Abstract:     Under the volatile, uncertainty, complex and ambiguous (VUCA) environment, organizations are facing paradoxical demands to achieve fast and steady growth. To achieve the above goals, organizational leaders play crucial roles in setting and leading the paradoxical goals. Earlier studies have mainly focused on one particular type of leadership in setting organizational goals (e.g., empowering leadership; authoritarian leadership) that may restrict the paradoxical organizational goals as they put more focus on one single aspect. Hence, leadership that can satisfy different even paradoxical organizational goals gained more research and practical attention. 
    To achieve the paradoxical demands, two newly developed leadership behaviors have gained much attention: ambidextrous leadership and paradoxical leadership. Ambidextrous leadership is defined as leaders who adjust their leadership behaviors to adopt two contradictory leadership behaviors based on the situations. Paradoxical leadership, however, is mostly studied in the Chinese context, which is rooted in the Yin-Yang philosophy. Paradoxical leadership is mostly shown as leaders taking two contradictory leadership behaviors to satisfy the paradoxical work goals. Although the root of ambidextrous and paradoxical leadership is different, both revealed that leaders should adopt their leadership behaviors even two contradictory leadership behaviors to satisfy the organizational competing goals. 
   Studies on ambidextrous and paradoxical leadership have consistently shown that both leadership behaviors can have positive relationships with innovation and task performance. However, there remains three main questions to be answered: first, research have shown different findings about the effect of ambidextrous and paradoxical leadership on task performance. For instance, studies on the relationship between ambidextrous leadership and task performance has ranged from negative to positive, and the relationship between paradoxical leadership and task performance has ranged from nonsignificant to significant. Second, since both ambidextrous and paradoxical leadership share similarities in their definition, understanding about which leadership style is more powerful in predicting employee performance is unknown. Third, it is also unknown about the role of leaders′ hierarchical level in affecting the effectiveness of ambidextrous and paradoxical leadership. For instance, prior studies have suggested that middle level managers′ paradoxical leadership are more likely to achieve effectiveness; and the effectiveness of ambidextrous leadership will also vary across different levels. Based on the above questions, we conducted our study with meta-analysis approach. 
   Different with the single-sample quantitative study, meta-analysis approach has several advantages. First, meta-analysis can resolve research dispute by integrating different research findings. With a large sample size, it can achieve more accurate estimation. Second, meta-analysis can also achieve sub-group analysis to show the different impact of ambidextrous and paradoxical leadership. Third, meta-analysis can also identify the conditional factors. As such, it can answer our third question about how leaders′ hierarchical level may affect their leadership effectiveness. 
    Before conducting meta-analysis, we firstly searched the database (i.e., CNKI, Web of Science, and ProQuest) on ambidextrous leadership and paradoxical leadership. We restricted the timeline to May, 2021. This led to a total of 249 papers. Second, we screened the literature and only included studies that meet our standards. For instance, we excluded conference paper and dissertation and excluded papers that did not report the correlations between leadership and performance outcomes. This led to 43 papers that satisfying all our requirements. Based on the 43 papers, we conducted our meta-analysis. Before coding the data, we trained all the coders about how to code the data to reduce misunderstandings and bias. According to the 43 papers, we got 50 samples with a total sample size of 13232. Applying R software, we found that: first, both ambidextrous leadership and paradoxical leadership had positive relationships with innovation and task performance. Second, subgroup comparison analysis results revealed the relationship between ambidextrous leadership and innovation performance was not significantly different from that between paradoxical leadership on innovation performance; and the relationship between paradoxical leadership and task performance was not significantly different from that between ambidextrous leadership and task performance. Third, subgroup analysis also showed that there was no moderating effect of leaders′ hierarchical level on the above relationships. 
    Our study has the following theoretical implications: First, through meta-analysis, it revealed the "real" effect of ambidextrous and paradoxical leadership on innovation and task performance, which could resolve the puzzle about the different impact of these two leaderships on the innovation and task performance outcomes. Second, through the comparative analysis between ambidextrous and paradoxical leadership and their outcomes, it could show the unique role of paradoxical leadership, which can offer further evidence on the studies about paradoxical leadership. Third, by focusing on the conditional factors in affecting the effectiveness of ambidextrous and paradoxical leadership, it offered insights for future studies to explore when and why these two leadership behaviors can have different impacts. Overall, our study contributed to the ambidextrous and paradoxical leadership studies by offering an integrated picture, which can also offer insights for future studies on these two leadership behaviors. 

Key words:  ambidextrous leadership, paradoxical leadership, innovation performance, task performance, meta-analysis