Acting on Leader’s Emotions: How and When Emotion Recognition Ability Motivates Voice?

Abstract

Purpose

Employee voice has been considered as an important means to understand the cutting-edge information, gain social status and performance advantage for leaders, employees and the organization, respectively. However, our knowledge about how and when employees’ emotions influence voice remains limited.

Design/Methodology/Approach

In order to better illustrate the role of emotion on voice, based on emotion as social information theory and similarity attraction theory, we proposed a research model through which emotion recognition ability affects voice via perceived ambidextrous leadership. A sample of 182 comprised of full-time employees and their 43 immediate supervisors was collected through questionnaires in China, and analyzed via hierarchical regression method.

Findings

We found that subordinate’s emotional recognition ability has a significant positive effect on promotive and prohibitive voice, and that perceived ambidextrous leadership plays a significant mediating role between subordinate’s emotional recognition ability and promotive voice, while no mediating role is found between subordinate’s emotional recognition ability and prohibitive voice. In addition, in contrast to leader-subordinate gender dissimilarity, leader-subordinate gender similarity is more effective in strengthening the impact of emotion recognition ability on perceived ambidextrous leadership, and thus promotes employee voice.

Originality/Value

This research not only advances our understanding of employee voice, but also provides specific reference for management practices from the perspective of gender.

Keywords:

• emotion recognition ability
• perceived ambidextrous leadership
• leader-subordinate gender similarity
• promotive voice
• prohibitive voice

Introduction

During the period of huge organizational transformation and adjustment, high environmental uncertainty make it inadequate for leaders to make accurate decision merely relying on information they grasp, which highlights the importance of subordinates’ voice.^1,2 Furthermore, numerous companies have regarded voice as an important indicator for employee promotion and salary increase, such as Huawei. Given the significance of voice, prior studies have examined its triggering mechanisms from multiple perspectives. For example, based on the constructive intention of voice, some scholars suggested that employee’s perceived insider status and psychological safety constitute the basis of voice.^3,4 In contrast, according to self-interested voice, there are studies stating that employee voice aims to gain high performance and social status.^5,6

Although researchers have attempted to demystify the antecedents of employee voice, existing studies seldom focus on the promotive or prohibitive effects of emotion on voice. In particular, employees’ emotional perception and emotional understanding of their leaders are lacking. Drawing on emotion as social information theory, emotions have a social signal function,⁷ and leaders’ emotional changing means a lot to employees. According to previous studies, the social reference effect of information enables to reduce employee’s behavioral decision-making deviation significantly,⁸ which further contributes to a win-win situation through voice.⁹ A study from Harvard Business Review suggests that “reading facial expressions” is a vital approach for employees to assess the risk of their voice.¹⁰ If employees capture the negative emotions of their leaders, they usually adopt a tepid work attitude owing to the consideration of psychological risk and resource loss.^11,12 Emotional recognition ability refers to an individual’s ability to perceive and understand other people’s emotions accurately, and thus guide his/her own behaviors.¹³

According to emotion as social information theory, it must be through a complex mechanism of emotional information process from “reading leaders’ expressions” to employee voice.¹⁴ A work from Harvard Business Review has unveiled the mechanism of leader emotions on employee behavioral decision-making.¹⁰ It stated that the open-loop character of cerebral limbic system enables employees to dig into the behavioral tendency from their leaders’ emotions, through which employees evaluate their voice risk. According to Meindl,¹⁵ leadership is in essence kind of perception constructed by subordinates subjectively, and resides in their mind for a long time. Wong and Law¹³ suggested that

employees’ recognition of their leaders’ emotional changing profoundly affects the quality of leadership behavior perception, and employees’ perceptions on leader behavior vary over time and contexts.

Thus, we assume that perceived ambidextrous leadership may bridge between “reading leaders’ expressions” and employee voice. Furthermore, due to gender differences, leader-subordinate gender (dis) similarity may have different impacts on the process from “reading leaders’ expressions” to employee voice.^16,17 Based on similarity attraction theory,¹⁷ subordinates having the same gender with their leaders are more likely to gain interaction opportunities and develop close working relationships with their leaders,¹⁹ which is helpful to grasp leaders’ behavioral and emotional intentions,²⁰ and thus enhance voice.²¹ Hence, our purpose is to examine the mediation effect of perceived ambidextrous leadership and the moderation effect of gender similarity in the relationship between employees’ recognition ability and voice.

This study makes three main contributions to the existing literatures. First, we combine emotion recognition ability with voice, which enriches the antecedents of voice. Although previous research primarily clarifies the mechanism of voice through employees’ constructive and instrumental intentions,^5,6 these studies focus on such cognitive factors as status perception, neglecting the roles of emotional factors. According to emotion as social information theory, emotion has the function of social signals. It offers reference information for employees’ behaviors, and can reduce their decision bias, thereby driving employees to make the right decision. Thus, our study extends the antecedents of voice from cognition field to emotion field, and provides new direction for the research on the antecedents of voice.

