When proactivity produces a power struggle: how supervisors’ power motivation affects their support for employees’ promotive voice

ABSTRACT

Previous research informs us about facilitators of employees’ promotive voice. Yet little is known about what determines whether a specific idea for constructive change brought up by an employee will be approved or rejected by a supervisor. Drawing on interactionist theories of motivation and personality, we propose that a supervisor will be least likely to support an idea when it threatens the supervisor’s power motive, and when it is perceived to serve the employee’s own striving for power. The prosocial versus egoistic intentions attributed to the idea presenter are proposed to mediate the latter effect. We conducted three scenario-based studies in which supervisors evaluated fictitious ideas voiced by employees that – if implemented – would have power-related consequences for them as a supervisor. Results show that the higher a supervisors’ explicit power motive was, the less likely they were to support a power-threatening idea (Study 1, N = 60). Moreover, idea support was less likely when this idea was proposed by an employee that was described as high (rather than low) on power motivation (Study 2, N = 79); attributed prosocial intentions mediated this effect. Study 3 (N = 260) replicates these results.

KEYWORDS:

• Promotive voice
• idea support
• power motive
• supervisor support
• proactivity

Today’s organizations are faced with a constant pressure to become more effective, and to adapt to changing internal and external demands. Drawing on employees’ ideas on how to improve day-to-day procedures and the functioning of the organization is propagated as a way to cope with these demands (e.g., Farr & Ford, 1990; Frese & Fay, 2001; Grant & Ashford, 2008; Morrison & Phelps, 1999). Despite the general praise of continuous improvement and incremental innovation, not every idea for constructive change may be readily embraced. One reason for this is that such ideas come along with changes in work processes and procedures (sensu West & Farr, 1990) that are likely to affect different stakeholders and evoke resistance (Klein & Sorra, 1996). Supervisors play a particularly central role in idea implementation as many organizations leave the initial evaluation of employees’ ideas to line managers (Leach, Stride, & Wood, 2006). Hence, supervisors may become a “bottle neck” in the process of change and improvement as they can use their power to move an idea forward to implementation, or to bring it to a halt.

Expressing ideas for constructive change is a form of proactive work behaviour (Frese, Teng, & Wijnen, 1999) that has been described as promotive (Liang, Farh, & Farh, 2012) or constructive voice (Maynes & Podsakoff, 2014). This behaviour is self-initiated by the employee and aims at improving organizational functioning. A large body of research informs us about variables that facilitate employees’ proactive work behaviours in general (e.g., Bindl & Parker, 2011; Hong, Liao, Raub, & Han, 2016; Parker, Williams, & Turner, 2006), and proactive suggestion-making (e.g., Frese et al., 1999) or voice behaviour (e.g., Liang et al., 2012; Maynes & Podsakoff, 2014; Morrison, 2011) in particular. Research on the evaluation of these discretionary employee behaviours mainly focused on how they reflect in employees’ general performance appraisal (e.g., Grant & Ashford, 2008; Howell, Harrison, Burris, & Detert, 2015; Thomas, Whitman, & Viswesvaran, 2010; Whiting, Maynes, Podsakoff, & Podsakoff, 2012). We want to extend this research perspective on the general appreciation of proactive work behaviour by investigating what affects supervisors’ support for a specific act of promotive voice – i.e., a specific idea voiced by an employee – and thereby shapes their intentions to support this idea, or to turn it down.

We propose that supervisors’ intentions to support a specific idea depend on the consequences the idea’s implementation has for the supervisors, in particular, their need for being in power. The implementation of ideas for improvement may have various implications for its different stakeholders. For supervisors, one major reason for denying support for an idea may be that it challenges the existing distribution of power or of other resources (Venkataraman, MacMillan, & McGrath, 1992). Research in social psychology demonstrates that threatening leaders’ position-based power enhances the likelihood of leaders’ self-interested actions to secure their power position, even if these actions point against the interests of their group and the wider organization (e.g., Maner, Gailliot, Butz, & Peruche, 2007; Maner & Mead, 2010; for a review, see Williams, 2014). Although scholars have acknowledged that such power struggle can be a crucial obstacle for proactivity, innovation and change (Grant, Gino, & Hofmann, 2011; Kanter, 1988), empirical test of this notion is scarce.

Previous research suggests that three groups of variables affect supervisors’ response to employees’ ideas: First, supervisors’ response to voice is affected by the nature of the voice act, i.e., whether they are challenging versus supporting the status quo. Challenging employee ideas are less likely to be endorsed by supervisors (Burris, 2012). Following current theorizing (Williams, 2014), challenge in the form of power threat should result in more self-interested responses of supervisors towards an idea. Second, supervisors’ characteristics seem to shape how they cope with the challenging nature of employees’ ideas (e.g., Chiaburu, Peng, & Van Dyne, 2015; Fast, Burris, & Bartel, 2014; Sijbom, Janssen, & Van Yperen, 2015a, 2015b). For example, supervisors low in managerial self-efficacy were found to discourage voice out of ego defensiveness (Fast et al., 2014). With regard to power, supervisors’ individual need for being in power is likely to enhance their sensitivity towards power threats in an idea and facilitate defensive responses (e.g., Maner & Mead, 2010). Third, characteristics of the employee are likely to affect the evaluation of their idea proposal (e.g., Sijbom et al., 2015a). In particular, perceiving their employee as a potential competitor to their power (Maner & Mead, 2010) should intensify supervisors’ defensiveness in responding to power-threatening ideas.

Taken together, we propose that a supervisor’s intention to support an idea depends on the idea’s capacity to satisfy or to threaten the supervisor’s power motive. The effects of threatening the supervisor’s power motive should be augmented if the idea-presenting employee is perceived to strive for gaining power at the supervisor’s cost. We investigate the levels of prosocial and egoistic intentions attributed to the idea presenter as mediators of this process. We test these hypotheses in three scenario-based studies.

The present research advances our theoretical and practical knowledge on employees’ proactive work behaviour in three ways: First, we investigate determinants of supervisors’ support for specific ideas for change in organizations. Thereby, we contribute to existing knowledge by shedding light on how specific acts of proactive work behaviour, particularly promotive voice, are perceived and evaluated by significant others. Second, we demonstrate under what conditions a type of behaviour that is conceptualized as seeking to improve organizational functioning may actually lead to a negative response. Third, we provide empirical evidence that challenges to the power distribution in an organization can bring a constructive change initiated by an employees’ proactive effort to a halt.

Theoretical framework and hypotheses

Supervisors’ power motive and their support for employees’ ideas

A supervisory or leadership position is strongly related to power, control and status because the core of supervisors’ activities lies in “the direction, coordination, delegation, and planning of other people’s actions” (Kazén & Kuhl, 2011, p. 320). Power is proposed to be such a desirable state that most individuals in powerful positions would generally avoid losing their power (e.g., Keltner, Gruenfeld, & Anderson, 2003). Accordingly, it has been argued that supervisors in organizations would be sensitive to power threats and respond in self-interested ways to secure their power position (Williams, 2014). The way in which individuals make use of the power their position offers differs as a function of individual differences (Chen, Lee-Chai, & Bargh, 2001) – in particular, supervisors’ personal power motivation. Power motivation describes an individual’s need for gaining superiority in terms of status, access to valued resources and social competence (McClelland, 1975, 1985, 1987). Based on their level of power motivation, individuals differ in their sensitivity towards power-related issues, in particular towards power threat (Maner et al., 2007; Williams, 2014). These differences also reflect in individuals’ decision-making: If individuals high in power motivation are put in a position that provides them with power (e.g., being assigned to a leadership role), they will make more conservative decisions in order to defend their power position (Maner et al., 2007). To achieve this, individuals were shown to withhold important information from their group or to use their power to eliminate potential competitors to their power position (Maner & Mead, 2010).

