Useful to know R?

[Señor Chang]

Is it useful for medical students and residents to know how to use statistical software like R? Assuming they are not super into research.
If a doc is on a research team, I imagine the stats is going to be done by a PhD researcher while the doc is probably in the clinic (gathering data), so is it really that useful? Note: I know that understanding stats is very useful, I just dont know if knowing how to personally do the analysis is.

---

Members don't see this ad.

 

[mimelim]

Is it useful for medical students and residents to know how to use statistical software like R? Assuming they are not super into research.
If a doc is on a research team, I imagine the stats is going to be done by a PhD researcher while the doc is probably in the clinic (gathering data), so is it really that useful? Note: I know that understanding stats is very useful, I just dont know if knowing how to personally do the analysis is.

If you do research in medical school or residency, you should know the basics of R, STATA or some other package. R is the new kid on the block in medicine and very few people use it currently.

I am a resident currently and we do all of our own analysis, it just is too expensive to justifying having our in house people do it. And we use R exclusively and have built our own medical packages for our field.

 

[TheLionheart]

While it certainly wouldn't hurt, I do not believe it necessary to know much beyond the "basics"--e.g. standard deviations, T test, ANOVA, etc.

Advanced statistics, like the kind you see in complex meta-analyses, is something I would best leave to someone who is a stats wizard. We enlist people like that to help us in the clinical projects I've participated in--but this is only in my limited experience on a handful of clinical projects.

 

[Pacna]

Soraka's going to mess your R up, anyhow.

 

[mimelim]

While it certainly wouldn't hurt, I do not believe it necessary to know much beyond the "basics"--e.g. standard deviations, T test, ANOVA, etc.

Advanced statistics, like the kind you see in complex meta-analyses, is something I would best leave to someone who is a stats wizard. We enlist people like that to help us in the clinical projects I've participated in--but this is only in my limited experience on a handful of clinical projects.

I agree. I would also add life-table analysis/Kaplan Meir to your basic skills list, but the more complex stuff should be done by someone with a reasonable amount of experience. I have a stats grad student and my brother (applied math/stats undergrad) who consult for us on anything that I don't feel comfortable with. If we had some really high end or complicated stuff, I'd find the funding to have our actual stats people do it.

 

[NoDakDok]

Soraka's going to mess your R up, anyhow.

Is that a LoL reference on SDN?

I agree. I would also add life-table analysis/Kaplan Meir to your basic skills list, but the more complex stuff should be done by someone with a reasonable amount of experience. I have a stats grad student and my brother (applied math/stats undergrad) who consult for us on anything that I don't feel comfortable with. If we had some really high end or complicated stuff, I'd find the funding to have our actual stats people do it.

I have a fundamental understanding of statistics and the usage of R as well as several other statistical analysis programs, but all the research projects I have worked on had their own data analysis specialists through the hospital's research system. Obviously this is anecdotal, but at least in my experience the understanding of statistics, what they mean, and the reasonable conclusions to be drawn from their output is more valuable as a clinician researcher than actually knowing how to do the statistical analyses themselves. However, obviously knowing how to do them helps significantly with the former.

Also, I think you gain a significant amount of respect from being able to do your own stats, even if they're only preliminary. As a student researcher, I know my PIs have all appreciated the preliminary analyses I've done so they have some indication of where the research is going previous to when the project finally gets put through the statisticians. Looking better in the eyes of your superiors is always a good thing, and this is a great way to 'brown nose' without feeling like you're brown nosing.

 

[Pacna]

Is that a LoL reference on SDN?

No, I'm talking about the statistical software Soraka. Wtf is LoL?

 

[NoDakDok]

No, I'm talking about the statistical software Soraka. Wtf is LoL?

I don't believe you. The only results I could find under Soraka are LoL-related.

 

[Pacna]

I don't believe you. The only results I could find under Soraka are LoL-related.

 

[Awared]

If you do research in medical school or residency, you should know the basics of R, STATA or some other package. R is the new kid on the block in medicine and very few people use it currently.