Second, our findings contribute to the ambidextrous leadership literature. Despite prior studies have examined the antecedents of voice from a single leadership’ perspective (eg, transformational leadership),²² leaders often present different leadership styles over time and space in the managerial practices.²³ In addition, compared with other leadership behaviors, ambidextrous leadership behavior is consistent with Chinese traditional thinking- “Yin-Yang Balance”. The extant research rarely focuses on the link among emotion recognition ability, ambidextrous leadership and voice. To fill this research gap, our study integrates ambidextrous leadership research with the research on voice and emotion.

Lastly, we also explore the role of the leader-subordinate gender similarity in the relation between emotion recognition ability and voice, and find that leader-subordinate gender similarity is more likely to promote employees’ voice. Although the existing research holds that leader-subordinate gender combination can significantly influence employees’ attitudes and behaviors,^19,20 it still confuses us that what role of leader-subordinate gender similarity plays in the process from “reading facial expressions” to voice, which hinders our understanding of employee behaviors from the perspective of gender. Our study realizes the integration of gender combination and voice, and provides new insights into the antecedents of voice.

Theoretical Background and Hypotheses

Emotion Recognition Ability and Employee Voice

Employee voice refers to employees’ behaviors to informally and voluntarily report problems at work to their leaders, and express their own suggestions to benefit organizational efficiency and transformation.²⁴ In light of its contents, voice can be divided into promotive voice and prohibitive voice. Promotive voice means that employees bring forward constructive suggestions toward process re-engineering and normative innovation to further advance organizational efficiency, emphasizing efficiency improvement while prohibitive voice refers to putting forward their own opinions regard problems impeding organizational efficiency such as inappropriate work flow and unreasonable organizational routines, aims at the improvement of organizational status quo. Since a meta-analysis study showed that promotive and prohibitive voice have a similar triggering mechanism,²¹ we, in this study, regard voice as a holistic construct.

Emotional recognition ability refers to an individual’s ability to perceive and understand other people’s emotions accurately, and thus guide his/her own behaviors.¹³ In this study, we focus on employees’ ability to assess and recognize their leaders’ emotions. According to emotion as social information theory, leader’s emotion is an interactive information, and has a social signal function.²⁵ It can be perceived and distinguished by employees through affective reactions process and inferential process.¹⁴ We argue that the influence of emotional recognition ability on employee voice is rooted in the transfer process from information reception to information function release. On the one hand, affective reactions process suggests that leaders’ emotional changing can directly affect employees’ emotions,²⁶ which, in turn, leads to employees’ quick recognition and assessment on their leaders’ emotions, engenders interpersonal effects at emotional level through emotional contagion.²⁷ It further makes a recurrence of these emotions in subordinates, and therefore has great influence on subordinate behaviors.²⁸ Employee voice is a typical extra-role behavior with high risk,²⁹ and has strong instrumental intention.^5,6 When employees capture the negative emotions of their leaders, they are inclined to keep silence owing to the consideration of psychological risk and resource loss.² On the other hand, inferential process indicates that employees view leaders’ emotional expressions (emotional information) as their own decision basis.¹⁴ Precious studies found that leaders are easy to achieve subordinates’ trust and voice when they show positive emotions.^8,30 Besides, a study from Harvard Business Review shows that when it is difficult to distinguish leaders’ emotions, employees will prioritize to button up their mouths for the purpose of resource reservation.¹⁰ Thus, we hypothesize:

H1 Employee emotional recognition ability positively influences their voice

The Mediating Role of Perceived Ambidextrous Leadership

In managerial practices, leaders often encounter two opposite tensions or ambidextrous leadership situations.³¹ It requires leaders to adjust, coordinate and integrate these two contrary and complementary leadership behaviors under a specific situation to benefit from its interaction effects.³² Scholars define these two behaviors as ambidextrous leadership behavior.^23,31 Referring to prior studies that tend to regard leadership behavior as a construct regarding subordinate perception,¹⁵ we define it as perceived ambidextrous leadership in this study. In contrast to any single leadership behavior, perceived ambidextrous leadership emphasizes leaders’ emotional changes over time and space.²³ For instance, adopting different leadership behaviors for employees with different positions,²³ or taking different leadership behaviors over time.³³ To date, the combinations of perceived ambidextrous leadership are mainly based on three perspectives: opening and closing leadership from the cognitive perspective, transformational and transactional leadership from the conventional perspective, and empowering and directive leadership from the perspective of power.³¹ The widespread use of transactional leadership in Chinese enterprises,³⁴ coupled with the highest efficacy of transformational leadership in contrast to other single leadership,³⁵ has led to adoption of transactional leadership and transformational leadership in this study.

