•  A basic survey (Agricultural-Livestock Census), conducted every ten years

•  A sample survey conducted on a two-year basis till 2010 and on a three-year basis since, in the period between Agricultural-Livestock Censuses.

Census surveys were carried out in the years: 1921, 1929, 1950, 1961 and every 10 years since. Sample surveys were carried out in: 1966/67, 1977/1978, 1983 and since then, every 2 years until 2010. From 2010 onward the sample survey is carried out every 3 years.

The purpose of the FSS is to determine the basic structural features of the agricultural and livestock holdings. Data are collected, according to Community legislation, on:

•  General characteristics

•  Utilized agricultural area

•  Livestock

•  Variables of special interest, such as labour force, rural development issues, management and cultivation methods.

The development of the agricultural holdings’ structure constitutes the main element upon which the National and Community policies in the Agricultural Sector are based.

The unit of the survey is the agricultural and/or livestock holding. The sampling frame, which was used for the 2016 FSS, was the updated Register of Agricultural and Livestock Holdings compiled by ELSTAT. The sampling method used is the single random stratified sampling, according to a stratification scheme based on the Regional Unit, the Typology and the Economic size, expressed through the Standard Output, of the holdings.

a) The private short-term contractors were selected from the Enumerators Register of ELSTAT that is built up every 8 months, after a public invitation launched through the mass media. The selection takes place in compliance with a pre-defined system of rules. The applicants fill in and submit their applications online and are ranked through a computerized system, on the basis of clear and objective criteria (e.g. availability/employment, previous experience in statistical works, level of education, etc). The applicants are selected on the basis of the ranking lists.

b) The employees of Municipalities and other public services were selected on the basis of their experience on statistical surveys in the agricultural sector, their knowledge of the territory and the local situation in agriculture as well as their agronomic background.

Upon signing of a contract, both categories of selected enumerators were bound to fulfill their obligations (to complete the questionnaires and to check their quality) within the defined time frame and in the best possible manner. They were also bound to observe the statistical confidentiality procedures of ELSTAT, for all collected data.

ELSTAT provided each enumerator with an “Associate” identification card issued by the Administrative Support Division of ELSTAT. 

GG            Government Gazette

EU             European Union

EC             European Commission

FSS           Farm Structure Survey

UAA          Utilised Agricultural Area

LSU           Livestock Units

OPEKEPE    Payment and Control Agency for Guidance and Guarantee Community Aid

IACS          Integrated Administration and Control System

OCR          Optical Character Recognition

ELSS          Hellenic Statistical System

SCC           Statistical Confidentiality Committee

-    Growing of non-perennial crops

-    Growing of perennial crops

-    Plant propagation

-    Animal production

-    Mixed farming

-    Support activities to agriculture and post-harvest crop activities

a) At least 0.1 ha of utilized land (A_3_1) or at least 0.05 ha of greenhouses (B_1_7_2+B_1_8_2+B_4_7), regardless of the production type, ownership, or the location of the holding, or

b) At least:

- 1 cow (C_2_6+C_2_99) or

- 2 other "large animals" of any type and age (oxen, horses, donkeys, mules) (C_1+C_2_1+C_2_2+C_2_3+C_2_4+C_2_5), or

- 5 "small animals" (sheep (C_3_1), goats (C_3_2), pigs (C_4)) of any age and type, or

- 50 poultry birds (C_5), or

- 50 female rabbits (C_6), or

- 20 hives of “domestic” or “European” bees (C_7) or

- 5 ostriches (C_5_3_4).

c) Cultivates mushrooms (B_6_1)

These holdings which are eligible to be included in the survey are:

- 512 holdings with fallow land and/or permanent grassland no longer in production purposes which is kept in good agricultural and environmental conditions,

- 2 holdings with only "other livestock" not economically valued. These are holdings with hares, deers and wild boars recorded under "other livestock".

- 2 holdings with no agricultural activities. These are two of the 52 ‘special common land units’ (see 8.1.4), namely the units corresponding to the predominantly urbanized NUTS 3 Regional Units: Kentrikos Tomeas Athinon (EL303), Notios Tomeas Athinon (EL304).” 

More specifically:

-  the Structure of Agricultural and Livestock Holdings Section/ Primary Sector Statistics Division had the overall responsibility for the survey and was directly responsible for every step except the ones mentioned below.

-  the Methodology, Analysis and Research Section/ Organization, Methodology and International Relations Division was responsible for steps 1b, 2a, 4e.

