these two dimensions of communication may have effects that are conditioned by more pervasive and powerful relational variables such as the giving of face. Indeed, future research could investigate whether threats to face trigger a hostile attribution bias: In the present context, with a supervisor showing little regard for a subordinate’s face, the combination of straightforward (spontaneous) and tentative (provisionalism) communicative acts may have been treated as suspicious, and may therefore have elicited attributions of malevolent intent. Future studies may also explore whether supervisors, like subordinates, respond vengefully when spoken to in spontaneous, tentative, but ultimately disrespectful, ways.

General Discussion

The overriding aim of the current research was to discover what makes conversations difficult and what can be done to make them less so. Findings can be summarized as follows. Difficult workplace conversations occur frequently and are generally dreaded. One common type of these conversations entails a superior and a subordinate discussing perceived deficiencies in the latter’s work performance. Such conversations often involve disagreement, defensiveness, and resistance. Outcomes may include attempts by the subordinate to withdraw from the work situation either physically or psychologically, or to retaliate with accusations of harassment and acts of defiance. The frequency of such difficult conversations, and of their adverse outcomes, appears to increase at times, and in places, of work stress. Nurse managers suggested many strategies that can be employed to achieve mutually satisfying outcomes, control damage, and/or protect their own position within the organization. A major contribution of the current research is the development of a conceptual scheme that gives structure on these various ideas. Greater focus on supportive communicative behaviors and a greater willingness to “give face” to the other person were highlighted as ways that may reduce defensiveness and resistance, and thereby improve conversation processes and outcomes. The adoption of such other-directed communication strategies is likely to depend on a range of personal and contextual factors including the salience of rival task- and self-focused goals.

The research drew conceptual links between the theories of communication and face-work proposed by Gibb (1961) and Goffman (1967), respectively. The findings provide some support for both theories. In Study 2, while aspects of supportive communication, especially empathy, equality, and description, were positively and uniquely associated with conversation satisfaction, other communication dimensions displayed less profound effects. The instrument used to differentially assess Gibb’s supportiveness dimensions lacked discriminant validity, suggesting that work is required to develop a tool that more precisely measures each of these behaviors. Study 3 findings pertaining to both spontaneity and provisionalism were equivocal. Study 3 did, however, provide unambiguous evidence in support of Goffman’s ideas as to the importance of face-giving in social interactions. While not the first research to show evidence of this kind, the study did so under experimental conditions in a setting different from the contexts in which many prior studies have been conducted. Moreover, face operated as a higher order moderator of the impact of two of Gibb’s less well established, and arguably less influential, communication variables.

A strength of this research was its use of multiple methods, each of which has complementary strengths. The research does, however, have limitations. The sample in Study 1 was restricted to nurse managers. It would have benefited from including working people from a wider range of businesses and professions. Study 2 was limited by inadequacies in the supportive communication scale. Future research should develop, validate, and use improved scales and should recruit a sample of sufficient size to enable the moderating role of diverse conversation variables (nature of the topic, relational context, etc.) to be more fully examined. Study 3 used a scenario-method which, despite its strengths in terms of feasibility, control, and economy, probably did not capture the full extent and depth of reactions that occur in real-life difficult conversations. In addition, the operationalization of face-work in these scenarios was (necessarily) unilateral, rather than reflecting the dialogical quality evident in many potentially difficult conversations (Clifton, 2012). Finally, findings from this study may be limited to the particular conversation type depicted in the scenario. Future research could use an expanded range of methods, including fine-grained analyses of recordings of actual conversations and laboratory studies that have confederates scripted to manipulate conversation variables. Interview studies that collect process and outcome data from both members of difficult conversation dyads would also be welcome.

All studies were conducted in Australia, using samples of predominantly European-background participants. Findings may not apply outside this setting and population. Voluminous research has demonstrated the relevance of cultural values and norms to communication processes and outcomes (Hofstede, 1980; Triandis, 1994). No doubt these same cultural variables influence how difficult conversations are dealt with. There is, for example, very good evidence of differences between Asian and Western cultures in preferred conflict management styles (e.g., Wang, Lin, Chan, & Shi, 2005) and strategies for dealing with face loss (e.g., Bjorkman & Lu, 1999). Further research is required to identify other, perhaps more subtle, cultural differences in the triggers, processes, and outcomes of difficult conversations.

