Great series, both explanations and production, thank you! Just a sidenote - at 4:40 you mention that the only real difference when subjecting the same person to treatment A and then B is the treatment A and B (as the subject is the same) - to that I would add the history of the subject as another factor (being subjected to A before receiving treatment B may "change the person"). This is essential in experimental design, may limit what experimental design is possible (or impossible and then decide for group comparisons) and involves the neccessity of establishing the baselines first (as described in Sidman:Tactics od scientific research). As our professor used to say: "You can boil a pot of water (treatment A), then let it cool down and you are where you were at the beginning (baseline) and then you may apply treatment B (add salt and observe the time it takes to boil). But if you boil an aquarium with fishes, you get fish soup (no return to baseline is possible because its a "historical system" / "system with memory" (it changes irreversibly with treatment so you can´t establish a baseline to which you can return after applying experimental manipulation). Please keep up the good work! :)
Yes, some good points. This video was too short to get into all of these, and washout periods, and all that, etc. experimental design can be very important, and definitely worth exploring
Hello Mike, I had a small doubt please help me out here. So, as you mentioned Bivariate Analysis is the study of effect of X(i.e Independent Var) on Y(i.e. Dependent Variable) My doubt lies when you mentioned Paired and Independent groups If we have 2 Paired Groups then what is X and what is Y? This same questions goes for 3+ Paired groups, 2 Independent Groups and 3+ Independent groups. I have actually done some ML(very basic) and I need to understand how a dataset would look like if I had Paired or Independent groups. What would the X and the Y be in a dataset? Your videos are great Mike! I really appreciate how you boil down these concepts and make it easier for a complete beginner like myself, easier to understand.
You are an incredibly good teacher! Your videos are so helpful! thank you, thank you, thank you! I have a question, sorry if it may sound very naive, I am a beginner: I compare voice and speech measures of two groups: a patient group with a certain motor neuron disease affecting dimensions of voice and speech (n = 18, all male) and a healthy group from a neurological point of view (n = 38, all male too). We tried to match the ages (approximately two controls for one patient within the same range of age, but not always the exact same age). In this case, should I consider that my groups are paired or independent? I would say independent because I don't have the same number of men in the two groups. And considering my small samples, I would choose a non parametric approach to compare them. Am I right? thank you again for your very precious help all over the world! (I'm French)
so these are of numericals ... avery nice and vivid explanation...what out categorical data lets say outcome is a yes and no or for 3+ yesterday , today and tomorrow.. anyone is free to respond.. thanks
This would be 2 categorical variables, and you could do things like chi-square test, or others. We have a video over viewing some different bivariate approaches here: ua-cam.com/video/_m8v77qbkBA/v-deo.html
Sure, you can use: Chi square test, Fishers Exact test, test of two proportions, or calculate Odds Ratio or Risk Ratio and test if these equal 1 or not. We have videos explaining most of these tests
Great series, both explanations and production, thank you! Just a sidenote - at 4:40 you mention that the only real difference when subjecting the same person to treatment A and then B is the treatment A and B (as the subject is the same) - to that I would add the history of the subject as another factor (being subjected to A before receiving treatment B may "change the person"). This is essential in experimental design, may limit what experimental design is possible (or impossible and then decide for group comparisons) and involves the neccessity of establishing the baselines first (as described in Sidman:Tactics od scientific research). As our professor used to say: "You can boil a pot of water (treatment A), then let it cool down and you are where you were at the beginning (baseline) and then you may apply treatment B (add salt and observe the time it takes to boil). But if you boil an aquarium with fishes, you get fish soup (no return to baseline is possible because its a "historical system" / "system with memory" (it changes irreversibly with treatment so you can´t establish a baseline to which you can return after applying experimental manipulation). Please keep up the good work! :)
Yes, some good points. This video was too short to get into all of these, and washout periods, and all that, etc. experimental design can be very important, and definitely worth exploring
Great series, I took some biostatistics courses at university but these videos are much much better, fantastic explanation, thanks a lot!
Great presentation. I love the whole series. Thanks for the great videos.
Nice overview
Hello Mike, I had a small doubt please help me out here.
So, as you mentioned Bivariate Analysis is the study of effect of X(i.e Independent Var) on Y(i.e. Dependent Variable)
My doubt lies when you mentioned Paired and Independent groups
If we have 2 Paired Groups then what is X and what is Y? This same questions goes for 3+ Paired groups, 2 Independent Groups and 3+ Independent groups.
I have actually done some ML(very basic) and I need to understand how a dataset would look like if I had Paired or Independent groups. What would the X and the Y be in a dataset?
Your videos are great Mike! I really appreciate how you boil down these concepts and make it easier for a complete beginner like myself, easier to understand.
Thanks so much for these helpful videos!
Thanks a ot for the videos! Could you recommend a good book on statistics in biomedicine?
You are an incredibly good teacher! Your videos are so helpful! thank you, thank you, thank you!
I have a question, sorry if it may sound very naive, I am a beginner:
I compare voice and speech measures of two groups:
a patient group with a certain motor neuron disease affecting dimensions of voice and speech (n = 18, all male) and a healthy group from a neurological point of view (n = 38, all male too). We tried to match the ages (approximately two controls for one patient within the same range of age, but not always the exact same age). In this case, should I consider that my groups are paired or independent? I would say independent because I don't have the same number of men in the two groups. And considering my small samples, I would choose a non parametric approach to compare them. Am I right?
thank you again for your very precious help all over the world! (I'm French)
thanks, i appreciate that! i dont see any question here, so not sure what it was you wanted to ask...
I think they're independent as well
What if you have two independent groups and the DV is categorical and the IV is numerical? What test should I run?
I’d use logistic regression there, but there are other approaches that would be reasonable, depending on context
so these are of numericals ... avery nice and vivid explanation...what out categorical data lets say outcome is a yes and no or for 3+ yesterday , today and tomorrow.. anyone is free to respond.. thanks
This would be 2 categorical variables, and you could do things like chi-square test, or others. We have a video over viewing some different bivariate approaches here: ua-cam.com/video/_m8v77qbkBA/v-deo.html
I have to run a bivariate analysis where by IV and DV are both categorical. Is this even possible?
Sure, you can use: Chi square test, Fishers Exact test, test of two proportions, or calculate Odds Ratio or Risk Ratio and test if these equal 1 or not. We have videos explaining most of these tests
Did you actually train yourself to write in the opposite way ?