Incubators That Select for Confidence Select for Men

AVODA has no gender-split figure from its own selection panels, whose scoresheets have never once been averaged by gender. The Youth Enterprise with Innovation in Nigeria business plan competition asked its applicants how confident they felt about nine business tasks, from hiring good employees to estimating the cost of a new project. Three years later, David McKenzie and Dario Sansone set those answers beside what had happened to the businesses, and the confidence index turned out statistically unrelated to survival, employment or profits. Confidence is also the trait that selection panels read differently in a man and a woman. An incubator that admits founders on how sure of themselves they seem is choosing on a signal that tracks gender more closely than it tracks the business.

AVODA has a stake in this question. In what alumni told us, a drop-out split from 99 alumni, the answers ran 15% registered, 25% informal and 45% without a business. Matching those answers to the scores our own panels gave at entry is the check recommended below, and our founders are owed it first.

Key Takeaways

In a large Nigerian business plan competition, a self-rated confidence index across nine business tasks was statistically unrelated to survival, employment or profits three years on, and judges’ business plan scores predicted little.
Identical pitches draw different verdicts by the presenter’s sex: in one experiment with 521 participants, 68% chose the version narrated by a man.
Ugandan women business owners score level with men on attitudes (88% each), so a Kampala panel’s sense of “confidence” mostly picks up the props around it: a registered company, a laptop deck, fluent English.
Across 318 accelerators in GALI’s data, all-women teams were accepted at roughly 22% against 16% for all-men teams. The gender gap sits earlier, in who applies, and later, in rooms where a pitch decides the money.
The Score Split compares men and women on two kinds of points, those earned by how an applicant came across and those earned by what she has done, and shows a program lead where her own panel rewards confidence.

Nine confidence questions on a Nigerian application

The authors call the Nigerian competition the world’s largest business plan competition, and McKenzie, of the World Bank, and Sansone, of Georgetown, followed 2,506 of its applicants. Each one had answered the confidence questions on a scale from “not at all confident” to “very confident.” The researchers also had the judges’ scores, ability tests and a grit scale. Their abstract states the first finding plainly: the judges’ business plan scores were “uncorrelated with business survival, employment, sales, or profits three years later.” Among the winners, the scores did carry a weak link to hiring. Elsewhere their predictive power was close to zero.

The confidence index did as badly. In the tables for survival, employment and profits, its coefficient is small and statistically insignificant for winners and for the rest. What did carry some signal was plainer: age, a reasoning test score, the sector, and a measure of optimism for those who lost.

Two details make this study useful for an incubator. The judges scored anonymized plans, so the applicant stayed unseen, and the authors set that apart from “pitching situations where judges see the applicants.” The early rounds also selected hard on schooling. Only 5.5% of Nigerian youth held a university education, yet 52% of applicants did, and 64% of those invited to the business plan training. A screen built from a written application and an online form chose the educated before any judge read a plan.

So the most carefully studied African competition offers two warnings. Expert scoring of a written plan predicts outcomes weakly, and self-rated confidence predicts them barely at all. Put a founder in front of a panel and a third variable enters the room.

The same pitch in a man’s voice

Alison Wood Brooks, Laura Huang, Sarah Wood Kearney and Fiona Murray ran an experiment in which participants watched business pitch videos and chose which venture to fund. The content was identical; only the narrating voice changed. Among 521 participants, 68% chose the pitch narrated by a man. The study, published in PNAS in 2014, also coded real pitch competitions and found the same preference.

A second line of evidence concerns the questions many application forms ask. Christine Exley and Judd Kessler ran experiments with more than 4,000 adults in online labor markets and more than 10,000 school pupils. Women described their own ability and performance less favorably than men who had performed equally well, and the gap stayed when every incentive to promote oneself was removed. It appears by the sixth grade. A form question asking applicants to score their own ability to lead a team (our invented example, not an item from these studies) is partly a survey of who was raised to answer generously.

The habit underneath has a name in Entrepreneurial Ecosystems: A Gender Perspective (Cambridge University Press, 2022), by Banu Ozkazanc-Pan and Susan Clark Muntean. Much of entrepreneurship, they write, rewards performance: “pitching well, exuding confidence, projecting excitement.” Their fieldwork in Asheville, North Carolina, in 2018 and 2019 found founders and support leaders agreeing that “men are perceived as worthy of support when exhibiting overconfidence and a great deal of aggressiveness, whereas women are viewed with suspicion” for the same traits. The same leaders placed the gender gap in other cities and described their own as fair.

Confidence, as a panel sees it, predicts little. Women report less of it for the same performance. When a woman displays it anyway, the evidence suggests a panel may read it as a warning. A criterion with all three properties sorts applicants by sex while appearing to sort them by promise.

