[Interview]: Dason - that guy that isn't a bot

CB

Super Moderator
#41
I've been a fan of Bayesian methods for quite some time so doing more with that kind of stuff would definitely be of interest to me.
How Bayesian are you these days? Do you think Bayesian analyses are best in most situations? And if so, why are we still using NHST so much?
 

Dason

Ambassador to the humans
#43
How Bayesian are you these days?
I'm typically as Bayesian as I can get away with. A lot of the nice methods for working with microarray data tend to be "empirical Bayes" type methods though. So I guess most of the time I'm an empirical Bayesian (whatever that means exactly). These methods typically seem Bayesian but you use the data to help specify your prior (on some of the parameters). It sounds like cheating but if done properly you can justify it in from a fully Bayesian point of view.

Do you think Bayesian analyses are best in most situations?
It's hard to define "best". I certainly like Bayesian methods and prefer the interpretations. I think it makes sense to incorporate prior information if we have any and I don't have a problem using "reference"/"noninformative" priors in the other situations. So I definitely prefer Bayesian methods. However, if you have trouble specifying a data model things get a little murky. I think nonparametric methods have their place as well. If you're not comfortable saying the data has a certain distribution given certain parameters then it's hard to justify certain Bayesian methods. If you're doing a lot of repeated testing (like in certain industrial processes) then the frequentist framework makes a lot of sense too. So there is a time and place for most methods I think. I tend to prefer the Bayesian methods if I can get away with it.

And if so, why are we still using NHST so much?
Like I said in the last question in some situations I think NHST makes a lot of sense. But I do think NHST gets done more often than necessary. Most of the time we know the null is false. We should probably focus more on confidence/credible intervals since I think those tend to be more interesting and get at the questions of interest more.
 

Dason

Ambassador to the humans
#44
what *is* your favorite firefly episode?
Out of Gas. Definitely. I loved learning the backstories for (some) of the characters. I also really liked the last episode - I don't remember the name right now but it's the one with Early - the bounty hunter. But really all of the episodes are great. There are some episodes I don't like as much as the others but I think there is something awesome in every episode.
 

Dason

Ambassador to the humans
#46
Who is Hamid. I don't think I know that poster.

EMP would influence the entire continent :p
Hamid's famous post.

And for the EMP - it depends on what type of EMP and the altitude it is at when detonated. But still - I don't want to be anywhere in the general vicinity (where the "general vicinity" means anywhere in the blast radius).
 

trinker

ggplot2orBust
#47
You seem to be a bot of science and most statisticians enjoy evidence for their beliefs and you say you believe in God thus I ask what is your evidence for a belief in God?

You asked for something hard...
 

Dragan

Super Moderator
#48
...
I would love to be teaching at a teaching university. I think I could work in industry for a while - but I don't think I want to end up at an R1 university where research is everything I do and teaching is just an annoyance on the side. But ultimately I want to end up teaching so if I could get there in 5 years that what I would love to be doing.
Dason, I think you're underestimating your talent (don't you?). In other words, I think you would make an excellent academic scholar in the future. Note that it's very difficult to continuously publish research articles in "top notch" journals e.g. JASA -- very few people who finish their Ph.D. have the ability to go on and do this. Why don't you strive to do this i.e. maximize your (obvious) talent? I suspect that after a number of years you would get bored (or worn out) with just teaching and with little or no interest in publishing research.
 

noetsi

No cake for spunky
#49
lol on the hamid thread. I was wondering if it was the guy who offered money for assistance :p

People who publish in elite journals are 1) either name professors already or 2) have some highly topical topic. At least in the journals I know. Maybe it is different in statistics.

Interestingly, in social sciences, people at elite universities don't get published in elite journals all that much. The answer I have been given is that they deliberately avoid this to work on books. Again statistics may be an exception to this, I dont know.
 

Dason

Ambassador to the humans
#50
You seem to be a bot of science and most statisticians enjoy evidence for their beliefs and you say you believe in God thus I ask what is your evidence for a belief in God?

You asked for something hard...
Empirical sciences can't do everything. Some questions science can't answer. What is the meaning of life? What happens after you die? Some things can't be answered using empirical evidence. Some of the reasons for my belief in God I'm sure could be explained away by science. Some of my reasons I know couldn't be explained away. But I don't see an inconsistency in believing in God and using science to attempt to answer some questions about how things work.
 

Dason

Ambassador to the humans
#51
Dason, I think you're underestimating your talent (don't you?). In other words, I think you would make an excellent academic scholar in the future. Note that it's very difficult to continuously publish research articles in "top notch" journals e.g. JASA -- very few people who finish their Ph.D. have the ability to go on and do this. Why don't you strive to do this i.e. maximize your (obvious) talent? I suspect that after a number of years you would get bored (or worn out) with just teaching and with little or no interest in publishing research.
Thank you for your compliments. Maybe I am underestimating myself but that doesn't change the fact that I prefer teaching. Maybe once I do more research I'll find that I enjoy that more - but I know I really enjoy teaching. That doesn't mean that I wouldn't continue to do some research - just that it wouldn't necessarily be my top priority.
 

trinker

ggplot2orBust
#52
I am in a similar mind set as dason with teaching university versus a research institution. I have people in my department saying this is a waste of talent, this is not as stressful as you think it is but the reality is Dason has to go were he's going to be useful and happy. Soon he'll have little bots running around and I think a teaching institution allows more flexibility with duties than a R1. (IMHO) Plus some people are born teachers, Dason obviously once had an interest in teaching; I find when you don't let a teacher teacher teach they become aloof. Plus think of the future scholars Dason may impact.

My advice and thoughts are you can be influential wherever you wind up and you can make an impact. Both types of institution would desire your skills. Go where you think it's best for you and your family and the God you believe in will work out the rest :)

Thanks for volunteering to go first; I learned a lot about you and enjoyed getting acquainted more with the Dason.
 

vinux

Dark Knight
#54
Dason, I think you're underestimating your talent (don't you?). In other words, I think you would make an excellent academic scholar in the future. Note that it's very difficult to continuously publish research articles in "top notch" journals e.g. JASA -- very few people who finish their Ph.D. have the ability to go on and do this. Why don't you strive to do this i.e. maximize your (obvious) talent? I suspect that after a number of years you would get bored (or worn out) with just teaching and with little or no interest in publishing research.
I agree with Dragan. Research( publications) is highly rewarded compare to a good teaching. Also a good research life opens lot of opportunity to teach bigger crowd and impact/contribute to society. I guess the tenure in US universities also depending on quality publications. In my case, Industry experience didn't add much value in my research life( It was helpful in teaching).

Although I don't have plan to follow the research life. The main reason is I take lot of time to write a paper(I can finish all theory/modeling work much faster). In my view good writing is most important part of research life.
 

Dason

Ambassador to the humans
#55
Thank you everybody for your interest and for participating over the course of the week. My active period is over so try to focus questions on the current person being interviewed. That doesn't mean you can't still ask questions about me here though. So if you have a question you're dying to ask me I'll still make an effort to answer them.