Farm Structure Survey 2009/2010 Survey on agricultural production methods 2009/2010

Transcription

1 Farm Structure Survey 2009/2010 Survey on agricultural production metods 2009/2010 National Metodological Report (NMR) According to Art. 12 of Regulation (EC) No 1166/2008 of te European Parliament and of te Council of 19 November 2008 publised in te Official Journal of te European Union L 321, p.14 of 1 December 2008 Member State: GREECE ELSTAT. - Hellenic Statistical Autority

2 FARM STRUCTURE SURVEY 2009/2010 SURVEY ON AGRICULTURAL PRODUCTION METHODS 2009/2010 NATIONAL METHODOLOGICAL REPORT CONTENTS SUMMARY CONTACTS SURVEY METHODOLOGY NATIONAL LEGISLATION CHARACTERISTICS AND REFERENCE PERIOD SURVEY ORGANISATION CALENDAR (OVERVIEW OF WORK PROGRESS)... 8 PHASE 1: ORGANIZATION AND PREPARATION OF THE SURVEY CALENDAR SURVEY DESIGN SAMPLING, DATA COLLECTION AND DATA ENTRY Drawing te sample for SAPM Data collection and data entry Use of administrative data sources SPECIFIC TOPICS Common Land Geograpical reference of te olding Volume of water used for irrigation Oter issues RESPONSE-BURDEN POLICY Estimation and sampling errors for SAPM Estimation and sampling errors for SAPM Symbolisms: Estimation process Metods for andling missing or incorrect data items. Control of te data EVALUATION OF RESULTS PUBLICATION AND PUNCTUALITY TIMELINESS AND PUNCTUALITY CONFIDENTIALITY AND SECURITY

3 SUMMARY Te Farm Structure Survey (FSS) is a survey of national interest, wic is carried out bot as a sample survey and as a census, in order to collect objective quantitative information relating to te structure of te farming sector. Te Hellenic Statistical Autority (ELSTAT) carried out te first sample survey of te Structure of Agricultural and Livestock oldings in 1966/67, wen Greece was still an associated member of te EU. Te next sample survey took place in 1977/78. After te accession of te country to te EU furter surveys were carried, out every two years from 1983 till today. Every ten years an exaustive survey (Basic FSS or Agricultural Census) is carried out. Te first Agricultural Census was conducted in 1950, after te Second World War. Since 1950 five censuses of agriculture and livestock farming ave been eld, in 1961,1971,1981,1991 and 1999/2000. From 1961 to 1991 censuses were conducted simultaneously wit te General Population and Housing Census. Te Agricultural Census of 1991 was te last census carried out at te same time wit te General Censuses for Population, Houseolds etc. Te Agricultural Censuses of 1999/2000 and 2009/2010 were carried out before te General Population Censuses of 2001 and 2011, respectively. Te purpose of FSS is to determine te basic structural features of te agricultural and livestock oldings, wic encapsulate te agricultural picture of Greece at te specific time. Te developments of te agricultural oldings structure constitute te main element for te National and Community policy drawing up in te Agricultural Sector. Terefore, te collection of objective and reliable data is absolutely necessary in order to draw up time series tables concerning te oldings caracteristics. During te last decade FSS took place as a sample survey in years 2003, 2005 and 2007 and as an exaustive survey (census) in year Te FSS 2009 was conducted simultaneously wit te sample Survey on Agricultural Production Metods 2009 (SAPM 2009), wic was carried out for te first time. Wereas te aim of FSS is to determine te basic structural features of te agricultural and livestock oldings, te aim of SAPM is to explore more tecnical aspects of te agricultural activity in Greece. Te Hellenic Statistical Autority (ELSTAT.) is te responsible body for te surveys implementation. Some of te Central Office responsibilities are te surveys organization and preparation, te tabulation and presentation of te results and finally te information dissemination. Te data collection and process are carried out by te Regional Statistical Offices of te 51 districts. Te data collection is carried out by interviewers, selected by te Regional Statistical Offices and recommended to ELSTAT. for appointment. On average, 100 agricultural units correspond to eac interviewer. Te Regional Statistical Offices, in carge of te surveys, supervise te activities of te interviewers. Te FSS 2009 and SAPM 2009 were planned for te end of year 2009 (October December 2009) but were finally carried out during te period July to October 2010 due to decisions made by te competent Minister. 3

4 Te processing of te data was carried out te period from 1st November 2010 to 31st Marc Te reference period for te FSS data in respect of crops, labour force and oter items, as well as for te SAPM 2009, was te cultivation period from 1 October 2008 to 30 September Te reference date for te FSS data, in respect of livestock, was 1 November 2009, and for Rural development was te last tree (3) years of te reference year Te public announcements for te agricultural census made troug te media and press (newspapers, radio, TV broadcasts), as well as te announcements released by te Municipalities underlined tat te reference period of te census was October 2008 to September Furtermore, te data collection as been made via personal interview wit te farm older. During te interview te interviewer made clear tat te data e ad to record concerned te reference period and not te period wen te interview was taking place. FSS 2009 was an exaustive survey so every agricultural, livestock or mixed olding of te Statistical Farm Register was surveyed. Te Farm Register includes oldings, te older of wic made use of: at least 0.1 a of utilized land or at least 0.05 a of greenouses, te olding s own animals, namely: one (1) or more cows or two (2) or more oter "large animals" of any type and age (oxen, orses, donkeys, mules), or five (5) or more "small animals" (seep, goats, pigs) of any age and type, or fifty (50) or more poultry birds, or twenty (20) or more ives of domestic or European bees or five (5) or more ostrices. SAPM 2009 was a sample survey wit te same tresold as for te agricultural census. Te sampling frame used was te Statistical Farm Register. Te bodies involved in te FSS 2009 and SAPM 2009 were: 1 Working Group at ELSTAT, 2 senior supervisors (two senior officials of te Central Office of ELSTAT), 52 supervisors at te 51 Regional Statistical Offices (Heads of te Regional Statistical Offices), 235 assistant supervisors at te 51 Regional Statistical Offices (officials in te Regional Statistical Offices), interviewers and 105 accompanying (interpreters) interviewers (in order to assist te main interviewers in areas were language problems existed). (ANNEX IV Decisions designating private interviewers) 550 local statistical correspondents for te common land questionnaires (witout carge) 1 expert to contribute to te survey design, processing, tabulation design and publication. Data was collected by a personal interview wit te older of te farm. Te filled in questionnaires were collected and cecked by te assistant supervisors. Te assistant supervisors cecked te questionnaires and deliver tem to te supervisors wo coordinated te wole work at te district of teir responsibility. 4

5 ELSTAT s personnel carried out te scanning of te questionnaires for te OCR (Optical Caracter Recognition), as well as te processing and correction pase. Te responsible department, at central level, carried out te quality controls. Te FSS and SAPM results were compared wit data coming from previous FSS surveys as well as data coming from oter sources (e.g. special annual agricultural surveys, administrative data etc). 1. CONTACTS Contact organisation Contact organisation unit Contact name Contact person function Contact mail address Contact address Hellenic Statistical Autority ELSTAT. Primary Sector Statistics Division Lemonia Dionysopoulou, Head of Primary Sector Statistics Division Metodology, database management, dissemination, etc. Pireos 46 & Eponiton, 18510, Pireas lemdiony@statistics.gr, primary1@statistics.gr Contact pone number SURVEY METHODOLOGY 2.1 National legislation ELSTAT is an independent autority acting under te supervision of te Greek Parliament. Te main statute concerning te ELSTAT is Law 3832/ In addition, a joint decision including matters relating to te proclamation as well as te approval of te surveys implementation and te duty delegation of te surveys to te responsible unit togeter wit details of implementation and processing is issued by te Ministers of Economic Affairs and Finance and te co-responsible Ministers. Furtermore, te joint decision sets out te time scedule, te organization and te cost of te surveys. Te above-mentioned national legislation deals wit te scope and te coverage of FSS and SAPM, assigns ELSTAT te responsibility for te surveys, determines te obligations of te respondents wit respect to te census and identification, as well as te protection and te obligations of enumerators. In addition, it includes administrative and financial provisions and provisions relevant to te rigt of access to administrative data. As far as te metodology for bot FSS and SAPM is concerned, ELSTAT fully complies wit te EU legislation. 2.2 Caracteristics and reference period Te caracteristics of te FSS 2009 and SAPM 2009 fully comply wit te EU Regulation and more specifically ver.7 of te Handbook on implementing te FSS and SAPM. Tere are no 5

6 caracteristics tat are surveyed only for national purposes, or caracteristics tat deviate from EU list or caracteristics not collected. In addition, tere are no canges of definitions of caracteristics and/or reference time and/or measurement affecting te comparability wit previous survey data. Te reference period for te FSS data in respect of crops, labour force and oter items, as well as for te SAPM, was te cultivation period from 1 October 2008 to 30 September Te reference date for te FSS data in respect of livestock was 1 November For RD was te last tree (3) years of te reference year. A copy, in Greek and in Englis, of bot questionnaires (FSS & SAPM) is provided in te Annex. 2.3 Survey organisation Te Hellenic Statistical Autority (ELSTAT) is responsible for te FSS and SAPM, and more precisely te Structure of Agricultural and Livestock Holdings Statistics Section of te Primary Sector Statistics Division. In particular, tis Section is responsible for te overall planning, organization, supervision and conduct of te surveys, as well as te processing and publication of te survey results, in collaboration wit oter co-responsible sections of ELSTAT, suc as te Divisions of Organization and Metodology, Informatics, Statistical Information and Publications, Administrative Support and Financial Administration. Te conduction and processing of te survey was decentralized and was in te ands of te Regional Statistical Offices of ELSTAT at 51 prefectures (nomi). Te bodies involved in te FSS 2009 and SAPM 2009 were: - 1 Working Group at te ELSTAT, - 2 senior supervisors, - 52 supervisors at te 51 Regional Statistical Offices, assistant supervisors at te 51 Regional Statistical Offices, interviewers and 105 accompanying (interpreters) interviewers (in order to assist te main interviewers in areas were language problems existed) (ANNEX IV Decisions designating private interviewers) 550 local statistical correspondents for te common land questionnaires - 1 expert to contribute to te survey design, processing, tabulation design and publication. Te composition, te responsibilities and te functions of te bodies involved in te survey were defined as follows: 1. Working Group at ELSTAT: In connection wit te organization, conduction and processing of te FSS and SAPM, a working group was set up and operated in te Central Office of ELSTAT. Te task of te Working Group was te effective planning and coordination of all te work relating to te organisation and conduction of te survey and te processing of te results. 6

