Statistics New Zealand Release Practices Policy
1 To preserve its position of independence and impartiality in the release of statistics, Statistics New Zealand aims to ensure that the main findings of statistical collections and compilations are widely known, easily accessible, that equal opportunity in access to statistics is enjoyed by all users or potential users of the statistics, and that methods and practices are easily available. For any statistical series, this policy provides for consistent access across all forms of statistical services, at the time of release.
The principles derive from the Statistics Act 1975, United Nations Principles of Statistics, New Zealand Information Policy, and the Official Information and Privacy Acts. The authority behind the collection of official statistics results in clear obligations for access and release practices.
Funding of the release practices of Statistics New Zealand comes from both public and private sector funding, as appropriate for the particular release. These principles apply regardless of the source of the funding.
2. Key Policies
All releases of information will be consistent with the legal authority used to obtain the information.
The legal basis for all information will determine data management, access and release practices
All official statistics will be released in a manner that promotes equality of access by users.
We ensure impartial access, by following standard procedures with all releases.
Statistics New Zealand (in essence - the Crown) holds the copyright over all official statistics it releases, so that there is no exclusive right of use of any official statistical measure or database.
All release arrangements will assert Statistics New Zealand expectations of acknowledgement of Statistics New Zealand as the data source.
The timing of release of all statistics is known in advance, by a six-monthly public notice.
For statistics or statistical databases where no market sensitive statistics exist, and a high public interest is unlikely, the release time will be marked by updating the main access services of Internet the time series databases, and delivery of any customised requests.
Unauthorised persons should not be allowed access to statistics prior to embargo time.
The rare occasions when key officials or Ministers should have, as judged by the Government Statistician, early access to market sensitive measures, will be met through known practices (see box at end of this document).
There is a process that provides an option for simultaneous briefing of relevant Ministers at the time of a media release, currently starting at 10:30 and finishing at 10:45 am. The link provides further information. Other statistics, in addition to the core economic statistics, can be treated this way with GS approval.
Where a survey sponsor has a publication programme for the results of a survey, then Statistics New Zealand will provide for a reasonable period of analysis by the sponsor before Statistics New Zealand publishes the results, or a joint publication is issued.
If a market sensitive statistic has been released early to someone by accident, the Government Statistician will make an immediate release of the key features.
Statistics are to be disseminated as soon as practicable after they have been compiled.
All statistical information released via media release will be released within one week of its availability, after all forms of summarising, validation and presentation are complete.
The period of analysis by a survey sponsor will not exceed three months from receipt of a valid dataset or data.
Joint publication of statistical information with other agencies will be consistent with all Statistics New Zealand's release practices.
Other authoring agencies will agree to Statistics New Zealand's publishing policy, before a joint publishing arrangement is confirmed.
Even if a joint publishing arrangement has not been made, any significant collaboration between Statistics New Zealand and other agencies must be consistent with Statistics New Zealand's release policy (if applicable).
Widespread and equal access to key findings and summary statistics will be promoted through distribution via major public information service providers.
Charges for access to Official Statistics will be consistent with the charging and cost recovery policies of Statistics New Zealand.
Usual public access channels for information will have the common standard free access to the products so designed for those channels, plus low cost or reasonable support services at no cost, before charging for specialist or customised additional services cuts in. This is the media, public libraries, the Internet and parliament. Specific policies are developed and modified for these users from time to time.
Statistical information will be presented clearly and impartially, without advocacy, alongside commentary and analysis to enable a wide understanding.
Relevant commentary will be provided with releases to the general public.Statistical information will be presented in an impartial and objective manner.
Statistical information will not be presented alongside advocacy, or unsubstantiated judgement.
Regularly updated statistical information will be delivered in consistent formats.Where different measures of similar phenomena exist, usually one will consistently be given emphasis, to reduce user confusion. The difference will be explained even if the release focuses on one measure or users will be directed to the availability of other measures, e.g. income.Writing about statistics for the community will follow the "Good Practice Guide".
Statistics New Zealand expects acknowledgement of Statistics New Zealand as the data source.
All information releases will protect the confidentiality of individual respondents.
Confidentiality rules will apply to all statistical estimates.
All holders of microdata files will meet Statistics New Zealand's confidentiality standards when publishing results.
