Evaluation of socioeconomic data collected from censuses

1
Evaluation of socioeconomic data collected from
censuses

• United Nations Statistics Division

2
What can be done to check data quality?

• Check the internal consistency of the data
• Whether plausible when tabulated or
  cross-tabulated with other characteristics
• Compare with other sources -indirect comparison
• Graph the derived indicators
• Cohort analysis of certain indicators
• Disaggregate by sex and age
• Compare with other sources- direct comparison
• Post enumeration surveys, where people were
  re-interviewed and content of the census
  responses is verified
• Matching records with other sources

3
Main types of socioeconomic characteristics from
censuses

• Demographic and social characteristics
• Age
• Sex
• Marital status
• Educational characteristics
• Literacy
• School attendance
• Educational attainment
• Economic characteristics
• Economic activity status (labor force
  participation)
• Occupation
• Industry
• Status in employment

Core topics from the Principles and
Recommendations for Population and Housing
Censuses, Rev. 2
4
Education

• Three core concepts
• Literacy ability to read and write a short,
  simple statement
• School attendance current, regular attendance
  at an accredited educational institution or
  program
• Distinguished from enrollment, which means that
  the student is officially registered at school,
  not necessarily that s/he actually goes to class
• Educational attainment highest grade completed
  within the most advanced level reached in the
  educational system

5
Education

• How to evaluate the quality of data
• Cohort analysis
• Comparison of education indicators calculated
  from the census and other sources of data
  household surveys, administrative registers
• Plausibility in the distribution of data

6
Cohort approach

• Use the demographic concept of birth cohort to
  generate time series data for people who were
  born in the same year/period by their completed
  level of education and literacy
• Analyze successive census data by cohorts - if
  they show similar level for certain level of
  education while age increases
• Assumption
• People do not continue education after reaching
  certain age - based on educational system in a
  country

7
Cohort approach

• Data requirement Consecutive population censuses
  by age groups, sex and literacy status and
  completed level of education
• Example for two
• successive censuses
• conducted in 1994 and 2004

Literate Population, Morocco, Male Literate Population, Morocco, Male Literate Population, Morocco, Male
Age group 1994 Census 2004 Census
15-19 1,001,617 1319162
20-24 862,136 1097725
25-29 606,808 884574
30-34 498,536 772151
35-39 436,522 595152
40-44 354,813 527018
45-49 192,203 448700
50-54 134,027 362123
55-59 99,384 184222
60-64 74,235 125762
65-69 40,602 81536
70-74 27,775 56054
8
Cohort approach
Literate Population, Morocco, Male Literate Population, Morocco, Male Literate Population, Morocco, Male
Age group 1994 Census 2004 Census
15-19 1,001,617 1319162
20-24 862,136 1097725
25-29 606,808 884574
30-34 498,536 772151
35-39 436,522 595152
40-44 354,813 527018
45-49 192,203 448700
50-54 134,027 362123
55-59 99,384 184222
60-64 74,235 125762
65-69 40,602 81536
70-74 27,775 56054
Re- organization of data by birth cohorts
Literate Population, Morocco, Male Literate Population, Morocco, Male Literate Population, Morocco, Male Literate Population, Morocco, Male
Age group 1994 Census 2004 Census Year of birth
15-19 1001617 1319162 1984-89
20-24 862136 1097725 1979-84
25-29 606808 884,574 1974-79
30-34 498536 772,151 1969-74
35-39 436522 595,152 1964-69
40-44 354813 527,018 1959-64
45-49 192203 448,700 1954-59
50-54 134027 362,123 1949-54
55-59 99384 184,222 1944-49
60-64 74235 125,762 1939-44
65-69 40602 81,536 1934-39
70-74 27775 56,054 1929-34
   
       
9
Cohort approach
Impact of out-migration among literate population
?
10
Cohort approach
  1994 Census 2004 Census Cohort survival ratio (2004/1994)
Year of birth Male Male Cohort survival ratio (2004/1994)
1974-79 1001617 884574 0.88
1969-74 862136 772151 0.90
1964-69 606808 595152 0.98
1959-64 498536 527018 1.06
1954-59 436522 448700 1.03
1949-54 354813 362123 1.02
1944-49 192203 184222 0.96
1939-44 134027 125762 0.94
1934-39 99384 81536 0.82
1929-34 74235 56054 0.76
Age misreporting?
Impact of international out-migration ?
Expected pattern at older ages because of the
effect of high mortality risk
11
Cohort approach
Educational attainment, Republic of Korea, 2000,
2005 and 2010 censuses
12
Cohort approach
Educational attainment, Republic of Korea, 2000,
2005 and 2010 censuses
13
Internal consistency
Egypt, 2006 census
Age 11
Age 11
Age 14
Age 14
14
Comparison with other sources

• The possible sources of data for comparisons
  are
• Population registers,
• Vital registration systems,
• School enrollment data,
• Citizen identification systems,
• Social security/health systems
• Existing household surveys
• Serious discrepancies between census and
  survey/administrative distributions for census
  topics are indicative of error in one or both
  sets of data - cause(s) should be investigated