The specific changes proposed by an employee’s idea may challenge the power distribution within an organization (Botero & Van Dyne, 2009; Venkataraman et al., 1992). In this case, the changes implied by an idea may involve power threat for a supervisor, for example, a loss of influence in decision-making authority. Following trait activation theory (Tett & Guterman, 2000), we assume that if an idea has power-related consequences for the supervisor, it activates supervisors’ power motivation and promotes behaviour that is in line with supervisors’ goals. Thus, supervisors are likely to respond differently to the same idea depending on their level of power motivation. Individuals tend to evaluate options more positively that are congruent with their motivational state (Biernat, 1989; Pang, Villacorta, Chin, & Morrison, 2009), and vice versa, respond aversively to situational threats to their motives (Fodor, 1984; McClelland, 1982). As mentioned earlier, highly power-motivated individuals are most likely to respond to situational power threats by averting the respective threat (e.g., Maner et al., 2007; Maner & Mead, 2010). We propose that this principle also applies to supervisors’ assessment of an employee’s idea and their intentions to support it: If an idea threatens a supervisor’s share in the power distribution in an organization, it is less likely to find support the higher the supervisor’s power motive is.

Hypothesis 1: A supervisor’s intention to support a power-threatening idea will be lower the higher his/her power motive is.

Subordinates’ power motive and supervisor-attributed intentions

We further suggest that supervisors’ support for an idea is affected by their attribution of the idea-presenting employee’s behaviour as being motivated by prosocial or egoistic intentions. People try to make sense of others’ actions by drawing inferences about the causes as well as the intentions underlying this behaviour, particularly if the behaviour is relevant to their self (Reeder, 2009). These attributed intentions in turn affect the appraisal of others’ behaviour and others’ responses towards that behaviour (Ferris, Bhawuk, Fedor, & Judge, 1995). Discretionary employee behaviours are likely to be attributed as dispositional, because individuals chose freely to engage in these behaviours (Grant & Ashford, 2008). Therefore, supervisors’ reward or punishment for such behaviours should depend on their judgements of intentionality, i.e., why employees engaged in the respective behaviour (Reeder, 2009). Previous research has distinguished two broader categories of intentions behind proactive work behaviours (Grant & Ashford, 2008), i.e., prosocial, other-serving intentions (e.g., prosocial values; Grant, Parker, & Collins, 2009) and egoistic, self-serving intentions (e.g., impression management; De Stobbeleir, Ashford, & de Luque, 2010). A specific act of proactive behaviour can serve both prosocial and egoistic purposes. Current proactivity research suggests that employees’ proactive behaviour is more likely to be welcomed if it is perceived as beneficial for others or for the entire organization. In contrast, it is evaluated less positively if the behaviour is perceived as self-serving or potentially harmful (e.g., Grant & Ashford, 2008; Grant et al., 2009). Studies on supervisor evaluations of organizational citizenship behaviour (Bolino, Varela, Bande, & Turnley, 2006; Halbesleben, Bowler, Bolino, & Turnley, 2010) yielded similar results. Accordingly, an employee’s idea should be more likely to find supervisors’ support if the presenting employee is perceived to pursue higher levels of prosocial, other-serving intentions and lower levels of egoistic, self-serving intentions with his/her idea.

Research on attributions suggests that the characteristics of the employee who presented the idea inform this attribution process (De Stobbeleir et al., 2010; Ferris et al., 1995). We propose that supervisors’ attributions depend on the extent to which an idea is perceived to predominantly serve the focal employee’s own needs and goals – specifically, his/her need for power – and the extent to which an idea is perceived to be particularly beneficial to others within the organization. Halbesleben and colleagues (2010) argue that employees’ organizational citizenship behaviour (e.g., helping out colleagues) is more likely to be attributed to self-serving reasons when the proximal consequences of their behaviour are beneficial for themselves rather than for others. Accordingly, an idea that is satisfying the power motive of the presenting employee should be perceived as rather self-serving and instrumental. As a consequence, supervisors should attribute higher levels of egoistic intentions to the employee’s behaviour, and they should be less willing to support his/her idea. If, however, the consequences of the idea did not serve the employee’s own motive but were predominantly beneficial to others, it should be more likely that supervisors perceive the employee’s activity as motivated by selfless reasons. As a consequence, they should be more willing to support the idea, because they attribute higher levels of prosocial intentions and lower levels of egoistic intentions. Altogether, we propose:

Hypothesis 2: (a) A supervisor’s intention to support a power-threatening idea will be lower if he/she perceives the idea to serve the presenting employee’s power motive.

(b) This effect is mediated by supervisor-attributed prosocial as well as egoistic intentions of the employee.

Supervisors’ and subordinates’ competition for power

Grant et al. (2011, p. 532) note that “proactivity has the potential to create a power struggle, as both leaders and employees seek to gain control”. We argue that this power struggle is most likely if a specific act of behaviour, i.e., an idea, challenges the power distribution between supervisor and employee, and if both supervisor and employee have a high need for power.

Interpersonal theories of personality propose that individuals’ dispositions are manifested in characteristic patterns of interpersonal behaviour. Complementarity theory (Kiesler, 1983) states that interactions between individuals are satisfying as long as their behaviour is complementary in terms of serving each other’s needs. Due to formal hierarchy, the complementarity dimension of dominance-submissiveness should be particularly salient in supervisor–subordinate dyads, such that the supervisor’s dominance requires submissiveness by the subordinate. Research lends support to the role of complementarity in individuals’ power motive configurations. Fodor, Wick, and Hartsen (2006) showed that individuals holding a high power motive respond with a stronger physical arousal when they are faced with others that challenge their power motive, i.e., by being assertive rather than compliant. The reason for this might be that they become more aware of the instability of their position-based power. As a consequence, supervisors who strongly value having power should be more likely to engage in self-interested actions to secure their power (e.g., Maner et al., 2007; Maner & Mead, 2010), e.g., by denying support for the employee’s idea.

Likewise, supervisors’ attributions of prosocial and egoistic employee intentions should become more extreme: When an idea threatens a supervisor’s power motive and, at the same time, personally benefits the employee who is threatening the supervisor, a supervisor should perceive this as an inappropriate attempt of the lower status employee to gain power at the supervisor’s cost. Such unjust interpersonal treatment resulting in an unfavourable outcome is related to blame attributions and negative responses towards the interaction partner (e.g., Barclay, Skarlicki, & Pugh, 2005; Blader & Chen, 2011). As a result, supervisors should attribute particularly high levels of egoistic intentions to the employee. At the same time, the attribution of prosocial intentions should be very low as the supervisor is not at all benefitting from the idea. Therefore, we expect:

Hypothesis 3: (a) The negative effect of threatening a supervisor’s power motive on his/her intention to support an idea (H1) is moderated by the idea-presenting employee’s power motive, such that the power threat effect will be stronger when the employee’s power motive is perceived to be high rather than low.

(b) This moderator effect is mediated by supervisor-attributed prosocial and egoistic intentions of the employee.

Overview of studies

To test our hypotheses, we conducted three studies with supervisors, using the experimental vignette methodology (Aguinis & Bradley, 2014) to simulate a situation of evaluating a specific idea. In all three studies, the described idea was designed to address a supervisor’s power motive by involving changes in decision authority between supervisor and subordinates. In Study 1, all participants read a power-threatening idea, and we tested Hypothesis 1 as the effect of supervisors’ power motive on their intention to support the idea. In Study 2, we introduced the idea-presenting employee’s power motive (high versus low) as an experimental factor, and additionally tested Hypotheses 2 and 3 (see Figure 1). In Study 3, we sought to replicate the previous results. Extending the design of Study 1 and 2, we also manipulated the idea’s consequences in terms of power threat versus power gain for the supervisor in order to investigate if supervisors high on power motivation are likewise (un-)supportive of ideas if they serve to satisfy their need for power. In all studies, the presenting employee was described as being male.

Figure 1. Conceptual model of Study 1 and Study 2. Paths in dashed lines were subject to Study 2.

