In deciding the form and content of a nutrition survey, the social structure of the community of an area must be considered. Social structures between people may be built on biological, economic, political, religious and/or geographical grounds. These social structures are assembled by various levels. Structures are built starting with the smallest survey unit (the individual) through families, communities, etc. up to the level of the population of a country or region (see table 2).
Great differences can also be found between urban and rural areas. Although in both urban and rural areas the smallest unit is the individual (survey level A), at the next higher level (survey level B) the social organizational structures start to differ from each another. In rural areas, the household unit is a family or extended family. In urban areas, households consist of single families or individual persons.
The complexity of the social structure is also important. The higher the survey level, the more complex the social organizational structure. In addition, these structures grow even more complex with increasing numbers of people in an interdependent living environment.
As shown in the following table, the quality of a higher survey unit is different from the sum of the subordinate survey units. In order to describe the quality of the whole, all levels must be described. If one wants to undertake a nutrition survey of a population, this principle also applies to the nutritional assessment of individuals. Not only should the sum of individual units be assessed, but the higher structural levels must also be assessed.
Therefore, a survey must allow for the special characteristics at the level of the individual, household, and village or suburb. The sum of information on individuals alone does not provide a complete picture of a family situation, and similarly the sum of information on families does not do so for villages or suburbs, etc.
Table 2. Observable characteristics
related to the survey level
|A||Individual||Individual||Age, size, sex|
|B||Household||Household||Size, income, religion|
|D||Valley (geographical unit)||City/town, part of a metropolis||Health structures, types of schools|
Information concerning survey levels A and B has to be obtained by direct interviews and data collection from the target group. Variables concerning individuals will be described in sub-chapters 3.2.10 - 3.2.13, and those concerning households in sub-chapters 3.2.1 - 3.2.9. Inquiries should be made at the appropriate agencies concerning the variables at survey levels C, D and E before household interviews are conducted. Part 2 discusses how to obtain this information.
In a nutrition survey the data collection is divided into
3.1.1 Initial interactions with community
To obtain an actual view of the nutritional situation of a target community and to lay the foundation for intervention, an essential requirement is that the survey is planned, carried out and evaluated only in cooperation with the community concerned.
This means that from the very beginning,
a dialogue is sought with representatives of the target population.
In this way the community will be informed as to the purpose of the survey
and its form and contents. Before any planning of the survey method and
its contents, time and cultural requirements must first be defined. If
parts of the survey cannot be carried out due to the wishes of the community,
and therefore, some information cannot be gathered, this objection must
be accepted and respected. For example, taking blood samples from children
or asking questions concerning income is often rejected by community members.
During, or at the latest upon completion of, a survey a dialogue must be initiated with the local population concerning the results of the survey. At this opportunity
An important principle in planning and evaluating a survey is to distinguish between the following community groups:
A distinction has to be made between nutritional physiological risk groups and social risk groups. Physiological risk groups are, for example, small children, school children, nursing and pregnant mothers, elderly people, and men and women who perform physically hard work. Examples of social risk groups are the landless, low income or other groups displaying particular forms of behavior (e.g., nutritional or health practices).
During planning and evaluation the distribution of individual groups among the other groups needs to be defined. From the baseline survey the proportion of the risk group to the total beneficiary group can be estimated and the proportion of the beneficiary group to the total risk group can similarly be estimated.
This is illustrated in the following
1. Planned degree of intervention:
Proportion of total beneficiary group that is at risk = a / (a + b) * 100
With a known degree of intervention, the efficiency of the assistance provided can be determined; in other words, the proportion of efforts that reach those who need them will be established.
2. Planned coverage of intervention:
Proportion of the total risk group that receives benefits = a / (a + c) * 100
The coverage of intervention will determine the proportion of the population at risk that will benefit from intervention.
As described earlier, a project would
be overtaxed both in terms of schedule and expertise if all risk groups
were to be considered in a nutrition survey, therefore:
The reasons for this are as follows:
Additionally, the type, severity and causes of risks differ with each age group of children.
3.1.3 Cross-sectional vs. longitudinal surveys
The survey design needs to be tailored in a way that the maximum amount of information can be collected regarding the impact of the planned intervention. For example, the impact of a oil palm plantation project on the nutritional situation of its smallholders/producers is studied according to table 3.
