WEBVTT

00:00:00.040 --> 00:00:02.390
The following content is
provided under a Creative

00:00:02.390 --> 00:00:03.680
Commons license.

00:00:03.680 --> 00:00:06.630
Your support will help MIT
OpenCourseWare continue to

00:00:06.630 --> 00:00:09.980
offer high quality educational
resources for free.

00:00:09.980 --> 00:00:12.820
To make a donation or view
additional materials from

00:00:12.820 --> 00:00:16.560
hundreds of MIT courses, visit
MIT OpenCourseWare at

00:00:16.560 --> 00:00:17.810
ocw.mit.edu.

00:00:20.635 --> 00:00:25.330
PROFESSOR: So we now have,
hopefully, I don't know.

00:00:25.330 --> 00:00:28.580
I've been going to some of the
groups and we seem to be there

00:00:28.580 --> 00:00:30.500
for at least one of the
groups I surveyed

00:00:30.500 --> 00:00:32.420
by the end of yesterday.

00:00:32.420 --> 00:00:34.420
We have a question.

00:00:34.420 --> 00:00:37.250
We know what we'd like to
evaluate, and we have some

00:00:37.250 --> 00:00:39.470
kind of a design.

00:00:39.470 --> 00:00:46.160
So we know how we are going
to get a sample.

00:00:46.160 --> 00:00:49.540
We know how we are going
to randomize.

00:00:49.540 --> 00:00:54.890
And that's a good start.

00:00:54.890 --> 00:00:59.700
So what we are going to do today
is getting to know that

00:00:59.700 --> 00:01:01.740
we have a way of randomizing.

00:01:01.740 --> 00:01:04.920
We're going to get into the
details of how we are actually

00:01:04.920 --> 00:01:08.120
going to go about collecting the
data for the evaluation.

00:01:08.120 --> 00:01:10.600
So it's not going to be so much
about the program, it's

00:01:10.600 --> 00:01:14.390
going to be about thinking
how we really

00:01:14.390 --> 00:01:15.880
want to do the survey.

00:01:15.880 --> 00:01:21.540
Thinking about what is the
sample size we need, how many

00:01:21.540 --> 00:01:26.980
data points we need, and what
data should we go about

00:01:26.980 --> 00:01:30.630
collecting, and how much is it
going to cost, and what are

00:01:30.630 --> 00:01:33.170
the trade-offs we are facing
when we are trying to address

00:01:33.170 --> 00:01:34.460
those questions.

00:01:34.460 --> 00:01:40.400
So we could be spending
from one week to one

00:01:40.400 --> 00:01:41.790
year on these questions.

00:01:41.790 --> 00:01:44.880
In particular, on the survey
design aspect.

00:01:44.880 --> 00:01:47.290
These are complicated
questions.

00:01:47.290 --> 00:01:50.530
Not many people who are much
more qualified, certainly,

00:01:50.530 --> 00:01:53.740
than me on this issue
of how to ask the

00:01:53.740 --> 00:01:55.260
right question correctly.

00:01:55.260 --> 00:01:58.820
So we are not going to
get into any of that.

00:01:58.820 --> 00:02:03.330
It's going to remain at a
somewhat abstract level of

00:02:03.330 --> 00:02:04.040
business one.

00:02:04.040 --> 00:02:07.360
It's going to be both abstract
and quite particular in a way.

00:02:07.360 --> 00:02:10.289
At the abstract level is how
do we even think about the

00:02:10.289 --> 00:02:12.000
type of data we want
to collect?

00:02:12.000 --> 00:02:15.180
And at the concrete level it's
we're going to do it for one

00:02:15.180 --> 00:02:19.280
particular example, which is
this Panchayat example which

00:02:19.280 --> 00:02:22.630
happens to be one that
I know pretty well.

00:02:22.630 --> 00:02:24.260
AUDIENCE: [UNINTELLIGIBLE]

00:02:24.260 --> 00:02:26.245
PROFESSOR: Not making
much progress, huh?

00:02:28.960 --> 00:02:36.440
So what are the questions we
need to know the answer to

00:02:36.440 --> 00:02:39.640
before getting ready to
start our evaluation?

00:02:39.640 --> 00:02:43.140
One is what data should
we collect?

00:02:43.140 --> 00:02:44.690
It's kind of a big question.

00:02:44.690 --> 00:02:46.930
What should go in the
questionnaire?

00:02:46.930 --> 00:02:49.530
Anybody among you who's
been involved in doing

00:02:49.530 --> 00:02:50.810
questionnaires knows
that you cannot

00:02:50.810 --> 00:02:53.240
ask a million questions.

00:02:53.240 --> 00:02:55.880
And so how do we choose
what to put in?

00:02:55.880 --> 00:03:00.580
And then, what systems we should
put in place to ensure

00:03:00.580 --> 00:03:02.380
that the quality of
the data is good?

00:03:05.690 --> 00:03:07.650
And what sample size
we need to plan.

00:03:07.650 --> 00:03:09.720
So not going to do sample
size this morning.

00:03:09.720 --> 00:03:16.640
It has its own little lecture
and a firm set of exercises.

00:03:16.640 --> 00:03:18.560
So that's going to go
this afternoon.

00:03:18.560 --> 00:03:21.280
We are going to be doing
a lot of that and

00:03:21.280 --> 00:03:24.400
a bit of that today.

00:03:24.400 --> 00:03:29.710
So let's start with the
Panchayat example.

00:03:29.710 --> 00:03:33.590
So let's talk a bit
about the setting.

00:03:33.590 --> 00:03:36.220
So the setting is a reservation
for women in the

00:03:36.220 --> 00:03:39.550
Panchayat region in India.

00:03:39.550 --> 00:03:48.110
So what's the Panchayat and
what's the reservation policy?

00:03:52.700 --> 00:03:54.300
So the rule of the
game, I don't

00:03:54.300 --> 00:03:56.540
ask rhetorical questions.

00:03:56.540 --> 00:03:59.910
So when I ask a question,
someone needs to answer

00:03:59.910 --> 00:04:00.550
otherwise we're stuck.

00:04:00.550 --> 00:04:03.793
AUDIENCE: It was a program
designed to decentralize the

00:04:03.793 --> 00:04:06.538
allocation of public resources
in villages

00:04:06.538 --> 00:04:08.534
to the lower level.

00:04:08.534 --> 00:04:11.650
And at the same time, you wanted
to ensure that meant

00:04:11.650 --> 00:04:16.000
minorities and scheduled
tribes in India got

00:04:16.000 --> 00:04:19.975
represented somehow, and the
preferences and opinions at

00:04:19.975 --> 00:04:22.420
this Panchayat level.

00:04:22.420 --> 00:04:23.380
PROFESSOR: That's right.

00:04:23.380 --> 00:04:28.390
So the policy is Panchayat
stands to

00:04:28.390 --> 00:04:30.460
consider five people.

00:04:30.460 --> 00:04:32.920
And maybe in some historical
India there was this council

00:04:32.920 --> 00:04:35.970
of five people who make
decisions for the villages.

00:04:35.970 --> 00:04:38.560
Or maybe not, but anyway,
that's where

00:04:38.560 --> 00:04:40.520
the name comes from.

00:04:40.520 --> 00:04:44.840
And the idea, and that's not
only India, it's something

00:04:44.840 --> 00:04:47.470
that you see in a lot of
developed and developing

00:04:47.470 --> 00:04:52.890
countries today, is that
decisions regarding local

00:04:52.890 --> 00:04:58.010
public goods such as drinking
water infrastructure, the

00:04:58.010 --> 00:05:00.340
local wards, the buildings--

00:05:00.340 --> 00:05:03.470
at least the buildings so, for
example, schools, health

00:05:03.470 --> 00:05:05.720
facilities and such things--

00:05:05.720 --> 00:05:12.310
would better represent people's
need if people had a

00:05:12.310 --> 00:05:14.690
say in what they wanted.

00:05:14.690 --> 00:05:16.770
So it's very difficult to know
if our bureaucrats sitting in

00:05:16.770 --> 00:05:20.280
Delhi that such and such village
in the middle of Bihar

00:05:20.280 --> 00:05:23.110
needs a road as opposed
to a water well.

00:05:23.110 --> 00:05:26.480
And so you can get this
vast misallocation of

00:05:26.480 --> 00:05:28.150
money in this way.

00:05:28.150 --> 00:05:33.410
So the idea with the Panchayat
is that now even though the

00:05:33.410 --> 00:05:36.960
revenue collection is still
pretty much not coming from

00:05:36.960 --> 00:05:40.120
the villages where there is
not much taxation ability,

00:05:40.120 --> 00:05:43.415
this one defending decision
should be taken increasingly

00:05:43.415 --> 00:05:48.790
at this local level to ensure
that better adequation, a

00:05:48.790 --> 00:05:52.830
better fit between what you
build and what people want.

00:05:52.830 --> 00:06:02.770
The problem with that, of
course, is that when you have

00:06:02.770 --> 00:06:05.660
local decision making it's
who is going to be

00:06:05.660 --> 00:06:07.950
controlling the power.

00:06:07.950 --> 00:06:10.680
So India, like many other
countries, perhaps even more

00:06:10.680 --> 00:06:14.200
than other countries, has this
history of these conditions

00:06:14.200 --> 00:06:16.230
again some particular group.

00:06:16.230 --> 00:06:18.820
So, for example, the scheduled
caste who are the former

00:06:18.820 --> 00:06:19.080
untouchable?

00:06:19.080 --> 00:06:22.410
The scheduled tribe who don't
even have a caste?

00:06:22.410 --> 00:06:29.450
And so the worries that there
are enough people, strong

00:06:29.450 --> 00:06:32.370
people, in these groups, that at
the national level they can

00:06:32.370 --> 00:06:37.210
organize to make sure that there
are policies that help

00:06:37.210 --> 00:06:38.980
these particular groups.

00:06:38.980 --> 00:06:39.930
Same thing for women.

00:06:39.930 --> 00:06:42.590
There definitely are very
strong women in India.

00:06:42.590 --> 00:06:48.890
But in any local villages, you
might not have a minority of

00:06:48.890 --> 00:06:54.780
scheduled casual tribal women
who are able to take

00:06:54.780 --> 00:06:57.240
responsibility to ensure that
their group is represented.

00:06:57.240 --> 00:06:59.910
So they might never be
represented, you might have

00:06:59.910 --> 00:07:02.370
some tyranny of the majority.

00:07:02.370 --> 00:07:06.660
So that's why the system also
introduced this reservation

00:07:06.660 --> 00:07:10.960
concept, which is to ensure
some representation.

00:07:10.960 --> 00:07:15.270
So how does a reservation
work?

00:07:15.270 --> 00:07:19.260
How did they put it in place?

00:07:24.170 --> 00:07:27.020
AUDIENCE: They reserved a
certain number of seats for

00:07:27.020 --> 00:07:30.281
women in each council.

00:07:30.281 --> 00:07:31.120
PROFESSOR: Right.

00:07:31.120 --> 00:07:35.710
So there is one reservation at
the level of the council.

00:07:35.710 --> 00:07:38.100
So we can say these are
various councils.

00:07:41.050 --> 00:07:44.570
Each council has about, say,
10-12 representatives which

00:07:44.570 --> 00:07:48.250
represent a population
of 10,000 to 12,000.

00:07:48.250 --> 00:07:50.960
And so the first reservation
is within each council.

00:07:50.960 --> 00:07:52.896
How many seats for women?

00:07:52.896 --> 00:07:53.730
AUDIENCE: One third.

00:07:53.730 --> 00:07:55.050
PROFESSOR: One third.

00:07:55.050 --> 00:07:59.130
So each council gets one
third for women.

00:07:59.130 --> 00:08:02.900
And what about SC and
ST What do they get?

00:08:02.900 --> 00:08:04.508
STUDENT: In proportion
to their population?

00:08:04.508 --> 00:08:06.480
PROFESSOR: In proportion
to their population.

00:08:06.480 --> 00:08:08.080
Not in that particular
village, but

00:08:08.080 --> 00:08:09.380
in the whole district.

00:08:09.380 --> 00:08:12.290
So, for example, in Birbhum,
there is about, I think, 30%

00:08:12.290 --> 00:08:13.810
of scheduled caste.

00:08:13.810 --> 00:08:19.910
So every village needs to have
30% of scheduled caste, even

00:08:19.910 --> 00:08:21.380
if they have much less.

00:08:21.380 --> 00:08:23.680
Except that if they have less
than five, they're exempted.

00:08:26.390 --> 00:08:29.020
If you have no scheduled tribe
in the village, you can not

00:08:29.020 --> 00:08:31.320
have a scheduled tribe
representative.

00:08:31.320 --> 00:08:33.870
So that's the first thing.

00:08:33.870 --> 00:08:39.570
So are we going to be able to
evaluate the impact of having

00:08:39.570 --> 00:08:46.050
this policy which has, within
each council, has one third of

00:08:46.050 --> 00:08:49.000
the seats to a woman.

00:08:49.000 --> 00:08:50.940
AUDIENCE: No, because
there's no control.

00:08:50.940 --> 00:08:51.690
PROFESSOR: Right.

00:08:51.690 --> 00:08:52.750
That's going to be
very difficult

00:08:52.750 --> 00:08:54.420
because there is no control.

00:08:54.420 --> 00:08:57.400
So certainly we won't be able to
look at the impact of this

00:08:57.400 --> 00:09:02.610
policy on what the council
does as a whole.

00:09:02.610 --> 00:09:06.640
You might ask whether the
particular village where she

00:09:06.640 --> 00:09:09.220
comes from, or segment of the
village where she comes from,

00:09:09.220 --> 00:09:11.270
gets different good, possibly.

00:09:11.270 --> 00:09:16.620
But certainly you're not going
to be able to compare the GPs

00:09:16.620 --> 00:09:18.170
since they are different.

00:09:18.170 --> 00:09:21.110
So what is the second layer of
the policy that's going to be

00:09:21.110 --> 00:09:22.690
more helpful?

00:09:22.690 --> 00:09:23.514
AUDIENCE: The Pradhan?

00:09:23.514 --> 00:09:25.490
PROFESSOR: At the level
of the Pradhan.

00:09:25.490 --> 00:09:28.015
And how does that work?

00:09:28.015 --> 00:09:29.795
AUDIENCE: It's like
a rotation?

00:09:29.795 --> 00:09:32.330
PROFESSOR: Right.

00:09:32.330 --> 00:09:36.768
So how does it work for any
particular election?

00:09:36.768 --> 00:09:38.895
AUDIENCE: It's like it gets
sorted or something?

00:09:38.895 --> 00:09:40.140
PROFESSOR: Right.

00:09:40.140 --> 00:09:46.340
So what the deal exactly,
is that they rank them.

00:09:46.340 --> 00:09:50.170
So these GPs have a serial
number, which is some number

00:09:50.170 --> 00:09:53.890
that they had forever
and ever.

00:09:53.890 --> 00:09:57.080
So they rank them by their
serial number and by

00:09:57.080 --> 00:09:58.840
geographic unit.

00:09:58.840 --> 00:10:03.310
So they first rank them by
block, and within each block

00:10:03.310 --> 00:10:05.980
they rank them by their
serial number.

00:10:05.980 --> 00:10:08.560
And then the constituent
three lists.

00:10:08.560 --> 00:10:14.540
One list of what they call
general, one list of SC, and

00:10:14.540 --> 00:10:18.120
one list of ST. So the general
list means it's not reserved

00:10:18.120 --> 00:10:20.360
for an SC Pradhan.

00:10:20.360 --> 00:10:23.900
Those SC lists means it's there
for a SC Pradhan, and ST

00:10:23.900 --> 00:10:25.610
means it's there for
a ST Pradhan.

00:10:25.610 --> 00:10:29.950
And the way I did this selection
is that they have to

00:10:29.950 --> 00:10:33.530
stay with a random number which
they use to say, so if

00:10:33.530 --> 00:10:35.750
you need to reserve it's
like a very long table.

00:10:35.750 --> 00:10:40.560
Because it tells you if you need
to reserve five GPs, you

00:10:40.560 --> 00:10:45.880
pick number 1, 15, 17,
and 21 for SC.

00:10:45.880 --> 00:10:48.840
So they do this list this way.

00:10:48.840 --> 00:10:53.570
And then this gives them three
lists, which are ordered by

00:10:53.570 --> 00:10:55.400
serial number.

00:10:55.400 --> 00:10:58.460
And in each list, they count
1-2-3, 1-2-3, 1-2-3, in the

00:10:58.460 --> 00:11:00.830
first election.

00:11:00.830 --> 00:11:03.715
All the GP number one
in all the lists got

00:11:03.715 --> 00:11:05.130
reserved for women.

00:11:05.130 --> 00:11:07.330
In the second election,
again this goes

00:11:07.330 --> 00:11:08.640
through the same process.

