Its so cool to find people that try to help without any money reward. Thanks for the video. I like how you really go to the point. I wanted to ask you what would be an example of a question on STEP1 for this topic?? Thanks in advance
Omar Ciano Thank you so much for the kind words! I really appreciate the encouragement. Here are some CI example questions I wrote An investigator is researching the link between obesity and myocardial infarction. A retrospective case-control study is designed. 10,000 patients who have been diagnosed with myocardial infarction within the last year have been identified and 10,000 similarly matched controls with no diagnosed myocardial infarctions are selected as controls. Retrospective data about the participants blood pressure over the last 5 years is then collected. The odds ratio for hypertension among the myocardial infarction group was 2.6 with a 95% confidence interval of 1.5-3.3. Based on these results, which of the following is most correct? A. Hypertension causes myocardial infarctions B. Myocardial infarction causes hypertension C. Myocardial Infarction does not have statistically significant association with hypertension D. Myocardial Infarction has a statisticallysignificant association with hypertension \ E. Hypertension has no correlation with Myocardial Infarction Correct answer is D A researcher is investigating the link between alcohol consumption and liver cirrhosis. A prospective cohort study is designed with 100,000 participants. The study population is divided into groups based on their self-reported daily consumption of alcohol. The participants are monitored for 20 years for the incidence of cirrhosis of the liver. The results are reported below. GroupNo alcohol consumption, RR 1.00, CI --- 1-2 drinks per day RR 1.05, CI 0.93-1.14 3-5 drinks per day RR 1.10, CI 0.97-1.19 6-10 drinks per day RR 1.98, CI 1.75-2.24 > 10 drinks per day RR 20.13, CI 16.4-25.3 Based on these results, which of the following statements is most accurate? A) Any alcohol consumption shows a statistically significant association with cirrhosis of the liver B) Consuming 1-2 drinks per day causes 5% more cases of liver cirrhosis when compared to individuals who do not consume alcohol C) Consuming 3-5 drinks per day causes 110% more cases of liver cirrhosis when compared to individuals who do not consume alcohol D) Alcohol consumption does not have a statistically significant association with cirrhosis of the liver E) Only alcohol consumption of > 5 drinks per day shows a statistically significant association with cirrhosis of the liver Correct answer E The number used in these questions are just for illustrative purposes and don't reflect real research results
I like your video (so far), but in the slide @ around 2:51, your interpretation of the CI is not quite correct. The confidence interval is NOT the probability that the true value of the population mean falls within the interval. Rather, if you repeated the experiment N times (constructing N confidence intervals), then you would expect that 95% of the confidence intervals (if you were constructing 95% CIs) would contain the population mean. In other words, the confidence level is not so much a statement about any particular interval, but pertains to what would happen if a very large number of like intervals were to be constructed. (Source: Probability and Statistics for Engineering and the Sciences, Jay Devore, Brooks Cole Publishing Company, Thomson Learning, 2000).
Thank you so much for you help. I am using your video's for step 3 as now step 3 has 12 out 42 questions per block on biostate. If you like, you can probably make addition to your current videos to cover additional topics of step 3 biostate(although none actually knows which topics they added as its still new). Overall, man you are getting lots of positive karma out of this!
Milind Joshi Thanks for the comment and kind words! I didn't realize how important Biostats was for Step 3 (I won't even take Step 2 until this year) so I'll definitely look into that in the future and see if I can potentially make videos more appropriate for that setting.
yah, if you want, you can make more extensive videos and sell them.... (since you mentioned in one of comment that you took year off to study public health, your knowledge must be quite in depth in biostate). Especially those dreadful drug adds!
raj mahi We have all been there at times. Especially with Biostats. It is so different than the rest of the stuff we have to learn, many of us don't have a strong background in this type of math and its so hard to study it for long periods because IT CAN BE SO BORING LOL. But just keep at it and I'm sure you will be cruising through this material in no time. Good luck and thanks for the comment!
@Stomp on Step 1 In correlation work, when the sample size decreases, does the size of the correlation that is needed to reach statistical significance; increase, decrease or stay the same?
