WEBVTT

00:00:05.280 --> 00:00:07.416
This is me at five years old,

00:00:07.440 --> 00:00:11.960
shortly before jumping
into this beautifully still pool of water.

00:00:12.840 --> 00:00:17.176
I soon find out the hard way
that this pool is completely empty

00:00:17.200 --> 00:00:19.896
because the ice-cold water
is near freezing

00:00:19.920 --> 00:00:22.200
and literally takes my breath away.

00:00:23.360 --> 00:00:25.256
Even though I already know how to swim,

00:00:25.280 --> 00:00:29.440
I can't get up to the water's surface,
no matter how hard I try.

00:00:30.040 --> 00:00:33.520
That's the last thing I remember
trying to do before blacking out.

00:00:34.520 --> 00:00:37.776
Turns out, the lifeguard on duty
had been chatting with two girls

00:00:37.800 --> 00:00:39.376
when I jumped in,

00:00:39.400 --> 00:00:41.296
and I was soon underwater,

00:00:41.320 --> 00:00:45.256
so he couldn't actually
see or hear me struggle.

00:00:45.280 --> 00:00:48.856
I was eventually saved
by a girl walking near the pool

00:00:48.880 --> 00:00:51.120
who happened to look down and see me.

00:00:52.160 --> 00:00:54.816
The next thing I know,
I'm getting mouth-to-mouth

00:00:54.840 --> 00:00:58.960
and being rushed to the hospital
to determine the extent of my brain loss.

00:00:59.880 --> 00:01:03.376
If I had been flailing
at the water's surface,

00:01:03.400 --> 00:01:06.640
the lifeguard would have noticed
and come to save me.

00:01:07.520 --> 00:01:10.456
I share this near-death experience
because it illustrates

00:01:10.480 --> 00:01:14.720
how dangerous things are
when they're just beneath the surface.

NOTE Paragraph

00:01:15.800 --> 00:01:19.616
Today, I study implicit
gender bias in start-ups,

00:01:19.640 --> 00:01:24.176
which I consider to be
far more insidious than mere overt bias

00:01:24.200 --> 00:01:25.800
for this very same reason.

00:01:26.440 --> 00:01:28.496
When we see or hear an investor

00:01:28.520 --> 00:01:31.856
behaving inappropriately
towards an entrepreneur,

00:01:31.880 --> 00:01:33.816
we're aware of the problem

00:01:33.840 --> 00:01:36.680
and at least have a chance
to do something about it.

00:01:37.480 --> 00:01:40.376
But what if there are subtle differences

00:01:40.400 --> 00:01:43.720
in the interactions
between investors and entrepreneurs

00:01:44.800 --> 00:01:46.936
that can affect their outcomes,

00:01:46.960 --> 00:01:50.056
differences that we're not conscious of,

00:01:50.080 --> 00:01:53.160
that we can't directly see or hear?

NOTE Paragraph

00:01:54.200 --> 00:01:57.216
Before studying start-ups
at Columbia Business School,

00:01:57.240 --> 00:02:01.656
I spent five years running
and raising money for my own start-up.

00:02:01.680 --> 00:02:05.976
I remember constantly racing around
to meet with prospective investors

00:02:06.000 --> 00:02:08.735
while trying to manage my actual business.

00:02:08.759 --> 00:02:11.656
At one point I joked
that I had reluctantly pitched

00:02:11.680 --> 00:02:15.456
each and every family member,
friend, colleague, angel investor

00:02:15.480 --> 00:02:17.680
and VC this side of the Mississippi.

00:02:18.640 --> 00:02:21.296
Well, in the process of speaking
to all these investors,

00:02:21.320 --> 00:02:23.536
I noticed something
interesting was happening.

00:02:23.560 --> 00:02:27.936
I was getting asked a very different
set of questions than my male cofounder.