-  the Regional Offices carried the burden for step 2f, 3a, 4a.

-  the Applications Development Section/ Informatics Division was responsible for step 5a, 5c. 

The Farm Register is a statistical register generated and updated periodically during the Agricultural Censuses. Furthermore, the Farm Register is updated from administrative sources (OPEKEPE), as well as other surveys conducted by ELSTAT such as the FSS (conducted every three years) and the specialized national annual agricultural surveys.

The Farm Register is a list frame.

The sampling frame for the 2016 FSS was based on the latest available version of the Farm Register. Data from specialized national annual agricultural surveys were compared and crosschecked to those of the Farm Register based on the identification data of the holder.

The sampling units were drawn randomly from the sampling frame. In detail, in each stratum the sample has been selected with equal probabilities by systematic random sampling from the population of holdings belonging to this stratum. The holdings were ordered within strata according to the total Standard Output (SO, in esu) of the holding.

• The Regional Unit of the holdings (74 Regional Units in Greece). The 74 regional units are a national level, in the most part equivalent to, but in some cases more detailed than NUTS 3. For the transition to NUTS 3 (52 regional units) some regional units are merged.
• The particular type of farming according to the technical and economic orientation of holdings
• The economic size of holdings was divided into 6 classes. The economic size has been defined by the total Standard Output (SO) calculated in ESU (1 ESU=1,200 Euro)

Table 1: Classes of holdings’ size determined by their total standard output (SO)

4,878 holdings included in the Register of ELSTAT, whose total SO is equal to zero, were stratified as follows:

• By Regional Unit by size class of holdings: In each Unit the crop holdings were stratified into 9 size classes, according to their size, determined by their area with crops, as follows:

Table 2: Classes of holdings’ size determined by their area with crops

• Holdings with economic size more than 65 ESU (6,084 holdings).
• Holdings breeding ostriches, included in the Register of the ELSTAT (27 holdings).
• Livestock and crop holdings that were included in the Register of Agricultural Holdings of  ELSTAT that had been enriched by OPEKEPE and other  ELSTAT surveys (3,982=3,467 holdings from OPEKEPE + 515 holdings from other ELSTAT surveys) for which their economic size and type of farming were unknown.
  

a)  At regional level (NUTS II), the relative standard error of the size of the arable land of a certain crop characteristic should be less than 5%, when the size of the land of this certain characteristic is greater than 7.5% of the Region’s utilized agricultural area.

b) At regional level (NUTS II), the relative standard error of the capital livestock units of a certain kind of livestock should be less than 5%, when the livestock units of this certain kind of livestock exceeds 7.5% of the total livestock units in the region, under the condition that the certain kind of livestock in the region exceeds 5% of the certain kind of livestock at country level.

where n is the overall sample size in each region (NUTS II), n_h is the sample size at stratum h , N_h is the population (number of holdings) of the stratum  h  and S_h is the standard deviation of the standard output (SO) of the holdings in the stratum h.

For the crop holdings belonging to the Register of the ELSTAT and of the Greek Ministry of Rural Development and Food, in the above formula, the value of S_h is the standard deviation of the arable land of the holdings belonging to the stratum h.

A complementary (reserve) sample was also drawn in order to replace holdings in the following cases:

• When the holding was sold/rented and merged with other holding
• When the holding had been temporarily or permanently closed down
• When the holding changed location and the new location is in a different Municipality.
• When the holder is unknown, could not be reached and/or the sampling unit has been misclassified in the agricultural sector.
• When the holder refused to give the required information (Unit non-response).

All data and information referred herein are used by ELSTAT and the other members of ELSS for the production of official statistics as specified in the Regulation of Statistical Obligations. 

Administrative data specifications are determined by the needs of the pertinent agency. ELSS members are not customarily able to intervene in these procedures; however there is the possibility to do so, usually within the frame of special interagency collaborations. Even in these cases, though, differences in the data characteristics imposed by legal requirements (national and/or EU) are not easy to bridge.

As mentioned in point 4.4 above, direct record matching was not feasible so common land is aggregated at NUTS 3 regional level into special common land units.

The survey questionnaire was designed in such a way so as to satisfy both national and Community needs for statistical information. It covered all variables stipulated in Regulation 1166/2008 which must be analyzed, thus helping drawing the Hellenic agricultural policy.

