2.10 Comparing modes of data acquisition#
It is of interest to list and discuss the main advantages and disadvantages of the three main data acquisition modes discussed above and the feasibility for the producers of official statistics in utilising them. Starting with surveys and survey-based censuses, the main advantage is that the data is in advance, defined for statistical purposes. The data collection is organised to respond precisely to the need for data and statistics about specific phenomena. It also follows that successions of surveys on specific matters carried out at specific intervals, monthly, quarterly, annually or at other intervals, can be kept fixed, entailing that the different surveys are basically comparable over time. Furthermore, new variables can be added to capture new issues, and some existing ones cut as they become obsolete. Against these advantages weighs the substantial cost of carrying out surveys and survey-based censuses.
Advantages |
Disadvantages |
|---|---|
- The main advantages of utilising administrative data are the relative ease and low cost of acquiring the data, given that the NSO and the other producers of official statistics are granted access to the administrative data or provided with the data regularly. |
- The main disadvantage of using administrative data is that it is collected for administrative use and may not be a good match for the statistical needs of given issues. Thus, the administrative variables often do not correspond to requested statistical variables and may not be immediately organised into the statistical classifications applied for the relevant issues. In such cases, the administrative data may not be sufficient and must be augmented by statistical surveys. Labour force statistics are a good example of this as available administrative data in many countries does not satisfy the data needs as agreed internationally and required for domestic monitoring and policy purposes. For this reason, labour force sample surveys are still carried out in most countries. |
A general requirement for using administrative data is that administrative systems have been developed and are available for statistical purposes. The more developed these are and the more embedded they are into the workings of the societies, the easier and more feasible it is to replace surveys with administrative data capture. Conversely, using administrative data for statistical purposes is less feasible in countries with poorly developed administrative systems. In a few countries, civil registration systems are operated based on unique identification numbers of persons that are used in the entire administrative system. Similarly, there are business registration systems that apply unique registration numbers of firms. In these countries, the utilisation of administrative data is greatly enhanced, particularly as this allows linking data from the different administrations for statistical purposes. It has to be borne in mind, however, that such data linking has to be exercised carefully and may be restricted for reasons of ensuring full confidentiality of the data and the need to respect requirements for privacy of individuals, households and businesses.
Their novelty and the richness trigger the interest in making use of other data sources (Big Data). This is thought to open up possibilities for acquiring data on new phenomena in various fields, such as in commerce, communication, and social media, that may allow new or extended analysis of economic and social matters. It has been found, however, that harnessing some of these sources is easier said than done. For statistical purposes, some sources are poorly defined, insufficiently structured, or lacking consistency and comparability. Another factor is that it has proven quite difficult in many countries to obtain permission from firms to access their databases and data streams, as the firms prefer to keep their business transactions confidential. Nonetheless, it seems likely that various types of new data sources will be further harnessed by developing novel methods, applications, and algorithms for this specific purpose.