Annual survey on industrial energy consumption 2015
EACEI 2015
EACEI 2015
Précision et fiabilité
Overall accuracy
The precision indicators (standard deviation, coefficient of variation, 95% confidence interval) take into account errors due to sampling, total non-response correction and calibration, but not the error related to partial non-response correction.
The variables used for the EACEI 2015 precision calculations are :
BR_QAUP (quantity purchased in electricity)
DR_QAUP (quantity purchased in natural gas)
JR_QAUP (quantity purchased in butane-propane)
LR_QAUP (quantity purchased in heating oil)
The indicators are as follows:
estimate of total
coefficient of variation in percentage = 100 * s / total
The results of the calculations are in the file "Calculs de précision pour les variables cibles".
- Precision calculations 2016 (pdf, fr, 151 Ko, 09/04/0028)
Non-sampling error and A4. Unit non-response - rate for U and A5. Item non-response - rate for U
The out-of-scope share in the sample is 2.9%. It is estimated at 3.4% in the sampling frame.
Non response error
Cf. "Effet de redressement" file, which summarizes the differences on the aggregate
coming from the adjustment for the target variables.
The total non-response rate is usually calculated from the variable ETAT by : Total
non-response rate = N/(R+N)*100. It stands at 19.6%. See "Taux de non réponse" This
document summarizes the distribution of units according to ETAT coding as well as
the non-response rates by grouping of sampling strata.
Cf. files "Taux de non-réponse partielle" and "Top variables de taux de non-réponse
partielle", which summarize on the one hand the overall partial non-response rates,
as well as by division of the French activity nomenclature (NAF), by workforce band,
and by region, for the target variables, and on the other hand the 10 highest partial
non-response rates, all variables combined.
- Top variables of partial non-response rate (EACEI 2015) (pdf, fr, 68 Ko, 01/01/2016)
- Non-response rate 2016 (pdf, fr, 50 Ko, 18/12/2018)
- Recovery effect 2016 (pdf, fr, 74 Ko, 17/12/2018)