census data questions

Discussion in 'CT4' started by swati, Aug 14, 2014.

  1. swati

    swati Member

    Hi, I am having difficulty understanding questions pertaining to census data
    Please guide
     
  2. John Potter

    John Potter ActEd Tutor Staff Member

    Hi swati,

    What particular aspect of which particular question can we help you with?

    John
     
  3. r_v.s

    r_v.s Member

    Hi Mr. Potter,

    In connection with problems on exposed to risk, would you pls explain how to proceed when census and death data have different age definitions?
    That's where I get most muddled :(

    Thanks!
    V S R
     
  4. swati

    swati Member

    Hello Mr Potter,
    I have difficulty in in linking census data and death data when they have two different definitions. I would like to quote a sum here which appeared in one of our Indian CT4 paper.

    A mortality investigation was held between 1.1.2007 and 1.1.2009.The following information was collected .The figures in the table below are the numbers of lives on each census date with specified age labels

    Date
    Age last birthday 1.1.2007 1.1.2008 1.1.2009
    48 3486 3384 3420
    49 3450 3507 3435
    50 3510 3595 3540
    During the investigation there were 42 deaths at age 49 nearest birthday.Estimate mu(49) stating any assumptions you make.

    Thanks and regards,
    Swati
     
  5. John Potter

    John Potter ActEd Tutor Staff Member

    Hi Swati,

    Here, mu(49) = Theta(49) / E49c = 42 / E49c

    We have to make the denominator correspond to the numerator. This is the PRINCIPLE OF CORRESPONDENCE.

    Here, theta*(49) = number of DEATHS aged 49 nearest BDay
    I invent P* so that it corresponds with the definition of Theta(49):
    P*(49,t) = number of PEOPLE aged 49 nearest BDay at time t

    This bit is easy, you just copy and write "people" instead of "deaths".

    Now, E49c = integral t = 0 to 2 of P*(49,t)

    which we approximate using trapezium rule as 0.5 P*(49,0) + P*(49,1) + 0.5 P*(49,2)

    All that remains is to approximate P*(49,t) (the data we want) from P(49,t) (the data we have)

    P*(49,t) = number of PEOPLE aged 49 nearest BDay
    P(x,t) = number of PEOPLE aged 49 last BDay

    The people aged 49 nearest birthday will be between the ages of 48.5 and 49.5. So, half of them are 48 on their last birthday and half of them are 49 on their last birthday:

    P*(49,t) = 0.5 {P(48,t) + P(49,t)}

    Copy this in for t = 0,1 2 above:

    E49c = 0.5*0.5{P(48,0) + P(49,0)} + 0.5{P(48,1) + P(49,1)}
    + 0.5*0.5 {P(48,2) + P(49,2)}

    = 0.5*0.5{3486 + 3450} + 0.5{3384 + 3507}
    + 0.5*0.5 { 3420 + 3435}
    = 6893.25

    So, mu(49) = 42 / 6893.25 = 0.0060929

    Check my numbers but I hope this is the answer.

    Good luck!
    John
     
    Last edited: Aug 20, 2014

Share This Page