What a Kampala panel is reading as confidence

The book’s evidence comes from Boston, St. Louis and Asheville, and much American advice treats the gap as women’s lower self-belief, to be repaired with confidence training. The Ugandan numbers point elsewhere.

The State of Entrepreneurship in Uganda 2024, a trade ministry study of more than 3,000 small businesses, scored each one on eight dimensions. Women-led and men-led businesses scored the same on attitudes, 88% each, and nearly the same on willingness to grow, 78% against 79%. The gaps opened on registration (35% against 44%), technology (33% against 38%) and linkages (22% against 26%). Among the barriers women met in registering a business, the study lists language and skills.

An older data point agrees. In the 2003 Global Entrepreneurship Monitor survey for Uganda, analyzed by Laura Orobia, Arthur Sserwanga and Gerrit Rooks, 75.9% of men and 75.3% of women starting a business were classed as willing to take risks, a difference too small to register statistically. Across the other countries in that year’s survey, pooled, the gap between men and women was statistically significant.

Here the American frame stops fitting. A Ugandan woman founder reports roughly the attitude of the man beside her. What she more often lacks is the kit a panel mistakes for confidence: a slide deck built on a laptop, a certificate of incorporation, practice pitching in English, a friend who has sat on a panel and coached her answer. A selection interview at a Kampala hub rewards all of these. Fewer women hold them, for reasons that run through registration offices and device prices more than through self-belief. Confidence training for her answers a gap the Ugandan data barely shows, and a Kampala panel that scores fluency and polish is mostly scoring papers, devices and schooling.

Acceptance rates across 318 accelerators

The Global Accelerator Learning Initiative (GALI) holds the largest dataset on who enters accelerators: 14,985 ventures that applied to 318 programs between 2013 and 2018, a quarter of them in Africa. Its 2020 brief on women-led startups reports that all-women founding teams “are slightly more likely to be accepted into the program to which they apply (roughly 22% vs. 16% for all-men teams)”. An IFC research snapshot drawing on GALI data reached the same verdict: male-led and female-led startups were selected at similar rates.

For this one stage of selection, the finding runs against the headline above. Across impact-oriented accelerators, the decision to admit from an application form treats women at least as well as men in aggregate. An incubator lead who reads the evidence fairly should start from there.

Yet the same documents locate the confidence premium on either side of that decision. Before it, only 13% of applicant teams were all-women. In programs where at least 45% of the selection committee were women, 56% of applicant teams included a woman founder, against 45% in the rest, and their cohorts showed the same gap. The book’s Boston interviews make the point from the inside: “Women need to be explicitly extended an invitation.” After admission, in the rooms where a pitch decides the money, the gap widened. Among ventures that raised funding, women-led teams raised nearly $100,000 less in equity, and the IFC team found that acceleration widened the equity gap.

GALI’s survey also leaves confidence unmeasured, as its own brief notes. So the witness settles one question and leaves another open. Form-based admission at these programs looks even-handed. Selection that runs through a live pitch, an interview or a demo day sits where the evidence of a premium is strongest, and that is the format many Kampala incubators use. The Score Split checks which of the two your own panel resembles.

The Score Split

The Score Split works on paper a program already holds: last intake’s scoresheets. It answers one diagnostic question, whether your panel is paying for confidence, and leaves the redesign of the rubric for later.

Sort every scored item into two piles. Room points come from how an applicant came across: delivery, presence, energy, “passion”, a self-rating of ability, the panel’s sense of who seems coachable. Record points come from something that happened and could be checked: months trading, repeat customers, a sales book, a product a panelist has used.
Total the two piles separately for every applicant you scored.
Average them by gender at three steps: everyone who applied, everyone shortlisted, everyone admitted.
Read the gap. Women level on Record points and behind on Room points means the panel is scoring confidence. Women behind on both points upstream, to who heard about the call and who chose to apply. Women level on both and still admitted less often means the cut-off or the tie-breaks need a look.

Consider a worked example with invented figures. Suppose Record points run out of 40 and Room points out of 60, and the shortlisted men and women both average 31 Record points. The men average 44 Room points and the women 36. With a cut-off at 70, the average man clears it at 75 and the average woman falls short at 67, and the whole of that difference was earned in the room.

The exercise takes a morning with a spreadsheet, and its output is two columns a board can read. What a better rubric should contain is the subject of a separate piece; the split only tells you whether you need one.