00:11:08.640 --> 00:11:11.400
And all the GP number two are
reserved, and the third

00:11:11.400 --> 00:11:14.020
election, all the GP number
three were reserved.

00:11:14.020 --> 00:11:16.960
So in principle there is a
rotation, though it is not

00:11:16.960 --> 00:11:21.330
necessarily one for one in the
sense that if a GP happens to

00:11:21.330 --> 00:11:26.210
be number one on this list in
the election in 2003, and

00:11:26.210 --> 00:11:28.450
number two in this list
in 2008, they get

00:11:28.450 --> 00:11:30.190
reserved twice in a row.

00:11:30.190 --> 00:11:34.110
Possibly if they become number
three in this list, they would

00:11:34.110 --> 00:11:35.290
do that three times in a row.

00:11:35.290 --> 00:11:37.240
That could happen, too.

00:11:37.240 --> 00:11:40.400
But twice in a row is pretty
frequent, actually.

00:11:40.400 --> 00:11:47.540
So this is more hopeful because
now we have some GPs

00:11:47.540 --> 00:11:48.890
that are reserved for women.

00:11:51.490 --> 00:11:55.000
That says those three are
reserved for women.

00:11:55.000 --> 00:11:57.620
And some who are not
in the same year.

00:11:57.620 --> 00:12:00.820
So now we can compare them.

00:12:00.820 --> 00:12:04.110
Compare the decisions that are
made in those places, compare

00:12:04.110 --> 00:12:07.090
anything we would like to
compare, and we have a chance

00:12:07.090 --> 00:12:09.060
to actually identify
the effects.

00:12:09.060 --> 00:12:09.470
Yeah.

00:12:09.470 --> 00:12:13.157
AUDIENCE: Can you compare if in
the non-reserve you elect a

00:12:13.157 --> 00:12:16.370
woman, does that become part
of the control group, or?

00:12:16.370 --> 00:12:20.830
PROFESSOR: So what's your
thinking on that?

00:12:20.830 --> 00:12:25.790
So two things could
happen, right?

00:12:25.790 --> 00:12:28.520
It could happen that in the
places that are reserved for

00:12:28.520 --> 00:12:31.510
women, they elect a man
anyway in defiance?

00:12:31.510 --> 00:12:33.040
So you don't have that
problem, actually.

00:12:33.040 --> 00:12:34.590
They elected a woman.

00:12:34.590 --> 00:12:36.400
And the other thing that could
happen is what you're talking

00:12:36.400 --> 00:12:41.710
about, is that some of those
places might decide to elect a

00:12:41.710 --> 00:12:42.870
woman anyway.

00:12:42.870 --> 00:12:45.960
In fact, about 7% of them do.

00:12:45.960 --> 00:12:49.012
So what do you think we
do with those guys?

00:12:49.012 --> 00:12:50.305
AUDIENCE: They're part
of the control.

00:12:50.305 --> 00:12:52.650
PROFESSOR: We keep them
in the control, right?

00:12:52.650 --> 00:12:55.900
Because we're going to spend
more time on that tomorrow,

00:12:55.900 --> 00:12:59.940
but if we did put them in the
treatment, then we don't have

00:12:59.940 --> 00:13:02.260
a random selection anymore
because now in the treatment

00:13:02.260 --> 00:13:04.050
we have the people who
are forced to have a

00:13:04.050 --> 00:13:05.410
woman that's on them.

00:13:05.410 --> 00:13:09.970
Plus the places that choose
that's not random.

00:13:09.970 --> 00:13:17.540
So we'll be keeping them here,
and this is going to mean that

00:13:17.540 --> 00:13:19.770
it's going to look a little
bit like an encouragement

00:13:19.770 --> 00:13:21.990
design like you saw yesterday.

00:13:21.990 --> 00:13:25.310
Which is, in the places that are
reserved for women, 100%

00:13:25.310 --> 00:13:27.720
of them have women.

00:13:27.720 --> 00:13:29.890
In the places that were
not reserved,

00:13:29.890 --> 00:13:32.200
7% have women anyway.

00:13:32.200 --> 00:13:35.540
So the difference is
not quite 100%.

00:13:35.540 --> 00:13:39.260
So tomorrow we'll deal with,
see, so here we are going to

00:13:39.260 --> 00:13:42.890
focus on the impact of being
reserved, and we're not going

00:13:42.890 --> 00:13:47.020
to be concerned so much about
the impact of having a woman.

00:13:47.020 --> 00:13:49.460
So we are still going to always
compare treatment

00:13:49.460 --> 00:13:50.460
versus control.

00:13:50.460 --> 00:13:53.680
Tomorrow you can see how, in
fact, we can do something

00:13:53.680 --> 00:13:57.080
about thinking about how we go
from the impact of being

00:13:57.080 --> 00:13:59.480
reserved, to the impact of
being a woman if we are

00:13:59.480 --> 00:14:03.970
willing to make a couple
more assumptions.

00:14:03.970 --> 00:14:05.920
So that's a better setting.

00:14:05.920 --> 00:14:12.440
We can now compare those places
and even though the

00:14:12.440 --> 00:14:15.860
Indian government didn't do this
beautiful experiment with

00:14:15.860 --> 00:14:18.460
the view of an evaluation
in mind.

00:14:18.460 --> 00:14:19.650
And they might have.

00:14:19.650 --> 00:14:21.740
And then they might contact
you and say, hey, we

00:14:21.740 --> 00:14:23.240
have this set up.

00:14:23.240 --> 00:14:28.060
How should we go about
evaluating it?

00:14:28.060 --> 00:14:29.430
And so that's the name
of the game.

00:14:32.570 --> 00:14:35.070
So I think we went through
this maybe?

00:14:35.070 --> 00:14:40.620
We can spend a little more
time on the goals.

00:14:40.620 --> 00:14:47.410
So what do we expect from
the Panchayati raj?

00:14:50.080 --> 00:14:56.210
What will give us a first sense
of what data we should

00:14:56.210 --> 00:14:59.990
think about collecting to know
what's the goal of the

00:14:59.990 --> 00:15:01.240
institution.

00:15:05.960 --> 00:15:09.350
This big untraditional amendment
took place in 1993.

00:15:09.350 --> 00:15:14.624
What was in the mind of the
policy makers at that time?

00:15:14.624 --> 00:15:19.096
AUDIENCE: That there would be
more accountability to the

00:15:19.096 --> 00:15:20.984
interest in priorities of
interested countries?

00:15:20.984 --> 00:15:23.988
PROFESSOR: Right.

00:15:23.988 --> 00:15:26.398
AUDIENCE: Greater women's
empowerment?

00:15:26.398 --> 00:15:31.550
PROFESSOR: So I don't know if
it's a goal of the Panchayati

00:15:31.550 --> 00:15:35.575
raj amendment per se, but we
could think it was something

00:15:35.575 --> 00:15:38.354
people had in mind in the
reservation policies.

00:15:38.354 --> 00:15:40.018
AUDIENCE: That program selection
would be more

00:15:40.018 --> 00:15:41.080
aligned with the local
preferences?

00:15:41.080 --> 00:15:42.140
PROFESSOR: Right.

00:15:42.140 --> 00:15:44.860
Program selection might reflect
local preferences.

00:15:44.860 --> 00:15:45.745
We should keep that in mind.

00:15:45.745 --> 00:15:48.189
AUDIENCE: To increase
the quality and

00:15:48.189 --> 00:15:48.890
quantity of public goods?

00:15:48.890 --> 00:15:49.285
PROFESSOR: Right.

00:15:49.285 --> 00:15:51.910
We can see that there is more
as a result of their

00:15:51.910 --> 00:15:55.620
accountability, there might be
better and more public goods.

00:15:58.562 --> 00:16:00.745
AUDIENCE: More participation
in public life?

00:16:00.745 --> 00:16:01.340
PROFESSOR: Right.

00:16:01.340 --> 00:16:05.550
We might also think that we
might value the participation

00:16:05.550 --> 00:16:07.530
for its own good.

00:16:07.530 --> 00:16:09.850
I would think it's a good thing
if people are involved

00:16:09.850 --> 00:16:10.865
in local democracy.

00:16:10.865 --> 00:16:11.956
AUDIENCE: They vote more?

00:16:11.956 --> 00:16:14.280
PROFESSOR: They vote more,
participate in meetings,

00:16:14.280 --> 00:16:15.680
things like that.

00:16:15.680 --> 00:16:19.440
So that's kind of the sum of the
objective of the Panchayat

00:16:19.440 --> 00:16:21.010
raj, generally.

00:16:21.010 --> 00:16:24.680
And within that there was
this reservation.

00:16:24.680 --> 00:16:31.610
And the reservation policy is,
was, still is in a sense,

00:16:31.610 --> 00:16:32.860
quite controversial.

00:16:35.210 --> 00:16:37.990
It is still controversial in the
sense that the next step

00:16:37.990 --> 00:16:41.300
they are asking themselves now,
and in fact I think has

00:16:41.300 --> 00:16:43.510
always been somewhat lurking in
the background, is should

00:16:43.510 --> 00:16:46.130
there be reservations
at other levels.

00:16:46.130 --> 00:16:48.850
For example, for
foreign peace.

00:16:48.850 --> 00:16:51.280
Like, should there be
reservations in the parliament

00:16:51.280 --> 00:16:55.760
modeled on the one
[UNINTELLIGIBLE] model, but

00:16:55.760 --> 00:16:57.760
akin to what we have
on the Panchayat.

00:17:00.410 --> 00:17:03.100
And that debate is
lively in India,

00:17:03.100 --> 00:17:09.170
but it's also elsewhere.

00:17:09.170 --> 00:17:11.950
Many countries have reservation
policies.

00:17:11.950 --> 00:17:15.349
Last I counted, hundreds of
some kind of some form of

00:17:15.349 --> 00:17:17.780
mandated representation
for women.

00:17:17.780 --> 00:17:24.099
So it is not a very prominent
policy, and some countries are

00:17:24.099 --> 00:17:25.859
very opposed to that.

00:17:25.859 --> 00:17:27.770
For example, the US.

00:17:27.770 --> 00:17:36.810
So what are the potential
pros and cons of

00:17:36.810 --> 00:17:40.040
a reservation policy?

00:17:40.040 --> 00:17:42.970
Since our evaluation is meant
in a sense to inform this

00:17:42.970 --> 00:17:47.030
debate, we should think about
what are the questions that

00:17:47.030 --> 00:17:50.285
are in people's minds?

00:17:59.100 --> 00:18:00.760
AUDIENCE: It's not democratic.

00:18:00.760 --> 00:18:02.370
It's not majority rule.

00:18:02.370 --> 00:18:02.900
PROFESSOR: Right.

00:18:02.900 --> 00:18:04.510
So that's a disadvantage.

00:18:15.820 --> 00:18:18.830
So it's not democratic, I'll
just put it that way.

00:18:18.830 --> 00:18:22.020
We are constraining
people's choices.

00:18:22.020 --> 00:18:25.000
They might not like it.

00:18:25.000 --> 00:18:27.833
So that reduces their utility.

00:18:27.833 --> 00:18:31.596
AUDIENCE: But there is this
structural inequality, so you

00:18:31.596 --> 00:18:34.530
need to empower your
women [INAUDIBLE]

00:18:34.530 --> 00:18:36.490
do it, [INAUDIBLE]

00:18:36.490 --> 00:18:38.420
this reservation
[UNINTELLIGIBLE].

00:18:38.420 --> 00:18:41.622
But at some point in history,
you need to do it to push

00:18:41.622 --> 00:18:43.005
women farther.

00:18:43.005 --> 00:18:46.820
PROFESSOR: So I see more than
one thing in that comment.

00:18:46.820 --> 00:18:51.270
One is that in the short run
we need to ensure woman's

00:18:51.270 --> 00:18:52.520
representation.

00:18:57.690 --> 00:18:59.460
That's the short run.

00:18:59.460 --> 00:19:03.504
And I think what I also hear
in your comment is that--

00:19:03.504 --> 00:19:05.722
AUDIENCE: So many noises
in the room.

00:19:05.722 --> 00:19:07.330
It's not working as it should.

00:19:07.330 --> 00:19:08.070
PROFESSOR: Right.

00:19:08.070 --> 00:19:10.650
So one of the reasons why you
might want to ensure woman's

00:19:10.650 --> 00:19:12.700
participation is
just because--

00:19:12.700 --> 00:19:13.470
just like that.

00:19:13.470 --> 00:19:17.030
Not even because you think it
will affect, again, like

00:19:17.030 --> 00:19:20.170
democracy here is an
end in itself.

00:19:20.170 --> 00:19:22.280
You might think, well, democracy
is not real because

00:19:22.280 --> 00:19:24.040
women are not participating.

00:19:24.040 --> 00:19:27.050
I'm interested in having women
participating, period.

00:19:27.050 --> 00:19:29.360
Even if it doesn't change
anything to the outcome,

00:19:29.360 --> 00:19:30.740
that's what makes
democracy real.

00:19:30.740 --> 00:19:32.430
So that could be one outcome.

00:19:32.430 --> 00:19:36.600
So we are interested in that
as an end in itself.

00:19:36.600 --> 00:19:40.460
And I think one thing that I
overhear in your comment that

00:19:40.460 --> 00:19:44.400
might not be there in reality,
is also that maybe that will

00:19:44.400 --> 00:19:47.440
help jump-start the process
where that would not be needed

00:19:47.440 --> 00:19:48.960
in the long run.

00:19:48.960 --> 00:19:51.830
So it would ensure a woman's
representation in the long

00:19:51.830 --> 00:19:56.195
run, and maybe change
voter's view.

00:20:00.090 --> 00:20:07.290
View is such that in the long
run, the effect would persist.

00:20:14.420 --> 00:20:19.990
In which case you might go to a
bigger and better democracy

00:20:19.990 --> 00:20:21.850
where everybody's actually
involved.

00:20:21.850 --> 00:20:25.620
In fact, it might be good
for men as well to avail

00:20:25.620 --> 00:20:28.030
themselves of the possibility
of half the population's

00:20:28.030 --> 00:20:33.640
talents which now, for some
reason, becomes shut.

00:20:33.640 --> 00:20:37.360
So that's a first set of things
which concerns the

00:20:37.360 --> 00:20:39.000
participation itself.

00:20:39.000 --> 00:20:41.485
AUDIENCE: So one of the
advantages of ensuring women's

00:20:41.485 --> 00:20:47.950
representation is that they
might then allocate more money

00:20:47.950 --> 00:20:52.270
to goods or services that
benefit women's [INAUDIBLE].

00:20:52.270 --> 00:20:53.790
PROFESSOR: So that is certainly

00:20:53.790 --> 00:20:56.930
something you hear a lot.

00:20:56.930 --> 00:21:05.190
So we allocate more money to
goods and services which,

00:21:05.190 --> 00:21:07.740
sorry, can you repeat the end?

00:21:07.740 --> 00:21:09.996
AUDIENCE: Which benefit women
in [UNINTELLIGIBLE].

00:21:09.996 --> 00:21:11.710
PROFESSOR: Which
benefit women.

00:21:15.020 --> 00:21:17.350
Which might be good if--

00:21:17.350 --> 00:21:19.820
you were saying that woman
were not represented

00:21:19.820 --> 00:21:24.436
adequately before, and all--

00:21:24.436 --> 00:21:24.912
yeah.

00:21:24.912 --> 00:21:26.578
AUDIENCE: And then there might
also, I mean, there's been

00:21:26.578 --> 00:21:29.160
some literature saying that
there's been a spillover

00:21:29.160 --> 00:21:31.605
effect for children and
families, but if [INAUDIBLE].

00:21:40.896 --> 00:21:43.200
PROFESSOR: That's also something
you hear a lot which

00:21:43.200 --> 00:21:47.050
is, so this is an argument in
favor of redistribution.

00:21:47.050 --> 00:21:49.280
To half the population there
is a group that doesn't get

00:21:49.280 --> 00:21:52.150
very much in the
normal system.

00:21:52.150 --> 00:21:54.660
They need to get some part
of the rent as well.

00:21:54.660 --> 00:21:56.830
And then that's sort of one
step further which is an

00:21:56.830 --> 00:21:58.050
efficiency argument.

00:21:58.050 --> 00:22:01.540
You say, as you do that,
you will also get a

00:22:01.540 --> 00:22:03.780
different type of goods.

00:22:03.780 --> 00:22:07.740
STUDENT: Increases a woman's
empowerment.

00:22:07.740 --> 00:22:08.540
PROFESSOR: Right.

00:22:08.540 --> 00:22:16.450
So as you do that you might
also increase a woman's

00:22:16.450 --> 00:22:17.920
empowerment.

00:22:17.920 --> 00:22:19.980
I think it's a double
positive, but I'll

00:22:19.980 --> 00:22:21.340
leave it like that.

00:22:21.340 --> 00:22:22.590
Empowerment.