Study 1

Method

Participants and procedure

Data were collected in one site of a large German company operating in communication services and logistics. Due to data protection issues, we relied on the site manager to circulate the URL to our online survey via e-mail among the company’s supervisors. Sixty supervisors completed the survey. Of these, 32% (n = 19) were female. On average, participants were 49.1 years old (SD = 5.75), had been working for that company for 25.1 years (SD = 7.96), and had been in a supervisory position for 16.5 years (SD = 7.87). Eighty-five percent of the participants held a university degree; 15% completed a professional training.

We first assessed participants’ explicit power motive. Then, they were asked to read a scenario containing a fictitious idea brought up by a subordinate employee. To make sure participants had read the scenario attentively, we asked them to take notes about the consequences the implementation of this idea would have for them personally. After this, we captured supervisors’ intention to support the idea, and asked for their demographics.

All participants were presented with the same fictitious idea. The idea was designed to threaten the satisfaction of a high power motive in supervisors if implemented. The presented idea concerned the annual performance review for employees. The written scenario described that, so far, employees’ annual performance review was solely based on the appraisal of each employee’s supervisor. Now, an employee suggested to additionally include co-workers’ appraisal in this performance review to make it more valid and reliable (see Appendix for the full scenario). If the idea was implemented, supervisors would – at least in part (i.e., to an extent of 50%) – have to give away control and allow others to take over their responsibility.

We performed a pilot study to test the power-threatening potential of the idea in a sample of 131 psychology students. Participants rated three items that described power-related consequences of the idea, i.e., “the supervisor would have to give away control”, “the supervisor would gain influence” (reverse coded) and “it would question the supervisor’s authority” (α = .60). The mean rating of these items (rated from 1 = not at all true to 5 = completely true) was M = 3.60 (SD = 0.70); compared to the theoretical scale mean (3), power threat can be considered as above average, t(130) = 9.84, p < .001.

Measures

Supervisors’ explicit power motive (α = .66) was assessed with the German version of the dominance subscale of the Personality Research Form (Jackson, 1984; Stumpf, Angleitner, Wieck, Jackson, & Beloch-Till, 1985). To reduce the demands placed upon respondents, we applied an 8-item version of the original 16-item scale (scale reduction was based on a separate pilot study with 266 participants; information can be obtained from the first author). Participants rated in a dichotomous response format whether the different statements hold true for them (1) or not (0). The sum of true ratings was used as an indicator of supervisors’ power motive, with higher values indicating a higher power motive (potential range: 0–8). Sample items are “I try to control others rather than permit them to control me” and “I feel confident when directing the activities of others”.

Supervisors’ intentions to support the idea (α = .84) were captured by four items adapted from previous research on idea evaluation in the peer context (Urbach, Fay, & Lauche, 2016). The items reflect supportive behavioural reactions that are likely to occur in response to an employee’s idea proposal in an organizational context. The target “colleague” in the original items was replaced by “employee”. The items were “I will encourage the employee to take his idea further”, “I will take my time to thoroughly listen to the employee’s idea”, “I will commend the employee for his initiative” and “I will seriously consider what has to be done to implement the idea”. Participants were asked to rate how likely they would show the respective behaviours in response to the idea. Items were rated on a seven-point scale ranging from 1 = not at all to 7 = very much.

Results and discussion

Descriptive statistics, intercorrelations and internal consistency reliabilities of all study variables are shown in Table 1.

Table 1. Descriptive statistics, intercorrelations, and internal consistency reliabilities (Study 1).

Variable		M	SD	1	2	3	4	5
1	Age	49.05	5.75	–
2	Gender^a	0.32	–	−.11	–
3	Tenure as supervisor	16.47	7.87	.64**	−.10	–
4	Supervisor power motive	5.33	1.93	.16	−.06	.16	(.66)
5	Idea support	5.11	1.52	−.14	.04	.07	−.29*	(.84)

N = 60. Cronbach’s alphas are given in italics in parentheses on the diagonal.

^a0 = male, 1 = female.

^†p < .10, *p < .05, **p < .01

We conducted a multiple regression analysis to test Hypothesis 1. First, age (β = −.31, p = .061), gender (β = .04, p = .775; 0 = male, 1 = female), and tenure as a supervisor (β = .28, p = .108) were entered as control variables (R² = .065, p = .284); second, supervisors’ power motive was entered. As expected, supervisors’ intention to support the idea was negatively related to their power motive (β = −.29, p = .026, ΔR² = .081).

This result provides initial support for our assumption that a power threatening idea is less likely to be supported the more supervisors valued having power (i.e., having a high power motive). However, the study has two limitations. First, a shortcoming of the power motive scale lies in the dichotomous response format that is used in the original scales. Thus, we could not evaluate the factor structure of the modified version of the scale as the sample size was too small to use estimators in a CFA that would account for the categorical nature of the data, such as WLSMV (Nussbeck, Eid, & Lischetzke, 2006). Moreover, the internal consistency reliability of the power motive scale was rather low. To improve the fit of the response format to the distribution of the trait, we increased the number of response categories in Study 2 (Bandalos & Enders, 1996). Second, as all supervisors worked for the same organization, the generalizability of results to other samples and organizations needs to be established.

Thus, the aims of Study 2 were twofold: First, we extended the experimental design in Study 2 in order to test the effects of the presenting employee’s power motive (Hypotheses 2 and 3). Second, we sought to address the limitations of Study 1.

Study 2

Method

Participants

Data were obtained from a convenience sample of German supervisors working in the banking, financial service and insurance industry. Participation was anonymous and voluntary; participants could take part in a raffle to win two shopping vouchers (worth €30 each). The survey URL was circulated among the supervisors of seven organizations (>100 employees). In total, 193 participants accessed the survey; we excluded participants that did not complete the survey. The final sample consisted of 79 supervisors, 35% of which were female. On average, participants were 41.9 years old (SD = 7.59), had been working for their companies for 14.5 years (SD = 7.42) and were in supervisory positions for 8.59 years (SD = 5.52). All participants held a university degree (65%) or had completed a professional training (35%).

Design and procedure

We used the same procedure as in Study 1, and extended it in two ways to address Hypotheses 2 and 3. First, we expanded the scenario used in Study 1 by including a manipulation of the power motive of the idea-presenting employee. This employee was described as being either high or low in power motivation in terms of his typical behaviour at work. The employee high in power motivation was described as someone who likes to control and supervise co-workers and seeks recognition from the supervisor, but who can also be very charismatic and motivating to others. The latter aspects were included to reduce the negative connotation of the control component of the power motive by also highlighting positive aspects that are related to the power motive.¹ The employee low in power motivation was described as an unassuming person who acts rather submissively within his team, who does not seek special attention and who is perceived as a typical follower (see Appendix). Participants were randomly assigned to one of the two employee descriptions (high power motive n = 36, low power motive n = 43); equivalent to the procedure in Study 1, all participants then read the same power-threatening idea as used in Study 1 (see Appendix). As the idea involved including co-workers’ appraisal in employees’ performance reviews, the implementation of the idea would not only imply a loss of power for the supervisor but also a gain of power for the presenting employee.

Second, to test Hypothesis 2, we further assessed the degree of prosocial and egoistic intentions the supervisor attributed to the idea-presenting employee.

Manipulation checks

Participants rated two items that captured the perceived degree of the employee’s power motive (“likes to keep things under control”) and the perceived loss of control regarding the idea (i.e., power threat, from 1 = not at all true to 5 = completely true). As intended, participants rated the employee described as highly power-motivated (M = 4.09, SD = 0.66) significantly higher in power motivation than the employee described as lowly power-motivated (M = 2.67, SD = 0.95), F(1, 77) = 55.37, p < .001, η² = .425. The perceived loss of control for the supervisor if the idea was implemented was moderate (M = 3.18, SD = 1.37).

Measures

Supervisors’ explicit power motive was assessed by the same items as in Study 1. To increase the sensitivity of the scale and to adjust the dichotomous response format to commonly used Likert scales, we used a four-point response format, ranging from 1 = not at all true to 4 = completely true. Internal consistency improved slightly (α = .75).

