YouTube Instagram TikTok How it works FAQ

The fastest way to turn talking videos into clean, searchable text. Free, private and no account required.

About

Built by Rishav Raj with one goal: free transcripts for everyone.

rishav@yttotranscript.com

Transcribe

YouTube to text
YouTube with timestamps
Instagram Reels
TikTok videos

Learn

How it works
FAQ
YouTube transcripts guide

More

Chrome extension
Contact
Privacy policy
Terms of service

© 2026 YTtoTranscript. Made for creators & researchers.

Not affiliated with YouTube, Instagram or TikTok. Respect creators’ copyright.

Sample Size and Power: Fundamentals Part 2 Part 5 — Transcript · YTtoTranscript

Home/Transcripts/YouTube

YouTube Transcript

Sample Size and Power: Fundamentals Part 2 Part 5

NIH VideoCast · 2,365 words · 12 min read · EN

Watch on YouTube

Share X LinkedIn WhatsApp

Below is the complete, readable transcript of Sample Size and Power: Fundamentals Part 2 Part 5 by NIH VideoCast on YouTube. Read the full text, copy any part you need, or generate a transcript for any video with our free tool.

so what are some changes if you have unequal numbers in each group the ratio of cases to controls you want to use this if you have say a certain lambda set of patients randomized to the treatment arm for every patient randomized the placebo arm you can also use this in a case control

setting so the idea here is that you're going to have different variances you're going to kind of use this ratio calculate the sample size for one group and then use that to calculate sample size for the other group the shortcut way of doing this let's say i'm going to use equal variance sample size

formula so i assume both groups have the same variance i want to look at the total sample size calculate that then increase it by this factor k so let's say i'm doing two to one randomization i calculate the total sample size if i had two equally sized study arms it's 26 so it's going to have 13 per arm

i say okay 26 times 2 plus 1 because that's my ratio squared divided by 4 times the 2 and i find out that i need 30 people in my study instead so i would have 20 in one group 10 in the other why might i do this well sometimes i have a device let's say i have a new

artificial knee and i want to know how many knees i need to make and i want to do an unequal randomization this might be a case where i use that sometimes i have a fixed number of cases i've already made the knees i only have so many of them to use so if my sample size calculation said i

needed 13 per arm but i only have 11 of the knees i want the same precision then i'm going to calculate how many people will be in my control arm back calculate that using this formula when you're doing survival analysis and other types of analyses you may have a cohort of say those exposed and unexposed people that

Transcribe another video

Paste any YouTube, Instagram or TikTok link to get a free transcript.

Free · No sign-up · Unlimited

Add to Chrome — transcribe in one click

we talked about you have this relative risk that you're interested in i know the prevalence and the unexposed population i need to figure out the number of events so the risk of an event in the exposed group divided by the risk of the event and the unexposed group and back calculate the number of events i think

i'm going to be seeing so i have the number of events in my unexposed group calculate the number of events in the exposed group then calculate how many people i need to see the events remember stability issues in logistic regression there's this big discussion and a whole bunch of simulations were done that you need at

least 10 subjects for every variable investigated you look at principle components analysis and a lot of other methods that get used you see everything from ten to a hundred it's all sorts of crazy some people don't say you have to use balanced designs other people say you have to use them i'm just going to say if you have

balance designs you have the same number of people in each study arm it's easier to figure out your power in your sample size it is easier to handle but you know i have a job to work on complicated designs so you can do simulations you can do pretty much anything you want as long as you can do the work to make

sure it's going to be rigorous and valid so what about multiple comparisons if you have four groups and you want to do all two-way comparisons of means you have six different tests but sometimes you say well i have four groups i have one placebo arm at three different doses and i actually just want

to compare each dose of the placebo i'm not going to compare the doses to each other so you have to figure out exactly how many tests you really are going to run the simple way people do it is this bonferroni test right they divide their alpha by the number of tests it's common

there's a long literature but it's super conservative and in some of your high throughput laboratory tests it can be worked out at corn i think it was work this out and maybe lisa or somebody else in that branch it said like there were cases that it was physically impossible for them to have any statistically

significant results after they accounted for multiple comparisons that's not an experiment worth running this comes up a lot in these like microarray proteomics other omics types of experiments many times people say just set your alpha and your beta stricter i tend to agree with this you know just set a really strict type

one error level that you're going to adhere to and test everything against that if that's what you want to think about for false pauses false negatives there's also like family wise error rates and all these other things but i think just set strict up front and plan for that so what are some of these rejected

statements again this um st george's hospital you know i'm looking at this link i think this is the correct link but i don't remember if i double checked it this morning because i keep moving some stuff around anyway this entire you can put the name into your favorite search engine and find it

st george's hospital medical school has one of the best statistics guides for research grant applicants i've ever seen like many times i was like can i just copy and reference it be done i think it's very applicable across any country that you go to so one of my favorite parts because i see it way too often it's

people who go me to me too it's like no no no you must justify your sample size in every single application so this is an example an exact quote a previous study in this area recruited 150 subjects and found highly significant results p-value equals .014 and therefore similar sample size should be sufficient here