The questionnaire was designed taking into consideration comments and observations made by the main data users (Ministry of Rural Development and Food, Ministry of Environments and Climate Change), as well as by other Divisions of ELSTAT (Division of Methodology and Organization, Division of Informatics, Division of Statistical Information and Publication, Division of National Accounts).

• Updating the Farm Register so as to have valid contact information
• Contacting the interviewees prior to the actual interview to ensure their presence.
• Training of the enumerators on the personal interview procedure
• Contacting non-respondents by telephone at a later date, and if possible completing the interview, especially for large holdings and holdings belonging to exhaustively surveyed strata.

For cases where the holder refused to provide information, the interviewer had instructions to insist and inform the holder about the Greek Statistical Law that obliges the surveyed person to provide the required statistical information. If the holder continued to refuse to cooperate then the interviewer had to inform the Supervisor in order to decide the proper action to be taken against the holder.

In cases in which it was impossible to collect statistical information from certain sampling units, units included in the complementary sample replaced these cases.

In case of a census 1=3+4+5

Only for sample survey, in which case 2=3+4+5

6,104

This item is a subset of 3.

4>4.1+4.2

5=5.1+5.2

5.1>=5.1.1+5.1.2

6=5.2+5.1.2+4.2

• Data Processing and Validation by the Regional Statistical Offices
• Quality Controls at Regional Unit level by the Central Service

During the Validation phase all values were checked for acceptability and consistency. The estimated gross error rate was 5 errors per questionnaire, including all types of errors from simple misspelling of a postal code or omission to fill-in a total to erroneous values being entered. The data were validated according to the following procedure:

1. Logical and completeness checks of the questionnaires in the Regional Statistical Office, in order to check their correctness and to correct any errors, if necessary. It should be noted that the external enumerators themselves had already performed such kind of checks before submitting the filled in questionnaires to the employees of the Regional Statistical Office.

2. Data entry by means of OCR and correction of the errors due to erroneous reading.

3. Validation of data after a series of checks which identified errors or notifications.

4. Checks for identifying double recordings. The questionnaires were checked in order to identify the holdings that had been enumerated twice.

During the Quality Control phase, even though performed at the Regional Unit level, corrections were attempted at the holding level mostly by identifying abnormally high or low values. Such corrections were relatively seldom. At this stage, some follow-up interviews were also considered necessary, resulting in a number of questionnaires being completed by phone interviews. More specifically the survey data were compared with the results of previous Censuses and previous Structure surveys, as well as with the results of the annual statistical surveys and with data from administrative sources. (Ministry of Rural Development and Food, etc). In case where major inconsistencies were identified for a specific variable, an in-depth study and analysis were carried out in cooperation with the respective Regional Statistical Office and the Ministry of Rural Development and Food.

The ABBYY FlexiCapture 11 software was used for OCR and the preliminary validation of the data. Then the data were exported for further validation in ELSTAT’s database where all software tools used are developed within the Oracle system and are custom made either by the staff of ELSTAT or by external contractors.

Data validation has been performed at all levels, according to the respective time frame. Assistant supervisors and interviewers performed data quality control and initial validation during the data collection period. Supervisors and experienced personnel at the Regional and Central Offices performed the final validation of the data after all data were collected and digitized, whereas specialised staff of the Central Office performed the final quality checks before the data were submitted to Eurostat.

In the design phase of the survey an initial weight (design weight) was given to each sampling unit (holding). This initial weight was estimated as the inverse of the probability of selection. More precisely, for the holding i that belongs to stratum h the initial weight is:

W_h  = N_h / n_h   

where,

N_h: population size according to the data of the Register of Agricultural Holdings

n_h: number of the respondent holdings in stratum h, excluding the extra holdings derived from splitting of other holdings

For the non-response cases, the initial weights were corrected by a factor that takes into account the response rates in each separate stratum. The essence of this correction is to increase the initial weights of the respondents, so that they represent the non-respondents. More specifically, the initial weight  in each stratum h is multiplied by the inverse of the response rate:

r_h = m_h / n_h

where,

m_h: is the number of respondents.

Sampling errors that derive from the application of the one-stage stratified random sampling, and were estimated through the calculation of the coefficients of variation.

Non-sampling errors that derive from any other reasons except sampling and arise during the planning, conducting, processing and final stages of estimation, in all surveys. Non-sampling errors cannot be estimated through the sampled data.

The main sources of errors are:

1. Cases of new holdings that had not been included in the Register of Agricultural and Livestock Holdings, thus creating under-coverage errors.