7 2. Senior supervisors: Te senior supervisors were two senior officials of te Central Office of te ELSTAT, Division of Primary Sector Statistics. More precisely, tey were te ead of te Division and te Head of te Structure of Agricultural and Livestock Holdings Statistics Section. Teir task was to organize and monitor all kinds of operations of te surveys, train te supervisors, supervise and coordinate teir work, monitor te organization, conduct te surveys in te prefectures, and deal wit any potential problems. 3. Supervisors: Te supervisors were mainly te eads of te Regional Statistical Offices, wo ave long experience in coordinating and implementing te structural surveys and te agricultural census, as well as oter statistical surveys. Teir task was to organize and complete, witin te prescribed time limits, te necessary work for preparing and conducting te surveys, in te area of teir responsibility (prefecture). More particularly, tey were responsible for: Informing all local-government bodies, public services, organizations and te public about te survey, Allocating all municipalities and rural districts in te prefecture to assistant supervisors, Selecting, on te basis of merit, te interviewers and recommending teir appointment by te ELSTAT, Training assistant supervisors and interviewers, Assigning work to te interviewers, Monitoring and coordinating te work of te assistant supervisors and interviewers trougout te conduct of te surveys and providing tem wit instructions and every possible assistance, and Supervising and taking responsibility for te successful conduct of te surveys, togeter wit te collection, cecking and processing of te questionnaires for teir district. 4. Assistant supervisors: Te assistant supervisors were mainly officials in te Regional Statistical Offices wo ave long experience to implement structural surveys, agricultural census and oter agricultural surveys as well as oter statistical surveys. Tey were responsible for assisting te supervisors in te work of organizing, conducting and processing te surveys, as described above. 5. Interviewers: Te interviewers were private collaborators, mainly unemployed, students and agriculturalists. Interviewers were selected by te Regional Statistical Offices, wic recommended teir appointment to ELSTAT. 7

8 Te interviewers were selected on te basis of teir experience on statistical surveys in te agricultural sector, teir knowledge of te territory and te local situation in agriculture as well as teir agronomic background. Te task of te interviewers was to complete te questionnaires and to ceck teir quality. 6. Accompanying interpreters: Te accompanying interpreters were also private collaborators. Teir task was to assist te interviewers in completing te questionnaires in areas were language problems existed. 7. Experts: Furtermore, oter ELSTAT s staff and experts outside ELSTAT were appointed for specific tasks of te survey and contributed to particular stages of te survey as: Survey design Survey processing IT application for OCR and data entry, automatic controls and programming development. In addition, a pilot study, funded by Eurostat (Grant Agreement No ), was carried out in collaboration wit te Agricultural University of Atens, as well as wit te National Institute of Agricultural Economics of Italy (INEA). Its scope was to provide a model for te estimation of te volume of water used for irrigation in agriculture and a pilot survey was conducted, in tree different prefectures, to support it. For more information, see te Final Report of te pilot study. 2.4 Calendar (overview of work progress) Te Farm Structure Survey 2009 and te SAPM 2009 were conducted in te period from July 2010 to October Te processing of te data was carried out in te period from November 2010 to Marc Te multiple operations for te FSS and SAPM, more particularly te preparatory work, te actual surveys taking and te post-survey work, were carried out in four pases, as detailed below: Pase 1: Organization and preparation of te survey. Te first pase comprised te organization activities and preparatory work for te surveys, more precisely te following actions were carried out: Farm registers update from Organic Farming and New Farmers register of Ministry of Rural Development and Food, Survey design, SAPM sample design, Design of questionnaires, manual of instructions and oter auxiliary documents, 8

9 Analysis, design and implementation of te IT application for OCR, data entry and automatic controls, Development of te database applications, Development of Eurofarm file and control tables, Appointment of senior supervisors, supervisors and assistant supervisors, Training of te above staff, in training centers assigned for te supervisors by te relevant Section of te Central Office of te ELSTAT and for te assistant supervisors by te supervisors, Delivery to te supervisors of te questionnaire instructions and oter auxiliary documents, Situating of te supervisors and assistant supervisors in teir posts, Contacts between supervisors and respective prefects and familiarization of te Regional Statistical Offices and all public services in te prefectures wit te purpose of te survey and te manner of conducting it, Division of eac supervision area into zones of responsibility for te supervisors and teir assistants, Selection and appointment of te interviewers for te conduct of te surveys, Selection and appointment of accompanying interpreters to assist interviewers in completing te questionnaires in areas were language problems existed. Updating in pase 1, during te preparation of te census. Te basic farm register tat was used for te 2009 census was te register from te 1999 census as tis ad been updated from te FSS surveys of 2003, 2005 and 2007, to a certain degree, and te special annual agricultural surveys (orcard survey, survey on areas under vine, survey on cereals production, survey on crop production oter tan cereals, survey on pigs livestock, survey on cattle livestock, survey on seep livestock, survey on goats livestock). Also for te update of te register in 2007, ELSTAT collaborated wit te Ministry of Rural Development and Food. In tis framework, ELSTAT made use of te specific registers of te Ministry of Rural Development and Food pertaining only to New Farmers (measure of te Programme for te Rural Development of Greece, wic is co-financed by te European Union by virtue of Council Regulation (EC) No 1698/2005 and Commission Regulation (EC) No 1274/2006) and Organic Farming, tat is it compared tose registers wit ELSTAT s basic register. No oter Registers eiter from te Ministry of Rural Development and Food were used, suc as registers for olive trees and te vineyards, because tere was no consistency wit te definition of te agricultural olding, or from any oter Administrative source. Te data from te basic Register and te two oter registers of te Ministry of Rural Development and Food (tat of te New Farmers and Organic Farmig), were compared and crosscecked on te basis of te identification data of te older. Tere were cases were te registers of te Ministry of Rural Development and Food were not fully complete, as some data were missing, suc as te date of birt date, te tax registration number, etc., and so tese could not be matced. Tose not matced cases wit te basic ELSTAT s register, were kept separately in two temporary file registers i.e., one for te new farmers and one for organic farming. Tat was te procedure to update te basic Farm Register at te preparatory pase. 9

10 Pase 2: Data collection Te second pase comprised te main work for conducting te surveys. In te course of tis pase te following operations were carried out: Training of interviewers, Allocation of te interviewers in teir sectors, distribution to tem of te questionnaires and te lists from te Farm Register wit te units and te sample units (for SAPM) and oter necessary documents, Conduct of te surveys (collection of statistical data), and monitoring and supervision of te operation from beginning to end by te assistant supervisors and supervisors, Ceck of te questionnaires, Delivery of te questionnaires by te interviewers to teir assistant supervisor, Collection of te questionnaires tat ad been cecked by te assistant supervisors by te supervisors of te questionnaires. Updating during te Census a) In te cases were some of te necessary caracteristics for te identification of te olding was still missing, making te identification impossible, te supervisors collaborated wit te departments of Agriculture located in te Prefectures for: a. obtaining te necessary missing information for te oldings. On tat basis, ELSTAT s basic register was updated wit te new information b. clarifying weter te oldings were in operation or not. b) During te census, te interviewers used apart from te basic register, te two temporary file registers (regarding te new farmers and te organic farming ). If tey found out a olding tat was in te basic register as well as in one of te temporary file register ten tey filled in te questionnaire for te olding of te basic register making a noteon te temporary file register indicating tat tese oldings ad been merged. Pase 3: Data processing (After te census) Te following operations were carried out: Scanning of te questionnaires, OCR, online processing (logical controls, consistency controls and automatic controls) by te ELSTAT personnel and creation of a database containing te survey data, Quality controls of te data by te Regional Statistical Offices, at NUTS III level, (te quality controls can be found in ANNEX IX Validation rules: error ) Validation in te Central Office of te data from te prefectures, Automatic controls of te data in te database at central level, (te quality controls can be found in ANNEX IX Validation rules: error 01) ANNEX VII Quality controls on te survey data (te quality controls can be found in ANNEX IX Validation rules: error ) 10

11 Processing of te data in accordance wit te Eurofarm programme-typology of oldings and creation of te Eurofarm file of individual data. Continuous cecking wit Farm Register and correcting according Updating after te Census (during pase 3) a) Te oldings of te temporary file registers tat were not included in te basic register and found to be in operation were added to te basic register. As it is said before a questionnaire ad been filled in during te census. b) Te quality cecks identified cases of duplication. (te quality controls can be found in ANNEX IX Validation rules : error ) c) Te quality cecks identified oldings tat did not fulfill te classification criteria of a olding, tus tey were deleted from te basic register. (te quality controls can be found in ANNEX IX Validation rules : error 01) Oter potential sources for updating te Basic Register In 1996 a micro-census took place before te 1999 census. For te 2009 census suc a microcensus was not conducted, someting tat could ave been used as an updating tool. However, a) Beside te effort undertaken to update te census register, some oldings were identified wic sould be in te register as operating oldings and tey were not or oldings wic were in te register at te time of te census wile tey sould not be included because tey were closed for a number of reasons suc as, te older was too old or e ad canged occupation not economic efficiency of te olding, merging, cange of land use (land to be developed), cultivations wic were burnt, etc. b) Oter critical issues are te sources wic are used for te update: for example te structural surveys, even toug tey take place every two years, tey are sample surveys (10% sample size). Terefore te updating tat takes place refers only to te sample size as no imputations are made pertaining to updating issues and consequently te register cannot be fully updated. c) ELSTAT as not yet developed a system for automatic and continuous update of te register, for instance, by linking its register wit te register of oter public services or government ministries suc as te register of te Ministry of Agricultural Development and Food and te register of OPEKEPE, etc. It sould be mentioned tat developing a system for automatic and continuous update is amongst ELSTAT s priorities and a relevant project as already been envisaged. d) Te discrepancies and differences identified before and after te census in te registers of ELSTAT, i.e. te register from te 2007 survey and te register from te 2009 census pinpoint issues pertaining to te quality of te applied metods for updating te register Pase 4: Evaluation of te results-publication and Dissemination At te final pase te following operations were/will be carried out: Qualitative analysis and documentation of te results, Production of national tables wit te final results, Preparation of press releases and a publication wit te final results. 11

