The calculation of the Austrian response burden analysis includes all compulsory surveys conducted by Statistics Austria.
Only a few and mostly very small compulsory surveys are excluded from the response burden barometer. Such surveys concern for example the freight transportation statistics by railway, air, inland water and pipeline and the annual statistics on overnight capacity in hotels and similar accommodation facilities.
Compulsory surveys conducted by the Austrian National Bank, the Austrian Federal Economic Chamber, governmental and other organisations are not included in the response burden barometer.
It is quite well known that one can distinguish between „perceived response burden“ and „actual response burden“. The first one is the subjective burden which will differ between the individuals and the second should be the real burden which in principle could be measured. Clearly, the measurement of perceived burden is out of scope of official statistics. Thus, only the actual burden can be the basis for the quantification of response burden. The actual burden can be measured by the time required to perform the task of completing a survey. The time required – measured in minutes and hours – can be compared between enterprises, surveys, over time, etc. and can also be aggregated to achieve totals.
The time required may not be viewed as the most appropriate indicator from an economic point of view. Time is an economic resource and economic resources are usually measured in terms of costs considering the different values of time. However, the time required would in any way be a starting point for any kind of transformation into costs. In the Austrian response burden barometer the time is the ultimate goal of measuring response burden.
The time required is the main indicator for the response burden. It is defined as the time that is directly needed to fill in the questionnaire, whether on paper or electronically. The time required thus includes the time needed to perform the necessary investigations into the book keeping, cost accounting or any other data system of the enterprise which contains the data requested, and the time required to make the necessary calculations. It also includes the time to actually fill in the questionnaire. If more than one person is engaged in the task of providing the requested data, the time required of all persons should be included as well as time needed for the coordination work within the enterprise.
In principle, only the time required caused by the statistical questionnaire directly, should be included. The time required in order to fulfil other administrative requirements should not be attributed to the statistical burden. For example, the calculation of the monthly turnover for the VAT tax payment should not be attributed to the statistical burden even if the statistical questionnaire will also ask the turnover figure. This time for the calculation of the monthly turnover figure should neither partly nor for a second time attributed to the statistical burden. This is quite evident when the statistical request comes after the deadline of the administrative request which is usually the case, because the reference in the statistical questionnaire to already existing data in the enterprise is an act of reducing the response burden.
The time taken to improve the data systems of the enterprise, such as book keeping, cost accounting, controlling etc. systems, should also not be allocated to the statistical burden, as these are multipurpose projects which are more of an investment character.
The aim is to measure the regular time necessary for the filling in of the survey forms. Therefore, it was decided not to include enterprises in the sample which have to report to a certain statistics for the first time or which use an electronic media for the first time. In both cases, the time required can be expected to be relatively higher than in cases where the survey form is already well known and data were provided earlier, as well as in cases where the enterprise is well acquainted with the electronic questionnaire.
In order to achieve a complete picture of response burden not only ongoing surveys are included in the calculation but also perennial surveys and new compulsory surveys. Thus, contrary to the principle of measuring regular completion time described in the preceding paragraph, also launching costs are included. Therefore the total response burden will be slightly overestimated.
For simplification reasons no differentiation according to the level of qualification of the respondent is made, making it impossible to measure any monetary burden.
Time captured does not refer to the reference period but to the year the burden actually accumulates. However, for simplicity, reporting efforts for surveys affecting different calendar years are assigned to one calendar year only. Response burden for monthly surveys is therefore aggregated for months January to December of one year despite the fact that the response burden for December affects the following year.
Completion time is measured at the level of the enterprise. If statistical surveys addresses establishments or local units response burden is summed up to the level of the enterprise.
Another conceptual question relating to the issue of coverage is whether the response burden should include the burden of all enterprises having delivered the questionnaire or of all enterprises that should have delivered the questionnaire. Even in compulsory surveys there is a small ratio of non-response. It was decided to use the latter concept which equals to the total burden imposed by the compulsory survey whether avoided or not. No distinction was made whether the declaration was made by the enterprise directly or whether a third party was authorised to complete the survey form (e.g. by commercial tax consultant offices).
It was evident that for a reliable measure of the time required for filling in the survey forms, it would not be sufficient just to ask a few enterprises or only experts. The aim was to measure the regular time necessary for completing the survey forms. Therefore it was decided to select small samples of enterprises randomly in order to obtain the information. Another option was to include a response survey form to the regular survey materials and to ask for completion and to return it with the regular questionnaire. This could be sent to all enterprises with a reporting task or just to the ones selected in the sample. The definition of the time required could also be stated quite clear and unambiguous.
However, both approaches were applied. In the first years when the response burden barometer was implemented it was decided to draw small random samples out of the survey sample and to make telephone interviews. When Statistics Austria has received a completed survey form of one of the surveys included, the person responsible for filling in the questionnaire in the enterprise was contacted immediately by telephone and was asked about the completion time. These telephone calls were done by the special staff members of the respective survey who quite often have to contact the person in the enterprise requesting clarifications on the data provided anyway. Thus, there are already certain contacts between the statistical office and the enterprise which are utilised and a short telephone interview will not be seen as a considerable burden.
Starting in 2015 a new methodological approach was chosen for the calculation of the response burden of selected surveys, including short term statistics, structural business statistics, material input statistics as well as the multiannual surveys. In contrast to the concept applied previously, i.e. surveying the response burden in a base year and keeping the mean response time per strata constant over time, the completion time for these surveys is now measured every time the survey is performed on a voluntary basis. The acceptance of these additional questions was considerably with response rates beyond 90 per cent of the sampling population; therefore the results of the response burden barometer are well-founded.
Subsequently the weighting of data is done after adjusting for outlying observations in order to reduce bias.
© STATISTICS AUSTRIA, Last Changed 19.07.2016