Transformational leadership has more research behind it than any other leadership construct, and the average effect is large. It also has four named problems, published in the field's most selective review journal by two senior scholars, and a dominant measurement instrument whose factor structure keeps failing to reproduce the theory it is supposed to measure. If you have bought leadership training in the last thirty years, you have almost certainly bought some version of this. You were told the first half.
Burns, 1978
Bass, 1985
Judge & Piccolo, 2004
Transactional leadership is an exchange: do the work, get the reward. Transformational leadership is supposed to be something else — a leader who raises what followers want in the first place, so that effort stops being a trade and becomes something people give.
In the standard formulation it has four components: idealized influence (the leader as someone worth emulating), inspirational motivation (a compelling picture of what the work is for), intellectual stimulation (challenging assumptions rather than issuing instructions), and individualized consideration (attending to people one at a time).
It is genuinely appealing, it matches most people's intuition about leaders they have admired, and it has been the organising idea behind leadership development programmes for three decades. None of that is evidence, which is the point of the rest of this page.
.27
transformational leadership and
leader job performance
against .71 for how satisfied
followers are with the leader
In 2004, Judge and Piccolo published a meta-analysis in the Journal of Applied Psychology pooling 87 studies and 626 correlations. Corrected for unreliability, transformational leadership correlated .44 with the criteria overall — a large figure by the standards of this literature, and the one you will see quoted.
The interesting part is the breakdown. The relationship is strongest with outcomes that are judgments about the leader, and weakest with outcomes that are things the organization produces.
| Criterion | Transformational | Contingent reward |
|---|---|---|
| Follower satisfaction with the leader | .71 | .55 |
| Rated leader effectiveness | .64 | — |
| Follower job satisfaction | .58 | .64 |
| Follower motivation | .53 | — |
| Leader job performance | .27 | .45 |
| Group or organizational performance | .26 | — |
Corrected correlations. Dashes mark values present in the original table that are not reproduced here; the full table is in the paper. Highlighted rows are the two where contingent reward came out ahead.
First, the ordinary thing works. Contingent reward is the unglamorous half of the model — be clear about what you expect, then actually reward it. Overall it scored .39 against transformational leadership's .44, and on leader job performance it came out ahead, .45 to .27. Nobody builds a two-day offsite around being clear about expectations.
Second, watch what is being predicted. A .71 correlation between "my leader is transformational" and "I am satisfied with my leader" is close to two ratings of the same underlying opinion, made by the same person at the same sitting. The authors report that both constructs looked stronger when leadership and outcome came from the same source — and, to their credit, that transformational leadership held up better than contingent reward under longitudinal designs and separate-source measurement.
That is not a debunking. It is the difference between this is associated with people admiring their boss and this makes the unit perform, and only one of those is what a training budget is being spent on.
van Knippenberg & Sitkin, 2013
Academy of Management Annals 7(1)
In 2013 Daan van Knippenberg and Sim Sitkin published a review with a subtitle that tells you where it lands: Back to the drawing board? They set out four problems with the theory and its research base.
confounds charismatic-transformational leadership with its effects.If a measure asks whether followers are inspired, it cannot then be news that inspired followers score highly.
fail to reproduce the dimensional structure specified by theoryand fail to separate empirically from other aspects of leadership.
The four problems above are the authors' own summary, taken verbatim from the published abstract on the Erasmus University repository record. The full sixty-page article is paywalled and I have not read it. No figure from it is quoted anywhere on this page, and none should be quoted from an abstract — including by me.
Batista-Foguet, Esteve &
van Witteloostuijn, 2021
PLOS ONE 16(7) — open access
The standard instrument for transformational and transactional leadership is the Multifactor Leadership Questionnaire. A 2021 open-access study examined its structure in two samples: 129 Catalan police officers and 300 US adults recruited online.
Their central argument is technical but the consequence is not. The questionnaire's scales are treated, in almost all published use, as if the items are interchangeable indicators of one underlying quality — the assumption behind reporting a Cronbach's alpha and averaging a subscale. The authors argue that most of these dimensions do not work that way, and that for such measures the usual internal-consistency statistics are, in their words, meaningless and inappropriate here.