Notwithstanding these limitations and suggestions, the research draws attention to the need for people within work contexts to possess and practice superior communication skills. Findings not only underscore the need for interventions that facilitate the conduct of difficult conversations, but also suggest content for inclusion in such training. Particularly indicated are foci on relational goals, on the development of skills in empathic face-giving, and on the provision of opportunities within hectic and demanding workplaces for these skills to be put into practice. Interventions may need to promote a negotiated or “contingency” approach to managing difficult conversations, rather than one that focuses on single simple solutions.

Our research concentrated on difficult conversations between workplace subordinates and their superiors. But difficult conversations occur in a wider range of business, and other, contexts than this. Just as the sources of conversation difficulty are likely to vary between issues, participants and settings, so too are the communication practices necessary for success. Future research should explore the extent to which defensive-supportive communication styles, and concerns over face-giving, have explanatory value in diverse kinds of difficult conversations. As alluded to in the introduction, many bodies of theory and research can assist with the understanding and management of difficult conversations. Notwithstanding the potential contribution of the recently published literature, the current work is a reminder of the value that can still be gleaned from classic works such as those of Gibb (1961) and Goffman (1967).

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Graham L. Bradley is an associate professor in applied social psychology at Griffith University, Australia. His research interests include social interaction processes, service encounters, and work and well-being.

Amanda C. Campbell is a postgraduate student in organizational psychology at Griffith University, Australia. Her research interests include organizational communication and occupational health psychology.

References

Adler

R. B.

Rodman

(2009). Understanding human communication (10th ed.). New York, NY: Oxford University Press.

Alexander

J. F.

(1979). The reduction of cognitive conflict. Journal of Conflict Resolution, 23, 120–138.

Baron

R. A.

(1988). Negative effects of destructive criticism: Impact on conflict, self-efficacy, and task performance. Journal of Applied Psychology, 73, 199–207.

Baron

R. A.

(1990). Countering the effects of destructive criticism: The relative efficacy of four interventions. Journal of Applied Psychology, 75, 235–245.

Bisel

R. S.

Messersmith

A. S.

Kelley

K. M.

(2012). Supervisor-subordinate communication: hierarchical mum effect meets organizational learning. Journal of Business Communication, 49, 128–147. doi: 10.1177/0021943612436972

Bjorkman

(1999). The management of human resources in Chinese-Western joint ventures. Journal of World Business, 34, 306–324.

Bradley

G. L.

Sparks

B. A.

Zapf

McColl-Kennedy

J. R.

Jimmieson

N. L.

(2013). Task-Relationship-Self: A framework for understanding service encounter behaviours. Psychology and Marketing, 30, 512-528. doi:10.1002/mar.20624.

Brehm

S. S.

Brehm

J. W.

(1981). Psychological reactance: A theory of freedom and control. New York, NY: Academic Press.

Brown

Hyatt

Benson

(2010). Consequences of the performance appraisal experience. Personnel Review, 39, 375–396. doi: 10.1108/0043481011030557

10.

Brown

Levinson

(1987). Politeness: Some universals in language usage. Cambridge: Cambridge University Press.

11.

Carson

C. L.

Cupach

W. R.

(2000). Facing corrections in the workplace: The influence of perceived face threat on the consequences of managerial reproaches. Journal of Applied Communication Research, 28, 215–234.

12.

Cialdini

R. B.

(2001). Influence: Science and practice (4th ed.). Boston, MA: Allyn & Bacon.

13.

Clifton

(2012). Conversation analysis in dialogue with stocks of interactional knowledge: Facework and appraisal interviews. Journal of Business Communication, 49, 283–311. doi: 10.1177/0021943612436974

14.

Cloud

Townsend

(2005). How to have that difficult conversation you’ve been avoiding. Grand Rapids, MI: Zondervan.

15.

Costigan

J. L.

Schmeidler

M. A.

(1984). Exploring supportive and defensive communication climates. In Pfeiffer

J. W.

Goodstein

L. D.

(Eds.), The 1984 handbook for group facilitators (pp. 112–114). San Diego, CA: University Associates.

16.

De Dreu

C. K. W.

Gelfand

M. J.

(2008). Conflict in the workplace: Sources, functions, and dynamics across multiple levels of analysis. In De Dreu

C. K. W.

Gelfand

M. J.

(Eds.), The psychology of conflict and conflict management in organizations (pp. 3–54). New York, NY: Lawrence Erlbaum.

17.

Deutsch

(1994). Constructive conflict resolution: Principles, training, and research. Journal of Social Issues, 50, 13–32.

18.

Dickson

(2006). Difficult conversations: What to say in tricky situations without ruining the relationship. London: Piatkus Books.

19.

