Standard error of the mean | Inferential statistics | Probability and Statistics | Khan Academy

Поділитися
Вставка
  • Опубліковано 21 гру 2024

КОМЕНТАРІ • 270

  • @shostawin
    @shostawin 11 років тому +256

    i've done a PhD in maths and I STILL come back to the Khan videos to refresh my basic stats. true story. what a dude.

    • @amarsirfmai
      @amarsirfmai 4 роки тому +4

      R u for real?

    • @ohmingfeng9351
      @ohmingfeng9351 4 роки тому +12

      Amarpratap singh when u have a phd your brain capacity fills up really fast and u tend to forget older stuff

    • @feurigerStern
      @feurigerStern 3 роки тому +5

      @@ohmingfeng9351 So true. I have had 3 statistics classes and now taking my 4th. My last class was 9 years ago. I am back here to review while I work on my graduate degree.

    • @CineriousJones
      @CineriousJones 3 роки тому +5

      oh thank god. I'm not alone. I've taken 4 advanced modeling and stat classes in grad school now and I still need to keep refreshing.

    • @paparokauli
      @paparokauli Рік тому

      Good for you doctor :D

  • @jimmyhaotran123
    @jimmyhaotran123 5 років тому +59

    Just in case people are having problem with some of the definitions which I have been looking for the past hour.
    sample mean == you take a sample of n data and finds the average.
    SAMPLING distribution of sample mean == basically do the sample of n data repeatedly many times, so you get many means, and use those means as your distribution, in another words, you get a normal distribution full of means, even the extreme numbers are one of the means.
    σ = population standard deviation.
    σ (with subs x bar) = standard deviation of the 'SAMPLING distribution' of the 'SAMPLE means'
    s = sample standard deviation.
    Hope it helps for those of you are still confusing with the naming conventions.

    • @jimmyhaotran123
      @jimmyhaotran123 5 років тому +4

      and of course, n is always referring to sample size (numbers of x), not the sampling size(numbers of x bar)

    • @s2productions242
      @s2productions242 3 роки тому +3

      He's the messiah

    • @eliizabeth7557
      @eliizabeth7557 3 роки тому +2

      @@s2productions242 mans the long lost saviour we been searching for

    • @sanchitakanta1018
      @sanchitakanta1018 3 роки тому

      @@eliizabeth7557 could you please say what will be the formula for standard error when we take just a single sample, hence just one mean.
      Will it be
      S.E = S.D of sample/sqrt(n)
      Like basically S.D of sample instead of population.

    • @kjy05
      @kjy05 9 місяців тому

      thank you so much God bless you

  • @semakapuszoglu7021
    @semakapuszoglu7021 6 років тому +13

    I have a midterm tomorrow ,started the day without knowing a thing and now I can even solve questions. I wish my lecturer were you, every person deserve quality education thanks Sal

  • @jp32123
    @jp32123 2 роки тому +3

    best voice, best personality, best teaching. I grew up watching u and here i am back again for my medical licensing exam (USMLE Step 2 CK) to study epidemiology with you. I love u so much. Thank u for everything!

  • @blackrobe2007
    @blackrobe2007 14 років тому +12

    I love how he explains everything using common sense! makes you able to visualize things easier therefore understanding things faster! Teachers nowadays just read equations off slides..they are useless might as well read equations from a textbook.

    • @kjy05
      @kjy05 9 місяців тому

      13-14 years later and I've gone through school for as many years as this comment has been up. In my second year of university right now. Totally true.

  • @k1mcheenoodle
    @k1mcheenoodle 4 роки тому +1

    "You know, sometimes this can get confusing because you are taking samples of averages
    based on samples. So when someone says sample size, you're like, is sample size the number of times i took averages or the number of things I'm taking averages of each time?
    ....
    Normally when they talk about sample size they're talking about n..."
    My goodness this was so clarifying My book doesn't really make this distinction clear or apparent, so it's always a guessing game to try and figure out what they mean by sample size, at least for me. But now it's clear. Thanks, Sal! Saving the day once again.