2. Cases where the Register of Agricultural and Livestock Holdings included holdings that were closed or holdings which had merged and which were identified during the conduct of the survey.

3. Counting errors that were identified and corrected by means of logical checks.

1)  For some holdings (mainly livestock holdings) there seems to be an inconsistency between the SO (provided from Eurostat) based on the Register's data and the SO based on the observed LSU from the survey’s results.

Undercoverage (%) = {(Refusals+Rest of not surveyed holdings)/(Respondents+Refusals+Rest of not surveyed holdings)}*100

Respondents = 85,746 holdings (includes also holdings that derived from splitting of other holdings and holdings that were used from the reserve sample)

Refusals = 2,338 holdings

Overcoverage rate (%)=100*(Closed holdings+Merged holdings+Duplicates)/Gross sample size

Where,

Gross sample size =  101,858 holdings (Holdings in Register + New holdings + Holdings arisen from the division of holdings + Reserve sample)

Closed holdings = 3,226 holdings (Holdings that do not operate permanently + Holdings that do not operate temporarily + out-of-scope holdings)

Merged holdings = 2,725 holdings

The most common problematic questions/characteristics identified during the quality control of the data were the following:

• Household consumption (item A_3_3_1), sometimes reported as "yes" for large holdings,  
• Kitchen gardens vs outdoor fresh vegetables (items B_2 and B_1_7_1_2),
• Permanent grassland vs common land, in some cases difficult to discern.

In the design phase of the survey an initial weight (design weight) was given to each sampling unit (holding), estimated as the inverse probability of selection. The initial weights were corrected by a factor that takes into account the change in sample size imposed by the holders that refused to respond. The essence of this correction is to increase the initial weights of the respondents, so that they represent the non-respondents.

Corrections of this type were not applied to exhaustively surveyed strata. 

During this validation phase all errors identified were corrected using as reference the Agricultural Register, the experience of ELSTAT's personnel and common sense. The estimated gross error rate was 5 errors per questionnaire, including all types of errors from simple misspelling of a postal code or omission to fill-in a total to erroneous values being entered.

During both the Validation and Quality control phases, corrections and/or completions deemed necessary were performed, in order of preference, according to:

•  the data already in the questionnaire (i.e. completion of missing totals),
•  logical conjecture based on the experience of the handler (mostly for minor errors),
•  telephone contact with the interviewee (mostly for holdings of a significant size).

There are detailed manuals describing the various tests and control procedures and only ELSTAT employees with sufficient experience are authorised to participate in the data Validation and Quality control process.

There are no differences between the national and EU definitions.

ELSTAT adopted the 3rd method as it has been converted into the 2nd one. So, according to the method used common land is reported as assigned to 52 special 'common land units' which represent the 52 NUTS 3 regions of the country. 

In the FSS data transmitted to Eurostat records referring to this type of units were marked as 'common land units'  in the field on the legal personality of the holding (A_2$holdingtype) and as 'common land' in the type of tenure for the utilised agricultural area (A_3_1_4$ha).

Field A08 of these holdings (holding identification number) begins with “99999” followed with the corresponding NUTS3 code.

The common land data were obtained from the Payment and Control Agency for Guidance and Guarantee Community Aid (OPEKEPE) which, in turn, has collected the data from the applicants for the Community Aid (farm holders), under its competence as the Integrated Administration and Control System (IACS) operator.

In both cases common land was reported using 52 special units, one for each NUTS 3 region of the country.

In cases where large variations –depending on the variable– were detected, an in-depth analysis was carried out in close cooperation with the regional offices and the Ministry of Rural Development and Food.

The results exhibit partial coherence at the NUTS3 level, with the Livestock and the Crop Production Statistical Surveys.

Concerning Crop Production Statistics (ACS) the following general remarks need to be considered:

ACS is conducted by the Ministry of Rural Development and Food (MRDF) according to Regulation (EC) No 543/2009. The methodology described in the Regulation differs from that of FSS. Examples of methodological differences include:

-    ACS data are not collected through a survey, but are obtained primarily from the regional Directorates of Agricultural Economy and Veterinary, of the country and are based on experts’ opinions. This raises the issue of accuracy/subjectivity of the estimations.

-    In ACS, utilized agricultural area (UAA) is counted more than once in the case of successive crops, leading to higher values of the related variables being reported by the ACS compared to the FSS.

-    Data validation and cross-checking with external sources, namely ELSTAT-FSS and OPEKEPE-IACS, is reported by the MRDF, however the relevant procedures are not documented and the results are not provided in the metadata.

