| Objectives of a Structured Survey |
The general objective of structured surveys in rural areas is to obtain quantitative data on rural life. Based on concrete objectives, different types of structured surveys can be distinguished:
|
Socio-economic surveys, or household surveys, aim at establishing the economic and social characteristics of the target population, including: demographic data, access to and use of services and infrastructure, sources of income, etc. |
|
|
Farm management surveys provide data on all aspects of the farming system, including: soils, fields and grazing areas, livestock numbers and types, agricultural and livestock inputs, technology, yields or production, markets, etc. |
|
|
Specific surveys: labour surveys aim at determining the amount of labour available in the area (e.g. to see if people can work on rural road projects during the dry season); health surveys can establish the number of people affected by certain diseases; etc. |
|
The data generated can be used to plan rural development projects or to monitor and evaluate them. For a proper monitoring and evaluation system a baseline survey is needed. This is similar to the socio-economic survey but focuses on data about those aspects of rural life that can be expected to be affected by a project. A baseline survey provides the background situation, or benchmark, which can then be used subsequently in the monitoring and evaluation of the project.
|
| Designing a Survey |
A structured survey aims at quantitative research, and at its simplest, is the collection of information via interviews based on a structured questionnaire. It generates quantitative information (data), which can be analysed statistically, and endeavours to remove any bias potentially originating from either the researchers or the respondents.
A fully structured questionnaire leaves no room for any changes during the process of interviewing. This makes it necessary to invest significant time and resources in the initial preparation and testing of the questionnaire.
|
Bias
In the context of quantitative research, bias is viewed as a systematic deviation from validity, caused by errors in measurement or in sampling procedures.
In order to avoid bias, quantitative methods emphasise the use of means of data collection that minimise the researcher's influence and guarantee a standardisation of responses.
|
|
Increasingly, it is possible to design structured surveys which include a series of open-ended questions alongside the more formal questions. This technique can provide valuable information that is difficult to obtain in any other way.
Although the content of a survey - i.e. the questions - can of course vary considerably, in general the following steps are necessary:
|
|
Establish clear objectives for the survey |
|
|
Collect and review secondary information |
Some of this may for example come from previous surveys, for national censuses, or from published research. |
|
Conduct out an Exploratory Survey |
One cannot ask relevant questions if one does not have a reasonable understanding of the situation in the area in question. An exploratory survey is crucial. The researchers must enter this with an open mind and are required to identify the key aspects that are relevant to the objectives of the survey. RRA or PRA techniques can most usefully be used for this important stage. |
|
Decide which data are needed |
Concentrate on one topic (e.g. smallstock) but always be aware of related topics. |
|
Decide the analytical procedure to be used |
One can do a "one-off" survey or repeat it every season, annually, or at some other regular time interval. This is sometimes called a longitudinal survey. Where more than one survey is planned, the same group of respondents can be kept and used repeatedly (a panel survey) or one can select a new group each time. It is also possible to mix these methods and for example keep half of the respondents.
At this stage one should also consider some of the statistical techniques to be used during analysis, and ensure that the data are collected in such a way that they can indeed be fully analysed. |
|
Develop the questionnaire |
Developing a questionnaire requires time and a variety of skills. The wording of the questions should be simple, relevant and unambiguous. Do they need to be translated into one or more local languages? If so, the translated versions need to be tested. It is important to define exactly what is meant by the words used, e.g. what exactly a "household" is, and that all terms are fully understood. |
|
Select the target area
and target group |
(depending on specific circumstances this may need to be done either before or after the development of the questionnaire itself). |
|
Decide on the best time of the survey |
Find out when the stress periods occur, e.g. of seasonal hunger and shortages, or their opposites. |
|
Test the questionnaire |
This is an essential step and usually leads to modifications to the questionnaire, and/or the procedures for conducting the survey itself. |
|
Train enumerators |
Training should concern itself both with communication techniques (establishing a good rapport with the respondents) and with a thorough discussion of the concepts used. |
|
Design the sampling procedures |
Random sampling is sometimes considered preferable from a statistical point of view, but is not always possible or practical. A systematic sample can usually be carried out with greater ease. Systematic sampling has an advantage over random sampling in that it is capable of producing information on spatial distribution as well as quantities.
If no maps or lists of inhabitants are available, or if the available maps or lists are inaccurate or biased in some way, e.g. excluding unregistered houses or settlements, it is preferable to use less rigid sampling methods. One can for example select particular areas and visit every third or fifth house. |
|
Conduct the interviews |
|
|
Analyse the results of the Survey |
|
|
Problems to be aware of |
For many reasons people may not provide accurate answers to questions asked by enumerators:
- They can be afraid of political complications.
- They can be short of time to explain everything.
- They can be afraid of having to pay taxes, e.g. based on the number of livestock they own.
- They can give "desirable" answers in order to please the enumerators, or provide them with what they think they want to hear. [This same criticism can also be raised in relation to participatory methods, where it can present significant problems in interpreting the results].
- They can give answers which they think will assist them to be among the beneficiaries of expected projects.
- They can be afraid to show they do not understand a question or know the answer, and so simply make up an answer.
- Sometimes people do not even provide their real name to enumerators, let alone the correct number of sheep, goats and chickens they have. However, in most situations it is not necessary to record any names - so don't request this information if it is not needed.
Collecting data from people and analysing them elsewhere may alienates the target audience. It denies the target group a say in its own development. In reaction to this problem, in the last decade some progress has been made in giving more feedback to the people or communities who provided the data. [The reports from participatory surveys also often fail to provide feedback to the communities concerned].
Poor rural areas have poor databases, and may have little in the way of reliable secondary data or background information. Poor people have many reasons not to tell the truth to enumerators. Poor people lead complex lives and have little time to explain this in terms of precoded questions. These poor databases often make it difficult to achieve a representative sampling of the people to be interviewed.
A common problem is the "invisibility" of women. In the design of surveys issues which are essential for women and their position must not be left out. In the actual implementation all to often women are underrepresented as enumerators and as respondents - where possible make sure that women are included in the team of enumerators.
|
|