  • @leek.3671
    @leek.3671 7 років тому +193

    My brain is gonna explode

  • @anasahmad6749
    @anasahmad6749 Рік тому +3

    Needed to revise this while studying biostats in my MBBS....came back to the legend.... Jazakallah

  • @belgiandadlethe
    @belgiandadlethe 12 років тому +2

    I have a statistic exam this Friday. Studied Japanese and now communication, but statistics is lodged in my curriculum for some weird and torturing reason. I still think I'm doomed, but I'm less doomed thanks to you guys. Greets from Belgium

  • @eliyalee8551
    @eliyalee8551 8 років тому +8

    Just gotta say, I really enjoy the videos. Thank you Mr. Khan.

  • @oli.bradshaw
    @oli.bradshaw 8 років тому +82

    clear as mud

  • @teachonline
    @teachonline 13 років тому +4

    Your approach of emphasizing a firm grasp of the CONCEPT - which is helped tremendously by your illustrations, examples and "friendly" narrative - before going to complex mathematical formulas (proofs) is an excellent one!
    I look forward to seeing more. (Hope you get into ANOVA).
    Many thanks for making this help available.

  • @VickyBro
    @VickyBro 10 років тому +10

    OMG! I never knew the formula. But I guessed it correctly before he revealed it. Amazing. I wonder if it is his way of teaching that instills the concepts into our brain so quickly.

  • @steshanplaci6373
    @steshanplaci6373 4 роки тому +7

    "let me use a different colour for that"
    -khan academy (the greatest teacher ever)

  • @AvinashSingh-bk8kg
    @AvinashSingh-bk8kg 3 роки тому +2

    This guy is a champion in optimizing any topic 🙌
    ♥️Love from India🇮🇳

  • @laurenking8792
    @laurenking8792 4 роки тому +2

    This video and his explanation is so clear and straightforward! If you watch all the videos before this and understand all of those concepts, you should be able to understand this easily. Great job; thank you

  • @manhoing7422
    @manhoing7422 8 років тому +4

    So clear about it
    15 mins can explain everything about this topic
    the professor in the university has spent lectures to explain nth

  • @johnr.timmers2297
    @johnr.timmers2297 4 роки тому

    I swear my stats class would be so much better if the crappy professor just played up Khan Academy. Fantastic organization totally deserve every dollar they get

  • @vasilis1963
    @vasilis1963 4 роки тому +7

    I swear, in half of the videos regarding this issue, they say that σ reffers to the the sd of the sample and the other half say that σ reffers to the whole population. They mess so badly with what each parameter in the SE=σ/√n formula stands for.

  • @besimmons
    @besimmons 12 років тому +1

    Standard deviation of the mean is a parameter describing the distribution of all sample means of a a given sample size n from a population.
    standard deviation of the mean = (pop. stdev)/sqrt(n)
    Standard error of the mean is an approximation of this value calculated from a single sample.
    standard error of the mean = (sample stdev)/sqrt(n)

  • @janekou5214
    @janekou5214 11 років тому +1

    to compare how precise is your sample data compare to your population data. if the standard error is large, that means your sample data is not a good representative data for the population, vice versa, if its small, it means the sample data is representative! hope that helps :)

  • @dipikad9501
    @dipikad9501 7 років тому

    Thinking of standard error as a standard deviation of sampling distribution is so simplifying. Thank you for this.

  • @mckenziewitz3096
    @mckenziewitz3096 8 років тому +44

    finally talks about standard error of sample mean at 7:30ish

    • @ASiddig
      @ASiddig 6 років тому

      !!!!

    • @joshswain1055
      @joshswain1055 4 роки тому

      you just save me about 7 and half mins

  • @danman6612
    @danman6612 8 років тому +3

    When I think of sample size, it means the number of observations per sample, hence the lowercase n.
    For example, if one sample is 1, 2, 3, 5, and 9, those are 5 observations, so the sample size is 5. Another way of looking at a sample is by thinking of it as a data set, and each individual observation is a data item. If you call the entire thing a sample and everything in it a sample also, it gets really confusing especially when you're trying to get your mind around the concept of the sampling distribution of the sample mean.
    Anyways, your videos are the only reason that I'm passing my business stat course, thank you so much!

