Learning stats

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

ru4real1666

Full Member
10+ Year Member
Joined
Feb 26, 2009
Messages
342
Reaction score
15
Anyone have any specific MOOCs (like coursesa etc) or other sites (youtube, etc) where you successfully learned stats for research projects?

Looking for some advice on approaching this!

Members don't see this ad.
 
Well, statistics can be a very complex subject and how to apply it for research project means that you must have some basic understanding of the question being asked before to can apply statistics appropriately. But anyway...

There is an interactive textbook and it has built in examples.
http://vassarstats.net/textbook/

Additionally, it might be worth buying a cheap, basic statistics book.
Amazon product
 
  • Like
Reactions: 1 user
Members don't see this ad :)
For the basics, I like OpenIntro Statistics--it's free, explains the concepts very well, and can help you run analyses yourself with R or SAS. It's supported by a few MOOCs on Coursera--formerly "Data Analysis and Statistical Inference" and now a "Statistics with R" series.
 
  • Like
Reactions: 1 user
I second the notion of getting an introductory statistics textbook (doesn't need to be in medicine or public health). I'll give two qualifiers, though. 1) Get an applied statistics book that has real data examples you can use, especially if you don't have the mathematics background. Do the problems in it with an analysis software and some by hand. 2) Make sure the author of the book has a Ph.D. in Statistics or Biostatistics (an MS would be okay too). Epidemiology/public health/psychology/nursing/medical degrees won't make the author an expert in statistics, despite that they may have experience with stats. I'm not trying to knock anyone without a statistics degree, but this is based on readily observable information. The literature is rife with examples of bad statistics from non-statisticians, including teaching things that are flat out incorrect.

If you have a book from a qualified author, you can be much more confident in what you're learning. If you have a truly good foundation, you can mobilize that knowledge from field to field (i.e. an applied stats book from business written by someone with a PhD in Stats can allow you to take those ideas and apply them to biomedical research). After you've nailed down some of the fundamentals, I would also recommend learning multiple linear regression (and simple, obviously). You can then move on to logistic regression or survival analysis. The latter two are very common in biomedical research also, but the former will make these easier to understand the others.
 
Second "Biostatistics for Dummies" as a good overview. Another is "Intuitive Biostatistics" by Motulsky

Johns Hopkins open courseware page has quite a few stats and biostats and epidemiology courses available. Some are pretty good. L
 
  • Like
Reactions: 1 users
Second "Biostatistics for Dummies" as a good overview. Another is "Intuitive Biostatistics" by Motulsky

Johns Hopkins open courseware page has quite a few stats and biostats and epidemiology courses available. Some are pretty good. L
While the Biostatistics for Dummies isn't terrible, it does a good job at botching some elementary concepts and introducing common misunderstandings as accurate statements. Without developing a big list, a few examples are: they misinterpret the meaning of a confidence interval (claiming a particular interval has a 95% chance of containing the true parameter value), they do an awkward job while botching the p-value, and they conflate a multivariable regression (multiple independent variables) with a multivariate regression (multiple dependent variables).

About JHU, I think they have some decent free videos on coursera under the biostats 2 boot camp, if I recall.

Edit: I actually went and took another look at it, and they definitely botched the p-value.
 
Last edited by a moderator:
Top