[saeranv]

Careful, this seems to be falling for the prosecutor's fallacy where the probabilities aren't normalized to account for base rates of the hypothesis occuring. So in this case, the percent of successful valuations that have a female founders will likely be extremely low simply because the gender ratio is disproportionaly male, and being successful is extremely rare.

For example, let's say probability of being successful is due to random chance, and effects both genders equally. Then ratio of successful women to successful men would just reflect the ratio of women to men. Now imagine if the actual valuation of successful companies is exponentially weighted, so that the top 1 or 2 companies make up the bulk of the total valuation. Again, these two companies are more likely to be founded by men given the base gender rate, and now the percent of valuation attributed to men would completely dominate the valuation percent.

To get the real effectiveness of female founders, we need to account for the low probability of success and low probability of being female.

I think both can be achieved with Bayes theorem:

  p($_high|f) = p($_high) p(f|$_high) / Sum_i[p($_i) p(f|$_i)]
  p($_high|m) = p($_high) p(m|$_high) / Sum_i[p($_i) p(m|$_i)]

The first equation gets the number of founders who are female and successful, and then divide that by the number of female founders in total, and the second one does the same for males. That should give you a apples to apples comparison.

Note: p($_i) represents a sequence of valuation probabilities subdivided to reflect low valuation, moderate valuation etc. I think this should account for the exponential distribution of valuation, but someone correct me if I'm wrong.

[YetAnotherNick]

I commented the same thing in another thread. I am not arguing the base rate or whether female founders experience discrimination, I am just arguing YC's female ratio is definitely not higher than other VC's which pg hints at due to Jessica. In fact I think other VCs have something in the range of 20% IIRC(would not mind to be corrected though).

[saeranv]

Okay fair enough, but why look at female founders in terms total valuation percent, then? I think my point about the distortion from the (likely) exponential distribution of total valuation still stands, although in this case with regards to comparison between VCs.

Isn't it straightforward to just look at percent of female founders?

[YetAnotherNick]

> Isn't it straightforward to just look at percent of female founders?

Totally agreed. I couldn't find the numbers though.

[saeranv]

To your credit, at least you got data to begin investigating your claim. I'm usually to lazy to do that...