Tabulation and Dual System of Estimation (DSE) Pres. 9

1
Tabulation and Dual System of Estimation
(DSE)Pres. 9
2
Tabulations Dual System of Estimation (DSE)

• Tabulations Initial tables
• Estimated number of non-movers
• Estimated number of out-movers
• Estimated number of matched non-movers
• Estimated number of matched out-movers
• Estimated in-movers
• Estimated number of erroneous inclusions in the
  census
• Estimated number of correctly enumerated persons

3
Basic Tabulations for Coverage Analysis

• Tabulations for coverage estimates (by age group
  sex)
• E-sample estimates of population enumerated in
  the census
• P-sample estimates of total population
• Census omissions census omission rates
• Census erroneous inclusions census erroneous
  inclusion rates
• Preliminary DSE of True Population
• Net census coverage error and net coverage rate
• Gross census coverage error rate
• Final DSE of True Population

4
Examples - Tabulations for Content Error Estimates

• Sex as reported in the census and as reported in
  the PES for matched persons
• Age as reported in the census and as reported in
  the PES for matched persons
• Family relationship as reported in PES for
  matched persons
• NOTE PES should check content errors only for
  those items
  that
  support PES procedures

5
Dual System of Estimation (DSE)

• Omissions not a complete story in evaluating a
  census
• Census, as earlier stated, can contain duplicates
• Census can erroneously enumerate some one who
  should have been enumerated elsewhere
• Fictitious persons - in some cases
• Net coverage error
• Therefore need to estimate True Population

6
Dual System of Estimation (DSE) (contd.)
In census Out of census
In PES
Out of PES
Total
7
Dual System of Estimation (DSE) (contd.)
)

• Chandrasekaran-Deming estimator (1949) assuming
  independence

8
Dual System of Estimation (DSE) (contd.)
)

• The form of DSE used in PES is as follows

Where C census count II number of persons
with insufficient information CE estimated
number of correct enumeration from the
E-sample Ne estimated number of people from
the E-sample Np estimated number of people
from the P-sample M estimated number of
persons from the P-sample population who
matched to the census
9
Dual System of Estimation (DSE) (contd.)

• Coverage rate
• This is the matched population relative to
  estimated PES population
• The DSE is usually applied within
  substrata/subgroups formed e.g. sex and age -
  groups

10
Dual System of Estimation (DSE) (contd.)

• Erroneous inclusions rate
• This is equal to the total number of persons
  erroneously included in census relative to the
  Census population estimated by E-sample

11
Dual System of Estimation (DSE) (contd.)

• Matched population
• Matched population Matched non-movers
  estimated matched out-movers (thus, in practice,
  the number of out-movers is assumed to be the
  same as that of in-movers)
• Census population estimate
• Census population Matched non-movers matched
  out-movers population erroneously included in
  the census population correctly enumerated in
  the census but missed in the PES

12
Dual System of Estimation (DSE) (contd.)

• Estimated PES population
• PES population Census omission Erroneous
• PES population Number of non-movers in-movers
• Omission population
• Omissions PES population - (Census erroneous)
  PES population - Census Erroneous

13
Dual System of Estimation (DSE) (contd.)

• Omission rate
• Is the missed population between the census and
  PES relative to PES population

14
Dual System of Estimation (DSE) (contd.)

• True population
• True Population estimated PES population
  multiplied by population from the census after
  subtraction of erroneous enumerations and divided
  by matched population between the census and the
  PES

15
Dual System of Estimation (DSE) (contd.)

• Net coverage error
• The difference between what should have been
  counted, thus True Population and what was
  counted in the census
• Net Coverage rate
• The measure is the total net error relative to
  the DSE estimate of the true population (an
  important indicator of the quality of census
  coverage)

16
Dual System of Estimation (DSE) (contd.)

• Gross coverage error
• The sum of Omissions and Erroneous Inclusions
• Gross coverage error rate per unit enumeration
• This is the absolute gross error relative to
  census enumerated population

17
Evaluation of Content Error

• Net difference rate (NDR)
• The NDR is the difference between the number of
  cases in the census and number of cases in the
  PES that fall under a category (e.g. age-group)
  relative to the total number of matched persons
  in all response categories

18
Evaluation of Content Error (contd.)

• Index of inconsistency
• It is a relative number of cases for which the
  response varied between the census and PES. In a
  n x n table, numbers in diagonal cells show
  consistency (agreement). Off diagonal numbers
  represent inconsistency. This is calculated for
  each response from a category

19
Evaluation of Content Error (contd.)

• Other measures include
• It is Aggregate index of inconsistency
• Gross difference rate
• Rate of agreement

20
How to interpret content error measures
Index Low Moderate High
Absolute value of NDR lt 0.01 0.01-0.05 gt0.05
Index of Inconsistency lt20 20-50 gt50
Aggregate Index of Inconsistency lt20 20-50 gt50
21

• Thank You!