Transformational, transactional or servant: pick by the challenge

In Judge and Piccolo's (2004) meta-analysis, transformational leadership has a pooled validity of 0.44 against the outcome criteria and contingent reward has 0.39. The two also correlate 0.80 with each other (Judge and Piccolo, 2004). Suppose you are a CEO who has just taken over a division that has lost money for three quarters running. Those two numbers say both approaches go with better outcomes. They do not say which one your division needs first.

How do the three approaches differ?

Transformational leadership works by inspiring, transactional leadership by exchanging rewards for results, and servant leadership by putting followers' needs first. The first two are set side by side in the library's comparison framework:

I read "transactional" here as contingent reward, the part of transactional leadership that has its own figures in the study I read; management by exception was inconsistently related to the criteria, so I leave it out (Judge and Piccolo, 2004). Lee and colleagues describe servant leaders as leaders who focus on their followers' needs (Lee et al., 2020). The publisher of Greenleaf's (1977) book describes the idea as replacing traditional autocratic leadership with an approach rooted in service to others (Greenleaf, 1977).

Situational leadership is a different kind of thing. It adjusts how much to direct and how much to support according to each person's competence and readiness, so I treat it as a diagnosis of the follower. The directive and participative behaviours measured by Post and colleagues are behaviours in an experiment, not synonyms for transactional or servant leadership (Post et al., 2022).

What do the pooled studies say about transformational versus transactional?

Transformational leadership comes out slightly ahead, and the two overlap heavily. Judge and Piccolo pooled 626 correlations from 87 sources (Judge and Piccolo, 2004):

The abstract adds that the transformational result held across longitudinal and multisource designs; that on several criteria contingent reward was the stronger of the two; and that once the other dimensions were controlled both generally still predicted the criteria, except that transformational leadership failed to predict the leader's own job performance (Judge and Piccolo, 2004).

What do these numbers tell a manager, and what do they not?

They give a general direction. They do not give causation, or how far it applies to your unit. There are three limits.

• Overlap. If you treat 0.80 as an ordinary correlation, squaring it gives about 64 per cent shared variance; 0.44 and 0.39 give about 19 and 15 per cent (my arithmetic). With these data a manager cannot easily separate "inspiring" from "clear rewards" (Judge and Piccolo, 2004).
• Who rated. The abstract does not say how many of the correlations come from one person rating both the leader's style and the outcome (Judge and Piccolo, 2004). The risk is real: in the experiment by Post and colleagues, independent observers saw leaders in the participative condition encourage participation more (d = 0.64), yet team members rated participative behaviour almost the same across the two conditions (3.80 and 3.77 on a 5-point scale); the authors suspect response bias (Post et al., 2022).
• Scope. What I read names no sector, firm size or country. For the cultural side, see Cultural intelligence for leaders.

What does servant leadership add beyond transformational?

In two meta-analyses by different teams it adds more than the other "ethical" variants, but I read abstracts only. Hoch and colleagues compared authentic, ethical and servant leadership with transformational leadership on nine outcomes: authentic and ethical correlate highly with transformational and add very little variance, while servant leadership looks more promising as a stand-alone approach (Hoch et al., 2018). Lee and colleagues pooled 130 independent studies: servant leadership has incremental predictive validity over transformational, authentic and ethical leadership, and its link with outcomes is partly explained by trust in the leader, procedural justice and leader–member exchange (Lee et al., 2020).

"Incremental" is a statistical idea: after transformational scores are accounted for, servant scores still explain something extra. It does not mean servant leadership beats transformational leadership on every challenge, and neither abstract gives an effect size, so I cannot say whether the gain is 1 per cent or 10.

Should a crisis team be directed or invited to participate?

It depends on whether the crisis is familiar or unfamiliar to the team. Post and colleagues (2022) ran a randomised experiment with 72 three-person teams, 216 undergraduates from a Belgian military academy and a Dutch university. Each team handled four simulated fire-and-chemical scenarios, two familiar and two unfamiliar, with at most 10 minutes per scenario. An unfamiliar scenario had an unusually large quantity of chemicals or lacked a resource the team had learned to count on (Post et al., 2022).

The authors conclude that neither style is better a priori. They recommend training leaders to switch between the two, and warn that people tend to believe a situation is more familiar than it is, then direct by rote until it is too late. In unfamiliar scenarios 30.6 per cent of teams used the full 10 minutes, against 4.2 per cent in familiar ones (Post et al., 2022).

The authors name their own limits: students, mostly men (73 per cent), simulated situations, and familiar versus unfamiliar as only two levels (Post et al., 2022). Each person was trained in a single role, so these were not seasoned specialists.

How far can Hersey–Blanchard situational leadership be trusted?

As a set of questions about the follower, reasonably; as a validated formula, not. The library framework sets out four styles by amount of direction and support.

Thompson and Vecchio (2009) tested three versions of the theory on a survey of 357 bank employees and 80 supervisors in 10 Norwegian financial institutions. According to the abstract, the revised 2007 version predicted worse than the original; the third version, which predicts that outcomes depend on the autonomy a leader gives combined with the employee's experience, offered promise, although it too added little to the variance explained (Thompson and Vecchio, 2009). The abstract of Graeff (1997) says the model's theoretical standing and practical usefulness are challenged by logical inconsistencies, conceptual ambiguity and confusion among its versions (Graeff, 1997).

So I keep only the most defensible part: ask about the follower before choosing how to lead. How many times has this person done the task, how confident are they, how much autonomy do they need? Those questions are mine, and no study I read validates them.

Where should you start for each kind of challenge? A decision table

The "Starting point" column is my reading; the other two follow the sources cited. It suits you if you use it to pick a starting point and review it after a few weeks. It does not suit you if you need one fixed answer for the whole year.

Going deeper: what did I read, what did I not, and where is the criticism?

I read one article in full: Post and colleagues (2022), the publisher's version made public in the Maastricht University repository. For Judge and Piccolo, Hoch and Lee, I read only the public abstracts. So this post has no effect sizes from Hoch or Lee, does not know which criteria favour contingent reward over transformational, and does not know how many correlations rest on same-source ratings. For those figures, read the full texts through the DOIs in the reference list.

The publisher's page for Thompson and Vecchio would not open; I read the full abstract on Mendeley's catalogue page. For Graeff I read the abstract on Academia.edu. For Greenleaf I read only the publisher's page for the 50th-anniversary edition. The descriptions of transformational and transactional leadership come from the library framework.

On criticism, Hoch and colleagues note that researchers have pointed to conceptual overlap between transformational leadership and the newer forms (Hoch et al., 2018); in my reading the 0.80 between transformational and contingent reward is another kind of overlap (Judge and Piccolo, 2004). The comparison of leadership theories reminds us that the streams of theory are not successive.