00:22:25.380 --> 00:22:31.230
Which may be good again as a
redistribution or because we

00:22:31.230 --> 00:22:35.426
think that women in the house
will do things differently.

00:22:35.426 --> 00:22:38.300
AUDIENCE: Maybe the same thing
as empowerment, but maybe more

00:22:38.300 --> 00:22:41.950
measurable is that it claims
to reverse the trend of

00:22:41.950 --> 00:22:44.645
increasing gender gap
in the population?

00:22:44.645 --> 00:22:47.100
PROFESSOR: So one of the
possible outcomes would be

00:22:47.100 --> 00:22:48.380
gender gap.

00:22:48.380 --> 00:22:50.400
You mean among children
or among adults?

00:22:50.400 --> 00:22:53.960
AUDIENCE: Eventually it will
be among the adults, but it

00:22:53.960 --> 00:22:56.250
can be noticed [INAUDIBLE]

00:22:56.250 --> 00:22:59.700
indicators would be like
reduction in girl infanticide.

00:23:06.765 --> 00:23:10.050
PROFESSOR: So do we think there
is a direct rule between

00:23:10.050 --> 00:23:15.130
what women policy makers can do,
in terms of what type of

00:23:15.130 --> 00:23:16.510
public goods they can provide?

00:23:16.510 --> 00:23:20.960
Or do you think it's going to be
all indirect to doing stuff

00:23:20.960 --> 00:23:23.540
that benefits women, which
increases the power of women,

00:23:23.540 --> 00:23:27.495
which means that they make
decisions that are better for

00:23:27.495 --> 00:23:29.462
girls in the household?

00:23:29.462 --> 00:23:31.366
AUDIENCE: And perception,
it will quickly

00:23:31.366 --> 00:23:32.990
change in our society.

00:23:39.062 --> 00:23:41.600
PROFESSOR: So there is an issue
of perception as well.

00:23:46.590 --> 00:23:49.584
AUDIENCE: [INAUDIBLE]

00:23:49.584 --> 00:23:55.516
that when you see [INAUDIBLE]

00:23:58.492 --> 00:24:02.590
it would be more
[UNINTELLIGIBLE] to follow.

00:24:02.590 --> 00:24:03.120
PROFESSOR: Right.

00:24:03.120 --> 00:24:05.790
So you might think that might
be something that people are

00:24:05.790 --> 00:24:08.950
also talked about is just,
again, not just the aspect of

00:24:08.950 --> 00:24:10.420
what she might be doing.

00:24:10.420 --> 00:24:14.150
Just having the woman as the
figurehead changes the image

00:24:14.150 --> 00:24:15.860
in a somewhat more
permanent way.

00:24:15.860 --> 00:24:19.580
You don't just think well, she's
there because of the

00:24:19.580 --> 00:24:20.130
reservation.

00:24:20.130 --> 00:24:23.560
You see a woman in a position of
power and that might change

00:24:23.560 --> 00:24:26.105
your view of what's possible.

00:24:26.105 --> 00:24:29.240
AUDIENCE: As a disadvantage, you
also have the potential of

00:24:29.240 --> 00:24:32.540
the women who have poor job and
so the perception actually

00:24:32.540 --> 00:24:36.747
becomes negative, if it's
perceived that the woman might

00:24:36.747 --> 00:24:38.480
be unqualified.

00:24:38.480 --> 00:24:43.140
PROFESSOR: So the first woman
might be unqualified, and that

00:24:43.140 --> 00:24:45.440
would worsen perception.

00:24:45.440 --> 00:24:49.520
And then the other thing that
your comment could be read as

00:24:49.520 --> 00:24:51.580
is maybe they're not
particularly unqualified, but

00:24:51.580 --> 00:24:53.850
they are perceived as
being unqualified.

00:24:53.850 --> 00:24:56.950
Because they are perceived as
being there because of the

00:24:56.950 --> 00:24:58.190
reservation policy.

00:24:58.190 --> 00:25:01.500
And, in fact, it's makes it
even more difficult for

00:25:01.500 --> 00:25:03.040
competent women to assert
themselves.

00:25:03.040 --> 00:25:05.350
That's something that has been
said a lot about affirmative

00:25:05.350 --> 00:25:07.920
action in the US.

00:25:07.920 --> 00:25:13.920
For blacks, for example, now
every time you see a black

00:25:13.920 --> 00:25:15.520
person succeeding you're
thinking, why?

00:25:15.520 --> 00:25:19.090
It's because they got some
unfair advantage somewhere,

00:25:19.090 --> 00:25:21.590
and so that makes
things worse.

00:25:21.590 --> 00:25:22.410
AUDIENCE: Backlash.

00:25:22.410 --> 00:25:25.730
PROFESSOR: So we can write
it as backlash.

00:25:25.730 --> 00:25:27.050
That's a very good point.

00:25:29.610 --> 00:25:32.010
So that's a question that
we can also ask.

00:25:32.010 --> 00:25:34.330
What's the perception of women,
and women in power

00:25:34.330 --> 00:25:35.706
before and after?

00:25:35.706 --> 00:25:38.090
AUDIENCE: We also create
this backlash.

00:25:38.090 --> 00:25:41.454
It's my perception of it, too,
but it's a different issue.

00:25:41.454 --> 00:25:44.328
It's not necessarily clear
that women will, in fact,

00:25:44.328 --> 00:25:45.786
represent women better.

00:25:45.786 --> 00:25:48.702
It is possible that women will
feel the need to compensate

00:25:48.702 --> 00:25:53.070
and therefore go out of their
way to identify with men's

00:25:53.070 --> 00:25:56.184
issues, and go out of their way
to not be perceived as a

00:25:56.184 --> 00:25:57.112
feminine candidate.

00:25:57.112 --> 00:25:57.970
PROFESSOR: Right.

00:25:57.970 --> 00:26:05.925
So I'm looking for a term for
that but I don't find one.

00:26:09.410 --> 00:26:15.790
Women may overcompensate
and not

00:26:15.790 --> 00:26:17.040
represent women's interests.

00:26:26.138 --> 00:26:31.217
AUDIENCE: As an advantage having
women on the councils

00:26:31.217 --> 00:26:35.720
that you bring more information
that may benefit

00:26:35.720 --> 00:26:37.960
women but also may just benefit
the entire society.

00:26:37.960 --> 00:26:40.340
Because you weren't getting this
channel of information

00:26:40.340 --> 00:26:42.932
either directly from their
communities, or from their

00:26:42.932 --> 00:26:45.528
constituents who don't feel
comfortable with women

00:26:45.528 --> 00:26:46.330
[INAUDIBLE].

00:26:46.330 --> 00:26:46.880
PROFESSOR: Right.

00:26:46.880 --> 00:26:48.140
You may be more information.

00:26:48.140 --> 00:26:52.020
For example, because women speak
up more, and so they

00:26:52.020 --> 00:26:55.140
give their opinion on stuff and
there is stuff they might

00:26:55.140 --> 00:26:57.400
just know better about, even
if they don't particularly

00:26:57.400 --> 00:26:59.500
care more about.

00:26:59.500 --> 00:27:02.310
Everybody may have the same
preferences so that might not

00:27:02.310 --> 00:27:03.150
be a conflict.

00:27:03.150 --> 00:27:06.010
If you never hear from the women
that the water well is

00:27:06.010 --> 00:27:09.920
blocked, then you may never
think of fixing it even though

00:27:09.920 --> 00:27:12.035
that's something that would
benefit everybody ultimately.

00:27:16.685 --> 00:27:19.748
AUDIENCE: An advantage would be
building capacity in women

00:27:19.748 --> 00:27:22.088
for leadership for the
higher levels.

00:27:22.088 --> 00:27:22.910
PROFESSOR: Right.

00:27:22.910 --> 00:27:29.380
So we are now talking about
spillover over time.

00:27:29.380 --> 00:27:33.120
That you have one woman one
time, and people understand

00:27:33.120 --> 00:27:36.640
that they are good and then
they can continue on.

00:27:36.640 --> 00:27:41.280
But it can also be, of course,
different levels while you're

00:27:41.280 --> 00:27:48.550
building up a cadre
of powerful women.

00:27:53.350 --> 00:27:55.938
AUDIENCE: I would think that you
could maybe get an income

00:27:55.938 --> 00:27:59.120
of that, right away, on women.

00:27:59.120 --> 00:28:02.220
We vote next year to put
in a water well.

00:28:02.220 --> 00:28:05.060
That means that there's a lot
more time to do my trading

00:28:05.060 --> 00:28:08.175
business, [UNINTELLIGIBLE]
business, so that right away I

00:28:08.175 --> 00:28:09.430
might make more money.

00:28:09.430 --> 00:28:10.070
PROFESSOR: Right.

00:28:10.070 --> 00:28:13.770
So one of the ways we can put it
into women empowerment, I'm

00:28:13.770 --> 00:28:17.900
going to add "to," so it
could be to income.

00:28:17.900 --> 00:28:20.810
It could be to time that you
generate, because now you

00:28:20.810 --> 00:28:22.340
spend less time collecting
water.

00:28:22.340 --> 00:28:25.730
Or it could be to the role
models as already discussed.

00:28:25.730 --> 00:28:27.950
But just the fact that I have
the public good that is

00:28:27.950 --> 00:28:30.440
convenient for me,
that I need.

00:28:30.440 --> 00:28:33.780
And presumably if women's power
increased, they're going

00:28:33.780 --> 00:28:37.710
to do some kind of public good
that's good for them which

00:28:37.710 --> 00:28:40.680
might be these types of things
that will free times for them.

00:28:43.680 --> 00:28:46.180
AUDIENCE: Kind of a counterpoint
to the last

00:28:46.180 --> 00:28:48.426
disadvantage.

00:28:48.426 --> 00:28:51.528
It's also possible that because
a woman doesn't have

00:28:51.528 --> 00:28:55.578
to compete against a man, they
don't have to pander to men,

00:28:55.578 --> 00:29:00.542
and in a sense, overcompensate
to men's preferences as much.

00:29:00.542 --> 00:29:01.310
PROFESSOR: Right.

00:29:01.310 --> 00:29:03.976
So in a reservation--

00:29:03.976 --> 00:29:07.584
we have too many advantages.

00:29:07.584 --> 00:29:11.000
AUDIENCE: You add to
disadvantage also which is

00:29:11.000 --> 00:29:15.990
that, I think, this may be more
indirect in the sense

00:29:15.990 --> 00:29:18.770
that when you're electing
someone on the premise that

00:29:18.770 --> 00:29:22.420
they'll able to deliver public
goods to a specific

00:29:22.420 --> 00:29:30.470
constituency, you risk perhaps
instituting a culture of

00:29:30.470 --> 00:29:34.990
paternalistic government that is
elected just to deliver the

00:29:34.990 --> 00:29:36.710
pork to a specific
constituency.

00:29:36.710 --> 00:29:37.470
PROFESSOR: Right.

00:29:37.470 --> 00:29:39.610
And that is something that
has been discussed a lot.

00:29:39.610 --> 00:29:42.850
For example, with the scheduled
caste is that you

00:29:42.850 --> 00:29:45.250
could have this shifting.

00:29:45.250 --> 00:29:48.870
So you go it's your turn, it's
our turn, it's your turn, it's

00:29:48.870 --> 00:29:54.600
our turn, and then people are
not really watching the

00:29:54.600 --> 00:29:58.170
scheduled caste guy because they
understand this is the

00:29:58.170 --> 00:30:01.790
scheduled caste guy's
time to take their

00:30:01.790 --> 00:30:03.640
turn, and vice versa.

00:30:03.640 --> 00:30:05.960
It might also go the other
way in the sense that the

00:30:05.960 --> 00:30:09.610
scheduled caste people
who were previously

00:30:09.610 --> 00:30:12.390
disenfranchised might now feel,
well, we might as well

00:30:12.390 --> 00:30:15.500
get this guy to deliver really
well while he is

00:30:15.500 --> 00:30:17.150
in power for us.

00:30:17.150 --> 00:30:18.360
Same thing for women.

00:30:18.360 --> 00:30:19.900
But the effect on
accountability

00:30:19.900 --> 00:30:21.060
is it's a good point.

00:30:21.060 --> 00:30:23.260
Because since the whole
Panchayat is about

00:30:23.260 --> 00:30:25.840
accountability to the people,
the effect of reservation

00:30:25.840 --> 00:30:29.660
system on accountability,
at best, ambiguous.

00:30:29.660 --> 00:30:32.715
So we can say this is related
to this first point which is

00:30:32.715 --> 00:30:34.600
it's not democratic anymore.

00:30:34.600 --> 00:30:39.230
So any tampering with the
democracy you might have story

00:30:39.230 --> 00:30:41.190
going either way.

00:30:41.190 --> 00:30:47.620
But that's a disadvantage
showing, effect on

00:30:47.620 --> 00:30:48.870
accountability.

00:30:55.220 --> 00:30:57.210
You're now accountable
to nobody.

00:30:57.210 --> 00:30:59.180
First, you are a lame duck.

00:30:59.180 --> 00:31:02.680
There is a very poor chance that
you get reelected again,

00:31:02.680 --> 00:31:04.170
so that's your point here.

00:31:04.170 --> 00:31:06.440
Which in a sense makes you free
to pander to your guys

00:31:06.440 --> 00:31:09.240
which is maybe what we wanted in
this case, but on the other

00:31:09.240 --> 00:31:13.720
hand, makes you maybe less
likely to deliver.

00:31:13.720 --> 00:31:14.838
Sorry, there was a--

00:31:14.838 --> 00:31:17.766
AUDIENCE: Another potential
disadvantage which is kind of

00:31:17.766 --> 00:31:20.478
still under whether or
not women are in fact

00:31:20.478 --> 00:31:23.510
representing, is that if there's
a culture of women not

00:31:23.510 --> 00:31:26.720
speaking up or not advocating,
then you might end up with no

00:31:26.720 --> 00:31:28.070
representation at all.

00:31:28.070 --> 00:31:28.870
PROFESSOR: Right.

00:31:28.870 --> 00:31:31.900
So there is a question
of who is in charge.

00:31:31.900 --> 00:31:36.420
So it might be that when it's an
election you elect someone,

00:31:36.420 --> 00:31:37.600
they're in charge.

00:31:37.600 --> 00:31:41.970
Here it's like maybe no woman
really wants to run, so you

00:31:41.970 --> 00:31:45.400
pick up any figurehead.

00:31:45.400 --> 00:31:48.750
It might go back to
elite control.

00:31:48.750 --> 00:31:54.720
If nobody who's democratically
elected is in a position to

00:31:54.720 --> 00:31:57.350
exercise the power, whoever
is a natural leader

00:31:57.350 --> 00:31:58.180
will take it back.

00:31:58.180 --> 00:32:00.100
It might be her husband, it
might be anybody else.

00:32:04.380 --> 00:32:07.940
It's a very pretty reasonably
recent effort that local

00:32:07.940 --> 00:32:09.560
democracy at that level.

00:32:09.560 --> 00:32:15.070
And it might be crazy to go
back to elite control.

00:32:15.070 --> 00:32:17.870
So who is in charge?

00:32:17.870 --> 00:32:20.250
And is there a risk
of elite control?

00:32:20.250 --> 00:32:25.479
Which is of course not why we
did this in the first place.

00:32:25.479 --> 00:32:28.972
AUDIENCE: I guess it builds
upon the last point maybe

00:32:28.972 --> 00:32:31.966
overall less efficiency in elite
administration functions

00:32:31.966 --> 00:32:34.128
because there may be resentment
against the women,

00:32:34.128 --> 00:32:35.958
so less cooperation
against them.

00:32:35.958 --> 00:32:37.954
So maybe there's more stalemate

00:32:37.954 --> 00:32:39.451
policies being passed.

00:32:39.451 --> 00:32:40.470
PROFESSOR: Yeah.

00:32:40.470 --> 00:32:42.300
So that's a risk of stalemate.

00:32:47.927 --> 00:32:51.270
So unless you have a burning
point, then we stop here

00:32:51.270 --> 00:32:53.140
because I have no space
anymore on the board.

00:32:53.140 --> 00:32:56.720
So that's it, that's budget
constraint for thoughts.

00:32:59.400 --> 00:33:07.280
So there is kind of a lot of
ideas and so somehow we are

00:33:07.280 --> 00:33:12.020
going to want them organized
in order to go

00:33:12.020 --> 00:33:14.130
and collect our data.

00:33:14.130 --> 00:33:17.020
So one thing we could think
of doing is that, oh let's

00:33:17.020 --> 00:33:17.835
postpone the big

00:33:17.835 --> 00:33:18.790
deorganization of the thoughts.

00:33:18.790 --> 00:33:23.620
We can see that it can go all
over the place, so let's go

00:33:23.620 --> 00:33:40.670
and collect a bunch of outcomes
and see how it goes.