According to emotion as social information theory, subordinate’s emotional recognition ability affects voice through affective reactions process and inferential process.²⁸ Subordinates draw on leaders’ emotions to judge leaders’ behavioral inclination, and the judgement differs over time and contexts.¹³ Therefore, we postulate that subordinate’s emotional recognition ability affects voice through perceived ambidextrous leadership. More specifically, subordinates with high emotional recognition ability are more likely to distinguish and evaluate their leaders’ emotions precisely,¹³ and process leadership behaviors via affective priming,³⁶ so as to appraise the risk of voice.⁴ For example, when subordinates discern such emotions as concern, support and hope, they are inclined to perceive transformational leadership featuring differential caring and intellectual enlightenment while perceiving transactional leadership featuring rule emphasis and explicit division of labor. Recent studies have indicated that the efficacy of ambidextrous leadership relies on subordinates’ perceptions.^23,37 Subordinates’ perceptions are influenced by subordinates’ recognition and assessment on their leaders’ emotions,¹³ and subordinates have a tendency to appraise the risk of voice through their perceptions.³⁸ By contrast, it is not easy for subordinates with low emotional recognition ability to sense leaders’ emotional changes, which undoubtedly weakens their understanding of leadership behaviors,¹³ and thus increases their psychological unsafety, and restrain voice.³ Therefore, we hypothesize as follows:

H2 Perceived ambidextrous leadership will mediate the relationship between emotional recognition ability and voice (promotive and prohibitive voice)

Leader-Subordinate Gender Similarity as a Moderator

Gender is a typical demographic characteristic.^39,40 Numerous studies have identified that there are role differences between male and female.^17,20 For instance, females are supposed to value helping, caring and kindness while males are regarded as confident and independent.^41,42 These gender differences are the determinants to comprehensively considering leader-subordinate gender matching when we conduct research on behaviors.¹⁹ Specifically, similarity attraction theory suggests that leader-subordinate similarity in the aspect of demographic characteristics, such as gender, age and race, can strengthen their mutual attraction,⁴³ and thus increase the willingness of interaction. Evidence has shown that the similarity attraction effect makes justification for why employees are easy to be attracted by leaders having similarity with themselves.¹⁸ Those employees try to figure out their leaders’ behavioral inclination by distinguishing their leaders’ emotional change in hopes of being “insider”.^3,44 After surveying the interview process, Graves and Powell⁴⁵ found that interviewers are more likely to interact with interviewees having the same gender. Frequent leader-subordinate interactions provide employees with more opportunities to convert emotional recognition into behavioral perception.⁴⁶ In contrast, leader-subordinate gender dissimilarity results in significant differences between them in terms of job role and recognition,⁴⁷ which is likely to induce working pressure and workplace ostracism,⁴⁸ and dampens employee enthusiasms of emotional recognition feedback and behavioral perception, thereby prohibitive voice.²² Based on the theoretical and empirical evidence presented above, we argue the following:

H3 When Leader-subordinates gender is consistent, the effect of subordinate’s emotional recognition ability on perceived ambidextrous leadership is stronger, thereby strengthening voice (promotive and prohibitive voice)

Figure 1 presents the research model suggesting the relationship of all the variables.

Figure 1 Shows the research model of this study. In the research model, we argue that subordinate’ s emotional recognition ability can influence both promotive and prohibitive voice through perceived ambidextrous leadership. Moreover, gender similarity (an leader and his/her subordinates have the same gender) can strengthen the effect of emotional recognition ability on perceived ambidextrous leadership, thereby enhancing promotive and prohibitive voice.

Method

Sample and Procedure

We collected our data from employees and their direct supervisors in four Chinese traditional manufacturing enterprises. At present, Chinese manufacturing enterprises are undergoing transformation and upgrading, therefore employees in manufacturing enterprises have strong voice desires.²³ To guarantee the quality of data and reduce common method bias, we handed out the questionnaires in three waves at monthly intervals. One day in advance, one of the authors contacted the general managers or the president of the boards of targeted enterprises, and communicated with department managers about the participant lists and survey sites. The questionnaires were coded before being distributed to facilitate subsequent matching, and the coding principles are confidential for outsider. All surveys were conducted in the form of paper questionnaires, and merely those collected on the spot were adopted in our analyses. In Phase 1, employees were asked to report their demographic characteristics such as gender, age and education, and emotional recognition ability. We collected 302 valid responses after removing 48 responses. In Phase 2, supervisors report their demographic characteristics such as gender, age and education, and employee voice. We obtained 50 direct supervisors of 281 employee (one supervisor with multiple subordinates). One month later, in Phase 3, employees were asked to assess perceived ambidextrous leadership. We finally got 254 employee responses due to the absence of 27 participants. Removing extremely incomplete and invalid responses, we finally gained a sample of 43 supervisors and 182 employees, with a response rate of 71.7%. As for the supervisor sample, 19.0% were female; Supervisor’s age was distributed as follows: 25~35 years (2.4%), 35~45 years (42.9%), 45~60 years (54.7%); For education, 61.9% of the supervisors had a bachelor degree, 21.4% had a master degree, 14.3% had a doctoral degree, and 2.4% had a college degree. Of 182 employees, 60.4% were male. Respondents’ age was distributed as follows: 25 years or below (1.1%), 25~35 years (20.9%), 35~45 years (50.5%), 45~60 years (27.5%). Most of the employees had a bachelor degree (63.7%), 4.9% had high school degree, 15.9% graduated from community college, 15.5% had a master or high degree.

