Due to administrative data not being collected for the sole purpose of statistical analysis, issues arise in the collection of the data that need to be considered in order to manage its quality. In some instances questions about the same idea or topic used on administrative forms may differ across jurisdictions. In these instances it is important to understand these differences and how they impact on the reporting of the information.
Testing of questions, peoples' understanding of the concepts and quality of the form design may not have been undertaken in the case of administrative data collections. Hence, the reported information may not contribute to valid measures of the desired concepts.
Interviewer or administrator biases and errors may occur depending on who supplies the information which is captured. In some cases a third party may be filling in the data and may or may not seek answers from the unit of interest.
Where the target population are required to complete a form or otherwise respond to data items, the administrative data may suffer from partial non-response. This is due to missing fields where respondents either haven't understood the question, accidentally missed the question or refused to answer the question. In these cases it is hard to know the reason behind the non-response, which may impact on the ability to analyse and use the data. An output editing strategy may need to be developed in advance of acquiring the data in order to have procedures in place for how to deal with such issues. A discussion with the data custodian up front about whether they can contact the reporting units and obtain data for these missing values through follow up procedures is an essential discussion to undertake. This will enable a decision to be made as to how potential missing data will be managed.
Administrative data can also be collected as either flow or stock data. Flow data are accumulated during the reference period. For example birth data is a flow series as the number of births are tallied throughout the year to provide the accumulated total of births for the year. Stock data are measured at the end of a reference period. For example company net profits as at the end of the financial year (30 June). Knowing what type of data are being acquired is important from a data quality management perspective (e.g. duplicate records).