So, just to make sure, my 95% CI that is 1.0659 to 3.5784 does not cross 1 and therefore shows a statistically significant association, right? (Also, P = 0.0303). Thanks! Results: Odds ratio: 1.9530 95 % CI: 1.0659 to 3.5784 z statistic: 2.167 Significance level P = 0.0303
Why keep people repeating this UTTERLY INCORRECT interpretation of the frequentist CI? From the guy who basically invented it: "It will be noticed that in the above description the probability statements refer to the problems of estimation with which the statistician will be concerned in the future. [...] Consider now the case when a sample, E', is already drawn and the calculations have given, say, [lower bound] (E') = 1 and [upper bound] (E') = 2. Can we say that in this particular case the probability of the true value of [the population parameter] falling between 1 and 2 is equal to [alpha]? The answer is obviously in the negative. The [population] parameter is an unknown constant and no probability statement concerning its value may be made [...]" (Neyman 1937, 349) This clearly contradicts your incorrect and dangerous interpretation of frequentist CIs! For what you are trying to do with CIs, you need Bayesian inference.
+korpustutorials thanks for the feedback and help correcting my error. My plan is to eventually replace this video with a new more accurate one, but until I have time to remake this video from scratch I have added a correction to the video description and a pop up note in the video directing people to the mistake.
That is something I would like to potentially investigate in the future, but unfortunately that isn't possible currently. Between my MS3 rotations and making new videos I don't have enough time to dedicate to tutoring. I have also never done any tutoring for any subject so I'm not sure how good I would be at doing it LOL. There are some other services online that offer this though so you should check out a site like this www.medschooltutors.com/usmle-exam-preparation
latesq1 The CI doesn't apply to how accurately your data represents the study population. In either case you have a very accurate measures of what is happening within the study. For example, if you are measuring BP, your BP cuffs are working well and calculating things like averages is simple and straight forward. The numbers you get represents the study population very well. Where the confidence intervals come in is estimating whether the numbers you got for the study population represent the population overall. In other words how likely the "truth in the study population" is to represent the "truth in the overall population". A 99% CI means that random chance is much less likely to have caused the results. therefore you can be more confident that what you are seeing in the study population also applies to the population overall
If you liked this video and want me to make more please let me know by commenting, liking this video or by subscribing to my RUclips channel. If you have a question, please don’t hesitate to ask and I’ll try to answer it ASAP.
Shooti Tamboo Thanks for the comment. I believe everything in the video is accurate, but there could be a typo I didn't catch. Is there something in specific you are wondering about? I'll do my best to answer I'm just not sure which part you have a question about
Stomp On Step 1 its very nice.....thanks very much......i have a question i am having the CI 95% and population of 4,081,243 what can be the sample size? in addition i have a resource to collect data from 2,000 participants what do you say?
Tewodros MUlatu Thanks so much for the comment. My answer to the question would be that we don't have enough information to figure it out. The Confidence Interval takes into account the "effect size" (how different the 2 groups are). When whatever is being measured is drastically different between groups the confidence interval shrinks because you are more confident the 2 groups are not the same. Since we only have the sample size (n) and the CI without any data on the effect size we can't come to a conclusion. Does that make sense?
Literally helping me with my homework right now because I missed class and needed to review this from YEARS ago! Thanks man!
+Shadow Roth I hope you Ace the assignment!
Its so cool to find people that try to help without any money reward. Thanks for the video. I like how you really go to the point. I wanted to ask you what would be an example of a question on STEP1 for this topic?? Thanks in advance
Omar Ciano Thank you so much for the kind words! I really appreciate the encouragement. Here are some CI example questions I wrote
An investigator is researching the link between obesity and
myocardial infarction. A retrospective case-control study is designed. 10,000
patients who have been diagnosed with myocardial infarction within the last
year have been identified and 10,000 similarly matched controls with no
diagnosed myocardial infarctions are selected as controls. Retrospective data
about the participants blood pressure over the last 5 years is then collected.
The odds ratio for hypertension among the myocardial infarction group was 2.6
with a 95% confidence interval of 1.5-3.3. Based on these results, which of the
following is most correct?
A. Hypertension causes myocardial infarctions
B. Myocardial infarction causes hypertension
C. Myocardial Infarction does not have statistically significant association with hypertension
D. Myocardial Infarction has a statisticallysignificant association with hypertension \
E. Hypertension has no correlation with Myocardial Infarction
Correct answer is D
A researcher is investigating the link between alcohol
consumption and liver cirrhosis. A prospective cohort study is designed with
100,000 participants. The study population is divided into groups based on
their self-reported daily consumption of alcohol. The participants are
monitored for 20 years for the incidence of cirrhosis of the liver. The results
are reported below.
