20
$\begingroup$

My name is Katey and I work in community development at Stack Exchange. I've been studying stats.se for a few weeks, trying to figure out ways to improve this (already pretty great) site. My coworker Brett White and I have been emailing users on sites we've identified as having great content and engaged, awesome users but less than stellar overall voting activity.

I emailed a few users last week and got some great feedback -- particularly that it can be difficult to know whether a question is good or bad if you aren't familiar enough with the subject, software or field that is being asked about. That makes sense, Statistics is a huge and varied field, and I wouldn't want to ask someone to vote on material they were wholly unfamiliar with. However, I don't think it's helpful to the site to be too hesitant to cast votes on questions just because we think we might not know enough, if that makes sense. I also don't think that voting for voting's sake is going to be helpful, but there may be a happy medium in there somewhere?

First, here is a document that Brett put together focusing on 90 low-vote questions on the site.

Essentially we want to make sure that questions that are getting a lot of search engine traffic are quality questions that are well-representative of the site and community. Take a look at these and see if they're questions you hadn't had a chance to decide on voting on either way. It's entirely possible that you've already looked at them, but this seemed like a good place to start evaluating.

Keep an eye on new questions as the come in, as well, since it's possible that some of these low-voted questions are just slipping through the cracks when they're first asked.

This is the first step in what we're doing to help make Cross-Validated an even more useful and awesome site, so please discuss other ideas to improve the site and community on meta and in your chat rooms!

| |
$\endgroup$
  • 11
    $\begingroup$ The issues are very well put in this question. Because the same discussion is occurring on another site and I don't want to repeat myself--I would post essentially the same answers here as there--I would like to refer interested readers to meta.gis.stackexchange.com/questions/859. $\endgroup$ – whuber Aug 24 '12 at 16:46
  • 3
    $\begingroup$ Hi Katey - since receiving your email I've looked over this list and what strikes me is that many (maybe half?) of these questions are rather off topic for the statistics stackexchange. I see several that are essentially computing related (which we try to migrate). In any case, I'm not quite sure where this list comes from (high views maybe?) but to figure out why certain good questions/answers don't have higher vote counts, we need some examples that are "objectively good" that still have low vote counts, which will require the opinion of multiple objective statistician(s). $\endgroup$ – Macro Aug 24 '12 at 17:11
  • 6
    $\begingroup$ We certainly need to find ways to encourage a little higher-volume (principled) voting. I find two things a bit surprising: (a) the large proportion of total reputation conferred by a very small number of voters and (b) the proportion of (otherwise) high-activity and high-reputation participants who vote comparatively very rarely. $\endgroup$ – cardinal Aug 24 '12 at 17:51
  • 3
    $\begingroup$ I have high rep and I do a lot of voting. But my questions seem to often get very little attention even when the OP gives my answer a check. On occasion I get checks and thank yous but zero votes. While there is too little upvoting, I think sometimes the downvoting is overdone. $\endgroup$ – Michael R. Chernick Aug 24 '12 at 18:30
  • 7
    $\begingroup$ Votes are not the best gauge of "attention," @Michael. A better indicator is to note the number of thread views when you post an answer and then see how many more views have occurred later. This presumes that everyone looking at a thread will have seen your answers, of course, but it's a not unreasonable assumption and the best we can do. By this measure I believe your answers have probably received more total attention than those of anyone else on this site. Downvoting is rare overall (less than 0.5% of all votes): the site analytics provide no support for the claim it is "overdone." $\endgroup$ – whuber Aug 24 '12 at 19:37
  • 5
    $\begingroup$ Although I am enjoined from releasing detailed analytics data, I am pleased to be able to report a visible changepoint in the slope of the daily voting time series: an annualized increase of 20% from the site's inception changed in January (when it was averaging about 130 votes daily) and has averaged an annualized 100+% since then, reaching 220 votes/day recently. Thank you to all of you who are making that extra effort! $\endgroup$ – whuber Aug 24 '12 at 19:53
  • 1
    $\begingroup$ @whuber: How does that compare to the growth in number of posts (questions and answers) per day? And, growth in number of users over, say, 200 in reputation (to get over a modest threshold of site participation)? $\endgroup$ – cardinal Aug 24 '12 at 20:13
  • $\begingroup$ @whuber I think it is a matter of opinion whether or not downvoting is overdone. It is not the statistics on how often it is done that matters to me, but rather the reasons for doing it. One thing I have notied recently is that i am starting to get votes here and there on old questions. I guess that may mean that some people are starting to look at the old questions in more detail. $\endgroup$ – Michael R. Chernick Aug 24 '12 at 20:29
  • 5
    $\begingroup$ @cardinal: New user growth is relatively steady, roughly doubling annually. (I can't filter by user reputation.) One might imagine a changepoint in total Q&A in January '12, but it's not strong. (There's a lot of noise even in the 28-day moving averages I'm inspecting.) So the voting increase persists after controlling for these other factors. $\endgroup$ – whuber Aug 24 '12 at 20:49
  • 5
    $\begingroup$ @Michael (1) I won't argue about the meanings or interpretations of vague characterizations like "overdone"; I'm just trying to share the facts with you. (2) It is true that old questions continue to deliver votes for a long time, but the halflife is pretty short. $\endgroup$ – whuber Aug 24 '12 at 20:51
14
$\begingroup$

Being a moderator on Stack Overflow (and sometime contributor here), I can definitely emphasize the importance of upvoting good content.

While voting for answers requires more detailed understanding of the question at hand (after all, voting on an answer is somewhat based on the veracity of the information in the answer), voting on questions doesn't carry such a high bar.

Questions imply that there is a lack of understanding on the part of the asker (and that's ok); what's important when voting on questions is whether or not the person put the effort into conveying their problem, their understanding of what it is they don't know, along with any other information needed to contribute to getting a quality answer.

This is why mods on Stack Overflow and other sites are qualified to instantly close questions when they don't have intimate knowledge of the topic; if we see a question that a user put obvious effort into creating with the purpose of getting a well-defined answer to a problem then it qualifies as a quality question and is not only deserving of staying open, but generally of an upvote as well.

| |
$\endgroup$
  • 16
    $\begingroup$ +1. In the same vein, as a general rule, if I answer a question, then by that token it must have been an understandable and worthy question, so, almost without exception, I will automatically vote that question up. I'm surprised at the number of questions that get more answers than votes... $\endgroup$ – whuber Sep 1 '12 at 22:12

You must log in to answer this question.

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