Thanks for sharing this! As a former astrophysicist who recently entered the field of medicine as a data scientist, I was quite shocked to discover how misunderstood and misused the p-value is in that field. I was not aware of this issue. Fortunately, in other scientific fields, the p-value is not so strongly abused, but the 'replication crisis' is becoming more and more of a concern in the scientific literature.
So, I'm currently struggling through a PhD stats course and I cannot tell you how valuable and helpful this was! To have somebody explain this in plain english has helped my own comprehension immensely! Thanks so much and on to the next vids!
I appreciate the effort being done by the BSI to evaluate scientific work. I need to push back a little on the premise of this. I don’t think there is anything inherently wrong with a p-value being utilized. The problem seems to be more in the interpretation of its value in regards to the findings. You also used the example of a detective and criminal evidence. In that example, of course you wouldn’t look at one piece of evidence to determine guilt or innocence. But when you engage in scientific studies, you are most often looking at only one aspect of the issue/topic. Change one variable and test and evaluate. Now change another, test and evaluate, etc. That’s why science has to build on itself and those “answers” take a very long time to get. There shouldn’t be any determination of truth or not based on one study. In reference to your BP example, yes we want to know if the drug works, but I’m not sure how you could “prove” that, outside of the standard scientific protocol over time? All that being said, there is absolutely a problem with data manipulation for the sake of a desired p value and a tendency, by both researchers and journals, to overemphasize the importance of a p value. I would just argue it’s not the statistical methods itself but rather the misinterpretation and inherent desire of people to want a quick, definitive answer. The onerous is on the reader to properly evaluate the study based on the design and statistics used and its eventual interpretation by the authors. Just my opinion! Thanks for the work being done to shed light on a system that needs optimized!
Thanks for sharing this!
As a former astrophysicist who recently entered the field of medicine as a data scientist, I was quite shocked to discover how misunderstood and misused the p-value is in that field. I was not aware of this issue. Fortunately, in other scientific fields, the p-value is not so strongly abused, but the 'replication crisis' is becoming more and more of a concern in the scientific literature.
So, I'm currently struggling through a PhD stats course and I cannot tell you how valuable and helpful this was! To have somebody explain this in plain english has helped my own comprehension immensely!
Thanks so much and on to the next vids!
I appreciate the effort being done by the BSI to evaluate scientific work. I need to push back a little on the premise of this. I don’t think there is anything inherently wrong with a p-value being utilized. The problem seems to be more in the interpretation of its value in regards to the findings. You also used the example of a detective and criminal evidence. In that example, of course you wouldn’t look at one piece of evidence to determine guilt or innocence. But when you engage in scientific studies, you are most often looking at only one aspect of the issue/topic. Change one variable and test and evaluate. Now change another, test and evaluate, etc. That’s why science has to build on itself and those “answers” take a very long time to get. There shouldn’t be any determination of truth or not based on one study. In reference to your BP example, yes we want to know if the drug works, but I’m not sure how you could “prove” that, outside of the standard scientific protocol over time? All that being said, there is absolutely a problem with data manipulation for the sake of a desired p value and a tendency, by both researchers and journals, to overemphasize the importance of a p value. I would just argue it’s not the statistical methods itself but rather the misinterpretation and inherent desire of people to want a quick, definitive answer. The onerous is on the reader to properly evaluate the study based on the design and statistics used and its eventual interpretation by the authors. Just my opinion! Thanks for the work being done to shed light on a system that needs optimized!
You are both saying the same thing.
Thank you Emily! You have helped me understand better the p-value....
.05 is not "significant" in any logical sense. More BS. Google the originator of "statistical significants"
en.wikipedia.org/wiki/Statistical_significance