00:02:27.960 --> 00:02:31.376
I got asked just about everything
that could go wrong with the venture

00:02:31.400 --> 00:02:33.536
to induce investor losses,

00:02:33.560 --> 00:02:37.376
while my male cofounder was asked
about our venture's home run potential

00:02:37.400 --> 00:02:39.776
to maximize investor gains,

00:02:39.800 --> 00:02:42.656
essentially everything
that could go right with the venture.

00:02:42.680 --> 00:02:45.856
He got asked how many new customers
we were going to bring on,

00:02:45.880 --> 00:02:49.816
while I got asked how we were going
to hang on to the ones we already had.

00:02:49.840 --> 00:02:53.216
Well, as the CEO of the company,
I found this to be rather odd.

00:02:53.240 --> 00:02:56.080
In fact, I felt like
I was taking crazy pills.

00:02:57.240 --> 00:02:59.416
But I eventually
rationalized it by thinking,

00:02:59.440 --> 00:03:01.896
maybe this has to do
with how I'm presenting myself,

00:03:01.920 --> 00:03:04.480
or it's something
simply unique to my start-up.

NOTE Paragraph

00:03:05.520 --> 00:03:09.256
Well, years later I made the difficult
decision to leave my start-up

00:03:09.280 --> 00:03:13.176
so I could pursue a lifelong dream
of getting my PhD.

00:03:13.200 --> 00:03:17.496
It was at Columbia that I learned
about a social psychological theory

00:03:17.520 --> 00:03:21.976
originated by Professor Tory Higgins
called "regulatory focus,"

00:03:22.000 --> 00:03:25.896
which differentiates between
two distinct motivational orientations

00:03:25.920 --> 00:03:27.480
of promotion and prevention.

00:03:28.400 --> 00:03:30.936
A promotion focus is concerned with gains

00:03:30.960 --> 00:03:35.096
and emphasizes hopes, accomplishments
and advancement needs,

00:03:35.120 --> 00:03:37.776
while a prevention focus
is concerned with losses

00:03:37.800 --> 00:03:41.480
and emphasizes safety,
responsibility and security needs.

00:03:42.200 --> 00:03:45.256
Since the best-case scenario
for a prevention focus

00:03:45.280 --> 00:03:48.096
is to simply maintain the status quo,

00:03:48.120 --> 00:03:51.776
this has us treading water
just to stay afloat,

00:03:51.800 --> 00:03:56.616
while a promotion focus instead
has us swimming in the right direction.

00:03:56.640 --> 00:04:00.640
It's just a matter
of how far we can advance.

NOTE Paragraph

00:04:01.280 --> 00:04:04.896
Well, I had my very own eureka moment
when it dawned on me

00:04:04.920 --> 00:04:06.376
that this concept of promotion

00:04:06.400 --> 00:04:09.936
sounded a lot like the questions
posed to my male cofounder,

00:04:09.960 --> 00:04:12.920
while prevention resembled
those questions asked of me.

00:04:14.200 --> 00:04:15.696
As an entrepreneurship scholar,

00:04:15.720 --> 00:04:18.976
I started digging into the research
on start-up financing

00:04:19.000 --> 00:04:21.336
and discovered there's an enormous gap

00:04:21.360 --> 00:04:25.400
between the amount of funds
that male and female founders raise.

00:04:26.360 --> 00:04:30.560
Although women found
38 percent of US companies,

00:04:31.640 --> 00:04:34.640
they only get two percent
of the venture funding.

00:04:35.680 --> 00:04:37.656
I got to thinking:

00:04:37.680 --> 00:04:41.656
what if this funding gap is not due
to any fundamental difference

00:04:41.680 --> 00:04:43.800
in the businesses
started by men and women?

00:04:44.680 --> 00:04:47.776
What if women get less funding than men

00:04:47.800 --> 00:04:51.200
due to a simple difference
in the questions that they get asked?

00:04:52.080 --> 00:04:54.736
After all, when it comes
to venture funding,

00:04:54.760 --> 00:04:59.136
entrepreneurs need to convince investors
of their start-up's home run potential.