    • @jamesleem.d.7442
      @jamesleem.d.7442 6 років тому

      You wrote, "If you call the entire thing a sample and everything in it a sample also, it gets really confusing especially when you're trying to get your mind around the concept of the sampling distribution of the sample mean".
      I cannot agree more strongly. This is a VERY common and totally avoidable error made in almost all of the online presentations by statistical evangelists.

    • @sanchitakanta1018
      @sanchitakanta1018 3 роки тому

      @@jamesleem.d.7442 Can you please say what will be the formula of standard error if we are working with a single sample of maybe 30 observations.
      Will it be
      S.E= S.D of the sample of 30 observations/ sqrt(n=30)
      Basically if we are working with just one sample we won't have standard deviation of sampling distribution of sample means, right?

  • @vaishnavidang8864
    @vaishnavidang8864 5 років тому +2

    I really really appreciate the way he teaches... so nice and practical. I was never been able to get these concepts but now its crystal clear . can I please know the educator's name?

  • @maryisbest
    @maryisbest 12 років тому

    I have my exam tomm. and i felt this as the most reliable source...May God bless you :)

  • @p.z.8355
    @p.z.8355 7 років тому

    its easy :
    n:= number of samples to compute the mean
    sigma:= true variance of the original distribution.
    sigma_x:= variance of the means ( computed from the n samples ), the square root is called standard error
    basic message of the video := sigma_x = sigma/n . Variance of sample means can be approximated with the true variance and the number of samples we take.
    However, I think the formula is more usefull the other way round, since most time people don't know the true variance :
    sigma_x*n=sigma

  • @clairewang6054
    @clairewang6054 8 місяців тому

    It makes me feel so much better that people are coming back to review, I don't know, AP math?

  • @analuizanb
    @analuizanb 4 роки тому

    Thanks so much for taking the time to make clear what you're talking about. I've left two linkedin stats courses unfinished because both times it got to the point where I just couldn't keep all the Greek letters straight nor could I always tell whether the instructor was talking about a sample, the population or a sample distribution. Thanks!

  • @sherlockholmes882
    @sherlockholmes882 5 років тому +1

    I liked the video. Don't know why so many here in comment section found it confusing. If someone had followed the previous video, it's quite simple and beautifully explained.

  • @barbaric37
    @barbaric37 11 років тому

    When your airhead professor goes over an entire chapter in 20 minutes, writing some ambiguous notations on the board, expecting everyone to understand his chicken scratch.... you go to Khan Academy. Thank you for actually TEACHING!!!

  • @nickc3053
    @nickc3053 10 років тому +32

    This guy confuses me so much yet everyone tells me he's the best. Think the colourful drawing spaz my brain out

    • @fahadtube1406
      @fahadtube1406 7 років тому +3

      Statistics isn't an easy topic.

    • @karalanctot
      @karalanctot 7 років тому

      its true. its the material. he says it just like my stats teacher - it helps to write out each term and see the differences.

  • @lotus93277
    @lotus93277 4 роки тому +1

    So helpful! Thank you so much for doing this!

  • @alonsoquijano6749
    @alonsoquijano6749 6 років тому

    Thank you very much for this video. Having studied statistics for so many years and never actually understood the intuition of this formula. You helped me a lot. BTW, I think you made others people's head explode by repeating so many times "standard deviation of the sampling distribution of the sample mean" xD

  • @besimmons
    @besimmons 12 років тому +2

    This video muddles the distinction between "standard deviation of the mean" and "standard error of the mean".

    • @knwr
      @knwr 4 роки тому +1

      he describes the "standard error of the mean" as the "standard deviation of the sampling distribution of the sample mean", which is what it actually is. And he does make the distinction between this and the "standard deviation of the mean". The word "standard deviation" is usually not used when describing the "standard error of the mean", however it's useful to understand how the operation to get to the "standard error of the mean" is related to the operation to get to the "standard deviation". It is really just inception of variance.

