This is hands down the clearest and most intuitive explanation that I've ever encountered, whether on YT or in print. It has always surprised me how folks who normally do a great job at explaining statistics in plain language, when it comes to this particular question, either just evade it or resort to mathematical formulas that I bet look obvious and crystal clear to the inititiated but won't help those who only begin to learn statistics and are simply curious about the details of a notion as fundamental as SD. You, sir, did a great job here and I will be forever grateful to you for this!
Wish I had you as my statistics teacher in school. Like so many others here have commented, you provided a very intuitive answer to something I have been wondering abt for a long long time. Thx Mr. Nystrom and please keep more maths videos coming!👍👍
I've watched a few different teachers on RUclips but I leave still feeling confused, your really good at explaining! Thank you! Keep doing what your doing!
Now I find hope again and I think I'll pass my stats course this year for sure. Thank you. 👏🏻👏🏻👏🏻 Universities should not be accepting any less professors than you, for real:))
I got this video recommended on Google and thought it would be absolutely unjust if I didn't hit the like button! Because this video is just sooo perfect! On point!! 👍👍
Another way to look at is as follows. Does calculating the standard deviation of a sample with n=1 make sense? Also when one is calculating s, one uses xbar. So if you have n values, and you know (n-1) values and the average of n values, your nth value is not independent anymore.
Very good explanation. I think it is more of an observation that an explanation. I would suggest you to include the following at the end "We are using sample mean instead of the actual mean for the sample standard deviation calculation. That is, the term Xi minus the average is not there. That specific Xi minus average is 0. That term is always missing from the sample. It is because we do not have that term, we divided by N -1 instead." It is very difficult to express. You have done a very good job. ☺️
Thank you sir, that made lots of sense and finally explains it for me. I'm not even studying stats or maths but just enjoy the topic and your videos are very clear and fun to watch. Keep it up!
Nice informative video. I request you to make one similar video on the importance of the use of the degree of freedom (n-1) intend of n for ANOVA calculation.
OUTSTANDING, a new video! Starved and in withdrawal, started calculating the probability of seeing a new video again: p=0.15 it wasn't looking good :-)
I don't get what you're saying around 5:56. If there is less variability with a smaller sample, why do your sample mean drawings have a more spread out distribution with the n=10 than the n=50. Aren't you saying the opposite here?
That was a nice video, but I gotta say that it was those 'couple of other mathematical reasons' that I am actually after. I have often heard the n-1 figure being referred to as the degrees of freedom, and what I understand from your explanation is that this is misleading - degrees of freedom have nothing to do with it and it is just coincidence that the numbers are the same. Or is the degrees of freedom thing one of those 'other mathematical reasons'...
The degree of freedom is basically the number of outcomes subtracted by 1, or n-1. So, it does have to do with the equation. However, I don't think he was trying to make the viewers believe otherwise. Instead, I think he was trying to make it clear for those unfamiliar with the degrees of freedom, with the use of fractions, so as to not get distracted by more new terms and definitions.
Please make a video on hypothesis testing for two sets of data (hopefully I'm saying that right). I have a test coming up and your videos have really helped me! :D
Hello guys ! I've just watched the video and its amazing thank you so much for that but I wonder why do we divide by (n-1) not n-2 or n-3 etc. ? I'd be so appreciate if someone response my question. Best and take care guys .
This explanation for n-1 makes sense to me, and the more abstract "degrees of knowledge" explanation kind of does too. But they seem like completely different explanations. How can I synthesize these two explanations into a single understanding of n-1?
Why can't we do "n-2" or "n-3" and so on to be on the safer side to estimate Std. Dev of population from sample? Making the number very small (n-2 or n-3) in the denominator will expand the Std. Dev more. If the sample size is too small (let's say 10), we can make n-3 (just saying) and n-1 for a little larger sample (let's say 100). Why is it not dynamic and just n-1 for all different sample sizes, given the fact that Std. Dev will also vary when the sample size changes?
