Has there been any research done on this? I have recently noticed that if CV notices suspicious upvotes, they will negate these upvotes. But, if a user gets say almost 1600 reputation points in upvotes, and the site negates exactly 1600 reputation points, isn't this actually the bare minimum of the reputation points they received? Shouldn't they be penalized beyond this? I believe this is the bare minimum since I think the reputation points have a snowball effect. As their reputation grows bigger, people are more likely to assume this person knows what they are talking about. I realize some of this might be for meta, at the same time, it could require some statistical analysis.

How would this be determined?


migrated from stats.stackexchange.com Sep 27 '12 at 15:22

This question came from our site for people interested in statistics, machine learning, data analysis, data mining, and data visualization.

  • 1
    $\begingroup$ Re: "As their reputation grows bigger, people are more likely to assume this person knows what they are talking about" - are you implying that people give upvotes to those with a high reputation on faith, even if they have no clue whether the answer was a good one? I can see this happening if the answer sounds good (in which case it probably deserves an upvote) but I'd hate to think people take one took at someone's reputation and upvote blindly... $\endgroup$ – Macro Mar 21 '12 at 4:46
  • 1
    $\begingroup$ Yes, that's what I'm implying. And, I'm not saying it happens all of the time, but it could happen sometimes. In fact, I would say that's part of the reason why people try to cheat. In elections, don't some people vote for the person they think will win? $\endgroup$ – Adam Mar 21 '12 at 5:39
  • 2
    $\begingroup$ Of possible interest (as a discussion and for statistical ideas): stats.stackexchange.com/q/13858/930. As it stands, I would recommend migrating this question to Meta as it is not obvious to me how statistical analysis might be involved at this point. $\endgroup$ – chl Mar 21 '12 at 7:07
  • 3
    $\begingroup$ I think there's more of a "waterfall" effect on individual questions than on users. That is, if a question/answer has a lot of votes, then I think it's more likely for someone to blindly upvote (or if a question/answer has a high negative rating, it's more likely to be blindly downvoted). On the user-level I'm not so sure. Sure, it is possible though, I guess, although I don't really understand it. $\endgroup$ – Macro Mar 21 '12 at 12:42
  • 2
    $\begingroup$ @Macro, FWIW, my impression of voting behavior is the opposite. Not that this has any scientific / substantive basis whatsoever, but I suspect there are many cases where someone thinks, 'this is a decent answer, but it has 11 upvotes already; I don't think it's worth 12'. My impression is almost all upvotes come at the beginning, such that the pace of upvotes in (say) the 1st 100 question views is much higher than in the QV's from 101 to 1k. $\endgroup$ – gung Sep 27 '12 at 20:04
  • 2
    $\begingroup$ My heuristic and purely empirical rule-of-thumb is that a very solid answer usually gathers about 3 upvotes per 100 views. My impression is that this ratio is surprisingly robust across the other sites on the network that I frequent. $\endgroup$ – cardinal Sep 27 '12 at 21:58

You must log in to answer this question.

Browse other questions tagged .