well they could have been lucky could have been random sampling variation could have been a lot of other things you need to give an estimate of what you're expecting for the variance for the difference what your type 1 error is going to be etc you have to give all those little elements it's good to know that some

study who had similar elements found something significant but honestly if a lot of studies have already found this why are you doing your study no prior information now this is always trouble because sometimes you do have no prior information and what you're trying to do is run the little baby first pilot but the quote in the application

for a substantial sample size application i should say sample sizes aren't provided because there's no prior information on which to base them all right we have to know how many people how many animals whatever you are enrolling in this study find something that's published previously and try to extrapolate from that conduct a small pre-study but if

this is your application to conduct the small pre-study then you probably in fact don't need sample size calculations don't tell me the sample sizes aren't provided tell me like this is where you are and that you're going to use this amount of information to then calculate appropriate sample sizes but this is what you're gathering in order to do it

but if i can easily find a bunch of information that could have informed your sample sizes i'm probably not going to have a very happy review for you variance no prior information on standard deviations well again this might be where you talk about detecting the size of difference in terms of the standard deviations this is

where those effect sizes that cohen and others and they're different effect size values for different regression models but this is where that can be useful again it's your starting step but it's not going to be what you use for a pivotal or final study now the number of available patients wendy weber and others are going to talk a

little bit more about this but you have to make sure you have patience to be in the study few things you will not get all patients to be in your study number one in fact you're lucky if like five percent agree to be in your study in most cases number two feasible is important it does not tell

me about your power so this is again a direct quote that they got at st george's the clinic sees around 50 patients a year of whom 10 may refuse to take part in the study all right i don't think that's realistic that only 10 percent will refuse but we keep moving on therefore over the two years of the

study the sample size would be 90 patients let's say a miracle occurs and they get their 90 patients i have no idea what the power is for this study i know nothing about the differences that they want to detect i don't even know what they want to measure i don't know about the variance i know nothing

it's okay to give this information and then say and with 90 patients our power will be x based on the following assumptions that would have been okay but that sentence never appeared in the application so you do have to balance feasibility and power but sample size isn't decided just on available patient numbers

so what are some of our resources and conclusions what impacts your sample size the differences the standard deviations or the variance your type one error that you're willing to tolerate your type two error that you're willing to tolerate number of arms or samples one or two-sided tests are you randomizing if you are what type of randomization are

you using what are the analysis plans if you don't have an estimate of the variance make sample size or power tables make graphs you should in fact always even when you think you have good estimates they may not be that good you're always going to want to make tables and graphs show a wide variety of possible standard

deviations of possible differences and protect yourself with a high sample size if it's at all possible so when i was reviewing grants for the nih the top 10 statistics questions i had to ask back to investigators the exact mechanism they were going to use to randomize their patients don't tell me you're just going to

randomize them dr joaquin went through a whole bunch of different ways you can randomize people tell me more information tell me when etcetera they would say they're going to stratify i'd be like why are you going to stratify what are you going to stratify on emas the european medicine agency actually provides different information

on something called dynamic allocation so if you're going to use different types of allocation schemes they may or may not be acceptable to different regulatory bodies worldwide or they may be acceptable in pediatrics but not acceptable in adults acceptable on oncology not acceptable in cardiovascular disease blinded or masked personnel who is doing your endpoint assessment

who is masked to what are they masked are you compromising your blinded study with something else that you are doing many times i had to ask people to list every single hypothesis the hypothesis test the specific analysis and the specific sample size or power justification for those hypothesis tests that should be very clear

in your grant application somebody shouldn't have to ask you for this should also be clear in the statistical analysis plans that go with it and it needs to match everything else in your protocols i would need to ask how or if they were adjusting for multiple comparisons some fields get very very picky about multiple

comparisons others say well if you're honest and you report everything you can say i'm not adjusting for multiple comparisons typically most of your regulators and pivotal trials will make you do what's called a statistical hierarchy or some way to control the type 1 error so it may or may not be technically adjusting for multiple comparisons but

you are kind of doing a conservation they call it of the type one error i would ask if they're testing for effect modification are you doing any interim analyses and if so what exactly are you analyzing when what's the error spending model and what are the stopping rules those error spending models are again

to conserve your type 1 error and i would ask if this was accounted for in the sample size calculations why all those ways that you do interim analyses impact your sample size calculation what is your expected dropout you know they would tell me about their sample size but not the expected dropout and did they plus up

in order to account for this how do you handle the dropouts and missing data in your analyses again those methods may also impact your sample size and your power and then i always had to ask about repeated measures in longitudinal data almost always are you using a linear mix model are you trying to use repeated measures

anova repeated measures anova is an awful method it's very antiquated it has a lot of assumptions the newer methods have a lot fewer assumptions they're a lot more robust maybe they're going to use generalized estimating equations but again what is your actual method how are you collecting the data what's the study design

what are the analyses in order to actually test that data all of that goes into figuring out your sample size and the analyses and conclusions that you can draw