Forward

G. L.

Czech

Lee

C. M.

(2011). Assessing Gibb’s supportive and defensive communication climate: An examination of measurement and construct validity. Communication Research Reports, 28, 1–15. doi: 10.1080/08824096.2011.541360

20.

Geddes

Baron

R. A.

(1997). Workplace aggression as a consequence of negative performance feedback. Management Communication Quarterly, 10, 433–454.

21.

Gibb

J. R.

(1961). Defensive communication. The Journal of Communication, 11, 141–148.

22.

Goffman

(1967). Interaction ritual: Essays on face-to-face behavior. Garden City, NY: Doubleday.

23.

Hofstede

(1980). Culture’s consequences. Thousand Oaks, CA: Sage.

24.

Ilgen

D. R.

Davis

C. A.

(2000). Bearing bad news: Reactions to negative performance feedback. Applied Psychology: An International Review, 49, 550–565.

25.

Kerssen-Griep

Trees

A. R.

Hess

J. A.

(2008). Attentive facework during instructional feedback: Key to perceiving mentorship and an optimal learning environment. Communication Education, 57, 312–332. doi: 10.1080/03634520802027347

26.

Kluger

A. N.

DeNisi

(1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119, 254–284.

27.

Kosmoski

G. J.

Pollack

D. R.

(2005). Managing difficult, frustrating, and hostile conversations: Strategies for Savvy administrators (2nd ed.). Thousand Oaks, CA: Sage.

28.

Lehman

W. E. K.

Simpson

D. D.

(1992). Employee substance use and on-the-job behaviors. Journal of Applied Psychology, 77, 309–321.

29.

Lizzio

Wilson

MacKay

(2008). Manager’ and subordinates’ evaluations of feedback strategies: The critical contribution of voice. Journal of Applied Social Psychology, 38, 919–946.

30.

MacDonald

(2004). Difficult conversations in medicine. Oxford: Oxford University Press.

31.

Madlock

P. E.

Kennedy-Lightsey

(2010). The effects of supervisors’ verbal aggressiveness and mentoring on their subordinates. Journal of Business Communication, 47, 42–62. doi: 10.1177/0021943609353511

32.

Meyer

E. C.

Sellers

D. E.

Browning

D. M.

McGuffie

Solomon

M. Z.

Truog

R. D.

(2009). Difficult conversations: Improving communication skills and relational abilities in health care. Paediatric Critical Care Medicine, 10, 352–359. doi: 10.1097/PCC.0b013e3181a3183a

33.

Miles

M. B.

Huberman

A. M.

(1994). Qualitative data analysis: An expanded sourcebook (2nd ed.). Thousand Oaks, CA: Sage.

34.

Neuman

W. L.

(2011). Social research methods: Qualitative and quantitative approaches (7th ed.). Boston, MA: Pearson.

35.

Pruitt

D. G.

(1983). Strategic choice in negotiation. American Behavioral Scientist, 27, 167–194.

36.

Stamp

G. H.

Vangelisti

A. L.

Daly

J. A.

(1992). The creation of defensiveness in social interaction. Communication Quarterly, 40, 177–190.

37.

Tata

(2002). The influence of managerial accounts on employees’ reactions to negative feedback. Group & Organization Management, 27, 480–503.

38.

Tesser

Rosen

(1975). The reluctance to transmit bad news. In Berkowitz

(Ed.), Advances in experimental social psychology (Vol. 8, pp. 193–232). New York, NY: Academic Press.

39.

Trees

A. R.

Kerssen-Griep

Hess

J. A.

(2009). Earning influence by communicating respect: Facework’s contributions to effective instructional feedback. Communication Education, 58, 397–416. doi: 10.1080/03634520802613419

40.

Triandis

H. C.

(1994). Culture and social behaviour. New York, NY: McGraw-Hill.

41.

Tynan

(2005). The effects of threat sensitivity and face giving on dyadic psychological safety & upward communication. Journal of Applied Social Psychology, 35, 223–247.

42.

Wang

C. L.

Lin

Chan

A. K. K.

Shi

(2005). Conflict handling styles in international joint ventures: A cross-cultural and cross-national comparison. Management International Review, 45, 3–21.

43.

Weiner

(1985). “Spontaneous” causal thinking. Psychological Bulletin, 97, 74–84.

44.

Willson

McNamara

J. R.

(1982). How perceptions of a simulated physician-patient interaction influence intended satisfaction and compliance. Social Science & Medicine, 16, 1699–1704.

45.

Winer

Majors

R. E.

(1981). A research note on supportive and defensive communication: An empirical study of three verbal interpersonal communication variables. Communication Quarterly, 29, 166–172.