Nur Syahirah Variance: the variability in Y explained by X. example: the variability in Bananas I eat in a day (Y)is explained 72% of the time by working out (X). Working out is 72% of the reason on why I eat Y amount of bananas . Standard Deviation: typical distance from the mean. Example: on average(mean) I eat 3 bananas a day. But sometimes I eat 5 or 1 (standard deviation of 2). Standard Deviation SQUARED is Variance. Standard Deviation finds typical distance from the mean Variance measures the correlation strength between X and Y Hope that helped
So if in a TP of chemistry I measure the PH of my solution 6 times and I do a mean, must I use the Bessel correction to calculate the standard deviation? I mean, have I a sample in this case, or a population?
Hey bro is it a universal fact that sample means will always be below the true mean? Hence, the need to divide by n-1? Why can’t the sample mean be above the true mean so instead we would need to n + 1?
The actual reason is that for sample standard deviation we use the mean of the sample rather than the mean of the population. There is no fiddling at all in the formula as wrongly suggested.
This has absolutely nothing to do with this video. And god I hope this works. Anyway! We both attended a concert 6-12-17 in KC. You took what looked like a GREAT selfie of you and your colleagues with and my daughter and I am dying for a copy of it. And hopefully a reminder of you companion names. Y'all were AWESOME! 🤞This works ☺
It'd be great to talk about this in terms of degrees of freedom. I feel like there has to be a better, mathematical explanation than: "meh, we kinda just messed with it a little"
why does the adjustment eliminate the bias, rather than just reduce it? ... Never mind; I think I've figured it out. As the sample size increases, the difference between the estimated variance you get when dividing by n vs. n-1 approaches zero (the definition of 'unbiased', which is a large sample property). Shout out to StatisticsisFun. Thanks, gents!
Makes no sense. Why do u assume the SD of samples will be on left of the population SD. Also, how did u calculate the SD of the samples? Did u use n_1? Then no wonder ur SD of samples will be away from g he SD of population
Holy shit this guys enthusiasm is infectious. The world needs more teachers like this guy.
the passion that you have, makes me want to learn
This is the best video that was able to explain why n-1 in sample SD makes sense by far
This is hands down the clearest and most intuitive explanation that I've ever encountered, whether on YT or in print. It has always surprised me how folks who normally do a great job at explaining statistics in plain language, when it comes to this particular question, either just evade it or resort to mathematical formulas that I bet look obvious and crystal clear to the inititiated but won't help those who only begin to learn statistics and are simply curious about the details of a notion as fundamental as SD. You, sir, did a great job here and I will be forever grateful to you for this!
The clearest way to explain this! Well done Sir!
Wish I had you as my statistics teacher in school. Like so many others here have commented, you provided a very intuitive answer to something I have been wondering abt for a long long time. Thx Mr. Nystrom and please keep more maths videos coming!👍👍
This is such a good and logical explanation I have ever heard about n-1 denominator. I love the way you teach.
Mate. You are my hero. If mathematics was taught mainstream the way you teach it, we would have colonised the galaxy by now.
I'm not even taking the piss.
I'd love to see an in-depth video on degrees of freedom. Happy to have you back!
This was awesome. Started an intro course for pob/stats two weeks ago and this was exactly what I needed. Love the energy!
I've watched a few different teachers on RUclips but I leave still feeling confused, your really good at explaining! Thank you! Keep doing what your doing!
Dude that saved my grades way back in the day is back!! Lots of love man!
You are freaking amazing . For several years I had this question and now i am so so so happy . You are freaking genius dude.
Yes sir, that made perfect sense! Thank you for your continuous effort to spread good education.
Now I find hope again and I think I'll pass my stats course this year for sure. Thank you. 👏🏻👏🏻👏🏻 Universities should not be accepting any less professors than you, for real:))
Perfect , brilliant .This helped to find why n-1 not n-2 with your example of proportion of 5/5 ,10/10 and so on.
Thanks a lot.
One of the best explanations out there. Crystal clear. Good job buddy.
I got this video recommended on Google and thought it would be absolutely unjust if I didn't hit the like button! Because this video is just sooo perfect! On point!! 👍👍
After seeing around 5-6 videos, finally I understood here. Thank you.
He's back! The legend returns!
Another way to look at is as follows.
Does calculating the standard deviation of a sample with n=1 make sense?
Also when one is calculating s, one uses xbar. So if you have n values, and you know (n-1) values and the average of n values, your nth value is not independent anymore.
Damn you're the best! That fraction example to explain how to correct for the bias is awesome.
You'r videos helped me pass my Stat class, i am grateful.