Measures

Participants were asked to report the items on a 5-point scale (1 = totally disagree to 5 = totally agree). We used a 4-item scale to check emotional recognition ability,¹³ which included items such as “I often capture my leader’s emotions from his/her behaviors”. The Cronbach alpha coefficient for emotional recognition ability was 0.885. According to Rosing, Frese, and Bausch,³¹ we measured perceived ambidextrous leadership using 12 items scale, with 7 items for transformational leadership and 5 items for transactional leadership. We adopted the interactive term of the means of each measurement.³² The Cronbach alpha coefficient for perceived ambidextrous leadership was 0.868. Employee voice was assessed using a 10-item scale from Liang, Farh, and Farh,²⁴ with 5 items for promotive voice and the other 5 items for prohibitive voice. Sample items were “my subordinates tell the truth on issues that may cost the company even though other people disagree” and “my subordinates positively make suggestions”. The Cronbach alpha coefficient promotive voice and prohibitive voice were 0.874 and 0.926, respectively. Referring to Li and Luo,⁴⁴ we set leader–subordinate gender similarity as a dummy variable where 0 = the same gender and 1 = different genders. Besides, we controlled age, education, leader-subordinate age similarity and leader-subordinate education similarity following prior studies.⁴⁴

Results

Descriptive Statistics

Table 1 displays the means, standard deviations, Pearson correlations for the variables used in our study. Table 1 presents that subordinate’s emotional recognition ability was positively associated with both promotive voice (r = 0.442, p<0.01) and prohibitive voice (r=0.430, p<0.01), and positively related to perceived ambidextrous leadership (r=0.499, p<0.01). Perceived ambidextrous leadership was positively related to both promotive voice (r=0.376, p<0.01) and prohibitive voice (r=0.288, p<0.01). Besides, gender similarity has no significant relations with other variable. The results are in accordance with our theoretical predictions.

Table 1 Means, Standard Deviations and Correlations of Variables

Variables	M	SD	1	2	3	4	5
1. SERA	4.091	0.728	0.863
2. PAL	4.089	0.621	0.499**	0.780
3. Promotive voice	3.747	0.799	0.442**	0.376**	0.818
4. Prohibitive voice	4.028	0.759	0.430**	0.288**	0.669**	0.879
5. Gender similarity	0.401	0.491	−0.017	0.068	−0.087	−0.099	NA

Notes: N = 43, n = 182; the value on the diagonal means the square root of AVE; **p < 0.01 (double tail).

Abbreviations: SERA, subordinate emotional recognition ability; PAL, perceived ambidextrous leadership.

Confirmatory Factor Analysis (CFA)

To check the construct validity of the four variables (subordinate’s emotional recognition ability, perceived ambidextrous leadership, promotive voice, and prohibitive voice) in our model, we conducted confirmatory factor analysis with Lisrel 8.7. The results are presented in Table 2. As shown in Table 2, the four-factor model fit the data well with χ²/df = 2.347, RMSEA = 0.086, SRMR = 0.071, CFI = 0.95, NFI = 0.92. Meanwhile, we contrasted the four-factor model (measurement model) against three-factor, two-factor and one single-factor models, respectively. The results in Table 2 show that the four-factor model offered better model fit indexes than any other models. Thus, it supported the discriminant validity of the main measures in our study. In addition, we also adopted factor loading, AVE, CR to measure the validity of the key constructs. The results presented that the interval of the factor loadings of emotional recognition ability, perceived ambidextrous leadership, promotive voice, and prohibitive voice were [0.805, 0.882], [0.563, 0.857], [0.781, 0.860] and [0.838, 0.911], respectively, exceeding 0.5; their CR were 0.921, 0.948, 0.910 and 0.944, exceeding 0.7. Besides, all the square root of the AVE of the four variables were over their correlations, supporting convergent validity and discriminant validity. To examine whether there is no response deviation within the questionnaire, we conducted a t-test on both valid and invalid questionnaires. Results showed that there is no significant difference in the aspect of age and education, which indicates that no response deviation exists.