00:04:59.160 --> 00:05:00.976
It's not enough to merely demonstrate

00:05:01.000 --> 00:05:03.296
you're not going to lose
your investors' money.

00:05:03.320 --> 00:05:07.656
So it makes sense that women
would be getting less funding than men

00:05:07.680 --> 00:05:08.936
if they're engaging

00:05:08.960 --> 00:05:12.776
in prevention as opposed to
promotion-oriented dialogues.

NOTE Paragraph

00:05:12.800 --> 00:05:15.336
Well, I got the chance
to test this hypothesis

00:05:15.360 --> 00:05:20.416
on companies with similar quality
and funding needs across all years

00:05:20.440 --> 00:05:23.856
at the funding competition
known as TechCrunch Disrupt

00:05:23.880 --> 00:05:26.576
Startup Battlefield has run
in New York City

00:05:26.600 --> 00:05:28.920
since its inception in 2010.

00:05:30.200 --> 00:05:34.016
TechCrunch is widely regarded as
the ideal place for start-ups to launch,

00:05:34.040 --> 00:05:38.536
with participants including start-ups
that have since become household names,

00:05:38.560 --> 00:05:40.496
like Dropbox, Fitbit and Mint,

00:05:40.520 --> 00:05:43.280
presenting to some of the world's
most prominent VCs.

00:05:44.240 --> 00:05:48.136
Well, despite the comparability
of companies in my sample,

00:05:48.160 --> 00:05:51.856
male-led start-ups went on
to raise five times as much funding

00:05:51.880 --> 00:05:53.200
as the female-led ones.

00:05:54.240 --> 00:05:58.400
This made me especially curious to see
what's driving this gender disparity.

NOTE Paragraph

00:05:59.400 --> 00:06:00.656
Well, it took a while,

00:06:00.680 --> 00:06:05.216
but I got my hands on all the videos of
both the pitches and the Q and A sessions

00:06:05.240 --> 00:06:07.840
from TechCrunch,
and I had them transcribed.

00:06:08.520 --> 00:06:10.456
I first analyzed the transcripts

00:06:10.480 --> 00:06:14.136
by loading a dictionary
of regulatory-focused terms

00:06:14.160 --> 00:06:18.216
into the Linguistic Inquiry
and Word Count software called LIWC.

00:06:18.240 --> 00:06:20.856
This LIWC software
generated the frequencies

00:06:20.880 --> 00:06:24.896
of promotion and prevention words
in the transcribed text.

00:06:24.920 --> 00:06:26.416
As a second method,

00:06:26.440 --> 00:06:30.376
I had each of the questions
and answers manually coded

00:06:30.400 --> 00:06:33.200
by the Tory Higgins
Research Lab at Columbia.

00:06:34.680 --> 00:06:36.936
Regardless of the topic at hand,

00:06:36.960 --> 00:06:40.816
an intention can be framed
in promotion or prevention.

00:06:40.840 --> 00:06:44.400
Let's take that topic of customers
I mentioned briefly earlier.

00:06:45.200 --> 00:06:47.416
A promotion-coded question sounds like,

00:06:47.440 --> 00:06:50.776
"How many new customers
do you plan to acquire this year?"

00:06:50.800 --> 00:06:53.016
while a prevention-coded one sounds like,

00:06:53.040 --> 00:06:56.360
"How do you plan to retain
your existing customers?"

00:06:57.240 --> 00:06:58.576
During the same time,

00:06:58.600 --> 00:07:01.016
I also gathered background information

00:07:01.040 --> 00:07:05.096
on the start-ups and entrepreneurs
that can affect their funding outcomes,

00:07:05.120 --> 00:07:08.176
like the start-up's age,
quality and funding needs

00:07:08.200 --> 00:07:10.376
and the entrepreneur's past experience,

00:07:10.400 --> 00:07:14.040
so I could use these data points
as controls in my analysis.

NOTE Paragraph

00:07:14.840 --> 00:07:16.976
Well, the very first thing that I found

00:07:17.000 --> 00:07:21.616
is that there's no difference in the way
entrepreneurs present their companies.

