1
00:00:00,060 --> 00:00:03,464
- [Yan Chen] It's good that we have
an army of enthusiasts

2
00:00:03,464 --> 00:00:05,390
writing Wikipedia articles,

3
00:00:05,960 --> 00:00:10,868
but sometimes when it concerns 
a disease that I might have,

4
00:00:10,877 --> 00:00:13,494
I really want the experts' input.

5
00:00:13,790 --> 00:00:15,574
♪ [music] ♪

6
00:00:26,647 --> 00:00:30,240
Wikipedia is one of the most 
important references

7
00:00:30,240 --> 00:00:31,932
for the general public

8
00:00:31,991 --> 00:00:36,818
It's actually one of the top five
most visited websites in the world.

9
00:00:37,051 --> 00:00:40,900
Everyone reads Wikipedia articles,
but sometimes you spot an error

10
00:00:40,900 --> 00:00:44,101
or you say, 
"Well, this is not really correct."

11
00:00:44,101 --> 00:00:45,418
But you move on

12
00:00:45,418 --> 00:00:48,219
and say, "Someone else might fix it."

13
00:00:51,108 --> 00:00:53,694
That's called the "free rider problem."

14
00:00:55,131 --> 00:00:58,702
The success of Wikipedia 
has been really surprising

15
00:00:58,702 --> 00:01:03,069
for economists because it relies 
purely on volunteer labor.

16
00:01:03,437 --> 00:01:07,969
The medical profession has found
that patients tend to bring printouts

17
00:01:07,969 --> 00:01:10,785
of Wikipedia articles 
to their doctor's office.

18
00:01:11,515 --> 00:01:14,223
Some of these articles 
are of low quality

19
00:01:14,224 --> 00:01:16,528
because they were not written by experts.

20
00:01:16,528 --> 00:01:21,411
We're trying to figure out 
what are the some of the motivators

21
00:01:21,411 --> 00:01:24,800
to get experts to contribute 
to high quality content.

22
00:01:25,064 --> 00:01:28,832
So we decided to do a field experiment
to tease out the causalities,

23
00:01:28,832 --> 00:01:32,332
to figure out what motivates people
to contribute to Wikipedia,

24
00:01:32,332 --> 00:01:34,927
whether it's social impact 
or private benefit

25
00:01:34,927 --> 00:01:39,081
or public acknowledgement,
or a combination of these factors.

26
00:01:42,286 --> 00:01:44,603
So in this study,
in this field experiment,

27
00:01:44,603 --> 00:01:47,970
we contacted about 4,000
academic economists.

28
00:01:47,970 --> 00:01:49,439
We have a generic message

29
00:01:49,439 --> 00:01:52,805
which says Wikipedia 
is a very valuable public good,

30
00:01:52,805 --> 00:01:57,620
and yet lots of the articles 
are inaccurate or not up to date.

31
00:01:57,620 --> 00:02:01,089
Would you spend 
10 to 15 minutes commenting

32
00:02:01,089 --> 00:02:02,871
on these Wikipedia articles?

33
00:02:02,871 --> 00:02:06,205
Then we vary the paragraphs 
depending on whether

34
00:02:06,205 --> 00:02:08,322
they're in the treatment 
or the control group.

35
00:02:08,861 --> 00:02:12,061
In the control group, 
we don't mention that the articles

36
00:02:12,061 --> 00:02:13,885
might cite your research.

37
00:02:14,838 --> 00:02:19,129
And in the private benefit condition 
we say they might cite your research,

38
00:02:19,129 --> 00:02:21,962
and we have another condition
which says, "We will publicly

39
00:02:21,962 --> 00:02:24,607
acknowledge your contributions."

40
00:02:27,153 --> 00:02:30,706
Simply asking the expert,
"Would you contribute?"

41
00:02:30,837 --> 00:02:33,938
you get a pretty high response rate,

42
00:02:34,027 --> 00:02:37,662
which is about 45% of the people
say, "Yes, I'm willing."

