Experimental statistics

About experimental statistics


These are new, alternative statistical calculation methods that have not yet been finally tested, methodologically harmonized, validated as official statistics and included in the Official Statistics Program. Experimental statistics are not suitable for legislation. Indicators can be compiled on the basis of administrative data and based on mathematical calculations, statistics based on forecasts.


How does it differ from Official Statistics?



There is no approved calendar yet, it is compiled/produced taking into account the need and possibilities

Published according to the approved Official Statistics Calendar

Administrative sources, mathematical forecasting methods, alternative information sources are used

Administrative sources, statistical surveys are used

Not included in the Official Statistics Program

Included in the Official Statistics Program

Not all indicators are comparable with data from other countries

The methodology is harmonized and often comparable with other countries

Methodologies are still being developed and tested

Approved methodologies


Why is it necessary to compile experimental statistics?

Experimental statistics are much needed to meet the growing needs of users. It can be even more accurate than the official statistics. For the purpose of publishing statistical information on the number of employees and job vacancies by municipality, persons employed are usually classified in the municipality where the enterprise or legal entity is registered. Thus, if a large trading company is registered in Vilnius, while its divisions operate in regions, the company's employees will be assigned to Vilnius City Municipality. In the development of indicators on an experimental basis, the number of employees is determined by the location of operational performance of enterprises. Under the experimental calculation method, employees are classified as local units of enterprises according to their place of registered residence.

Experimental statistics may meet the need for information that is not available at all in administrative data sources because such statistical surveys are not performed. This is the case, for example, with indicators for the officially non-observed/undeclared economy (ONE), which cannot be calculated in a usual manner because, of course, the ONE is not reflected in official data sources. However, the range of available information and mathematical models can be used to estimate the share of the non-observed economy (ONE) in GDP.

Experimental statistics may be more operational, more efficient than usual, i.e. official statistics, because they also use prognostic methods - in the absence of administrative data, they are forecasted according to the indicators of the previous year. For example, experimental poverty indicators are published almost a year earlier than usual poverty statistics. Forecasting models based on survey data on income and living conditions for the last few years and relevant information from administrative sources are used to compile these statistics.

Alternative ways of obtaining certain data may also be applied. Administrative data were used experimentally to compile job vacancies, which were combined with job advertisement information retrieved from the Internet. Later, the results obtained are summarized and published.

Obviously, the production of experimental statistics requires a great deal of creativity on the part of professionals: how to obtain certain information sooner, faster, or how to calculate indicators that are not recorded anywhere.