Thank you for covering the statistical part of this study as well. Too often the statistical measures are not mentioned at all.
You are viewing a single comment's thread from:
Thank you for covering the statistical part of this study as well. Too often the statistical measures are not mentioned at all.
I appreciated seeing the p-values but unfortunately they completely invalidate the study in my opinion. A p-value of 0.08 is not statistically significant. The highest p-value I have seen considered statistically significant is 0.05. I personally prefer 0.01. That is influenced by having to prove statistical tests while working on my math degree in college and realizing they are based on multiple layers of assumptions.
You also have to keep in mind that statistical tests say nothing about causation, but can only reveal a probability of correlation. The article seems to imply that losing your teeth is a precursor to dimentia (With a p-value of 0.2 I reject the correlation altogether). However, what if it was some change in body chemistry that would later lead to dimentia that caused tooth loss?
You're welcome. I can only cover the basics though since I am not a statistician! I still come across studies that use methods that I have not even heard of and I have to go and try to figure out what they mean - luckily there are lots of online resources these days to help:)