12 CALENDAR Survey pases Unit in carge Period Draft design and oter preliminary work Central Office / Agricultural Unit Survey design And Metodology/IT April 2009 June 2009 /Information and Publication Unit Questionnaire September 2008 May 2009 Manuals of instructions for survey conduct and September 2008 May 2009 processing. Sample design, sample selection Central Office / Metodology Unit Sample design April 2009 July 2009 Sample selection August 2009 September 2009 Production of Survey materials Central Office / Agricultural Unit Production of questionnaires June 2009 September 2009 Field work materials June 2009 September 2009 Advertising posters June 2009 September 2009 General computer programming IT Unit IT application for data entry and automatic controls September 2009 December 2010 Development of te database applications November 2010 Marc 2011 Development of Eurofarm file and control tables May 2011 July 2011 Coordination, support, control and monitoring by Agricultural Unit Central Office / Agricultural Unit Training session for survey Supervisors and assistant October 2009 supervisors Training courses for Interviewers July 2010 August 2010 Distribution of materials to te regional statistical June October 2010 offices Guidance during data collection July 2010 October 2010 Survey collection Regional Offices Field work July 2010 October 2010 Survey processing Central Office / Agricultural Unit and IT Unit Logical cecks and cecks on te completeness of October 2010 June 2011 te questionnaires Questionnaire Optical Reading Data entry and April September 2011 automatic controls Corrections on Optical Reading Data (data validation, May February 2012 verification) Data input in Oracle database October February 2012 Quality controls at NUTS III level January 2012-Marc 2012 Evaluation of te results-publication and Dissemination Central Office / Agricultural Unit and Metodology Unit Validation in te Central Service of te data from te February 2012-Marc 2012 prefectures Automatic controls of te data in data base at Central February 2012-Marc 2012 level Quality controls on te survey data February 2012-Marc 2012 Processing of te data in accordance wit te January Marc 2012 Eurofarm programme-typology of oldings and creation of te Eurofarm file of individual data Qualitative analysis and documentation of te results And IT unit Marc 2012 Production of national tables wit te final results Marc September 2012 Preparation of press releases and a publication wit And Metodology/Information June September 2012 te final results and Publication Unit 12

13 2.5 Population and frame Te definition of agricultural olding in te Farm Register is consistent wit te definition stated in Regulation 1166/2008 article 2.a. 571/88 article 5 a. It sould be noted tat te definition of agricultural olding is exactly te same wit te definition stipulated in Regulation 571/88 (ANNEX VII page 12-13). Te survey was conducted in all districts of Greece and te target population is all te agricultural, livestock or mixed oldings, te olders of wic made use of: a) at least one (1) stremma (0.1 a) of utilized agricultural area or at least alf a stremma (0.05 a) of greenouses, regardless of te type of crop, te ownersip of te land or te location, or b) te olding s own animals, namely: one (1) or more cows or two (2) or more oter "large animals" of any type and age (oxen, orses, donkeys, mules), or five (5) or more "small animals" (seep, goats, pigs) of any age and type, or fifty (50) or more poultry birds, or twenty (20) or more ives of domestic or European bees or five (5) or more ostrices. Te Sampling Frame, wic was used in tis survey, was te updated Register of Agricultural Holdings of ELSTAT (Updating in pase 1-During te preparation of te census and pase 2- During te census of te Item 2.4. Calendar) Te total number of te sampling frame accounts to oldings ( oldings from te basic Register of ELSTAT and oldings from te Registers of te Greek Ministry of Rural Development and Food) for te agricultural census and for te SAMP survey. ELSTAT made use of te registers of te Ministry of Rural Development and Food only concerning te New Farmers and Organic Farming and it compared tose registers wit te register of ELSTAT. Afterwards, te data of te registers were compared and crosscecked on te basis of te identification data of older. However, tere were some cases were te registers of te Ministry were not complete and some of te data were missing, suc as te date of birt, te tax registration number, etc. Cases tat could not be matced wit ELSTAT s farm register were kept separately in two temporary file-registers. During te conduct of te agricultural-livestock census, tese temporary file-registers were made available to te interviewers togeter wit te basic register. 2.6 Survey design Te Sampling Frame is te updated basic Register of ELSTAT as mentioned above. (Updating in pase 1-During te preparation of te census and pase 2 During te census) Te Farm Structure Survey 2009 was carried out as a census in accordance wit te EU legislation (EC) No 1166/2008, and te SAMP survey as a sample survey. Te total number of te sampling frame accounts to oldings ( units from te Register of te 13

14 ELSTAT and units from te Register of te Greek Ministry of Rural Development and Food) for te agricultural census and for te SAMP survey. Te oldings of te Ministry of Rural Development and Food, wic entered in te sampling frame of te FSS year 2007, were 33,783. From tese, only 5,538 oldings were selected in te sample of te FSS, year As a result, te register of ELSTAT was only updated, wit regard to te 5,538 oldings, after te completion of te specific survey. For te rest of te oldings of te Ministry, tere as been an examination for duplicated or closed oldings. After te completion of te Agricultural Census, te remaining (not duplicated and open oldings) ave been verified to be Tese oldings did not ave te necessary auxiliary variable for te determination of teir size. As a result, te design of te study SAMP was based on te register of ELSTAT tat included all te necessary variables. To sum up, we did not select sample from te oldings of te Ministry, for te survey SAMP. However, during te production of te results, tese oldings were used for te calculation of te extrapolation factors, since te necessary information became available from te Agricultural Census. In addition, te register, between 2007 and 2008, was updated only by te sample data of te annual surveys on livestock and crop capital statistics. It is to our belief tat te fact of no sample selection from te oldings for te SAMP creates bias. However, tis bias is almost negligible, because in te weigting process for estimating te results of SAMP, we took under consideration all te oldings tat tey were finally collected from te Agricultural Census (bot ELSTAT and Ministry of Rural Development and Food registers). Tere were oldings tat were discovered and recorded during te field work. Tese new oldings were not included in any register (ELSTAT or Ministry), because tey were unknown. Tat is te reason for not being included in te sample of SAMP. Data was collected by personal interview wit te older of te farm. 2.7 Sampling, data collection and data entry Drawing te sample for SAPM Te sampling metod tat ELSTAT applied for te conduct of SAPM 2009 was te one-stage stratified random sampling wit sampling unit te agricultural, livestock or mixed olding belonging to te target population. Te initial sample size amounts to oldings (sampling fraction=7,3%). Te decision for determining te sample size was based on financial criteria and on several precision criteria as follows: 14

15 a. At regional level (NUTS 2), te relative standard error of te size of te arable land of a certain crop caracteristic sould be less tan 10%, wen te size of te land of tis certain caracteristic is greater tan 10% of te Region s utilized agricultural area. b. At regional level (NUTS 2), te relative standard error of te capital livestock units of a certain kind of livestock sould be less tan 10% (annex IV Precision requirements), wen te capital livestock units of tis certain kind of livestock exceed 10% of te total capital livestock units in te region, under te condition tat te capital livestock units in te Region exceeds 5% of te total capital livestock units (country level). Te sample of oldings was selected from te Register of Agricultural Holdings of ELSTAT, wit reference year According to tis sampling sceme (one-stage stratified random sampling) and for oldings included in te above Register of ELSTAT, te strata were created by te combination of te following stratification criteria: 1. NUTS 3 (54 areas in Greece - 50 Departments and Department of Attiki, wic is divided into 4 areas). 2. Te economic size of oldings (6 classes). Te Economic Size as been defined by te Standard Gross Margin (SGM) calculated in ESU (1 ESU=1.200 Euro). Table 1: Classes of Economic Size Class Boundaries 1 Less tan 2 ESU 2 From 2 to less tan 5 ESU 3 From 5 to less tan 10 ESU 4 From 10 to less tan 19 ESU 5 From 19 to less tan 38 ESU 6 Greater tan or equal to 38 ESU 7 Economic size is not specified 3. Te following categorization of te general type of farming, according to te tecnical and economic orientation of oldings: Table 2: General Type of Farming Serial number Code of type Type of Farming 15

16 1 T10 T1 2 T21 T201 3 T22 T202+T203 4 T30 T3 5 T41 T41 6 T44 T42 7 T51 T51 8 T52 T52 9 T53 T53 10 T60 T6+T7+T8+T0 11 T90 BIO 1 12 T00 Not specified type Not specified eiter te type or te economic size Te above codes stand for te following general types of farming: a) Holdings belonging to te Register of Agricultural Holdings of ELSTAT (Codes of types T10- T90) - Specialist field crops (T10) - Specialist market garden vegetables (T21) - Specialist flowers and ornamentals (T22) - Specialist permanent crops (T30) (Specialist vineyards, specialist fruit and citrus fruit, specialist olives). - Specialist bovine animals (T41) - Specialist seep and goats (T44) - Specialist pigs (T51) - Specialist poultry (T52) - Specialist pigs and poultry (T53) - Mixed cropping (T60) 1 BIO: Holdings tat follow organic farming. Altoug tis group is not a specific class of organic farming on te typology, owever it is considered a specific design domain wit omogeneous population and tere no overlapping wit te rest of te domains. 16

17 (Mixed crops- livestock) - Specialist organic farming (T90) - New crop oldings tat neiter te type of farming nor te economic size are specified (10) - New oldings tat te type of farming is not specified (T00) 4. Te crop oldings belonging to te types T00 (new oldings tat are included in te Register of Agricultural Holdings of ELSTAT), were stratified as follows: By Region (NUTS 3) By size class of oldings: In eac Region (NUTS 3), te crop oldings were stratified into 9 size classes, according to teir size, determined by teir area wit crops, as follows: Table 3: Classes of oldings size determined by teir area wit known type of crops Class Code of Area wit crops in ectares class 1 11 Less tan 1 ectare 2 12 From 1 to less tan 2 ectares 3 13 From 2 to less tan 3 ectares 4 14 From 3 to less tan 5 ectares 5 15 From 5 to less tan 10 ectares 6 16 From 10 to less tan 20 ectares 7 17 From 20 to less tan 30 ectares 8 18 From 30 to less tan 50 ectares 9 19 Greater tan or equal to 50 ectares Te following categories of oldings ave been surveyed exaustively: Holdings wit economic size more tan 38 ESU (9.061 oldings) class 6 from table 1. Crop oldings (code type T00) included in te Register of Agricultural Holdings of ELSTAT, for wic teir classes of economic size and types of farming are not specified and teir size class is greater tan or equal to 50 ectares (30 oldings) class 9 from table3. Livestock oldings (size class 07 from table 1) tat included in te Register of Agricultural Holdings of ELSTAT (164 oldings), for wic eiter teir classes of economic size or types of farming were not specified. Te variable used for te construction of size classes, te size class boundaries and te number of classes were determined as follows: 17

