ABARE Fuel and Electricity Survey, Financial Year 2001/2002
Contents
I. Data Processing, Estimation, and Analysis
I1 Follow-up activities
What steps will be taken to ensure the expected response rate is achieved? (I1)
Note. Should detail what follow-up procedures are planned (such as, reminder letters, telephone follow-up of non-respondents, non-contacts, and refusals); and when and how many follow-ups will be attempted.
Sampled businesses will initially be sent a letter seeking to determine the most appropriate contact within the business for provision of energy usage data. All business contacts (including the best contact available for businesses that have not specified one) will be mailed a survey questionnaire. All non-responding businesses will be followed up via a mail reminder in the first instance and subsequently, by telephone, if necessary, to elicit cooperation and a response.
Return to top
I2 Performance measures
What data collection and processing performance measures have been defined? (I2)
Note. This item is concerned with processes which are in place to ensure that the timetable deadlines will be met, for example, tracking non-contacts and refusals, responses received by date, percentage of data captured by date, percentage of data edited by date, and the like.
A timetable of the key milestones of the survey has been developed, against which the actual survey performance will be measured.
In addition, the FES database on which the current survey data will be stored, will be undergoing significant redevelopment as part of the FES project, including the development of statistical reports identifying non-respondents, refusals, responses received by date and percentage of data captured by date.
Return to top
I3 Data entry and coding
What quality control procedures will there be for data entry and coding? (I3)
Note. Examples of quality control procedures include: clerical scrutiny of forms, on-line edits, and the like.
The quality control procedures which have been implemented for the data entry and coding component of the survey include:
- streamlining of the survey forms to minimise and/or eliminate manual coding;
- development and implementation of conversion tables within the FES database to eliminate the manual conversion of collected data;
- clerical scrutiny of forms prior to data entry;
- development and implementation of on-line editing within the FES database to correct data entry errors and respondent data errors which were not identified in the initial phase of clerical scrutiny; and
- the production of regular detailed statistical reports of data entered by respondent, refusals, responses received by date and percentage of data captured by date and summary reports by fuel classifications, by equipment classifications and industry classifications.
Return to top
I4 Outliers
Will outliers be identified, and, if so, how will they be handled? (I4)
Note. This item refers to the treatment of unusual or suspicious responses. It outlines treatment for responses which are found to be correct, and for responses which are found to be incorrect.
Data outliers will be identified using clerical scrutiny, comparison of usage estimates with known energy production, import/export and stock change data and adjustment, as appropriate.
Return to top
I5 Aggregation of data
Will data be aggregated into statistical tables, and if so what are the estimation formulae for the principal output data items?
(I5)
Note. Should include: how adjustments for outliers, imputation and frame undercoverage factors have been incorporated into estimation formulae; the variance formulae for variances that are to be published; and the weighting strategy.
Initial weights will be calculated as simple expansion weights at stratum level (eg. stratum population/number of respondents). Information on energy consumption and production is collected from both energy suppliers/producers and energy end-users. Information available from suppliers (from whom data will be completely enumerated, or collected as an aggregate from alternative sources), will be used to calibrate information collected from end-users (after being adjusted for exports). The supply information provides the total demand by fuel by sector (therefore the error is small) and the data provided by the end-user provides detailed information about how the energy is used. Weighted data is then aggregated to form output tables.
Return to top
I6 Non-sampling error
What allowances will be made for potential sources of non-sampling error in estimation? (I6)
Note. Should include: allowances to be made for non-response (such as imputation procedures, analysis of non-respondents); allowances to be made in estimation for frame deficiencies (eg. use of new business provisions); and other allowances to be made for other potential sources of non-sampling error in estimation.
Energy supply data is available from several sources at a variety of levels of disaggregation and is considered to be accurate. Energy consumption data is then reconciled with the supply data, with the end-use detail calibrated against the reconciled totals. This calibration process will control both sampling and non-sampling level error at calibration cell level. The potential for remaining non-sampling error to effect the accuracy of estimates will be discussed in the non-response report undertaken at the conclusion of the survey.
Return to top
I7 Analysis of results
What statistical data analysis techniques will be performed? (I7)
Note. Examples of statistical data analysis techniques include: significance testing, regression analysis, factor analysis and the like.
The survey results are analysed against a number of criteria, historical trends, current situation and projected requirements. The results of the survey are comprehensively reported in an ABARE research report against key contemporary energy policy issues. The results are also used extensively in a number of other areas, such as in the analysis of energy intensity trends, emissions of greenhouse gases and energy policy briefing.
Statistical data analysis may take place on the data at a later stage, subject to requirements of clients for any such analysis. Techniques that may be used on the data include regression at unit level and time series analysis at aggregate level. Econometric techniques will be used to project aggregate energy usage forward, incorporating aggregate predictions of future usage as supplied in this survey.
Return to top
I8 Availability of results
How will the results of the survey be explained / presented to the major users? (I8)
Note. If applicable, should include: tables; analyses; unit record data; cautionary advice to assist users in interpreting the data (eg suppressing data due to high sampling variability, presenting data in the right context especially in cases of low response rates or low participation rates); and graphs.
As per question D1, one of the principal outputs to be produced from the 2001 FES to be presented to major users, will be the publication of a comprehensive set of statistical tables on the amount of energy consumed and produced in Australia, by industry, by fuel type and by State for the 1998-99 to 2005-06 periods. The proposed publication is expected to be in a similar format to previous ABARE 'Australian Energy' reports.
A copy of the publication, 'Australian Energy - Market Developments and Projections to 2014-15' (ABARE Research Report 99.4) was one of the principal outputs from the 1999 survey, and provides an example of one of the principal outputs to be produced from the current survey. The report contains a snap shot of the type of data to be produced, whilst more detailed data will also be made available to clients in electronic format (on diskette, CD and e-mail).
In addition, the data produced by the survey will also be used in a number of other research and commodity streams within the bureau, including other publications.
Return to top
I9 Systems Testing
How have the data processing, aggregation and dissemination systems been tested? (I9)
Note. This item is concerned with the testing of systems to reduce non-sampling error.
New survey estimation systems and databases are currently under development. The systems will be tested on historical and/or dummy data to ensure readiness for use in processing the 2001 FES.
Return to top
-->
