Determining sample size based on confidence and margin of error | AP Statistics | Khan Academy

Поділитися
Вставка
  • Опубліковано 17 гру 2024

КОМЕНТАРІ • 56

  • @zhushuncai2642
    @zhushuncai2642 4 роки тому +33

    For people who don't know why p hat needs to be 0.5. The issue is we do not know the value of p hat before the researcher conducts the study. It could be 0.4, 0.3, or 0.01 based on the result. So if we want the margin of error to be no more than +-2%, we need to calculate the worst-case scenario in which p hat is 0.5. For reason I don't know, 0.5 * 0.5 is bigger than any other combination such as 0.4 * 0.6, 0.2 * 0.8, etc.
    Let's say we use 0.4 as p hat, and we get a sample size of 1537. We use this sample size and conduct the study, and we find out the p hat is 0.5. We do not have the sample size we need which is 2401 and the margin of error is more than +-2% which is not the result we want.

    • @qitang3813
      @qitang3813 3 роки тому +3

      The reason which you don't know may surprise you: A square has the largest rectangular area among all kinds of rectangles given a fixed circumference value.

    • @redpotter4074
      @redpotter4074 3 роки тому

      @@qitang3813 always interesting how arithmetic and geometry can come together

    • @qitang3813
      @qitang3813 3 роки тому +1

      @@redpotter4074 That's right my brother.

    • @thiagoazevedo9539
      @thiagoazevedo9539 3 роки тому

      The curve of the function f(p) = p(1-p) = -p²+p is a parabola, because it is a second-degree polynomial. A parabola always has a maximum or a minimum. In the case of p(1-p), its curve has a maximum, i.e., there is a value of p which yields the maximum possible value of f(p) = p(1-p). It happens that, in every parabola of a second-degree polynomial, the point of maximum or minimum is always in the average of the roots of the polynomial. Since the roots of p(1-p) are 1 and 0, the average value, which is the point of maximum, is 0.5! Everything I told becomes clearer if you study calculus, more specifically derivative. With calculus, you can better understand, among many other things, the process of finding the maximum or minimum of polynomials. Stay safe! God bless you. =)

    • @MrAdy0207
      @MrAdy0207 3 роки тому +2

      You can also compute the derrivative of -p^2 + p which is -2p + 1 and make it zero to get maximum resulting in p = 0.5

  • @jgonz1325
    @jgonz1325 5 років тому +22

    Why does this guy know everything

  • @FM-kz8vq
    @FM-kz8vq 5 років тому +16

    I didn’t learn ANY of this in class & it’s on my stats test 🤦🏻‍♂️ thank you !!

  • @wtenho1518
    @wtenho1518 6 років тому +1

    Neat trick to calculate square numbers near 50: use (a + b)^2 = a^2 + 2ab + b^2, so (50 + b)^2 equals 2500 + 100b + b^2. In this case b = -1 => 49^2 = 2500 - 100 + 1 = 2401

  • @jonhadley8387
    @jonhadley8387 7 років тому +3

    because we don't have P-Hat, just use N = 1 / ME^2
    pick the closest answer it will be off but close

  • @irakozenaftari4163
    @irakozenaftari4163 4 роки тому +4

    Thank you for great content but the formula could have been simplified to this n=(Z^2*p*q)/E^2 (1.96^2*0.5*0.5)/.02^2 = n=2401

    • @heyitsbbiii
      @heyitsbbiii 4 роки тому

      Does p hat 0.5 still be used even though I have a 99% level of confidence? Why or why not? I hope you could help me

    • @heleenpieloor8270
      @heleenpieloor8270 4 роки тому

      Wow, this makes it so much easier. Thank you

    • @irakozenaftari4163
      @irakozenaftari4163 4 роки тому

      @@heyitsbbiii Yes, if you don't know the value p, always use the .5 because that is where p hat is maximized.

  • @RongrongTang-u7u
    @RongrongTang-u7u Рік тому

    Awesome explanation! THX!

  • @noutalenezi8475
    @noutalenezi8475 2 роки тому

    Thank you so much this makes a lot more sense!!!!!

  • @stitchedtogetherbunny
    @stitchedtogetherbunny 5 років тому +9

    Why does p hat equal 0.5??

    • @littleriley007
      @littleriley007 4 роки тому +4

      The confidence interval is .95 ( I doubt you need this now though, sorry :c)

  • @laurencooper9076
    @laurencooper9076 5 років тому +16

    confusing introduction of p=0.5. why??