Table 3. Survey design of nutrition
|Intervention Group||Group for comparison|
(start of project)
|1. Follow up survey||(C)||(D)|
|2. Follow up survey||(E)||(F)|
A baseline survey at the start of the project (A) gives information about the nutritional situation of the project population. After several years a follow-up survey will be carried out within the same area or population (C) to study the project's impact on the nutritional situation. However, the comparison of the nutritional situation between (A) and (C) cannot be used alone to measure the impact of a project because several additional factors that have changed during the time of the implementation of the project may have influenced the nutritional situation of the population. It is difficult to attribute the project's impact to the change of the nutritional situation. If the general condition of the population in the area, but not in the project, deteriorated during the intervention period, for instance due to climatic or political factors, no change of the nutritional situation of those in the project area has to be regarded as a success because the nutritional status of the rest of the project's surrounding population has decreased. Therefore, it is recommended to include in the baseline survey a population that will not be included in the project (B). Often projects are carried out in phases and start with a smaller population group for a later expansion. The group of comparison (B) could be identified as a future expansion area. During the follow-up survey the population of the group for comparison will be surveyed again (D) and this survey can be used as their baseline survey. After several years a follow-up survey will be carried out within the same population (F) to study the project's impact on the nutritional situation of group (D).
If no comparison group is available, the results must be analyzed with extreme caution because it is difficult to separate changes due to the project from those independent of the project. If, for example, no improvement in the nutritional situation is observed within a specific time, it cannot be definitely concluded that the measures have not worked. It is possible that general worsening conditions have caused a general decline in the nutritional situation of the entire region. Under these conditions, if the original nutritional situation has been maintained, the results have been positive. Alternatively, observed improvements in the nutritional situation may not have much meaning by themselves as long as they are not compared to the overall situation. It is conceivable that improvements in nutritional indicators can be attributed to the general situation in the region and not to the measures undertaken in the project.
To further complicate matters, comparison groups are never fully identical to intervention groups. This problem can be reduced statistically by studying many individuals. However, this significantly increases the cost of a survey.
It can be said that in principle:
It has to be recognized that the assessment of the nutritional situation of population groups for the purpose of comparison may create ethical problems, since no immediate actions are planned to improve their situation (see section 126.96.36.199).
There is a special type of long-term
study undertaken with the same individuals over the period of the survey.
This survey method can be more economical, but in certain cases, more difficult
to carry out. The primary advantage of this method is that a smaller number
of individuals provides statistically valid results. Such a study cannot
always be carried out because it is necessary to ensure that the same individuals
will be available throughout the entire study. However, at the outset of
an intervention this cannot be guaranteed for such reasons as migration,
death, etc. This type of study also raises concerns about whether the results
are due to the intervention or to the repeated surveys. Experience has
shown that surveys carried out without any implementation of intervention
measures can lead to an improved nutritional situation for the survey group.
188.8.131.52 Baseline survey
Each nutrition-related project should start with a nutrition survey. This first (baseline) survey is much more extensive than a follow-up survey, not only because the nutritional condition of the community has to be assessed, but also the possible causes (such as inappropriate weaning practices or child nutrition, frequent diarrheal diseases, etc.) and important determining factors must be identified. Only if the causes of undernutrition are known, can sensible and equitable intervention measures be planned and carried out. The success of intervention can be checked and evaluated by establishing the determining causes of the nutritional condition. The analyses of causes and determinants will be discussed later in further details (see Chapter 4.7).
Before each baseline survey there must be a comprehensive collection of background information (see part 2 of the guidelines). Because the surveys are carried out in developing countries, where it is often difficult to obtain scientific publications, literature research should be conducted in advance. In addition, at least one week should be allocated for the collection and analysis of information in the project country.
Most projects start with an orientation phase in which specific sector data for the project are assessed to analyze the starting situation. If during the orientation phase data are obtained which are also relevant to the nutritional baseline survey (e.g., socioeconomic data), perhaps both surveys can be combined. If the nutrition baseline survey is a part of an overall socioeconomic survey, the assessment of socioeconomic data can also be carried out by other persons in the survey team. Care should always be taken that the community is not overburdened with unnecessary and redundant questions.
The relationship between the amount of information collected and the quality of the information collected is not linear. Frequently, the amount of information climbs sharply with the first few variables, but the law of diminishing returns takes effect with less and less additional information obtained from increasing numbers of variables. Finally a saturation point is reached and from there on less information actually results from the growing number of variables. A basic underlying principle is that the patience of the interviewee and the accuracy of the enumerator declines with increasing numbers of variables, and thus the likelihood of clean analysis and interpretation of data is reduced. A decision must therefore be made at some point concerning the benefits in terms of information to be derived from further variables.
Only those variables that are relevant
to the objective of the survey (e.g., selection of intervention measures,
determination of intervention groups, measurement of project results) should
be assessed. Therefore, not every variable that appears interesting should
be included in the survey.