18 Te ideal variable used for te creation of size classes of oldings belonging to te Register of Agricultural Holdings of ELSTAT is te standard gross margin (SGM) y of te oldings, as te value of y in combination wit te type of farming is igly correlated wit all te survey caracteristics. If we could stratify te oldings by te value of y in Regions and type of farming, tere would be no overlap between strata, and te variance witin strata would be muc smaller tan te overall variance, particularly if tere are many strata. Given te number of strata, in order to determine te best size class boundaries under Neyman allocation, te CUM f (y) rule was applied. Taking into consideration te frequency distribution of y, it was found tat te CUM f (y) rule creates te same stratum boundaries wit te Dalenious-Hodges rule, wic is rougly equivalent to making W S constant (W is te weigt of te size class, S is te standard deviation of y in te size class, = 1,2,...,6 ). Te question relevant to a decision about te number of size classes is at wat rate does te variance of Y st decrease as L (number of size classes) increases? ( Y st is te estimated value of y in stratified sampling, given te sampling size). So, applying te CUM f (y) rule, te oldings were stratified into L=4 to 8 strata, and subsequently, given te sample size in eac separate case, te variance ( Y V ) of st Y st was calculated. As L increased, te values of ( Y V ) were decreased. As very little reduction in variance appeared beyond L=6, we decided st tat te ideal number of te size classes sould be equal to 6. In eac separate Region (NUTS 2), te sample belonging to sampling strata was allocated to te strata following te Neyman allocation. More specifically, te following formula was used for te allocation of te sample units in eac separate stratum: n = n 08 N N were n is te overall sample size in eac Region (NUTS 2), is te sample size at stratum, n 08 N S 08 S is te population (number of oldings of te year 2008) of te stratum and S is te standard deviation of te standard gross margin (SGM) of te oldings in te stratum. For te new crop oldings belonging to te Register of Agricultural Holdings of ELSTAT (Type T00), were te type of farming and te economic size are not specified (T00), in te above formula, te value of S is te standard deviation of te arable land of te oldings belonging to te stratum. Te following table 4 depicts te distribution of te sample of oldings, wose teir types are available (Types T10 T90)., 18

19 Type of farming Table 4: Distribution of te sample of oldings Total Classes of economic size of te oldings 0 * ** T T T T T T T T T T T Total * Holdings tat follow organic farming and wose economic size is not specified ** Holdings tat te economic size is not specified Te following table 5 depicts te distribution of te sample of oldings, wosetypes and economic sizes are not available (Types T00 and 10). Table 5: Distribution of te sample of oldings by size classes Code of class Area wit crops in ectares Sample size 11 Less tan 1 ectare From 1 to less tan 2 ectares From 2 to less tan 3 ectares From 3 to less tan 5 ectares From 5 to less tan 10 ectares From 10 to less tan 20 ectares From 20 to less tan 30 ectares From 30 to less tan 50 ectares Equal to or greater tan 50 ectares 35 7 Not specified area Not specified area 163 Total 1302 Te number of respondent sampling units is oldings (response rate = 71,9%), wic is te 6,3% of te total number of te agricultural and livestock oldings all over te Country, according to Census data. From te non respondents (6857 oldings), te 43,0% is out of scope units (closed oldings, merged oldings etc) and te rest are refusals. Corrective measures for non-response of te SAMP were taken during te process of compiling te extrapolation factors for te estimation of te survey caracteristics. 19

20 Drawing te Sample a) Te sample of oldings was selected from te Register of te basic register of ELSTAT, wit reference year Tis Register was compiled using te data of te 1999/2000 census and it was updated wit te data of te sample units of te Farm Structure Surveys of te years 2003, 2005 and 2007, as well as te Livestock and Crop Capital Surveys conducted by ELSTAT till te year Concerning te new oldings, te Register was updated using te Register of te Greek Ministry of Rural Development and Food. More specifically, te Register of ELSTAT was updated wit regard to te sample oldings of FSS 2007 using te Register of te Ministry. Regarding te economic size of te oldings, tis was specified by data of te base year 1999/2000, and it was updated wit te data of sampling units from te Farm Structure Surveys of te years 2003, 2005 and b) Regarding te specialist organic farming, all oldings of tis type were surveyed on a census basis in te Farm Structure Survey of te year 2007, aiming at te identification of te economic size of tese oldings. Consequently, ELSTAT ad te necessary information to create te strata of specialist organic farming using te criterion of te economic size of te oldings. Most of te organic oldings, coming from te Ministry, were surveyed on a census basis in te FSS 2007 and as a result tey were included in te Register of ELSTAT. c) Te oldings of te Ministry of Rural Development and Food entered te surveyed population in FSS of year 2007 and reaced 33,783. From tese, 5,538 oldings were selected in te sample of te FSS, year As a result, te register of ELSTAT was updated after te completion of te specific survey. NUTS2 regions wit more tan oldings Crop caracteristics: Precision requirements Field codes NUTS2 regions Number of oldings in te NUTS2 region UAA, a of te NUTS2 region Area of cereals in a in te NUTS2 region A_3_1 B_1_ , , , , , , , ,1 % Cereals in te UAA of te NUTS2 region 46,5 52,6 58,8 46,1 Area of potatoes and sugar beet in B_1_3 + a in te NUTS2 region B_1_ , , , ,1 % potatoes and sugar beet in te UAA of te NUTS2 region 2,4 1,9 1,4 0,5 B_1_6_4 + B_1_6_5 + B_1_6_6 + Area of oilseed crops in a in te NUTS2 region B_1_6_7 + B_1_6_ , ,8 689,3 198,8 % oilseed crops in te UAA of te NUTS2 region 6,7 1,4 0,3 0,1 20

21 Area of permanent outdoor crops in a in te NUTS2 region B_4 - B_4_ , , , ,4 % permanent outdoor crops in te UAA of te NUTS2 region 5,8 14,4 3,7 12,1 Area of fres vegetables, melons, strawberries, flowers in a in te NUTS2 region B_1_7 + B_1_ , ,5 937, ,7 % fres vegetables, melons, strawberries, flowers in te UAA of te NUTS2 region 2,0 1,7 0,4 1,8 Area of temporary grass and permanent grassland in a in te NUTS2 region B_1_9_1 + B_ , , , ,8 % temporary grass and permanent grassland in te UAA of te NUTS2 region 8,6 11,2 19,6 12,6 NUTS2 regions wit more tan oldings Crop caracteristics: Precision requirements Field codes NUTS2 regions Number of oldings in te NUTS2 region UAA, a of te NUTS2 region Area of cereals in a in te NUTS2 region A_3_1 B_1_ , , , ,0 9489, , , ,7 % Cereals in te UAA of te NUTS2 region 9,1 3,2 20,3 26,5 Area of potatoes and sugar beet in B_1_3 + a in te NUTS2 region B_1_4 551,8 544, , ,9 % potatoes and sugar beet in te UAA of te NUTS2 region 0,5 0,7 1,7 0,7 B_1_6_4 + B_1_6_5 + B_1_6_6 + Area of oilseed crops in a in te NUTS2 region B_1_6_7 + B_1_6_8 46, ,8 104,0 % oilseed crops in te UAA of te NUTS2 region 0,1 0 0,1 0,0 Area of permanent outdoor crops in a in te NUTS2 region B_4 - B_4_ , , , ,7 % permanent outdoor crops in te UAA of te NUTS2 region 27,7 49,5 37,5 26,4 Area of fres vegetables, melons, strawberries, flowers in a in te NUTS2 region B_1_7 + B_1_8 638,6 392, , ,2 % fres vegetables, melons, strawberries, flowers in te UAA of te NUTS2 region 0,6 0,5 3,2 2,4 Area of temporary grass and permanent grassland in a in te NUTS2 region B_1_9_1 + B_ , , , ,8 21

22 % temporary grass and permanent grassland in te UAA of te NUTS2 region 52,9 40,0 27,2 25,2 NUTS2 regions wit more tan oldings Crop caracteristics: Precision requirements Field codes NUTS2 regions Number of oldings in te NUTS2 region UAA, a of te NUTS2 region Area of cereals in a in te NUTS2 region A_3_1 B_1_ , , , , , , , , , ,5 % Cereals in te UAA of te NUTS2 region 5,5 11,8 7,3 6,1 1,0 Area of potatoes and sugar beet B_1_3 + in a in te NUTS2 region B_1_ ,0 148,3 284, , ,2 % potatoes and sugar beet in te UAA of te NUTS2 region 0,6 0,3 0,2 1,1 0,4 B_1_6_4 + B_1_6_5 + B_1_6_6 + Area of oilseed crops in a in te NUTS2 region B_1_6_7 + B_1_6_8 124,8 20,1 427,4 0,9 0,4 % oilseed crops in te UAA of te NUTS2 region 0,04 0,04 0,3 0,0 0,0 Area of permanent outdoor crops in a in te NUTS2 region B_4 - B_4_ , , , , ,4 % permanent outdoor crops in te UAA of te NUTS2 region 69,6 63,7 36,8 20,5 41,0 Area of fres vegetables, melons, strawberries, flowers in a in te NUTS2 region B_1_7 + B_1_ , ,9 434, , ,6 % fres vegetables, melons, strawberries, flowers in te UAA of te NUTS2 region 1,2 5,0 0,3 1,2 0,9 Area of temporary grass and permanent grassland in a in te NUTS2 region B_1_9_1 + B_ , , , , ,7 % temporary grass and permanent grassland in te UAA of te NUTS2 region 17,1 10,0 52,7 49,7 54,5 Livestock caracteristics: NUTS2 regions Precision requirements Field codes LSU in te NUTS2 region , , , ,2 22

23 Bovine animals (all ages) Seep and goats (all ages) Pigs Poultry Number of Bovine animals in te NUTS2 region, in LSU C_2_1*0.4 + C_2_2*0.7 + C_2_3*0.7 + C_2_4 +C_2_5*0.8 + C_2_6 + C_2_99* , , ,3 % of te LSU in te NUTS2 region 36,9 35,5 31,7 26,1 % of national sare of bovine animals in LSU 16,4 29,5 6,3 16,3 Number of Seep and goats in te NUTS2 region, in LSU C_3_1*0.1 + C_3_2* , % of te LSU in te NUTS2 region 48,5 34,7 57,6 52,7 % of national sare of seep and goats in LSU 7,9 10,6 4,2 12,0 Number of Pigs in te NUTS2 region, in LSU C_4_1* C_4_2*0.5 + C_4_99* , ,3 5940, ,3 % of te LSU in te NUTS2 region 9,4 9,1 6,1 16,4 % of national sare of pigs in LSU 8,3 15,3 2,4 20,5 Number of Poultry in te NUTS2 region, in LSU C_5_1* C_5_2* C_5_3* , ,9 4417, ,7 % of te LSU in te NUTS2 region 5,2 20,6 4,5 4,7 % of national sare of poultry in LSU 3,4 25,2 1,3 4,3 Livestock caracteristics: NUTS2 regions Precision requirements Field codes Bovine animals (all ages) Seep and goats (all ages) Pigs LSU in te NUTS2 region , , , ,8 C_2_1*0.4 + C_2_2*0.7 + C_2_3*0.7 + C_2_4 Number of Bovine +C_2_5*0.8 + animals in te NUTS2 C_2_6 + region, in LSU C_2_99* ,7 4195, , % of te LSU in te NUTS2 region 17,5 12,5 14,8 12,1 % of national sare of bovine animals in LSU 9,2 0,9 8,7 4,1 Number of Seep and goats in te NUTS2 region, in LSU C_3_1*0.1 + C_3_2* , ,6 % of te LSU in te NUTS2 region 39,0 72,7 67,5 55,4 % of national sare of seep and goats in LSU 7,5 1,8 14,6 6,9 Number of Pigs in te NUTS2 region, in LSU C_4_1* C_4_2*0.5 + C_4_99* ,6 1092, ,2 % of te LSU in te NUTS2 region 11,6 3,2 11,5 16,1 23

