Data quality policy

6 November 2013 Revised: 20 April 2018

Data quality policy


Background

As the national agency for Australian vocational education and training statistics, NCVER strives to ensure the accuracy, quality, integrity and confidence of its published statistics.

Aim

To promote public confidence in its published statistics.

Policy

NCVER has put stringent quality assurance procedures in place to assure the integrity of data in the national VET administrative collections and surveys. These comprise the following elements:

Setting statistical standards

NCVER oversees the Australian VET Management Information Statistical Standard (AVETMISS) which ensures the consistent and accurate capture and reporting of VET information about students, providers, courses and training outcomes.

Data validation

NCVER provides a free online data validation tool AVETMISS Validation Software (AVS) for RTOs to validate their data before submission and to satisfy the AVETMISS requirements.

Data quality checks

NCVER encourages data submitters to take responsibility for providing high quality data. To enable State Training Authorities and Board of Studies to assess their own data quality prior to submission, NCVER provides data quality checks and reports for these stakeholders in AVS.

Management of fieldwork contractors

NCVER follows sound principles of practice in selecting and managing fieldwork contractors to collect survey data. This includes a panel to select the fieldwork contractor based on merit.

Resolution of data queries

NCVER thoroughly checks data before using it for any output. We also raise any queries directly with data submitters, document any revisions made, and can replicate these changes if required.

Automation of data output

NCVER produces its regular statistical output and most data requests without having to manually manipulate data. This process reduces the possibility of human error. We have developed carefully checked software code and data cubes to provide these services.

Cross-checking of data output before release

NCVER cross-checks all data before it is published. A person other than the originator of the data always cross-checks the output and signs it off as appropriate. Where possible, we also cross-check data using different means from those used to generate it initially.

Documentation of sources, concepts and caveats

  • NCVER gives information in all publications on how the data were collected and what they cover, defining key terms and noting any issues.
  • NCVER staff and users are encouraged to report errors in published statistics if they do occur. We then correct the public record as soon as possible and review our quality assurance procedures to prevent the error recurring. Specifically, we undertake the following measures:
    • revise any products on the NCVER portal and clearly indicate where products have been changed
    • advise relevant stakeholders and, where necessary, issue a media release to correct the public record
    • report to the NCVER board
    • institute an internal review to identify how the error occurred, and refine our quality assurance arrangements as necessary.