The best teacher at BHS hands down
I'll be 59 this year (2017)... and I'm still learning through your videos. Than you sir.
You make stats understandable and fun. Thank you!
Very good explanation. I think it is more of an observation that an explanation. I would suggest you to include the following at the end
"We are using sample mean instead of the actual mean for the sample standard deviation calculation. That is, the term Xi minus the average is not there. That specific Xi minus average is 0. That term is always missing from the sample. It is because we do not have that term, we divided by N -1 instead."
It is very difficult to express. You have done a very good job. ☺️
Thank you sir, that made lots of sense and finally explains it for me. I'm not even studying stats or maths but just enjoy the topic and your videos are very clear and fun to watch. Keep it up!
Thanks for your explanation! Everything becomes more clearer
The best 12 minutes of my life so far ❤️❤️ thanks a ton for the amazing explanation
Finallly i understood after watching so many videos !!!!!!!!! THANKSSS
Nice informative video. I request you to make one similar video on the importance of the use of the degree of freedom (n-1) intend of n for ANOVA calculation.
OUTSTANDING, a new video! Starved and in withdrawal, started calculating the probability of seeing a new video again: p=0.15 it wasn't looking good :-)
Its easy to understand your explanation. Good work
As in your other vids you explain it so one can really get the idea behind the thing. Awesome teacher! and fun to watch :)
I don't get what you're saying around 5:56. If there is less variability with a smaller sample, why do your sample mean drawings have a more spread out distribution with the n=10 than the n=50. Aren't you saying the opposite here?
MrNystrom your lessons are amazing, explanations are intuitive. Keep up the good work. Also looking forward to your new videos.
Excellent. I wish I saw this 35 years ago.
Thank you, no one ever explained this to me.
I like the way you explain with pretty cool nature cracking jokes. You made stat not only easy but also funny. Thanks a lot
That was a nice video, but I gotta say that it was those 'couple of other mathematical reasons' that I am actually after. I have often heard the n-1 figure being referred to as the degrees of freedom, and what I understand from your explanation is that this is misleading - degrees of freedom have nothing to do with it and it is just coincidence that the numbers are the same. Or is the degrees of freedom thing one of those 'other mathematical reasons'...
The degree of freedom is basically the number of outcomes subtracted by 1, or n-1. So, it does have to do with the equation. However, I don't think he was trying to make the viewers believe otherwise. Instead, I think he was trying to make it clear for those unfamiliar with the degrees of freedom, with the use of fractions, so as to not get distracted by more new terms and definitions.
So one is used when you have the full set of raw data, the other is when you are sampling less than a full set from the raw data?
The Tom Brady of stats
Shields21 G.O.A.T.
This is due to the df adjustment. It would be great if you explain the intuition behind this.
This is a totally excellent video!!!! I love you sooooo much dear mr Nystrom !
Great clarity in this presentation for standard deviation.
Please make a video on hypothesis testing for two sets of data (hopefully I'm saying that right). I have a test coming up and your videos have really helped me! :D
4:18 What a reaction. Just loved the video :-)
On spot.... keep it up💥💥💥💥💥💥
Hello guys ! I've just watched the video and its amazing thank you so much for that but I wonder why do we divide by (n-1) not n-2 or n-3 etc. ? I'd be so appreciate if someone response my question. Best and take care guys .
Thank you for your time and effort on these, I am using them!
This explanation for n-1 makes sense to me, and the more abstract "degrees of knowledge" explanation kind of does too. But they seem like completely different explanations. How can I synthesize these two explanations into a single understanding of n-1?
could you please also tell us what is the difference between mean deviation from mean and standard deviation thanks ya!!
Awesome explanation!
Brilliant explanation, thanks man!
Why can't we do "n-2" or "n-3" and so on to be on the safer side to estimate Std. Dev of population from sample? Making the number very small (n-2 or n-3) in the denominator will expand the Std. Dev more. If the sample size is too small (let's say 10), we can make n-3 (just saying) and n-1 for a little larger sample (let's say 100). Why is it not dynamic and just n-1 for all different sample sizes, given the fact that Std. Dev will also vary when the sample size changes?
what's the difference between variance and standard deviation then??
Nur Syahirah I also have the same question.