00:07:21.640 --> 00:07:24.736
In other words, both male
and female entrepreneurs

00:07:24.760 --> 00:07:28.056
use similar degrees
of promotion and prevention language

00:07:28.080 --> 00:07:29.976
in their actual pitches.

00:07:30.000 --> 00:07:33.016
So having ruled out this difference
on the entrepreneur's side,

00:07:33.040 --> 00:07:35.336
I then moved on to the investor's side,

00:07:35.360 --> 00:07:38.056
analyzing the six minutes of Q&A sessions

00:07:38.080 --> 00:07:41.320
that entrepreneurs engaged in
with the VCs after pitching.

00:07:42.440 --> 00:07:45.336
When examining the nearly 2,000 questions

00:07:45.360 --> 00:07:47.960
and corresponding answers
in these exchanges,

00:07:49.320 --> 00:07:52.456
both of my methods
showed significant support

00:07:52.480 --> 00:07:56.256
for the fact that male entrepreneurs
get asked promotion questions

00:07:56.280 --> 00:07:59.200
and female entrepreneurs
get asked prevention questions.

00:08:00.720 --> 00:08:05.176
In fact, a whopping 67 percent
of the questions posed

00:08:05.200 --> 00:08:07.696
to male entrepreneurs
were promotion-focused,

00:08:07.720 --> 00:08:12.320
while 66 percent of those posed to female
entrepreneurs were prevention-focused.

00:08:13.440 --> 00:08:15.176
What's especially interesting

00:08:15.200 --> 00:08:18.616
is that I expected female VCs

00:08:18.640 --> 00:08:21.320
to behave similarly to male VCs.

NOTE Paragraph

00:08:23.520 --> 00:08:27.736
Given its prevalence in the popular media
and the venture-funding literature,

00:08:27.760 --> 00:08:32.176
I expected the birds-of-a-feather
theory of homophily to hold here,

00:08:32.200 --> 00:08:35.456
meaning that male VCs
would favor male entrepreneurs

00:08:35.480 --> 00:08:37.216
with promotion questions

00:08:37.240 --> 00:08:40.880
and female VCs would do the same
for female entrepreneurs.

00:08:41.640 --> 00:08:47.056
But instead, all VCs displayed
the same implicit gender bias

00:08:47.080 --> 00:08:50.656
manifested in the regulatory focus
of the questions they posed

00:08:50.680 --> 00:08:52.720
to male versus female candidates.

00:08:53.400 --> 00:08:57.496
So female VCs asked
male entrepreneurs promotion questions

00:08:57.520 --> 00:09:01.176
and then turned around and asked
female entrepreneurs prevention questions

00:09:01.200 --> 00:09:02.840
just like the male VCs did.

NOTE Paragraph

00:09:04.360 --> 00:09:07.496
So given the fact
that both male and female VCs

00:09:07.520 --> 00:09:10.576
are displaying this implicit gender bias,

00:09:10.600 --> 00:09:14.480
what effect, if any, does this have
on start-up funding outcomes?

00:09:15.080 --> 00:09:18.560
My research shows
it has a significant effect.

00:09:19.320 --> 00:09:21.816
The regulatory focus of investor questions

00:09:21.840 --> 00:09:25.496
not only predicted
how well the start-ups would perform

00:09:25.520 --> 00:09:28.176
at the TechCrunch Disrupt competitions

00:09:28.200 --> 00:09:32.560
but also how much funding the start-ups
went on to raise in the open market.

00:09:33.840 --> 00:09:36.936
Those start-ups who were asked
predominantly promotion questions

00:09:36.960 --> 00:09:40.056
went on to raise
seven times as much funding

00:09:40.080 --> 00:09:42.040
as those asked prevention questions.

00:09:43.440 --> 00:09:45.056
But I didn't stop there.