15
Comparison with other sources

• Close investigation on the followings
• The definitions used for socioeconomic data can
  have a significant impact on results make sure
  that definitions used are consistent or
  investigate how differences in definitions would
  affect
• Consistency in how the question(s) is phrased -
  the terms used for the questions can be key
  factor in comparison with other sources
• Consistency in reference time-investigate how the
  differences in reference time would affect the
  results of evaluation

16
Comparisons with existing household surveys

• The rationale for using existing household
  surveys to evaluate a census on a non-matching
  basis lies in the fact that
• sample surveys are often less affected by
  nonsampling error than censuses due to the better
  operational control, qualified field staff with
  intensive tranning programme
• therefore, survey data on population
  characteristics also measured by a census often
  are considered to be more accurate than the
  census results

17
Comparison with other sources

• Census results indicate slightly lower level than
  the DHS
• Census results are highly consistent with the
  results of DHS

18
Economic activity

• Activity status a persons relationship to
  economic activity during a short reference period
  (typically a week)
• Employed a person who worked a defined, minimum
  amount of time over the reference period (may be
  as little as an hour)
• Unemployed a person who did not work the
  minimum amount of time during the reference
  period but was willing and able to work and
  looking for a job
• Inactive (out of labor force) a person who did
  not work the minimum amount of time during the
  reference period and did not want to work/was not
  looking for work
• Both the employed and the unemployed are
  economically active!
• Employed unemployed labor force

19
Economic activity

• Difficulties
• What does work mean?
• Goods and services produced for the market
• Goods produced for own-use (replacing need to buy
  on market)
• In practice, this is quite difficult to measure,
  especially in areas with large agricultural or
  informal economies
• Womens home-based production in particular is
  often undercounted in surveys
• What does looking for a job mean?
• E.g. some persons may have registered at a
  government labor office months ago, but done
  nothing else. Are they looking for a job?

20
Additional economic characteristics

• Occupation type of work that the person
  performs (most recent ISCO classification)
• Industry kind of production that the
  establishment in which the person works engages
  in (most recent ISIC classification)
• Status in employment type of contract the
  person has with his place of work (ICSE from the
  ILO)
• Recommended categories Employee Employer
  Own-account worker Contributing family worker
  Members of producer cooperatives Persons not
  classifiable by status

Source Principles and Recommendations for
Population and Housing Censuses, Rev.2, United
Nations, 2008
21
Example Egypt labor force measurement (2006)
Egypt Labor Market Panel Survey (ELMPS) 2006
vs. Egypt Census 2006
22
Comparison with survey data
Egypt Labor Market Panel Surveys and Censuses
Data source Minnesota Population Center.
Integrated Public Use Microdata Series,
International Version 6.1 (IPUMS International,
with CAPMAS Egypt) and Egypt Labor Market Panel
Survey (ELMPS, Economic Research Forum, Egypt)
23
Comparison with survey data
Census result is consistent with ELMPS-market,
but not with ELMPS-extended reasons?
24
Comparison with survey data

• Two definitions of the labor force
• The market labor force includes all those who are
  either engaged in economic activity for purposes
  of market exchange or seeking such work
• The extended labor force includes those engaged
  in the production and processing of primary
  products, whether for the market, for barter, or
  for their own consumption the production of all
  other goods and services for the market and, in
  the case of households that produce such goods
  and services for the market, the corresponding
  production for their own consumption. (ILO,
  1982)
• The distinction between the two definitions is
  particularly salient for women in Egypt, many of
  whom engage in animal husbandry and the
  processing of dairy products for purposes of
  household consumption and are thus counted as
  employed in the extended definition of the labor
  force

Source Economic Research Forum, 2007, Labor
Supply, Employment and Unemployment in the
Egyptian Economy, 1988-2006, Ragui Assaad,
working paper no. 0701
25
Marital status

• Marital status is the personal status of each
  individual in relation to the marriage laws or
  customs of the country. The categories of marital
  status should at least include the following
• Single (never married)
• Married
• Married but separated-comprise both legally and
  de facto separated
• Widowed and not married
• Divorced and not married

26
Internal consistency control

• Expected pattern
• smooth declining with ages
• females are getting married in younger ages than
  males
• males are getting married slightly younger ages
  in rural areas compared to urban areas

27
Internal consistency control
1994 Population and Housing Census of Tunisia
Proportion of widowed men
28
Comparison with survey data

• Very similar patter between the census and DHS
• Slight difference in the last age group

29
Singulate Mean Age at First Marriage (SMAFM)

• Technique for estimating the mean age at first
  marriage when actual dates of marriage are not
  available
• Is a period measure (uses a synthetic cohort)
• Very simple data requirements
• Total number of women by 5-year age groups
• Total number of ever-married women by 5-year age
  groups

30
SMAFM calculation (1)
Source Wachter, Ken, Essential Demographic
Methods, Data from United Nations Demographic
Yearbook
PEMult
31
SMAFM calculation (2)
32
Some remarks

• Census results are closely connected with other
  areas of statistical activities, therefore
  supplementary information about the census should
  be provided to the users for clarification of
  comparability of census data with previous
  censuses and other data sources
• Further analysis for interpreting the findings