24 Poultry % of national sare of pigs in LSU 12,2 0,4 13,6 11,0 C_5_1* Number of Poultry in C_5_2*0.014 te NUTS2 region, in + LSU C_5_3* ,4 3648, , ,5 % of te LSU in te NUTS2 region 31,8 10,9 6,0 16,2 % of national sare of poultry in LSU 24,6 1,1 5,2 8,1 Livestock caracteristics: Precision requirements Field codes LSU in te NUTS2 region , , , , ,7 C_2_1*0.4 + C_2_2*0.7 + C_2_3*0.7 + C_2_4 Number of Bovine +C_2_5*0.8 + animals in te NUTS2 C_2_6 + region, in LSU C_2_99* ,6 3562,8 8535, , % of te LSU in te NUTS2 region 8,9 5,2 12,7 22,3 0,7 % of national sare of bovine animals in LSU 2,7 0,7 1,7 3,0 0,4 Bovine animals (all ages) Seep and goats (all ages) Pigs Poultry Number of Seep and goats in te NUTS2 region, in LSU C_3_1*0.1 + C_3_2* , , , % of te LSU in te NUTS2 region 68,1 17,5 79,3 63,7 89,2 % of national sare of seep and goats in LSU 7,7 0,9 4,0 3,1 18,8 Number of Pigs in te NUTS2 region, in LSU C_4_1* C_4_2*0.5 + C_4_99* ,8 1477,3 2020,5 4167, ,7 % of te LSU in te NUTS2 region 13,4 2,2 3,0 6,4 4,1 % of national sare of pigs in LSU 8,3 0,6 0,8 1,7 4,7 Number of Poultry in te NUTS2 region, in LSU C_5_1* C_5_2* C_5_3* , ,8 3347,4 4857, ,8 % of te LSU in te NUTS2 region 9,4 74,7 5,0 7,4 5,6 % of national sare of poultry in LSU 4,2 15,3 1,0 1,5 4, Data collection and data entry Data collection Te data collection of te FSS and SAPM was carried out troug personal interviews wit te farm olders. 24

25 Te Regional Statistical Offices were responsible for te data collection. Te ead of eac Office was in carge of organizing and coordinating te wole work of te survey in te particular prefecture. A team of officials of te regional statistical offices (assistant supervisors) assisted te supervisor. Te supervisor and te assistant supervisors trained te interviewers, assigned te units to tem (approximately 100 units per interviewer) and supervised teir work. Prior to te interview date, wenever possible, te interviewers ad a first contact wit te farmers in order to arrange te interview date. Te interviews generally took place at te older s residence, altoug some interviews were conducted at municipality offices. Te interviewer conducted te interviews and completed te questionnaires wit data supplied by te older. Te completion time per questionnaire was approximately 30 minutes. In te case of te older s absence, te interviewer ad to make a second visit or to obtain te required information from anoter person, able to give accurate information about te olding i.e. a member of older s family, or an employee of te olding (e.g. foreman). If a sample unit was found split in two or more oldings te interviewer sould fill in a questionnaire for eac new olding, oter tan te one included in te original sample unit, reporting te new status of te previous olding. Te interviewer ad to report to is/er assistant supervisor every week about te process of is/ers work and to deliver te completed questionnaires. Te assistant supervisors gatered te completed questionnaire in order to ceck te quality of te data collected. If te completed questionnaires did not fulfill te requirements of te survey tey sould be returned to te interviewer to correct tem. Data entry Te data entry was done almost exclusively by OCR and only in some special cases, were it was impossible to scan te questionnaires, by entering te data manually Use of administrative data sources Administrative sources were used a) for quality controls of te results of FSS and SAPM survey (compare teir results wit special annual agricultural surveys, data from te Ministry of Rural Development and Food etc), and b) for te updating of te basic register. Tus, all te necessary data, included in te census questionnaires, were collected troug te census interviews and not taken from administrative files. For instance, data suc as equipment used for renewable energy production are collected from te census and not from administrative sources. 25

26 2.8 Specific topics Common Land Common Lands in Greece are usually roug grazing of permanent grassland used as pasture for cattle, seep and goat. Arable land and permanent crops are not part of Common Lands. Common land is te area tat used jointly by several oldings and it is not possible to assign a specific section to eac farmer. In line wit te decision of te 21-23September 2009 FSS WG meeting, common land sould be recorded using one of te tree recommended propositions. ELSTAT adopted te 3rd metod (Handbook on implementing te FSSS and SAMP definitions FSS WG eld in September 2011) tat indicates te most relevant geograpic level (e.g. NUTS III), of te total area of common permanent grassland. Te data for te common land collected troug te census survey using a special questionnaire (ANNEX VII). Using a dedicated questionnaire collected census survey data for te common land. Te unit was te local district t(lau 2). Te questionnaires were filled in at te level of Communal Department (LAU 2) by te statistical correspondents in te Municipalities in cooperation wit te staff of ELSTAT s Regional Statistical Offices. Regarding only te item Common Land, te questionnaires were filled in by te local statistical correspondents (ELSTAT collaborates wit tem for te Annual Agricultural Survey) wit te collaboration of te staff of te Regional Statistical Office. Te unit was te local district (LAU 2). At te end, te results of te common land survey (regarding te permanent grassland area) were compared wit te data from te Annual Statistical Survey. On te basis of te Annual Agricultural Statistical Survey we ave aggregated data on te total areas of grassland. Neverteless, tese data are not broken down by type of grassland (permanent meadows-roug meadows), tey do not specify weter tese are used for grazing or not and finally tese data are not broken down by teir tenure status. Tat is wy te results of te common land survey togeter wit te results of te agricultural census were compared wit te results of te Annual Agricultural Survey. Altoug te results tat were transmitted are compliant wit te Regulation requirements and tey are accepted, Eurostat suggested tat te results sould be transformed according to te second metod and be transmitted as special records. Tis suggestion was made in order for te results to be able to be publised. Following Eurostat s suggestions, we are going to modify te transmission mode of te data- as some oter countries, wic ad adopted te same metod, did- and we are going to send tese records in te next days. Results: Common land area ,53 a Geograpical reference of te olding Te current situation in Greece is te following: 26

27 o Te National Cadastral Register is not yet finalized so it is impossible to use it for te geo-reference of te olding. o Te Ministry of Rural Development and Food and its supervised organizations keep various registers tat are not yet completed, as far as te location of te olding is concerned. Neverteless, even wen tey will be completed, tey will not ave te appropriate format for ELSTAT to use, as tere is a difference in te definition of agricultural oldings between te Ministry of Rural Development and Food and ELSTAT. So, bot te National Cadastral Register and te various agricultural registers of Ministry of Rural Development and Food and its supervised organizations are valuable administrative sources tat could be used in te future, for example for te FSS of 2013, but it is impossible to be used for te Census of Facing tese issues, ELSTAT (NSSG at tat time) asked te Commission to supply more information, tecnical assistance and any oter support on te relevant subject. For tis reason, a Eurostat expert, Mrs Marjo Kasanko, visited ELSTAT in 19/2/2008 and te above-mentioned problems were discussed. After te visit tere was a written consultation between Eurostat and ELSTAT referring to alternative metods of providing data on te location of te olding. Eurostat, after studying our data of te minimum, maximum and average size of te local departments, te total number of local department for eac NUTS 3 area in Greece and te number of localities wit area more tat a in eac local department, suggested te following intermediate solution: ELSTAT will ave to provide Eurostat wit te geograpic coordinates of te central points of te locality, were te farm is located. Tis suggestion was accepted by ELSTAT and for te Census of 2009 Greece provided te coordinates of te locality were eac olding is located instead of te coordinates of te olding itself. For tese reason we used te National Geodetic Reference System (Greece 87) EPSG Wit tis system identify te ead quarter in te case of legal person or older s residence in te case of natural person. Locality is a subdivision of LAU2. Eac LAU2 consists of one or more settlements or localities. Tere are data concerning latitude and longitude for eac locality code. Tere are different localities and tere are oldings in of tem (ANNEX VI) Volume of water used for irrigation ELSTAT received a Grant Agreement titled Development of a geograpic information system for te estimation of irrigation demands at farm level in order to produce a model by wic te volume of water used for irrigation could be estimated. ELSTAT collaborated wit te Agricultural University of Atens, as well as wit te National Institute of Agricultural Economics of Italy (INEA). Te project analyzed te different approaces found in international literature on tis issue and ten, after conducting a pilot survey, finalized and proposed a model based metodology. Special attention was given in te Greek particular features of te agricultural sector suc as te large number of smalloldings, te fragmentation of te oldings, te polymorpism of te oldings from te standpoint of production branc. 27

28 Te starting point of te project was an in dept analysis of te existing literature and metodologies from national, European (JRC, EEA) and international agencies. Enormous amount of knowledge was extracted from te above sources regarding te estimation of irrigation water at farm level for various crops. Te agricultural researc community as focused muc on te agronomic side of water use, wic is reflected in te very large body of Evapotranspiration (ET). Most of tis researc concentrates on finding te water requirements for different crops under certain field conditions related to soil, climate, and te groundwater table. Additionally data on water distribution infrastructures and management practices were included in many similar researces. A second point was te determination of te availability and te nature (scale, spatial reference, analogue or in digital form) of te necessary data (specific climatic and soil variables) in te Greek territory. Furtermore, te availability of te necessary data related to land uses and irrigation metods systems. Te main outputs of te above work were te selection and adjustment of a proper model(s) and te detailed description of te needed computational and spatial statistical and matematical analysis of te necessary data. Te above-mentioned model combines data regarding land uses, crop water needs, irrigation metods, meteorological and soil data. Te sources of tese data are FSS and SAPM, meteorological data from te Hellenic National Meteorological Service (HNMS) and data from soil analyses of te relevant institute, NAGREF. Due to te spatial feature and te spatial complexities of te aforementioned basic data (soil, clime, land use) Geograpic Information System was te most efficient tool platform to develop and andle an irrigation water estimation model. A GIS-based approac allowed considering local variations in cropping, soil and climate and to spatial analyzing and interpolating te initial data. Te selected model was tested in a sample of agricultural oldings troug a pilot survey. A cost/benefit analysis was also implemented. According to te definition by tis model we estimated te volume of water, wic used for irrigation in agricultural for all cultivate except te kitcen gardens and greenouses Oter issues No oter issues were confronted. 2.9 Response-burden policy In order to increase te response rates te following measures were taken: a. In larger cities, were tere is an increased difficulty in arranging a meeting between te interviewer and te older, te interviewer made a prior pone-call to te older, in order to arrange an appointment. b. If te interviewer couldn t find te older at is residency, e would leave a note, including is name and pone number, in order to arrange an appointment for a different day. c. Priority was given to important olding, for example large farms. d. Extra care was given in training interviewers in andling difficult respondents and in cases tat it was considered necessary te ELSTAT personnel contacted directly te respondents in order to persuade tem to cooperate. 28