The timing of the baseline survey can be very important. In many rural areas there is a seasonal influence on the nutritional situation of children, because in agricultural areas the nutritional situation is often under greatest strain before the harvest season.
In urban areas a similar situation can be found with the families of employees. With incessant inflation, wages often do not increase gradually, but are only adjusted to the price of necessities after a long period. The nutritional situation is undoubtedly worse before such an adjustment in wages than afterwards.
It is very informative if a survey can be carried out at a critical time such as the period before harvest, before an expected general increase in wages, or after a period of drastic increases in food prices. Such timing will undoubtedly bias results negatively, however, it is far better to obtain information on the situation during "bad" than during "good" times, as it is not the "good" times one needs to worry about. Also, in an agricultural community, the period before harvest is a desirable time to carry out a baseline survey because family members are free from work in the fields and therefore are available to spend time participating in a survey.
As the timing of the survey depends
on the progress of the project (see chapter 1.4), it is not always possible
however to carry out the survey precisely during a critical phase.
184.108.40.206 Follow-up Survey
If a follow-up survey for evaluation is to be carried out, the relationship between the effect of the project on the nutritional status and the contributing poverty situation of the target group should be assessed.
A follow-up survey is faster than a
baseline survey because most of the parameters, such as the survey area,
the beneficiary group, the infrastructure, etc. have already been established
in the baseline survey. It is also unnecessary to conduct any preliminary
sampling as the questions used in earlier survey(s) should be used again.
Ideally, the former survey staff will be available to conduct the follow-up
survey and therefore can apply their earlier training. Consequently, only
a short retraining period will be necessary.
This means for example, variables with the same coding as in the baseline survey must be used, and that as far as possible the survey design (such as sampling procedure) remains constant. To achieve this, the report of the baseline survey must be carefully documented and reviewed as it must serve also as a basis of the follow-up survey..
Occasionally a project or intervention
area will extend into a former comparison area. In this case the follow-up
survey in the former comparison area can also serve as the baseline survey
for a future evaluation phase. A new comparison area should be found.
220.127.116.11 Groups for comparison
As described at the beginning of this sub-chapter, when possible and ethically justifiable, comparison groups (or "control groups") should be employed in nutrition baseline surveys. A comparison group differs from the survey group only in that no project intervention measures have been carried out there. The comparison area should be selected so that it lies outside the influence of the activities of the project as much as possible.
There is an ethical aspect in comparison
group surveys. Because no intervention is carried out on the comparison
group these individuals are left with their problems for the sake of comparison.
Before arriving at a survey scheme, this point has to be discussed in depth
with as many project representatives as possible and decided in all seriousness.
No survey without service!
This means, for example:
18.104.22.168 Frequency of the surveys
Determination of the frequency of surveys is a compromise between the requirements of accuracy in the M+E (monitoring and evaluation) system and the completion of the project within the deadline.
The interval between the baseline survey and the follow-up survey, as well as the frequency of the follow-up surveys, depends on:
As a rule, it is impossible and unnecessary to investigate all members of the target community to determine the prevalence of malnutrition is present in the community. A representative sample is studied and the results are extrapolated to the entire community. A sample is not representative if sampling systematically favors certain groups of people or makes systematic errors. As this can happen unknowingly, unintentionally, and is only detected during the analysis after the completion of the survey or even never, random sampling is essential. This means that all eligible persons have an equal chance to be included in the survey.
It is rare that nutritional deficiencies are evenly distributed between men and women or among age groups. When one wishes to determine the frequency of a certain problem, one should also define and describe the groups to be investigated, whether by age, sex, place of residence or social status, etc. In the selection of a sample, this means that the sampling should only be done in these groups.
In large survey regions it is difficult
to secure that every person in the group under investigation will be equally
likely to be selected as part of the sample. To ensure this, a complete
list of all inhabitants and their characteristics would be necessary. Since
such a list is rarely available, a solution to this problem is the division
of the area into clusters (sampling units). Clusters of individuals
often arise naturally (e.g., classrooms) or they may be formed artificially
(e.g., geographic clusters). The clusters are then selected at random
and within each cluster investigations are carried out on every member
of the community concerned. If the cluster is still too large, further
random sampling can then be done.
22.214.171.124 Sample size
Calculation of sample size depends on the objective of the nutrition survey. If the sampling is to be done for a survey in a nutrition-related project, in which the prevalence of undernutrition is to be established in a baseline survey, where specifically aimed nutritional intervention is not to be undertaken, the necessary sample size for the survey of anthropometric data can be derived using the following formula:
n = (4 x p x (100 - p)) / 25
where: p = expected prevalence of undernutrition.
