# What statistics would be interesting for a usage report?

Suppose MathOverflow were to start having quarterly or annual usage reports. What statistics would you like to see reported there?

Some are obvious — new questions, new answers, new users, etc. — but some are much less obvious. For example, the number of new votes on posts more than three months old is perhaps an interesting indicator of how much people use MathOverflow as a knowledge base rather than a place to ask questions.

• How many upvoted answers from all users with a name beginning with user. Sep 28, 2013 at 18:35
• @Bill: 685 - data.stackexchange.com/mathoverflow/query/139655/… - 894 if you include CW answers - data.stackexchange.com/mathoverflow/query/139656/… Sep 28, 2013 at 18:59
• Perhaps some measure of migration to & from MSE. The to may be easier to quantify than the from, as the latter generally occurs via reposting (e.g., this digit-sum posting). Sep 29, 2013 at 14:04
• Most questions, I suspect, are asked, up/downvoted, viewed, possibly edited and answered within a relatively short time, and then all such activity ceases until somebody stumbles upon it later. How true or typical is this description? I would be interested in seeing statistics relating to views/votes/answers but measured against the time since the question was first posed. Oct 1, 2013 at 17:15
• The histogram of time duration between a response getting upvoted and the next upvote on the question (or vice versa). The time duration between when the user lands on a page and when he casts his vote there. votes on a post vs the number of links it has. No of views of the page vs the modal reputation it has.
– ARi
Oct 2, 2013 at 11:25

## Number of new users per month in 2013

Generated using this query.

      All  >100  >200  >500 (rep)
Jan   406    76    14     4
Feb   442    66    11     2
Mar   450    70    11     2
Apr   497    55    12     1
May   558    79    21     6
Jun   905   282    14     4
Jul  1458   434    16     4
Aug  1278   307    22     5
Sep  1229   263    11     6
Oct  1383   278     5     1
Nov  1269   239     4     1
Dec  1328   304     4     1

• Perhaps it's worth tweaking it, and add >101 column as well.
– Asaf Karagila Mod
Jan 12, 2014 at 9:44
• Does >500 mean greater than 500 when joining? or greater than 500 as of the time of the query? Jan 12, 2014 at 13:54
• @GerryMyerson: At the time of the query. I'm trying to think of a better query here but it's not easy to correctly tally the votes. Jan 12, 2014 at 19:15

## Number of closed questions per month in 2013

Generated using this query, giving the number of questions asked during a certain month that were at the time of the query closed questions.

Jan 26
Feb 23
Mar 22
Apr 29
May 22
Jun 44
Jul 33
Aug 32
Sep 49
Oct 50
Nov 49
Dec 69


• I can't say I'm not perplexed but the steep incline could be due to the auto-deletion roomba. Jan 12, 2014 at 4:25
• Thanks for all the data in this thread! But I think for this one the heading should really be 'closed and not (yet) deleted' or something like this. There should be a huge difference between question that 'were closed at some point' and those showing up here (likely, fully explaining the increase towards the end). I am sure we had a lot more than one closed question per day at the beginning of last year.
– user9072
Jan 12, 2014 at 14:57
• @quid: I don't agree since the title describes the query very precisely. Personally, I think it's better to discuss visible features of the data, such as side-effects of auto-deletion, than to include them in the title. Jan 12, 2014 at 19:18
• Yes you are right after all it says "closed questions" and not "questions closed." Still I think it is important to be aware that these are completely different numbers.
– user9072
Jan 12, 2014 at 20:16
• @quid: Yes, I agree. Feel free to add explanations to the answer. This one is especially tricky to understand. Jan 12, 2014 at 20:47

## Nonpositive score posts per month in 2013

Generated using this query.

     Q's  A's
Jan  174  115
Feb  205  135
Mar  238  121
Apr  240  151
May  263  167
Jun  170  101
Jul  106  116
Aug   93  115
Sep   98  125
Oct  129  158
Nov  113  110
Dec  180  143


## Number of answered questions per month in 2013

Generated using this query.

      All Answered  Accepted
Jan   961      722       453
Feb  1009      714       427
Mar  1031      723       429
Apr  1148      786       486
May  1188      799       501
Jun  1033      715       432
Jul   962      693       403
Aug   891      639       384
Sep   938      648       371
Oct  1010      659       380
Nov   977      680       396
Dec   913      601       329


• Any conjectured explanation for the clear downward trend? Jan 12, 2014 at 19:03
• @JosephO'Rourke: I suspect the dip at the end reflects the fact that not all questions are answered immediately. I suspect in a few months the tail will be much flatter. The first two January numbers are more surprising - I wouldn't expect nearly all questions to have answers! This may be a side-effect of auto-deletion. Jan 12, 2014 at 19:13
• @JosephO'Rourke: Actually, something was wrong with the first column. I updated the query and data and it seems the effect is mostly due to a decline in the total number of questions. Jan 12, 2014 at 19:38
• Thanks for investigating so carefully! But now the decline in the total number of questions could use explanation (although the trend is not a prominent). Naively I would expect the opposite trend... Jan 12, 2014 at 19:45
• @JosephO'Rourke: I don't have a good explanation yet but it's interesting to compare with the relatively steady number of questions with positive score. It appears that the decline is mostly in questions with score zero or less. This is a head scratcher! Jan 12, 2014 at 20:10
• Ah, I see! Thanks. Jan 12, 2014 at 21:17

## Number of posts with positive score per month in 2013

Generated using this query.

    All   Q's   A's
Jan 2003  787  1216
Feb 2036  804  1232
Mar 1984  793  1191
Apr 2186  908  1278
May 2241  925  1316
Jun 2002  863  1139
Jul 1998  846  1152
Aug 1827  776  1051
Sep 1838  813  1025
Oct 1912  852  1060
Nov 1927  867  1060
Dec 1807  733  1074