29 ACCURACY AND RELIABILITY OF THE DATA COLLECTED 3.1 Data processing, analysis and estimation Estimation and sampling errors for SAPM ACCURACY AND RELIABILITY OF THE DATA COLLECTED 3.1 Data processing, analysis and estimation Estimation and sampling errors for SAPM Te sampling tecnique tat was applied in tis survey was te One-Stage Stratified Random Sampling. Regarding te extrapolation factor te procedure tat was followed is te following: a) In te design pase of te survey an initial weigt (design weigt) was given to eac sampling unit (olding). Tis initial weigt was estimated as te inverse of te probability of selection. More precisely, for te olding i tat belongs to stratum te initial weigt is =1/Prob wi (selected unit i in te stratum ). As in eac separate stratum, te sampling units were n selected wit equal probabilities, te initial weigts for all sampling units belonging to te stratum are equal to: w N n = ( N : population size according to te data of te Register of Agricultural Holdings wit reference year 2008) b) Te population size N in eac stratum was estimated from te sample s information and farm structure survey data, as follows: N ( N 08 = + ) (1) c n were: : Te number of out of scope population (closed oldings, non- target population units) N c N n : Te number of new units N : Factor, wic adjusts te population sizes in strata to make te totals in te NUTS 3 areas to t conform to te population totals wic are based on data from te farm structure survey 2009/2010 tat was conducted on a census basis. c) Te estimation of out of scope oldings N is based on te sample s information by applying c te formula: N c were: n c : Te number of sample units being non-population units (closed oldings, non- target population units) = N n 08 n N c (2) t 29

30 Te fraction n N 08 represents te inverse of te initial inclusion probabilities of te initial sample in stratum. n d) Concerning te estimation of te new oldings N, in eac NUTS 3 area, te total number n of new oldings based on farm structure survey data was allocated to strata proportionally to te 08 population size ( N N ). c e) Te extrapolation factorw in stratum is calculated as follows: were: : Te final sample size in stratum m w = N m (3) Te variance estimation and te calculation of te coefficient of variation was carried out as follows: Symbolisms: In eac stratum (let ): y i : Te value of te caracteristic y of te olding of order i belonging to te stratum Y : Te total of te variable y for all oldings in te stratum Y : Te total of te variable y for all oldings in all strata. Tat is: Y = Y Estimation process Te estimation of Y and Y is given by te following formulae: Y Y = N m = m i= 1 Y y i (4) (5) Te variance estimation of Y and Y is given by: were: V N ( ) ( N m ) Y = m S 2, (6) 30

31 S 2 = m 1 2 yi m 1 i 1 V ( Y ) = V ( Y ) m i= 1 y m = i 2, (7) (8) Te coefficient of variation of total estimation Y is given by: CV V ( Y = ) (9) Y ( Y ) Te following tables depict te estimation of te basic crop caracteristics (in a) and teir Relative Standard Errors (RSE) by Region (NUTS 2). Table 3.1 Cereals for te production of grain, weat and spelt, durum Regions (NUTS 2) weat, rye, barley, oats, Potatoes and sugar beet grain maize, rice and oter cereals Estimation RSE (%) Estimation RSE (%) Wole Country / Greece Total ,9% ,6% Eastern Macedonia &Trace ,3% ,5% Central Macedonia ,6% ,6% Western Macedonia ,8% ,5% Tessaly ,1% ,1% Epirus ,8% ,0% Ionian Islands ,7% ,0% Western Greece ,7% ,9% Central Greece ,0% ,7% Peloponnesus ,3% ,2% Attica ,6% ,4% Nortern Aegean ,6% ,2% Soutern Aegean ,9% ,3% Crete ,0% ,4% Table 3.2 Regions (NUTS 2) Oilseed crops Permanent outdoor crops 31

32 Estimation RSE (%) Estimation RSE (%) Wole Country ,9% ,6% Eastern Macedonia &Trace ,9% ,7% Central Macedonia ,6% ,7% Western Macedonia ,5% ,4% Tessaly ,1% ,7% Epirus ,8% ,6% Ionian Islands ,5% Western Greece ,1% ,1% Central Greece ,1% ,0% Peloponnesus ,0% ,2% Attica 14 0,0% ,3% Islands of Nortern Aegean ,8% Islands of Soutern Aegean ,6% Crete ,5% Table 3.3 Regions (NUTS 2) Fres vegetables, melons, strawberries, flowers and ornamental plants Temporary grass and permanent grassland Estimation RSE (%) Estimation RSE (%) Wole Country ,7% ,6% Eastern Macedonia &Trace ,7% ,4% Central Macedonia ,0% ,1% Western Macedonia ,7% ,4% Tessaly ,9% ,6% Epirus ,4% ,8% Ionian Islands ,1% ,0% Western Greece ,1% ,2% Central Greece ,2% ,4% Peloponnesus ,9% ,8% Attica ,9% ,2% Islands of Nortern Aegean ,5% ,1% Islands of Soutern Aegean ,5% ,8% Crete ,2% ,8% Te following tables 3.4 and 3.5 depict te estimation of te livestock caracteristics (number of eads) and teir Relative Standard Errors (RSE) by Region (NUTS 2). Table

33 Regions (NUTS 2) Bovine animals Seep and goats Estimation RSE (%) Estimation RSE (%) Wole Country ,6% ,2% Eastern Macedonia &Trace ,1% ,5% Central Macedonia ,4% ,6% Western Macedonia ,6% ,2% Tessaly ,0% ,5% Epirus ,6% ,1% Ionian Islands ,3% ,3% Western Greece ,5% ,6% Central Greece ,1% ,8% Peloponnesus ,0% ,2% Attica ,0% ,7% Islands of Nortern Aegean ,7% ,2% Islands of Soutern Aegean ,8% ,5% Crete ,5% ,9% Table 3.5 Pigs Poultry Regions (NUTS 2) RSE RSE Estimation Estimation (%) (%) Wole Country ,7% ,6% Eastern Macedonia &Trace ,1% ,2% Central Macedonia ,1% ,6% Western Macedonia ,0% ,7% Tessaly ,6% ,8% Epirus ,7% ,7% Ionian Islands ,7% ,3% Western Greece ,0% ,7% Central Greece ,8% ,6% Peloponnesus ,6% ,3% Attica ,1% ,4% Nortern Aegean ,9% ,6% Soutern Aegean ,5% ,4% Crete ,4% ,5% In addition, we present te following tables in order to evaluate te precision of te estimates of te above caracteristics. Particularly, according to Regulation (EC) No 1166/2008 te RSE for te crop caracteristics in Regions sould be less tan 10%, in te case were te percentages (%) of UAA of te crops' areas in te Regions are equal or more tan 10%. Concerning te livestock caracteristics, te RSE in te Regions sould be less tan 10%, in te case were te percentages (%) of te livestock units of animal categories in te Regions are equal or more tan 10% and te percentage (%) of national sare of livestock units for animal categories is 33

34 less tan 5%. Te data of te following tables derive from te Agricultural and Livestock Census of 2009/2010. Table 3.6: Percentage (%) of UAA of te crops' areas by Region Area of: Regions (NUTS 2) Cereals Potatoes and sugar beet Oilseed crops Permanent outdoor crops Fres vegetables, melons, strawberries, flowers and ornamental plants Temporary grass and permanent grassland Wole Country 29,3% 3,5% 1,0% 27,3% 1,6% 21,6% Eastern Macedonia &Trace 46,5% 3,7% 6,7% 5,8% 2,0% 3,1% Central Macedonia 52,6% 3,4% 1,4% 14,4% 1,7% 7,7% Western Macedonia 58,8% 12,9% 0,3% 3,7% 0,4% 13,2% Tessaly 46,1% 5,4% 0,1% 12,1% 1,8% 8,7% Epirus 9,1% 0,7% 0,0% 27,7% 0,6% 40,8% Ionian Islands 3,2% 0,3% 0,0% 49,5% 0,5% 39,7% Western Greece 20,3% 2,4% 0,0% 37,5% 3,2% 20,7% Central Greece 26,5% 3,0% 0,0% 26,4% 2,4% 21,6% Peloponnesus 5,5% 1,1% 0,0% 69,6% 1,2% 16,6% Attica 11,8% 2,0% 0,0% 63,7% 5,0% 9,9% Nortern Aegean 7,3% 5,3% 0,0% 36,8% 0,3% 52,3% Soutern Aegean 6,1% 3,1% 0,0% 20,5% 1,2% 49,4% Crete 1,0% 0,5% 0,0% 41,0% 0,9% 54,4% Table 3.7: Percentage (%) of te livestock units of animal categories by Region Regions (NUTS 2) Bovine animals Livestock of: Seep and goats Pigs Poultry Wole Country 19,4% 55,6% 10,1% 13,8% Eastern Macedonia &Trace 35,7% 48,5% 9,4% 5,2% Central Macedonia 34,7% 34,8% 9,2% 20,4% Western Macedonia 30,5% 57,6% 6,1% 4,5% 34

35 Tessaly 25,2% 52,7% 16,4% 4,7% Epirus 17,1% 39,1% 11,5% 31,8% Ionian Islands 10,7% 72,7% 3,3% 10,9% Western Greece 14,1% 67,6% 11,5% 6,0% Central Greece 11,4% 55,4% 16,1% 16,2% Peloponnesus 7,4% 68,1% 13,4% 9,4% Attica 4,7% 17,6% 2,2% 74,6% Nortern Aegean 8,5% 79,3% 3,0% 5,0% Soutern Aegean 18,4% 63,7% 6,4% 7,4% Crete 0,5% 89,2% 4,1% 5,6% Table 3.8: Percentage (%) of national sare of livestock units for animal categories by Region Livestock of: Seep Regions (NUTS 2) Bovine and animals goats Pigs Poultry Wole Country 100,0% 100,0% 100,0% 100,0% Eastern Macedonia &Trace 16,6% 7,9% 8,4% 3,4% Central Macedonia 30,3% 10,6% 15,3% 25,0% Western Macedonia 6,4% 4,2% 2,4% 1,3% Tessaly 16,5% 12,0% 20,6% 4,3% Epirus 9,4% 7,5% 12,2% 24,6% Ionian Islands 0,8% 1,8% 0,4% 1,1% Western Greece 8,7% 14,6% 13,6% 5,3% Central Greece 4,0% 6,9% 11,0% 8,1% Peloponnesus 2,4% 7,7% 8,3% 4,3% Attica 0,7% 0,9% 0,6% 15,3% Nortern Aegean 1,2% 4,0% 0,8% 1,0% Soutern Aegean 2,6% 3,1% 1,7% 1,5% Crete 0,3% 18,8% 4,7% 4,7% In te Survey on Agricultural Production Metods (SAPM), te ig values of te RSE of te estimated number of pigs and especially of poultry are due to te fact tat certain oldings ave an unusually ig number of pigs or poultry. In te sample design of SAMP te above-mentioned oldings were considered of small economical size and as a result tey belong to sample strata (not in take-all strata. However, after te data collection, in strata tat contain tese oldings, te element variance of pigs and poultry appears to be ig. Terefore, tese variances inflate te RSE of te estimated number of pigs and poultry. 35