00:33:40.670 --> 00:33:47.800
So given all of what we have
discussed here, what are the

00:33:47.800 --> 00:33:53.096
possible things that might be
affected by this policy?

00:33:53.096 --> 00:33:55.520
I'd like to have a board.

00:33:55.520 --> 00:33:58.200
I can just write.

00:33:58.200 --> 00:34:01.810
Given all the discussion we had,
what are the things that

00:34:01.810 --> 00:34:03.750
we think might be affected
by having

00:34:03.750 --> 00:34:07.380
this reservation policy?

00:34:07.380 --> 00:34:09.659
A lot of these things already
came in, but we can make a

00:34:09.659 --> 00:34:10.699
little list for ourselves.

00:34:10.699 --> 00:34:12.924
AUDIENCE: Choice of which
public goods--

00:34:12.924 --> 00:34:15.810
PROFESSOR: So one is definitely
the public goods.

00:34:15.810 --> 00:34:20.460
So the disadvantage
I'm going to--

00:34:20.460 --> 00:34:22.449
so outcomes--

00:34:22.449 --> 00:34:25.670
oh, wow.

00:34:25.670 --> 00:34:28.670
I don't usually use boards,
so I'm going to--

00:34:28.670 --> 00:34:31.400
ta da.

00:34:31.400 --> 00:34:32.650
That has to be MIT.

00:34:34.460 --> 00:34:36.530
So we have a whole
[UNINTELLIGIBLE], so one is

00:34:36.530 --> 00:34:41.270
clearly public good,
and potentially we

00:34:41.270 --> 00:34:42.159
have lots of them.

00:34:42.159 --> 00:34:45.147
What are the public goods that
we can see in villages?

00:34:45.147 --> 00:34:45.901
AUDIENCE: Water.

00:34:45.901 --> 00:34:46.530
PROFESSOR: Water.

00:34:46.530 --> 00:34:47.205
AUDIENCE: Roads.

00:34:47.205 --> 00:34:48.019
PROFESSOR: Roads.

00:34:48.019 --> 00:34:48.770
AUDIENCE: Schools.

00:34:48.770 --> 00:34:49.690
PROFESSOR: Schools.

00:34:49.690 --> 00:34:50.520
AUDIENCE: Hospitals.

00:34:50.520 --> 00:34:51.350
AUDIENCE: Health centers.

00:34:51.350 --> 00:34:52.989
PROFESSOR: Small health
centers, yeah.

00:34:52.989 --> 00:34:53.980
AUDIENCE: Young centers.

00:34:53.980 --> 00:34:55.946
PROFESSOR: Young centers.

00:34:55.946 --> 00:34:57.254
AUDIENCE: Large buildings.

00:34:57.254 --> 00:35:03.985
PROFESSOR: Large buildings,
irrigation, biogas.

00:35:03.985 --> 00:35:04.895
AUDIENCE: Electricity.

00:35:04.895 --> 00:35:05.490
PROFESSOR: Sorry?

00:35:05.490 --> 00:35:06.480
AUDIENCE: Electricity.

00:35:06.480 --> 00:35:08.000
PROFESSOR: Electricity,
potentially.

00:35:08.000 --> 00:35:09.470
Sanitation.

00:35:09.470 --> 00:35:14.670
So like a long list of public
goods could go either a way,

00:35:14.670 --> 00:35:15.920
so that's a long list.

00:35:19.230 --> 00:35:20.480
Where else?

00:35:25.825 --> 00:35:26.890
AUDIENCE: Perceptions
about women.

00:35:26.890 --> 00:35:28.427
PROFESSOR: Perception
of women.

00:35:33.910 --> 00:35:38.180
So if you want we'll talk a bit
more about how we measure

00:35:38.180 --> 00:35:39.490
perception about women.

00:35:39.490 --> 00:35:40.480
AUDIENCE: Participation.

00:35:40.480 --> 00:35:42.000
PROFESSOR: Political
participation.

00:35:46.240 --> 00:35:50.550
So attendance at meetings,
voting.

00:35:50.550 --> 00:35:52.280
And of course we have
men and women.

00:35:52.280 --> 00:35:54.530
That might be different.

00:35:57.185 --> 00:35:59.130
AUDIENCE: Better governance?

00:35:59.130 --> 00:36:01.650
PROFESSOR: So that's kind
of the same thing.

00:36:01.650 --> 00:36:03.110
So better governance.

00:36:03.110 --> 00:36:06.760
In practice that's going to be
whether there is graft, maybe

00:36:06.760 --> 00:36:08.100
budget utilization.

00:36:08.100 --> 00:36:10.860
So some measure of corruption.

00:36:10.860 --> 00:36:13.510
Bribes.

00:36:13.510 --> 00:36:15.480
You didn't say it actually
in the advantages or

00:36:15.480 --> 00:36:17.160
disadvantages.

00:36:17.160 --> 00:36:19.250
The women are less corrupt,
more corrupt.

00:36:19.250 --> 00:36:22.420
I guess it came up in
the accountability.

00:36:22.420 --> 00:36:22.845
What else?

00:36:22.845 --> 00:36:24.716
AUDIENCE: Household income.

00:36:24.716 --> 00:36:28.060
PROFESSOR: Household income.

00:36:28.060 --> 00:36:29.820
And while we are in
the household?

00:36:29.820 --> 00:36:31.452
AUDIENCE: Health education.

00:36:31.452 --> 00:36:32.755
PROFESSOR: Health education.

00:36:36.700 --> 00:36:39.940
Any gender differences in these
things, both for the

00:36:39.940 --> 00:36:42.660
household and for the kids.

00:36:42.660 --> 00:36:44.340
We were talking about gender

00:36:44.340 --> 00:36:46.110
discrimination within the household.

00:36:46.110 --> 00:36:49.370
So again it's like a long
list of household stuff

00:36:49.370 --> 00:36:51.270
potentially.

00:36:51.270 --> 00:36:53.670
Was a woman participating
in savings,

00:36:53.670 --> 00:36:54.920
groups, blah, blah, blah?

00:37:05.250 --> 00:37:08.600
There is both perception of
women politicians, and we also

00:37:08.600 --> 00:37:13.150
discussed about perception
of women in general.

00:37:13.150 --> 00:37:15.020
What else could we need?

00:37:15.020 --> 00:37:17.450
AUDIENCE: Maybe greater
employability of women?

00:37:17.450 --> 00:37:18.130
PROFESSOR: Right.

00:37:18.130 --> 00:37:21.730
So that's going to be maybe an
income and then employment.

00:37:21.730 --> 00:37:25.590
That's part of the long
list of stuff we might

00:37:25.590 --> 00:37:26.840
collect in a household.

00:37:29.810 --> 00:37:32.006
AUDIENCE: Social cohesion.

00:37:32.006 --> 00:37:32.710
PROFESSOR: Right.

00:37:32.710 --> 00:37:35.100
So that's maybe political

00:37:35.100 --> 00:37:38.660
participation and social cohesion.

00:37:38.660 --> 00:37:41.720
Don't know how you measure
that, but we can

00:37:41.720 --> 00:37:42.970
think about it later.

00:37:46.722 --> 00:37:48.189
AUDIENCE: Sustainability.

00:37:48.189 --> 00:37:50.030
Or sort of decentralization
of power.

00:37:54.340 --> 00:37:57.153
The majority that's been in the
government, it's like a

00:37:57.153 --> 00:37:57.485
vicious cycle.

00:37:57.485 --> 00:38:00.467
They'll do everything to keep
the minorities out of the

00:38:00.467 --> 00:38:04.443
governance, so if you had jump
started this process by

00:38:04.443 --> 00:38:06.440
introducing 30% of women--

00:38:06.440 --> 00:38:07.360
PROFESSOR: Right,
right, right.

00:38:07.360 --> 00:38:13.890
So it's perception of women
politicians is in a woman's

00:38:13.890 --> 00:38:17.560
future electoral success.

00:38:17.560 --> 00:38:21.090
Is it the case that once a place
is reserved for a woman,

00:38:21.090 --> 00:38:22.550
obviously you are the woman.

00:38:22.550 --> 00:38:25.650
How about the next time when
it's not reserved anymore, do

00:38:25.650 --> 00:38:27.130
you have more woman
candidates?

00:38:27.130 --> 00:38:30.170
Do you have a woman elected?

00:38:30.170 --> 00:38:32.040
Women's future electoral
successes.

00:38:34.660 --> 00:38:36.670
That's one of the first things
we had discussed.

00:38:39.480 --> 00:38:41.645
More?

00:38:41.645 --> 00:38:45.170
I think it's a pretty
long list already.

00:38:45.170 --> 00:38:48.380
And so now that we have this
long list, the issue

00:38:48.380 --> 00:38:51.200
is what do we do?

00:38:51.200 --> 00:38:52.870
Suppose you had no
money problem.

00:38:56.090 --> 00:38:59.250
Suppose budget was
not a problem.

00:38:59.250 --> 00:39:01.800
Would you just take
this long list?

00:39:01.800 --> 00:39:06.040
So one possible thing is to say,
well, we'll think about

00:39:06.040 --> 00:39:07.320
this thing later.

00:39:07.320 --> 00:39:11.490
That makes a long list
of outcomes, and

00:39:11.490 --> 00:39:13.550
go and collect data.

00:39:13.550 --> 00:39:15.320
So we're going to do a household
survey and a

00:39:15.320 --> 00:39:20.450
community survey, and then an
audit of what's there in the

00:39:20.450 --> 00:39:24.490
villages, and we are going to
collect a bunch of data.

00:39:24.490 --> 00:39:26.470
And then this data will
come back, and we'll

00:39:26.470 --> 00:39:28.250
start to look at it.

00:39:28.250 --> 00:39:31.930
So what are the pluses and
minuses of that approach?

00:39:31.930 --> 00:39:34.350
AUDIENCE: [INAUDIBLE]

00:39:34.350 --> 00:39:35.802
statistical significance.

00:39:35.802 --> 00:39:38.948
If you collect so much data,
eventually you'll randomly

00:39:38.948 --> 00:39:41.448
stumble upon a result, and maybe
that's the result that

00:39:41.448 --> 00:39:42.094
you end up reporting.

00:39:42.094 --> 00:39:43.090
PROFESSOR: Right.

00:39:43.090 --> 00:39:48.180
So in particular, since we have
seen at least some of the

00:39:48.180 --> 00:39:51.740
groups starting to discuss
power, so you've seen

00:39:51.740 --> 00:39:52.990
hypothesis testing.

00:39:52.990 --> 00:39:56.070
So if I ran a hundred regression
and looked for

00:39:56.070 --> 00:39:59.570
significant results and then
independently looked for

00:39:59.570 --> 00:40:01.830
results in each of them, how
many of them would count

00:40:01.830 --> 00:40:03.720
significant at the 5% level?

00:40:03.720 --> 00:40:04.626
AUDIENCE: Five.

00:40:04.626 --> 00:40:06.320
PROFESSOR: Five.

00:40:06.320 --> 00:40:09.730
So if I run 20, I
would get two.

00:40:09.730 --> 00:40:12.505
So by example, suppose I collect
20 public goods, which

00:40:12.505 --> 00:40:15.720
is not such a large number.

00:40:15.720 --> 00:40:21.970
And we find that water wells
go up in places which have

00:40:21.970 --> 00:40:28.320
more women, and irrigation goes
down, and anything else

00:40:28.320 --> 00:40:29.570
doesn't change.

00:40:33.770 --> 00:40:38.090
What can I conclude if I have
just gone on to this big

00:40:38.090 --> 00:40:40.832
fishing expedition?

00:40:40.832 --> 00:40:43.302
AUDIENCE: [INAUDIBLE]

00:40:43.302 --> 00:40:45.460
PROFESSOR: Well, you don't know

00:40:45.460 --> 00:40:46.730
whether it was just random.

00:40:46.730 --> 00:40:48.680
You can definitely
make up a story.

00:40:48.680 --> 00:40:51.300
What is a story you could easily
make up on the basis of

00:40:51.300 --> 00:40:52.550
those results?

00:40:52.550 --> 00:41:00.197
AUDIENCE: That women invest in
wells because collecting water

00:41:00.197 --> 00:41:04.508
used to be a woman's job, and
men invest in irrigation

00:41:04.508 --> 00:41:07.861
because farming is a
man's occupation.

00:41:07.861 --> 00:41:08.850
PROFESSOR: Exactly.

00:41:08.850 --> 00:41:11.960
So you could say, well, women
are not going to benefit from

00:41:11.960 --> 00:41:14.230
irrigation anyway, they're going
to benefit from drinking

00:41:14.230 --> 00:41:15.400
water a lot.

00:41:15.400 --> 00:41:18.450
So this is quite consistent with
what we were saying about

00:41:18.450 --> 00:41:22.300
women leaders investing in the
goods women want, and so

00:41:22.300 --> 00:41:24.130
that's great.

00:41:24.130 --> 00:41:28.260
The only problem is, if we've
done that ex-post, someone

00:41:28.260 --> 00:41:31.450
else could say, well, or an
alternative interpretation is

00:41:31.450 --> 00:41:34.020
you've run 20 regression, you've
found two significant,

00:41:34.020 --> 00:41:37.680
and you're making up
a story ex-post

00:41:37.680 --> 00:41:39.490
to explain the results.

00:41:44.170 --> 00:41:47.730
It's not that it's morally
wrong to do that, but if

00:41:47.730 --> 00:41:52.920
you're not sure it is always
going to be some suspicion.

00:41:52.920 --> 00:41:55.090
And then it's a little bit sad
to have spent so much money

00:41:55.090 --> 00:42:00.760
collecting so much data and
not to be very sure how to

00:42:00.760 --> 00:42:02.290
interpret the results.

00:42:02.290 --> 00:42:06.130
So maybe that's not the
right approach.

00:42:06.130 --> 00:42:09.615
Maybe we need to do something
slightly different, and that

00:42:09.615 --> 00:42:13.680
is we need to try and put a
little bit more thought into

00:42:13.680 --> 00:42:16.820
why we are collecting each piece
of data and where it's

00:42:16.820 --> 00:42:23.060
going to fit in our global
explanation of the results.

00:42:23.060 --> 00:42:30.360
So if you take this specific
example, instead of doing what

00:42:30.360 --> 00:42:34.000
we just suppose we did, which
is making this long list of

00:42:34.000 --> 00:42:36.880
results and hope for the best.

00:42:36.880 --> 00:42:39.810
Instead, if we had said we are
going to go after this one

00:42:39.810 --> 00:42:43.260
question, potentially we can
have we can have more than one

00:42:43.260 --> 00:42:44.380
big question.

00:42:44.380 --> 00:42:46.850
But we are at least going to
go after this one question

00:42:46.850 --> 00:42:50.580
which is, what is it the case
that the women leader do what

00:42:50.580 --> 00:42:52.640
women want?

00:42:52.640 --> 00:42:58.130
And we had given ourselves the
means to first find out what

00:42:58.130 --> 00:43:01.840
women want for real, and not
making it up from the results.

00:43:01.840 --> 00:43:05.690
The key is to try to not be in
a position where you're going

00:43:05.690 --> 00:43:08.760
to have to reverse engineer
your explanation

00:43:08.760 --> 00:43:11.920
of the result ex-post.

00:43:11.920 --> 00:43:15.850
If you want me to be completely
honest, it's very

00:43:15.850 --> 00:43:17.110
difficult not to do that.

00:43:17.110 --> 00:43:20.140
Because once you see the results
you always want to

00:43:20.140 --> 00:43:22.960
explain them to yourself,
and explain to others.

00:43:22.960 --> 00:43:27.980
So I'm not about to tell you
that from a position of where

00:43:27.980 --> 00:43:31.400
I sit and say well, don't you do
that or you will be damned

00:43:31.400 --> 00:43:32.360
and go to hell.

00:43:32.360 --> 00:43:35.320
But the truth is that we do it,
but the truth is the less

00:43:35.320 --> 00:43:37.410
we have to do it the better.

00:43:37.410 --> 00:43:42.330
And you have to do it less if
you've been thinking ex-ante

00:43:42.330 --> 00:43:47.330
about what it is that you want
to test, and what are the

00:43:47.330 --> 00:43:52.760
steps that align themselves
in order to get to

00:43:52.760 --> 00:43:54.210
where you want to be.

00:43:54.210 --> 00:43:59.680
And this result might have been
totally fine if we had a

00:43:59.680 --> 00:44:02.210
good way to say this is what
woman are really interested

00:44:02.210 --> 00:44:06.050
in, and we had collected
data on that.

00:44:06.050 --> 00:44:12.940
And then where our test will not
be good by good, is it the

00:44:12.940 --> 00:44:15.150
case that women do
different things.

00:44:15.150 --> 00:44:17.750
But we ask the questions
that generally do

00:44:17.750 --> 00:44:20.330
women go in that direction.

00:44:20.330 --> 00:44:21.490
And what is the case?

