@amoeba, it's a good thread. Most of what I would say has been said by @captain_ahab. I'll think about whether I have anything extra to offer.

@Glen_b I just saw this now even though it's two weeks old. Congratulations on the Legendary badge! Probably one of the most difficult, if not the(!) most difficult, badges to achieve. Great work :-)

@Andy: and everybody else seems to be more than three times behind @Glen_b, because even the Epic silver badge has only been awarded to him (50 times vs 150 times for Legendary), or am I confused?

@amoeba, no one else is even close. @whuber is at 33; @chl is at 9; I'm at 5, e.g. (data here).

@gung Although I wasn't keeping track, I am glad to see that my number is so small: I have kept it down in order to concentrate on moderation duties. I try to average no more than one answer per day, which will keep my daily reputation accrual far below 200 points most of the time. This does not mean I think all people should hold to the same discipline; and I am particularly glad that someone like @glen_b who consistently posts high-quality answers is so active.

I had a sense that was your policy, @whuber. I'm also sorry to hear that moderation is burning you out (I suppose it isn't surprising, though). I certainly appreciate the effort you've put into building this site into what it is: a courteous, productive community curating high-quality content. As your moderator duties ease up, I (selfishly) hope you won't keep to that policy; I would certainly benefit from more of your answers.

@Andy Thanks. That one was a surprise actually, I thought I was some distance from getting it yet. It's one of those badges you don't really do anything in particular to earn, it just marks a milestone. For me the Marshall badge seems harder - rarely do I even see a post that could be flagged where I can't do something other than flag (like nominate for closure). I don't think I see more than one post a week where I feel "no, a moderator must deal with this outside the usual channels".

@Andy Generally speaking I wasn't seeking to pass 200 (except, I admit, when I'd get back on and discover myself on 180 or 190 with plenty of time to go, sometimes I'd be tempted to answer a question to see if it would go over). Usually I'd just come back next day and see "oh, hey, that went over 200 yesterday, which answer did people like?". Last year I think I averaged about 3 answers a day, and I was usually pretty consistent, but I was getting over about ten times a month.

It's not typical for me that more answers get more points -- you can get over by answering lots of questions (we have a bunch of users who have answered more than 10 a day, some many times and the maximum is 26 - I've never answered more than 10 and regard the times when I got anywhere near it as overdoing it). On the days I have some extra time and find several questions I'd like to answer, I look for other things to so - recheck the review queues, ...

... polish questions that could be better, add tags where they seem needed, or just make my answers better. If there's still time, then I will a couple more answers, but to pass 200, I find that 4 or so answers will often be plenty. It typically takes one answer people like, plus a couple that get the usual few upvotes, plus a couple of votes from the day before.

Often I can't guess which answers will get two votes and which ones will get ten.

Even though it's often kind of random, I do enjoy getting "Nice Answer" more than I enjoy hitting that 200.

Accepted answers make a big difference. On days where I've been active several days in a row (so getting points from earlier days), then if I get a third acceptance in a day, it usually makes 200 -- because that's nearly 50 points right there.

@whuber I'd also benefit from more of your answers. I hope as the new moderators learn the ropes there's enough time for you to answer more questions. I get so much out of your answers. Indeed, sometimes one of your brief comments has more packed into in it than the whole set of answers combined.

@amoeba You might find this query helpful, but you'll notice it only has me at 148, so it must occasionally undercount in some subtle way. Indeed, I can't find any query that gives me the number the system thinks I'm at -- I think I even went back to count by hand, I was unable to find the 150th. Probably something to do with points on a deleted question or something.

(Even allowing for the fact that the database is only updated once a week and adjusting for that, there's still an undercount)

On significance testing, I do think it's heavily, heavily overused. Often I find when I talk to people about their analysis, what they're actually interested in finding out is something that could be better answered by something other than a significance test (sometimes, a confidence interval, sometimes actually a display answers the question they ask), but for some reason it's felt that without a p-value, somehow it doesn't "count" -- even when that adds nothing we don't already know.

I find significance stars completely counterproductive. I was quite annoyed when they crept into R.

==

I'm wondering what I'm missing here. Can anyone guess what distinction the user is making between "not interested in shape" ... and "interested in whether it resembles a uniform". What?

@Glen_b: I certainly could not agree more regarding significance tests "raw" vs. confidence intervals -- but that's more a less commonplace I believe, with everybody sensible agreeing that the focus should be on the effect size / parameter of interest with confidence intervals or smth like that. If this is then supplemented by a p-value and even some stars on the figure -- personally, I don't mind (and perhaps even find helpful). But the current discussion is about trashing CIs too...

@amoeba Yeah, CIs, Bayesian stats (or largely so by the look of it), in fact just about the whole of inference.

Now descriptive stats matter, and it's important to do them well, but sometimes you need some form of inference.