6
$\begingroup$

Sorry for the long-winded question title. During interviews for a junior data scientist position I have noticed that almost every interviewer I have talked to (not recruiters or "talent managers", but real life data scientists including the head of data science quite often) know very little statistics. I am very interested to hear if anyone has real-life experience in a data science setting, where a lack of statistical knowledge has caused serious problems?

Would such a question be off-topic? If not then what tag(s) might I use?

$\endgroup$
6
  • 11
    $\begingroup$ That doesn't sound on topic to me (for many reasons: lack of focus, no definite answer, opinionated, etc.), but you certainly could discuss it in chat. We could even create a dedicated chat room for the discussion if that would help. $\endgroup$
    – whuber Mod
    Jan 21 at 18:07
  • 1
    $\begingroup$ @whuber Thank you and understood. I thought that might be the case, hence why I asked here first, and now I'm especially pleased that I did since I didn't know about there were chat rooms here - that sounds like a great idea ! Thanks again. JK. $\endgroup$
    – Joe King
    Jan 22 at 10:36
  • 3
    $\begingroup$ Community wiki? $\endgroup$
    – Dave
    Jan 22 at 15:16
  • 6
    $\begingroup$ I'm wondering if I entirely agree with other comments. For instance, we do have a popular question about "statistical wins and horror stories", which hasn't been considered as off-topic so far, and feels similar to what you're talking about: stats.stackexchange.com/questions/434128/… . I say "I'm wondering", because I don't have a final opinion on the subject. In particular, a problem I see with asking about personal experiences is that they could be unverifiable, contrary to other types of questions. $\endgroup$
    – J-J-J
    Jan 23 at 8:08
  • 4
    $\begingroup$ (cont'd) On the other hand, StackExchange seems to incite people to share things that "that happened to [them] personally" stackoverflow.blog/2010/09/29/good-subjective-bad-subjective . It may be relevant on other SE websites, but I wonder if it's the case on CrossValidated. Relating an experience may illustrate an explanation about statistical issues, but making it the gist of an answer can be problematic due to unverifiability (e.g. if we get some obscure anecdote by an anonymous person on how a statistical mistake supposedly bankrupted a company, we only have one side of the story). $\endgroup$
    – J-J-J
    Jan 23 at 9:24
  • 1
    $\begingroup$ @J-J-J, that's a good remainder to read that old blog yet accurate stance towards opinionated subjective queries. Thanks for the link. $\endgroup$ Jan 23 at 11:37

2 Answers 2

7
$\begingroup$

As whuber noted, these types of discussions aren't suitable for SE - just ask yourself, whether there can be any single, definitive answer to such a question, if it were posted, leading to an inevitable closure on the grounds of "primarily opinion based". In fact, if you can't find any tag that fits your query, chances are it might be not an on-topic question on SE (not always, of course). SE is a q&a site and not a forum/chat, unfortunately.

You can definitely, as whuber suggested, visit our primary chat room Ten Fold where folks from diverse areas often drop by and might have valuable knowledge and experiences to share with.

$\endgroup$
2
  • $\begingroup$ Thank you ! I've taken a look at Ten Fold. @whuber mentioned setting up a dedicated room there for it which sounds great but I don't want to jump the gun or run before I can walk, so, since it's my first time using chat rooms here, should I just post my question in the "main room" (if that's the right term) ? $\endgroup$
    – Joe King
    Jan 22 at 11:46
  • 4
    $\begingroup$ @JoeKing, there isn't really a "main room" exactly. *Ten fold" is the name of the generic chat room for CV. Rooms can be set up fairly easily. Other rooms are usually created for specific conversations, most often comments on a Q or an A that go on too long or wander off topic. A room could be set up for you, or you could just ask in Ten fold. There isn't too much of a protocol. You can just jump in. $\endgroup$ Jan 22 at 12:31
7
$\begingroup$

We have a number of questions that don't meet the strict standards we expect, but are nonetheless valuable. Many are closed but some are tolerated and turned into community wikis. The decision about whether to do this is probably (and necessarily) a bit ad hoc, but I would support doing the same for a question like this. It's an interesting one and no worse than several others we have done the same for.

$\endgroup$
4
  • 2
    $\begingroup$ I seem to somewhat agree with you (J-J-J and Dave). Problem, I feel, is this specific question might have the balance tilted more towards opinion than experience. I am seeing it more as a Quora/Reddit query rather than a proper SE community wiki post. But that's my take only. $\endgroup$ Jan 23 at 11:35
  • 1
    $\begingroup$ @User1865345 It's certainly a grey area, but I think asking for specific examples (as the post does) puts it squarely in the realm of experience. It can still be off-topic for other reasons, of course. $\endgroup$
    – mkt
    Jan 23 at 11:58
  • 1
    $\begingroup$ Thanks for your input :) I have started a conversation at Ten Fold, but I would very much rather see the discussion on the main site, for posterity etc. I don't know what the procedure is to change a post into a community wiki, but now I'm leaning towards posting a question on the main site and hoping that someone (mods?) can turn it into a wiki rather than closing it ? $\endgroup$
    – Joe King
    Jan 23 at 19:48
  • 1
    $\begingroup$ @JoeKing Per my other comments under your question (stats.meta.stackexchange.com/questions/6631/…), if you were to ask such a question, I'd recommend you to restrict its scope to famous (or at least well-documented) examples. This may reduce (but not eliminate) the risk of having your question closed. About converting a question to a community wiki, my understanding is that this is a technical operation that only moderators can perform, so probably nothing to worry about too much. $\endgroup$
    – J-J-J
    Jan 23 at 23:02

You must log in to answer this question.

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