00:44:21.490 --> 00:44:27.480
Yes, we're missing a moral
explaining to us why it is.

00:44:27.480 --> 00:44:30.740
How do we go from this
hypothesis we have that women

00:44:30.740 --> 00:44:34.220
are going to do what women want,
to how is it going to

00:44:34.220 --> 00:44:36.630
translate into the
public goods.

00:44:36.630 --> 00:44:46.230
So the hypothesis to test must
be defined before the

00:44:46.230 --> 00:44:49.070
beginning of the experiment or
we don't know how to assist

00:44:49.070 --> 00:44:50.330
their validity.

00:44:50.330 --> 00:44:53.010
And what we missed when we did
this big list of outcomes, or

00:44:53.010 --> 00:44:56.120
what we would have missed if we
just did this big list of

00:44:56.120 --> 00:44:58.770
outcomes, and then go out and
collect the data and then

00:44:58.770 --> 00:45:02.410
think about how to interpret
them, is these steps.

00:45:02.410 --> 00:45:07.060
Which is go from a discussion
which is very likely at an

00:45:07.060 --> 00:45:09.700
implicit level at the
back of our minds.

00:45:09.700 --> 00:45:13.130
In fact, now it's very much in
the front of our minds since

00:45:13.130 --> 00:45:14.510
we just had it.

00:45:14.510 --> 00:45:17.120
But usually when you do an
evaluation, it's everybody

00:45:17.120 --> 00:45:22.420
sort of had that in mind, but
it might remain implicit and

00:45:22.420 --> 00:45:24.630
it's much better to
make it explicit.

00:45:24.630 --> 00:45:27.230
First because you're more likely
to collect the data

00:45:27.230 --> 00:45:29.040
that you actually will need.

00:45:29.040 --> 00:45:31.290
For example, here we might have
missed collecting women's

00:45:31.290 --> 00:45:31.730
preferences.

00:45:31.730 --> 00:45:34.660
It's not in the list.

00:45:34.660 --> 00:45:37.280
It's not in the list of
data that's here.

00:45:37.280 --> 00:45:37.650
Why?

00:45:37.650 --> 00:45:39.240
Because it's not an outcome.

00:45:39.240 --> 00:45:42.570
So if were just thinking
about, oh, what are the

00:45:42.570 --> 00:45:45.280
effects, and we forget to
collect women's preferences.

00:45:45.280 --> 00:45:48.130
But then we have no good way to
interpret the preferences

00:45:48.130 --> 00:45:51.690
in the context of that model
if it was the model.

00:45:51.690 --> 00:45:54.650
So if we don't do this thinking
implicitly, we might

00:45:54.650 --> 00:45:56.220
be missing a key step.

00:45:56.220 --> 00:45:59.560
In particular, in what we call
the intermediate variable that

00:45:59.560 --> 00:46:03.130
might be needed not as a measure
of the impact of the

00:46:03.130 --> 00:46:07.390
program, but as what is going
to help us interpret the

00:46:07.390 --> 00:46:09.860
impact of this program.

00:46:09.860 --> 00:46:13.620
So you need to try and define
the hypotheses before the

00:46:13.620 --> 00:46:15.300
beginning of the experiment.

00:46:15.300 --> 00:46:19.050
And this is actually an exercise
that's very useful in

00:46:19.050 --> 00:46:24.150
my experience working with
implementation partners.

00:46:24.150 --> 00:46:26.820
It's very useful
for both sides.

00:46:26.820 --> 00:46:32.710
Because from the side of the
evaluation team, strive to

00:46:32.710 --> 00:46:35.200
understand what is the program
you're evaluating.

00:46:35.200 --> 00:46:38.410
How it connects to what you know
about, say, developments,

00:46:38.410 --> 00:46:41.540
poverty, et cetera.

00:46:41.540 --> 00:46:47.310
From the side of the partner,
many of you have more

00:46:47.310 --> 00:46:48.600
experience than me on that.

00:46:48.600 --> 00:46:52.350
At least the half that are
actually into implementation.

00:46:52.350 --> 00:46:55.250
It's like why are
we doing this?

00:46:55.250 --> 00:47:02.282
Sometimes the answer is not as
forthcoming as you might hope.

00:47:02.282 --> 00:47:05.590
AUDIENCE: The other thing that
was noticeable to me when we

00:47:05.590 --> 00:47:09.232
were doing this is that there
are different levels maybe of

00:47:09.232 --> 00:47:11.802
importance, or that
a lot of them are

00:47:11.802 --> 00:47:12.825
subsets of other things.

00:47:12.825 --> 00:47:14.260
PROFESSOR: Exactly.

00:47:14.260 --> 00:47:19.810
So one of the things that you
might do in this process is to

00:47:19.810 --> 00:47:23.890
prioritize what is sort of the
big news, sort of the million

00:47:23.890 --> 00:47:25.390
dollar question?

00:47:25.390 --> 00:47:27.690
What are subsidiary?

00:47:27.690 --> 00:47:30.530
What are things that are
going to enlighten

00:47:30.530 --> 00:47:32.620
whatever impacts you find?

00:47:32.620 --> 00:47:35.920
So you might also, we are going
to do that in a minute,

00:47:35.920 --> 00:47:39.476
is to think depending on how
much money you have--

00:47:39.476 --> 00:47:41.312
is it that good?

00:47:41.312 --> 00:47:42.548
So sorry.

00:47:42.548 --> 00:47:46.070
We are trying to do something.

00:47:46.070 --> 00:47:49.370
Given how much money you have,
you might be thinking of some

00:47:49.370 --> 00:47:52.130
small things, or some
things that might--

00:47:52.130 --> 00:47:55.190
for example, some of the things,
like say, one of the

00:47:55.190 --> 00:47:58.940
outcomes we are thinking is
relative mortality of girls.

00:47:58.940 --> 00:48:01.640
You might think realistically
two years after the women's

00:48:01.640 --> 00:48:04.635
empowerment is less
likely to happen.

00:48:04.635 --> 00:48:06.895
So if you had an infinite amount
of money, you might

00:48:06.895 --> 00:48:09.275
still collect it as a subsidiary
outcome, but you

00:48:09.275 --> 00:48:12.190
wouldn't want it in a long list
up there with whether or

00:48:12.190 --> 00:48:15.850
not they invest in
water wells.

00:48:15.850 --> 00:48:20.730
There's no way to do this
prioritization unless you are

00:48:20.730 --> 00:48:25.490
in good cause the exercise of
thinking to your causal model,

00:48:25.490 --> 00:48:30.150
linking whatever intervention
you have to the results.

00:48:30.150 --> 00:48:34.280
AUDIENCE: In this discussion I
believe you also raised the

00:48:34.280 --> 00:48:38.340
possibly of doing like a factor
analysis to create an

00:48:38.340 --> 00:48:42.547
index of different factors so
that we could reduce the

00:48:42.547 --> 00:48:45.281
number of variables.

00:48:45.281 --> 00:48:46.090
PROFESSOR: Right.

00:48:46.090 --> 00:48:49.590
So that's a very good idea.

00:48:49.590 --> 00:48:53.760
But on the other hand, you have
to think about how to--

00:48:53.760 --> 00:48:56.400
your instinct is the right one,
which is to say how are

00:48:56.400 --> 00:48:59.800
we going to combine
this stuff?

00:48:59.800 --> 00:49:02.390
So if you take, for example,
the 20 public goods.

00:49:02.390 --> 00:49:05.350
If you're like, but these women
are more efficient,

00:49:05.350 --> 00:49:07.440
they're going to build
more goods.

00:49:07.440 --> 00:49:12.150
Then you might want to do that,
or you might want to sum

00:49:12.150 --> 00:49:14.630
them or to average them somehow,
which is what making

00:49:14.630 --> 00:49:16.900
an index is.

00:49:16.900 --> 00:49:20.340
To see whether in general
all go positive.

00:49:20.340 --> 00:49:25.270
But if your model is along this
line which is women are

00:49:25.270 --> 00:49:28.880
doing what women want, then you
might not get more goods.

00:49:28.880 --> 00:49:32.800
You may get some of some, and
less of others, so you might

00:49:32.800 --> 00:49:38.190
need something else than an
index to deal with this mess.

00:49:38.190 --> 00:49:40.970
But you're exactly right with
the idea that what we want is

00:49:40.970 --> 00:49:47.410
a way to not have many, many,
many hypothesis, but fewer.

00:49:47.410 --> 00:49:49.250
So, for example, it can be one,
which is a woman does

00:49:49.250 --> 00:49:50.910
what women want.

00:49:50.910 --> 00:49:52.580
That would be the first
one, and then what

00:49:52.580 --> 00:49:56.090
comes out of that.

00:49:56.090 --> 00:49:59.420
If that's your hypothesis, the
index you create would be

00:49:59.420 --> 00:50:02.050
something different than
what you would get

00:50:02.050 --> 00:50:03.720
from a factor analysis.

00:50:03.720 --> 00:50:07.060
If, on the other hand, it's an
education intervention, and

00:50:07.060 --> 00:50:12.500
you have a math result, and an
English result, and science

00:50:12.500 --> 00:50:14.490
result, and [UNINTELLIGIBLE]
results, and geographic

00:50:14.490 --> 00:50:20.010
results, then you know that
they should all up.

00:50:20.010 --> 00:50:21.260
Or at least that's your
hypothesis they should all go

00:50:21.260 --> 00:50:25.410
up, then you can do something
like that to average the

00:50:25.410 --> 00:50:28.311
effect across the outcomes.

00:50:31.137 --> 00:50:34.630
AUDIENCE: I understand why you
don't want to throw in 20

00:50:34.630 --> 00:50:39.830
different outcomes.

00:50:39.830 --> 00:50:45.650
But in terms of how you
prioritize the outcomes along

00:50:45.650 --> 00:50:48.540
with their causal model, there
are some fields where it's

00:50:48.540 --> 00:50:49.446
going to be fairly clear.

00:50:49.446 --> 00:50:51.545
You know, in the health field,
they would say, OK, well, our

00:50:51.545 --> 00:50:54.286
objective is to reduce
malaria, so we

00:50:54.286 --> 00:50:55.953
didn't reduce malaria.

00:50:55.953 --> 00:50:59.390
But if something that has more
intermediate steps along the

00:50:59.390 --> 00:51:00.640
way, such as a--

00:51:04.330 --> 00:51:05.580
PROFESSOR: What was
that about?

00:51:10.289 --> 00:51:12.650
AUDIENCE: --such as like a
conflict mitigation program,

00:51:12.650 --> 00:51:24.193
where you're saying, OK, we are
going to get children into

00:51:24.193 --> 00:51:29.980
clubs because we believe that
this is going to create more

00:51:29.980 --> 00:51:33.724
solidarity among ethnic groups,
and that this is going

00:51:33.724 --> 00:51:36.610
to lead to conflict reduction.

00:51:36.610 --> 00:51:39.440
When you have, like, six or
seven different steps along

00:51:39.440 --> 00:51:43.180
this causal model, how and at
what point do you prioritize,

00:51:43.180 --> 00:51:46.810
this is the impact that we're
looking for, versus these are

00:51:46.810 --> 00:51:50.170
the steps that we have take
to get to this impact?

00:51:50.170 --> 00:51:52.830
PROFESSOR: So that's a great
question, and you can answer

00:51:52.830 --> 00:51:54.620
it at two levels in a way.

00:51:54.620 --> 00:52:01.860
One is, where and when you're
going to see an impact first.

00:52:01.860 --> 00:52:04.250
For example, in this case, you
might think that ultimately

00:52:04.250 --> 00:52:06.750
what we care is not the
number of water wells.

00:52:06.750 --> 00:52:11.400
Ultimately what we care is the
growth of the Indian economy

00:52:11.400 --> 00:52:13.150
through these complicated
channels.

00:52:13.150 --> 00:52:17.320
But this is not what we are
trying to do, because we think

00:52:17.320 --> 00:52:18.250
it's going to take a
bit more time to

00:52:18.250 --> 00:52:19.490
arrive to this question.

00:52:19.490 --> 00:52:23.010
Or ultimately what we care about
is girl's mortality.

00:52:23.010 --> 00:52:25.380
And we're thinking that through
this complicated path

00:52:25.380 --> 00:52:28.210
we're going to get there
eventually, but this is not

00:52:28.210 --> 00:52:30.430
the yardstick by which you
measure the success of the

00:52:30.430 --> 00:52:34.030
program because it's in too long
time, and after too many

00:52:34.030 --> 00:52:36.230
other things have
diluted this.

00:52:36.230 --> 00:52:38.010
But at least you want to know
that you're going in the right

00:52:38.010 --> 00:52:42.550
direction, so you might decide
to stop by the water wells for

00:52:42.550 --> 00:52:44.470
the beginning.

00:52:44.470 --> 00:52:47.540
Likewise in your program, you
might say you're going to

00:52:47.540 --> 00:52:50.950
produce into groups
to work together.

00:52:50.950 --> 00:52:55.400
You might say, well our first
measure of success with this

00:52:55.400 --> 00:53:00.230
program is whether or not the
use opinion of the other guys

00:53:00.230 --> 00:53:04.040
has changed in a way that
we can reliably measure.

00:53:04.040 --> 00:53:05.830
And then you're going to ask
yourself the question of the

00:53:05.830 --> 00:53:17.260
measurement, and then you can
take it at various steps.

00:53:17.260 --> 00:53:21.020
And one is, am I still going
to find an effect after all

00:53:21.020 --> 00:53:23.270
the things that have happened?

00:53:23.270 --> 00:53:26.080
So is it fair to my program?

00:53:26.080 --> 00:53:28.800
And the other is what
can I measure?

00:53:28.800 --> 00:53:31.460
Some things are easier and
harder to measure, and they

00:53:31.460 --> 00:53:33.530
might be at different
levels in the chain.

00:53:33.530 --> 00:53:36.060
It might be that what is
happening first is easier to

00:53:36.060 --> 00:53:37.710
measure, sometimes it's what's
happening a little later

00:53:37.710 --> 00:53:38.730
that's easier to measure.

00:53:38.730 --> 00:53:41.870
For example, perceptions might
be very hard to measure.

00:53:41.870 --> 00:53:45.530
But whether or not people are
willing to work together to

00:53:45.530 --> 00:53:50.370
build something might be very
easy to measure, so you might

00:53:50.370 --> 00:53:50.900
go for that.

00:53:50.900 --> 00:53:52.690
AUDIENCE: And at the same time,
you do want to make sure

00:53:52.690 --> 00:53:55.962
that you're doing something
further enough along the chain

00:53:55.962 --> 00:53:58.243
that there isn't actually
impact, and not just an output

00:53:58.243 --> 00:53:58.558
measure, or--

00:53:58.558 --> 00:53:59.660
PROFESSOR: Exactly.

00:53:59.660 --> 00:54:04.520
So this is where this

00:54:04.520 --> 00:54:06.580
conversation's going to be helpful.

00:54:06.580 --> 00:54:12.100
What is our program hoping
to do, and how.

00:54:12.100 --> 00:54:14.330
And sometimes you realize in
that conversation is that what

00:54:14.330 --> 00:54:18.260
is your program hoping to do
so squishy that maybe we

00:54:18.260 --> 00:54:20.560
should do something else.

00:54:20.560 --> 00:54:22.770
But not the case of your
program that you just

00:54:22.770 --> 00:54:26.360
described which sounded actually
quite specific in

00:54:26.360 --> 00:54:27.800
what it was hoping to achieve.

00:54:36.350 --> 00:54:37.710
So here's our example.

00:54:37.710 --> 00:54:41.000
So we have to define
the hypothesis.

00:54:41.000 --> 00:54:43.750
So, for example, we can take
here, we could take several.

00:54:43.750 --> 00:54:46.850
One possibility is to take this
one which is public goods

00:54:46.850 --> 00:54:51.540
favored by women are more likely
to be chosen by women.

00:54:51.540 --> 00:54:57.080
So to test that, we need to know
what women want, and then

00:54:57.080 --> 00:54:58.810
we need to collect the
data on public goods.

00:54:58.810 --> 00:55:01.640
And then the one thing we are
going to do is, we are

00:55:01.640 --> 00:55:05.100
creating this index of wanted
by women more than by men,

00:55:05.100 --> 00:55:07.390
which is the way we
would aggregate.

00:55:07.390 --> 00:55:10.310
And then the key test would be
are the public goods moving

00:55:10.310 --> 00:55:12.460
towards what women want?

00:55:12.460 --> 00:55:15.900
And so the one little snag is
how are we going to measure

00:55:15.900 --> 00:55:17.790
women's preferences?

00:55:17.790 --> 00:55:20.080
What is it that women want?

00:55:20.080 --> 00:55:22.350
So what do you guys think?

00:55:22.350 --> 00:55:26.770
What would be possibilities to
measure a woman's preferences?

00:55:26.770 --> 00:55:28.710
AUDIENCE: Household surveys?