This can include individual persons, families, businesses, iwi, etc. as defined in Statistics New Zealand's confidentiality standards.
Knowledge of all statistical methods and practices will be available to users of official statistics.
Methods and practices will be documented and widely disseminated through appropriate access media.
Significant changes in statistical methods will be advised to users in advance of the release of new series.
Advice on use of statistics will be given through special publications where a need is recognised.
Statistical standards will be followed where they exist.
Statistics New Zealand's standards for graph presentation will be followed.
Statistics New Zealand's table presentation standards will be followed.
Statistics New Zealand media release practices will be followed.
Standard output classifications for variables will be used in published tables.
Statistics New Zealand will notify users of errors in releases, as soon as they are known.
Errors in key aggregates will be advised by correcting release.
Revisions to series will be presented consistent with Statistics New Zealand's revisions policy.
There is scope for exemptions from Release Policy Practice on providing access to statistics before they are released. The approval of the Government Statistician or Acting Government Statistician is required. For example, on rare occasions such as the release of a significant speech or public document, access to information by Ministers or key officials has been given up to two days in advance of its public release. Examples include the Prime Minister obtaining the latest migration statistics with Asia on the eve of her visit to Japan.
This does not apply to key market sensitive indicators, although special circumstances exist, particularly the publication of the Budget, where the Government Statistician is happy to forward series to Treasury in advance of their release.
This must be done on the following basis:
The series has gone through all of Statistics New Zealand's clearance processes.
Any passing on of the information from the Treasury recipient is in a document whose market sensitivity is at least as great as that for the GDP release.
This is not regarded by Treasury as early access for purposes other than the report.
We will provide the Hot Off The Press, but not the media release.
In the accompanying notes on our practices in the relevant Hot Off The Press, we will note that"This Hot off the Press was provided on embargo to the Treasury on ........, in order to be included in the 'e.g. Budget forecasts'."
A third situation is where there may be significant risk of national confusion if the PM or Minister of Finance is not adequately briefed on key economic statistics. The conditions for this are in point 4 in the link .
Policy for the First Release of Statistics
In the release of statistics it is the aim of Statistics New Zealand to maintain a position of impartiality and to ensure that all users and potential users of its products and services enjoy equal opportunity in access to the department's data.
Information is released under the basic principle of the community's "right to know".
2. General Policy
* The release of statistics data should be equally accessible to all users.
* Statistics should be disseminated to users, in Hot Off The Press form, as soon as practicable after the data on which they are based is available. This must precede
or coincide with the release of data via other media, namely INFOS, the Internet and the
Reuters/Telerate on-line release (in normal circumstances all four releases will be simultaneous).
* All Hot Off The Press releases are subject to an embargo.
* Selected Hot Off The Press releases are released by way of a media conference.
* Dates for the release of data should be set well in advance and publicised according to rules set by the Publishing and Media Services Division.
* When a Hot Off The Press release is the first or only publication to be issued for a collection, no statistics from the collection may be released prior to the embargo being lifted (with exceptions outlined in sections 4 and 10 below).
* All Hot Off The Press releases are released via the Publishing and Media Services Division in Wellington.
* At least one Media Release should accompany each Hot Off The Press release, other than for nominated releases. Current exceptions are:
- Estimate of Incomes of Persons
- Overseas Cargo.
In the event of statistics being released with a significant delay, a media release may not necessarily be produced.
3. Embargo System
* All statistics released by way of media conference are released at 10.45am. In general these statistics are monthly or quarterly time series that describe economic activity. They are of the highest importance for government and business policy formation, and for economic and financial analyses.
* All other statistics have an embargo of 3.00pm.
* If the preparation of a release is running behind schedule, it should still be released upon completion, even though the embargo time may have passed.
4. Revisions of Data
Revisions of past data will normally be released simultaneously with a Hot Off The Press release of current data. They should have the same embargo as the new data.
A significant error in any data will be announced by the release authority for the original
release at the time it is identified.
See also 'Revision of Data' policy number 3.1
5. Contract Surveys
* An advance copy of the results of a Contract Survey could be provided to any sponsoring body when the statistics are available, depending on the arrangement made in the contract at the time the survey is undertaken, and depending on the publishing commitment of the sponsoring body.
