ABOUT THIS RELEASE
This working paper will assess the effect of using a Census form based on a self-coding rather than write-in format by comparing coded responses from the two forms used in the July 1992 Census test. The Optical Mark Recognition (OMR) form required responses to be self-coded while the Optical Character Recognition (OCR) form required responses to be written in. This is the first time that extensive analysis of the difference between the pattern of self-coded and write-in responses has been possible using parallel forms. Previously, data for the 1986 Census, which was mostly write-in, was compared with test data, which was mostly self-coded. The July 1992 Test was conducted with a view to assessing the possibility of using OCR technology in the 1996 Census but it has since been decided that OMR technology will be used as in 1991. The data will still be useful, however, in comparing written and self-coded responses. It should be noted that there were several problems affecting the quality of the test data. These are documented in 'July 1992 OCR Test, Report on Form Design Issues'.