May 2021

Some odds and ends I have come across in the last month:

Interesting article on reproducibility.  It doesn't only hit the social sciences:  Reproducibility issues in physics:  https://www.nature.com/articles/d41586-021-00954-8?fbclid=IwAR0qbalwMzWgvY0_4NS4gDsJ2YRr6nwZU1j1xtRglJBcI1AyJFjzuoFKlA0

Magic mushrooms for depression:  https://www.sciencemediacentre.org/expert-reaction-to-phase-2-trial-comparing-psilocybin-and-escitalopram-for-depression/?fbclid=IwAR35Zad_LqV9E_ZOh13xDms20vpOosEgWs2aLjWPEpkcFVcXtV64m7KRrB0

Good overview of many of the things that are often done wrong in presenting graphical information: https://getpocket.com/explore/item/3-questions-to-ask-yourself-next-time-you-see-a-graph-chart-or-map?utm_source=pocket-newtab

Another article on presentation of information: https://getpocket.com/explore/item/designers-and-statisticians-disagree-on-what-makes-a-good-information-graphic

Last month there was a request to review Odds Ratios and related calculations.  After giving it some thought and realizing that these are topics that are best presented with visual aids, I thought it best to curate some of the better videos I have found on the topic rather than trying to present the material in a meeting that most seem to attend via phone.  The links below cover Odds Ratios, Relative Risk, Risk Difference, and Specificity and Sensitivity.  All are basically contingency table calculations but different approaches are best used to answer different questions.

If you want to understand what is going on in calculating odds ratios, Mike Marin's videos are a good resource. He actually presents things in terms of probability and shows how the  traditional A-B-C-D table is set up.

In video #30 Marin discusses Odds Ratios, Relative Risk, and Risk Difference, what they are and how they relate to each other.
In video 31 he looks at the application of contingency table analysis to case control study calculations.

#30 - Odds Ratio, Relative Risk, Risk Difference : https://www.youtube.com/watch?v=JmuciUfCJ_w&t=151s  Number needed to treat gets a brief mention(at time  6:05), where he notes that it is essentially 1over the Risk Difference.This is a good overview  of the relationships among the various calculations.

#31 - Case Control Study and Odds ratio: https://www.youtube.com/watch?v=NqsSun9HZfI&t=450s

Specificity and sensitivity (Not Marin, but a good video from the Clinical Information Sciences YouTube Channel: https://www.youtube.com/watch?v=Doa_QqtAexU&list=WL&index=9

Unfortunately Marin does not have a video on Number Needed to Treat. TheNTT.com does, however, have a good explanation. https://www.thennt.com/thennt-explained/

If you only need to do the arithmetic (easy enough by hand, but computers make many fewer mistakes in basic calculations), this link goes to several calculators, one of which is NTT.  https://clincalc.com/Statistics/

Here is an interesting (but kind of lengthy) article on p-values: https://surfdrive.surf.nl/files/index.php/s/gi8KrNL7vjTzOmB

A little rusty on basic statistical concepts? https://www.youtube.com/watch?v=kyjlxsLW1Is This sit is also a good resource for general statistical concepts.  

Statistical software:  Depending on the background of your undergraduate statistics professor, you likely learned to use SPSS, Excel, SAS. or R.  

SPSS seems to be the most commonly used package but costs money to obtain.  Most of my students will never use a statistics package enough (or need the sophisticated analysis capabilities), so it isn't one I tend to use.

Excel tends to be used in business statistics courses.  Professional feel researchers use SPSS, but most students have access to Excel and will continue to have access to it when they get out into the business world.  Real statisticians hate excel, but it is probably good enough for most purposes.

SAS is another professional level package but does not seem to be as readily available as SPSS.  I grew up with SAS and have always preferred it but haven't had reason to use it for years.

R is a free, cross platform (Mac, PC, and even Linux) software package and is the one I use for my undergraduate classes.  Combined with an IDE such as RStudio, it is reasonably easy to learn for basic analyses and, with add-on packages (also generally free) can be used for most any analyses you might need to run.  The more advanced features are also more difficult to implement, but basic analyses are not too bad at all.

JASP and Jamovi - Are both free, R based packages that add a GUI to the basic functionality of R.  This makes them act more similar to SPSS but also somewhat limits the functionality to those things that are programmed into the interface. I have not used JASP and have not used Jamovi for a while (last time I downloaded it, the program claimed all of my data files as its own and they all opened automatically in Jamovi.  The only way I could reclaim them was to remove the program.  As I was using RStudio to teach my classes it seemed a good idea to remove Jamovi.

Both JASP and Jamovi get good reviews and are probably worth a try if you want to avoid being made fun of for using Excel but don't want to spend the money to get SPSS or SAS.  JASP is particularly liked by people who are in to Baysian statistical analyses.

Links:
https://www.r-project.org/ for the R software and

https://www.rstudio.com/products/rstudio/download/ for the RStudio IDE (Get the free desktop version.)

https://jasp-stats.org/  to download the JASP software

https://www.jamovi.org/  for Jamovi