Indirect Estimation

1
Indirect Estimation

• Infant, Child and Life Expectancy using Census
  data

16th September 2005 Fourth Sub-regional Training
Course/Workshop on Statistics for MDG Indicators
Roko Koloma National Statistical Office Papua New
Guinea
2
Why use indirect estimation for mortality
indicators in PNG?

• The usual sources such as the Civil Registry
  Office and the Health Information System are
  unable to produce reliable and accurate data for
  the direct estimation in PNG.
• The procedure of indirect estimation has been
  used since 1980

3
What is the next best option available?

• The population censuses conducted every ten years
  (1980, 1990 and 2000).
• The DHS conducted as inter-censal periods.
• These two sources then become the available means
  to generate the required estimates on mortality.

4
What were the questions used in the obtaining the
required data?

• Q20 How many children have you given birth to?
• Q21 How many of your own children are living in
  this house?
• Q22 How many of your own children are living
  elsewhere?
• Q23 How many of your own children are dead?
• Q7 Is the persons own mother still alive? (For
  the maternal orphan-hood technique)

5
Possible Errors or limitations to the data used.

• Omission of lifetime fertility of never married
  women.
• Report by proxy respondents
• Adoption bias
• Reporting of live births as still births or vice
  versa
• Non-response
• Zero error

6
Methodology

• Contrary to births and migration, death occurs
  only once to every person as part of the
  demographic process.
• Using information on Children Ever Born,
  Surviving and Dead by age and sex.
• The Average number of children ever born (Pi),
  surviving (Si) and Dead (Di) were calculated for
  women in 5 year age group. (see table A1)
• Then the Di values are then converted in
  life-table probabilities (qx)
• To convert to life-table probabilities, the
  modified brass method was used in the conversion.
  This was done using the CEBS program using the
  Mortpak Software program.
• The output produces values for each of the
  life-table models.
• Brass method on Logit Transformation is used to
  adjust lx values for the women age groups 20-34
  years.

7

• Once all the lx values are adjusted then the
  analysis of the adult mortality is done.
• The Brass/Hill Maternal Orphan-hood Technique
  which requires several iterations to calculate
  the alpha and beta values.
• These values are then used to solve to
  simultaneous equations, which are required to
  construct the two-parameter Logit life-tables.
• The remaining procedure requires the use LIFTB
  program in Mortpak to generate the require
  life-table.

8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
Thank you for your attention