Data quality

In order to measure the quality of the data contained in the BPE, several types of control are carried out each year.

Data checks are carried out throughout the production process : comparisons with data from the previous year, quality control of geolocation, carrying out a field quality survey, checks carried out in the office, etc.

• Office checks
  They are conducted on comparison with other available sources. For example, the exploitation of the SIRUS directory makes it possible to control information from other sources and vice versa. These checks relate to the presumption of the absence or wrongful presence of equipment in the BPE, and do not allow any assertion as to its completeness.

In a special case, the maternities contained in the BPE take into account the check carried out each year with the INSEE birth file.

In addition, each year the BPE Division carries out expertise on the consistency and evolution of data, classifications or sources over several vintages, in order to test the robustness of the data published.

• Field checks
  Each year, about fifteen types of equipment are the subject of a quality survey, carried out in the municipalities with 400 to less than 10,000 inhabitants, concerned by the population census campaign.

This survey makes it possible to identify any surpluses (equipment present in the BPE and declared non-existent by the municipality) or deficits (equipment absent from the BPE and reported to exist by the municipality) of the BPE.

The main results of this survey are published on insee.fr. In 2020, it was not carried out following the cancellation of the census collection due to the health context.

This survey is being redesigned and was not conducted in 2021.

Quality of coordinates (x, y)

The coordinates (x, y) made available for each piece of equipment are accompanied by a quality indicator which can take the following forms :

• good : the maximum deviation of the coordinates (x, y) provided with the reality of the ground is less than 100 m;

• acceptable: the maximum deviation of the coordinates (x, y) provided with the reality of the ground is between 100 m and 500 m ;

• bad : the maximum deviation of the coordinates (x, y) provided with the reality of the ground is greater than 500 m and random imputations could be made ;

• not geolocated : the coordinates (x, y) are not available because the geolocation did not succeed for the equipment. Equipments from Mayotte are concerned.

Regarding the quality thresholds, INSEE does not guarantee total compliance with these : deviations from reality can be greater in a certain number of cases, and conversely, prove to be of better quality than the documentation does not indicate it.

For the 2021 BPE categories, the distribution of equipment according to the quality of the coordinates (x, y) provided is established as follows:

• good quality: 88.5 %

• acceptable quality: 4.3 %

• poor quality: 6.9 %

• non-geolocated equipment: 0.2 %