Leading for Innovation
The role of leadership in enhancing innovation is critical and important (Tierney, 2008). However, this has been found to be highly heterogeneous (Hammond, Neff, Farr, Schwall, & Zhao, 2011; Rosing et al., 2011). For instance, for follower creativity and innovation, Gong, Huang, and Farh (2009) and Shin and Zhou (2003) found a significant positive relationship with transformational leadership, whereas Jaussi and Dionne (2003) found a negative correlation which was not significant, whereas for team creative performance, it was negative and statistically significant. This heterogeneity extends to other leadership theories as well, Williams (2004) found an insignificant relationship between leader preference for initiating structure and follower creative performance, though Osborn and Marion (2009) found a positive relationship between initiating structure behaviours of leader on innovation performance of alliances. They also found a significant positive relationship between leader consideration behaviours and innovation performance though these two behaviours have been linked to differing outcomes by Judge, Piccolo, and Ilies (2004). Supervisor support was not found to be linked to creativity by Tierney and Farmer (2002) and Oldham and Cummings (1996) but was found to be linked to creativity and innovation by Ohly, Sonnentag, and Pluntke (2006). For transformational leadership, in particular, it is found to be related to explorative behaviours by followers but not to exploitative behaviours (Mainemelis et al., 2015) . Kahai, Sosik, and Avolio (2003) go so far as to suggest that transactional leadership might have certain advantages in the leadership of creative efforts compared to transformational leadership. For team innovation, EisenbeiE and Boerner
(2010) found a curvilinear relationship between transformational leadership and team innovation.
This heterogeneity in the relationship between leadership and innovation has resulted in a search for moderators of the relationship. Herrmann and Felfe (2013) have explored task novelty and personal initiative, Keller (1989) consider employees that face an uncertain environment and are professional to place a higher value on leader-initiating structure behaviours, Gilmore, Hu, Wei, Tetrick, and Zaccaro (2013) have shown that positive trait affectivity substitutes for the influence of transformational leadership when considering creative performance, Pieterse, van Knippenberg, Schippers, and Stam (2010) found that transformational leadership was effective in encouraging innovative behaviour only when followers were psychologically empowered, and finally Rosing et al.
(2011) suggest that different leader behaviours moderate each other.
Though Woodman et al. (1993) suggest that leadership (a contextual variable) would interact with personal characteristics to predict creative behaviour, there is no clear guidance on what interactions can be expected. Therefore, the interactionist perspective, though supported by current findings, needs to be enriched to create higher predictive value. Specific to the relationship of leader behaviours and innovation, Anderson et al. (2014) suggest that the heterogeneity could be due to different leadership styles and behaviours being important at different stages of the innovation cycle. In another recent review, Kesting, Ulhoi, Song, and Niu (2016) propose a similar argument of different stages of innovation placing differing demands on leadership. Amabile (1996b) suggest that certain tasks in the innovation cycle are heuristic whereas others are algorithmic. Heuristic tasks are those that cannot be solved by following a recipe, whereas as algorithmic tasks have a single process of solution. The differing nature of these tasks, suggests that different leadership behaviours may have differing effects during the innovation process and behaviours that facilitate one type of task might inhibit the other type, thus resulting in the overall effect being heterogeneous. As we argued in the previous section, the iterative and non-linear nature of innovation precludes us from insisting on a specific sequence of stages and conceptualizing follower behaviours at a finer level of granularity into explorative and exploitative behaviours would be a more beneficial approach. This is the approach taken by Rosing et al. (2011), and the approach that we build upon.