36 Notes: a) Regarding te standard errors presented in tis document, ELSTAT as double-cecked te oldings tat included pigs and poultry in strata, wic presented extreme variance. Extreme variance was noted wen we recorded ig values for te number of pigs and poultry. Tis review ceck revealed erroneous data in four oldings: tree in poultry and one in pigs. After te corrections, te RSE of te estimated number of poultry at wole country canged from 12.5% to 6.6%, wile te RSE of te estimated number of pigs remained almost uncanged. In addition, a similar ceck was performed for temporary grass and permanent grassland in Regions wit ig proportion of grass land ( 10 of UAA) and wit ig relative standard errors. Tese cecks did not result in any canges. b) Regarding te ig Relative Standards Errors tat are outside te precision requirements of te R1166/2008, we would like to inform you te following: Concerning te cereals in Attica te cultivated land wit cereals is more tan 10% (11.8%) of te total Utilized Agricultural Area (UAA).Tis is a rater ig percentage for te specific area (igly urban), but for te wole Country it covers only 0.59% of te cultivated land wit cereals. Concerning te temporary grass and permanent grassland, te RSE is outside te precision requirements of te R1166/2008 in two Regions mainly due to te fact tat tis land appears eterogeneity in te design strata. We notice te economic size of livestock oldings (wic is used as stratification variable) does not ave ig correlation wit temporary grass and permanent grass in many cases. Concerning pigs and poultry, te RSE is outside te precision requirements of te R1166/2008 in some Regions mainly due to te fact tat te number of animals presents ig eterogeneity in te design strata. According to te Register of ELSTAT, tese livestock oldings appeared small size and tey belong to sampling strata. However, at field work it was revealed tat tey were large scaled oldings and terefore tey sould ave been at take-all strata Non sampling errors Non-sampling errors arise mainly due to misleading definitions and concepts, inadequate frames, unsatisfactory questionnaires, defective metods of data collection, tabulation, coding, incomplete coverage of sample units etc. Tese errors are unpredictable and not easily controlled, and tey arise from te initial stage wen te survey is being planned and designed to te final 36

37 stage wen data are processed and analysed. Unlike in te control of sampling error, tis error may increase wen sample size is increased. If not properly controlled, non-sampling error can be more damaging tan sampling error for large-scale business surveys. Te non-sampling errors tat appear in all statistical processes can be categorised as: Coverage errors Measurement errors Processing errors Non-response errors In practice, te non-sampling error can be decomposed into variable error (or variable component) and systematic error (or bias). Variable error arises from random factors affecting different samples and repetitions of te survey, wereas bias refers to systematic errors tat affect any sample taken under a specified survey design wit te same constant error. All variable errors (sampling and no sampling) are incorporated into te variance of te estimates. Coverage errors Coverage errors (or frame errors) arise due to existing divergences between te target population and te frame population. We can distinguis te following types of coverage error: Over-coverage stems from te fact tat tere are units accessible via te frame but tey do not belong to te target population. In agricultural surveys, te over-coverage mainly as to do wit oldings tat were included in te farm register, tey were selected in te sample, but tey did not actually exist at te time of te survey (e.g. oldings tat do not operate permanently or temporarily, oldings fully turned over and merged wit anoter olding etc). Tese oldings actually reduce te initial sample size. Te decrease of te number of sampling units from te initial to te actual size inflates te variance of te survey caracteristics. Survey on Agricultural Production Metods By using te sample data, te over-coverage rate (%) of closed and merged oldings as been estimated and it amounts to 12,4% By using te sample data, te over-coverage rates (%) of closed and merged oldings ave been estimated and tey are depicted in te tables 6 and 7. Table 6:Over-coverage rates(%) by type of farming and economic size of oldings Classes of economic size of te oldings Type of farming Total 0 * ** Total 12,1 20,4 28,6 17,3 12,3 8,0 4,9 2,2 8,5 T10 12,4 0,0 42,8 23,5 14,2 9,0 5,4 2,0 7,7 T21 9,4 0,0 44,5 30,0 17,1 11,6 3,2 2,2 0,0 T22 5,6 0,0 66,7 17,6 9,4 6,8 4,0 3,4 0,0 T30 13,1 0,0 21,7 14,2 11,0 8,0 5,9 3,0 0,0 T41 8,9 0,0 48,8 27,6 19,4 8,9 5,6 2,4 0,0 37

38 T44 12,1 0,0 43,5 26,9 17,4 8,3 3,7 1,7 12,3 T51 9,0 0,0 13,3 36,0 24,2 12,5 11,5 4,0 0,0 T52 10,9 0,0 40,5 16,7 12,5 25,0 10,8 3,8 0,0 T53 20,0 0,0 21,4 30,4 13,3 20,0 14,3 0,0 0,0 T60 10,3 0,0 35,3 15,2 10,4 5,6 4,3 1,4 0,0 T90 5,3 20,4 16,1 10,5 4,5 4,9 2,4 2,4 0,0 * Holdings tat follow organic farming and wose economic size is not specified ** Holdings tat te economic size is not specified Table 7: Over-coverage rates (%) by size classes Code % Area wit crops in ectares of class Total 25,5 11 Less tan 1 ectare 38,4 12 From 1 to less tan 2 ectares 29,0 13 From 2 to less tan 3 ectares 27,1 14 From 3 to less tan 5 ectares 22,4 15 From 5 to less tan 10 ectares 19,8 16 From 10 to less tan 20 ectares 18,2 17 From 20 to less tan 30 ectares 17,8 18 From 30 to less tan 50 ectares 11,3 19 Equal to or greater tan 50 ectares 11,4 7 Not specified area 12,9 10 Not specified area 41,1 Census According to census data, te over coverage rate (%) amounts to18,4%, as follows: Closed oldings + Merged oldings + Duplicates Over cov erage rate (%) = 100 = Initial size = 100 = 18,4% were: a) Initial size= Holdings in Register + New oldings + Holdings arisen from te division of oldings= oldings oldings oldings = oldings b) Closed oldings=holdings tat do not operate permanently + Holdings tat do not operate temporarily= oldings oldings= oldings 38

39 c) Merged oldings= oldings in te table tese are recorded as oldings wit cange of te manager. Tose merged oldings ave at te same time and cange in manager. Tis is now depicted in te table page 43. d) Duplicates in te Register=5.193 oldings Te above information concerns te census over-coverage. Misclassification stems from te fact tat te auxiliary information provided by te frame may be inaccurate for some population units (e.g. wrong economic size or olding s type of production). Due to problems of misclassification, te coefficient of variation of te produced statistics of SAMP is iger tan te coefficient of variation based on te initial sample design. Under-coverage refers to units missing from te sampling frame. As a result, te undercoverage problem underestimates te produced statistics. Corrections and weigting for under-coverage is difficult, because it cannot be obtained from te sample itself, but only from external sources. According to census data, oldings were not covered by field enumeration, because teir olders refused to answer (refusal rate=0,9). In addition, according to census data, oldings were not covered, because teir olders were eiter unknown or temporary absent, in ospital etc. Due to refusals and te rest not surveyed oldings, about 6,4% of oldings were not covered by field enumeration and te appeared under-coverage of te total utilized agricultural area is about 3,5%, according to te istorical data from te Register of ELSTAT. Te above under coverage rate as been calculated as follows: Re fusals+ Rest not surveyedoldings Under coveragerate(%) = 100= Respondents+ Re fusals+ Rest not surveyedoldings = 100= 6,4% were: a) Respondents= oldings b) Refusals = oldings c) Rest not surveyed oldings (olders were unknown, temporary absent etc)= oldings A suitable imputation tecnique was applied to elp maintain coverage and compensate for missing data from te not surveyed oldings (refusals and rest not surveyed units). A ot deck approac was used for oldings since tey are known to be under operation, because most of tese oldings participated in te sample of FSS of te previous years. Here, a not surveyed olding was matced wit a similar responding olding and all te relevant variables of te responding olding were ascribed to te not surveyed oldings. Areas and capital livestock units for not surveyed units were estimated on tat basis. More specifically, Random imputation witin classes was applied. In tis ot-deck metod, a respondent (donor) is cosen at random witin an imputation class and te selected respondent s values are assigned to te not surveyed unit (recipient). Te respondents are as omogeneous as possible witin eac class. For eac missing value, a reported value is imputed wic is in te same class. Tus, te assumption is made tat witin eac class te non-surveyed units follow te same distribution as te respondents. Te auxiliary variables, used to define te imputation classes for oldings, were: a) Municipality/Commune, b) Type of farming and c) Economic size. For bot te donors and 39

40 te recipients te values used to determine te imputation classes, in wic tey belong, were taken by te istorical data of ELSTAT Register. After te implementation of te imputation tecnique, te under-coverage rate is 1,6% at national level, in terms of number of farms. Te under-coverage rate is 1.6% for bot te Census and te SAPM. Tis was acieved due to te fact tat te extrapolation factors were calculated so as te estimated number of oldings of te Census to coincide wit tat of te SAPM, at NUTS 3 level (Perfecture) Measurement errors Measurement errors occur during data collection and make te recorded values of variables to be different tan te true ones. Generally, tey can be regarded as random errors, wic increase te variance wit contributions, wic enter automatically in te calculations of te variance. Processing errors Once data ave been collected, a range of processes is performed before te production of final estimates (e.g. coding, editing, weigting and tabulating etc.). Errors tat arise at tese stages are called processing errors. Processing errors can be regarded as random errors, wic increase te variance Non-response errors Non-response refers to te failure of collecting data from some or all variables of te population units designated to obtain information in a sample or complete enumeration. Te difference between te statistics computed from te collected data and tose tat would be computed if tere were no missing values is te non-response error. Tere are two types of non-response a) unit non-response, wic occurs wen data can not be collected from all te designated population units and b) item non-response, wic occurs wen te information is not gatered on all survey variables from te designated population units. Survey on Agricultural Production Metods Te response rate of te SAMP was defined as te fraction of actual sample size divided by te initial sample size. Te initial sample size of te SAMP contains respondents, refusals, closed, merged or out of scope oldings, as well as oldings wit olders unknown, temporary absent, in ospital, etc. Te response rate of te SAMP amounts to 71,9%. A basic problem is tat te response rate is not directly related to bias, tat is, te main problem caused by non-response. In principal, it is possible non-response rates to be low and bias to be ig and vice-versa. In SAMP, re-weigting was applied to amend suitably te extrapolation factors, by taking into account te response rates in all final strata. It compensates for non-responses, and reduces te absolute bias of te estimation of te survey caracteristics. Te aim is to remove non response bias but, in practice, tis is unlikely to be fully acieved. Te effect of non-response on te produced statistics is tat it increases teir variability and introduces bias. Bias is introduced by te fact tat non-respondents may be different tan respondents in teir values of some survey caracteristics. Variability increases due to decreased sample size and weigting adjustments tat are used to compensate for unit non-response. Te tables 8 and 9 depict te unit response rates (%), total and broken down by type and sizes of oldings. 40

