Statistical software in clinical programs...

[ndlek21]

---

Members don't see this ad.

Simple question -- anyone want to share what your program uses for statistical software?

I ask because my undergrad institution has a renowned quantitative psych program and they are insistent that more and more researchers are gravitating towards R. In fact, I'm enrolled in an Advanced Statistics course right now...and I was hoping we'd be learning more advanced syntax in SAS or SPSS, but instead it's all about R. Is this going to be relevant or useful at all in graduate school and beyond? Do most programs still use only SAS/SPSS/Stata?

 

[Ollie123]

I'd say SPSS is still far and above the most common, but the faculty are right that R is becoming extremely popular.

I think more important than that is just developing the basic programming skills so that you "can" learn several languages. Once you get it, writing syntax should be the easy part. I don't use SAS much anymore, but when I need to it is extremely easy for me to just look at the manual, google things, and generate the necessary syntax. The only exception is if you want to do something incredibly novel (i.e. "invent" a new type of statistical technique), in which case it obviously helps to "really" know a program in-depth. The vast majority of folks won't be doing that and as long as you are planning to use the canned, developed techniques - once you get a feel for scripting it is quite simple to move between software.

 

[PsychPhDone]

It depends on the lab here. I mostly use SAS and occasionally SPSS, and in our intro stats sequence we used SAS and R. I think it's good to know a few programs, especially for analyses that are more intuitive in one program over another.

I found that knowing one statistical program well wasn't important in terms of being able to run analyses right away when I started my program. What did help was knowing I knew how to write syntax and that skill is something that you can apply to lots of programs. You just learn the quirks of each program, but that ability to think in a "coding" way helps.

 

[ndlek21]

Thanks so much, that answers my question perfectly.

I learned basic SPSS in my first stats class and the lab I worked in over the summer...and I used SAS for my thesis. So I guess I'll know R now too and I'll be familiar with all three. 👍

 

[becomethelight]

SPSS is pretty crummy and I would view knowing R as a strength. SAS is very good and I prefer it well above SPSS. The great thing about R is that it has so much greater functionality than either SAS or SPSS. The downside is there are no guarantees about the reliability of the software as far as bugs go, but then again it's free and most bugs end up being worked out in time. Bottom line - knowing R will be extremely useful in the future and would be a very marketable skill in applying to grad programs.

 

[futureapppsy2]

IIRC, R is becoming more proprietary, so that may affect it's popularity.

If you are doing any structural equation modeling, Mplus is WELL worth the cost, IMO. I like it so, so much better than AMOS.

 

[ResearchGirlie]

SPSS is expensive and isn't as flexible as R or SAS. However, SPSS is still the easiest to use with little to no training.

 

[psychstudent5]

Let me put my ignorance on display here: what is R? I've never heard of it. I'm most familiar with SPSS. Are you able to run all the same tests R that you can in SPSS?

 

[AcronymAllergy]

Let me put my ignorance on display here: what is R? I've never heard of it. I'm most familiar with SPSS. Are you able to run all the same tests R that you can in SPSS?

It's a statistical syntax language. Or more precisely, per the program's website: "R is a language and environment for statistical computing and graphics."

I've heard good things about its flexibility and such once you've learned it, and as an added bonus, it's free.

 

[Pragma]

IIRC, R is beginning more proprietary, so that may affect it's popularity.

If you are doing any structural equation modeling, Mplus is WELL worth the cost, IMO. I like it so, so much better than AMOS.

Seconded. Much better program. But you also have to learn syntax for it...

 

[bmedclinic]

IIRC, R is beginning more proprietary, so that may affect it's popularity.

If you are doing any structural equation modeling, Mplus is WELL worth the cost, IMO. I like it so, so much better than AMOS.

R is open source.
It's not popular (IMO) because it's not backed by someone and promoted. For example, SPSS is promoted/owned by IBM. It's also "point and click" (has a GUI). R is a language (you can use a variety of different GUI/Front ends with it... but it doesnt come with a 1800 number or as much help (as is typically the case with open source software). Kinda a "figure it out yourself, you're smart enough" atttitude.

BTW, not knocking R. I've used it and SAS and of course SPSS.
I'm a huge open source advocate, too... hence the above perspective.

 

[futureapppsy2]

R is open source.
It's not popular (IMO) because it's not backed by someone and promoted. For example, SPSS is promoted/owned by IBM. It's also "point and click" (has a GUI). R is a language (you can use a variety of different GUI/Front ends with it... but it doesnt come with a 1800 number or as much help (as is typically the case with open source software). Kinda a "figure it out yourself, you're smart enough" atttitude.

BTW, not knocking R. I've used it and SAS and of course SPSS.
I'm a huge open source advocate, too... hence the above perspective.

I know R has historically been open source, but last year-ish, I read some material from reputable sources that suggested that they were moving towards making the program at least partially proprietary. Perhaps they backed out of that, but it was the plan at some point in the recent past.

 

[Ollie123]

I think it would sink overnight if they made it proprietary so I hope they've abandoned that. S-plus has been around for years and is essentially a proprietary version of R so I fail to see what would be gained by having two of them unless they were going to try and undercut S-plus.

 

[psychstudent5]

It's a statistical syntax language. Or more precisely, per the program's website: "R is a language and environment for statistical computing and graphics."

I've heard good things about its flexibility and such once you've learned it, and as an added bonus, it's free.

Thanks for the clarification! I will check it out.

 

[bmedclinic]

Another thing worth mentioning in here is how your department is oriented. My masters in clinical was in a department that was mostly oriented towards SAS. We used both SAS and SPSS in our stats courses. When I showed up to the dept where I did my phd work, they pretty much only used SPSS. I dont know this for sure, but I bet if I sent my major advisor SAS output, he'd write me back and say "do it in SPSS and send it back to me- I cant find anything in there" or something to that effect. In essence, my suggestion is learn multiple programs for you, but make sure you know how to do it in SPSS. 🙂

 

[Telemachus]

SAS, Mplus, R. R is what's being taught in the intro grad stats sequence. The more gung-ho quant people in the department are pushing for more people to use R, and I agree with them.

 

[becomethelight]

I know R has historically been open source, but last year-ish, I read some material from reputable sources that suggested that they were moving towards making the program at least partially proprietary. Perhaps they backed out of that, but it was the plan at some point in the recent past.

There are several companies that have taken the base R package and then added some features and made it look prettier then slapped a price tag on it. They also have tech support and that sort of thing. S-Plus is one, R-evolution (Revolution?) is another.

I agree with some other posters here that Mplus is quite good, but I have to say that sort of like SPSS it makes things so easy I am wary about what it is doing to make them so easy and I get concerned that people may be using techniques inappropriately because it's so easy to do so.