|Page tools: Print Page Print All RSS Search this Product|
Census Working Paper 96/2
This report outlines the form design testing program undertaken for the 1996 Census and consolidates the earlier reports submitted after every testing. It reviews the objectives of the program, describes the form design testing of each topic or issue examined and provides recommendations to be considered for the next census in 2001. The objectives of this report are to provide readers with a better understanding of the way the 1996 Census form design was developed and to document the 1996 Census testing program for future uses. Further, the suggestions made for future improvements may be helpful for the next census cycle of form design tests.
The testing program was aimed to maximise the accuracy of responses for the 1996 Census questions. The assessment of the performance of the questions was generally based on:
For the first time, as part of the Census testing program, cognitive studies (eg. observation studies and focus groups) were conducted, for selected topics. These studies gives valuable first hand insight into respondent understanding which helps in the question design. Moreover, it helped in assessing the level of reliability of data, for further clarification of the results and also for further analysis of the questionnaire design.
It should be noted that due to the voluntary nature of census tests, the non-response rate is generally higher in the tests than in the actual census. For each of the following tests, two forms were used. These forms were handed out to alternate houses in the testing areas. The form which had the experimental questions was tested against the control form, which was generally similar to the 1991 Census form. Wherever possible the 1991 Census results were used as a bench mark.
1996 Testing Program
The 1996 Census testing program was undertaken to develop the most suitable form designs, field and processing procedures possible for the 1996 Census. Seven field tests and a dress rehearsal were conducted. Further details on the whole program are given in Appendix 1.
This test was conducted on Tuesday, 21 July 1992 in Brisbane. 4,974 dwellings were enumerated. The test was primarily aimed to assess respondent oriented aspects of the viability of Optical Character Recognition (OCR) as the processing system for the 1996 Census. OMR (Optical Mark Recognition) and OCR forms were used for this test. The OMR form was the same as that used for the 1991 Census with only minor changes. The OCR form was a five-person household form in landscape format similar to the OMR form in content and question order, but with a different format for many of the questions. A Personal OCR form was also used for this test. For a number of reasons, including a lower degree of apparent respondent acceptance it was decided that the 1996 Census and future test forms would continue to use OMR.
The second test was held in Adelaide on Tuesday, 30 March 1993. 983 dwellings were enumerated. Two forms (Form 1 and Form 2) were used to test alternative approaches. Form 1, the control form, was the same as that used for the 1991 Census with some minor changes, and Form 2 included the experimental question designs. No Personal Forms were used in the test.
The third test was held in Melbourne on Tuesday, 24 August 1993. 1,816 dwellings were enumerated. Two forms (Forms 3 and 4) were used to test alternative approaches. Form 3 was the same as that used for the 1991 Census with slight changes, and Form 4 included the new question formats being tested.
This test was held in Melbourne and 1,644 households were enumerated. After the release of '1996 Census of Population and Housing: ABS Views on Content and Procedures' in February 1993, the ABS received 48 submissions disagreeing with its recommendation to exclude a disability topic from the 1996 Census. After further discussions on census content, the ABS agreed to test a question on disability, and measured the quality of the data obtained by comparing it with information gathered by personal interview. Form 5 was used to test the two questions on disability.
The fifth test was conducted on 22 March 1994. Approximately 1,250 dwellings were enumerated in the Newcastle area. Forms 6 and 7 were used to test alternative approaches. Form 6, the control form, was similar to the 1991 Census form. Form 7 was the experimental form which incorporated the design and question wording changes being tested. A Personal Form, Form 8, was also used in the test and question designs were conceptually similar to Form 7.
The Major Test for the 1996 Census was held in Perth on Tuesday, 23 August 1994 and 11,732 dwellings were enumerated. Forms 9 and 10 were used in this test. Form 9 was similar to that used for the 1991 Census with some changes. Form 10 included alternative question design. Household and Personal forms, for both form types were used. A Summary Form for Non-private Dwellings was also tested.
Interview forms, known as Special Indigenous Forms, used to enumerate indigenous people in remote areas, were tested in late March 1995, in the Aboriginal community of Yarrabah in North Queensland. It was a small test which mainly utilised observation studies, for qualitative analysis of the forms. Three draft form types were used in the skirmish: the Community List, Special Indigenous Household Form and Special Indigenous Personal Form.
The Dress Rehearsal for the 1996 Census was held in parts of Sydney, Bathurst and some areas in the Northern Territory on Tuesday, 8 August 1995. Only the Special Indigenous Forms were tested for form design since the 1996 mainstream forms were finalised by this stage. Field collection and data processing procedures were tested for all forms. Accordingly, there is no form design report for the Dress Rehearsal. It should be noted that some tables have been randomly adjusted to meet ABS confidentiality policy.
These documents will be presented in a new window.