We utilized two four-item scales developed by Urbach et al. (2016) to measure attributed prosocial and egoistic intentions of the idea-presenting employee. We instructed participants to think of the idea-presenting employee (instead of the “colleague” targeted in the original items) and to rate to which extent the given intentions were true for this person. Sample items are “The employee wants to help improving everyone’s situation” (prosocial intentions) and “The employee wants to safeguard his/her own interests” (egoistic intentions). Ratings were made on a five-point scale (1 = not at all true to 5 = completely true). The means of the two sets of four items were used as measures for attributed prosocial and egoistic intentions, respectively.

Supervisors’ intention to support the idea was measured as in Study 1. To avoid confusion with the terms for attributed prosocial and egoistic intentions, we will in short refer to supervisors’ intention to support the idea as “idea support” in the following.

Results and discussion

Descriptive statistics, internal consistency reliabilities, and intercorrelations of all study variables are presented in Table 2. We conducted multiple regression analyses to test Hypotheses 1, 2a, and 3a; results are summarized in Table 3. In all analyses, we controlled for supervisors’ age, gender and tenure as a supervisor.

Table 2. Descriptive statistics, intercorrelations, and internal consistency reliabilities of the study variables (Study 2).

Variable		M	SD	1	2	3	4	5	6	7	8
1	Age	41.92	7.59	–
2	Gender^a	0.35	–	−.06	–
3	Tenure as supervisor	8.59	5.52	.58**	.16	–
4	Supervisor power motive	2.68	0.47	−.07	−.12	−.18	(.75)
5	Employee power motive^b	0.46	–	.05	−.03	−.02	.33**	–
6	Prosocial intentions	3.31	0.76	.09	−.07	.05	−.28*	−.43**	(.73)
7	Egoistic intentions	2.92	0.99	−.09	.05	−.23*	.27*	.28*	−.38**	(.86)
8	Idea support	4.25	1.65	.16	.04	.14	.09	−.23*	.34**	−.08	(.87)

Pearson correlations; sample sizes differ due to pairwise deletion of missing values (N = 76–79). Cronbach’s alphas are given in italics in parentheses on the diagonal.

^a0 = male, 1 = female.

^bExperimental factor: 0 = low power motive, 1 = high power motive.

^†p < .10, *p < .05, **p < .01.

Table 3. Multiple regression results (Study 2).

		Prosocial intentions		Egoistic intentions		Idea support
Step	Predictors	β	∆ R^2	Β	∆ R^2	β	∆ R^2
1	Age	.09		.07		.11
	Gender^a	−.05		.07		.04
	Tenure as supervisor	.01		−.28^†		.10
2	Supervisor power motive (SM)	−.31**	.091**	.25*	.062*	.11	.012
3	Employee power motive (EM)^b	−.40**	.138**	.24*	.051*	−.28*	.071*
4	SM×EM	−.34*	.060**	−.23	.028	−.37*	.074*
	Total R^2	.303		.199		.194

N = 76 due to listwise deletion of cases with missing values.

^a0 = male, 1 = female.

^bExperimental factor: 0 = low power motive, 1 = high power motive.

^†p < .10, *p < .05, **p < .01.

As in Study 1, all participants read a power-threatening idea. Thus, Hypothesis 1 suggests a negative effect of supervisors’ power motive on their idea support (see Figure 1). In contrast to Study 1, we did not find this effect here (β = .11, p = .342, ΔR² = .012).

In support of Hypothesis 2a, we found that supervisors’ idea support was lower when the power motive of the idea-presenting employee was described as high rather than low (β = −.28, p = .020, ΔR² = .071). Hypothesis 2b proposed that this effect is mediated by the level of prosocial and egoistic intentions attributed to the employee. We conducted a multiple mediation analysis to test for the significance of both indirect effects at a time (using the PROCESS macro by Hayes, 2013); we additionally controlled for supervisors’ power motive. As proposed, the employee’s power motive significantly predicted the level of attributed prosocial (B = −0.67, SE = 0.19, p < .001) and egoistic intentions (B = 0.50, SE = 0.24, p = .042). However, only attributed prosocial intentions were in turn related to supervisors’ idea support (B = 0.69, SE = 0.25, p = .008; egoistic intentions: B = 0.11, SE = 0.19, p = .582). Accordingly, only the indirect effect via attributed prosocial intentions was significant, bootstrapped ab = −0.45, SE = 0.24, bias corrected and accelerated 95% CI [−1.10; −0.10]; attributed egoistic intentions: ab = 0.06, SE = 0.12; 95% CI [−0.14; 0.39]. Hence, Hypothesis 2b was partly supported.

Hypothesis 3a proposed that supervisor’s idea support would be lowest when the idea’s consequences threatened the supervisor’s power motive and, at the same time, served to satisfy the employee’s power motive. The proposed two-way interaction between supervisors’ power motive and the employee’s power motive was significant (β = −.37, p = .014, ΔR² = .074; see Table 3). The plot of this interaction is depicted in Figure 2 (diagram on the left). Simple slope analyses revealed that the supervisor’s power motive had a negative effect on supervisors’ idea support when the idea-presenting employee was described as being high in power motivation, B = −0.60, t(69) = −2.35, p = .022, but not when he was described as being low in power motivation, B = 0.35, t(69) = 1.80, p = .077.

Figure 2. Supervisors’ and employees’ power motivation interact to predict supervisors’ evaluation of an idea and prosocial intentions attributed to the employee (Study 2).

In Hypothesis 3b, we proposed that this moderation effect is mediated by the prosocial and egoistic intentions attributed to the employee. Conditional process analysis (Hayes, 2013) lends support to this hypothesis for attributed prosocial intentions (indirect effect of the Supervisor Power Motive × Employee Power Motive interaction, B = −0.24, SE = 0.18, 95% CI [−0.73, −0.02]), but not for attributed egoistic intentions (B = −0.00, SE = 0.08, 95% CI [−0.21, 0.15]). As depicted in Figure 2 (diagram on the right), the attribution of prosocial intentions was lowest when both supervisor and employee were highly power motivated. The level of prosocial intentions attributed to the employee was in turn positively related to supervisors’ intention to support the idea. Accordingly, the conditional indirect effect of supervisors’ power motive on idea support was significant when the employee was described as high on power motivation, boot-strapped ab = −0.22, SE = 0.14, 95% CI [−0.61, −0.02], but was not significant when the employee was described as low on power motivation, boot-strapped ab = 0.02, SE = 0.08, 95% CI [−0.10, 0.26].

The results of Study 2 support our proposition that a power struggle between supervisor and employee can affect supervisors’ intention to support an idea: A power-threatening idea was less likely to find support if the presenting employee was not perceived to hold prosocial intentions, but was trying to gain power himself. This effect was particularly strong when supervisors also had a high power motive.

In contrast to Study 1, we did not find a main effect of supervisors’ power motive on idea support here. A possible reason for this is the manipulation of the employee’s power motive, such that supervisors only perceived the idea as threatening when the presenting employee was described as highly power-motivated, but not when there was no reason to believe that the employee was striving for power himself. To provide a post-hoc test of this assumption, we re-analysed the manipulation check item that assessed perceived loss of control when the idea was implemented. Indeed, the manipulation of the idea presenter’s power motive significantly affected supervisors’ loss of control perception regarding one and the same idea (high employee power motive, M = 3.53, SD = 1.34; low employee power motive, M = 2.86, SD = 1.32; F(1, 77) = 4.94, p = .029, η² = .060). Thus, the power-threatening potential of the idea was dependent on the employee’s power motive.

While there was evidence that attributing prosocial intentions to the presenting employee affects supervisors’ intention to support an idea, we did not find support for attributed egoistic intentions as a mediator. Even though supervisors attributed higher levels of egoistic intentions to the highly power-motivated presenter, this effect did not carry through to their intention to support the idea. Overall, the results underscore that attributed intentions are a meaningful process variable here.

Limitations of this study are threefold. First, despite the modification of the response format of the power motive scale, its internal consistency was only acceptable. As a consequence, we applied a different measure in Study 3.