-    In the FSS survey there is threshold of 0.1 hectare for crops, as opposed to the ACS where there is no threshold.

The definitions of the examined variables, according to the respective EC regulations, differ between the ACS and the FSS. Some examples are:

-    The FSS includes communal pastures in the UAA whereas ACS doesn’t.

-    The FSS includes Fallow land in the Arable Land, whilst ACS does not.

-    The ACS includes fodder plants in Pulses, while FSS considers these plants as a separate category

-    Areas reported in ACS do not include young cultivations that are not yet productive, whereas in FSS all cultivated land is reported.

-    Minor differences can be noted in many sectors such as Pistachio trees not being included in ACS, mushroom cultivations being included in UAA whereas FSS treats them as a separate category, etc.

Concerning Livestock Statistics conducted also by ELSTAT, the design of both surveys, FSS and Livestock Statistics, was based on the total number of animals, in each main animal category, and not on the subcategories within the main categories of animals. Therefore, the total number of animals is more or less consistent in the two surveys (bovine 11,9% more in FSS, sheep 5,8% less in FSS, goats 8,9% less in FSS, and pigs 3,5% more in FSS), while larger differences are only observed in the subcategories . In addition,FSS 2016 has covered a number of holdings coming from OPEKEPE which have not been covered by the Livestock Statistics.  In the case of bovine, for example, half of the difference in total bovine animals is due to these OPEKEPE holdings, with the overall difference falling to 5%, should these be removed.

Results were published on the website of ELSTAT (free of charge) in the form of detailed tables (national series of tables) accompanied by the relevant metadata files.

A news release was published on the 26th of April 2018.

The results were released in the second half of 2018 (free of charge). 

a) an appropriate request together with a detailed research proposal in conformity with current scientific standards have been submitted;

b) the research proposal indicates in sufficient detail the set of data to be accessed, the methods of analyzing them, and the time needed for the research;

c) a contract specifying the conditions for access, the obligations of the researchers, the measures for respecting the confidentiality of statistical data and the sanctions in case of breach of these obligations has been signed by the individual researcher, by his/her institution, or by the organization commissioning the research, as the case may be, and by ELSTAT.

Users can request access to microdata by submitting an application to the Hellenic Statistical Authority, Statistical Information and Publications Division, 46, Pireos & Eponiton Str, P.O.Box 80847, GR-18510, Piraeus (tel (30)213-1352022, FAX: (30)213-1352312, e-mail: data.dissem@statistics.gr.

• Regulation (EC) No 1166/2008 of the European Parliament and of the Council, of 19 November 2008, on farm structure surveys and the survey on agricultural production methods (http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32008R1166&from=EN)
•  Regulation (EU) No 715/2014 of 26 June 2014 amending Annex III to Regulation (EC) No 1166/2008 of the European Parliament and of the Council on farm structure surveys and the survey on agricultural production methods, as regards the list of characteristics to be collected in the farm structure survey 2016 (http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014R0715&from=EN)
• Regulation (EC) 2015/1391 of 13 August 2015 amending Regulation (EC) No. 1200/2009 of the  the European Parliament and of the Council on farm structure surveys and the survey on agricultural production methods, as regards livestock unit coefficients and definitions of the characteristics. (http://eur-lex.europa.eu/legal-content/LV/ALL/?uri=CELEX%3A32015R1391
• Regulation (EC) No 223/2009 of the European Parliament and of the Council, of 11 March 2009, on European statistics and repealing Regulation (EC, Euratom) No 1101/2008 of the European Parliament and of the Council on the transmission of data subject to statistical confidentiality to the Statistical Office of the European Communities (http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32009R0223&from=EN
• European Statistics Code of Practice (http://ec.europa.eu/eurostat/web/quality/european-statistics-code-of-practice)
• Quality Assurance Framework of the European Statistical System (http://ec.europa.eu/eurostat/documents/64157/4392716/qaf_2012-en.pdf/8bcff303-68da-43d9-aa7d-325a5bf7fb42)
• ESS Quality Glossary (https://ec.europa.eu/eurostat/documents/3859598/6651706/KS-GQ-15-003-EN-N.pdf)
• ESS Handbook for Quality Reports (http://ec.europa.eu/eurostat/documents/3859598/6651706/KS-GQ-15-003-EN-N.pdf)
• Principles Governing International Statistical Activities (http://unstats.un.org/unsd/methods/statorg/Principles_stat_activities/principles_stat_activities.pdf)