41 Table 8:Unit response rates (%) by type of farming and economic size of oldings Type of Classes of economic size of te oldings Total farming 0 * ** Total 71,9 33,3 54,4 64,9 71,4 77,0 80,9 82,5 72,0 T10 77,0 48,1 65,3 75,6 80,3 84,9 86,4 76,9 T21 67,9 37,0 50,0 61,6 69,4 72,9 72,8 0,0 T22 73,1 33,3 52,9 68,8 72,7 82,7 72,5 0,0 T30 65,8 57,4 63,4 67,8 71,5 74,3 77,9 88,9 T41 77,5 39,5 61,9 71,8 77,1 81,2 82,5 100,0 T44 75,3 45,2 62,2 69,2 80,5 82,9 84,3 65,8 T51 80,5 80,0 60,0 69,7 80,0 80,8 83,4 0,0 T52 76,5 48,6 66,7 81,3 69,4 73,8 82,8 0,0 T53 61,1 57,1 56,5 60,0 70,0 57,1 85,7 0,0 T60 75,9 52,6 72,1 75,9 80,7 81,6 83,2 85,7 T90 76,1 33,3 68,6 73,0 78,1 78,5 78,5 79,7 72,7 * Holdings tat follow organic farming and wose economic size is not specified ** Holdings tat te economic size is not specified Table 9: Response rates (%) by size classes Code of Area wit crops in ectares (%) classes Total 55,2 11 Less tan 1 ectare 39,1 12 From 1 to less tan 2 ectares 46,2 13 From 2 to less tan 3 ectares 54,3 14 From 3 to less tan 5 ectares 61,9 15 From 5 to less tan 10 ectares 60,9 16 From 10 to less tan 20 ectares 64,7 17 From 20 to less tan 30 ectares 65,8 18 From 30 to less tan 50 ectares 69,4 19 Equal to or greater tan 50 ectares 62,9 7 Not specified area 67,7 10 Not specified area 41,7 Census Te response rate of te census is te fraction of respondents divided by te initial population size (respondents, closed, merged, refusals etc). Te response rate of census amounts to 76,5%, taking into account also te non-active oldings. Te non response of te census creates under coverage tat amounts to 6,4%. However, te under-coverage rate as been reduced to 1,6% at national level, after te implementation of imputation tecnique.. Item non-response 41

42 Item non-response was not appeared in te oldings included in te survey, and as a result no imputation was applied Metods for andling missing or incorrect data items. Control of te data Follow up interviews were conducted in cases were missing or incorrect data were detected. In most cases telepone was used. Item imputation was also conducted. Te imputation procedure was based on te usage of relevant questions in te questionnaire. Te processing and cecking of te data was carried out in five pases as follows: Pase 1 Te first pase encompassed te processing of te questionnaires by officials of te Regional Statistical Offices (supervisor and assistant supervisors). Te assistant supervisors cecked manually all te questionnaires in order to define teir completeness and consistency and correct tem accordingly. It is noted tat te interviewers, before delivering te questionnaires to teir assistant supervisor, ad to ceck if tey were complete and consistent. Pase 2 Te questionnaires were scanned and te second pase of controls followed. Tese controls included data verification (meaning te verification of caracters not written very clearly by te interviewer), verification (meaning te verification of specific values, sums, etc), batc integrity (meaning correction of oldings surveyed twice) etc. Te data ad to pass troug tese different types of corrections in order to be transmitted to te main database. Pase 3 A new set of controls was conducted to te imported data to te database. Tere were two types of errors: Te first type of errors was spotted by warnings, for example a very big (unusual) value for te number of animals. Tis type of error did not necessarily need correction. Te second type concerned errors tat ad to be corrected eiter immediately or later, depending on te availability of te correct answer. Te control of te questionnaire could be continued. Later te user sould print all te incorrect cases in order to solve te problems after consultation wit te interviewer or even te older. Pase 4 Questionnaires were cecked again in order to detect errors of te integrated data, for example double or multiple entry oldings etc. 42

43 Pase 5 Quality controls related to te aggregates at NUTS III level were made. Te quality controls were carried out in order to ensure te quality of te final file and te aggregates at regional level. 3.2 Evaluation of results Survey FSS (excl. OGA in case of sample survey) OGA (if sample survey) SAPM (if sample survey) Initial list of units Initial sample Number of oldings wit completed questionnaires (incl. Eventual imputed questionnaires): Number of units under te tresold applied * NA NA Duplicated oldings Holdings wit ceased activities: (If information is available) of wic definitely ceased, i.e. te land is abandoned, of wic: no more operated (due to abandon, canges of use, etc temporary no operated (If information is available) of wic oldings wit cange of te manager due to merge wit oters oldings, of wic: fully transfer and merged wit oter olding rented or it was turned over under anoter legal status and merged wit anoter olding Unit Non-response : Refusals not corrected of wic: unknown older te older refuse to answer oter (temporary absence, in ospital, impossible to contact te older e.t.c) Refusals corrected (imputed) oldings, wic are: old oldings (from te Farm Register) plus new oldings tat ELSTAT identified as operating during te census. From te latter, are new oldings coming from te splitting of old oldings tat were registered in te Farm Register, and are new oldings oldings, wic are: old oldings (from te Farm Register) plus new oldings coming from te splitting of old oldings tat existed in te Farm Register. 43

44 Number of records transferred to Eurostat * Common land units (A_2_1) * Units tat do not meet te national tresold criteria (in some countries tere could be completed questionnaires for tem, in oters not). In case were it's impossible to provide tis information, a sort explanation about te reasons is to be provided. Comments on major trends from FSS 2007 to FSS Comments must be given in case tere is a cange of more tan 10% at national level between FSS 2007 and FSS 2010 for any of te groups below: From FSS 2007 From FSS 2010 % Difference in Number of oldings; ,9% UAA (A_3_1), a; , ,0-14.7% Arable land, a; , ,5-16,6% Permanent grassland (B_3), a; , ,1-8,4% Permanent crops (B_4), a; ,3-15,6% Wooded area (B_5_2), a; , ,3-16,6% Unutilised Agricultural area (B_5_1), a; , ,2-11,5% Fallow land (B_1_12_1 + B_1_12_2), a; , ,9-28,2% LSU in LSU; ,9-8,4% Cattle (C_2), ead; ,2% Family Labour force - in persons; ,8% Family Labour force - in AWU; ,3% Non family labour force - in persons; ,4% Non family labour force - in AWU ,0% Concluding Remarks According to te results of te 2009 Agricultural and Livestock Census (FSS 2010) and after comparison wit te results of te 2007 Survey on Agricultural-Livestock Holdings (FSS 2007), as tese are sown above, tere appears to be a significant decrease during te period in te number of agricultural oldings (-15,9%) and in te utilized agricultural areas (-14,7%) as well as in all variables tat appear in te above table. Tese findings were torougly studied and cecked by ELSTAT. Te views of te Ministry of Agricultural Development and Food and of oldings duplicates wit ceased activities refusals = special records for common land in NUTS 3 level. 44

45 te Payment and Control Agency for Guidance and Guarantee of Community Aid (OPEKEPE) were also sougt. Te potential factors for te above noted decrease include te following: 1. Up to now, te Farm Register is being updated basically by te agricultural census and to a certain extent by sample agricultural surveys and administrative sources. It as not been possible to record and depict all te canges in te structure of te agricultural oldings for a number of reasons. For example, a number of oldings tat were not operational any more were sown in te Farm Register as active. Tus te number of oldings included in te Farm Register, wic is te sampling Frame for te conduct of te FSS 2007 is, in tat particular case, potentially greater tat te real one. We consider tat tis led to an overestimation of te FSS 2007 variables, explaining to a significant degree te apparent large decreases in oldings (and utilized land) between te 2007 FSS and te 2009 Census. ELSTAT will consider te possibility of revising te 2007 FSS data and previous FSS data, taking into account inter alia te results of te 2009 and 1999 Censuses. Te issues raised regarding te quality of te farm register wit regard to te sources used for its updating are presented analytically above. As it as been also mentioned above, ELSTAT as already taken actions to improve te metods and te tecniques of te update of te register. In te Framework of te Joint Overall Statistical Greek Action Plan (JOSGAP) an external consultancy as already taken place and ways for improving te register ave been indicated. ELSTAT envisages implementing a project wit te aim of linking its farm register wit te registers kept by oter Greek Autorities, wic will enable ELSTAT to update te farm register continuously and more fully. 2. It sould also be noted tat te 2009 Census took place at a significantly later date tan originally planned and would ave been optimal. Te reason for tis cange of te date of te conduct of te census ad to do wit te crisis in wic te ten National Statistical Service of Greece found itself in (triggered by te problems in te area of government finance statistics) in te Fall of In tis context, te time tat te 2009 Census could finally be conducted was te Summer-early Fall of 2010, and tis may ave conceivably contributed to some underenumeration of oldings and teir owners beyond wat would ave taken place under oter circumstances, altoug tere is no direct evidence of an under enumeration and te latter cannot can be estimated. 3. Finally, it is quite possible tat canges in economic incentives and oter objective conditions may ave ad an actual effect on developments in te agricultural sector, wic te Census results properly captured. 45

46 Quality Controls In variables were big variations were detected, (te percentage of variation depends on te kind of variable), ten an in dept analysis was carried out in close cooperation wit te regional offices and Ministry of Rural Development and Food. 4. PUBLICATION AND PUNCTUALITY 4.1 Publication From te database of te Agricultural Census 2009, te Eurofarm file was compiled, wit individual data for eac olding. Tis file was sent to Eurostat in Marc Preliminary Results will be publised on te web site of NSSG in te second semester of 2012 (free of carge). Te final results of agricultural Census 2009 at national level will be presented in te second alf of 2012 in te form of: Detailed tables (national series of tables), An electronic and a ard copy publication containing statistics and related analyses, togeter wit maps sowing te geograpical distribution of te various survey caracteristics (not free of carge). All publications contain meta-data. Access to individual data for users is not possible at all. 46