Agricultural Integrated Pilot Survey 2018 - Ghana

Abstract
---------------------------

The AGRIS Ghana Pilot test was implemented in 4 districts of the Ashanti Region (Ahafo Ano South, Asante Akim North, Ejura Sekye Dumase, and Sekyere Afram Plains) in February 2018, to collect information on:
- Crop and livestock production as well as data on farm characteristics, diversification and structures;
- Farm revenues and expenses;
- Type of labour used by the agricultural holding;
- Farming practices and their linkages with the natural environment;
- Farm machinery, equipment and assets.

The general objective of the pilot was to customize AGRIS instruments and methodologies for adoption as a standard tool to efficiently gather relevant and reliable agricultural data for policy making and monitoring the Sustainable Development Goals (SDGs).

The specific objectives of the AGRIS Ghana pilot were as follows:
- Elaborate the overall set up of AGRIS in Ghana;
- Customize the content of the AGRIS questionnaire to the Ghanaian context;
- Assess the overall efficiency of the customized, integrated questionnaires and their feasibility in terms of length, flow, use of Computer Assisted Personal Interviewing (CAPI), and integration of core and rotating modules;
- Assess the difficulty and relevance of each question, each section and each generic questionnaire for different types of holdings;
- Test the use of Survey Solutions software to implement CAPI data collection, and the current version of the CAPI questionnaires;
- Assess the relevance of the training material developed to train survey enumerators and supervisors.

Geographic coverage
---------------------------

District level coverage. The 4 district covered by the survey were:
- Ahafo Ano South (CORE+PME)
- Asante Akim North (CORE+MEA)
- Ejura Sekye Dumase (CORE+LABOUR)
- Sekyere Afram Plains (CORE+ECO)

Analysis unit
---------------------------

Agricultural holdings in the household sector

Universe
---------------------------

All households, agricultural or not, in the 4 surveyed districts.

Kind of data
---------------------------

Sample survey data [ssd]

Sampling procedure
---------------------------

Definition of Agricultural Holding
As stated in the manual of the World Programme for the Census of Agriculture (FAO, 2015), an agricultural holding is defined as an economic unit of agricultural production under single management comprising all livestock and poultry kept, and all land used wholly or partly for agricultural production purposes, without regards to title, legal form, or size. Single management may be exercised by an individual or household, jointly by two or more individuals or households, by a clan or tribe, or by a juridical person such as a corporation, cooperative or government agency (FAO, 2015).

1. The Sampling Frame
The initial plan for the pilot survey was to consider as statistical units, agricultural holdings covering both the household and the non-household sectors, as proposed in the AGRIS methodology.
For holdings in the household sector, no updated list of agricultural households in the country was available, and therefore a sampling frame needed to be established. To do so, the 2010 Population and Housing Census (PHC) was used to build a frame of EAs which were the primary sampling units (PSUs) of the adopted sampling design. After selecting the sample of PSUs in the four districts of interest, a complete list of holdings in the selected EAs was built. All households, agricultural or not, present in the selected EAs were listed.

Holdings in the non-household sector are by definition, economic units such as commercial farms and government institutions engaged in agricultural production. GSS and MoFA provided a list of these holdings to be used as sampling frame. Therefore, the plan was to use as the overall sampling frame a multiple frame composed of the two lists described above (one for the household sector and one for the non-household sector). However, after further discussion and evaluation, it was determined that the list of holdings in the non-household sector could not be considered as a reliable sampling frame for the targeted units. As a consequence, the data collected for the 80 non-household units could not be analysed to represent holdings in the nonhousehold sector.

2. The Sampling design
A stratified two-stage sampling design was used for the holdings in the household sector. The PSUs were the EAs and the secondary sampling units (SSU) were the agricultural households.

