What the Evidence Has Actually Established

For women-founded startups in the Indian sample, two findings now sit side by side: women founders have more measured potential educational and professional connections, while the measured founder–board, founder–angel, and board–angel relationships are less common. The useful judgment at this point is not to choose one finding and discard the other. It is to identify what those observations establish, what they leave open, and what evidence is needed before moving from network structure to capital.

Two facts, one conversion gap

The first fact concerns broad potential centrality. After controls for firm age, tier-1-city location, state, and sector, the woman-founder coefficients are positive for any connection, educational connection, and professional connection. The estimates are 0.0408, 0.0439, and 0.0214 for the indicator outcomes. The corresponding count coefficients are 18.09, 18.03, and 0.0692. These are associations on constructed network measures in a sample of 54,585 firms. They do not mean that every woman founder has a fixed number of useful contacts, but they do reject the simple premise that the observed difference begins with women having fewer potential links.

The second fact concerns relationships around a specific venture. With state and sector fixed effects, the woman-founder coefficients are −0.0764 for a founder–board link, −0.0519 for a founder–angel link, and −0.0733 for a board–angel link. The relevant outcome means are approximately 0.454, 0.333, and 0.360. In plain terms, the study finds fewer of these measured role relationships for women founders even while finding more broad potential centrality.

The two facts describe different layers of a network. A background match can widen the field of possible contact without producing a relationship with someone who has a board or investment role in the venture. A role link can be created through a route that the broad education and experience measures do not capture. The evidence therefore points to a conversion gap: potential proximity and consequential relationship do not move together. “Gap” here names the difference between observed layers. It does not, by itself, name the cause.

A useful way to test the combined finding is to imagine four firms with the same broad potential-network score. One may have a founder–board link but no angel link; another may have a founder–angel link but no board link; a third may have neither role-specific link; and a fourth may have both. The broad score alone cannot distinguish these routes. Nor can the role flags tell us whether the relationship was active, beneficial, or created after the firm had already attracted capital. The study’s contribution is to show that the distribution across these layers differs by founder gender, not to assign a value to every possible configuration.

That matters for interpretation because the word “conversion” can hide two different judgments. The first is descriptive: a potential match is less often accompanied by a measured role link for women founders in the observed sample. The second is causal: some process has prevented or discouraged that match from becoming a role link. The first is supported by the tables; the second requires evidence about timing, decisions, and alternatives that the tables do not contain. Keeping those judgments apart is the difference between a useful diagnosis and an accusation disguised as an explanation.

This is the point at which a common reading goes wrong.