Living Standards Survey III 1991-1992 - Ghana

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

Following the pattern set in the first two rounds of the Ghana Living Standards Survey (GLSS), the questionnaire used for the third round again covered a wide spectrum of topics such as education, health, housing, employment, income and expenditure, which affect the living standards of households. GLSS III thus provides data on various aspects of Ghanaian household economic and social activities, which are of help for monitoring the impact of the Government's Economic Recovery Programme. GLSS III differed from the two previous rounds, however, in concentrating particularly on the income, consumption and expenditure of households at a much more disaggregated level than previously. As a result, GLSS III provides more accurate estimates of income and expenditure, including the imputed value of home produced food which is consumed by households. The data on household expenditure are also being used to derive the weights needed for rebasing the Consumer Price Index. The GLSS data on income, consumption and expenditure, together with other individual, household and community level data collected in GLSS III, will also provide a valuable database for national and regional planning purposes. Detailed anthropometric data had been collected in GLSS I and GLSS II, involving the need to include an anthropometric in each survey team. This topic had to be dropped from GLSS III, so that the expanded income, consumption and expenditure data could be collected.

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

National

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

Households

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

Sample survey data [ssd]

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

A multi-stage sampling technique was used in selecting the GLSS sample. Initially, 4565 households were selected for GLSS III, spread around the country in 407 small clusters. in general, 15 households were taken in an urban cluster and 10 households in a rural cluster. The actual achieved sample was 4552 households. Because of the sample design used, and the very high response rate, the sample can be considered as being self-weighting, though in the case of expenditure data, weighting of the expenditure values is required.

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

Face-to-face [f2f]

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

The data collected in this survey were entered directly onto microcomputers which had been installed in the eight regional capitals. Kumasi and Accra had two PCs each, while Tamale, Sunyani, Koforidua, Ho, Cape Coast and Sekondi/Takoradi had one each. Special interactive software programs had been prepared for data entry and checking, using the software package Rode-PC. Data entry was done in two rounds. In both urban and rural clusters interviewers completed Part A of the questionnaire by the end of the fifth visit to each household; and after checking them, the supervisor took these questionnaires straight away to the regional capital, where the data entry operator began keying in. Once Part B had been completed, the supervisor took these questionnaires to the regional capital, and returned with the Part A questionnaires, plus detailed printouts showing what errors had been discovered by the editing program during the keying in operation. These errors were then corrected in the field. By the time the data entry operator had finished keying in the second batch of questionnaires (Part B), the team would have moved from those clusters to the next set of clusters. However, the next set of clusters were very close to the previous ones, so going back to correct errors detected in the second round involved travelling only a short distance. This arrangement made field reconciliation fairly easy. In addition, each set of clusters had been chosen close together so as to make supervision relatively easy. Finally, clusters in areas that were hardly accessible during the rainy season were scheduled to be covered during the dry season. At regular intervals during the fieldwork the diskettes containing the GLSS III data for each completed cycle were returned to the headquarters in Accra. Final tabulations were produced using the SAS software package.

摘要
---------------------------
遵循前两轮加纳生活标准调查（GLSS）所设定的模式，第三轮的调查问卷再次覆盖了广泛的议题，如教育、卫生、住房、就业、收入与支出等，这些议题均影响着家庭的生活水平。因此，GLSS III 提供了关于加纳家庭经济和社会活动各个方面的数据，这些数据有助于监测政府经济复苏计划的影响。然而，GLSS III 与前两轮调查的主要区别在于，它特别关注家庭收入、消费和支出的细分程度，远超以往。因此，GLSS III 提供了更精确的收入和支出估计，包括家庭消费的自产食品的推算价值。家庭支出数据还被用于推导构建消费者价格指数所需的权重。GLSS III 关于收入、消费和支出的数据，连同其他在GLSS III中收集的个人、家庭和社区层面的数据，也将为国家和区域规划提供宝贵的数据库。在GLSS I和GLSS II中，已经收集了详细的体质测量数据，需要在每个调查团队中包含一名体质测量人员。这一议题在GLSS III中被取消，以便收集扩展的收入、消费和支出数据。
地理覆盖范围
---------------------------
全国
分析单元
---------------------------
家庭
数据类型
---------------------------
样本调查数据 [ssd]
抽样程序
---------------------------
在GLSS样本的选取中采用了多阶段抽样技术。最初，为GLSS III选择了4565个家庭，分布在407个小集群中，遍布全国。一般来说，在城市集群中选取15个家庭，在农村集群中选取10个家庭。实际达到的样本量为4552个家庭。由于采用的样本设计以及极高的响应率，该样本可以被认为是自我加权的，尽管在支出数据的情况下，需要支出值的加权。
数据收集方式
---------------------------
面对面 [f2f]
数据清理操作
---------------------------
在本调查中收集的数据直接输入到安装在八个区域首府的微型计算机上。库马西和阿克拉各有两台PC，而塔马莱、索尼阿尼、科佛杜亚、豪、凯普科斯特和塞孔蒂/塔科拉迪各有台。特别为数据输入和检查准备了交互式软件程序，使用Rode-PC软件包。数据输入分两轮进行。在城市和农村集群中，访谈员在第五次访问每个家庭时完成问卷的第一部分；检查无误后，督导员立即将这些问卷带到区域首府，数据录入操作员开始录入。一旦完成第二部分（B部分），督导员将这些问卷带到区域首府，并带回第一部分的问卷，以及详细的打印输出，显示在录入操作期间编辑程序发现的错误。然后，在实地对这些错误进行纠正。当数据录入操作员完成第二批问卷（B部分）的录入时，团队已经从这些集群转移到下一组集群。然而，下一组集群与前一组非常接近，因此，在第二阶段发现错误的回访只需要短途旅行。这种安排使得现场核对相对容易。此外，每个集群的选择都相对集中，以便相对容易地进行监督。最后，计划在雨季难以到达的地区在旱季进行覆盖。在实地工作的定期间隔期间，包含每个完成周期GLSS III数据的软盘被返回到阿克拉总部。最终的总表使用SAS软件包生成。