3. The Sampling Size
For holdings in the household sector, the calculation of sample size was performed fixing the minimum degree of precision required for the final estimates of main variables of interest. The variable considered to determine the sample size was the area of the agricultural land owned by the households. This information had been collected during the 2012-2013 Ghana Living Standards Survey 6 (GLSS6). Therefore, data from this survey was used to estimate the coefficient of variation (CV) of the variable of interest in the chosen four districts.
It should be noted that the estimation domain of the GLSS6 was the region. For that survey, a two-stage sampling design was used and the PSUs (EAs) were selected in each region with the probability proportional to size (PPS). The measure of size was given by the number of individuals in each region, provided for the chosen districts for the AGRIS-Ghana pilot survey by the GLSS6. For the estimation of the CV of the households' agricultural land, it was assumed that the EAs sampled in GLSS6 and located in the target districts were selected in these districts with the same method of selection (PPS). Thus, the households included in the sample were supposed to have been selected with a two-stage sampling design.

The formula for the computation of the sampling size can be consulted in the final report of the survey.

The number of households to be surveyed in each PSU is fixed to 10. Therefore, the size of the sample of PSU is the size of the sample of the households divided by 10.

4. Agricultural holding definition
As stated in the manual of the World Programme for the Census of Agriculture (FAO, 2015), an agricultural holding is defined as an economic unit of agricultural production under single
management comprising all livestock and poultry kept, and all land used wholly or partly for agricultural production purposes, without regards to title, legal form, or size. Single management
may be exercised by an individual or household, jointly by two or more individuals or households, by a clan or tribe, or by a juridical person such as a corporation, cooperative or
government agency (FAO, 2015).

Sampling deviation
---------------------------

As mentioned in the sampling procedure section, holdings in the non-household sector were not included in the survey, as per initial plan, due to a problem in the listing frame provided by the Ghana Statistical Service.

Mode of data collection
---------------------------

Computer Assisted Personal Interview [capi]

Research instrument
---------------------------

The AGRIS Core module integrates with the Production Methods and Environment module (Core+Pme) collected information on household and holding characteristics, agricultural production and practices used for agricultural production by holdings.

A full appraisal of the contents of the questionnaires can be get by downloading the questionnaires in the documentation section.

Cleaning operations
---------------------------

The first level of data quality checks was implemented through Survey Solutions, which allows the programmer to develop a questionnaire containing skips and validation rules to minimize errors and inconsistencies in the collected data.

After data collection, data editing and preparation was performed using STATA.

Response rate
---------------------------

Out of 370 households planned for interview, 366 were interviewed (98.91% response rate).

Sampling error estimates
---------------------------

Details on the estimates of sampling error are provided in the final survey report.

摘要
---------------------------
AGRIS加纳试点测试于2018年2月在阿散蒂地区的四个区（阿哈福阿诺南、阿桑特阿基姆北、埃祖拉塞基杜马塞和塞克里阿弗拉姆平原）实施，旨在收集以下信息：
- 作物和畜牧业生产，以及农田特征、多元化结构和数据；
- 农场收入和支出；
- 农业经营单位所使用的劳动力类型；
- 农业实践及其与自然环境的联系；
- 农田机械、设备和资产。
试点研究的一般目标是将AGRIS工具和方法定制化，作为一项标准工具，以高效收集与政策制定和监测可持续发展目标（SDGs）相关的可靠农业数据。
AGRIS加纳试点研究的具体目标如下：
- 制定AGRIS在加纳的整体架构；
- 将AGRIS问卷内容定制化以适应加纳的国情；
- 评估定制集成问卷的整体效率和可行性，包括问卷长度、流程、计算机辅助个人访谈（CAPI）的使用以及核心模块和旋转模块的整合；
- 评估每个问题、每个部分和每个通用问卷对不同类型经营单位难度和相关性；
- 测试使用Survey Solutions软件实施CAPI数据收集以及当前版本的CAPI问卷；
- 评估为培训调查员和监督员开发的培训材料的适用性。
地理覆盖范围
---------------------------
区级覆盖范围。调查涵盖的四个区为：
- 阿哈福阿诺南（核心+PME）；
- 阿桑特阿基姆北（核心+MEA）；
- 埃祖拉塞基杜马塞（核心+LABOUR）；
- 塞克里阿弗拉姆平原（核心+ECO）。
分析单元
---------------------------
家庭农业部门中的农业经营单位

