The Takeaway
Three peer-reviewed studies point to a clear conclusion: coaching produces measurable improvements in leadership, behaviour and personal outcomes. The evidence isn’t perfect—but for women in STEM, it’s stronger than you might expect.
The Evidence
I’m an engineer. When someone suggests coaching, my first question is the same as yours: where’s the evidence?
Not testimonials. Not “86% of clients said they felt better.” Actual, rigorous, independently conducted research. The kind you’d want to see before you trusted any other intervention.
I went looking. Here’s what exists.
The gold standard
In 2023, a team of researchers published a meta-analysis in the Academy of Management Learning & Education. They reviewed 37 randomised controlled trials (RCTs) of workplace coaching, the same methodology used to test whether a drug works. Across 2,528 participants, coaching produced a moderate, statistically significant effect on leadership and personal outcomes (g=0.59 for those stats geeks amongst us).
In plain terms: a measurable, replicable improvement. Not a miracle. Not a placebo. A real effect, tested properly.
The researchers were honest about the caveats. They found signs of publication bias (positive results get published more). Self-reported outcomes showed larger effects than observed ones — both matter. But even accounting for these, the effect held.
One more thing from that analysis: the effects were somewhat larger for female coachees. Not dramatically (the authors’ word is “somewhat”), but the right direction for the people this piece is written for.
The study that tested it on women in STEM
This is the one I want you to know about.
In 2022, a study was published in PNAS (the Proceedings of the National Academy of Sciences). It followed a coaching and peer-group intervention with 44 early-career women at a North American biotechnology company, with baseline measures collected just before the COVID lockdowns and a matched control group for comparison.
Women who received the coaching rated their own soft skills more than 9% higher than at baseline. The matched control group’s ratings fell by about 3.5% over the same period; a treatment effect of nearly 13%, or about 2% for each month of the programme. The intervention didn’t just improve things; it stopped a decline that hit everyone else.
One thing to be clear about. That decline isn’t normal. The study ran through the first year of Covid, with everyone working remotely, and the authors put the control group’s drop down to that. Nobody has measured what happens in an ordinary year. What holds either way is the gap between the two groups. That is the finding.
And here’s the part that can’t be dismissed as people being generous about themselves: the company’s own managers rated the coached women’s performance higher than both control groups.
I should be straight about the limits, because they matter. The women weren’t randomly assigned, the authors say so themselves, and warn that causal inferences should be treated with caution. It’s 44 women at one company, 35 of whom finished the programme. And the headline number is what they thought of their own skills, not an objective test.
It is still the best evidence that exists on coaching women in STEM specifically. That tells you something about the state of the research, and it’s why I won’t oversell it.
What coaching changes
A separate meta-analysis in Frontiers in Psychology (2023) sought to look at this by researching what exactly does coaching shift?
Their finding: coaching’s biggest effect is on behaviour: what people do, and above all whether they reach the goals they set. It also moves things you’d expect to be fixed: self-belief, resilience, even personality traits. Less, but measurably.
Read that again. Coaching changes what you do, but, more importantly, it changes who you are in ways you might not expect.
For a mid-career woman in STEM weighing up whether to push for promotion, move sideways, or leave, the quality of that decision, and the confidence to act on it, is worth a great deal.
What I’ll be honest about
Most coaching ROI statistics you’ll find, like the “7x return” numbers and the “86% of organisations” claims, come from the coaching industry itself. ICF commissions them. Coaching bodies cite them. They point in the right direction, but they’re not independent evidence, and I won’t pretend they are.
The three studies above are different. All three are peer-reviewed. Two of them, the meta-analysis and the Frontiers review, are also independent of anyone selling coaching.
The PNAS study I should be straight about. It was run in partnership with the platform whose programme was being tested, which is exactly the kind of arrangement I’ve just been sceptical about. Three things stop me from discounting it: the researchers were academics at Stanford, not the vendor; they declare no competing interests; and they published their data and analysis code for anyone to check. That is better practice than almost anything the coaching industry produces about itself. It still isn’t the same as independent.
They’re also, honestly, rarer than they should be. The coaching profession needs more research like this, not less.
If you’re the kind of person who needs to see the evidence before you commit, I respect that. I’m the same way. The evidence is moderate, consistent, and — for women in STEM specifically stronger than most people expect.
Sally Purdie CEng FIMMM — Chartered Engineer | STEM Retention Coach
References
de Haan, E. and Nilsson, V.O. (2023) ‘What can we know about the effectiveness of coaching? A meta-analysis based only on randomized controlled trials’, Academy of Management Learning & Education, 22(4). doi: 10.5465/amle.2022.0107.
Melin, J.L. and Correll, S.J. (2022) ‘Preventing soft skill decay among early-career women in STEM during COVID-19: Evidence from a longitudinal intervention’, Proceedings of the National Academy of Sciences, 119(32), e2123105119. doi: 10.1073/pnas.2123105119.
Nicolau, A., Candel, O.S., Constantin, T. and Kleingeld, A. (2023) ‘The effects of executive coaching on behaviors, attitudes, and personal characteristics: a meta-analysis of randomized control trial studies’, Frontiers in Psychology, 14, 1089797. doi: 10.3389/fpsyg.2023.1089797.
