Leadership training changes what leaders do back at work, with a corrected effect of 0.82. That figure comes from Lacerenza and colleagues, who in 2017 pooled 335 independent samples, restricted themselves to data from employees rather than students, and tested fifteen features of programme design against four outcomes. It remains the largest evidence base on the question, and its answer is more favourable than the reputation of the field would suggest.
Before it, the honest answer was that nobody knew. Earlier reviews produced effects from negligible to large depending on which studies they let in, and the practitioner literature settled into a comfortable stalemate. Vendors cited the optimistic reviews, sceptics cited the pessimistic ones, and both sides had a reference.
All four outcomes improve, and behaviour at work improves most
The meta-analysis reports corrected effect sizes across the four levels that training evaluation has used since Kirkpatrick set them out in 1959.
| Outcome | What it measures | Corrected effect |
|---|---|---|
| Reactions | Whether participants valued the programme | 0.63 |
| Learning | Whether they know more afterwards | 0.73 |
| Transfer | Whether their behaviour at work changed | 0.82 |
| Results | Whether team or organisational outcomes moved | 0.72 |
By the usual convention an effect of 0.5 is moderate and 0.8 is large. All four sit in that band. The largest is transfer, the one that matters and the one everyone assumes is weakest. The authors are explicit that their estimates come out higher than earlier work, and they attribute the difference to a broader and more recent body of primary studies.
The caveats are real. These are effect sizes from studies that were designed to be evaluated, which is a different population from programmes in general. And within the results criterion the number of primary studies is small enough that several moderators could not be tested at all.
Data from 335 independent samples suggest that leadership training is substantially more effective than previously thought.
A formal needs analysis has the largest single effect
Of the fifteen design features tested, the one with the largest single effect is the least glamorous. On transfer, nothing else comes close.
Programmes preceded by a formal needs analysis produced significantly stronger learning and significantly stronger transfer than programmes without one. The test statistic on transfer is the largest in the paper by a distance. There was no significant difference on results, and reactions could not be tested because too few studies measured them.
A needs analysis here means establishing what the leaders in question cannot yet do before deciding what to teach them. It is routinely skipped, on the grounds that it delays the start and that everyone already knows what the problem is. The evidence says it is the difference between a programme people enjoy and a programme that changes what they do on Monday.
Practice beats explanation, and the combination beats both
The meta-analysis separates three delivery methods: information, demonstration and practice. Programmes that used all three produced significantly stronger learning than programmes using practice alone or information alone, and significantly stronger transfer than programmes using only information and practice, or only demonstration and practice.
For the results criterion the picture is thinner. All three methods beat information alone, but there was no significant advantage over programmes that combined information and practice, or over practice alone.
A lecture with a case study is the weakest common format. The addition that does the work is watching the behaviour done well and then doing it badly a few times with someone watching. That takes hours, and hours are what employers have been cutting. Spending per trained person has fallen about a quarter in real terms since 2005 while the number of people trained has risen.
Spaced sessions change behaviour, a single block does not
Programmes spread across several sessions produced significantly stronger results and significantly stronger transfer than programmes delivered in one block. For learning there was no significant difference.
The pattern is consistent and slightly uncomfortable. If the only thing measured is whether participants know more at the end, a one-day intensive performs as well as six weekly sessions. The difference shows up in whether the behaviour survives contact with the job. The end-of-course questionnaire does not measure that.
Two findings are cited more confidently than the paper supports
Face-to-face against virtual. Face-to-face programmes produced significantly stronger transfer than virtual ones. For learning there was no significant difference, and for results and reactions the comparison could not be made because there were too few studies. The paper is frequently cited as showing that virtual leadership training is less effective, without the qualification. It showed that in one of four outcomes, on data collected before 2017, before most organisations learned how to run anything online.
Feedback. Programmes including feedback produced significantly stronger transfer than those without. On reactions, learning and results there was no significant difference. The comparison vendors sell hardest, 360-degree feedback against single-source feedback, showed no significant advantage on any of learning, transfer or results. That finding has its own treatment in what the evidence says about 360-degree feedback.
Junior leaders transfer more than senior ones
The moderator that should change budget decisions is the level of the participants.
Programmes delivered to low-level leaders produced significantly stronger transfer than programmes delivered to high-level leaders, and significantly stronger than programmes delivered to middle-level leaders. There was no significant difference between high and middle. On learning, no level differed significantly from any other.
Senior and junior leaders learn the material equally well. The junior ones are markedly more likely to change their behaviour because of it. Most leadership budgets are allocated the other way round, with cost per head rising steeply with seniority. The evidence does not say executive development is worthless. It says the same money spent one or two levels down is more likely to show up as changed behaviour, which matters given that manager engagement is falling faster than anyone else's.
Voluntary programmes transfer more, mandatory ones move results
Voluntary programmes produced significantly stronger transfer than mandatory ones. Mandatory programmes produced significantly stronger results than voluntary ones. There was no significant difference on learning.
Both can be true at once, and the explanation is probably selection. People who choose to attend are more likely to act on what they learn. When attendance is compulsory the whole unit goes through it, so whatever effect there is applies to everyone at the same time, and team-level outcomes pick that up. It is a genuine trade-off rather than a contradiction. Whether selection is the whole explanation is the one question here the meta-analysis could not test.
Common questions
How effective is leadership training? Across 335 independent samples, corrected effect sizes were 0.63 for participant reactions, 0.73 for learning, 0.82 for behaviour change at work and 0.72 for organisational results. All four count as moderate to large by the usual convention.
Does virtual leadership training work as well as face-to-face? On learning there was no significant difference. On behaviour change at work, face-to-face performed significantly better. The comparison could not be made for organisational results or participant reactions because too few studies measured them, and the underlying data predates 2017.
Does 360-degree feedback improve leadership training outcomes? Not measurably, in this evidence. Programmes with 360-degree feedback showed no significant advantage over programmes with single-source feedback on learning, on behaviour change or on results.
Should leadership training be one intensive block or spread over weeks? Spread over weeks, if the goal is behaviour change. Spaced programmes produced significantly stronger transfer and results than single-block programmes. For knowledge alone the two formats were indistinguishable.
Is a training needs analysis worth the delay? The evidence says yes, and by a wide margin on behaviour change. It is the strongest single moderator in the analysis for transfer, and it also predicted significantly stronger learning.
Who benefits most from leadership training? Junior leaders. Programmes delivered to low-level leaders produced significantly stronger behaviour change than those delivered to middle or senior leaders, while learning was equivalent across levels.
What is the difference between learning and transfer in these figures? Learning is whether participants know or can do more when the programme ends, usually measured by a test or an assessment. Transfer is whether their behaviour at work is different afterwards, usually measured by ratings from people who work with them. A programme can score well on the first and not at all on the second.
Sources
- Lacerenza, Reyes, Marlow, Joseph & Salas (2017), Leadership training design, delivery, and implementation: A meta-analysis, Journal of Applied Psychology 102(12)
- Full text of the meta-analysis, Doerr Institute for New Leaders, Rice University
- Kirkpatrick's four levels of training evaluation, as used in the meta-analysis