总体
---------------------------
调查涵盖的四个区的所有家庭，无论是否从事农业。
数据类型
---------------------------
样本调查数据 [ssd]
抽样程序
---------------------------
农业经营单位的定义
如世界农业普查计划手册（FAO，2015年）所述，农业经营单位被定义为在单一管理下，包括所有饲养的牲畜和家禽，以及全部或部分用于农业生产的土地的经济生产单位，不考虑所有权、法律形式或规模。单一管理可以由个人或家庭、两个或更多个人或家庭共同、部落或族群，或法人实体如公司、合作社或政府机构行使（FAO，2015年）。

1. 抽样框
试点调查的初步计划是将涵盖家庭和非家庭部门的农业经营单位视为统计单位，如AGRIS方法所建议。
对于家庭部门的经营单位，由于国家缺乏农业家庭最新清单，因此需要建立抽样框。为此，使用了2010年人口和住房普查（PHC）来构建一个框架，该框架是采用抽样设计的初级抽样单位（PSU）。在选定四个感兴趣地区的PSU样本后，构建了所选EAs中经营单位的完整清单。在所选EAs中，列出了所有农业或非农业家庭。
非家庭部门的经营单位按定义是经济单位，如商业农场和从事农业生产的政府机构。GSS和MoFA提供了这些经营单位的清单，用作抽样框。因此，计划使用由上述两个清单组成的总体抽样框（一个用于家庭部门，一个用于非家庭部门）。然而，经过进一步讨论和评估，确定非家庭部门经营单位的清单不能作为针对目标单位的可靠抽样框。因此，收集的80个非家庭单位的数据无法分析以代表非家庭部门的经营单位。

2. 抽样设计
对于家庭部门的经营单位，使用了分层两阶段抽样设计。PSU是EAs，次级抽样单位（SSU）是农业家庭。

3. 抽样量
对于家庭部门的经营单位，计算样本量时固定了所需的最小精度程度，以用于最终估计主要变量的估计值。用于确定样本量的变量是家庭拥有的农业土地面积。这些信息是在2012-2013年加纳生活标准调查第6次（GLSS6）期间收集的。因此，使用了此调查的数据来估计所选四个地区中感兴趣的变量的变异系数（CV）。应注意的是，GLSS6的估计领域是地区。对于该调查，使用了两阶段抽样设计，并在每个地区以规模成比例的概率（PPS）选择了PSU（EAs）。规模衡量标准是每个地区提供的选定地区的人数。为了估计家庭农业土地的变异系数，假设GLSS6中采样的EAs位于目标地区，并使用相同的方法在这些地区进行选择（PPS）。因此，包括在样本中的家庭应假设是使用两阶段抽样设计选出的。
样本量计算的公式可以在调查的最终报告中查阅。
每个PSU要调查的家庭数量固定为10。因此，PSU样本量是家庭样本量的十分之一。

4. 农业经营单位定义
如世界农业普查计划手册（FAO，2015年）所述，农业经营单位被定义为在单一管理下，包括所有饲养的牲畜和家禽，以及全部或部分用于农业生产的土地的经济生产单位，不考虑所有权、法律形式或规模。单一管理可以由个人或家庭、两个或更多个人或家庭共同、部落或族群，或法人实体如公司、合作社或政府机构行使（FAO，2015年）。

抽样偏差
---------------------------
如抽样程序部分所述，由于加纳统计局提供的清单存在问题，非家庭部门的经营单位未包括在调查中，按照最初计划。
数据收集方式
---------------------------
计算机辅助个人访谈 [capi]
研究工具
---------------------------
AGRIS核心模块集成了生产方法和环境模块（Core+Pme），收集了家庭和经营单位特征，以及经营单位用于农业生产的农业生产和技术信息。
可通过在文档部分下载问卷来全面评估问卷内容。
数据清洗操作
---------------------------
通过Survey Solutions实施了第一级数据质量检查，该软件允许程序员开发包含跳过和验证规则的问题卷，以最大限度地减少收集数据中的错误和不一致性。
在数据收集后，使用STATA进行了数据编辑和准备。
应答率
---------------------------
在计划采访的370个家庭中，有366个家庭接受了采访（应答率为98.91%）。
抽样误差估计
---------------------------
抽样误差估计的详细信息可在最终调查报告中找到。