Census Planning and Management

1

• Census Planning and Management
• United Nations Statistics Division

2
Reference Material for Presentation

• Handbook on Census Management for Population and
  Housing Censuses, United Nations Publication,
  Sales No. E.00.XVII.15.Rev.1, New York, 2001
• Principles and Recommendations for Population and
  Housing Censuses, Revision 2 (forthcoming)
• Both are available on the United Nations
  Statistics Division web-site www.unstats.un.org/u
  nsd and on the CD for this workshop

3
Overview of Presentation

• Census Planning
• Quality Assurance
• Evaluation
• Attributes of a good statistical output

4
Census planning what is it?

• Census planning is the core process linking the
  different phases of the census cycle
• Preparation
• Field operations
• Processing
• Dissemination
• Evaluation
• Census planning is the most critical process to
  conducting a successful census

5
Census planning (contd).

• Aim of the planning process is to ensure that
• Each phase is properly resourced and organized
• The output of each phase is of sufficient quality
  for all subsequent phases
• All dependencies between the different phases are
  identified
• Due to long duration of census cycle, planning
  should not remain static but be dynamic and
  flexible to take into account changes that occur

6
Census planning (contd.)

• Each phase of the census cycle is dependent on a
  preceding phase
• The quality of the output from each phase has a
  direct effect on the success of the next phase

7
Census Planning (contd.)

• Issues that require careful consideration when
  planning a census
• Specifying the role of the census
• The role of Government
• Setting goals
• Developing project plans
• Monitoring project
• Developing a budget

8
Census Planning- The role of the census

• Issues to consider
• The census should be considered as part of the
  larger National Statistical Programme
• The prime role of a census is usually to provide
  an accurate count of the total population for
  each of the administrative regions of a country
• The key strength of a census is the ability to
  provide data for small geographic areas and for
  small population groups
• When data are not required at this level of
  detail, other statistical methodologies more
  cost-effective than censuses should be adopted

9
Census Planning The role of Government

• To provide
• The legal framework for the census
• Funding for the census
• Logistical support for the census

10
Census Planning Setting goals

• Needs and requirements of stakeholders should be
  used to establish the census goals, taking into
  account
• Costs
• Data quality considerations
• Logistical implications
• Need to maintain public cooperation and
  confidence

11
Census Planning Setting goals (contd.)

• Census goals generally revolve around
• (a) Topics on which to collect data
• (b) Confidentiality
• (c) Timeliness of data release
• (d) Data quality
• (e) The nature of the output
• (g) The total cost of the census
• Goals interact with one another. So priorities
  need to be set as there could be trade-offs,
  e.g., between what topics can be collected and
  costs
• Once goals are established, they should be
  communicated to the staff, and appropriate
  strategies should be devised

12
Census Planning Developing project plans

• Once goals have been established and strategies
  identified to implement them, more detailed
  planning begins
• A census is a large project broken down into a
  series of related projects that are dependant on
  one another

13
Census Planning Developing a project plan

• To understand the interdependencies, need to
  develop a framework with a hierarchical structure
• Projects (planning, preparation, field
  operations, etc.)
• Phases (Field mapping, training, evaluation,
  etc.)
• Activities (Enumeration area design, map
  production, etc)
• Tasks (Review previous census methods, procedures
  and outcomes Prepare enumeration area design
  manual)
• The plan should also include Milestones Specific
  points in time at which key outcomes are expected
  (to measure projects progress)

14
Census Planning Developing a project plan

• Some issues associated with each activity and
  task that need to be taken into account
• Timing
• Resources
• Risk management
• Goals
• Planning process involves identification of
  issues and adopting a consistent approach in
  listing them

15
Census Planning Risk management

• Project plans should also deal with risk
  management
• Risks are all possible events that could occur
  and have a negative impact on the success of the
  census
• Risks with significant likelihood should be
  managed explicitly by developing fully detailed
  plans parallel to the census plan
• Risk management is essential because of the
  importance of the census and the fact that it is
  an infrequent exercise
• The success or failure of the census may depend
  on the implementation of the plans associated
  with these risks if they occur

16
Census Planning Monitoring project plans

• Developing a good census project plan is
  important, but not sufficient
• The project plan must be monitored closely, and
  feedback delivered to all levels of management
• Results should be reviewed on a regular basis
• Most important components to track
• Time for completing a task
• Resource usage per task
• Cost per task
• Milestones

