EVALUATION OF THE 2003 POPULATION CENSUS DATA THE GAMBIA BY MR. ALIEU SARR PRESENTED AT THE UN REGIONAL WORKSHOP ON CENSUS EVALUATION AND POST-ENUMERATION SURVEYS, ADDIS ABABA, ETHIOPIA, 14TH

1
EVALUATION OF THE 2003 POPULATION CENSUS DATA
THE GAMBIABY MR. ALIEU SARRPRESENTED AT
THE UN REGIONAL WORKSHOP ON CENSUS EVALUATION AND
POST-ENUMERATION SURVEYS, ADDIS ABABA, ETHIOPIA,
14TH 18TH SEPTEMBER 2009
2
EVALUATION OF AGE

• Age misreporting is a common phenomenon in
  African censuses
• This may be due to Ignorance of actual age
• Deliberate attempt to increase or decrease age
  for one reason or the other
• Why evaluate the quality of age reporting?
• To establish the consistency of the data.
• Reveal past trends in fertility.
• Identify errors in reported ages.
• Determine the effects of migration on age
  structure.

3
EVALUATION OF AGE (cont.)

• Indices of Evaluating Age and Sex
• Sex Ratios The larger the departure of the Sex
  Ratio from 100 the larger the possibility of
  errors in the data.
• Age Ratios All Age Ratios should be closed to
  100 where fertility has not fluctuated a lot
  during the past and international migration has
  not been significant.
• Age-sex Accuracy index

4
Population Pyramid of The Gambia 2003
Male
Female
5
Population by Age and Sex, The Gambia, 2003
6
Sex Ratio by LGA
7
Myers Index of Age Heaping by LGA
8
Summary of Indices Measuring the Accuracy of
Data, 1993 and 2003 Censuses, The Gambia
Index Reported, 1993 Census Reported, 2003 Census
Sex ratio score 15.5 9.4
Male age ratio score 10.6 8.8
Female age ratio score 22.4 15.9
Accuracy index 79.4 52.9
The UN defines the values of the index as
follows lt 20 Accurate, 20-40
Inaccurate, 40 highly inaccurate
9
Key Findings

• Sex-ratio in urban areas is above 100, whilst
  that of rural areas is below 100
• Under-reporting of under-five population
• Age data are relatively more accurate in urban
  than in rural areas
• Age data for males are relatively more accurate
  than that of females
• Age-sex data in 2003 is relatively more accurate
  than that of 1993
• Level of accuracy of age data below recommended
  standards

10
Correction of Age Misreporting

• Methods of Smoothing Age Misreporting
• Smoothing without modifying the totals of each
  age-group
• Smoothing modifying the totals of each age-group
• Smoothing of age structure can be done using
  spreadsheets such as SINAGE, AGESEX and AGESMTH

11
(No Transcript)
12
(No Transcript)
13
Evaluation of Data on Children Ever Born

• Questions Asked to Determine Parity for Females
  aged 12 years and Over
• Of the children ever born alive to you, how many
  are living in this household?
• Of the children ever born alive to you, how many
  are living elsewhere?
• Of the children ever born alive to you, how many
  have died?

14
Evaluation of Data on Children Ever Born (cont.)

• How do we evaluate the data on children ever
  born?
• Average parities should increase by age of women.
  Average parities for the 2003 census
  progressively increased with age of women showing
  some consistency in the data.
• When parity not stated is less than 5 per cent,
  they can either be ignored or added to the
  denominator since their inclusion or exclusion
  will not affect the estimates. The parity not
  stated for the 2003 Census data was 2.5 per
  cent.
• Coale-Demeny and Brass empirical formulae to
  compare the results with the average parity for
  women 45-49 or P7. If the average parity for
  women 45-49 is lower than that estimated from the
  empirical formulae, then this can be an
  indication that there was under-reporting or
  omissions of children for women 45-49 years.

15
Average Parities, The Gambia 2003
Age Group Index Average Parity
15-19 1 0.190
20-24 2 1.107
25-29 3 2.487
30-34 4 3.884
35-39 5 4.873
40-44 6 5.482
45-49 7 5.609
16
Evaluation of Data on Average Parities

• Using the data from Table 2.1 above, the
  Coale-Demeny empirical formula is as follows
• (P3)2/P2 (2.487)2/1.107 5.588
• The Brass empirical formula yields the following
  result
• (P2)(P4/P3)4 (1.107) (3.884/2.487)4 6.581
• P7 or reported average parity for women 45-49
  from Table 2.1 above is 5.609
• Coale-Demeny formula compares favourably with the
  parity for women 45-49, which means there were no
  under-reporting or omissions of children by older
  women. However, the estimates from Brass formula
  indicate that there were under-reporting of
  children. The Brass formula provides a more
  robust estimate.

17
Evaluation of Data on Average Parities (cont.)

• Evaluation of data on children born during the
  year preceding the census showed some
  inconsistencies in the data with an
  over-reporting of births in the year preceding
  the census.
• General Conclusion
• There has been a general under-reporting of
  children of women in the older ages.

18
Deaths in the Year Preceding the Census

• Methods of Evaluation Data
• Growth Balance Method developed by Brass
• Preston-Coale (PRECOA)

19
Deaths in the Year Preceding the Census (cont.)

• Results of Evaluation
• Reporting of deaths more accurate in the older
  ages than the young
• Reported male deaths more accurate than female
  deaths
• Female deaths under-reported by 12.3 per cent
• Male under-reporting negligible

20
Recommendations

• Improve on enumerator training to improve
  accuracy of estimating age using existing
  techniques
• Minimize data collection through proxy interviews
  during census taking
• Institute stricter measures to ensure the
  recruitment of qualified enumerators and
  supervisors
• Put in place measures to improve on the quality
  of supervision during census taking

21
Thank you for the attention.