Table 2 Results of Confirmatory Factor Analyses

Model	χ^2	df	χ^2/df	RMSEA	SRMR	CFI	NFI
Four-factor model	687.56	293	2.347	0.086	0.071	0.95	0.92
Three-factor model	934.83	296	3.158	0.109	0.081	0.94	0.90
Two-factor model	1284.30	298	4.310	0.135	0.100	0.90	0.87
Single-factor model	3306.47	299	11.058	0.236	0.150	0.81	0.79

Notes: Four-factor model: subordinate emotional recognition ability, perceived ambidextrous leadership, promotive voice and prohibitive voice; Three-factor model: subordinate emotional recognition ability, perceived ambidextrous leadership, promotive voice + prohibitive voice; Two-factor model: subordinate emotional recognition ability + perceived ambidextrous leadership, promotive voice + prohibitive voice; Single-factor model: subordinate emotional recognition ability + perceived ambidextrous leadership + promotive voice + prohibitive voice.

Hypotheses Testing

We used hierarchical regression analysis method to test the main, mediating and moderating effects in this study. Table 3 outlines the results. After controlling age, education, leader-subordinate age similarity and leader-subordinate education similarity, the results in M1 presents that subordinate’s emotional recognition ability positively and significantly affects promotive voice (β=0.439, p<0.01). When introducing perceived ambidextrous leadership into the model (M2), we found that perceived ambidextrous leadership has a positive and significant effect on promotive voice (β=0.225, p<0.01) while the impact of subordinate’s emotional recognition ability on promotive voice is weaker (β=0.329, p<0.01). This means that perceived ambidextrous leadership partly mediates the effect of subordinate’s emotional recognition ability on promotive voice. In a similar vein, the results in M3 and M4 indicate that subordinate’s emotional recognition ability positively and significantly influences prohibitive voice (β=0.428, p<0.01) while the influence of perceived ambidextrous leadership on prohibitive voice is not significant (β=0.095, n.s.). This indicates that perceived ambidextrous leadership has no mediating effect in the relationship between subordinate’s emotional recognition ability and prohibitive voice. Therefore, H1 is supported and H2 is partly supported.

Table 3 The Results of Hierarchical Regression Analysis

Variables	Promotive Voice		Prohibitive Voice		Promotive Voice				Prohibitive Voice
	——		——		Gender Similarity		Gender Dissimilarity		Gender Similarity		Gender Dissimilarity
	Model1	Model2	Model3	Model4	Model5	Model6	Model7	Model8	Model9	Model10	Model11	Model12
Employee age	−0.038	0.036	−0.141	−0.111	−0.104	0.023	0.047	0.056	−0.213	−0.115	−0.095	−0.113
Employee education	−0.109	−0.093	0.009	0.016	−0.146	−0.115	−0.097	−0.098	−0.015	0.011	−0.042	−0.040
Age similarity	−0.143	−0.108	−0.268**	−0.254*	−0.264*	−0.171	0.042	0.038	−0.336*	−0.267*	−0.194	−0.187
Education similarity	−0.041	−0.040	0.047	0.049	−0.069	−0.083	−0.002	0.003	−0.010	−0.016	0.142	0.132
SERA	0.439**	0.329**	0.428**	0.381**	0.366**	0.203*	0.536**	0.512**	0.344**	0.209*	0.541**	0.586**
PAL		0.225**		0.095		0.337**		0.049		0.271*		−0.093
R^2	0.227	0.262	0.222	0.228	0.223	0.299	0.276	0.277	0.179	0.228	0.307	0.313
F	10.261**	10.282**	9.917**	8.514**	5.863**	7.196**	5.101**	4.221**	4.399**	4.915**	5.935**	5.004**

Notes: *p < 0.05; **p < 0.01.

Abbreviations: SERA, subordinate emotional recognition ability; PAL, perceived ambidextrous leadership.

Since leader-subordinate gender similarity is a dummy variable, to check its moderating role, we tested the mediating effect of perceived ambidextrous leadership under different gender similarity situations referring to Li and Luo.⁴⁴ The results are displayed in M5-M12. According to M5 and M6, when leaders and subordinates have the same gender, subordinate’s emotional recognition ability positively and significantly affects promotive voice (β=0.366, p<0.01). When perceived ambidextrous leadership was introduced into the model, perceived ambidextrous leadership has a positive and significant effect on promotive voice (β=0.337, p<0.01) while the impact of subordinate’s emotional recognition ability on promotive voice is weaker (β=0.203, p<0.05). This means that perceived ambidextrous leadership partly mediates the effect of subordinate’s emotional recognition ability on promotive voice. Similarly, the results in M9 and M10 indicate that subordinate’s emotional recognition ability and positively and significantly influences prohibitive voice (β=0.344, p<0.01), and perceived ambidextrous leadership has a positive and significant impact on prohibitive voice (β=0.271, p<0.05) when perceived ambidextrous leadership entering the model. Meanwhile, the influence of subordinate’s emotional recognition ability on prohibitive voice is weaker (β=0.209, p<0.05). This indicates that perceived ambidextrous leadership has no mediating effect in the relationship between subordinate’s emotional recognition ability and prohibitive voice.