17
Census Planning Monitoring project plans

• Useful tool for planning and monitoring the
  Gantt chart

18
Census Planning Gantt Charts

• Each task has name, duration, start, finish and
  timescale
• Horizontal bars represent duration of tasks
  relative to each other
• Dependencies and milestones can be included
• Separate Gantt charts can be prepared for each
  sub-task and all activities to be completed
• Provides a visual perspective on work-loads
• Automated systems are easy to use and speed up
  planning process

19
Census Planning Census budget

• Census budget must be planned well in advance and
  cover all known activities
• Take into account that the census budget is
  highly cyclical (peaks during enumeration and
  processing)
• Sufficient resources must be allocated to each
  phase
• Resource needs for the dissemination phase need
  to be realistically assessed
• Funds allocated and used effectively on planning
  and preparation will result in savings in all
  other phases, namely in enumeration and
  processing operations

20
Census Planning Census budget (contd.)

• Monitoring the census budget is fundamental
• Monitor regularly (quarterly or even monthly)
  expenditures against funding for each project
• Estimates of expenditures for all years of the
  census cycle should be prepared in advance and
  reviewed yearly
• This would allow to identify on time possible
  shortfalls, and take appropriate measures

21
Census Planning - Administrative report

• What is it?
• Its the census historical memory a report
  where all census experiences are recorded
• It is not a detailed description of the process
  (documentation)
• What is the purpose?
• To retain as much as possible the skills and
  knowledge acquired in developing the census, and
  use them at the time of the next census
• Why is needed?
• Because after census results are released, most
  census staff usually move to other duties

22
Census Planning Administrative Report (contd.)

• When it should be prepared?
• As soon as the decision to take a census is made!
  ? Evaluation and recording should not be left
  until the end of the census process
• How?
• The structure of the report could be similar to
  the structure of the project plan, but it could
  be modified (group tasks or create sub-tasks when
  appropriate)
• Keep record of resources used (staff years and
  funds) and of changes to the planned schedule of
  activities (what changes, and why?)

23
Census Planning Administrative Report (contd.)

• How should it used?
• To plan subsequent censuses or other large scale
  statistical activities (i.e. lessons learned)
• To create synergies in the current census cycle
  planning and management tasks
• To transfer knowledge and share experiences with
  other countries
• To provide advice and technical assistance to
  countries that need them (relevant for donors and
  development partners)

24
Quality Assurance

• Due to the size and complexity of census
  operations, it is likely that errors may arise at
  any stage of the census
• To minimize and control errors, it is good
  practice to devote a part of the budget to
  quality assurance and control programmes

25
About Quality Assurance

• The objective of quality assurance is to provide
  standards and controls so that decisions can be
  taken quickly to correct or change on-going
  census operations
• There is no single quality assurance system that
  can be applied to all censuses. They need to be
  specified for each census and for each stage of
  the census operations

26
Whats Quality Assurance?

• Four attributes to quality
• relevance
• cost
• timeliness
• data accuracy
• Achieving a quality outcome is essentially about
  balancing cost, timeliness, accuracy and
  relevance - relevance usually determined early
  when topics and output determined

27
Measuring Quality

• The 3 attributes cost, timeliness and data
  quality are inter-linked. Higher quality data can
  be obtained for higher cost whilst timeliness
  increases the relevance and utility of data
• Deficiencies in quality are usually the results
  of deficiencies in the process rather than the
  actions of staff
• Key to achieving a quality outcome is to
  regularly measure the cost, timeliness and
  accuracy so that the process can be improved -
  using the Quality Assurance Circle
• Quality is relative, and based on what is
  acceptable, rather than a concept of achieving
  absolute perfection

28
Quality Assurance Circle
Measure Quality
Implement Corrective Action
Identify Root Causes of Problem
Identify Most Important Quality Problem
29
Quality Assurance

• People undertaking the process are in a good
  position to identify problems and suggest
  improvements
• Quality therefore relies on
• established, documented procedures
• systems to monitor outcomes
• active encouragement by management to involve
  staff in identifying and resolving quality issues
• Managers play a key role in achieving quality
• establish a culture of focusing on quality
• giving staff responsibilities to allow them to
  achieve
• ensure staff understand the philosophy of quality
• providing the opportunity for staff to contribute

30
Quality Control versus Quality Assurance

• Quality Control finding errors and fixing them
• relies on ability to find all errors
• can add significant cost
• fixing errors can itself introduce error
• puts the responsibility in the hands of the
  inspector
• Quality Assurance/Continuous Quality Improvement
  emphasis on improving the process rather than
  just fixing the error
• recognizes there will be errors in the process
• aims to improve the process as it proceeds
• gives staff a responsibility in improving the
  process