2.      Cochran, W.G. (1977). Sampling Techniques, New York: John Wiley and Sons

3.      Dalenious T., and Hodges, J.L (1959). Minimum variance stratification. JASA, 54,88-101

4.      Deming, W.E. (1953). On a probability mechanism and the bias of non-response. JASA, 48, 743-772 

5.      Evans, W.D. (1951). On stratification and optimum allocation. JASA, 46, 95-104

6.      Hansen, M.H., Hurwitz, W.N., Madow, W.G. (1953). Sample Survey Methods and Theory. Vol. I, New York: John Wiley and Sons

7.      Hess, I, Sethi, V.K., and Balakrishnan,T.R (1966). Stratification: A practical investigation. JASA, 61, 74-90

8.      Holt, D., and Elliot, D. (1991). Methods of weighting for unit non-response. The Statistician, 40, 333-342

9.      Kalton, G (1983). Models in the Practice of Survey Sampling. International Statistical Review, 51, 175-188  

10.   Kalton, G. and Kasprzyk, D. (1986). The Treatment of Missing Survey Data. Survey Methodology, 12, 1-16.

11.   Kalton, G. and Flores – Gervantes, I. (2003). Weighting Methods. Journal of Official Statistics, 19, 81-97.

12.   Kish, L., (1965). Survey Sampling, New York: John Wiley and Sons

13.   Kish, L., and Frankel, M.R. (1974). Inference from complex samples. Journal of the Royal Statistical Society, A, 139,80-95

14.   Kish, L., and Anderson, D.W. (1978). Multivariate and multipurpose stratification. JASA, 73, 24-34

15.   Kish, L., (1987). Statistical Design and Research, New York: John Wiley and Sons

16.   Kish, L., (1988). Multipurpose Sample Designs. Survey Methodology, 14, 19-32

17.   Kish, L., (1989). Sampling Methods for Agricultural Surveys, Rome: Food and Agricultural Organization of the United Nations

18.   Kish, L., (1992). Weighting for Unequal Pj. Journal of Official Statistics, 8, 183-200

19.   Kish, L (1995).Questions/Answers (1978-1994), Paris: INSEE, International Association of Survey Statisticians 

20.   Little, R.J.A. (1982). Models for non-response in sample surveys. JASA, 77, 237-250

21.   Little, R.J.A. (1986). Survey non-response adjustments for estimates of means. International Statistical Review, 54, 139-157 

22.   Madow, L. H (1946). Systematic sampling and its relation to other sampling designs. JASA, 41, 207-214

23.   Murthy, M.N. and Rao, J.T.(1988). Systematic sampling with illustrative examples. In Handbook of Statistics, Vol. 6, (Eds. P.R. Krishnaiah and C.R. Rao). Amsterdam: Elsevier Science, 147-185

24.   Royall, R.M and Herson, H.J. (1973). Robust Estimation in Finite Populations I.  JASA, 68, 880-889 

25.   Royall, R.M and Herson, H.J. (1973). Robust Estimation in Finite Populations II: Stratification on a Size Variable. JASA, 68, 890-893 

26.   Sarndal, Swensson, and Wretman (1992). Model Assisted Survey Sampling. New York: Springer-Verlag

27.   Thomsen, I. (1977). On the effect of stratification when two stratifying variables are used. JASA, 72, 149-153 

• Summary quality report for users, Farm Survey Structure, Year 2016
• Single Integrated Metadata Structure (SIMS), Farm Survey Structure, 2016, Greece

SIMS is the format in which metadata are made available through the ELSTAT web page. It is quite similar to the present NMR (in ESQRS format). The Summary quality report is a 2-page summary of the basic points in the SIMS report.

The NMR is not made directly available to the public through the ELSTAT web page (the SIMS report is provided instead). However, the NMR, after validation, could also be made available after submitting an application to: http://www.statistics.gr/en/provision-of-statistical-data”

More precisely ELSTAT disseminates the statistics in compliance with the statistical principles of the European Statistics Code of Practice and in particular with the principle of statistical confidentiality: http://www.statistics.gr/en/statistical-confidentiality?inheritRedirect=true

• for sample survey data, aggregation of micro-data to a minimum level of NUTS3,
• primary cell suppression on the aggregated data, using minimum frequency rules, according to the recommendations of the SCC, and
• secondary cell suppression with full singleton handling.

The above procedures are implemented using the τ-argus software.