  • @akshittuli1211
    @akshittuli1211 7 років тому

    Hey why so many "its too confusing" comments, watch his previous videos, or get a background of the topic....you are going to love it...
    Worked for me😃

    • @louieafk8424
      @louieafk8424 7 років тому

      Could you share that video? I am still confused about why the standard error should converge to 0 as the sample size grows instance of converging to the standard deviation of the population. Thank you

  • @chatsociety
    @chatsociety 11 років тому +1

    This sent shivers of my stats classes.

  • @renaudlunardo2164
    @renaudlunardo2164 7 років тому

    Super video. Good to see how much clearer the potentially confusing notion of standard error of the mean is. Thanks a lot for making it clear and exciting.

  • @ankurrawat8658
    @ankurrawat8658 4 роки тому +1

    Simulation made it clear. I was so confused before that.

  • @greenangelssky
    @greenangelssky 7 років тому +1

    Thank you, thank you, thank you, this was actually very very helpful.

  • @rdcruick
    @rdcruick 11 років тому

    I think eventually that is what will happen. People will make videos on subjects that are so concise and easy to understand that they will be the most efficient and productive means of teaching. With the following generations having to know more and more (as humanity itself learns more), it just makes sense that education will go in this direction. :)

  • @ssshukla26
    @ssshukla26 5 років тому

    The standard error (SE) of a statistic (usually an estimate of a parameter) is the standard deviation of its sampling distribution or an estimate of that standard deviation. If the parameter or the statistic is the mean, it is called the standard error of the mean (SEM) - Wikipedia
    So for this video, a. Standard Error of the Mean is equal to the Standard Deviation of the Sampling distribution i.e. (Standard Deviation of the original distribution / √n) where n is no. of samples.

  • @ColdByrdz
    @ColdByrdz 13 років тому +4

    6:32 "I'm just making that number up." ...No problem, I do that all the time. =p

  • @aabhaanekar9990
    @aabhaanekar9990 2 роки тому

    Thank you so much! Explained very beautifully! 👌🏼👌🏼

  • @beeshyak
    @beeshyak 14 років тому +1

    Great! I finally understand medical stats!

  • @Tommah1000
    @Tommah1000 5 років тому

    this is much better than Duke's coursera class on central limit theorem

  • @taladiv3415
    @taladiv3415 3 роки тому

    Thank you Sal for this explanation video and simulation.

  • @SureshM-uh5xf
    @SureshM-uh5xf 4 роки тому

    Your videos are really very very helpful for me , please if u have do some video regarding complex,differential and calculus icant able to understand it clearly

  • @yessicawie
    @yessicawie 8 років тому

    Thank you very much for all you videos. It really helps me out in my statistic courses!

  • @ulvund
    @ulvund 14 років тому

    @nazirdjon The probability distribution is the probability of taking a certain value in a sample space. e.g. for a die you have 1/6 probability for each outcome 1-6.
    The standard deviation is the average of the squared difference between the mean and the observations.

  • @jyotikavarmani4454
    @jyotikavarmani4454 7 років тому

    You bring the formulae so close home by making them so logical. Thanks for all your efforts :)

  • @nocturnalvisionmusic
    @nocturnalvisionmusic Рік тому

    12:24 remember that calculator from years past 📟😇

  • @andresvanmelkebeke2237
    @andresvanmelkebeke2237 9 років тому

    Good job !! Thanks! (from a Belgian UGent student)

  • @ecarlson
    @ecarlson Рік тому

    Awesome! Is the simulation available? I want to try it out.

  • @samruai3875
    @samruai3875 10 років тому +4

    Would the distribution become a vertical straight line when n -> infinity? Since the standard deviation -> 0
    But how come you said the curve become perfect normal when n-> infinity?

    • @saurabh75prakash
      @saurabh75prakash 6 років тому

      When n = size of population (from which samples are taken) the distribution will be a straight line/peak. He is incorrect, when n approaches the size of population, the standard error becomes 0.

    • @nidavelliir
      @nidavelliir 3 роки тому

      N can't be infinity. N can only increase up to the population size.