"Standard Deviation" is the square root of the average squared distance to the mean while "Variance" is the average squared distance
Nur Syahirah
Variance: the variability in Y explained by X. example: the variability in Bananas I eat in a day (Y)is explained 72% of the time by working out (X). Working out is 72% of the reason on why I eat Y amount of bananas .
Standard Deviation: typical distance from the mean. Example: on average(mean) I eat 3 bananas a day. But sometimes I eat 5 or 1 (standard deviation of 2).
Standard Deviation SQUARED is Variance.
Standard Deviation finds typical distance from the mean
Variance measures the correlation strength between X and Y
Hope that helped
what if we use n-2 or some other integer or decimal value more/less than 1. Mainly my question is why 1?
yeah same doubt... why not n-2,n-3 ??
This is such a good video, thanks man
So if in a TP of chemistry I measure the PH of my solution 6 times and I do a mean, must I use the Bessel correction to calculate the standard deviation? I mean, have I a sample in this case, or a population?
Fantastic explanation! I enjoyed it very much.
Hey bro is it a universal fact that sample means will always be below the true mean? Hence, the need to divide by n-1? Why can’t the sample mean be above the true mean so instead we would need to n + 1?
It doesnt matter if its above or below,, u square it anyway, its about the dispersion being less thas why its on the left.
The actual reason is that for sample standard deviation we use the mean of the sample rather than the mean of the population. There is no fiddling at all in the formula as wrongly suggested.
Thanks for your video! It help me a lot.
Really love this guy
you are really helpfull man. please keep doing continues videos for a levels
love ya
You are an angel. Thank you.
Beautiful explanation....Thank you so much
DUDE...NICE VIDEOS..LOVED THEM.. WHY U STOPPED MAKING THEM???
you're a life saver thank you so much man
This has absolutely nothing to do with this video. And god I hope this works.
Anyway! We both attended a concert 6-12-17 in KC. You took what looked like a GREAT selfie of you and your colleagues with and my daughter and I am dying for a copy of it.
And hopefully a reminder of you companion names. Y'all were AWESOME! 🤞This works ☺
Thanks so much for the easy explanation.
holy shit you're alive
amazing....simply amazing...thanku
It'd be great to talk about this in terms of degrees of freedom. I feel like there has to be a better, mathematical explanation than: "meh, we kinda just messed with it a little"
Genius! Thank you sir
why does the adjustment eliminate the bias, rather than just reduce it? ... Never mind; I think I've figured it out. As the sample size increases, the difference between the estimated variance you get when dividing by n vs. n-1 approaches zero (the definition of 'unbiased', which is a large sample property). Shout out to StatisticsisFun. Thanks, gents!
on a scale of 1 to 10 ... how stoned are you exactly? ... eyes wide open :-D -- kidding bro, great video.
very clear cut explanation
Could you perform p-value explanation?
Holly Harry ruclips.net/video/-MKT3yLDkqk/видео.html
Great video man! Thank you so much!
Thanks for the videos! Could you make one explaining Type of error I and II and power :)
This was great. Could you do one about Cohen's d?
great video my guy
Helped a lot, thanks!
why not 2? why not 3? why not 0.5?
thank you for saving my life
Sweet! Can you do data mining next? :-D
Thank you! That helps a lot
hi. are you and the beardless one the same person. my whole class is debating over this. we need answers.
Annie Reynolds i will answer that with a question.. if P(A and B) = P(A)*P(B), are A and B independent?
MrNystrom Is the answer yes you are the same person or yes you and this supposed brother of yours are independent of each other 🤔
Thank god for this Vid!!!
Thank you very much
Thank you so much! I can go to sleep now!
but why the hell do we have to use (n-1) and not (n-2) or (n-3) !!!!!!!!!!!!!
awesome!!
Dammmmmmmmmm....Super!!
I love your videos! Can you make a video of t-distribution?
Wow man🙌
Makes no sense. Why do u assume the SD of samples will be on left of the population SD. Also, how did u calculate the SD of the samples? Did u use n_1? Then no wonder ur SD of samples will be away from g he SD of population
Kowthar Hassan we know because sample spreads tend to be less than population spreads.
MrNystrom but we said that population spread is less hence the distribution looks narrower rather than wider? So what am I missing?
brilliant.
This is brilliant... answering a question I had never had properly explained for years. Cheerful examples are the best way :)