18

This now deleted question (10k only) stated that questions by new users are generally received worse than other questions. It was based on anecdotal (n=1) evidence, but are there some statistics from e.g. SEDE to support this observation or not?

20

Yes; questions by users which are newly registered (less than one month) are less well received than questions asked by users who have more experience on the site. This could be research by Captain Obvious, but if you dig deeper, there seems to be an interesting trend. According to this SEDE query, the average question score on Stack Overflow decreases when you look at accounts older than a year:

enter image description here

Questions by users with an account of about 4 years are (on average) just as 'good' as questions by 'newbies'. If you 'zoom in' to the first year (with this SEDE query), you see that already after a few weeks, the score is nearly equal to the 2.25 average for the rest of the first year.

enter image description here

It's also interesting to investigate the behaviour on other sites. Mathematics does not observe a decrease after a year. On English Language & Usage there's a small downward trend but not as much as Stack Overflow (the zigzag at the end is caused by the small number of data points). On Meta the trend is upwards.

But all sites exhibit the same behaviour for new (< 1 month old) users, which is not surprising since it takes some time to 'know' the Stack Exchange culture and how to phrase your question. Note that the SEDE query does not include posts which were deleted (often by Roomba) due to their low quality. New users are more likely to post these, but those have already been filtered out.

Actually, upon further investigation, the trend on Stack Overflow seems to be skewed by old, highly upvoted questions. Those were only posted by users which were at that time new. This SEDE query shows that more recent questions follow the same trend as e.g. Meta Stack Exchange:

enter image description here

Another case of Lies, damned lies and statistics?

  • 1
    Thanks @M.A.R.ಠ_ಠ - something went wrong in refactoring that sentence. – Glorfindel Apr 25 '18 at 20:54
  • A controlled experiment could be to post very similarly structured questions from a medium rep participant and a very low rep participant. These could help us tease out the difference between perceptions affecting voting trends, and the learning curve, i.e. participants' questions improving objectively in quality of questions as the participants gain experience with the site in general and also with writing questions. – aparente001 Apr 26 '18 at 5:28
  • 1
    Interesting trend on the Meta query. Could this suggest newbies have a harder time here than on SO? – DarkCygnus Apr 26 '18 at 5:29
  • 1
    Also, seems that in the case of Answers, it is a different story: for SO it is almost the same, while Mathematics and Meta show some clear trends. – DarkCygnus Apr 26 '18 at 5:37
  • Good stuff Glorfindel. But I think the important issue is that highly-upvoted questions and answers tend to get more upvotes. And the high-rep user is more likely to have buddies who will immediately upvote his question or answer regardless of its quality. I don't know how stats could show that. – John Duffield Apr 26 '18 at 7:14
  • 2
    @JohnDuffield - You have absolutely no proof that "high reputation users" have "buddies" who vote on their contributions. – Ramhound Apr 26 '18 at 10:14
  • 3
    @JohnDuffield oh please, there is nothing as much fun on SO as voting down a bad Q/A from a high-rep user, especially if you 'know' them! Sadly, doesn't happen very often, whereas finding a bad question from a new user just means opening a tag and poking at the screen with your eyes shut. That' not entertaining at all:( – Martin James Apr 26 '18 at 12:05
  • If anything, it's low-rep users who are most likely to have umm... 'upvoting buddies'. Has there ever been a case of a skilled,. experienced and respected high-rep user running sock-puppets and/or voting-rings to inflate their rep? – Martin James Apr 26 '18 at 12:11
  • @MartinJames yes (that question does not contain all details, though) – Glorfindel Apr 26 '18 at 12:31
  • The stats make me wonder if the FAQ and help for new users is ok to start with, as with users with experience in the site methodology get better score. I guess users learn how to write a good question to get result, as SO vs a normal forum is really not the same – yagmoth555 Apr 26 '18 at 12:53
  • @Ramhound : let's not go into "gaming the system" today. Your comment noted Martin James. Another good one Glorfindel. – John Duffield Apr 26 '18 at 15:42
  • @JohnDuffield - I never once made mention of Martin James So I am not sure, if you confused me with somebody else, but please keep track of whom directing comments to. – Ramhound Apr 26 '18 at 16:53
  • The shown statistics don't say anything about the quality of questions asked by new users, anyway. – kiamlaluno Apr 26 '18 at 18:45
  • @Ramhound : I think you misread my comment. – John Duffield Apr 27 '18 at 7:13

You must log in to answer this question.

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