* Statistics New Zealand must make the information publicly available within two weeks of handing the advanced copy to the client, generally via a Hot Off The Press release.
* Agreement may be reached with the client to share the revenue from sales of contract survey data by Statistics New Zealand.
6. Joint Reports
* Approval of the Government Statistician must be obtained before any commitment is made to publish a joint report.
* The release of joint reports should follow the Release Policy of Statistics New Zealand, and the collaborating agency must comply with this.
* Statistics New Zealand retains final responsibility for joint reports including the content of the report.
* All joint reports must be written in a style that reflects the impartiality of Statistics New Zealand as a partner.
Guidelines for Publishing Data Subject to Error
Acknowledgement: This set of guidelines is based on Statistics Canada's (1987) "Guidelines on the documentation of data quality and methodology"
1. Agencies producing official statistics have a responsibility to inform users of the concepts and methodologies used in collecting and processing the data, the quality of the data produced, and other features of the data that may affect their use or interpretation. This obligation is summarised in Principle 10 of the Protocols for Official Statistics:
Be open about methods used; documentation of methods and quality measures should be easily available to users to allow them to determine if the data is fit for their use.
Why do users need this information?
2. Data users have to be able to verify that the concepts they have in mind are the same as, or sufficiently close to, those employed in collecting the data. To do this, a knowledge of the underlying conceptual framework and definitions used in the data collection is required.
3. Users generally recognise that data are subject to error and therefore need to know whether the data are sufficiently accurate to be useful to them. To make this assessment, they need to be informed of the likely principal sources of error and, where possible, the size of the error. They also need to know of unusual circumstances which might influence the data.
4. Given that indicators of data quality cannot be expected to be comprehensive, data users also require a knowledge of the data collection and processing methodology in order to verify whether the data adequately approximate what they wish to measure and whether the estimates they wish to use were produced with tolerances acceptable for their intended purpose.
Purpose of these guidelines
5. These guidelines supplement the discussion of principle 10 in the Protocols for Official Statistics, expanding on parts of that discussion. They are intended as a guide for agencies when the documentation and publication of results of a statistical programme are being planned or reviewed.
6. The level of detail to be provided in documentation on data quality or methodology will depend to a considerable extent on the type of data collection, how the results are being published, the range and impact of uses of the data, and the total budget of the collection/production process. This document provides only a guideline, and judgment is required in determining the level of detail appropriate for a given data set.
Guidelines on the description of data quality
7. Data quality is generally described in terms of sampling and non-sampling errors.
8. Unexpected events which influence the data should be flagged for users to help them in interpretation of the data.
9. It is not generally possible to provide comprehensive measures of data quality. Rather one should aim to identify what are thought to be the most important sources of error and provide quantitative measures where possible or qualitative descriptions otherwise. The result should be a balanced discussion which addresses itself to specific sources of error or bias and is therefore informative to users.
10. For censuses, surveys or administrative data surveys, the description should cover as many as possible of the elements described in paragraph 12 below.
11. Statistics derived from administrative data or from data collected by another agency can also be dealt with under the guidelines of paragraph 12, but it is likely that the information available will be less detailed. Nevertheless, important issues such as coverage, response errors and comparability over time should be discussed.
Data quality descriptions
12.The following aspects of data quality are regarded as basic and, subject to constraints of cost and feasibility, some indication of their level should be provided or made available, where applicable, for every statistical product:
a. Coverage: the quality of the survey frame or list (for surveys or censuses) or source files (for administrative data) as a proxy for the desired universe should be addressed (including gaps, duplications and definitional problems).
b. Sampling error: if the survey is based on a random sample then estimates of the sampling error of tabulated data based on the sample should be provided, together with an explanation of how these standard error figures should be used to interpret the data. The method of presentation may vary from explicit estimates of sampling error to use of generalised tables, graphs or other indicators. If the survey is based on a non-random sample, the implications of this on inferences that might be made from the survey should be stated.
c. Response rates: the percentage of the target sample or population from which responses or usable data were obtained (on a question by question basis if appropriate) should be provided. Any known differences in the characteristics of respondents and non-respondents should also be described as well as a brief indication of the method of imputation or estimation used to compensate for non-response.
