|
In
June 1997 there was a major landmark in the development programme
for the 2001 Census of Population in Northern Ireland with the
conduct of a Census Test, involving the enumeration of some 9,000
households in a number of pre-specified areas.
The
1997 Test was designed to examine three key design variables:
 |
Form design -
a conventional matrix
style form, as used in the 1991 Census, was compared with a new page
per person style form to determine their respective effects on
completion rates and general public reaction;
|
|
Form content -
forms containing a question on income were compared with forms which did not contain an income
question in order to assess the acceptability of including such a
question, primarily through comparing response rates for the two
types of form; and |
|
Collection methodology
- conventional enumerator collection of Census Test forms was compared
with a new postback method
and respective response rates assessed. |
In
addition, the Census Test also examined the key design variables in
the context of the socio-economic characteristics of the sampled
enumeration areas, viz the rural / urban circumstances of the areas,
their level of deprivation (deprived vs non-deprived) and their
religion circumstances (predominantly Protestant, mixed or
predominantly Roman Catholic).
Census Test Evaluation Survey
Following
on from the Northern Ireland Census Test a Census Test Evaluation
Survey (CTES) was conducted, the aim of which was to examine the
quality of the information given in the Census Test form by
repeating the Census Test questions with a sample of respondents. A
number of questions were also asked to determine the acceptability
of the form in terms of its overall appearance and to find out
whether there were any questions which had caused particular
difficulties or which respondents found unacceptable.
A
number of addresses were selected from 24 of the Enumeration Areas
chosen for the Census Test, such that each of the area variants and
form variants which made up the design variables of the Test were
sampled in a balanced fashion. One person per responding household
was scheduled for interview.
A
further sample of 216 addresses (9 per enumeration area) were
randomly selected from households who did not respond to the Census
Test.
The
CTES questions asked of the Census Test responding households fell
into three categories:
|
questions to assess coverage of people within the household; |
|
questions on attitude to the questionnaire and to sensitive
questions and about the burden imposed on householders; and |
|
questions on quality of answers to, and understanding of, the other
Census Test questions. |
The
questions asked of Census Test non-responding households were used
to discover reasons for non-response, and particularly whether this
was associated with one (or more) of the key design variables of the
Test. In the case of no contact, basic observation data on whether
or not the house looked occupied, house type
and Enumeration Area were collected.
Full
reports
The
full reports on the 1997 Census Test are available below in PDF
format
|