00:09:45.080 --> 00:09:49.736
I then moved on to analyze entrepreneurs'
responses to those questions,

00:09:49.760 --> 00:09:53.696
and I found that entrepreneurs
are apt to respond in kind

00:09:53.720 --> 00:09:55.296
to the questions they get,

00:09:55.320 --> 00:09:58.456
meaning a promotion question
begets a promotion response

00:09:58.480 --> 00:10:01.616
and a prevention question
begets a prevention response.

00:10:01.640 --> 00:10:05.456
Now, this might make
intuitive sense to all of us here,

00:10:05.480 --> 00:10:09.880
but it has some unfortunate consequences
in this context of venture funding.

00:10:10.520 --> 00:10:12.056
So what ends up happening

00:10:12.080 --> 00:10:15.416
is that a male entrepreneur
gets asked a promotion question,

00:10:15.440 --> 00:10:19.096
granting him the luxury
to reinforce his association

00:10:19.120 --> 00:10:23.216
with the favorable domain
of gains by responding in kind,

00:10:23.240 --> 00:10:26.576
while a female entrepreneur
gets asked a prevention question

00:10:26.600 --> 00:10:29.376
and inadvertently
aggravates her association

00:10:29.400 --> 00:10:33.496
with the unfavorable domain
of losses by doing so.

00:10:33.520 --> 00:10:38.096
These responses then trigger venture
capitalists' subsequent biased questions,

00:10:38.120 --> 00:10:42.136
and the questions and answers
collectively fuel a cycle of bias

00:10:42.160 --> 00:10:45.176
that merely perpetuates
the gender disparity.

00:10:45.200 --> 00:10:46.760
Pretty depressing stuff, right?

NOTE Paragraph

00:10:47.960 --> 00:10:51.080
Well, fortunately, there's
a silver lining to my findings.

00:10:51.960 --> 00:10:55.576
Those plucky entrepreneurs
who managed to switch focus

00:10:55.600 --> 00:10:59.696
by responding to prevention questions
with promotion answers

00:10:59.720 --> 00:11:02.976
went on to raise 14 times more funding

00:11:03.000 --> 00:11:05.256
than those who responded
to prevention questions

00:11:05.280 --> 00:11:06.520
with prevention answers.

NOTE Paragraph

00:11:07.360 --> 00:11:10.336
So what this means
is that if you're asked a question

00:11:10.360 --> 00:11:13.136
about defending
your start-up's market share,

00:11:13.160 --> 00:11:15.816
you'd be better served
to frame your response

00:11:15.840 --> 00:11:19.736
around the size and growth potential
of the overall pie

00:11:19.760 --> 00:11:24.016
as opposed to how you merely plan
to protect your sliver of that pie.

00:11:24.040 --> 00:11:25.936
So if I get asked this question,

00:11:25.960 --> 00:11:27.216
I would say,

00:11:27.240 --> 00:11:29.976
"We're playing in such a large
and fast-growing market

00:11:30.000 --> 00:11:32.416
that's bound to attract new entrants.

00:11:32.440 --> 00:11:35.616
We plan to take
increasing share in this market

00:11:35.640 --> 00:11:38.216
by leveraging our start-up's
unique assets."

00:11:38.240 --> 00:11:43.600
I've thus subtly redirected this dialogue
into the favorable domain of gains.

NOTE Paragraph

00:11:44.520 --> 00:11:48.576
Now, these results are quite compelling
among start-ups that launched at TechCrunch

00:11:48.600 --> 00:11:52.576
but field data can merely tell us
that there's a correlational relationship

00:11:52.600 --> 00:11:55.016
between regulatory focus and funding.

00:11:55.040 --> 00:11:58.896
So I sought to see whether
this difference in regulatory focus

00:11:58.920 --> 00:12:01.536
can actually cause funding outcomes

00:12:01.560 --> 00:12:03.776
by running a controlled experiment

00:12:03.800 --> 00:12:06.160
on both angel investors
and ordinary people.