Second, while we overcame a limitation of Study 1 and collected data of a more heterogeneous sample in Study 2, the sample size we obtained was rather small. Although the significant results indicate that statistical power was not a problem here, the generalizability as well as the stability of results needs to be established in a larger sample.

Third, a further refinement of our methodological approach is warranted. In Studies 1 and 2, we argued that the content of the scenario poses a potential threat to the supervisor’s power motive. Actually, the scenario contains two potential sources of power threat: First, as intended, the described idea implies changes in work procedures that would threaten a supervisor’s position power. Second, however, the mere fact that a subordinate advanced an idea on changes to work procedures could be perceived as power-threatening by supervisors (Grant et al., 2011). This alone could have accounted for the negative relationship between supervisors’ power motive and their support intentions. Thus, to test whether this effect was observed due to the content of the idea, we need to investigate whether supervisors’ power motive similarly is related to their intention to support an idea that has the potential to enhance their power, such that supervisors are more likely to support it. To address this limitation, we extended the experimental design in Study 3, and systematically varied the power-related consequences of the idea (power threat vs. power gain for a supervisor). Figure 3 illustrates the conceptual model tested in Study 3.

Figure 3. Moderated mediation model tested in Study 3. Note that for reasons of parsimony, the paths testing Hypothesis 2a and Hypothesis 2b were not included in this figure.

Study 3

Method

Participants

Participants for this study were recruited via a professional German access panel in order to obtain a large sample of working individuals who hold supervisory positions. The access panel provider distributed the URL to our online survey and compensated participants after completing the survey. The initial sample consisted of N = 284 participants who completed the survey. To ensure high data quality, we checked several indicators of potential careless responding (Huang, Curran, Keeney, Poposki, & DeShon, 2012). At the end of the survey, participants rated how earnestly and conscientiously they had filled in the questionnaire (from 1 = not at all to 5 = fully); we excluded 16 participants with ratings of three and lower. To ensure that participants had read the scenarios carefully, we checked how much time they had spent on the respective website. Based on average adult reading rates of 200–250 words per minute (e.g., Lewandowski, Codding, Kleinmann, & Tucker, 2003), we excluded 8 participants who spent less than 60 s with their scenario (which had a 269 to 301 word count, plus taking notes as in previous studies).

The final sample of this study was N = 260. Of these, 26% were female; the mean age was M = 44.5 years (SD = 9.45), ranging from 18 to 65. Participants were recruited in medium to large organizations (>250 employees), and worked in various industries, e.g., retail and touristic services (19.2%), IT and media (14.2%), production (13.8%), health and social services (11.2%), administration (8.5%), transport and logistics (6.5%), and others (26.6%). On average, participants had worked for their current employer for 15.3 years (SD = 10.3) and held supervisory positions for 9.2 years (SD = 7.5). Almost all participants held a vocational qualification (28.5%) or a university degree (71.2%), and worked full time (96.2%).

Design and procedure

The study procedure was the same as in the previous studies. We extended the design of Study 2 by manipulating a second factor, i.e., the power-related consequences of implementing the idea as power threat versus power gain. The scenarios were based on the idea presented in Studies 1 and 2, which dealt with the annual performance review based on supervisors’ and peers’ performance ratings. To create a situation of power gain and power threat, respectively, we altered the percentage of supervisor’s and colleagues’ share in the annual performance reviews. In the power threat condition, we described a change in the shares in the performance reviews from “60% supervisor appraisal/40% co-worker appraisal” to “80% co-worker appraisal/20% supervisor appraisal”. In the power gain condition, we described a change in shares from “40% supervisor appraisal/60% co-worker appraisal” to “80% supervisor appraisal/20% co-worker appraisal”. To support this manipulation, we also offered different arguments that would speak for such a change in procedures (see Appendix for full scenarios).

As in Study 2, we manipulated the idea-presenting employee’s power motive at two levels (high vs. low). This yields a 2 (power-related idea consequences: power gain vs. power threat) × 2 (employee power motive: high vs. low) factorial design; supervisors’ power motive was treated as a continuous moderator. Random assignment to conditions yielded a slightly skewed distribution of participants (n₁ = 55, n₂ = 59, n₃ = 80, n₄ = 66).

Manipulation checks

Participants rated the same manipulation check items as described in Study 2. As intended, participants anticipated a higher loss of influence in the power threat (M = 3.73, SD = 1.24) compared to the power gain condition (M = 2.05, SD = 1.37), F(1, 256) = 104.11, p < .001, η² = .289, and they perceived the employee described as highly power-motivated (M = 3.91, SD = 0.91) to be higher in power motivation than the presenter described as lowly power-motivated (M = 3.43, SD = 0.95), F(1, 256) = 55.37, p < .001, η² = .061.

To further ensure the discriminant validity of our manipulations, we assessed whether the scenarios also evoked other than power-related associations. As the power motive is related to the achievement (r = .35) and the affiliation motive (r = .15; e.g., based on the PRF measure, Schönbrodt & Gerstenberg, 2012), we assessed potential achievement- and affiliation-related consequences of the idea, using one item each. Moreover, workload reduction has been found as a predictor of individuals’ support for ideas (Urbach et al., 2016); thus, we assessed possible changes in workload for the supervisor with one item. Albeit much smaller in effect size than the power-related consequences, these characteristics differed between conditions: The idea described in the power gain condition was associated with higher levels of perceived need to demonstrate competence (gain: M = 3.58, SD = 1.21; threat: M = 2.91, SD = 1.16), F(1, 256) = 18.66, p < .001, η² = .068, and a higher potential to become unpopular as a supervisor if one supported the idea (gain: M = 2.66, SD = 1.13; threat: M = 2.36, SD = 1.15), F(1, 256) = 4.47, p = .035, η² = .017. The idea described in the power threat condition was associated with a slightly higher level of workload reduction for the supervisor (threat: M = 2.89, SD = 1.21; gain: M = 2.51, SD = 1.22), F(1, 256) = 6.78, p = .010, η² = .026. To account for these differences in the scenarios, we statistically controlled for these idea characteristics in the hypothesis tests.

Moreover, two items captured the perceived achievement and affiliation motivation of the idea-presenting employee. There were no significant differences between conditions (both η² < .005); the average ratings were moderate (achievement motive: M = 3.84, SD = 0.94; affiliation motive: M = 3.27, SD = 1.05).

Measures

Participants’ power motive was captured by two subscales taken from a German leadership motivation inventory (Felfe, Elprana, Gatzka, & Stiehl, 2012). The striving for influence motive component measures the tendency to seek positive outcomes associated with power, i.e. appreciating to be in control; e.g. “I feel comfortable with being in control of what happens in my surroundings”. The fear of losing control motive component assesses the tendency to avoid negative outcomes of power, i.e., being responsible, or worrying not to meet expectations; e.g. “When others have to rely on me, I am strongly concerned about making a mistake”. Each motive component was captured by six items; participants rated to what extent the respective item was true for them (from 1 = not at all true to 5 = completely true). The results on the fear of losing control motive component are not reported here but can be obtained as an online supplement to this manuscript.

In addition to the measure applied in Studies 1 and 2, supervisors’ idea support intentions were assessed by the 5-item idea endorsement scale developed by Burris (2012). Sample items are “How likely is it that you will take this person’s comments to your supervisors?” and “I think this person’s comments should be implemented” (rated from 1 = very unlikely/strongly disagree to 5 = very likely/strongly agree).

The intentions attributed to the employee were measured as in Study 2.

Results and discussion

Descriptive statistics, intercorrelations and internal consistency reliabilities of all study variables are shown in Table 4. The high correlation of our idea support measure (r = .85) with the idea endorsement scale supports the validity of our measure.

Table 4. Descriptive statistics, intercorrelations, and internal consistency reliabilities of the study variables (Study 3).