M7 and M8 present the results when leaders and subordinates have different genders. The results show that subordinate’s emotional recognition ability positively and significantly affects promotive voice (β=0.536, p<0.01). However, when perceived ambidextrous leadership was introduced into the model, the influence of perceived ambidextrous leadership on promoting voice is not significant (β=0.049, n.s.). This means that perceived ambidextrous leadership has no mediating effect in the relationship between subordinate’s emotional recognition ability and promotive voice. In a similar way, according to M11 and M12, subordinate’s emotional recognition ability and positively and significantly influences prohibitive voice (β=0.541, p<0.01). Yet, when perceived ambidextrous leadership enters the model, it has no significant effect on prohibitive voice (β=−0.093, n.s.). This indicates that perceived ambidextrous leadership has no mediating effect in the relationship between subordinate’s emotional recognition ability and prohibitive voice. Together, perceived ambidextrous leadership has a stronger mediating effect when leaders and subordinates have the same gender rather than different genders, supporting H3.

To check the robustness of our conclusions, we further used bootstrapping method to confirm the mediating and moderating effects in the research model. The results are outlined in Table 4. The results show that the direct effect of emotional recognition ability on promotive voice is 0.360, with 95% CI of [0.197, 0.524], excluding 0; the indirect effect via perceived ambidextrous leadership is 0.121, with 95% CI of [0.035, 0.239], excluding 0; and the total effect is 0.481, with 95% CI of [0.337, 0.625], excluding 0. This means that perceived ambidextrous leadership partly mediates the relationship between subordinate’s emotional recognition ability and promotive voice. Similarly, we find that the direct effect of emotional recognition ability on prohibitive voice is 0.400, with 95% CI of [0.236, 0.557], excluding 0; the indirect effect via perceived ambidextrous leadership is 0.049, with 95% CI of [−0.037, 0.155], including 0. This means that perceived ambidextrous leadership has no mediating effect in the relationship between subordinate’s emotional recognition ability and prohibitive voice, supporting previous conclusions. In addition, we also used Sobel test to reinforce the robustness of the mediating effect. The results showed that perceived ambidextrous leadership mediates the relationship between emotional recognition ability and promotive voice (2.676, p<0.01), and has no mediating effect between emotional recognition ability and prohibitive voice (1.156, n.s.). H2 is further verified.

Table 4 The Robustness Test on Perceived Ambidextrous Leadership as a Mediator

IV	Sobel Test	Effect	Coeff	se	95% CI
					LLCI	ULCI
Promotive voice	2.676**	Indirect effect	0.121	0.052	0.035	0.239
		Direct effect	0.360	0.083	0.197	0.524
		Total effect	0.481	0.073	0.337	0.625
Prohibitive voice	1.156	Indirect effect	0.049	0.049	−0.037	0.155
		Direct effect	0.400	0.081	0.236	0.557
		Total effect	0.446	0.070	0.308	0.583

Note: **p < 0.01.

Table 5 depicts the results of robustness test on the moderating effect of leader-subordinate gender similarity. Under the situation of leader-subordinate gender congruence, the results indicate that the direct effect of emotional recognition ability on promotive voice is 0.259, with 95% CI of [0.032, 0.487], excluding 0; the indirect effect via perceived ambidextrous leadership is 0.196, with 95% CI of [0.081, 0.358], excluding 0. This means that perceived ambidextrous leadership partly mediates the relationship between subordinate’s emotional recognition ability and promotive voice. Under the situation of leader-subordinate gender incongruence, the results indicate that the direct effect of emotional recognition ability on promotive voice is 0.499, with 95% CI of [0.261, 0.736], excluding 0; the indirect effect via perceived ambidextrous leadership is 0.016, with 95% CI of [−0.088, 0.184], including 0. This means that perceived ambidextrous leadership has no significant mediating effect in the relationship between emotional recognition ability and promotive voice.

Table 5 The Robustness Test on Leader-Subordinate Gender Similarity as a Moderator

IV	Condition	Sobel Test	Effects	Coeff	se	95% CI
						LLCI	ULCI
Promotive voice	Gender similarity	2.933**	Indirect effect	0.196	0.069	0.081	0.358
			Direct effect	0.259	0.115	0.032	0.487
	Gender dissimilarity	0.264	Indirect effect	0.016	0.069	−0.088	0.184
			Direct effect	0.499	0.119	0.261	0.736
Prohibitive voice	Gender similarity	2.551*	Indirect effect	0.156	0.085	0.018	0.359
			Direct effect	0.204	0.107	−0.008	0.415
	Gender dissimilarity	−1.050	Indirect effect	−0.068	0.061	−0.205	0.042
			Direct effect	0.609	0.122	0.367	0.851

Notes: *p < 0.05; **p < 0.01.