  • @harivgl
    @harivgl 4 роки тому

    When do we have to use the estimated population standard deviation from the sample, using Bessel's correction, instead of the sample standard deviation?

  • @MagnusAnand
    @MagnusAnand 8 років тому +2

    QUESTION:
    When you make the simulation and the simulator tells that the standard deviation (sd) of the sample mean is 2.33. That was calculated with the standard error formula? Or was calculated with the standard deviation formula?

  • @meanmanturbo
    @meanmanturbo 14 років тому

    Great stuff...but we need confidence intervalls as well, pretty please with sugar on top? Also, queue theory would be great as well, I realy need to brush up on my birth-death processes

  • @vishalsrivastava8613
    @vishalsrivastava8613 7 років тому

    This was excellent. Thanks, Sal!!!

  • @romanlawson9619
    @romanlawson9619 6 років тому

    The people at Khan Academy need to learn to spit it out.

  • @TheZaratustra12
    @TheZaratustra12 5 років тому +1

    what I find confusing is the word "sample" - when you say "sample" in English - do you mean one observation or do you mean an action where you take (e.g. randomly) several observations which, together, build one sample?

    • @zhengxijiang4087
      @zhengxijiang4087 5 років тому

      now find his wording is confusing in statistics he is pretty good with many things for example calculus , but I realised that he is not perfect. people has to be objective.

  • @shadmanmartinpiyal4057
    @shadmanmartinpiyal4057 6 років тому

    this video was awsome... clear concept

  • @bhavikganatra117
    @bhavikganatra117 10 років тому

    Thankkss

  • @haibinli5386
    @haibinli5386 11 років тому +7

    school time = 6 months
    khan time = approx. 15 min
    knowledge acquired = SAME SHIT

    • @uncommon2848
      @uncommon2848 5 років тому +7

      hey buddy you wrote this 6 year ago . How are you right now? How's is Life ?

  • @AlejandroErickson
    @AlejandroErickson 9 років тому

    This was great! How do you estimate the SEM when you do not know the SD of the original distribution? Any good techniques for determining bounds?

  • @shekarsubramanian9562
    @shekarsubramanian9562 6 років тому

    what is 'n' here n the formula... n is 1. number of samples for 1 standard deviation calculation or 2. is it the number of time that the calculation of standard deviation that was repeated ???

  • @saurabhverma1691
    @saurabhverma1691 5 років тому

    Hello Sal
    I just got super confused by the two topics in the section of Sampling Distributions. Those two topics are-
    1) Sampling Distribution of Sample proportion
    2) Central Limit Theorem
    They both seem to give a normal distribution shaped distribution with large sample size. The formula of mean is same for both but the formula of std deviation of the sample is different for them.
    It would be a great help if you could explain the basic difference between them?

  • @pawekopytek7596
    @pawekopytek7596 4 роки тому

    Can anyone tell me where the "n" in the denominator came from at 6:20?

  • @Atul_Nigam
    @Atul_Nigam 12 років тому

    Khan.... You rock...

  • @srajalbajpai8547
    @srajalbajpai8547 4 роки тому

    What is this app you are using for plotting graphs?????

  • @RImAPhan
    @RImAPhan 11 років тому

    Well put and exactly describes my teacher as well.

  • @zhtea
    @zhtea 13 років тому

    thanks khan!

  • @deen2dam
    @deen2dam 14 років тому

    good understanding video. it's better you can do a video on proof of the standard deviation of sample mean becomes that formula.

  • @amartyagupta8255
    @amartyagupta8255 7 років тому

    If you plot the mean of all the possible combinations of sample n from the population, you will get the normal distribution with the mean equal to the population. You don't have to reach infinity.

  • @sayantanghosh6714
    @sayantanghosh6714 2 роки тому

    Thank you so much sir!

  • @pg1282
    @pg1282 10 років тому +1

    So if I understand everything correctly, you can only estimate the standard error if you are performing multiple sampling, e.g. asking 100 people (always other ones), 100 times about, say, how long they have slept last night. Is it possible to estimate standard error using only one trial/measurement ? Or in that situation, the only thing we can obtain is the standard deviation ? So one would expect values like mean+-SEM to be more typical for population studies, gene expression analyses of numerous group of patients, but not in a single gene expression analysis where you compare, say, 5 control samples vs. 5 experimental samples ?