    • @sudiptasaha738
      @sudiptasaha738 5 років тому +4

      p*(1-p) is maximum at p=0.5

    • @uniformstudy1294
      @uniformstudy1294 4 роки тому +1

      @@sudiptasaha738 if we take p = 0.5 then the sample which we will get will not be the smallest sample size.

    • @uniformstudy1294
      @uniformstudy1294 4 роки тому

      Please explain why we are taking 0.5 ?
      We require smallest sample size

  • @MrArtbyart
    @MrArtbyart 3 роки тому +1

    I would expect that the sample size would vary with the population size. What if the total population is less than 2,000? Thanks

  • @jeevanrana4076
    @jeevanrana4076 5 років тому +2

    Can we take margin of error 2% when confidence interval 95%?

  • @valdesito98
    @valdesito98 4 роки тому +1

    Hello, does anyonw know if there is a similar way to calculate stratified sample size? (without knowing the population size for each stratum)

  • @noellecook3660
    @noellecook3660 4 роки тому

    Why does p hat have to be maximized? And why does .5 maximize it? Would using .5 apply to any margin of error percentage in which the p hat is unknown?

  • @jerome450
    @jerome450 7 років тому +9

    Why did you use 0.5?

    • @TheRGuy
      @TheRGuy 6 років тому +8

      You mean as large as possible.

    • @nastyhaterz
      @nastyhaterz 6 років тому

      Because a normal distribution is assumed here

  • @bambino53
    @bambino53 4 роки тому +1

    Thanks my man you rule

  • @jaytums2818
    @jaytums2818 3 роки тому

    bruh ur a hero

  • @SpittingBritTeaEarlGrey
    @SpittingBritTeaEarlGrey 7 років тому

    Needed this

  • @ToyaTeeTee
    @ToyaTeeTee 2 місяці тому

    where does the 1.96 come from?

  • @drew75123
    @drew75123 6 років тому

    Is this Cochrans formula?

  • @callmecarpio
    @callmecarpio 6 років тому +2

    Where did you get the 1.96?

    • @girleatsbread
      @girleatsbread 6 років тому +1

      The t-table, if you look at the bottom, 95% confidence gives a z score of 1.96

    • @yashhooda293
      @yashhooda293 5 років тому +1

      you can also use your calculator menu - 5 - 5 - 3 and then put area in as .025 and leave mean to 0 and standard deviation to 1 and you will get -1.95996 which is the same thing as 1.96 according to my AP Stats teacher, hope this helps :)

  • @fikilefm9221
    @fikilefm9221 2 роки тому

    how did get 2/196?

  • @junglecat4480
    @junglecat4480 6 років тому

    Who made the formula? what is it called?

  • @zuhayranasrin7270
    @zuhayranasrin7270 5 років тому +1

    you save meeeeee

  • @joaovitorcampos6629
    @joaovitorcampos6629 6 років тому +2

    what if my population is lesser than 2401??

    • @Gabrielabc42
      @Gabrielabc42 5 років тому +1

      There's a different formula for a finite population

  • @fallout6067
    @fallout6067 4 роки тому

    What if there's no margin of error available?

    • @chrispbacon3326
      @chrispbacon3326 4 роки тому

      Then you'd use z*(a/2) times StDev of the square root of n to get Margin of Error.

    • @chrispbacon3326
      @chrispbacon3326 4 роки тому

      StDev divided by the square root of n*

  • @adeyeyeoluwafemi8403
    @adeyeyeoluwafemi8403 4 роки тому

    please what is my samlpe size if my population is 3500

  • @HoodSleepy2850
    @HoodSleepy2850 7 років тому +1

    NICE FIRST

  • @lpsglitterpaws8536
    @lpsglitterpaws8536 5 років тому +3

    very confusing!

  • @marianpalko2531
    @marianpalko2531 4 роки тому

    Might be a stupid question, but is this not at all affected by the total number of all of the community members?

  • @wildzach
    @wildzach 4 роки тому +1

    what

  • @anthonycarbonaro7890
    @anthonycarbonaro7890 5 років тому

    You made a total mess out of something very simple.
    n=(z alpha/2 * theta/E)squared.

  • @pelasgeuspelasgeus4634
    @pelasgeuspelasgeus4634 Рік тому

    So, you found the sample without caring about the population size. Does it make sense to you? It sounds like pseudoscience to me.