last day (33 days later) » 

Corone
Corone
1:06 PM
Ok, I'm optimistically starting a new chat room for discussing digging out these old questions and improving our answer rate
I think it would be a great idea if people joined me in looking back at the old unanswered questions and seeing if we can clear a few of them away. This could be great for us newbies as we may bring new skills/answers to questions from before we joined (also we get +2 rep for edits! :-) )
Many of these very old questions are orphaned - that is the original asker has long since disappeared. This means you will never get an accept for answering the question.
On the other hand it does mean you can be really brutal in bashing the question into an answerable form, and then answering it.
Of course the thing this needs, is people to upvote any good answers that we give to these orphaned questions
And so... hence this chat room.
Use this room to draw attention to an edit/answer of an old question that you think is good, and then other archeologist can go look at it, and up vote if we think you did a good job.
We can also discuss what should be done with some questions, where the best course isn't clear
First off, some meta questions relevant to this:
17
Q: Improvement Drive - Cross Validated

Katey HWGreetings from Stack Exchange HQ! We have been assigned to evaluate Cross Validated and recommend some methods for making it even better. Please note that this is a more hands-on evaluation, and is not related to the Community Quality Evaluations already done on sites. This is a distinct evaluati...

discussion
5
Q: Why is our answer rate so low?

CoroneHas anyone else noticed that we are 5th from bottom with an answer rate of only 76%? On the one hand there are some questions that appear "answered" in the comments, like this one: Is the percent of total deviance explained a useful model summary? These should probably try be tidies up and "an...

discussion unanswered-questions
17
Q: Comments that are actually answers

ParburyRelated discussion on Meta.SO. Is there some kind of way that comments that are infact answers to questions be moved from a comment section into the answer section? Ideally the user who asks the question should be able to do this or a mod. This would possibly decrease the number of unanswered q...

feature-request status-bydesign
And now some initial digging:
I feel this question was answered in the comment, so I CW'd it, it either needs an upvote or a better answer:
0
A: Undefined link function in gamma distribution

CoroneDropping a zero is usually not a good idea unless the dataset is so large that losing an extreme value will make no difference in the analysis. Treating it as very small is good, but then the result can be sensitive to the value chosen. One could check for that in various ways. A more formal appr...

I heavily edited this orphan to make it answerable, and attempted to answer it myself:
0
A: Species Richness, Dominance and Diversity Differences

CoroneI think neither of these responses fit perfectly into any of the standard GLM link functions. Taking a pragmatic approach it is probably sufficient to pick a link function that is broadly doing the right thing. Your raw data is from a multinomial distribution, and Richness, $R$, is the number o...

Ok, so that's me started things off. I suspect I will now get slated for "asking for votes", or some such, or be completely ignored, but I'm just trying to make things better. If lots of people join in then this could really work.
 
Andy W
Andy W
1:51 PM
Nice work, some sites have community organized clean up events where they do similar (get a list of questions, and then either attempt to answer them or close them if they are poorly worded).
 
Corone
Corone
2:15 PM
Thanks Andy! Here's a question I'm less sure what to do with:
7
Q: Specifying a covariance structure: pros and cons

Jack TannerWhat are the benefits of specifying a covariance structure in a GLM (rather than treating all off-diagonal entries in the covariance matrix as zero)? Aside from reflecting what one knows of the data, does it improve goodness of fit? improve predictive accuracy on held-out data? allow us to est...

generalized-linear-model covariance
It has seven upvotes - which suggests it is interesting for readers. It isn't badly worded, although it is a little open ended. It seems like an important enough topic that it would be really good if someone could provide an answer for this. The one comment so far starts us off, but isn't really enough.
 
Andy W
Andy W
2:47 PM
IMO the comment on the question is sufficient, and we can encourage Stephen to place it as an answer. Assuming no co-variance is specifying structure, a particularly restrictive one. The only question unanswered I see is "Is it possible to determine the right covariance structure empirically, or is this something that depends on your knowledge of the data-generative process?"
I know you can check residuals for particular auto-correlation, so you can determine if the covariance structure is in-sufficient, but I don't know how you empirically compare different ones (maybe a model selection type question).
 

  last day (33 days later) »