GroupNo alcohol consumption, RR 1.00, CI ---
1-2 drinks per day RR 1.05, CI 0.93-1.14
3-5 drinks per day RR 1.10, CI 0.97-1.19
6-10 drinks per day RR 1.98, CI 1.75-2.24
> 10 drinks per day RR 20.13, CI 16.4-25.3
Based on these results, which of the following statements is
most accurate?
A) Any alcohol consumption shows a statistically significant association with cirrhosis of the
liver
B) Consuming 1-2 drinks per day causes 5% more cases of liver cirrhosis when compared to
individuals who do not consume alcohol
C) Consuming 3-5 drinks per day causes 110% more cases of liver cirrhosis when compared to
individuals who do not consume alcohol
D) Alcohol consumption does not have a statistically significant association with
cirrhosis of the liver
E) Only alcohol consumption of > 5 drinks per day shows a statistically significant
association with cirrhosis of the liver
Correct answer E
The number used in these questions are just for illustrative purposes and don't reflect real research results
May I suggest you take the video down and upload the corrected version. If someone misses the warning sign they potebtially build a wrong concept.
I like your video (so far), but in the slide @ around 2:51, your interpretation of the CI is not quite correct. The confidence interval is NOT the probability that the true value of the population mean falls within the interval. Rather, if you repeated the experiment N times (constructing N confidence intervals), then you would expect that 95% of the confidence intervals (if you were constructing 95% CIs) would contain the population mean. In other words, the confidence level is not so much a statement about any particular interval, but pertains to what would happen if a very large number of like intervals were to be constructed. (Source: Probability and Statistics for Engineering and the Sciences, Jay Devore, Brooks Cole Publishing Company, Thomson Learning, 2000).
+Dan Hillman You are right. Unfortunately, this video and many other sources (online and offline) keep repeating this mistake.
+Dan Hillman . Thank you. And you are correct. As Kostas mentioned it is a very common mistake.
thats true. what my professor was stressing in lecture today
Thank you so much for you help.
I am using your video's for step 3 as now step 3 has 12 out 42 questions per block on biostate. If you like, you can probably make addition to your current videos to cover additional topics of step 3 biostate(although none actually knows which topics they added as its still new). Overall, man you are getting lots of positive karma out of this!
Milind Joshi Thanks for the comment and kind words! I didn't realize how important Biostats was for Step 3 (I won't even take Step 2 until this year) so I'll definitely look into that in the future and see if I can potentially make videos more appropriate for that setting.
yah, if you want, you can make more extensive videos and sell them.... (since you mentioned in one of comment that you took year off to study public health, your knowledge must be quite in depth in biostate). Especially those dreadful drug adds!
Thank you so much...a struggling nursing student
raj mahi We have all been there at times. Especially with Biostats. It is so different than the rest of the stuff we have to learn, many of us don't have a strong background in this type of math and its so hard to study it for long periods because IT CAN BE SO BORING LOL. But just keep at it and I'm sure you will be cruising through this material in no time. Good luck and thanks for the comment!
Finally I found a good video to understand CI. Thanks a lot.
Ashwiinii Nandesshwar thanks for the comment!
Thank you for the simplicity.
Juliet Brown Thanks for the comment!
I need to know please how to report the 95%CI interval in a research article. Thank you
Awesome video! Thank you
Eliane Coffler Thanks for commenting!
Brilliant dude.keep up this good work
+Kershal Ramchurren Thanks :)
Thank you so much. finally understood the confidence intervals..
Thanks for the comment!
@Stomp on Step 1 In correlation work, when the sample size decreases, does the size of the correlation that is needed to reach statistical significance; increase, decrease or stay the same?
Finally.. now I know what my professor has been barking about all semester... #ThankYou
RyanByAir Thanks!
Thanks so much, well presented.
thanks you so much for these videos. they are so helpful..
Gajendra Maharjan I'm glad I could help. Thanks for the feedback
very well done congrats!
MsAmour2009 Thanks for watching!
This should be like a 7 not a 4. It always shows up.
loved it! thanks.
So, just to make sure, my 95% CI that is 1.0659 to 3.5784 does not cross 1 and therefore shows a statistically significant association, right? (Also, P = 0.0303). Thanks!