Seven sons before Samuel

Samuel arrives in Bethlehem with a horn of oil and a commission to anoint one of Jesse’s sons. Eliab walks in first, and Samuel is sure at once: “Surely the Lord’s anointed stands here before the Lord.” He has done what most panels do, reading height and bearing as fitness. The correction comes immediately: “People look at the outward appearance, but the Lord looks at the heart.” Seven sons pass, and each is passed over. Set beside the evidence, the scene names both halves of the problem. Our judges predict poorly, which Scripture states as a limit on human sight, and a rubric can at most stop a panel paying for outward signs. Our calls also miss people, and there the text lays down a duty. Samuel asks a question scoresheets rarely prompt: “Are these all the sons you have?” The youngest is out with the sheep, left behind while his brothers were presented. Samuel’s answer keeps the whole feast waiting for the absent one: “Send for him; we will not sit down until he arrives.”

Frequently Asked Questions

Do accelerators accept fewer women-led startups than men-led ones?

Across GALI’s data on 14,985 ventures applying to 318 accelerators between 2013 and 2018, all-women teams were accepted at roughly 22%, against 16% for all-men teams. The gap appears before and after admission: only 13% of applicant teams were all-women, and accelerated women-led ventures raised less equity than men-led ones.

Does a founder’s confidence predict whether her business will succeed?

The best African evidence says weakly at most. In the Youth Enterprise with Innovation in Nigeria competition, a self-rated confidence index across nine business tasks was statistically unrelated to survival, employment or profits three years later. Judges’ scores also predicted little. Age, a reasoning test and the business sector carried more information than either.

Why do investors favor pitches delivered by men?

Experiments point to the presenter, with the content held constant. In a 2014 study by Brooks, Huang, Kearney and Murray, 521 participants watched identical pitch videos, and 68% chose the version narrated by a man. Fieldwork in American cities found overconfidence read as strength in men and treated with suspicion in women.

Are Ugandan women entrepreneurs less confident than men?

The survey data suggest the attitude gap is small. The 2024 State of Entrepreneurship in Uganda study found women-led and men-led businesses level on attitudes at 88% and close on willingness to grow. Gaps opened in registration, technology and business linkages, which shape how polished a pitch looks.

How can an incubator check whether its selection rewards confidence?

Take last intake’s scoresheets and split each applicant’s points into those earned by how she came across and those earned by checkable history, such as months trading or repeat customers. Average both by gender at the applied, shortlisted and admitted steps. A gap only in the first kind points to a panel scoring confidence.

Next Step

Pull last intake’s scoresheets, sort every item into Room and Record points, and average both by gender at the applied, shortlisted and admitted steps. If you chair selection for an incubator or accelerator, bring those two columns to a strategy call with Joshua. We will read the gap together and decide whether your next call for applications, your interview format or your cut-off needs changing before the next panel sits, so the next cohort is chosen on what founders have done.

Sources and Evidence

McKenzie and Sansone, “Predicting Entrepreneurial Success is Hard: Evidence from a Business Plan Competition in Nigeria”, Journal of Development Economics, 2019 (World Bank Open Knowledge manuscript): 2,506 applicants; judges’ scores and the self-confidence index against outcomes three years later; anonymized plans; the education screen.
MIT News, “Study says attractive men fare best in gaining venture capital”, on Brooks, Huang, Kearney and Murray, PNAS 2014: identical pitches, 521 participants, 68% chose the male-narrated version.
Exley and Kessler, “The Gender Gap in Self-Promotion”, NBER Working Paper 26345 (Quarterly Journal of Economics, 2022): women describe equal performance less favorably, with or without incentives.
Ozkazanc-Pan and Clark Muntean, Entrepreneurial Ecosystems: A Gender Perspective, Cambridge University Press, 2022 (DOI 10.1017/9781009023641): chapter 3 on performativity; chapter 5 on Asheville and Boston support organizations.
Global Accelerator Learning Initiative, “Accelerating Women-led Startups”, knowledge brief, March 2020: 14,985 ventures, 318 programs; acceptance rates of 22% and 16%; 13% all-women applicant teams; the equity gap after acceleration.
IFC, We-Fi, World Bank Africa Gender Innovation Lab, GALI and Village Capital, “Venture Capital and the Gender Financing Gap: The Role of Accelerators”, 2020: similar selection rates; women selectors and applicant pools (56% against 45%); acceleration widens the equity gap.
Ministry of Trade, Industry and Cooperatives, “The State of Entrepreneurship in Uganda 2024: Performance of Women-led Businesses” (Ichuli and Ipsos): attitudes 88% for both; registration, technology and linkages by gender.
Orobia, Sserwanga and Rooks, “Risk Taking and Start-up Capital: Exploring Gender Differences in Uganda”, Journal of Economics and Behavioral Studies, 2011: GEM Uganda 2003; risk attitudes of 75.9% and 75.3%.
1 Samuel 16:1-13 (NIV), Bible Gateway: Samuel at Jesse’s house.