6. Accuracy and reliability |
Top |
|
|
6.1. Accuracy - overall |
Main sources of error |
Main sources of error are sampling errors, over-coverage, non-response and measurement errors. |
|
6.2. Sampling error |
Method used for estimation of relative standard errors (RSEs) |
The method for estimation of RSEs was SAS PROC SURVEYMEANS procedure. We calculated standard errors and coefficients of variation, by using general SAS programs that are used in most of the SURS surveys. |
|
6.2.1. Sampling error - indicators |
1. Relative standard errors (RSEs) - in annex
2. Reasons for possible cases where precision requirements are applicable and estimated RSEs are above the thresholds |
The coefficients of variation are provided in annex. We meet all precision requirenments stipulated in Annex IV "Precision Requirements" of the Regulation 1166/2008. |
Annexes: 6.2.1-1.Sampling error |
6.3. Non-sampling error |
See below |
6.3.1. Coverage error |
1. Under-coverage errors |
The probability of under-coverage in the FSS is very low since there are not many new agricultural holdings. All important new farms are included in administrative registers and were consequently included into the list. All new agricultural holdings from administrative sources were added just before the survey started. |
2. Over-coverage errors |
We updated the Statistical Register of Agricultural Holdings (we excluded ineligible family farms from the frame, which turned out that they didn’t belong to the target population). Overcoverage is adjusted with the help of the extrapolator factor. Weighting factors were calculated on the basis of eligibility status of agricultural holdings, with the formula (responses + nonresponses)/responses - on the level of strata. |
2.1 Multiple listings |
There were less than five agricultural holdings listed twice in the FSS 2016. They were treated as ineligible. |
3. Misclassification errors |
There are no significant misclassification errors. Only major misclassifications were modified before extrapolation factor was calculated. In the responses, we found some units obviously belonging to different (bigger) size class and for them we updated the strata assignment and adapted weighting factors. |
4. Contact errors |
The initial list of agricultural holdings in FSS 2016 was 15 592 agricultural holdings. There were 1,2% of all agricultural holdings without telephone number. There were altogether 14% of not contacted farms (the holder could not be reached - there was no telephone number or no answer). The non-contacting of the person which could give response has been taken into account when calculating extrapolation factor. |
5. Other relevant information, if any |
Not available. |
|
6.3.1.1. Over-coverage - rate |
Over-coverage - rate |
The initial list of agricultural holdings in FSS 2016 was 15 592 agricultural holdings. The share of units that were included in the frame and it turned out that they didn’t belong to the target population was 2.7%. |
|
6.3.1.2. Common units - proportion |
If the data was available in administrative register, then the data was not collected from agricultural holdings. |
6.3.2. Measurement error |
Characteristics that caused high measurement errors |
We are aware of measurement errors and we try to avoid this kind of errors by training interviewers, supervisors, by data checking and validation process. Where inconsistency or extreme values were discovered, the data were checked with possible administrative data or there was also a “call-back” to the farmers, and the data were checked again. So extreme values of variables were checked and corrected if necessary. Since the data was inserted directly into the data entry program (controls were included), there was likely to have less mistakes caused by interviewer.
Eurofarm variable |
Variable describtion |
Difficulties |
A_3_3_1 |
More than 50% of production self-consumed by the holder |
Very difficult to assess for farmers - subjective estimation. |
A_3_3_2 |
More than 50% of sales are direct sales |
Very difficult to assess for farmers - subjective estimation. |
B_5_3 |
Other land |
Respondents' inability to provide accurate answers. |
E_1_x |
Farm work for each of the persons (AWU) |
Sensitivity of the characteristic and subjective estimation. |
F_2_1 |
Importance of other gainful activities directly related to the holding |
Sensitivity of the characteristic. Also quite difficult to assess for farmers - subjective estimation. |
M_6_5_1 |
Broadcast application of manure with no incorporation |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_6_5_2 |
Broadcast application of manure with incorporation within 4 hours |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_6_5_3 |
Broadcast application of manure with incorporation after 4 hours |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_6_6_1 |
Bandspread application of manure with trailing hose |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_6_6_2 |
Bandspread application of manure with trailing shoe |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_6_7_1 |
Injection of manure on a shallow or open slot |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_6_7_2 |
Injection of manure on a deep or closed slot |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_1_1 |
Tillage: conventional |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_1_2 |
Tillage: conservation |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_1_3 |
Tillage: zero |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_2_1_1 |
Soil cover: normal winter crop |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_2_1_2 |
Soil cover: cover or intermediate crop |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_2_1_3 |
Soil cover: plant residues |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_2_1_4 |
Soil cover: bare soil |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
M_2_1_5 |
Outdoor arable land areas which are covered by multi-annual plants |
New variable, introduced in 2016 and since it is not based on the area of crop (reported in IACS), it is quite difficult to assess for farmers. |
|
|
6.3.3. Non response error |
1. Unit non-response: reasons, analysis and treatment |
Unit non-response was treated with re-weighting and imputation. The main reasons for non-response were the following:
- holders consider themselves as “non-agricultural holding”,
- dissatisfaction with the current agricultural policy in Slovenia,
- problems with unsolved ownership (official procedures regarding succession can be very long),
- general refusal because of low economic conditions of living.