If the prevalence of undernutrition in a survey area is estimated at 40%, the sample size (n) should be:
n = (4 x 40 x 60) / 25 = 384
In other words, the survey must be conducted on 384 individuals.
The estimated prevalence of undernutrition can be derived from literature study, data from health services, or ideally from the pilot study. The formula can only be used if:
In self-standing nutrition projects, where intervention is to be undertaken, a different method is used to determine sample size. The following information must be available to ascertain the required sample size for comparison with anthropometric data in a follow-up survey:
The formula used to calculate the sample size is relatively complicated and cannot be understood without an extensive knowledge of statistics. Sample size can more easily be determined using graphs (see the following page), based on Fleiss' illustrated calculations (Fleiss, J.L. (1981) Statistical methods for rates and proportions. Second edition, John Wiley & Sons, New York, pp 38-42).
First one looks for the exponential curve that corresponds to the expected rate of nutritional problems for the surveyed community before intervention. Then on the lower horizontal axis the expected prevalence of malnutrition after intervention is identified and a perpendicular line is followed from this point to the exponential curve. From where these lines intersect one goes horizontally to the left or right vertical axis of the graph, and one can find the respective number of persons needed for sampling. If the prevalence of undernourished children under five is 30% before intervention and the desired goal is to halve this number, i.e., to 15%, slightly fewer than 300 children should be investigated before and after the intervention.
Figure 2. Expected prevalence
(Sample size for the comparison of proportions, a = 0.01; 1-b = 0.05!)
126.96.36.199 Implementation of sampling
Two methods are available for selecting the clusters to sample. The selection of the method depends on the availability of maps of the area.
If 500 children are to be surveyed and in each village there are at least 30 families with an average of 1.5 children under 6 years (i.e., 45 children/village), then 500/45 = 11 clusters must be selected. If there is more than one village or urban district within a single rectangle, one village or urban district in that rectangle must be selected by draw. This can be done by drawing lots, throwing dice, or tossing a coin.
It is known that in developing countries the under-five age group of children generally makes up 18% to 22% of the total population. Consequently, the number of children in each cluster can be estimated. In the following table there are least 77 inhabitants per cluster (Mayama, the least populated cluster) or at least 15 children under five years of age. If a total of 300 children is to be surveyed, we can calculate that 20 clusters need to be surveyed, i.e., 300/15 = 20.To decide which cluster villages should be surveyed, all villages can be assigned numbers and from these twenty can be selected at random. However, this does not consider the differing sizes of the clusters. Therefore, the following procedure is preferred:
The total number of inhabitants in the survey area is divided by the selected number of clusters and the mean number of inhabitants is calculated (In the example from table 4: 99,756/20 = 4,988). Now, a number below the mean is randomly selected from a table (e.g., 507). With this random number a series of numbers is constructed by addition of the mean to this randomly selected number and subsequently to each sum (in the example: 507, 507 + 4,988 = 5,495; 5,495 + 4,988 = 10,483; 10,483 + 4,988 = 15,471, etc.). Using the second column of the cluster list (cumulative population) this series of numbers can be used to identify the villages to be surveyed, that is, all villages that against the selected cumulative number of inhabitants have the lowest difference from the numbers in the series.
Table 4. List of communities
|Cluster or Community||
|T o t a l||99 750|
In our example the clusters to be surveyed are:
After the individual clusters, i.e., urban districts or villages, are selected, the households within the clusters need to be selected. Again, either of the two methods described above can be used for selection of households.
If an accurate city map is available, in which all residential areas are shown, the grid method can be used. The size of the rectangle should be selected so that it contains about 4-8 houses. All rectangles are numbered and the numbers of the rectangles to be sampled are selected at random. In each of these areas (rectangle) one household with at least one child will be surveyed. If individual houses are shown on the city map, the houses in the sampling area should be numbered at random and assigned numbers in sequence starting at number 1. The survey team starts the survey at house 1. If there is no family with children under 5 years, they go on to house 2. If there is still no household with children under 5, the team continues to house 3 etc. When a family with at least one child under 5 is found, the survey is completed in that area and the survey team proceeds to the next sampling area.
If there is no city map, or the map is very unreliable in showing houses, the following option can be used: a central location is chosen in the selected area (intersection, marketplace, etc.), and a coin is tossed twice, once to select north/south and once for east/west, or a number between 1 and 4 is picked from a randomized number table (see chapter 6.5) to choose north/south and east/west. From the same table a number can than be selected between 1 and 25, and the survey will commence at the house corresponding to this number. From this house, sampling will be undertaken at the next 15 houses with children under five years. If the edge of the village is reached before there is the opportunity to visit 15 houses, one starts again in the central location to continue the investigation on the next street in the clockwise direction. A third possibility arises if there is a list of households with children in a village, the listed children can then be selected by draw.