00:12:06.760 --> 00:12:09.536
Simulating the TechCrunch
Disrupt environment,

00:12:09.560 --> 00:12:13.816
I had participants listen
to four six-minute audio files

00:12:13.840 --> 00:12:16.536
of 10 question-and-answer exchanges

00:12:16.560 --> 00:12:19.816
that were manipulated
for promotion and prevention language,

00:12:19.840 --> 00:12:22.079
and then asked them
to allocate a sum of funding

00:12:22.103 --> 00:12:23.863
to each venture as they saw fit.

NOTE Paragraph

00:12:24.360 --> 00:12:28.976
Well, my experimental results
reinforced my findings from the field.

00:12:29.000 --> 00:12:32.976
Those scenarios where entrepreneurs
were asked promotion questions

00:12:33.000 --> 00:12:35.536
received twice the funding allocations

00:12:35.560 --> 00:12:38.440
of those where entrepreneurs
were asked prevention questions.

00:12:39.480 --> 00:12:41.456
What's especially promising

00:12:41.480 --> 00:12:44.896
is the fact that those scenarios
where entrepreneurs switched

00:12:44.920 --> 00:12:48.576
as opposed to matched focus
when they received prevention questions

00:12:48.600 --> 00:12:52.840
received significantly more funding
from both sets of participants.

NOTE Paragraph

00:12:54.000 --> 00:12:56.856
So to my female entrepreneurs out there,

00:12:56.880 --> 00:12:59.576
here are a couple
simple things you could do.

00:12:59.600 --> 00:13:02.976
The first is to recognize
the question you're being asked.

00:13:03.000 --> 00:13:05.776
Are you getting a prevention question?

00:13:05.800 --> 00:13:09.896
If this is the case, answer
the question at hand by all means,

00:13:09.920 --> 00:13:12.656
but merely frame
your response in promotion

00:13:12.680 --> 00:13:16.880
in an effort to garner higher amounts
of funding for your start-ups.

NOTE Paragraph

00:13:17.880 --> 00:13:20.336
The unfortunate reality, though,

00:13:20.360 --> 00:13:23.256
is that both men and women
evaluating start-ups

00:13:23.280 --> 00:13:26.856
display the same implicit
gender bias in their questioning,

00:13:26.880 --> 00:13:30.320
inadvertently favoring
male entrepreneurs over female ones.

00:13:31.360 --> 00:13:33.456
So to my investors out there,

00:13:33.480 --> 00:13:37.296
I would offer that you have
an opportunity here

00:13:37.320 --> 00:13:40.816
to approach Q&A sessions
more even-handedly,

00:13:40.840 --> 00:13:43.296
not just so that you
could do the right thing,

00:13:43.320 --> 00:13:47.280
but so that you can improve
the quality of your decision making.

00:13:48.480 --> 00:13:52.176
By flashing the same light
on every start-up's potential

00:13:52.200 --> 00:13:54.056
for gains and losses,

00:13:54.080 --> 00:13:57.016
you enable all deserving
start-ups to shine

00:13:57.040 --> 00:13:59.600
and you maximize returns in the process.

NOTE Paragraph

00:14:00.800 --> 00:14:03.936
Today, I get to be that girl

00:14:03.960 --> 00:14:05.936
walking by the pool,

00:14:05.960 --> 00:14:07.496
sounding the alarm

00:14:07.520 --> 00:14:10.120
that something is going on
beneath the surface.

00:14:11.040 --> 00:14:14.776
Together, we have the power
to break this cycle

00:14:14.800 --> 00:14:17.240
of implicit gender bias
in start-up funding.

00:14:18.200 --> 00:14:21.016
Let's give the most promising start-ups,

00:14:21.040 --> 00:14:24.376
regardless of whether
they're led by men or women,

00:14:24.400 --> 00:14:28.136
a fighting chance to grow and thrive.

NOTE Paragraph

00:14:28.160 --> 00:14:29.456
Thank you.

NOTE Paragraph

00:14:29.480 --> 00:14:35.040
(Applause)