Similarly, under the situation of leader-subordinate gender similarity, we find that the direct effect of emotional recognition ability on prohibitive voice is 0.20, with 95% CI of [−0.008, 0.415], including 0; the indirect effect via perceived ambidextrous leadership is 0.156, with 95% CI of [0.018, 0.359], excluding 0. This means that perceived ambidextrous leadership partly mediates the relationship between emotional recognition ability and prohibitive voice. Under the situation of leader-subordinate gender incongruence, the direct effect of emotional recognition ability on prohibitive voice is 0.609, with 95% CI of [0.367, 0.851], excluding 0; the indirect effect via perceived ambidextrous leadership is −0.068, with 95% CI of [−0.205, 0.042], including 0. This means that perceived ambidextrous leadership has no mediating effect on the relationship between emotional recognition ability and prohibitive voice. H3 is further confirmed.

To test the mediation and moderation effect of emotional recognition ability on voice, we construct the structural equation model after multiple revisions according to the T value. As shown in Figure 2, various indexes provide good fits (χ²=786.22, df = 296, RMSEA = 0.095, CFI = 0.94, NFI = 0.91). Results show that perceived ambidextrous leadership has a mediation effect on the relation between emotional recognition ability and voice, while gender similarity significantly and negatively moderates the relationship between emotional recognition ability and perceived ambidextrous leadership (β=−0.44, p<0.01), that is, leader-subordinate gender similarity is more inclined to reinforce the impact of emotional recognition ability on perceived ambidextrous leadership.

Figure 2 Shows the results of the structure equation model. Various indexes provide good fits (χ²=786.22, df=296, RMSEA=0.095, CFI=0.94, NFI=0.91, **p < 0.01). The results indicate that subordinate’ s emotional recognition ability have a positive impact on perceived ambidextrous leadership (β=0.62, p < 0.01), and thus promote promotive voice (β=0.38, p < 0.01) and prohibitive voice (β=0.45, p < 0.01). In addition, gender similarity negatively affects the first stage mediation effect (β=−0.44, p < 0.01). That is, when an leader and his/her subordinates have the same gender, the mediation effect of perceived ambidextrous leadership will be stronger.

Discussion

Our study has three main theoretical contributions. First, this study integrates subordinate’s emotional recognition ability and employee voice. Previous research tend to demystify the triggering mechanism of employee voice based on employees’ constructive and instrumental intentions,^5,6 overlooking the effect of emotion on voice. A survey from Harvard Business Review indicates that “reading facial expressions” is a primary determinant for employee voice.¹⁰ Emotion recognition ability reflects subordinates’ recognition and assessment on leaders’ emotions,¹³ and aroused the concerns of scholars in voice field. The findings of this study show that subordinates with high emotion recognition ability are more likely to conduct voice behaviors. One logical explanation may be that these subordinates are able to make accurate judgement on their leaders’ expectation of voice.²² This study supports emotion as social information theory, that is, leader emotion is a critical interpersonal interaction information, and it can motivate employees’ personal effect emotionally, and thus drive employees to make behavioral decision based on information feedback.²⁵ Additionally, we introduce emotion as social information theory into voice research, which not only advances our understanding of voice, but also provides a new perspective for voice research.

Second, this study sheds light on the mediating role of perceived ambidextrous leadership in the relationship between subordinate’s emotional recognition ability and employee voice. Based on emotion as social information theory, the process of subordinates handling their leaders’ emotional information can affect their behavioral decision-making.¹⁴ Study from Harvard Business Review shows that subordinates with high emotion recognition ability are more inclined to evaluate the risk of voice through perceived leadership behaviors.¹⁰ Of numerous leadership behaviors, ambidextrous leadership behavior is considered to be with time and space flexibility. Compared with other leadership behaviors, ambidextrous leadership behavior is consistent with Chinese traditional thinking-“Yin-Yang Balance”.²³ Hence, this study introduces perceived ambidextrous leadership into the research model regarding subordinate’s emotional recognition ability and employee voice. This study finds that perceived ambidextrous leadership plays a significant mediating role between subordinate’s emotional recognition ability and promotive voice, a no mediating role is found between subordinate’s emotional recognition ability and prohibitive voice. One possible reason is that prohibitive voice may do harm to employees. Because some studies have confirmed that leaders are more likely to show negative emotions when employees point out problems in the current organization or leader.^2,49 The conclusion shed light on the application situations of emotion as social information theory. That is, the transmission of emotional information is more likely to promote employees to put forward suggestions for future development, but cannot encourage them to come up with suggestions on improvement toward existing problems. This, on the one hand, sheds light on the specific application situations of emotion as social information theory in the field of organizational behavior, and, on the other hand, expand the voice research based on emotion, and clarifies the difference of the mechanism of different voice, which provides important reference for the further research on voice.