31
Activities to Assist in Assuring Quality of the
Census

• Testing the form design
• involving the public to learn how they understand
  the questions being asked
• involving stakeholders to ensure the results
  obtained are as expected, such as the Processing
  team to ensure the form works with the processing
  systems, and the subject matter specialists to
  ensure the questions are being answered as
  expected
• Testing the field operations
• How are the procedures implemented, does the
  training provide staff with sufficient skills,
  what suggestions do staff have to improve the
  process
• Testing the Processing, Dissemination and
  Evaluation
• do the systems perform as expected, how well do
  the documented procedures work, how accurate is
  the data obtained during the test
• It is ok for tests to fail - the purpose of
  testing is to learn and improve

32
Quality Improvement and the Census

• The quality circle can be applied to the entire
  census cycle with
• Performance in the previous phase being evaluated
  at any given level of detail
• Problems with quality ranked in order of
  importance
• Root causes identified and corrective action
  implemented

33
Quality Circle Dependency Chart
34
Evaluation

• Evaluation may be considered as the last stage of
  the census cycle or the first step in the next
  census cycle
• All aspects of the census program should be
  evaluated (strengths and weaknesses)
• Evaluation of the accuracy of the census data
  should be undertaken through
• comparing the census results with similar data
  from other sources such as surveys, previous
  census or analytical methods assessing the
  quality or degree of accuracy of the data, or the
  coverage of the population

35
Purpose of Evaluation of Data Accuracy

• To inform users of the quality of the current
  census data and to assist in future improvements,
  through
• Improving processes
• Establishing performance benchmarks against which
  the quality of the data from future censuses can
  be measured

36
Evaluation of Coverage

• This aims at estimating the extent to which
  people have been missed or duplicated
• The best way of estimating undercount or
  over-count is through a post enumeration survey
  (PES)
• Undercount may result from missing out households
  due to difficult access, people in transit,
  highly mobile persons, inaccurate mapping,
  misunderstanding census instructions etc.
• Over-count may result from duplication in
  overlapping E.As, long enumeration periods,
  misinterpretation of coverage instructions etc.
• Usually more people are missed than double
  counted

37
Coverage Post enumeration survey

• The PES should be independent from the census, as
  it provides an independent validation of the
  census count
• The PES must be representative of the whole
  country and of all population groups and should
  be conducted as close to the census as possible,
  without interfering with the census
• The census/PES matching greatly benefits from
  accurate recording of names and addresses during
  both exercises. Other matching characteristics
  include age, sex, birthplace and relationships

38
Accuracy

• Accuracy involves assessing the reliability of
  the data. Benchmarks can be established for each
  variable to assess the contributions of
  non-response, editing and imputation to data
  quality.
• Tables can be created comparing non-response
  rates between censuses. The tables can show
  responses before and after processing, edits
  done, and imputations made, in the compilation of
  data for each variable. This will show the impact
  of the processing system on the original
  responses.
• Other data sets (eg. a labour force survey) may
  give indicative information to benchmark the
  census on the size of the labour force

39
Detailed Analysis

• In-depth analysis should include comparisons
  between censuses and other data from surveys or
  administrative sources. The surveys need to be
  based on compatible standards and should be
  recent
• Census data on births and deaths, in most African
  countries, can only be investigated through
  indirect methods, as vital registration systems
  are deficient. These may include study of the
  age-sex distribution, stable population analysis,
  etc.

40
Attributes of a Good Statistical Output

• Relevance
• Meeting the needs of users
• Value as a level of demand for data
• Completeness
• Full coverage of the phenomenon
• An extension of relevance meeting users needs
  as completely as possible

41
Attributes of a Good Statistical Output (contd.)

• Accuracy
• Distance between the estimated and the true value
• Usually expressed in terms of errors coverage,
  sampling, non-response, response, processing and
  dissemination
• Comparability
• Across space (countries, regions of a country,
  districts)
• Across time (different time periods)

42
Attributes of a Good Statistical Output (contd.)

• Coherence
• Logical and substantive connection between
  outputs of different statistical exercises
• National Statistical System
• Uniformity of classification and other coding
  schemes
• Timeliness
• Elapsed time between release of data and the
  reference period

43
Attributes of a Good Statistical Output (contd.)

• Punctuality
• Degree to which pre-announced release dates are
  met
• Clarity
• Degree to which statistics are understandable for
  non-expert users

44
Attributes of a Good Statistical Output (contd.)

• Accessibility
• Degree to which statistical data can be obtained
  with ease
• Metadata
• Availability of information describing sources,
  definitions, methods
• Crucial for enriching the numerical information
• Indispensable for assessing comparability

45

• Thank you.