Variable		M	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1	Age	44.51	9.54	. –
2	Gender^a	0.26	.–	−.19**	. –
3	Tenure as supervisor	9.17	7.53	.61**	−.18**	. –
4	Achievement-related idea consequences	3.28	1.24	−.04	.03	−.05	. –
5	Affiliation-related idea consequences	2.53	1.15	−.30**	.08	−.21**	.26**	. –
6	Workload reduction	2.67	1.23	−.08	−.05	−.09	.21**	.30**	. –
7	Power-related idea consequences^b	0.56	.–	.07	.12^†	.07	.27**	.13*	−.15*	. –
8	Employee power motive^c	0.48	.–	−.01	.09	.02	−.14*	.00	−.11^†	−.07	. –
9	Supervisor power motive - striving for influence	3.79	0.70	.01	.13*	−.01	.30**	.02	.13*	.07	−.01	(.85)
10	Supervisor power motive - fear of losing control	2.54	0.99	−.37**	.12^†	−.22**	.17**	.30**	.29**	−.05	−.13*	.11	(.90)
11	Idea support	4.49	1.60	−.07	−.04	−.06	.36**	.09	.32**	.14*	−.22**	.16*	.29**	(.90)
12	Idea endorsement	3.04	1.24	−.08	−.02	−.09	.39**	.11^†	.32**	.18**	−.21**	.06	.28**	.85**	(.95)
13	Prosocial intentions	3.17	1.03	−.09	−.02	−.07	.32**	.14*	.40**	.02	−.34**	.11^†	.35**	.69**	.66**	(.89)
14	Egoistic intentions	3.54	0.84	−.09	.08	.01	.14*	.17**	.04	.13*	.20**	.22**	.05	−.09	−.11^†	−.15*	(.80)

N = 260. Cronbach’s alphas are given in italics in parentheses on the diagonal.

^a0 = male, 1 = female.

^bExperimental factor: 0 = power threat, 1 = power gain.

^cExperimental factor: 0 = low power motive, 1 = high power motive.

^†p < .10, *p < .05, **p < .01.

In all analyses, we controlled for supervisors’ age, gender and tenure as a supervisor. We further controlled for the perceived achievement-, affiliation- and workload-related idea consequences.² Results are summarized in Table 5.

Table 5. Multiple regression results (Study 3).

		Prosocial intentions		Egoistic intentions		Idea support		Idea endorsement
Step	Predictors	β	∆ R^2	β	∆ R^2	β	∆ R^2	β	∆ R^2
1	Age	−.08		−.11		−.08		−.05
	Gender^a	−.02		.07		−.04		−.03
	Tenure as supervisor	.01		.12		.00		−.04
	Achievement-related idea consequences	.26**		.11^†		.32**		.35**
	Affiliation-related idea consequences	−.05		.13^†		−.09		−.08
	Workload reduction	.35**	.222**	−.02	.054*	.28**	.204**	.26**	.217**
2	Power-related idea consequences (IC)^b	.02		.10		.14*		.18**
	Employee power motive (EM)^c	−.25**		.21**		−.12*		−.10^†
	Supervisor power motive - fear of losing control	.22**		−.02		.20**		.19**
	Supervisor power motive - striving for influence (SM)	−.02	.108**	.19**	.083**	.03	.068**	−.09	.073**
3	IC×EM	−.06		.09		.02		−.01
	IC×SM	.07		−.15		.11		.04
	EM×SM	−.00	.003	−.21*	.028*	.07	.006	.03	.001
4	IC×EM×SM	.22*	.012*	.15	.006	.25*	.015*	.24*	.014*
	Total R^2		.345		.171		.293		.306

N = 260. Results on fear of losing control component of supervisors’ power motive are available in an online supplement to this article.

^a0 = male, 1 = female.

^bExperimental factor: 0 = power threat, 1 = power gain.

^cExperimental factor: 0 = low power motive, 1 = high power motive.

^†p < .10, *p < .05, **p < .01.

In Hypothesis 1, we proposed that a supervisor’s intention to support a power-threatening idea should be lower the higher his/her power motive is. To test Hypothesis 1 in Study 3 where we have manipulated the idea consequences (power gain vs. power threat), we computed the effects of the Idea Consequences × Supervisor Power Motive interaction on idea support and idea endorsement. Effects were not significant (see “IC × SM” in Table 5).

Hypothesis 2 stated that (a) supervisors’ idea support would be lower if the employee was perceived to be highly power-motivated, and thus serve his/her own motives with the proposed idea, i.e., gaining valued power him-/herself; (b) attributed intentions should mediate this effect. Results on the Idea Consequences × Employee Power Motive interaction neither confirm this for idea support and endorsement nor for the attributed intentions (see “IC × EM” in Table 5).

Hypothesis 3a proposed a complex interplay of the idea consequences, the supervisor’s power motive and the employee’s power motive, such that the negative effect of threatening a supervisor’s power motive should be stronger if the employee was seen as striving for influence. In line with this assumption, results revealed significant three-way interaction effects on idea support and idea endorsement (see “IC × EM × SM” in Table 5). The respective interaction effect on idea endorsement is plotted in Figure 4 (diagram on the left). The graph shows that the employee’s power motive indeed affected the pattern of the Idea Consequences × Supervisor Power Motive interaction: When the idea-presenting employee was described as high on power motivation (see slopes with white end marks), supervisors’ high on power motivation were far less likely to support the power threat than the power gain idea (simple slope, B = 0.81, SE = 0.28, p = .005); supervisors low in power motivation did not differ in their idea support for the idea involving power threat or power gain, respectively (B = 0.07, SE = 0.27, p = .801). This is in line with Hypothesis 3a. When the idea-presenting employee was described as low on power motivation (see slopes with black end marks), supervisors high on power motivation did not differ in their support for the power threat versus power gain idea (B = 0.22, SE = 0.27, p = .435); supervisors low on power motivation, however, were more likely to support the power gain than the power threat idea (B = 0.74, SE = 0.28, p = .010). The latter effect goes beyond our predictions in Hypothesis 3a. The result pattern for the dependent variable idea support (not depicted) was the same, with the exception that the unexpected effect in the low employee and low supervisor power motive condition missed conventional levels of significance (B = 0.63, SE = 0.37, p = .087). To sum up, in line with Hypothesis 3a, idea endorsement and idea support were lowest when both the supervisor and the idea-presenting employee were high in power motivation, and the proposed idea threatened the satisfaction of the supervisors’ power motive.

Figure 4. Three-way interaction effects of the idea consequences, the supervisor’s power motive and the employee’s power motive on supervisors’ idea endorsement and the prosocial intentions attributed to the employee (Study 3).

Hypothesis 3b stated that this complex three-way interaction effect was mediated by the level of prosocial and egoistic intentions attributed to the employee. We computed a conditional process analysis (Hayes, 2013) to test this mediated moderation hypothesis. Results revealed a marginally significant indirect effect of the Idea Consequences × Supervisor Power Motive × Employee Power Motive interaction via attributed prosocial intentions on idea support, B = 0.44, SE = 0.25, 90% CI [0.01, 0.86], and idea endorsement, B = 0.32, SE = 0.18, 90% CI [0.01, 0.62]. Indirect effects via attributed egoistic intentions were not significant (both p > .10). Figure 4 (diagram on the right) illustrates the three-way interaction effect on attributed prosocial intentions (β = .22, p = .035, Table 5). The result pattern is the same as described for idea endorsement with regard to Hypothesis 3a (diagram on the left). When the employee was described as high on power motivation (see slopes with white end marks), we can observe differences in supervisors’ attribution of intentions relating to their own power motive (conditional effect of the Idea Consequences × Supervisor Power Motive interaction, B = 0.33, SE = 0.16, p = .037). In particular, the higher supervisors’ own power motive was, the more prosocial intentions they attributed to the employee for the idea involving power gain for the supervisor; vice versa, supervisors attributed lower levels of prosocial intentions when the idea involved power threat for them. There was no such effect when the employee was described as low on power motivation (see slopes with black end marks, B = −0.15, SE = 0.16, p = .360). As noted above, attributed prosocial intentions were in turn positively related to idea support and idea endorsement. With regard to attributed prosocial intentions, these results lend cautious support to Hypothesis 3b.

