New subscriber here! I am in a Statistics course for Nursing for Evidence-Based Practice, and you are life-saver. Your explanation of these difficult to grasp concepts are very effective in helping me gain an understanding of some of the concepts. I look forward to browsing through your playlist to see if you have covered a concept in my course that my book does not do well in teaching.
Thanks for the explanation, you did it quite well. I'm a Colombian student of economics, and I've learned a lot of statistics with this channel. It's interesting how in economics people tend to think that expensive policies despite being expensive and yields few results should remain, practical significance matters a lot.
You just earned yourself a subscriber!!! Thank you so much for a superb video! The examples were excellent! I will be taking quantitative research next semester and I am confident your videos will be useful! Again, thank you!
So statistical significance is a quantitative question or, in other words, a numbers question. Numbers, formulas, math, etc. And practical significance is a values question. If we determine that the action being studied has statistical significance (e.g., we determine that a vaccine actually reduces disease incidence and that the lower case numbers aren’t simply a product of chance), then we need to ask ourselves if that action is something that we, as a community, want to implement. The latter question, i.e., the practical-significance question, is about who we are, who we want to be, and how we want to allocate our limited resources. We might find it odd or unreasonable, but we can imagine a community that decides to dedicate enormous resources to lowering the incidence of a certain disease by only one percent. Different communities can answer the practical-significance question differently even though, in contrast, and because it is a quantitative question, they must answer the statistical-significance question in the same way. Is this a fair statement, or have I wandered off the path?
I agree with everything you said. I would go further, perhaps, by saying that more or less everything in the human world is correlated with everything else, and therefore it only takes a big enough sample size to have a fifty (50) percent chance of showing statistical significance. Also, the interesting results are the ones where a large sample did *not* show a statistically significant correlation in either direction. This is because such a result shows an exceptionally small effect. For example, if there is no detected correlation, with a large sample, between life expectancy and eating a certain food, we might conclude it is safe to eat it, and reasonably. On the other hand if there is a statistically significant correlation with lowered life expectancy, that on its own (without knowing the effect size) tells us essentially nothing, or if it does tell us anything, it is only that the food does not increase life expectancy if the sample is large, which is the same as what finding no correlation either way tells us. Does any of that make sense?
@@DataDemystified If you have time, please read and comment on my recent Medium article "What Statistical Significance Really Means" that got accepted by Nerd For Tech magazine. If you want to give me feedback privately, DM me on Twitter. I am @bartshmatthew on the latter.
New subscriber here! I am in a Statistics course for Nursing for Evidence-Based Practice, and you are life-saver. Your explanation of these difficult to grasp concepts are very effective in helping me gain an understanding of some of the concepts. I look forward to browsing through your playlist to see if you have covered a concept in my course that my book does not do well in teaching.
Very well explained, I finally got it after 1.5 years!
Happy to hear that!
Thanks for the explanation, you did it quite well.
I'm a Colombian student of economics, and I've learned a lot of statistics with this channel.
It's interesting how in economics people tend to think that expensive policies despite being expensive and yields few results should remain, practical significance matters a lot.
This was so interesting! Thank You for putting in the effort to make this content - it has honestly piqued my interest!
Thank you. Glad you found it interesting!!
very informative!
This is such a helpful video. I subscribed. I am so happy I found your channel to help me this semester.
You just earned yourself a subscriber!!! Thank you so much for a superb video! The examples were excellent! I will be taking quantitative research next semester and I am confident your videos will be useful! Again, thank you!
So happy you found them useful! Happy holidays!
So statistical significance is a quantitative question or, in other words, a numbers question. Numbers, formulas, math, etc. And practical significance is a values question. If we determine that the action being studied has statistical significance (e.g., we determine that a vaccine actually reduces disease incidence and that the lower case numbers aren’t simply a product of chance), then we need to ask ourselves if that action is something that we, as a community, want to implement. The latter question, i.e., the practical-significance question, is about who we are, who we want to be, and how we want to allocate our limited resources. We might find it odd or unreasonable, but we can imagine a community that decides to dedicate enormous resources to lowering the incidence of a certain disease by only one percent. Different communities can answer the practical-significance question differently even though, in contrast, and because it is a quantitative question, they must answer the statistical-significance question in the same way. Is this a fair statement, or have I wandered off the path?
I agree with everything you said.
I would go further, perhaps, by saying that more or less everything in the human world is correlated with everything else, and therefore it only takes a big enough sample size to have a fifty (50) percent chance of showing statistical significance.
Also, the interesting results are the ones where a large sample did *not* show a statistically significant correlation in either direction. This is because such a result shows an exceptionally small effect.
For example, if there is no detected correlation, with a large sample, between life expectancy and eating a certain food, we might conclude it is safe to eat it, and reasonably. On the other hand if there is a statistically significant correlation with lowered life expectancy, that on its own (without knowing the effect size) tells us essentially nothing, or if it does tell us anything, it is only that the food does not increase life expectancy if the sample is large, which is the same as what finding no correlation either way tells us.
Does any of that make sense?
I think that's a very clear understanding of statistical vs. practical importance! Thank you for the comment!
@@DataDemystified If you have time, please read and comment on my recent Medium article "What Statistical Significance Really Means" that got accepted by Nerd For Tech magazine. If you want to give me feedback privately, DM me on Twitter. I am @bartshmatthew on the latter.
so touching for an excellent video
Hi Prof , Thank you so much for posting this great video. Could you please make a video on effect size as well ?
Much Thanks.
Thanks! Not exactly what you're asking for, but in the ballpark; ruclips.net/video/95MyDVxWrO8/видео.html
@@DataDemystified Thank you:)
Thank you!
Thank you!!!! It finally is clicking. Whoa!!