43
00:02:37,662 --> 00:02:41,510
When we sent out the links,
it turns out a third of the people

44
00:02:41,510 --> 00:02:45,532
actually contributed,
and we look at what are the features

45
00:02:45,532 --> 00:02:47,525
that predict contributions.

46
00:02:48,011 --> 00:02:53,450
It turns out that if the article is 
really well-matched

47
00:02:53,720 --> 00:02:55,866
to their research expertise,

48
00:02:56,107 --> 00:02:58,506
they're much more likely 
to contribute,

49
00:02:58,587 --> 00:03:00,887
and they're contributing 
higher quality content.

50
00:03:01,371 --> 00:03:04,629
So good matching is really 
important for volunteering.

51
00:03:04,945 --> 00:03:08,646
We also try to figure out
are people more motivated

52
00:03:08,646 --> 00:03:12,194
by the private benefits,
what they get out

53
00:03:12,194 --> 00:03:13,946
of the contributions.

54
00:03:13,946 --> 00:03:16,695
So we do that by telling 
the treatment group

55
00:03:16,695 --> 00:03:19,710
that we'll send you articles
to comment on

56
00:03:19,710 --> 00:03:21,578
that might reference your research.

57
00:03:21,578 --> 00:03:25,984
So it turns out that knowing
that you might be cited

58
00:03:25,984 --> 00:03:30,894
increases the positive response rate
by about 13%.

59
00:03:31,594 --> 00:03:35,195
We also find 
that the public acknowledgement,

60
00:03:35,195 --> 00:03:37,962
saying that we will post 
your contributions

61
00:03:37,962 --> 00:03:40,923
and acknowledge 
your contributions publicly,

62
00:03:40,923 --> 00:03:44,794
people are more likely 
to provide high quality content.

63
00:03:46,011 --> 00:03:50,095
And public impact --
you know we vary the views

64
00:03:50,095 --> 00:03:52,861
of the Wikipedia articles
that we sent.

65
00:03:52,861 --> 00:03:57,444
We either say on average,
a Wikipedia will get 426 views.

66
00:03:57,444 --> 00:04:01,934
But we'll send articles which have 
at least 1,000 views to you.

67
00:04:01,934 --> 00:04:07,623
People are most motivated
when the private benefit

68
00:04:07,623 --> 00:04:09,773
is combined with the social impact.

69
00:04:09,773 --> 00:04:12,745
The social impact 
by itself actually doesn't

70
00:04:12,745 --> 00:04:14,706
quite have the same effect.

71
00:04:18,923 --> 00:04:21,790
I think if we replicate it 
in other fields

72
00:04:21,790 --> 00:04:24,831
we'll have more confidence
that private benefits,

73
00:04:24,831 --> 00:04:28,474
such as citation benefits,
would get people interested

74
00:04:28,474 --> 00:04:31,219
in contributing, 
and citation benefits

75
00:04:31,219 --> 00:04:33,102
in combination 
with social impact

76
00:04:33,102 --> 00:04:35,502
would have a larger effect.

77
00:04:35,604 --> 00:04:38,653
We need to push it 
to other fields as well

78
00:04:38,653 --> 00:04:42,621
to see if they're robust
across different communities.

79
00:04:43,270 --> 00:04:45,388
- [Narrator] Want to see more
economists in the wild?

80
00:04:45,388 --> 00:04:46,760
Check out our playlist.

81
00:04:46,760 --> 00:04:47,843
Are you a teacher?

82
00:04:47,843 --> 00:04:49,877
Here's some related material 
for your classroom.

83
00:04:49,877 --> 00:04:51,225
Want to dive deeper?

84
00:04:51,225 --> 00:04:54,377
Wikipedia is what economists 
call a "public good."

85
00:04:54,377 --> 00:04:56,278
Learn more by watching this video.

86
00:04:56,698 --> 00:04:58,231
♪ [music] ♪