3

Compared with other SE boards, there seems to be somewhat of a deficient use of voting on both questions and answers.

Is this accurate? Are there any available statistics to do this comparison available to users or admins?

Is this a symptom of a more discerning (or perhaps persnickety) audience?

Is this a symptom of a lack of voting incentive and engagement in this specific community?

  • 2
    Can you give some examples of the problem you're seeing, or explain it in more detail? Just saying that it's deficient doesn't give us a lot of information. – Keelan Jun 3 '17 at 6:54
  • 1
    The trouble is possibly a rather narrow definition of what constitutes a good question/answer. It must be about a referenced philosopher, not so simple as to be easily found on the SEP, yet not so complex that it invites too much discussion, not so basic that it doesn't interest the academic philosophers but not so advanced that not enough people have written about it to cite. It's not an easy balance to strike, I'm not surprised there are few votes cast. – Isaacson Jun 3 '17 at 13:05
  • @Keelan I don't have the usage statistics and it would be difficult to manually collect the data. I don't know how many users there are here, how many questions are posed, etc. But it seems like there are only 200 questions with more than 30 votes up. That seems low compared to some of the other site on SE. Does anyone do statistical reviews on the successfulness of the voting apparatus? I would be interested to do it, but I don't have the requisite reputation. – PV22 Jun 3 '17 at 13:15
6

I had a look at some similar sites using the overview of sites, ordered by traffic. Here is an overview of the sample (I looked at sites with a similar amount of traffic, and tried to choose topics somewhat related to philosophy, although not all sites below are). For every site, I list the number of questions, users, visits/day and site age. Then, the scores of the five highest-scoring questions.

  • Philosophy (8.2k Q, 19k U, 4k v/d, 6y): 149, 131, 89, 78, 78
  • Politics (4.4k Q, 12k U, 5.2k v/d, 4y6m): 186, 117, 115, 111, 105
  • Project Management (3.8k Q, 18k U, 4.5k v/d, 6y4m): 118, 97, 69, 62, 57
  • Mi Yodeya (22k Q, 7.7k U, 4.3k v/d, 6y1m): 106, 75, 62, 49, 47
  • Law (6.3k Q, 8.7k U, 3.9k v/d, 2y1m): 76, 54, 53, 49, 46
  • Cognitive Sciences (4.5k Q, 12k U, 2.8k v/d, 5y5m): 146, 75, 57, 54, 52
  • Linguistics (5.1k Q, 9.7k U, 2.8k v/d, 5y9m): 57, 49, 45, 43, 36

Using the SE Data Explorer and this query, I collected the number of posts that have 0-9 votes, 10-19, 20-29, etc. for the same sites. Dividing those numbers by the total amount of posts gives the part of the posts that has this score. These numbers can be found in the figure below. E.g., all sites have 0-9 votes for almost all posts. Philosophy has the second-least 10-19 votes (after Law). In higher bins, Philosophy is comparable to Project Management, Cognitive Sciences and Linguistics.

enter image description here

A second query looks at scores between -4 and 9. The chart for the same sites shows Philosophy comparable to Project Management and Law, and to a lesser degree Linguistics. The other sites tend to vote more, which is seen in lower bars for scores of 0 and 1, and higher bars for higher scores. It could be interesting to try to get users to vote more, such that good content is easier to find.

