Jump to content

Wiki Guides

From Meta, a Wikimedia project coordination wiki

To do

[edit]

Needs normalized charts for days on study.

Introduction

[edit]

Daily reporting

[edit]

Snapshot metrics

[edit]

As always, more detailed metrics are available on our workspace, and analysis follows.

Date Mentored edits (all) Control edits (all) Mentored edits (mainspace) Control edits (Mainspace)
2/26/2011 185 229
2/27/2011 117 65
2/28/2011 245 50
3/1/2011 148 198
3/2/2011 104 39
3/3/2011 64 98
3/4/2011 68 54 21 38
3/5/2011 15 42 10 13
3/6/2011 13 63 13 33
3/7/2011 119 15 108 3
3/8/2011 40 29 32 23
3/9/2011 614 853 336 429
3/10/2011 118 273 83 126
3/11/2011 212 225 166 148
3/12/2011 194 138 125 88
3/13/2011 189 149 136 49
3/14/2011 140 143 96 88
3/15/2011 170 119 67 51
3/16/2011 112 128 56 74
3/17/2011 75 216 47 120
3/18/2011 146 142 74 73
3/19/2011 126 240 109 42
3/20/2011 103 221 56 42
3/21/2011 174 185 75 84
3/22/2011 64 127 38 59
3/23/2011 66 92 53 33
3/24/2011 79 117 47 14
3/25/2011 48 110 34 46
3/26/2011 117 154 96 82
3/27/2011 94 153 58 87
3/28/2011 114 95 64 72
3/29/2011 82 145 58 96
3/30/2011 86 160 76 57
3/31/2011 55 134 34 30
4/1/2011 55 35 28 6
4/2/2011 518 99 278 54
4/3/2011 376 69 207 27
4/4/2011 600 183 362 146
4/5/2011
4/6/2011
Total 5845 5587 3043 2333

Analysis

[edit]

On April 4, we began to look intensively at the statistics. Our mentees are compared to a control group.

We found some interesting things, which we'll go through here, but the key points are:

  • First, we can make a difference in editing behavior. We're seeing that mentored users are contributing more to the encyclopedia, and spending less time on user and user talk space.
  • We've got to get to them the day that they create their account. Waiting three days is too long, and we lose them.


The first metric we looked at was the number of edits in each group, per day. Initially, we narrowed to just edits within the Wikipedia namespace, as on this chart:

We charted the difference between the number of edits in the mentored group and the number of edits in the control. In this case, numbers below zero reflect more edits by the control. Positive numbers are better. The average is +22.

Obviously, this is good for our mentored users! They're editing the mainspace more.

So let's look now at the same trending for the User and User talk namespaces:

Notice that aberration at the front of the graph? We'll get to that in a second.

And again, the difference:

The average is -0.4. Our mentored users are editing User and User talk space LESS.

Remember the aberration we talked about on the previous graph? Here's why:

By the 3rd day of their editing, we'd already lost many of the new users. This graph is from our first cohort. But the one from the second cohort is almost the same.

This means we've got to get to them faster to offer mentorship: before they lose interest.

Blocks