Finally, we investigate the role of leader-subordinate gender similarity between subordinate’s emotional recognition ability and employee voice, and find that leader-subordinate gender congruence can strengthen employee voice. Although prior studies have emphasized the importance of leader-subordinate gender similarity for employee attitudes and behaviors,^19,20 it still confuses us that what role of leader-subordinate gender similarity plays in the process from “reading facial expressions” to voice. Drawing on similarity attraction theory, this study finds that leader-subordinate gender congruence can strengthen employee voice. This finding supports Riordan’s⁴³ conclusions that gender congruence can increase mutual attraction and develop close working relationship easily. This conclusion verifies the similarity attraction theory, that is, leader-subordinate gender similarity can strengthen their attraction,^18,43 and thus enhance their willingness to interact with each other. This not only provides explanation for voice in terms of gender, but also offers new directions for the antecedents of perceived ambidextrous leadership from the demographic perspective.

Our study also has some practical implications. First, when leaders encourage employees to voice, they should take emotional recognition ability into full consideration, and give priority to the employees with high emotional recognition ability, because those employees are more likely to provide high-quality suggestions. Second, in order to promote employee voice, leaders ought to adopt different leadership styles over time and functions rather than adopt one-size-fits-all approach. For example, transformational leadership for R&D personnel, and transactional leadership for production personnel; transformational leadership in the early stage of a project while transactional leadership in the late stage. The switching of leadership styles can help employees to perceive their leaders’ respect to them, and regard themselves as “insider”. Third, except for such implicit factors as emotional recognition ability, some explicit factors such as gender are as well important for voice and perceiving leader behaviors. Therefore, leaders can pay more attention to those employees having the same gender with them when seeking advice, because those employees are more likely to provide high-quality suggestions.

This study is subject to some limitations. First, although we collected data three-wave collection data from employee and supervisor through three-wave survey, it still belongs to cross-section data. Our data fail to unveil the dynamic evolution of our research model, which is a limitation of this study. We encourage future research to examine the dynamic causal relation among the main variables by collecting data through multiple time points and experience sampling method. Second, demographic indicators provide new perspectives for research on organizational behaviors. This study examines the boundary condition of gender similarity, leaving education similarity and age similarity undeveloped. Therefore, future research should focus on aforementioned similarity indicators. Third, the third limitation of our study is that we merely emphasize the importance of subordinate’s emotional recognition ability. Although previous research has suggested that subordinate’s emotion recognition ability is more likely to enhance employee voice in contrast to leader’s emotion recognition ability.⁴⁴ However, leader’s emotion recognition ability probably significantly affect voice acceptance because leaders can well judge employees’ intentions of voice if they are able to discern their subordinate’s emotional precisely. Hence, leader’s emotion recognition ability would be a potential direction for voice research.

Conclusions

Drawing on social information theory and similarity attraction theory, we proposed a research model clarifying the mechanism and boundary of subordinate’s emotional recognition ability influencing employee voice. We then tested our research model with data from 182 employee and 43 supervisors using hierarchical regression analysis method and bootstrapping method. The results showed that subordinate’s emotional recognition ability has a significant positive effect on both promotive and prohibitive voice; perceived ambidextrous leadership plays a significant mediating role between subordinate’s emotional recognition ability and promotive voice, a no mediating role is found between subordinate’s emotional recognition ability and prohibitive voice. Additionally, in contrast to leader-subordinate gender dissimilarity, leader-subordinate gender similarity is more likely to strengthen the influence of subordinate’s emotional recognition ability on perceived ambidextrous leadership, thereby motivating voice.

Ethical Statement

The present study adhered to the Declaration of Helsinki and was approved by the Ethics Committee on Human Experimentation of Shandong University of Finance and Economics. All participants received the first-wave questionnaire in an envelope with an introduction of the study purposes as well as a written informed consent form. We explained that this study welcomed voluntary participation, and complied with the principle of confidentiality, and is only for research purposes. Before response to the first-wave questionnaire, all participants provided informed consent, claimed their understandings of the study purposes and they would like to participate in the study voluntarily.

Acknowledgments

The present study adhered to the Declaration of Helsinki and was approved by the Ethics Committee on Human Experimentation of Shandong University of Finance and Economics. All participants received the first-wave questionnaire in an envelope with an introduction of the study purposes as well as a written informed consent form. We explained that this study welcomed voluntary participation, and complied with the principle of confidentiality, and is only for research purposes. Before response to the first-wave questionnaire, all participants provided informed consent, claimed their understandings of the study purposes and they would like to participate in the study voluntarily. All authors contributed to the work equally.

Disclosure

The authors report no conflicts of interests in this work.

Additional information

Funding

This research was funded by Natural Science Foundation of Shandong Province (ZR2020MG026).