Taken together, the results of Study 3 underscore that idea evaluation is a complex process, involving the interplay of the characteristics of the idea, the evaluating supervisor and the idea-presenting employee. Extending the results on the power-threatening idea presented in Study 1 and Study 2, Study 3 shows that supervisors high on power motivation also responded more positively to the possibility of gaining power. This differential effect of supervisors’ power motive leads us to conclude that the negative effects of supervisors’ power motive on idea support in Study 1 and Study 2 were due to the power-threatening content of the idea, not that the idea has been voiced by a subordinate overstepping his powers.

An interesting finding of Study 3 is that when both supervisor and employee scored low on power motivation, supervisors would be more likely to support the power gain than the power threat idea. Although we did not expect this effect, it does not necessarily contradict Hypothesis 3a: Following trait activation theory (Tett & Guterman, 2000), power-related issues should not be salient to supervisors low on power motivation, particularly when the supervisors’ assessment of the employee also does not contain power-related cues. Under these circumstances, supervisors are likely to focus on the content of the idea, and to see that the employee may indeed pursue an urgent request with his proposal. The power gain idea used in this study implied that the supervisor gains in say in performance reviews. This could have been interpreted as a subordinate’s demand for the supervisor to become more active, resulting in enhanced idea endorsement or support.

A limitation of Study 3 relates to the manipulation of power gain and power threat. While it makes the description of an idea proposal more realistic to include arguments why the employee suggests certain changes, these arguments could not be held constant for the two opposing ideas presented. This additional source of variance may have produced differences between the two experimental conditions that go beyond power gain or power threat.

General discussion

Previous research points to the role of supervisor support for the successful implementation of employees’ ideas for constructive change (e.g., Axtell et al., 2000; Škerlavaj, Černe, & Dysvik, 2014) and other change-oriented employee behaviours (for a review, see Chiaburu, Lorinkova, & Van Dyne, 2013). The central role of supervisors lies in their power to formally approve of ideas, and to provide resources necessary for their implementation. We contribute to this research by pointing out reasons why supervisors might not be willing to support employees’ ideas. Our research suggests that supervisors’ intention to support employees’ ideas may be undermined by their personal preferences, i.e., their power motivation. Across all studies, power threatening ideas were less likely to be supported the more supervisors valued having power (i.e., having a higher power motive). In Studies 2 and 3, this effect was dependent on the presenting employee’s power motive. We assume that this finding is based on a power struggle between supervisor and employee, which can be a reason for denying support for employees’ ideas.

Theoretical and practical implications

We contribute to research on proactive work behaviours by adding a new perspective to existing research paradigms: Even though employees’ contribution of ideas is often wished for in organizations, good ideas may be turned down for reasons that are not based on the organization’s benefit. Thus, going beyond existing research that focused on factors stimulating employees’ ideas, we identified characteristics of the social context that can benefit or hinder the actual implementation of these ideas. Our results lend support to the long-standing notion that “‘political variables’ may play a larger role [in idea evaluation], especially the acquisition of ‘power tools’ to move the idea forward” (Kanter, 1988, p. 186). We explored the role of supervisors’ personal power motivation to explain individual differences in their responses to ideas. In line with various theories of human motivation (e.g., Lazarus, 1991; McClelland, 1985; Tett & Guterman, 2000) as well as research on leadership and power (e.g., Maner et al., 2007; Maner & Mead, 2010; Williams, 2014), deciding to support versus to reject an idea was related to supervisors’ personal goals. In line with dominance complementarity theory (Kiesler, 1983), we demonstrate that if an employee shows rather dominant behaviour (e.g., by proposing a power-threatening idea) or appears to be a dominant person, this may not complement a supervisor’s need for dominance; thus, supervisors might be less willing to support an idea proposed by such an employee. If we compare the results of Study 1 with the results of Studies 2 and 3, we see that the effect of threatening a supervisor’s power motive was dependent on the employee’s power motive as soon as this piece of information was provided. Thus, the social stimulus that one could lose influence to a power-striving subordinate was central in determining the power-threatening potential of the idea.

Moreover, our findings highlight the role of supervisors’ attributions of the employee’s behaviour in determining their support for an idea. If supervisors perceived the idea’s consequences to serve the employee’s own goals, supervisors attributed higher levels of egoistic intentions (Study 2) and lower levels of prosocial intentions to the employee (Study 2 and 3). However, only the level of attributed prosocial intentions carried through to supervisors’ intention to support the idea. It appears that an idea voiced by someone who seeks to serve others and might harm his or her own interests lends itself to the cognitive shortcut that the idea must be “good”. In contrast to this, an idea attributed to egoistic intentions does not necessarily seem to be evaluated as “bad”. Our results on the role of attributed prosocial intentions are in line with previous research on the evaluation of proactive and other work behaviour (e.g., De Stobbeleir et al., 2010; Grant & Ashford, 2008; Grant et al., 2009). Halbesleben et al. (2010) likewise report a positive attributional asymmetry for OCB, such that the negative relationship of self-serving impression management motives with performance ratings seemed to be less strong than the positive relationship of prosocial motives with performance ratings. Lam, Huang, and Snape (2007) investigated supervisor-attributed performance enhancement versus impression management intentions of a feedback-seeker. They found that feedback-seeking was positively related to leader-member exchange quality only when supervisors attributed employees’ feedback-seeking behaviour to performance enhancement intentions. Although both kinds of intentions can be perceived as self-serving, performance enhancement aims at individual goals that, in the long run, correspond to the organization’s goals of enhanced performance. In a similar way, an idea may seem self-serving at first glance, but have indirect or deferred benefits for the team or organization, too. If the supervisors recognized both outcomes, this should have both enhancing and diminishing effects on their idea support. This would explain the null relationships of attributed egoistic intentions and idea support we found in our studies.

We further extend the perspective of research on power in organizations by emphasizing the role of individual differences in power dynamics. The majority of psychological research on power in organizations has focussed on how having power or being put in a powerful position affects individuals’ affect, cognition, and behaviour (for recent reviews, see Anderson & Brion, 2014; Sturm & Antonakis, 2015). This literature suggests that being in power is, in general, a desirable state (e.g., Keltner et al., 2003). Accordingly, we observed a higher likelihood of supervisors to support an idea involving power gain rather than power loss (see Table 5, Study 3). This effect explained 1.6% of variance in idea support and 2.6% of variance in idea endorsement. Thus, threatening a supervisors’ position power seems to be a general drawback for an idea. However, our results underscore the incremental validity of taking supervisors’ personal striving for power and their perceptions of a power-challenging subordinate into account as moderators of position power effects on supervisors’ behaviour. In Study 3, all two- and three-way interactions involving these moderators accounted for an additional 2.1% of variance explained in idea support, and 1.6% explained in idea endorsement.

From a practical perspective, our results suggest that organizations may face the risk of losing out on worthwhile ideas for constructive change in case their consequences do not serve the motives of the supervisor deciding on its implementation. Interestingly, participants of Study 1 worked in an organization with a very sophisticated idea capture scheme, and a proclaimed interest in utilizing employees’ ideas and contributions. Thus, even though supervisors might have some level of expertise in dealing with employees’ ideas, this does not automatically produce disinterested decisions. Leach et al. (2006) recommend idea management schemes that are characterized by short distances between idea-presenting employee, the field of implementation, and the person deciding about the idea’s implementation. This recommendation allows for a robust test of an idea in terms of practicality and feasibility; at the same time, as our results indicate, it enhances the likelihood that supervisors’ personal agenda can affect their idea support. Employees who want to bring their ideas to the attention of their supervisor should also be aware of these processes, and try to reflect on potential implications of their idea for their supervisor. If their idea would interfere with their supervisor’s power position, it might be advisable that this idea is brought forward by an employee who is not perceived as a particularly dominant person. This may reduce the risk that a supervisor feels threatened by the proposed changes, and therefore turns the idea down.