I am a resident currently and we do all of our own analysis, it just is too expensive to justifying having our in house people do it. And we use R exclusively and have built our own medical packages for our field.

Each med student at my COM gets 4 free hours with the stats people for our research. It's awesome 😀

 

[Narmerguy]

It's pretty simple to use R for basic stuff (means, standard deviation, linear regressions, multiple linear regressions, t tests, etc). You could probably be up and running doing that with a 2 hour youtube video that would tell you everything from how to open files, run the analysis, and make your plots. Most people won't be doing anything sophisticated with R unless you're doing gene expression analysis and wizzy stuff like that.

 

[mimelim]

No, I'm talking about the statistical software Soraka. Wtf is LoL?

I have no idea what you are talking about or what Soraka is. Google didn't reveal anything remotely related to this thread.

 

[phunky]

I would argue it is extremely useful if you want to do any sort of research. I've been a research assistant for a couple of years, and I got sick of waiting weeks or months for one of the statisticians to have time to work on my projects, so I took the Data Science specialization on coursera and learned R.

I certainly can't do nearly as much as the biostatisticians I work with, but a lot of biostats is really quite simple. You can at least preform your exploratory analyses, formulate crude association, and create basic statistical models without a PhD. As long as your work is later checked by a real statistician.

 

[ridethecliche]

I use sigmaplot for analysis. It's a bit of a PITA to get everything imported in there, but the analysis is sound and it's relatively idiot proof as it tests for normality and suggests better tests for the data depending on that.

 

[operaman]

I use SPSS and/or Prism for most of my basic stuff. Prism is kinda like a stats package for dummies and really helps walk you through a lot of it. Not as powerful as others, but nice for basic stuff and great for learning. I haven't really done much with R but think it would probably be worth learning if you have time and an inclination toward research. Even aside from that, being good with stats will help you a lot. It amazes me how many people in medicine don't really understand even the basics (and my knowledge is fairly weak compared to real stats guys).

 

[NoDakDok]

I have no idea what you are talking about or what Soraka is. Google didn't reveal anything remotely related to this thread.

He was evidently making a joke, but this a pretty odd demographic to make the joke to. That's why I was unfunnily skeptical. LoL is a video game and "Soraka" is a character from it.

 

[Narmerguy]

😕 OK I guess...

 

[surg]

If you have any intention of doing research, being able to perform your own basic stats is probably the most important skill you could pick up. While you should liberally use statisticians whenever feasible, being able to do correlations, comparisons, regressions, and life-table analysis will get you a long way. Statisticians are always in demand and overworked, so people who can act as amateur statisticians end up on a lot of extra papers because they get brought into a lot more projects both at the design stage and the analysis stage.

Essential though to work within your knowledge base and have a professional statistician check your work until you are very secure in your techniques. The new packages are getting easier to use, and thus easier to abuse by applying the wrong tests, or the right tests but in the wrong way. As a reviewer, I often find myself having to send back papers because I find their stats don't add up properly.


Sent from my iPad using Tapatalk

 

[WillburCobb]

He was evidently making a joke, but this a pretty odd demographic to make the joke to. That's why I was unfunnily skeptical. Just as a fun cultural fact for you:

Basically, Soraka is one of the heroes (called "champions") in the world's most played video game, League of Legends (aka "LoL"). "R," named after the keystroke used to activate the ability, is a reference to any hero's "Ultimate," the best ability for any particular champion. By saying "Soraka's going to mess up your R, anyhow," he's just saying that she is going to screw over your ultimate ability. The most recent numbers state that nearly 70 million people play League of Legends every month (about a quarter of the population of the US, though spread over the entire world). It is well known by a significant portion of the 16-25 year old demographic. It isn't a stats program, and that's why you're getting such strange results in Google.

 

[NoDakDok]

😕 OK I guess...

Don't worry, it wasn't funny before I explained it either.

@WillburCobb, aren't you supposed to be sick? 😉