So some companies appeared six different ways (some of them typos). They were shocked by that. How did they not already know? You just sort a list of uniques and you can spot them in a few moments. 600 companies takes only a few minutes to look through for near duplicates.
They had no clue about any of the information in it.
Some of the assertions I could see were false by looking at the spreadsheet of the raw data before I even read it into R. "This column is only 0 and 1" "Then why is there something that's neither 0 nor 1 in the second row?"
On the other hand, I get far more more "people helped" each week than I've ever met in my life. Even if only say 5% percent of them are actually people helped, a few months of it vastly outstrips all the people I've had lectures, conference talks, seminars and so on I've ever done, all put together.
People just don't edit as much, I suppose. You'd probably have a more robust measure by skipping the first 2 people. I have voted ~2x for each edit. That might be a ratio to explore.
@gung I'm often a bit nervous while I'm editing because I realize how little I know. I try to stick to grammar, spelling, and formatting, but even then I worry. One question I almost changed logit to logic. I ended up looking up the term because it was used too many times to be a casual spelling error.
Spelling, grammar, punctuation, formatting, etc., is a huge help. If you accidentally change "logit" to "logic" someone will change it back. Go ahead and edit.
My assumption is statements like "Please help" at the end of a post should be removed aggressively like "Thanks" and signatures aka "John Doe" Is that accurate?
@Erik What I do is by no means definitive, but my habit is to always remove things like that if a post has recent activity, or if I am editing an old post (e.g. to add a tag or correct a typo)
but when you do, it's worth tidying up anything else (tags, spelling, grammar) that you can spot
I have to admit that if I see an old post with low views, and nothing that needs doing except removing a brief "thanks" at the end, I'll leave it there
But that's because I'm a bit paranoid about clogging up the "active questions" list with trivial edits, and probably needlessly so
If it's a recent question, so I know the author is reasonably likely to see my message, then after editing I'll leave a brief comment explaining why I removed the "thanks" - it seems a bit rude and impersonal, but we like to get to the point on this site because it helps the people who come to read this question later
@Glen_b Thanks for sharing your story about the not-so-clean data. I quite often find myself telling people that allowing custom text entry on questionnaires and surveys is a recipe for trouble. Also, it's surprising how often people select contradictory responses at different stages in a survey...
Probably the best thing to do in an ideal world would be to ask questions which don't give someone the chance to disagree with themselves later! But some sort of validation that points out to someone filling in their survey that the response they just selected disagrees with an earlier answer, and giving the opportunity to amend either or both, might not be a bad idea (dunno if this is actually implemented on any major survey software).
On inspection, it looks like many "late answers" from years ago have been added to the review queue.
Is this a change to, or an error in, the algorithms that select which answers get added to the queue?
In situations like that, should I just try starting a close vote and see if other people agree? Flag it? (I've generally avoided flagging duplicates since I got over the rep threshold to let me click on "close" instead?)
It's a curious situation where if they were regarded as duplicates I think the threads should probably be merged. Perhaps onto the newer question of the two, if only because that one is the slightly more general of the two. But I'm not even 100% sure whether they are dupes.
I have opted for the neutral approach of mentioning it on chat and seeing what other people think!