00:55:28.710 --> 00:55:30.970
PROFESSOR: So we could ask them,
in household surveys.

00:55:30.970 --> 00:55:33.860
That would be one way.

00:55:33.860 --> 00:55:35.110
What they care about.

00:55:37.936 --> 00:55:39.186
Any other ideas?

00:55:42.162 --> 00:55:46.210
AUDIENCE: If there were
communities where women were

00:55:46.210 --> 00:55:49.200
well represented, what were
the choices of those

00:55:49.200 --> 00:55:50.660
communities, historically?

00:55:50.660 --> 00:55:52.980
PROFESSOR: Well, it's slightly
circular because--

00:55:55.560 --> 00:55:56.020
AUDIENCE: [INAUDIBLE]

00:55:56.020 --> 00:56:02.440
that have a majority of women
on the local council have

00:56:02.440 --> 00:56:03.990
higher percentage of wells.

00:56:03.990 --> 00:56:05.010
PROFESSOR: Well, yeah.

00:56:05.010 --> 00:56:12.200
That's a little bit circular,
because that means that what

00:56:12.200 --> 00:56:14.930
we can do that only, A, if we
believe that, in fact, it is

00:56:14.930 --> 00:56:18.470
true that women better do what
women want, which is what we

00:56:18.470 --> 00:56:19.440
are trying to test.

00:56:19.440 --> 00:56:22.570
And B, if we think there was no
selection in which village

00:56:22.570 --> 00:56:27.890
elected a woman, and we don't
believe that which is the

00:56:27.890 --> 00:56:29.190
whole reason why we
are going through

00:56:29.190 --> 00:56:30.510
this tortuous exercise.

00:56:30.510 --> 00:56:31.648
So that might--

00:56:31.648 --> 00:56:36.379
AUDIENCE: I know in the exercise
it talked about the

00:56:36.379 --> 00:56:39.328
transcripts from some of the
meetings, so you could look to

00:56:39.328 --> 00:56:42.674
see if there were patterns for
what issues were raised.

00:56:42.674 --> 00:56:45.542
Whether or not the council voted
on them, it was what

00:56:45.542 --> 00:56:46.976
they brought to the table.

00:56:46.976 --> 00:56:47.740
PROFESSOR: Right.

00:56:47.740 --> 00:56:49.070
So you could do that.

00:56:49.070 --> 00:56:51.560
In fact, this is what
we did in this case.

00:56:51.560 --> 00:56:54.960
Cheaper than household survey,
is to say, in general, what do

00:56:54.960 --> 00:56:58.220
women ask about in
those meetings?

00:56:58.220 --> 00:57:00.930
Preferably, not in places where
the woman is the head,

00:57:00.930 --> 00:57:03.100
because that might have changed
the dynamic, but in

00:57:03.100 --> 00:57:05.480
places which are not reserved.

00:57:05.480 --> 00:57:07.690
What are women talking about?

00:57:07.690 --> 00:57:09.220
What are men talking about?

00:57:09.220 --> 00:57:13.120
And the advantage of a household
survey is that it's

00:57:13.120 --> 00:57:15.760
a little bit costly to go and
talk about something.

00:57:15.760 --> 00:57:18.980
In a meeting, you speak up and
other people look at you.

00:57:18.980 --> 00:57:22.850
If you have to go to the leader
and have a written

00:57:22.850 --> 00:57:25.670
complaint it takes some time,
and people wouldn't do that

00:57:25.670 --> 00:57:27.160
unless they cared about it.

00:57:27.160 --> 00:57:28.870
So maybe that's one
way to proceed.

00:57:28.870 --> 00:57:31.550
But household surveys is also
a good way to proceed.

00:57:31.550 --> 00:57:33.490
AUDIENCE: But if you're
measuring whether women

00:57:33.490 --> 00:57:38.630
represent women, like you've
assumed at that point, but if

00:57:38.630 --> 00:57:40.630
you're looking at the discussion
within the council,

00:57:40.630 --> 00:57:42.093
you've assumed at that
point that the

00:57:42.093 --> 00:57:43.605
woman that was selected--

00:57:43.605 --> 00:57:44.855
PROFESSOR: Alright.

00:57:44.855 --> 00:57:45.645
AUDIENCE: --for women.

00:57:45.645 --> 00:57:46.350
PROFESSOR: Alright.

00:57:46.350 --> 00:57:49.645
So I was thinking of not the
small council, but the big

00:57:49.645 --> 00:57:53.030
council when they have
this big meeting.

00:57:53.030 --> 00:57:55.500
What she was referring to
is the transcript of the

00:57:55.500 --> 00:57:58.140
[UNINTELLIGIBLE] meetings.

00:57:58.140 --> 00:58:02.320
Or even when women have gone to
the office of the Pradhan

00:58:02.320 --> 00:58:03.570
and asked them.

00:58:05.600 --> 00:58:09.510
So that's different ways,
but the key is, we

00:58:09.510 --> 00:58:10.600
need to have that.

00:58:10.600 --> 00:58:12.430
Because then we can
see it doesn't

00:58:12.430 --> 00:58:13.990
go into that direction.

00:58:13.990 --> 00:58:16.950
And now we don't have 20 public
goods, we have one

00:58:16.950 --> 00:58:17.600
hypothesis.

00:58:17.600 --> 00:58:20.609
AUDIENCE: Is it necessarily
critical that what women are

00:58:20.609 --> 00:58:23.062
speaking about when they go
visit to large meetings that's

00:58:23.062 --> 00:58:25.278
what they want, and what the men
are speaking about, that's

00:58:25.278 --> 00:58:25.980
what they want?

00:58:25.980 --> 00:58:27.384
What if a man is speaking
on behalf

00:58:27.384 --> 00:58:28.830
of his wife's interests?

00:58:28.830 --> 00:58:29.320
PROFESSOR: Right.

00:58:29.320 --> 00:58:32.290
So it is not entirely clear.

00:58:32.290 --> 00:58:37.920
And so this measure of women's
and men's preference might not

00:58:37.920 --> 00:58:39.920
be the best you can think of.

00:58:39.920 --> 00:58:42.100
The key is to have a good one.

00:58:42.100 --> 00:58:45.290
I think you're right, which is
it could be that women never

00:58:45.290 --> 00:58:48.110
speak for example, or never
complain about anything, and

00:58:48.110 --> 00:58:50.390
they would still have
preferences so what would have

00:58:50.390 --> 00:58:53.990
to figure something
out to get them.

00:58:53.990 --> 00:58:57.940
To the extent that both genders
speak, then you might

00:58:57.940 --> 00:59:00.840
think that they speaking about
what they care about as long

00:59:00.840 --> 00:59:09.890
as the house is not a fully
harmonious unit.

00:59:09.890 --> 00:59:11.720
Or you might think that it's
not, and then for example,

00:59:11.720 --> 00:59:14.450
women tend to talk about water
just because they know better

00:59:14.450 --> 00:59:16.990
about water as someone
suggested earlier.

00:59:16.990 --> 00:59:21.430
And it doesn't represent needs,
it just represents

00:59:21.430 --> 00:59:24.120
relative advantages.

00:59:24.120 --> 00:59:27.790
So that is something that this
particular wealth measuring

00:59:27.790 --> 00:59:31.250
preferences might not be
the best one, but the

00:59:31.250 --> 00:59:34.590
key is to have one.

00:59:34.590 --> 00:59:37.042
More than one would
be even better.

00:59:37.042 --> 00:59:40.300
So in this case, that's the only
when we had so it was a

00:59:40.300 --> 00:59:42.300
bit of a gamble.

00:59:42.300 --> 00:59:46.600
But the key thing is that
we want to do that.

00:59:46.600 --> 00:59:49.010
Then there are other things
we might be interested in.

00:59:49.010 --> 00:59:53.860
So once we have that, do women
invest more in public goods,

00:59:53.860 --> 01:00:00.430
then we can ask the next
question which is do investing

01:00:00.430 --> 01:00:04.530
more once we find this that
kind of can be our study

01:00:04.530 --> 01:00:05.610
number one.

01:00:05.610 --> 01:00:09.165
Which is yes, we are showing
that women invest more in the

01:00:09.165 --> 01:00:10.660
public goods that
women prefer.

01:00:10.660 --> 01:00:12.970
Then there was this other
question that also came about,

01:00:12.970 --> 01:00:16.160
a completely different question,
which is, does

01:00:16.160 --> 01:00:19.590
having a woman as a policy maker
change the perception of

01:00:19.590 --> 01:00:21.810
women as policy makers.

01:00:21.810 --> 01:00:25.470
And that we can think of a
separate hypothesis altogether

01:00:25.470 --> 01:00:27.180
that can be tested separately.

01:00:27.180 --> 01:00:32.210
And it could be that women as
leaders do different things

01:00:32.210 --> 01:00:36.030
and do what women want, but
that there is no lingering

01:00:36.030 --> 01:00:39.330
effect of having a woman because
that doesn't affect

01:00:39.330 --> 01:00:41.345
people's preferences or because
there is a backlash.

01:00:45.410 --> 01:00:48.390
So we can think of these
two as different bins.

01:00:48.390 --> 01:00:50.160
We can even do one
study and not the

01:00:50.160 --> 01:00:53.090
other, of both of them.

01:00:53.090 --> 01:00:57.050
But they are kind of separate
tracks to which to go.

01:00:57.050 --> 01:01:02.830
So if we take a minute on
people's perception, do women

01:01:02.830 --> 01:01:08.380
affect the perception of voters
of women leaders?

01:01:08.380 --> 01:01:10.480
How would you go about
thinking about this?

01:01:10.480 --> 01:01:15.090
How would you go about trying to
measure people's political

01:01:15.090 --> 01:01:16.861
preferences?

01:01:16.861 --> 01:01:19.506
AUDIENCE: Asking questions about
the role of women in the

01:01:19.506 --> 01:01:21.190
household, or women
in the community.

01:01:21.190 --> 01:01:23.490
PROFESSOR: So you can
ask questions.

01:01:23.490 --> 01:01:25.340
So what is an issue
is this asking

01:01:25.340 --> 01:01:27.185
questions in this context?

01:01:27.185 --> 01:01:29.420
Do you like women leaders?

01:01:29.420 --> 01:01:30.994
What do you think of--

01:01:30.994 --> 01:01:32.965
AUDIENCE: People might feel like
there's a right answer.

01:01:32.965 --> 01:01:35.580
PROFESSOR: People may feel
there is a right answer.

01:01:35.580 --> 01:01:38.820
And what's the right answer, do
you think, in this context?

01:01:38.820 --> 01:01:41.280
AUDIENCE: I see that you can
tell the story right away that

01:01:41.280 --> 01:01:46.230
it's important to believe in
equality between genders.

01:01:46.230 --> 01:01:52.400
Or that if you're a male, in
front of your other male

01:01:52.400 --> 01:01:54.916
friends you want to appear
like a tough guy.

01:01:54.916 --> 01:01:56.302
AUDIENCE: Or who's asking
the questions.

01:01:56.302 --> 01:01:58.180
PROFESSOR: Or who is asking
the questions.

01:01:58.180 --> 01:02:00.240
Or you might think that it's
the right time to send a

01:02:00.240 --> 01:02:04.800
message to say those people
from the capital who are

01:02:04.800 --> 01:02:06.920
imposing this woman on us,
better tell them that really

01:02:06.920 --> 01:02:09.170
we don't like it.

01:02:09.170 --> 01:02:11.680
So it could really go either
way, we don't know.

01:02:11.680 --> 01:02:14.300
And in a sense that is something
that is interesting

01:02:14.300 --> 01:02:21.500
as well, is to ask which way is
what people are willing to

01:02:21.500 --> 01:02:22.480
reveal to you?

01:02:22.480 --> 01:02:25.170
In which direction is it best?

01:02:25.170 --> 01:02:29.636
So then we would need some
measure of "to" preferences.

01:02:29.636 --> 01:02:35.460
To willingness to consider
women as policymakers.

01:02:35.460 --> 01:02:38.148
AUDIENCE: Their voting
preferences in the next cycle?

01:02:38.148 --> 01:02:38.920
PROFESSOR: Yeah.

01:02:38.920 --> 01:02:42.100
So it seems that the litmus
test here would be voting.

01:02:42.100 --> 01:02:46.370
Which is after one cycle of
reservation, or two cycles of

01:02:46.370 --> 01:02:50.930
reservation, are women more
likely to be elected.

01:02:50.930 --> 01:02:54.880
And that seems to be that it
would be like the place easier

01:02:54.880 --> 01:02:57.060
to start or at least to end,
which is does it make a

01:02:57.060 --> 01:02:57.730
difference?

01:02:57.730 --> 01:03:03.590
And along the way, if you find
that you can think, well,

01:03:03.590 --> 01:03:05.550
maybe this is because
of various things.

01:03:05.550 --> 01:03:09.000
Maybe this is women started
to develop their networks.

01:03:09.000 --> 01:03:14.970
Maybe women figured out that
they could do this.

01:03:14.970 --> 01:03:18.310
So if we wanted to know that
it's really true, the change

01:03:18.310 --> 01:03:21.630
in the perception of women as
policy makers, you would try

01:03:21.630 --> 01:03:23.310
and get a measure
of perception.

01:03:23.310 --> 01:03:27.720
Again, more to eliminate the end
line result, and to try to

01:03:27.720 --> 01:03:31.908
get the measure of pure,
just one sort of--

01:03:31.908 --> 01:03:34.255
AUDIENCE: Could you also--

01:03:34.255 --> 01:03:36.790
so there are women elected
to these councils.

01:03:36.790 --> 01:03:37.570
PROFESSOR: Yeah.

01:03:37.570 --> 01:03:42.420
AUDIENCE: Could you also though
measure the women's

01:03:42.420 --> 01:03:46.785
ascendancy to other kinds of
more management posts, like at

01:03:46.785 --> 01:03:53.125
school level, or at community
groups or something like that,

01:03:53.125 --> 01:03:56.645
as stepping stones on
the way to that?

01:03:56.645 --> 01:03:57.900
PROFESSOR: Right.

01:03:57.900 --> 01:04:00.350
That goes into this margin
on women's participation.

01:04:02.910 --> 01:04:04.690
So you're thinking
[STUDENT NAME]

01:04:04.690 --> 01:04:07.070
earlier was thinking up, like
do we see more women--

01:04:07.070 --> 01:04:07.865
AUDIENCE: At higher levels.

01:04:07.865 --> 01:04:09.390
PROFESSOR: --going
at higher level.

01:04:09.390 --> 01:04:12.300
Or you could say do we see
more women in position of

01:04:12.300 --> 01:04:13.350
power within the village.

01:04:13.350 --> 01:04:14.150
AUDIENCE: Sink or swim.

01:04:14.150 --> 01:04:17.190
PROFESSOR: School headmasters,
things like that.

01:04:17.190 --> 01:04:20.020
School council, other
things like that.

01:04:20.020 --> 01:04:25.280
Going back to the perception
question, do you have an idea

01:04:25.280 --> 01:04:29.210
of how we could go about
measuring people's perception

01:04:29.210 --> 01:04:30.460
of women as policymakers?

01:04:33.195 --> 01:04:36.153
AUDIENCE: We could have an index
of the satisfaction they

01:04:36.153 --> 01:04:39.357
have with investments that women
have made, and try to

01:04:39.357 --> 01:04:44.041
relate it to how much they like
or dislike the woman who

01:04:44.041 --> 01:04:45.027
decided about this investment?

01:04:45.027 --> 01:04:47.700
PROFESSOR: So we could try
satisfaction of the goods.

01:04:47.700 --> 01:04:51.860
Thinking that in particular if
we had an objective quality of

01:04:51.860 --> 01:04:54.480
the goods, we could say that if
their satisfaction of the

01:04:54.480 --> 01:04:59.100
goods is lower even though the
goods are the same, it

01:04:59.100 --> 01:05:01.550
indicates that they
like women less.

01:05:01.550 --> 01:05:04.360
But that requires having a
very good measure of the

01:05:04.360 --> 01:05:06.120
quality of the goods.

01:05:06.120 --> 01:05:09.760
Otherwise, someone could always
say they are worse in

01:05:09.760 --> 01:05:12.330
some dimension that you
didn't observe.

01:05:12.330 --> 01:05:16.030
AUDIENCE: We could do some kind
of hypothetical question

01:05:16.030 --> 01:05:20.402
that involves psychology,
or you're trying to--

01:05:20.402 --> 01:05:22.280
PROFESSOR: So yeah, great.

01:05:22.280 --> 01:05:25.474
Continue in this direction.

01:05:25.474 --> 01:05:30.310
AUDIENCE: Where you're
interviewing people or doing

01:05:30.310 --> 01:05:36.850
surveys, and you asked them to
imagine a scenario where you

01:05:36.850 --> 01:05:41.450
have two different candidates,
and you give them the

01:05:41.450 --> 01:05:44.720
backgrounds and one's a man and
one is a woman, and ask

01:05:44.720 --> 01:05:46.620
them who they would pick.

01:05:46.620 --> 01:05:49.740
I mean, that would be a poorly
disguised question--

01:05:49.740 --> 01:05:53.420
PROFESSOR: If you have the two
candidates and you ask them to

01:05:53.420 --> 01:05:56.150
compare, you might get
into the same issue.

01:05:56.150 --> 01:05:58.690
But just go one step further
with this same idea.

01:06:03.080 --> 01:06:03.810
You're almost there.

01:06:03.810 --> 01:06:06.080
Or at least that's one
way of doing it.

01:06:09.778 --> 01:06:13.970
AUDIENCE: You could present two
hypothetical CVs to people

01:06:13.970 --> 01:06:16.110
and say, would like to
vote for this person.

01:06:16.110 --> 01:06:19.030
But on one of the hypothetical
CVs, make it up as a woman,

01:06:19.030 --> 01:06:22.915
and the other make it up as a
man, but they have all the

01:06:22.915 --> 01:06:23.365
same qualifications.

01:06:23.365 --> 01:06:26.200
And just see if there's a sense
that people would vote

01:06:26.200 --> 01:06:27.830
more for one versus the other.

01:06:27.830 --> 01:06:28.255
PROFESSOR: Exactly.

01:06:28.255 --> 01:06:28.960
You are [? interpreting ?]

01:06:28.960 --> 01:06:31.290
values to different people.

01:06:31.290 --> 01:06:35.050
Nothing forces you to present
the same questionnaire to

01:06:35.050 --> 01:06:36.540
every person.

01:06:36.540 --> 01:06:39.060
You could randomize within your
questionnaire what you're

01:06:39.060 --> 01:06:40.980
going to show them.

01:06:40.980 --> 01:06:43.120
So you're going to show them,
for example, exactly the same

01:06:43.120 --> 01:06:45.940
CV and ask them, are you going
to vote for this person?

01:06:45.940 --> 01:06:49.990
Or you could present a scenario
saying, so this and

01:06:49.990 --> 01:06:53.290
this happened, and there was a
choice to be made, and they

01:06:53.290 --> 01:06:55.610
decided to do this and
what do you think?

01:06:55.610 --> 01:06:57.350
Was it a good decision?

01:06:57.350 --> 01:07:01.040
And in one case it's like Mr. So
and So decided to do this,

01:07:01.040 --> 01:07:04.790
and in one case Mrs. So and
So decided to do this.

01:07:04.790 --> 01:07:06.860
If you ask both, you're
exactly right,

01:07:06.860 --> 01:07:07.860
it was poorly disguised.

01:07:07.860 --> 01:07:10.010
But if you ask only one, people

01:07:10.010 --> 01:07:13.950
will answer the question.

01:07:13.950 --> 01:07:16.820
And they don't have enough
information to really judge

01:07:16.820 --> 01:07:18.260
the person fully.

01:07:18.260 --> 01:07:20.520
Especially if you give them a
small scenario and then ask

01:07:20.520 --> 01:07:22.260
them, will you vote for them?

01:07:22.260 --> 01:07:25.280
So then we would bring in,
presumably, whatever else

01:07:25.280 --> 01:07:28.020
their other views
on the person.

01:07:28.020 --> 01:07:32.280
So then you will have compare
across surveys whether people

01:07:32.280 --> 01:07:40.160
tend to rank more highly the
survey that has Mr. So and So,

01:07:40.160 --> 01:07:43.850
versus the survey that
has Mrs. So and So.

01:07:43.850 --> 01:07:46.390
The way we did it in the follow
up study to the one

01:07:46.390 --> 01:07:50.090
that is in the case is we
actually taped speeches.

01:07:50.090 --> 01:07:53.410
So we had speeches that someone
had given, and we had

01:07:53.410 --> 01:07:56.930
a bunch of women record
the same speech.

01:07:56.930 --> 01:08:00.840
And we then played a tape and
said, so what do you think?

01:08:00.840 --> 01:08:03.080
So the advantage is, we don't
even have to insist on that

01:08:03.080 --> 01:08:04.880
it's Mr. So and So, it's
just would you

01:08:04.880 --> 01:08:05.820
listen to this speech.

01:08:05.820 --> 01:08:09.590
And the voice immediately tells
you and then you can

01:08:09.590 --> 01:08:11.800
compare people's answers.

01:08:11.800 --> 01:08:14.765
And so now it becomes you have
to compare people's answers to

01:08:14.765 --> 01:08:18.460
this question in the villages
that were reserved and the

01:08:18.460 --> 01:08:21.060
villages that were not reserved,
and see whether any

01:08:21.060 --> 01:08:22.790
gap changes.

01:08:22.790 --> 01:08:25.779
An interesting fact is that then
you can correlate that to

01:08:25.779 --> 01:08:30.160
what they actual tell you and
see whether, for example, it

01:08:30.160 --> 01:08:33.359
might be that the real gap
doesn't change, but what they

01:08:33.359 --> 01:08:35.650
tell you narrows, or it
could be the opposite.

01:08:35.650 --> 01:08:40.590
If they're trying to signal
their dislike of the policy or

01:08:40.590 --> 01:08:41.790
something like that.

01:08:41.790 --> 01:08:44.290
And at the end of the day, all
of this leads to sort of I'll

01:08:44.290 --> 01:08:47.819
find a hypothesis of did
this work, in a sense.

01:08:47.819 --> 01:08:50.029
Which would be what are
the vote shares

01:08:50.029 --> 01:08:52.370
for women in politics?

01:08:52.370 --> 01:08:55.819
Or if you were less ambitious,
something like what are the

01:08:55.819 --> 01:08:58.220
share of women who are
represented in positions of

01:08:58.220 --> 01:09:02.770
power within the village as
you suggested earlier.

01:09:02.770 --> 01:09:04.960
So that's something for day
two because we were on

01:09:04.960 --> 01:09:06.100
measurement of outcomes.

01:09:06.100 --> 01:09:11.130
I just wanted to give you one
example so we don't lack of

01:09:11.130 --> 01:09:14.189
various ways we can
collect outcomes.

01:09:14.189 --> 01:09:15.430
That's one way.

01:09:15.430 --> 01:09:22.620
Going back to the example of
do women leaders better

01:09:22.620 --> 01:09:26.270
represent the preferences
of women.

01:09:28.930 --> 01:09:33.470
So what we try and do with an
evaluation is to try to find

01:09:33.470 --> 01:09:37.200
out whether the program's just,
whether the program is

01:09:37.200 --> 01:09:42.950
effective, but also why it's
really more helpful?

01:09:42.950 --> 01:09:46.710
Both because well, we have a
richer understanding, and

01:09:46.710 --> 01:09:53.670
because it will enrich our
understanding of the program

01:09:53.670 --> 01:09:57.190
and also make it easier to draw
more general lessons.

01:09:57.190 --> 01:10:00.800
Because for example, this
program you evaluate in West

01:10:00.800 --> 01:10:02.580
Bengal and someone
can say, well we

01:10:02.580 --> 01:10:03.890
need to work in Rajasthan.

01:10:03.890 --> 01:10:07.360
So one way is to go and
do it in Rajasthan.

01:10:07.360 --> 01:10:09.070
But then you can say,
well, but it worked

01:10:09.070 --> 01:10:10.200
in Rajasthan, too.

01:10:10.200 --> 01:10:13.200
We need to work in South India
and it cannot replicate

01:10:13.200 --> 01:10:14.970
everywhere, everywhere.

01:10:14.970 --> 01:10:19.910
So eventually you want to have
the ability to say something.

01:10:19.910 --> 01:10:24.870
It's not necessarily we'll go
with some hypothesis, but to

01:10:24.870 --> 01:10:27.160
say something that what is your
take at the end of the

01:10:27.160 --> 01:10:31.815
day of the reasons why this
program will have such and

01:10:31.815 --> 01:10:32.830
such effect.

01:10:32.830 --> 01:10:37.220
So if you say, well what I think
is what's happening is

01:10:37.220 --> 01:10:41.510
when women are elected, when
there is a reservation for

01:10:41.510 --> 01:10:44.540
women they are doing stuff
that women prefer.

01:10:44.540 --> 01:10:47.470
Then you can say, well, in West
Bengal what they want is

01:10:47.470 --> 01:10:49.470
water so that's what
they'll do.

01:10:49.470 --> 01:10:52.580
In, say, Tamil Nadu our water's
not so important--

01:10:52.580 --> 01:10:55.310
well, it is very important to
me-- so in any place where

01:10:55.310 --> 01:10:58.210
it's not so much of an issue,
I guess that wouldn't be

01:10:58.210 --> 01:11:02.120
India, then they'll do that.

01:11:02.120 --> 01:11:05.000
So as long as you tell me what
women care about, I can tell

01:11:05.000 --> 01:11:08.360
you that it's going to
go in this direction.

01:11:08.360 --> 01:11:10.730
It has the advantage that it
makes the replication more

01:11:10.730 --> 01:11:15.480
interesting because if you're
doing North Bengal and then

01:11:15.480 --> 01:11:21.690
you say, well now some goods
go up, some goods go down.

01:11:21.690 --> 01:11:24.930
And then say you replicate in
Rajasthan a new level of tests

01:11:24.930 --> 01:11:27.100
that you are willing to
subject yourself to.

01:11:27.100 --> 01:11:31.190
If some goods go up, some goods
go down, it's like well,

01:11:31.190 --> 01:11:33.090
maybe you don't even need to
do the evaluation because

01:11:33.090 --> 01:11:34.600
probably you'll find it.

01:11:34.600 --> 01:11:40.860
Whereas after the West Bengal
study you say, whatever I find

01:11:40.860 --> 01:11:44.870
from this however imperfect way
to collect preferences is

01:11:44.870 --> 01:11:48.500
what women prefer, this is the
same direction it's going to

01:11:48.500 --> 01:11:50.100
move in Rajasthan.

01:11:50.100 --> 01:11:51.770
So, in fact, in here
in this case it's

01:11:51.770 --> 01:11:52.710
exactly what happened.

01:11:52.710 --> 01:11:56.990
Because we first did West
Bengal, and we find that women

01:11:56.990 --> 01:12:01.500
preferred water, according to
the measure of preferences,

01:12:01.500 --> 01:12:03.290
and goods went to water.

01:12:03.290 --> 01:12:06.360
And men prefer schools, which
was surprising to us and

01:12:06.360 --> 01:12:07.840
others, but that's the
way it is, and the

01:12:07.840 --> 01:12:09.740
goods went to schools.

01:12:09.740 --> 01:12:12.050
So now for Rajasthan, we say,
well we are going to do the

01:12:12.050 --> 01:12:12.500
same thing.

01:12:12.500 --> 01:12:15.370
Find out what women want, what
men want in the same way.

01:12:15.370 --> 01:12:17.200
It's going to go in
this direction.

01:12:17.200 --> 01:12:19.530
And we find that there in
Rajasthan, also women prefer

01:12:19.530 --> 01:12:23.100
water, but men love roads.

01:12:23.100 --> 01:12:25.030
So we should have less roads.

01:12:25.030 --> 01:12:26.470
More water, less roads.

01:12:26.470 --> 01:12:29.330
While in West Bengal, women also
likes the road, for the

01:12:29.330 --> 01:12:32.290
reason that they work on them.

01:12:32.290 --> 01:12:35.620
They are the people who do the
roads, so it's employment

01:12:35.620 --> 01:12:37.070
activities for them.

01:12:37.070 --> 01:12:41.460
So it's interesting because we
have different predictions.

01:12:41.460 --> 01:12:44.000
Our prediction is that the road
will go up in West Bengal

01:12:44.000 --> 01:12:46.490
where you have women
reservation, and they will go

01:12:46.490 --> 01:12:48.380
down in Rajasthan.

01:12:48.380 --> 01:12:50.580
We know why, because it's
related to the thing.

01:12:50.580 --> 01:12:54.500
And in fact we can make this
leap of faith beforehand.

01:12:54.500 --> 01:12:58.380
And so that makes it much more
powerful once you replicate,

01:12:58.380 --> 01:13:02.840
rather than I'll replicate
as I can.

01:13:02.840 --> 01:13:08.240
To say I will replicate with
a good sense of what I'm

01:13:08.240 --> 01:13:10.960
expecting to find.

01:13:10.960 --> 01:13:13.130
Which brings me to the same
thing which is our saying it's

01:13:13.130 --> 01:13:18.170
a very difficult ex-post not to
use the data to learn more

01:13:18.170 --> 01:13:22.210
than what the hypothesis was at
the beginning because, you

01:13:22.210 --> 01:13:23.660
know, it's sad.

01:13:23.660 --> 01:13:25.840
So you could think about it in a
more constructive way, which

01:13:25.840 --> 01:13:28.760
is you could think well, this is
what was my hypothesis was

01:13:28.760 --> 01:13:30.740
in the beginning, this
is what I find.

01:13:30.740 --> 01:13:34.490
In addition, I have also these
interesting, tantalizing

01:13:34.490 --> 01:13:36.690
tidbits of results.

01:13:36.690 --> 01:13:39.460
I'm putting it in front if you
admitting that it was not my

01:13:39.460 --> 01:13:43.130
hypothesis to start with, but I
am contesting that it should

01:13:43.130 --> 01:13:44.540
be the hypothesis of
the next study.

01:13:47.310 --> 01:13:50.340
I'll give you one very
good example of that.

01:13:50.340 --> 01:13:56.330
There was one project by David
McKenzie from the World Bank,

01:13:56.330 --> 01:14:00.570
and Suresh de Mel, who works
in Sri Lanka, and Chris

01:14:00.570 --> 01:14:04.860
Woodruff in UCSD, and they were
interested in the return

01:14:04.860 --> 01:14:07.720
to capitol for very small
entrepreneurs.

01:14:07.720 --> 01:14:11.730
So what they did is they gave
people in Sri Lanka a grant.

01:14:11.730 --> 01:14:15.060
Small entrepreneurs, people who
had about $200 of working

01:14:15.060 --> 01:14:17.460
capital, that's just
what we call a

01:14:17.460 --> 01:14:19.650
helicopter drop of money.

01:14:19.650 --> 01:14:23.420
So you get $100 grant
or $200 grant.

01:14:23.420 --> 01:14:26.260
And they did that and they found
that at first you get

01:14:26.260 --> 01:14:28.780
the average and they found
great returns to capital.

01:14:28.780 --> 01:14:30.590
Very high returns to capital.

01:14:30.590 --> 01:14:33.340
So very beneficial to give
people of the other

01:14:33.340 --> 01:14:35.100
5% percent a month.

01:14:35.100 --> 01:14:37.850
So very high return
to capital.

01:14:37.850 --> 01:14:39.320
Great.

01:14:39.320 --> 01:14:42.800
And then they decided to do it
separately at women and men.

01:14:42.800 --> 01:14:46.530
And oh, surprise, they found no
return whatsoever for the

01:14:46.530 --> 01:14:50.760
women, and huge return
for the men.

01:14:50.760 --> 01:14:55.350
So you can say, sorry, it was
not in your original design.

01:14:55.350 --> 01:14:58.430
It was not stratified by gender,
so we have really no

01:14:58.430 --> 01:15:00.010
intents that you are--
so we have to

01:15:00.010 --> 01:15:02.090
throw this result away.

01:15:02.090 --> 01:15:03.910
Of course, we don't want to
throw this result away because

01:15:03.910 --> 01:15:06.690
that's so surprising and
striking that we kind of want

01:15:06.690 --> 01:15:08.360
to think about it.

01:15:08.360 --> 01:15:09.910
So what's the idea?

01:15:09.910 --> 01:15:13.140
You write this up being very
explicit that we found this

01:15:13.140 --> 01:15:16.570
ex-post, but it seems
like really robust.

01:15:16.570 --> 01:15:19.700
We are going to go and do a new
experiment, so we could

01:15:19.700 --> 01:15:22.630
redo it in Sri Lanka or do
it in somewhere else.

01:15:22.630 --> 01:15:27.730
And in this case, our hypothesis
is the age zero is

01:15:27.730 --> 01:15:29.950
the return to capital are the
same for men and women, that's

01:15:29.950 --> 01:15:31.750
what you are trying to reject.

01:15:31.750 --> 01:15:34.230
So these scores are good
to think about.

01:15:34.230 --> 01:15:38.090
These evaluations are part of
a process, we are not alone.

01:15:38.090 --> 01:15:39.790
A lot of people are working
on this, there will be

01:15:39.790 --> 01:15:42.570
replication either by
you or by others.

01:15:42.570 --> 01:15:47.870
And being explicit up front
about what was your first

01:15:47.870 --> 01:15:49.990
hypothesis and your
current model.

01:15:49.990 --> 01:15:53.050
And was is it that you
found out as well?