Limitations and future directions

We applied a scenario-based, quasi-experimental approach that allows for drawing causal inference on the effects of the manipulated variables. Furthermore, the simulation-based design of our studies enabled us to make up ideas that explicitly address supervisors’ power motive, which was necessary to provide a sound test of our hypotheses. Moreover, we could present the same ideas to all participants, ensuring that the ideas to be evaluated did not differ with regard to aspects other than the intended, e.g., the kind and scope of changes proposed, or other characteristics of the employee or the organizational context. At the same time, this design comes along with limitations of the external validity of our results (Grant & Wall, 2009). Therefore, future research needs to replicate the obtained results investigating ideas in a more naturalistic context. We used a manipulation to convey information on the subordinates’ personality; this does not take place in a naturalistic work setting. In the work context, supervisor and employee may have developed assumptions about the other’s usual behaviour and general values based on day-to-day observations (Grant et al., 2009). It is likely that these perceptions would be much more nuanced than presented in our scenarios. Field studies may further include other variables that are likely to influence supervisors’ perceptions of the idea-presenting employee and their intentions to support an idea, such as the relationship quality between supervisor and employee in terms of their mutual levels of trust, respect and liking (e.g., Graen & Uhl-Bien, 1995). Supervisors’ willingness to delegate responsibility to an employee and to take risks is higher when relationship quality is high (e.g., Bauer & Green, 1996). Accordingly, an idea might be evaluated more positively when relationship quality is high.

Another limitation to the scope of our studies lies in the exclusive focus on power-related consequences of an idea for change. Not all ideas will have power-related consequences to the same extent; when power-related attributes such as roles, responsibilities or procedures are not subject to change, then supervisors’ power motive will not be predictive of supervisors’ idea support. However, in this case, similar processes could take place with regard to affiliation- or achievement-related idea consequences and their propensity to satisfy supervisors’ need for achievement or affiliation (e.g., Urbach et al., 2016).

While we can draw causal inferences on the effects of the manipulated factors on the dependent variables, the relationships proposed among the dependent and mediator variables do not allow for this. All dependent variables were captured by explicit, direct measures through participants’ self-report; declaring attributed intentions as proximal outcomes and supervisors’ supportive behavioural intentions as distal outcomes was solely based on theoretical considerations. We cannot rule out that the correlations between these variables reflect common-method variance, but the fact that relationships are differential and that we also have zero effects speaks against a uniform inflation of correlations (Spector, 2006).

Further limitations refer to the procedures and measures we applied. Although behavioural intentions are seen as the best predictor of actual behaviour (Ajzen, 1991), future studies should capture a supervisor’s actual behaviour following an employee’s idea proposal. Moreover, in Study 2 and Study 3, we did not explicitly assess whether the changes presented in the context of the scenarios where actually new in the organizational realities of our participants. If participants already practiced peer-participated performance reviews, this is likely to have affected their responses to the scenarios. Another limitation is based in the rehearsal instructions we gave participants after they had read the scenarios, i.e., focussing on idea consequences for them personally. Future research needs to investigate whether instructing supervisors to reflect on the consequences of an idea for the organization can reduce or even eliminate the rather self-interested decision-making our results suggest. There is experimental research in social psychology that suggests that leaders can be distracted from self-interested actions (for an overview, see Williams, 2014).

Another limitation relates to gender differences with regard to the idea-presenting employee, who was presented as being male. There is reason to assume that ideas might be perceived differently depending on employees’ gender. While women are expected to be communal, men are expected to be agentic (e.g., Abele, 2003); accordingly, agency is not compatible with female role expectations (Eagly & Karau, 2002). Thus, coming up with an idea that aims at augmenting the employee’s power (agentic behaviour) might be perceived as even more detrimental when the employee is female as compared to male. Such gender-differential effects have been reported for OCB (Heilman & Chen, 2005).

Overall, we suggest that we need to broaden our research perspective on employees’ proactive work behaviours. Future research should move on from identifying predictors of the generation of these behaviours to exploring processes and conditions that might determine whether organizations can actually benefit from these behaviours.

Acknowledgments

We appreciate the support of Lydia Ristau and Grzegorz Szarowski in collecting the data for Study 1 and Study 2.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental data

Supplemental data for this article can be accessed here.

Notes

1. We conducted an additional study to test whether these positive aspects produce other than power-related differences in participants’ perceptions of the employee described as low versus high on power motivation. This was not the case. Detailed results can be obtained from the first author.

2. We reran the analyses of all studies reported in this manuscript without including control variables. Results did not change with the exception that the p-values of the three-way interactions reported in Study 3 (Table 5) dropped to p < .10. However, the result pattern was the same as depicted in Figure 4.

Appendix

Scenario Applied in Study 1

“So far, the employees’ annual performance review is solely based on our line manager’s appraisal. However, we spend most of our worktime with our colleagues and not with the supervisor. Therefore, I think our colleagues are a much better source of information on how staff gets their work done, how everyone contributes to the team, and how they handle difficult situations. For this reason, the annual performance review should not be based on the manager’s appraisal only. For example, 50% of the performance review could be based on the supervisor’s appraisal, and 50% could be based on our colleagues’ appraisal. We could also consider recording each individual colleague’s strengths and capabilities. This information could be used when vacant positions are to be filled. Then our appraisal would also affect replacement decisions. That way, the annual performance review would not only become more detailed and more precise, but it would also be easier to accept.”

Scenario Applied in Study 2

Description of the idea-presenting employee

Highly power-motivated employee:

“You have been working together with your subordinate Mr. M. for quite some time. Mr. M. always completes his tasks according to your instructions. He is an employee who enjoys tasks that involve managing or – to a certain extent – guiding his co-workers. Several times you have noted how well Mr. M. can motivate his colleagues for their jobs. You feel that Mr. M. seeks to impress you, because he is quite keen to talk to you about himself and his work. Colleagues around Mr. M. have an ambivalent attitude towards him: On the one hand, they tend to feel swayed by Mr. M. as he wants to impose his way of working on them; on the other hand, they appreciate him because he seems to understand them, and there is this inspiring streak in him.”

Lowly power-motivated employee:

“You have been working together with your subordinate Mr. D. for quite some time. Mr. D. is a quiet and unassuming employee who always accomplishes his tasks according to your instructions. Several times you have noted that Mr. D. has no problems at all to subordinate to others, and this way he has blended very well into the team. It does not seem like Mr. D. seeks to particularly impress you, because he accomplishes his work quietly and self-reliantly. The other employees describe him as a quiet fellow, but also as a pleasant person who acts with discretion, and who does not impose on others.”

Scenario Applied in Study 3

Description of the idea

Power gain:

“Once a year, we receive our performance review. So far, 40% of this review is based on our line manager’s appraisal, and 60% is based on colleagues’ appraisal. In my opinion, this weighting system is no good. I am convinced that line managers can assess their staffs’ performance much better than the co-workers can. Line managers are more experienced in appraising staff. And because they oversee quite a number of people they have a better standard of comparison. Apart from that, I am concerned that the co-workers’ performance appraisal is strongly biased by personal likes or dislikes. For these reasons, I would suggest that the line manager’s appraisal should have a bigger share in the total performance review, e.g., 80% (instead of currently 40%), and that the share of colleagues’ appraisal reduces, e.g., to 20% (instead of currently 60%).”

Power threat:

“Once a year, we receive our performance review. So far, 60% of this review is based on our line manager’s appraisal, and 40% is based on colleagues’ appraisal. In my opinion, this weighting system is no good. We spend most of our worktime with our colleagues and not with the supervisor. Therefore, I think our colleagues are a much better source of information on how staff gets their work done, and how everyone contributes to the team. Moreover, colleagues have a more realistic view of all the challenges one is faced with at times. Apart from that, I am concerned that the line managers’ performance appraisal is strongly biased by personal likes or dislikes. For these reasons, I would suggest that the colleagues’ appraisal should have a bigger share in the total performance review, e.g., 80% (instead of currently 40%), and that the share of the supervisors’ appraisal reduces, e.g., to 20% (instead of currently 60%).”