Results:
Odds ratio: 1.9530
95 % CI: 1.0659 to 3.5784
z statistic: 2.167
Significance level P = 0.0303
Yep :)
Loved this video.
What about the magnitude,Validity, and precision using the last example?
What about a hazard ratio? if it crosses one does that mean it is not significant?
Well explained sir, thanks.
+tamara mkandawire thanks for the comment Tamara!
thanks u look like expert.
Why keep people repeating this UTTERLY INCORRECT interpretation of the frequentist CI? From the guy who basically invented it: "It will be noticed that in the above description the probability statements refer to the problems of estimation with which the statistician will be concerned in the future. [...] Consider now the case when a sample, E', is already drawn and the calculations have given, say, [lower bound] (E') = 1 and [upper bound] (E') = 2. Can we say that in this particular case the probability of the true value of [the population parameter] falling between 1 and 2 is equal to [alpha]? The answer is obviously in the negative. The [population] parameter is an unknown constant and no probability statement concerning its value may be made [...]" (Neyman 1937, 349)
This clearly contradicts your incorrect and dangerous interpretation of frequentist CIs! For what you are trying to do with CIs, you need Bayesian inference.
+korpustutorials thanks for the feedback and help correcting my error. My plan is to eventually replace this video with a new more accurate one, but until I have time to remake this video from scratch I have added a correction to the video description and a pop up note in the video directing people to the mistake.
thanks for this!
Thanks for the feedback!
nice...very helpful.
Thank you!!
i love the video.
so if a confidence interval for a relative risk captures 1, then the result is NOT statistically significant?
+David Patterson correct
"if you find any mistakes or typos in my videos please post a comment at the bottom."***
Thanks for d video..:-)
alex thanks for watching and please tell your friends!
Thannnk uuuu sooo much :*
Divya Lovesyou thanks for the comment. Please tell your friends!
could you please do a video on cardiac cycle ? thanks for all biostats vids
DrJotsna Naidu I'm working on the psych section now, but I plan on doing cardio soon so it shouldn't be too long
do you give online tutoring where me and my friends could pay n get lectures and learn ?
That is something I would like to potentially investigate in the future, but unfortunately that isn't possible currently. Between my MS3 rotations and making new videos I don't have enough time to dedicate to tutoring. I have also never done any tutoring for any subject so I'm not sure how good I would be at doing it LOL. There are some other services online that offer this though so you should check out a site like this www.medschooltutors.com/usmle-exam-preparation
But if u have any specific questions just post a comment under the appropriate video and Ill do my best to answer
Thank you sir
Hi. May I please get permission to download the basic videos which talk about P-value and how to interpret it? Thanks
Is a 99% confidence level a more rigorous estimate of where a parameter is, than a 95%confidence level? Explain
latesq1 The CI doesn't apply to how accurately your data represents the study population. In either case you have a very accurate measures of what is happening within the study. For example, if you are measuring BP, your BP cuffs are working well and calculating things like averages is simple and straight forward. The numbers you get represents the study population very well. Where the confidence intervals come in is estimating whether the numbers you got for the study population represent the population overall. In other words how likely the "truth in the study population" is to represent the "truth in the overall population". A 99% CI means that random chance is much less likely to have caused the results. therefore you can be more confident that what you are seeing in the study population also applies to the population overall
Thanks
you made a few typos around 7:07.
Haha
If you liked this video and want me to make more please let me know by commenting, liking this video or by subscribing to my RUclips channel. If you have a question, please don’t hesitate to ask and I’ll try to answer it ASAP.
Very nicely explained. I would like to be sure if this is true for the 95% CI for the odds ratio. Kindly reply.
Thanks
Shooti Tamboo Thanks for the comment. I believe everything in the video is accurate, but there could be a typo I didn't catch. Is there something in specific you are wondering about? I'll do my best to answer I'm just not sure which part you have a question about
Stomp On Step 1 its very nice.....thanks very much......i have a question i am having the CI 95% and population of 4,081,243 what can be the sample size? in addition i have a resource to collect data from 2,000 participants what do you say?
Tewodros MUlatu Thanks so much for the comment. My answer to the question would be that we don't have enough information to figure it out. The Confidence Interval takes into account the "effect size" (how different the 2 groups are). When whatever is being measured is drastically different between groups the confidence interval shrinks because you are more confident the 2 groups are not the same. Since we only have the sample size (n) and the CI without any data on the effect size we can't come to a conclusion. Does that make sense?