APPENDIX 1 METHODOLOGY
Methodology
The 2005 Innovation Survey was conducted by the ABS after incorporating the results of a review into the survey scope, questionnaire content and procedures adopted in the 2003 Innovation Survey. As a result, changes were made to the 2005 Innovation Survey. The scope of the 2005 survey was all businesses in Australia with employment recorded on the ABS Business Register of five or more employees, except those classified to the following institutional sectors or industry divisions:
- SISCA* 3000 General Government
- SISCA 6000 Rest of the World
- ANZSIC** Division A Agriculture, Forestry and Fishing
- ANZSIC Division M Government Administration and Defence
- ANZSIC Division N Education
- ANZSIC Division O Health and Community Services
- ANZSIC Division Q Personal and Other Services.
- Standard Institutional Sector Classification of Australia (SISCA)
- Australia and New Zealand Standard Industrial Classification (ANZSIC)
These institutional sectors and divisions were excluded from the innovation survey because of statistical and cost constraints. In particular, the ABS rationale was that different survey questionnaires would be required for these industries in addition to significant work being undertaken to align different types of statistical units, ie government units versus business units. Although Agriculture, Forestry and Fishing is generally seen as an innovative industry, it is predominantly made up of micro businesses (ie those with less than five employees) and innovation activity is normally conducted through Rural Research and Development Corporations. In addition, market forces do not necessarily apply for entities operating in the Government Administration and Defence, Education and Health and Community Services industries due to significant government involvement in the decision making process. The Personal and Other Services industry division was excluded due to high levels of provider burden.
The frame for the 2005 Innovation Survey was taken from the ABS Business Register (ABSBR). The ABSBR provides a comprehensive list of employing businesses, primarily based on their registration for the Australian Taxation Office's (ATO) Pay As You Go Withholding (PAYGW) tax scheme. The frame is updated quarterly to take account of the entry of new businesses, businesses which have ceased employing, changes in employment levels, changes in industry and other general business changes.
The 2005 Innovation Survey was conducted by mail and under the authority of the Census and Statistics Act (1905). It was based on a random sample of approximately 6,800 businesses which was stratified by industry, State/Territory and number of employees and with weighting factors used to construct the whole-population estimates. All businesses with 200 or more employees in scope of the survey were included in the sample, with the exception of manufacturing industries, where all businesses with 500 or more employees were included in the sample. The response rate of the 2005 survey was 93%. The conceptual definitions and guidelines of the 2005 Innovation Survey were based on the Oslo Manual (1997 and 2005 editions). Based on the framework of the Oslo Manual, the 2005 Innovation Survey was extended to include any innovative activity that was abandoned during the reference period or was incomplete at the end of the reference period. However, for this study only innovative activities completed during the reference period have been considered as the core measure of innovation.
The difference between the information obtained from a sample of businesses and the information that would have been obtained if data had been obtained from all businesses is known as the sampling error. Standard errors can be used to measure sampling error. They indicate the degree to which information may vary from the value that would have been obtained if all businesses in the population had been included in the survey. In this study, the sampling variability is measured by relative standard errors (RSEs). The RSE is a useful measure in that it provides an immediate indication of the percentage of errors likely to have occurred due to sampling variability and avoids the need to refer to the size of the estimate.
Any RSEs greater than 50% are considered by the ABS to be too unreliable for general use and they should be treated with extreme caution. They cannot be used for input into any decision making processes.
The 2005 Innovation Survey sample design was intended to produce acceptable standard errors at the one-digit ANZSIC division level, with the exception of the Manufacturing and the Property and Business Services industry divisions where acceptable standard errors for output at the two-digit sub-division level have been included in the sample design.
The survey scope includes businesses with employment of five or more people. The sample was designed to provide reasonable standard errors for the following broad business size bands:
- 5-19 (small businesses);
- 20-99 (medium businesses); and
- 100 or more (large businesses).
Although the survey scope includes businesses with five or more employees, some businesses had less than five people employed by the business by the time they completed the survey questionnaire. For practical reasons, these businesses were included in the data set and contributed to final estimates. For the purpose of analysis in this study, businesses with less than five employees are included in the "5-9 employees" category. This approach aligns with the data published in the ABS publication Innovation in Australian Business, 2005 (ABS Cat. No. 8158.0).
Inaccuracies or non-sampling errors may occur as a result of:
- the way in which businesses report data on the survey;
- the design of the survey questionnaire; or
- ABS data processing techniques.
However, every effort was made by the ABS to reduce non-sampling error by taking the following measures:
- Careful design and testing of the questionnaires and data processing systems;
- Providing detailed instructions to providers on how to respond to questions; and
- Detailed checking of reported data to ensure that it is logical, consistent and complete.
In addition, the data input editing process was supported by a post enumeration survey (PES), which was conducted by the ABS with several data providers in different States and Territories. It was used to identify problems with the quality of reported data. Quality issues were then targeted during the output editing stage of the survey to minimise survey bias.
This study only examines the qualitative aspects of innovation in Australian businesses. A number of comparisons have been carried out with the results of the previous study. However, since the previous study reported proportions relating to innovation on a three year calendar period (noting that innovation in 2001 and 2002 were reported on a combined basis), some modelled estimates were developed to determine innovation proportions over the calendar years 2002 and 2003. These estimates enabled a comparison over two calendar year periods (2002 and 2003 and 2004 and 2005) for some selected broad categories. Statistical hypothesis testing was also used to compare the relationships between data items for some selected population groups.
Print Page
Print All