enter image description here

  • Thanks! I hope it wasn't too time consuming. – PV22 Jun 5 '17 at 20:19
  • @Lampy not at all, I enjoy collecting data and making charts :-) – Keelan Jun 5 '17 at 20:22
  • Two suggestions if you wanted to get some more descriptive data. The drop off from 0-9 to 10-19 is large (the largest of all the groups on average). This means you're not getting as statistically significant a result in terms of comparison than you would comparing sites within the 0-9 category (i.e. Individual scores from 0 through to 9). Second, if there was an underutilisation problem, then site subject might be one of the causes and you would have masked that factor by selecting similar sites, maybe just sites of similar traffic alone as a criteria would be more revealing. – Isaacson Jun 6 '17 at 12:53
  • Sorry, I'm a statistics nerd, this stuff is like nicotine. – Isaacson Jun 6 '17 at 12:54
  • @Isaacson feel free to create other charts using the SE Data Explorer I linked to; the rest is copying the values to Excel and having it generate a chart. – Keelan Jun 6 '17 at 13:40
  • I'm not familiar with the Data Explorer's syntax, I simply misinterpreted "I enjoy collecting data and making charts" from your previous comment and thought you might have an interest in correcting some of the statistical biases in your original. Maybe one day I will have a play with it though, it does look fun. – Isaacson Jun 6 '17 at 16:09
  • 1
    @Isaacson there are no statistical biases; these data just give other information than you expected. I think a comparison with other sites is not that interesting, because the number of votes you can expect depends on the type of site. This is a knowledge-based site; users tend to upvote answers when they know they are correct. On other sites, like Stack Overflow, answers can be easily verified by non-educated users, leading to more votes. With this comparison I'm trying to say: if you believe the voting is defunct here, you are also saying that it is defunct on these other sites as well. – Keelan Jun 6 '17 at 16:15
  • 1
    If I could award a bounty to a meta answer, I would. – Dave Jun 6 '17 at 21:49
  • @Keelan I meant no offence by the use of the term bias. In statistics, bias is introduced every time the operator uses their judgement to select groupings not otherwise determined by the data, your method does that and so may hide/exacerbate the very feature you're trying to investigate. The question was if voting is underutilised in Philosophy.SE, your figures show that most voting is in the 0-9 category anyway, so if Philosophy.SE generally has votes in the 1's and 2's where other sites are in the 8's and 9's that would be a significant result for what you're trying to find out. – Isaacson Jun 7 '17 at 9:02
  • @Isaacson no offence taken. I mostly considered Lampy's comment: "it seems like there are only 200 questions with more than 30 votes up. That seems low compared to some of the other site on SE." That is why I chose the bins like this. I agree that lower scores are interesting as well; I'll have a look at it and update the answer. – Keelan Jun 7 '17 at 9:05
  • Thanks, I wish I could be of more help, I will have to get my head round the Data Explorer when I next have time. I tried just making some intuitive edits to your query to see what happened (like just deleting the "group by" bit), but it didn't work. – Isaacson Jun 7 '17 at 9:08
4

Thanks to Keelan doing the Data Explorer work for me I've been able to produce the graphs for the other aspect to this question that might be missed from the groupings selected. Below is a graph without selecting groups, it's just the three sites ahead of Philosophy.SE and the three behind it in terms of traffic. The graph shows the spread of votes as a percentage of the total score given out by that site's users (from -4 to 9). enter image description here

As can be seen, both Philosophy.SE and Personal Productivity have an anomalous spike for low vote answers compared to other sites of similar traffic. As I've made this graph using percentage of total votes cast (from -4 to 9) it could theoretically be done with sites of any traffic and still be comparable. Maybe a job for another day. Still, this shows that the OP's instincts are not wrong at least for the few sites of similar traffic, Philosophy.SE does have an anomalous spike of low votes, but a few other sites do too.

Edit - just thinking about possible causes, traffic doesn't seem to be one because Mathematics has a similar spike despite high traffic, neither does subject matter (I can think of no better polemic of subjectivity than Mathematics and Personal Productivity, yet both have spikes).

As the community that's providing the "lower quality" answers is the same one that's determining them to be "low quality" I can only guess that communities with these spikes are ones where there's a lot of disagreement for whatever reason and the votes available are being split between widely different answers or answers that some deem suitable (the people doing the answering presumably) others do not. On sites with a more even spread there must be more general agreement on what constitutes a good answer.

  • I am surprised by the low vote spike on Math SE, you'd think that if there is a general agreement on what is correct it would be there. My guess is that the reason on Math SE is different from that on Phil SE or PP SE, for which your explanation seems plausible. I suspect that at any given time it has a very large number of inexperienced users, who produce low quality questions and answers that do not get much viewing. – Conifold Jun 16 '17 at 3:47
  • Yes, that's certainly what I thought, which is why I checked Maths, to see if subjectivity might be an issue. Of course there's no reason to believe the low vote spikes are caused by the same factor on all sites that experience them, but presuming for now, they are, I think low interest in each particular question might be an issue, but I'm inclined to think it's because voting might be seen as unnecessary on something like Maths. Given that answers are more likely to simply be right, there might be fewer competing answers and so users do not so much feel the need to vote the best one higher. – Isaacson Jun 16 '17 at 6:11

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .