00:00:00
PROFESSOR: So what I would like
you to do is to answer
00:00:04
the following question.
00:00:06
"Do you have..?" oh, now
wait a second -- this
00:00:10
says polling closed.
00:00:11
Let's see whether we can
get polling to open.
00:00:14
Do you...?" -No, that
didn't work.
00:00:16
Oh, this is the most desperately
horrible thing
00:00:18
that I've ever had happen.
00:00:21
I've had horrible things happen
that are worse than
00:00:23
this, I'm sorry.
00:00:25
I don't want to belittle all
the really horrible things
00:00:27
that have ever happened to me.
00:00:29
So let's try again.
00:00:31
Click practicalities
blah blah.
00:00:33
OK. "Do you have
a clicker yet?
00:00:36
Polling is open." OK.
00:00:39
And there's ten seconds left.
00:00:41
You'll see this little
timeline.
00:00:43
Eight, seven, six, five, four--
ninety two of you, so
00:00:46
many of you have clickers.
00:00:48
Let's see what we learned.
00:00:51
It looks like 97% of
you have clickers.
00:01:00
I actually began with this
exercise to make a point about
00:01:07
psychological research.
00:01:10
We just made a classic and
dangerous mistake, a mistake
00:01:19
known as sampling bias.
00:01:23
We used a measure which doesn't
give us accurate
00:01:28
information about what we
wanted to find out.
00:01:31
We wanted to find out, what
percentage of people in the
00:01:35
room had clickers.
00:01:37
And what we found out instead
was, what percentage of people
00:01:40
in the room with clickers
had clickers.
00:01:45
I don't know who you guys are.
00:01:49
This error is an incredibly
dangerous one, and it could
00:01:55
Suppose I then ask you a
question whether you're from
00:01:58
the class of 2014, '13, '12,
or '11, and discovered that
00:02:04
80% of you were freshmen
and sophomores.
00:02:08
There, then, I would inherited
exactly the sampling bias
00:02:12
error that we just observed.
00:02:15
Freshmen and sophomores
have handed in
00:02:18
their schedules already.
показать еще
русский▼
English
