Testing of data, a criteria of population surveyThere is no denying the fact that after setting
questionnaire and process of regular data manipulation, the plan of operation
is virtually prepared. The plan sets forth the countrywide survey publicity
arrangements and other operational components of the survey plan. Survey
publicity uses every avenue of nation-wide publicity particularly media, loud
speaker, drum beating, and other traditional means. Other operational
components include scheduling of field training of EAs and SAs delivered
through a hierarchy of master trainers, and
the logistical arrangements for survey materials including dispatch of
books containing survey questionnaire and return of the filled in questionnaires
and record schedule. Specifying dates for training including mock record by all
enumerators and supervisors and the dates for actual record around the
specified survey night are the other key elements of the plan of operation.The testing of the various aspects of the operation plan
including record is carried out a few months prior to holding the actual survey
in all EAs comprising a Thana in
an operation known as Pilot Survey. The Pilot survey tests the operations
planned for all stages of survey including record, processing and evaluation of
results and provide information on the adequacy of field organization, quality
control measures and training program and help identify the procedural
weaknesses so that they can be rectified on time. By using OPTICAL MARK READER schedule for
survey record has simplified and cut down by at least two thirds the data
capture time, while insuring an almost error free data capture process. This in
turn has reduced substantially the overall data processing and tabulation time.
In fact full tabulation of the survey is completed within one year of survey
record which usually took around five to six years in pre- OPTICAL MARK READER
survey. Use of OCR and other state-of –the –art data capture equipment are
likely to reduce further the data processing time in future survey. Survey results
are tabulated hierarchically up from mauza [and villages comprising mauza and
mahalla and ward for urban areas], union, Thana, zilla, division and finally aggregated for the
country as a whole. For the purpose of dissemination survey publication
includes several volumes including the National Analytical Report and District
Reports for each district of the country.
These are supplemented by special
reports on Post record quality check survey, Sample survey, Analytical report
on urban areas and an administrative report on the survey operations. In
addition to hard copies, all survey data are available in machine-readable
forms-in diskettes and CDs.
Survey
Evaluation: Post Record Quality Check: A unique feature of Bangladesh survey is post record quality check survey that is
conducted immediately [usually 15 days to a month] after the main count. The
purpose is to check the overall quality of the survey main count by estimating
the degree of coverage error i.e. omission or duplication of persons and
content error or the errors in responding to survey questionnaire. Typically
coverage error in Bangladesh survey relate to missing of persons and entire
households from the main count. This arises largely from the high density of
population and the propensity to live everywhere, not only in dwelling but also
in every type of structures. The growing number of floating population and slum
dwellers particularly in urban and peri-urban areas adds to the problem.
Content or response errors relate primarily to age record, to literacy and
marital status. PEQC is conducted by completely enumerating a smaller number of
randomly selected survey record areas both in rural and urban areas through a
completely independent agency engaging a different set of field enumerators.
Results of the PEQC survey are matched with the survey record to estimate the
coverage and response errors of survey. The survey results however, are not
corrected for these errors except that on the basis of net coverage error the
global count of population for the country is adjusted and published as an
estimate allowing for undercount.