# DocBuckets

less info
reputation
18
bio website mic.wustl.edu location St. Louis, MO age member for 1 year, 10 months seen Aug 6 at 15:41 profile views 37

Intermediate statistics, mostly self taught so don't take me too seriously.

# 53 Actions

 Oct2 awarded Nice Answer Jun27 awarded Yearling Jun11 comment Regression for an outcome (ratio) between 0 and 1 Back when I wrote this comment, I used REGRESSION after log-transforming the data. Since then I've written a more sophisticated version that uses GLM. I deal with light emission measurements and my testing suggested gamma regression with a log-link was the least prone to runaway uncertainty on the parameters. For most of my real data, the answers from using normal, negative-binomial, and gamma with log-link were all really similar (at least to the precision I needed) May30 revised How to code separate data for a regression parameter in nonlinear regression? Adding another example so people will address the actual question. May29 comment How to code separate data for a regression parameter in nonlinear regression? Alright you got me, I was rushing to think of an example between lab work and got sloppy... pretend I had an example on hand that didn't simplify. I'm asking about a theoretical statistical method, not about a specific equation. May29 answered Bonferroni adjustment in SPSS - what does it do? May29 comment How to code separate data for a regression parameter in nonlinear regression? I used that equation because I'm familiar with it, not because it's a specific equation giving me trouble. I've updated my question with a less convenient example. May29 revised How to code separate data for a regression parameter in nonlinear regression? clarified and expanded question based on comments May29 asked How to code separate data for a regression parameter in nonlinear regression? May23 comment Combining two confidence intervals/point estimates If going back to confidence intervals from the pooled SE, what would the degrees of freedom for the T distribution be? Would this change if combining more than 2 confidence intervals? May20 comment Confidence error bars and “central point”: Should we emphasize the median? @whuber After I wrote the edit, I realized that I was looking at it all wrong. Indeed I got away from the definition of the CI and was confusing the distribution of the parameter and the distribution of the repeated estimate of that parameter. Thanks for setting me back on track. May20 accepted Confidence error bars and “central point”: Should we emphasize the median? May20 comment Confidence error bars and “central point”: Should we emphasize the median? @whuber: I realized that a few minutes after I submitted the edit. I think it clicked. May20 comment Confidence error bars and “central point”: Should we emphasize the median? I did my best to clarify in the body of the question. @whuber: confidence limits are defined much more like the median than the mean, which is why I bring median into the discussion in the first place. As confidence goes to 0, the confidence interval goes to a median value (which is also the mean, when the distribution is symmetrical e.g. normal). Perhaps I am making a mistake in trying to push concepts easily understood in symmetrical distributions into asymmetric ones. May20 revised Confidence error bars and “central point”: Should we emphasize the median? added context to question per discussion, made equations easier to read. May20 asked Confidence error bars and “central point”: Should we emphasize the median? May18 comment What does the Scale parameter mean in linear regression? Does this apply to models that don't use probit or logit? My example specifically uses two continuous variables. Does scale still just account for heteroscedasticity? May17 asked What does the Scale parameter mean in linear regression? May15 awarded Commentator May15 comment What are common statistical sins? I try to be statistically literate and still fall for this one from time to time. What are the alternatives? Change your model so the old null becomes $H_1$? The only other option I can think of is power your study enough that a failure to reject the null is in practice close enough to confirming the null. E.g. if you want to make sure that adding a reagent to your cells won't kill off more than 2% of them, power to a satisfactory false negative rate.