Computational strata (for variance estimation)
In the cases where a country stratifies the sample by (for example) size class (using the standard output), and within each size class, the country uses systematic sampling by ordering the holdings by UAA, then Eurostat defines the "computational strata" by pairing the holdings considering their order (by UAA) within each original strata constructed by size class, ensuring that each "computational stratum" has at least two holdings. The relation between original strata and 'computational' strata is one to many. The "computational strata" are recorded in a dataset field, additionally to the original strata field.
Choosing between extrapolation factors
When tabulating the data from different modules (e.g CORE, LAFO, RDEV) with distinct scope of coverage for the purpose of establishing time series, it remains a challenging task to choose between the available weighting methods. To guide data producers, we highlight different situations for the case where data is applied from CORE and LAFO module.
To illustrate the case, we show a table with only a subset of holdings of the sampling design and extrapolation factor fields for the case of census for CORE and sample for LAFO.
HLD_ID | EXTPOL _FACT1 _CORE | EXTPOL _FACT2 _CORE | EXTPOL _FACT3 _CORE | STRA _ID_CORE | STRA_ IDF_CORE | PSU _CORE | PSUF _CORE | SSU _CORE | SSUF _CORE | OSU _S1_ CORE | OSU _SF1_ CORE | EXTPOL _FACT1 _LAFO | EXTPOL _FACT2 _LAFO | EXTPOL _FACT3 _LAFO | STRA _ID _LAFO | STRA _IDF _LAFO | PSU _LAFO | PSUF _LAFO | SSU _LAFO | SSUF _LAFO | OSU _S1_LAFO | OSU _SF1 _LAFO |
1 | 1 | _Z | _Z | _Z | _Z | _Z | _Z | _Z | _Z | |||||||||||||
2 | 1 | _Z | _Z | _Z | _Z | _Z | _Z | _Z | _Z | |||||||||||||
3 | 1 | _Z | _Z | _Z | _Z | 1.707865 | 110 | 1 | _Z | _Z | _Z | |||||||||||
4 | 1 | _Z | _Z | _Z | _Z | 1.707865 | 110 | 1 | _Z | _Z | _Z | |||||||||||
5 | 1 | _Z | _Z | _Z | _Z | 1 | 10 | 1 | _Z | _Z | _Z | |||||||||||
6 | 1 | _Z | _Z | _Z | _Z | _Z | _Z | _Z | _Z | |||||||||||||
7 | 1 | _Z | _Z | _Z | _Z | _Z | _Z | _Z | _Z | |||||||||||||
8 | 1 | _Z | _Z | _Z | _Z | _Z | _Z | _Z | _Z | |||||||||||||
9 | 1 | _Z | _Z | _Z | _Z | 1 | 10 | 1 | _Z | _Z | _Z | |||||||||||
10 | 1 | _Z | _Z | _Z | _Z | 1.707865 | 110 | 1 | _Z | _Z | _Z | |||||||||||
11 | 1 | _Z | _Z | _Z | _Z | 2.065753 | 170 | 1 | _Z | _Z | _Z | |||||||||||
12 | 1 | _Z | _Z | _Z | _Z | _Z | _Z | _Z | _Z |
The choice between EXT_CORE and EXT_LAFO depends on "Total variables" and on the "Dimension variables":
Case 1 | Totals of CORE variables broken down by CORE dimensions | Use EXT_CORE for Totals of CORE variables |
Case 2 | Totals of LAFO variables broken down by LAFO dimensions We use only the holdings with LAFO data, meaning EXTPOL_FACT1_LAFO ne “.”. | Use EXT_LAFO for Totals of LAFO variables |
Case 3 | Totals of LAFO variables broken down by CORE dimensions We use only the holdings with LAFO data, meaning EXTPOL_FACT1_LAFO ne “.”. These holdings have both LAFO and CORE data. These are holdings common in CORE and LAFO. | Use EXT_LAFO for Totals of LAFO variables |
Case 4 | Totals of CORE variables broken down by LAFO dimensions We use only the holdings with LAFO data, meaning EXTPOL_FACT1_LAFO ne “.”. These holdings have both LAFO and CORE data. These are holdings common in CORE and LAFO. | Use EXT_LAFO for Totals of CORE variables (if EXT_CORE is used then CORE variables are underestimated, because it considers only the holdings in LAFO) |
Eurostat tables EF_M_FARMANG and EF_M_FARMLEG use Case 1 and Case 3. That is, they use only CORE dimensions.
We use in Eurostat EXT_CORE and EXT_LAFO in order to designate the product of the 3 extrapolation factors of CORE, respectively of LAFO. For the case shown above, as only one extrapolation factor is filled in for each CORE and LAFO, then EXT_CORE= EXTPOL_FACT1_CORE and EXT_LAFO= EXTPOL_FACT1_LAFO
Choosing between coverage (MAIN FRAME plus FRAME EXTENSION or only MAIN FRAME)
Few countries (see 9.7 How to interpret the figures from Eurobase tables) conducted in 2020 the CORE on MAIN FRAME + FRAME EXTENSION, and LAFO on MAIN FRAME only.
As shown in the case above, Cases 2, 3 and 4 cannot be applied on FRAME EXTENSION because in the FRAME EXTENSION there are no holdings with LAFO.
Only CASE 1 can be applied on FRAME EXTENSION.
Therefore, for the comparison 2020 with 2016 or 2023, the time series comparison can only be exerted in a meaningful way for Totals of CORE variables broken down by CORE dimensions.
If dat producers strive to publish a table including besides Case 1, also e.g. Case 3, then it is recommended adding a metadata info that there is a break in time series for Totals of LAFO (In 2020, the Totals of LAFO variables are computed on a reduced scope e.g. on the holdings exceeding at least one of the thresholds from Annex II of Regulation (EU) 2018/1091).