d. Comparability over time: it may be appropriate to discuss comparability with the results of the same activity for a previous reference period, especially if there has been a change in methodology, concepts or definitions. If such a change would affect comparability from one time period to another, a quantitative estimate of this effect should be made whenever possible.
e. Benchmarking and revisions: the effects of benchmarking or revisions on comparability over time should be described. Guidance on the possible impact of future benchmarking should be given based on past experience.
f. Comparability with other data sources: if similar data from other sources exist they should be identified. Where appropriate, a reconciliation should be attempted describing how the data sets differ and the reasons for these differences. Comments on quality of the other data should be provided if an evaluation is available.
g. Other important aspects: there may be other aspects of data quality that are of prime importance given the objectives of a specific activity. These should be included with the basic indicators of data quality. Examples are: unusual collection problems, misunderstandings of the intended concepts by respondents, incidents such as disasters and strikes which impact on key variables, changes in classification or in its application, response based on financial years that do not correspond to a fixed reference period. In different surveys and at different levels of aggregation, different sources of error may predominate. Subject to cost limitations, the most important sources of error should be evaluated periodically, and the results made available to users in the most convenient form.
h. Non-response bias: an assessment of the effect of non-response on the statistics should be provided if possible.
i. Response bias: evidence of response bias problems stemming from respondent misunderstanding, questionnaire problems, or other sources, should be provided if available.
j. Edit and imputation effect: the effect of editing and imputation on the quality of data should be assessed.
k. Any other error sources: if there are particular sources of error or unforeseen events which are relevant to the series or occasion, these should be described.
Guidelines on the description of methodology
13. While all users should be provided with some appreciation of the methodology, some will require greater detail. Therefore, two levels of documentation should ideally be available:
a. General user reports that are prepared for a wide audience in order to assist them in interpreting the data and in deciding on their appropriateness for a particular purpose;
b. Technical reports that are definitive and exhaustive and give full and detailed information on methods underlying the data.
14. The amount of detail covered in methodology documentation will vary with the type of data collection (census, sample survey, administrative data survey), the medium of dissemination, the range and impact of uses of the data, and the total budget of the program. A reference to available documentation may be sufficient, especially when it is only to accompany a short response to a special request, or a summary report.
15. For data resulting from surveys or censuses, the methodology reports should provide at least an outline of the main steps in conducting the survey and should provide more detailed information on those aspects of survey methodology which have a direct impact on the quality and applicability of the data produced from the survey. The following topics should be covered where applicable:
a. Objective of the survey;
b. The target universe and any differences between this and the survey frame actually used;
c. The questionnaire(s) used and all important concepts and definitions (the discussion of concepts and definitions may well be very lengthy and require a separate document, or they may be included with the survey results; in the former case a reference to the separate document should be made);
d. The sample design and estimation procedures;
e. The method used for collecting the data (e.g. interview, telephone, mail, etc. and details of any follow-up procedures for non-respondents;
f. Any manual processing (e.g., coding) that takes place prior to data capture;
g. The method of data capture;
h. Quality control procedures used in connection with operations (e)-(g) above;
i. Procedures for editing the data and for handling non-response and invalid data;
j. Benchmarking and revision procedures used;
k. Seasonal adjustment methodology used;
l. The form in which the final data are stored and the tabulation or retrieval/query system, including confidentiality protection procedures;
m. A brief summary of the results of any evaluation programs; and
n. Any other special procedures or steps that might be relevant to users of the survey data.
16. For administrative data, or data sourced partially from sources outside your agency, the ideal is the information in paragraph 15. The following topics at least should be covered:
a. The data sources
b. The purposes for which the data were originally collected;
c. The merits and shortcomings of the data for the statistical purpose for which they are being used (e.g., in terms of conceptual and coverage biases);
d. How the data are processed after being received and what, if anything, is done to correct problems in the original data set; and
e. The reliability of the estimates, including caveats where necessary.
It will be noted that items (c) and (e) should be covered in a discussion of quality.
17. Where the results of statistical modelling and analysis are being reported, the methods used should be described, and quality statistics appropriate to that sort of modelling should be reported, e.g. goodness of fit statistics for statistical models.