01:15:53.050 --> 01:15:56.070
Our goal is to not get mixed up,
and at the same time not

01:15:56.070 --> 01:15:57.930
to lose the information
that is going to be

01:15:57.930 --> 01:15:59.180
useful in the future.

01:16:05.610 --> 01:16:10.010
So we stop at 12:00 whenever
we started, or what are the

01:16:10.010 --> 01:16:11.565
social norms in this?

01:16:11.565 --> 01:16:12.912
AUDIENCE: As long as you want.

01:16:12.912 --> 01:16:16.050
PROFESSOR: Oh.

01:16:16.050 --> 01:16:18.080
I have a phone call at
12:30, otherwise--

01:16:18.080 --> 01:16:19.280
AUDIENCE: It's 12:15.

01:16:19.280 --> 01:16:23.510
PROFESSOR: I'd probably be
finished before that.

01:16:23.510 --> 01:16:26.840
I don't have time to finish
what's on the slides anyway.

01:16:26.840 --> 01:16:31.460
So I just wanted to give you
a sense of what might be a

01:16:31.460 --> 01:16:32.710
causal model in this case.

01:16:35.670 --> 01:16:43.100
The whole perception and goal is
not there, it's the public

01:16:43.100 --> 01:16:45.056
goods thing.

01:16:45.056 --> 01:16:49.350
And it's a way of disciplining
all of the outcomes, as well

01:16:49.350 --> 01:16:52.090
as the various things
we spoke about.

01:16:52.090 --> 01:16:54.800
So you start from reservation,
so one thing that reservations

01:16:54.800 --> 01:16:56.360
definitely do is that
they will lead

01:16:56.360 --> 01:16:59.520
to more women Pradhan.

01:16:59.520 --> 01:17:02.790
And then the question is whether
or not having more

01:17:02.790 --> 01:17:08.420
women Pradhan will change the
public good, and in what way?

01:17:08.420 --> 01:17:13.120
And there are really two
channels to change the

01:17:13.120 --> 01:17:15.190
preferences, which we
have discussed.

01:17:15.190 --> 01:17:20.830
One is the women as the Pradhan
do what women want.

01:17:20.830 --> 01:17:23.590
And you've not really discussed
that, but that comes

01:17:23.590 --> 01:17:26.840
with its own set of assumptions
which is on the

01:17:26.840 --> 01:17:31.680
one hand the Pradhan are not
representing the majority.

01:17:31.680 --> 01:17:33.750
As always, the majority
hasn't changed.

01:17:33.750 --> 01:17:39.200
We shouldn't see a difference
because even if you are

01:17:39.200 --> 01:17:42.680
saying, well, you don't have to
be accountable to the men.

01:17:42.680 --> 01:17:45.660
That's not true, because they
still do it for you.

01:17:45.660 --> 01:17:45.830
Ex-ante.

01:17:45.830 --> 01:17:53.500
Several woman compete, and the
issue is what platform are you

01:17:53.500 --> 01:17:55.930
going to run on?

01:17:55.930 --> 01:17:59.020
Well you should be running on
the platform that is going to

01:17:59.020 --> 01:18:00.790
get you elected.

01:18:00.790 --> 01:18:03.440
And whether or not you're a man
or a woman, you're elected

01:18:03.440 --> 01:18:05.540
by the same group of people.

01:18:05.540 --> 01:18:09.700
So in a totally stand out
model where democracy is

01:18:09.700 --> 01:18:12.720
perfect, who is in charge
doesn't matter.

01:18:12.720 --> 01:18:17.320
Because who is in charge is
representing the desire of the

01:18:17.320 --> 01:18:20.580
majority, what we call
the median voter.

01:18:20.580 --> 01:18:23.510
So if you had perfect democracy,
that channel would

01:18:23.510 --> 01:18:27.680
be killed, and we wouldn't
see an impact.

01:18:27.680 --> 01:18:31.950
On the other hand, if all the
decisions were made by a group

01:18:31.950 --> 01:18:36.260
of elite villagers, that again
wouldn't matter because who is

01:18:36.260 --> 01:18:38.180
in charge doesn't matter.

01:18:38.180 --> 01:18:44.290
So the identity of the Pradhan
is going to make a difference

01:18:44.290 --> 01:18:46.330
only in some

01:18:46.330 --> 01:18:48.810
middle-of-the-road kind of [? ward ?]

01:18:48.810 --> 01:18:53.280
where the politician has some
control over what is going on.

01:18:56.570 --> 01:18:59.170
It is not completely controlled
by a bureaucracy or

01:18:59.170 --> 01:19:01.600
by elite, and is not completely

01:19:01.600 --> 01:19:03.550
accountable to the people.

01:19:03.550 --> 01:19:05.190
But who is he?

01:19:05.190 --> 01:19:09.100
He cannot fully commit, for
example, to a platform.

01:19:09.100 --> 01:19:10.660
That doesn't seem unrealistic.

01:19:10.660 --> 01:19:12.660
People make electoral promises
and sometimes

01:19:12.660 --> 01:19:14.770
they go against them.

01:19:14.770 --> 01:19:18.820
Almost never, but sometimes.

01:19:18.820 --> 01:19:21.920
But that's something.

01:19:21.920 --> 01:19:27.750
So if we do learn that the
public good prefers to change,

01:19:27.750 --> 01:19:30.770
we have learned something
broader than just impact to

01:19:30.770 --> 01:19:32.410
this program.

01:19:32.410 --> 01:19:37.790
We have learned something about
politics in India, which

01:19:37.790 --> 01:19:39.760
is OK, there is some
democracy.

01:19:39.760 --> 01:19:40.960
Some people are contesting
that.

01:19:40.960 --> 01:19:46.390
Some people say the Panchayat
is just a face.

01:19:46.390 --> 01:19:49.510
There is no democracy really.

01:19:49.510 --> 01:19:52.570
So if you find a difference by
the identity of the Pradhan,

01:19:52.570 --> 01:19:57.050
it shows you by the by that
there is some reality in the

01:19:57.050 --> 01:19:59.100
democratic system, but it's
not perfect democracy.

01:19:59.100 --> 01:20:01.100
So if we've learned something,
it can be broader than a new

01:20:01.100 --> 01:20:02.540
program as well.

01:20:02.540 --> 01:20:06.810
And that is a lesson you
can take elsewhere.

01:20:06.810 --> 01:20:12.300
Another channel by which a woman
having reservation can

01:20:12.300 --> 01:20:16.500
influence the representation
of women is to more

01:20:16.500 --> 01:20:17.160
representation.

01:20:17.160 --> 01:20:19.620
For example, we were talking
about more political

01:20:19.620 --> 01:20:22.290
participation of women if
the woman is the head.

01:20:22.290 --> 01:20:25.920
So one thing that could be
true is that they're more

01:20:25.920 --> 01:20:27.450
likely to show up in meetings,
that they're more

01:20:27.450 --> 01:20:28.900
likely to speak up.

01:20:28.900 --> 01:20:31.530
For example, because the women
have said that the Pradhan has

01:20:31.530 --> 01:20:34.510
to be at the village meeting,
so she better put it at the

01:20:34.510 --> 01:20:36.040
time where she can go.

01:20:36.040 --> 01:20:38.580
So not in the middle of
the night in a field.

01:20:38.580 --> 01:20:43.080
And so as long as she can
go, a woman can also go.

01:20:43.080 --> 01:20:53.060
So to either of these channels,
you'd have the fact

01:20:53.060 --> 01:20:55.710
that the public good will
reflect better women's

01:20:55.710 --> 01:20:57.210
preferences.

01:20:57.210 --> 01:21:00.500
We have to add another
assumption, is that women have

01:21:00.500 --> 01:21:02.720
different preferences.

01:21:02.720 --> 01:21:05.770
And if that's the case, then
the public good will be

01:21:05.770 --> 01:21:10.180
different in a specific way,
which is towards those

01:21:10.180 --> 01:21:12.150
different preferences.

01:21:12.150 --> 01:21:17.280
And then you might have
different outcomes.

01:21:17.280 --> 01:21:22.080
More income for the women,
better health and education

01:21:22.080 --> 01:21:25.130
outcome, if it comes out that
it's what women care about.

01:21:25.130 --> 01:21:29.720
And you'll be able to follow
exactly the trace that, you

01:21:29.720 --> 01:21:31.960
know, if you find like in
West Bengal more water.

01:21:31.960 --> 01:21:34.410
Maybe you're going to be
interested in diarrhea.

01:21:37.490 --> 01:21:39.800
If you find less schools, you're
going to be interested

01:21:39.800 --> 01:21:42.430
in education to see whether
education goes down, and

01:21:42.430 --> 01:21:43.930
things like that.

01:21:43.930 --> 01:21:46.120
So now we have the
complete channel.

01:21:46.120 --> 01:21:49.630
And we can now think about all
of our variables that we had

01:21:49.630 --> 01:21:52.780
collected, and we're going to
slot them into, well, what are

01:21:52.780 --> 01:21:55.040
they going to do for us?

01:21:55.040 --> 01:21:57.400
So all the public
goods go here.

01:21:57.400 --> 01:22:00.020
We need to collect women's
preferences

01:22:00.020 --> 01:22:02.810
somewhere, it will go here.

01:22:02.810 --> 01:22:08.180
We want to know as a woman are
empowered, so we are going to

01:22:08.180 --> 01:22:11.700
be collecting all of these.

01:22:11.700 --> 01:22:13.730
Whether people come to the
meeting, whether they speak

01:22:13.730 --> 01:22:16.420
up, how they are answered
to once they speak up.

01:22:16.420 --> 01:22:19.270
This is all going to go here.

01:22:19.270 --> 01:22:23.130
So now, if we have an infinite
amount of money we're still

01:22:23.130 --> 01:22:26.910
going to collect a large amount
of data, but we know in

01:22:26.910 --> 01:22:29.420
advance what it is we are
going to do with them.

01:22:29.420 --> 01:22:32.050
We can write it down, put it
in an envelope, send it to

01:22:32.050 --> 01:22:37.550
your grandmother, and this is
the thing that really gives a

01:22:37.550 --> 01:22:39.340
lot of credibility to what
you're going to do.

01:22:42.100 --> 01:22:44.340
Another version of sending it
to your grandmother that we

01:22:44.340 --> 01:22:48.340
are going to try and implement
here at J-PAL is to allow you

01:22:48.340 --> 01:22:53.590
to put it on a website, to
upload it somewhere where

01:22:53.590 --> 01:22:56.085
nobody can see it but you,
but it's secure,

01:22:56.085 --> 01:22:57.500
and the data is mapped.

01:22:57.500 --> 01:23:02.690
So this is whatever it was your
analysis plan at the time

01:23:02.690 --> 01:23:04.580
of the beginning
of your study.

01:23:04.580 --> 01:23:08.650
So you are tying your hand
behind your back.

01:23:12.458 --> 01:23:17.761
AUDIENCE: Excuse me, but
why would you do that?

01:23:17.761 --> 01:23:19.340
PROFESSOR: Because
we want to--

01:23:19.340 --> 01:23:22.500
or maybe someone can answer
that question.

01:23:22.500 --> 01:23:23.570
Why do we do that?

01:23:23.570 --> 01:23:26.520
AUDIENCE: I guess it's just so
people aren't relying on your

01:23:26.520 --> 01:23:27.550
personal integrity.

01:23:27.550 --> 01:23:30.000
You're saying this was
our hypothesis.

01:23:30.000 --> 01:23:33.086
We didn't ex-post change
our hypothesis.

01:23:33.086 --> 01:23:34.040
PROFESSOR: Right.

01:23:34.040 --> 01:23:37.710
So the reason why you want to
say what was your hypothesis

01:23:37.710 --> 01:23:40.480
in advance, is because then
you can attest it.

01:23:40.480 --> 01:23:42.570
Whereas if you take something
you can always reverse

01:23:42.570 --> 01:23:43.480
engineer it.

01:23:43.480 --> 01:23:48.040
And again, I have no big issue
with that, with reverse

01:23:48.040 --> 01:23:48.780
engineering.

01:23:48.780 --> 01:23:51.000
Personally, I think it's useful,
but it needs to be

01:23:51.000 --> 01:23:54.010
very clear what was there before
and what was reverse

01:23:54.010 --> 01:23:56.810
engineered after, otherwise
we have no notion

01:23:56.810 --> 01:23:59.096
of statistical tests.

01:23:59.096 --> 01:24:01.510
AUDIENCE: I'm just pushing out
a little bit further, too.

01:24:01.510 --> 01:24:04.070
What if you just had a question
where you just don't

01:24:04.070 --> 01:24:04.890
know, right?

01:24:04.890 --> 01:24:07.490
You're like, wow, we really
think there might be an effect

01:24:07.490 --> 01:24:10.500
of this on that, or we're not
sure whether the effect would

01:24:10.500 --> 01:24:11.590
be upward or downward.

01:24:11.590 --> 01:24:15.250
Do you still have to just pick
one for the sake of having a

01:24:15.250 --> 01:24:16.190
hypothesis that you're
testing?

01:24:16.190 --> 01:24:22.000
PROFESSOR: I think you wouldn't
want to embark in an

01:24:22.000 --> 01:24:26.390
evaluation without at least
having a sense of why it would

01:24:26.390 --> 01:24:28.380
go up and why it
would go down.

01:24:28.380 --> 01:24:30.990
So take this specific example.

01:24:30.990 --> 01:24:33.900
To start with, you shouldn't
really know whether the water

01:24:33.900 --> 01:24:36.460
wells are going to go up and
down, because it's going to

01:24:36.460 --> 01:24:40.510
depend on what women want.

01:24:40.510 --> 01:24:42.105
So I'm not making a stance.

01:24:45.500 --> 01:24:50.200
It would be silly to write
to my grandmother.

01:24:50.200 --> 01:24:54.580
I'm betting that the women
water wells will go up.

01:24:54.580 --> 01:24:56.210
Because they could go
up or down depending

01:24:56.210 --> 01:24:56.990
on what women want.

01:24:56.990 --> 01:24:58.880
Of course, you may have a very
strong prior that it's what

01:24:58.880 --> 01:25:05.500
women want, but the statement
would be of the form, if it is

01:25:05.500 --> 01:25:08.450
the case that women have a
strong preference for water,

01:25:08.450 --> 01:25:10.520
water should go up.

01:25:10.520 --> 01:25:12.400
So if you have some uncertainty,
it's probably

01:25:12.400 --> 01:25:14.250
because there is an if somewhere
that you're not

01:25:14.250 --> 01:25:18.200
thinking about that you're
expliciting now.

01:25:18.200 --> 01:25:20.480
I think if thinking sufficiently
hard about

01:25:20.480 --> 01:25:23.340
something, you can know
in what condition you

01:25:23.340 --> 01:25:24.310
would go up and down.

01:25:24.310 --> 01:25:26.310
There are a lot of programs
which could go up and down,

01:25:26.310 --> 01:25:28.830
that's why we evaluate them.

01:25:28.830 --> 01:25:32.930
I mean, sometimes you think that
they should really go up,

01:25:32.930 --> 01:25:34.550
but it could also be zero.

01:25:34.550 --> 01:25:40.210
And it's good to know, if this
and this happened then this

01:25:40.210 --> 01:25:41.520
effect would be expected.

01:25:41.520 --> 01:25:44.240
If this and this doesn't happen,
then I wouldn't see

01:25:44.240 --> 01:25:45.610
this effect, so you
could write that.

01:25:48.620 --> 01:25:51.470
And in fact, I think these
types of statements, in a

01:25:51.470 --> 01:25:54.670
sense, are almost
more informative

01:25:54.670 --> 01:25:57.720
than this will happen.

01:26:00.360 --> 01:26:01.810
So let me stop here.

01:26:01.810 --> 01:26:07.217
What's in the rest of
the slides is kind

01:26:07.217 --> 01:26:10.010
of little bit random--

01:26:10.010 --> 01:26:13.080
not randomized, but not random
in the sense of randomized,

01:26:13.080 --> 01:26:18.500
but random in the sense of every
which way I pass out the

01:26:18.500 --> 01:26:23.460
advice on how to collect data
and how to enter data and

01:26:23.460 --> 01:26:24.510
things like that.

01:26:24.510 --> 01:26:27.750
For those of you who are going
to IPA training after this,

01:26:27.750 --> 01:26:30.200
you're going to be sick of it
by the end of the three days

01:26:30.200 --> 01:26:31.360
so that's not needed.

01:26:31.360 --> 01:26:38.330
For the other ones, it's in the
slides and it's really--

01:26:38.330 --> 01:26:40.750
that's the problem of it
being too short anyway.

01:26:40.750 --> 01:26:45.330
So it is pretty self-explanatory
and

01:26:45.330 --> 01:26:49.910
relatively common sense and not
sufficient anyway, but a

01:26:49.910 --> 01:26:51.660
starting point.

01:26:51.660 --> 01:26:52.910
Thank you very much.