Q 3.32

Discussion in 'CT3' started by dextar, May 1, 2013.

  1. dextar

    dextar Member

    Hi
    I have two doubts here
    1) I'm not able to understand the part 1(b)
    CAn anyone explain me how did they calculate this expression

    pl=P(Bin(10,pl)>=8)

    I'm not able to understand >=8 sign. They have mentioned the exact value of 8.
    Also, out of curiosity how do we get the pu (the upper part of the interval)?


    2) In Example 11.3, we have to estimate the interval of q and we have estimated q only to be 45/250 and went to calculate q. Shouldn't we use different subscript for these? I can spot similar thing for q3.32 (ii) b where the interval of Y has to be estimated, we have again put Y=1/2 to calculate same Y=526.01?
     
    dextar, May 1, 2013
    #1
  2. John Lee

    John Lee ActEd Tutor Staff Member

    When doing an "exact" CI for binomial or Poisson with observed data x, then for a 95% 2-sided CI we'd have:

    P(X<=x) = 2.5% and P(X>x) = 2.5%

    In this case we have a 1-sided 95% CI, so this is:

    P(X>=x) = 5%

    For 2 -sided CI see above. For 1-sided it would be:

    P(X<=x) = 5%

    Personally, I would have put hats on the q's in the denominator.

    This is not Example 11.3.
     
    John Lee, May 3, 2013
    #2
  3. dextar

    dextar Member

    Thanks for the explanation. However, I can't get why it is one sided CI. CI are supposed to be 2 sided right? Otherwise what is the purpose of the word "interval"? Kindly let me know I think I'm missing somehow here.

    This is Q3.32 ii (b) in question bank where Y has to be calculated. should we put a hat on y =1/2 as well?
     
    Last edited by a moderator: May 7, 2013
    dextar, May 7, 2013
    #3
  4. John Lee

    John Lee ActEd Tutor Staff Member

    They can be both - see last CR paragraph of page 3 of Chapter 11.

    It's still an interval - for example (0.62, infinity).

    No as Y is the observed value not an estimate.
     
    John Lee, May 7, 2013
    #4
  5. dextar

    dextar Member

    Understood thanks

    Just one clarification. we used here one sided CI because of P(a)>1/2 in question?
     
    dextar, May 8, 2013
    #5
  6. John Lee

    John Lee ActEd Tutor Staff Member

    No, because it says "a lower 95% confidence limit".
     
    John Lee, May 8, 2013
    #6

Share This Page