A Guide To Odds Ratios: What They Are and How To Communicate Them Clearly
ฝัง
- เผยแพร่เมื่อ 4 มิ.ย. 2024
- Despite being rather counter-intuitive Odds Ratios are frequently used to report the results of research in the medical and social sciences. Professor David Spiegelhalter explains how they're calculated, and how to turn them into something more meaningful for effective communication.
🔗🔗 Links🔗🔗
RealRisk: a tool for science communicators 👉👉 realrisk.wintoncentre.uk
The Winton Centre's website: 🌐 wintoncentre.maths.cam.ac.uk/ 🌐
Risky Talk the podcast with David Spiegelhalter: 🗣️🎙️ riskytalk.libsyn.com/
David Spiegelhalter is a statistician, author and broadcaster. His book THE ART OF STATISTICS is a worldwide bestseller. He was the Winton Professor for the Public Understanding of Risk at the University of Cambridge for many years, and served a term as the President of the Royal Statistical Society.
From the way this guy talks I trust him as a teacher
He does a good job communicating risk and odds, that after a decade of reading reports about RR and OR, I can explain what they really mean. Thanks for saying not to communicate odds.
This was so nicely explained! you could sense the wisdom of the man from the way he was teaching.
I do my diploma in clinical research , and this video is really informative and recommended for who are interested
Thank you Professor Spiegelhalter for this excellent and pedagogical explanation.
Awesome explanation! Thank you.
This is brilliant! Thank you.
Дякую! Робіть більше таких відео. У Вас є талант пояснення концепцій. :) (Ukraine \ Irpine)
Thank you very much, Prof.
Thank you so much! This helped a lot!
Thank you, Prof.
Thank you from France !
Thank you so much!
Thank for video
Thank you❤❤
I loved it !
This is great
Thanks for the explanation of OR.. that paper on depression.. according to the authors who tried to associate use of cannabis to depression.. what if those who use cannabis where already depressed..?
I (student) was asked what they were in an interview and explaining clearly really stumped me. Do experienced statisticians have this problem too?
Maybe because they are simple or unimpressive (when coming from anovas and regressions) and are more part of formatting and etiquette. (thats why i watch the video, because im missing that cool feeling of understanding and want it to be more complex than it really is). so my motivation would just be "oh i have to learn it, because i need to put it in my paper before handing it in"
Professor Spiegelhalter…this is a fantastic presentation. I have shared to my students and residents from all walks of life. A question arose that I was hoping you could help with…and it's silly…what kind of "white" board are you writing on? I'm renovating one of our teaching rooms and I love the way that looks. Thanks!
Great to hear it, thanks Evan! Re board - we used one of the teaching rooms in the University of Cambridge maths dept. They're basically chalkboards, but we used liquid chalk markers for a smoother colour. Hope that helps!
@@TheWintonCentre Thanks so much! I appreciate the tip. Take care and stay safe!
Isn't there an imbalance between the two groups since there are only 200ish in the depressed group and nearly 2500 in the non-depressed group? Is this why the odds ratio in the actual paper is adjusted?
Can someone help me out please? When he calculated the RR, for the depressed group, how did the % become 15, when for the non depressed it was 5% if using the formula of RR?
34 / (34+187) = 0.15 = 15% and 132 / (132+2366) = 0.05 = 5% and 15% / 5% = RR of 3
i cant understand the difference [ odds - proballity ] from dies to come six ..i know the probality is 1 from 6 ..but what does 1 out of 5 means ?
The Odds of an event is the number of outcomes of interest over the number of *other possible* outcomes. Whereas the Probablity of an event is the number of outcomes of interest over the number of *all possible* outcome. So, for rolling a die, there's one outcome you're keen on and 5 other possible outcomes. Odds = 1/5. But in probability, there's one outcome you want but 6 possible outcomes in total!
How do you add multiple odds ratios? Let’s say a patient has multiple risk factors for developing schizophrenia, each risk factor with a 2 fold possibility. (Ie having an older father, smoking cannabis, birth complications). Would you add them all together or multiple them?
This video is like a box of chocolates, you never know what you're gonna get
Handwriting ✅
34 from132 ~ 1/4 or ~25% of smokers r depressed & 187 from 2336 ~ 1 /12,5 or 8% non smokers r not depressed
i believe is the same numpers ? butwhy i find the above way more logical to calculate
Either way to look at the association is valid I believe. In general you'd want to present the association based on the variable you're interested in, not the one you're controlling for ... I think... Like if this is a study of cannabis and you want to see how many are depressed, you'd do it your way. However if this is a study of depression and you want to see if there is a link with cannabis, you'd present it his way...
Good point to raise though. I'd appreciate feedback on my take (as I'm unsure if I'm correct).
@@danbolser5913 Yes, to my mind Spiegelhalter calculates the figures in the wrong direction - across the rows rather than down the columns, though the number comes out the same in the end anyhow. My background is in Social Science, where the convention is for the independent (causal) variable at the top and columns to add to 100%. So in this case, cannabis use influencing depression not the other way around. I would frame the question as such: What are the odds of being depressed if a regular cannabis user - divided by, the odds of being depressed if not a regular user. The calculation would then be: (34/132)/(187/2366) = 3.3. If a person smoked cannabis, they would be 3.3 more times likely to be depressed.
@@shamusenright5387 We don't know if depression causes increased cannabis use, or vice versa. Like many things in social science, it's an association but we're not sure of the causal relationship. Plus this was an observational study - so there is no independent variable.
@@TheWintonCentre Thanks for the reply. I agree that in this case we don’t know which variable causes the other. But is that strictly true regarding no independent variables in observational studies? Say for example, you were looking at sex and student education levels. You could say sex is influencing education levels but it wouldn’t make sense to say the opposite, education influences a person’s sex. Sex is determined before education and therefore the independent variable is it not?
shouldnt the percent for depressed be 18 percent? how come is 15 percent?
No - the *odds* is 0.18, the *probability* is 0.15, which is the same at 15%. Odds = 34/187 where as the probability is 34/(34+187).
@@TheWintonCentre thanku