In the US sample, fitting the conventional model returned a correlation of .99 between the transformational and transactional second-order factors — an improper solution. The authors read that as the model being misspecified rather than as proof that the two styles are the same thing. Either reading is bad news for a report that scores a manager separately on both.
This matters more than it sounds. Almost every transformational leadership programme ends by handing someone a profile with a score on each dimension and a development plan attached to the lowest one. That step assumes the dimensions are distinguishable. Thirty years of factor analyses have struggled to show that they are.
Everything above concerns the construct. Whether a programme that teaches it changes anything is a separate question, and the evidence base for it is far thinner than the citation count suggests. This is the distinction worth carrying into any vendor conversation: does your evidence cover the idea, or the intervention you are selling me?
There is at least one serious attempt at the harder question — a randomized field experiment across Danish public organizations, published in Public Administration Review in 2022, testing transformational training, transactional training and a combination against a control. It is the right design for the question.
I could not obtain the full text. So it is listed in the references and nothing from it is characterized here. If you have institutional access, it is the first thing I would read on this topic, and I would like to hear what it says.
I do not use transformational leadership, and this page is not a sales argument for the thing I do use. Servant leadership sits in the same family of follower-rated leadership constructs and inherits versions of the same problems: cross-sectional designs, self-report at both ends, and a real risk of confounding the leadership measure with the outcome. That page says so, and my own dissertation's limitations chapter says so at greater length.
A disclosure. The Multifactor Leadership Questionnaire is published by Mind Garden — the same publisher that licenses the psychological capital questionnaire this practice pays to use. I have no interest in their instruments doing badly, and one of the critiques on this site lands on an instrument I buy from them. That seemed worth saying on the page rather than leaving for someone else to notice.
And this is not aimed at practitioners. Plenty of good people have built careers delivering this material, and the leaders they worked with often did get better. The argument here is with the evidence claim, not with the room. If a programme helped, it helped; what it did not do is establish that the model behind it was measuring what it said.
A measure whose dimensions separate cleanly and repeatedly across samples, with the item content stripped of followers' evaluations of the leader. Plus randomized intervention studies using outcomes the organization already tracks for its own reasons, measured months after the training rather than at the end of it.
Both are doable. Neither is what the field has mostly spent thirty years doing.
Judge, T. A., & Piccolo, R. F. (2004). Transformational and transactional leadership: A meta-analytic test of their relative validity. Journal of Applied Psychology, 89(5), 755–768. Source of every correlation on this page, read in full text: 87 studies, 626 correlations; overall corrected validity .44 with an 80% credibility interval of .10 to .78; the criterion breakdown; and the contingent reward comparison.
van Knippenberg, D., & Sitkin, S. B. (2013). A critical assessment of charismatic-transformational leadership research: Back to the drawing board? Academy of Management Annals, 7(1), 1–60. The four problems, quoted from the published abstract. Full text paywalled and not read; no figures taken from it.
Batista-Foguet, J. M., Esteve, M., & van Witteloostuijn, A. (2021). Measuring leadership: An assessment of the Multifactor Leadership Questionnaire. PLOS ONE, 16(7), e0254329. Open access, read in full. Source of the two samples, the quoted statement on internal-consistency statistics, and the .99 improper solution.
Jacobsen, C. B., Andersen, L. B., Bøllingtoft, A., & Eriksen, T. L. M. (2022). Can leadership training improve organizational effectiveness? Evidence from a randomized field experiment on transformational and transactional leadership. Public Administration Review, 82(1), 117–131. Listed because it is the right design for the intervention question. Full text not obtained; nothing from it is characterized on this page.
Burns, J. M. (1978). Leadership. Harper & Row. Bass, B. M. (1985). Leadership and Performance Beyond Expectations. Free Press. The origin of the idea and of the transformational–transactional distinction; neither is a source of figures here.
Reviewed September 2026. If something on this page is wrong, write and say so — corrections are published with the date.
The same family of follower-rated constructs, including the one used here.
Small gains · downside riskWhere the per-dimension profile problem shows up in practice.
Useful vocabulary · thin evidenceAnother widely taught framework held to the same standard.
Every construct gets the same treatment, including the ones this practice uses and the ones that make it look good.