Agricultural enterprises: According to the National Programme of Statistical Surveys, reporting of data is obligatory for the enterprises (and voluntary for family farms). Weighting factors were calculated with the formula (responses+nonresponses)/responses - on the level of strata. The distribution of non-response across holdings' categories was checked and no significant discrepancies were found. |
2. Item non-response: characteristics, reasons and treatment |
In the process of data validation, we considered national rules as well as validation rules for EUROFARM. There were no specific units discovered which had not responded to a particular item. |
|
6.3.3.1. Unit non-response - rate |
Unit non-response - rate |
If the response rate is considered as the share of response among all eligible family farms, then the response rate is 65%. The non-response rate is thus 35%. |
|
6.3.3.2. Item non-response - rate |
Item non-response - rate |
In the process of data validation, we considered national rules as well as validation rules for EUROFARM. There were no specific units discovered which had not responded to a particular item. Since we had a computer assisted telephone interview, all questions had to be answered, otherwise the application did not allow to go further. |
|
6.3.4. Processing error |
1. Imputation methods |
- Method of logical imputations (if some values were inconsistent with other values (we discovered there was clearly a typing error), we imputed the values with the “Method of logical imputations”).
- Hot deck method (if we had only some data from administrative registers and no data for some variables (from logical point of view), then we used the “Hot deck method” to get the data from similar farms (same UAA, same region, etc.)).
- Structural hot deck method (if we had data from administrative data only for totals, then we used the “Structural hot deck method” to get all the subcategories. The proportions were taken from similar farms (same UAA, same region, etc.)).
- Method of cut average (if the data were missing (from logical point of view), there was a possibility to input the mean value within a given variable (e.g. intra-regional or intra-county), whereby a certain percentage of the maximum and minimum values are removed from the average computation).
|
2. Other sources of processing errors |
FSS was conducted as computer assisted telephone interview. The application had some controls already implemented in the data entry application. After the telephone interview we made corrections and imputations based on administrative sources and logical controls. Final extrapolation factor/weight is a product of sampling weight, non-response weight and calibration weight. The descriptions of imputations were written (established) by methodologists in the Department for Agriculture, Forestry, Fishery and Hunting (SURS). They were based on national rules, validation rules in Eurofarm and different calculations. The actual imputation was also made in SURS, in the Department for General Methodology and Standards. |
3. Tools used and people/organisations authorised to make corrections |
In the process of data validation, we considered national rules as well as validation rules for EUROFARM. Validations and imputations were done by SAS. |
|
6.3.4.1. Imputation - rate |
Imputation - rate |
The data set relating to labour force and gainful activities on agricultural holdings is methodologically complex. We therefore believe that for an adequate level of data quality it is not enough to put direct questions prescribed by Regulation into the questionnaire. For this reason we included more detailed and explicit questions into the questionnaire in order to obtain high-quality basic information on which further calculations of Eurofarm variables are based. It would therefore be incorrect for this set of variables to calculate imputed value of the shares of the Eurofarm variables in the same manner as for other variables which are collected directly from the data sources (primary or administrative). The range of imputed shares is from 0% to 30%, depending on the single variable. Values taken from administrative data are not counted as imputed values. 58 agricultural holdings were imputed.
DESCRIPTION OF THE CODE OR SECTION |
RATIO OF THE CORRECTED VALUE (in %) |
COMMENT |
Legal personality of the holding |
0 |
|
Data on organic farming |
0 |
Data is taken directly from administrative source. |
More than 50% of production self-consumed by the holder |
22 |
Very difficult to assess for farmers - subjective estimation. |
More than 50% of sales are direct sales |
8 |
Very difficult to assess for farmers - subjective estimation. |
Production animals |
0 |
The number of livestock was gathered from the agricultural holdings or from administrative sources (it is not counted as imputed value). Variables on livestock sent to Eurostat were imputed (corrected) in less than 1%, except the distribution of cows (dairy cows and other cows). The total of cows is gathered from the administrative source, but the distribution into dairy cows and other cow was imputed (ratio of cca. 10%). |
Land section |
0 |
The area of land was gathered directly from the agricultural holdings or from administrative sources (it is not counted as imputed value). Most variables on land section sent to Eurostat were not imputed (or were imputed in less than 1% of the total), except the ones listed bellow. |
Potatoes |
less than 1% |
|
Peas, field beans and sweet lupines |
less than 1% |
|
Fodder roots and brassicas |
1% |
|
Fresh vegetables, melons, strawberries |
less than 1% |
|
Flowers |
less than 1% |
|
Total irrigable area |
less than 1% |
|
Farm work non-family members non-regularly employed |
less than 1% |
|
Data on support for rural development |
0 |
Data is taken directly from administrative source. |
Turnover from other gainful activity |
20 |
Very difficult to assess for farmers - subjective estimation. |
|
|
6.3.5. Model assumption error |
[Not requested] |
6.4. Seasonal adjustment |
[Not requested] |
6.5. Data revision - policy |
Data revision - policy |
SURS has an internal revision policy. It is available on: http://www.stat.si/dokument/5151/Navodilo_revizije.pdf All publications and revisions, due to provisional data, are planned. Data for FSS are firstly published as “provisional data” due to users' needs for timely information. The data became final 15 months after the publication of provisional data. |
|
6.6. Data revision - practice |
Data revision - practice |
Provisional data for FSS were published 29.9.2016 (2 months and a half after the survey was conducted). Final data (without typology and economic size) was published on 29.6.2017. Final data on typology and economic size was published on 20.12.2017. |
|
6.6.1. Data revision - average size |
The differences in the key statistics was less than 1%. |