  • @raphaeldayan
    @raphaeldayan 3 роки тому

    I LOVE YOU KHAN!!!!!!

  • @kempisabel9945
    @kempisabel9945 5 років тому

    understood! thanks a lot!!

  • @MaxAgapoff
    @MaxAgapoff 8 років тому

    Khan is awesome

  • @InLoveWithFunkyPanda
    @InLoveWithFunkyPanda 9 років тому +132

    This was so confusing :S

    • @cu99460
      @cu99460 7 років тому +2

      Yeah I know exactly what you mean. But when you learn the central limit theorem, statistics makes allot more sense. Try and learn this first and maybe it might help.

  • @saanikagupta1508
    @saanikagupta1508 4 роки тому

    An amazing explanation!

  • @bob-ho3mu
    @bob-ho3mu 5 років тому

    where can I find the simulation used in this video?? Thanks!!

  • @sarrae100
    @sarrae100 8 років тому

    Makes it nice & clear !

  • @random-se1ep
    @random-se1ep 3 роки тому

    Question - doesn't central limit theorem apply only to larger samples (>30)?

  • @himanshumishra9284
    @himanshumishra9284 8 років тому +2

    Which software you're using in example

    • @uniksalman123
      @uniksalman123 6 років тому

      himanshu mishra Smooth Draw Software

  • @farvision
    @farvision 15 років тому

    Good for you buzwazfuz! Not everybody can handle that. You could teach yourself perhaps - if so you could zoom along at your own pace. If you think you can do that talk to your teacher about doing independent study.

  • @nbultman_art
    @nbultman_art 12 років тому

    14:36 sampleception!

  • @salgsalgglas
    @salgsalgglas Рік тому

    Hello, shouldn't the numerator be SD of the sample, rather than SD of the population divided by sample observations,..apologies if i am missing something?

  • @Ketanaut
    @Ketanaut 13 років тому

    THANK YOU!

  • @simeonhorozov9
    @simeonhorozov9 4 роки тому

    When we take the samples from a population, do the previous samples taken affect the choice of the next. Can a data point be taken twice in different samples?

    • @rajatsharma5309
      @rajatsharma5309 4 роки тому

      Yes, a data point can be taken twice or any number of times in different samples. This is because the samples are RANDOM SAMPLES.

  • @user-wetenweten
    @user-wetenweten 6 років тому

    thank you for your video, it helps my comprehension about this concept

  • @acinomknip
    @acinomknip 5 років тому

    this is an awesome video! can I ask a question tho why is there a formula says the standard error = sqrt(n)* the standard deviation of the original sample?? as n is the draw number. I thought it should be the standard deviation/sqrt(n) like you said??

  • @littlegiant101
    @littlegiant101 7 років тому

    What software did he use for sampling disruption

  • @wademartinishere
    @wademartinishere 6 років тому

    This was awesome!

  • @vikasns4603
    @vikasns4603 6 років тому

    very helpful.!

  • @sachinsaha3108
    @sachinsaha3108 4 роки тому

    Sir, please help in solving this....
    for the following data, calculate the standard error for mean when the sample is drawn under without replacement scheme. number of motor cycle accidents is 4.1 in a random examination of 8 cases out of 2500 with standard deviation being 0.9

  • @rescuemay
    @rescuemay 9 років тому

    Brilliant video, thanks so much!

    • @zhengxijiang4087
      @zhengxijiang4087 5 років тому

      I really cannot say his videos of statistics are brilliant. he muddles many things together, maining working quite confusing for people are learning.

  • @clemiboi
    @clemiboi 10 років тому

    awsome..........!

  • @sm8874
    @sm8874 8 років тому

    thanks a lot!

  • @yasminfatima5948
    @yasminfatima5948 4 роки тому

    When calculating std error of sample distribution we take square root of sample size in denominator. Why is it not sample no.s?