useful to know STATA or SAS (stat packages) to do research?

[waxedapathetic]

I imagine it is NOT necessary, but would it be useful? i am planning to study HOW to do health services research next year before starting a Gen Surgery residency.

I can take courses on doing stats using these packages. i'm just curious if it will be an asset for me as a future researcher / resident researcher doing health outcomes. will i be able to use and market these skills? (or do surgery MDs just leave the data up to their statisticians to do it for them?)

thank you in advance for advice.

---

Members don't see this ad.

 

[rddoms]

Look at the latest version of Graphpad Prism. That software is the best I've seen for research statistics. It can do anything in regarding statistics, and automatically makes the graphs for you.

Knowing SAS may be helpful, but it is such a pain to use.

 

[deleted89854]

I've only found SAS useful for graphing and plotting statistics. For the Fisher's, T-test, ANOVA, etc I've used SPSS. It can be a real pain until you figure out how to use it.

 

[Lance]

Learn SPSS. I learned all three of the major packages (SAS, STATA and SPSS) and use spss almost exclusively for research. It's not as strong of a package for advanced statistics, but will cover anything you need to do for surgical research, and can be used as a menu driven program instead of needing to remember any syntax, but will still generate an easily reproducible program if you want it to. Plus the student version is pretty cheap.

 

[SocialistMD]

I've been using SPSS for the last two years, but I just got GraphPad Prism and it is rapidly replacing SPSS.

 

[surg]

Pick any package you like and learn it well. I personally found SPSS the easiest to learn for stuff. You will want the GradPack version since you are still a student. It is cheap and does virtually everything you would need as a non-statistician.

SAS on the other hand, is the work horse of statisticians everywhere. If I could do it over again, I might have taken the time to learn SAS because some of the bigger datasets in the world come with SAS code to unpack it, which is a pain for me, since I don't have a copy. It requires a fair bit of training and practice to keep up with it though. (Think using UNIX v. using Windows as your operating system. UNIX gives you better access to the basic code, and is faster once you know it, but Windows is way easier to learn).

STATA is preferred by a number of health researchers as well. I've only used it a few times, but it has some nice features. Negatives are: the ugly looking interface, and some limitations on the size of the dataset. On the other hand.

As to your other questions, being able to do your own basic stats is a god send. You will end up needing a statistician reviewer for really important studies and difficult things, but being able to do your own stats will make you a popular person in your research group!

 

[waxedapathetic]

Thanks so much for the advice!

 

[tbo]

My quick answer is also SPSS. With time, it will be the best balance of power and ease-of-use.

My more involved run-down is that SAS and R are the leaders with serious statisticians. Both of them are designed for the programmer and so the interface seems scant at best. However, this allows the user to design whatever statistical test or battery of tests they choose, giving them pretty much free reign over their data.

Next is SPSS and STATA. 95% of the tests available to the above 2 "power programs" are available in SPSS, but with a much easier interface to use. SPSS alongside a good textbook (I recommend Altman's Practical Statistics for Medical Research) will help you navigate intelligible statistical analysis. Edit: I forgot to note, the comments above are good, but SPSS (I know for sure) and STATA (I believe) all can importa .sas files in SAS format, so that shouldn't be an obstacle in deciding which to learn.

The other stuff like GraphPad or SigmaStat are not highly regarded particularly in statistical circles, so if you intend to include the statistical package used in your publication, perhaps think twice, or get your stats verified by someone in a Stats department with more rigorous programs. Additionally, I've gone through peer-review comments where they specifically get a statistician or someone well-versed in statistics to critique the data, so the rigor of your analysis (and consequently, your statistical package) is not trivial.

The many things to consider when getting into statistics. And good luck. There are very few physicians that really know statistics well, so you would be a prized commodity.