This type of sampling can be undertaken when there is a homogeneous target community. If a survey has to be undertaken with a heterogeneous community group, however, clusters representing certain population groups are the basis of selection. In this way the specific nutritional conditions of that group can be determined. In this case, however, it is not possible to make a statement on the nutritional situation of the total population.
In rural areas village communities
do not always understand why only some of their community should be surveyed.
It may therefore be necessary to conduct the survey including all families,
and consequently correspondingly fewer villages will be surveyed, or more
time will be needed to complete the survey.
3.1.5 Training and supervising of survey personnel
Beyond generating information on the nutritional situation, the implementation of surveys on nutritional status also has a second important task. The importance of nutritional problems as a major constraint of living conditions in developing countries should be emphasized by involving technicians and communities.
The training of technical personnel
for nutrition survey implementation carries a special importance beyond
that of the survey itself.
188.8.131.52. Job description of enumerators and supervisors
There are two distinct groups of technical personnel for conducting the survey:
Supervisors are responsible for the technical quality of the surveys for which either they or the survey teams under their charge are conducting. Furthermore, they function as a liaison between institutions with an interest in the survey and representatives of the target groups.
Knowledge of local government structures and sociocultural characteristics are prerequisites for effective coordination. Only if the supervisor has the trust of the target groups, their representatives, and local authorities, can the survey proceed as planned. Efforts should be made to draw survey personnel from the health sector, or from the agricultural or educational sectors. Thus, it is possible for the respective sectors to gain greater understanding of, and motivation for intervention measures.
If the survey teams are required to
investigate families in their own homes, and the families are widely scattered
over the project area, the supervisors will be able to oversee only a small
number of survey teams. The assignment of survey teams to a supervisor
must be done randomly. The randomized number tables can be used for this
purpose (see Appendix 6.5).
The duties of a supervisor include
advice and control of the accuracy of the survey data.
If differences are evident between
the data collected by the supervisor and those collected by the enumerators,
then obviously the causes must be found and specific measures taken to
prevent further discrepancies.
It is important to maintain constant feedback with representatives of the target community about problems arising during implementation in order to prevent failure and friction.
The organizational structure of a survey is presented in figure 3. The survey leader should be chosen from a sector in charge of the nutritional situation of the population in that country. This is normally the health sector. The training workshop is carried out by the survey leader. The survey leader is responsible for hiring the supervisors, and the supervisors for hiring the survey teams. In smaller surveys, the survey leader may be able to undertake direct supervision of the survey teams, thus eliminating one organizational level. One or two additional persons are responsible for data entry into the computer. Data entry should occur during the implementation of the survey. This has the following advantages:
Figure 3. Organizational chart of field personnel participating in a nutrition survey
Data entry personnel (S01 - S24: Enumerators)
184.108.40.206 Personnel recruitment
The progress of a survey depends just as much on human qualities as on the technical qualifications of the personnel taking part in the survey. Completion of elementary education is a minimum requirement for an enumerator. An enumerator must have reading and writing capabilities, as well as the mastery of basic types of calculations. Previous work experience can replace the minimum formal educational requirements.
Besides education and work experience, an accurate knowledge of the local language is required. People who speak the same language, know the local customs and are familiar with the problems of the target community will more easily win the confidence of the target groups necessary for conducting interviews. As far as possible, the enumerators should originate from the same sociogeographical and cultural area as the target groups. However, it is also possible that foreign enumerators are more accepted than local interviewers. Therefore, in each survey the possibility that distinctions such as religion, ancestry, caste or sex may influence the result of the survey has to be examined.
In all societies there are varying degrees of distinct roles for the sexes. These distinctions have to be considered in the recruitment of enumerators. As the responsibility for household nutrition generally lies with women, it is recommended that women be engaged as nutritional enumerators.
Personality attributes should be considered when selecting an enumerator. Therefore, it is important to study the personalities of the enumerators during the training program to identify potential problems. In addition, potential enumerators should be assessed in relation to the following easily recognizable attributes:
The requirements set for enumerators apply also to supervisors, but also additional educational qualifications should be required. This could include education at a vocational school, or studies continued toward a specific goal after elementary education. Once again, successful previous work experience should be rated over formal education.
Finally, personnel engaged in a survey
should also be motivated through an adequate level of remuneration.
Training of survey personnel
Enumerators must be trained intensively to carry out carefully the following tasks:
The training should stimulate interest through teaching aids. To achieve this, the following materials should be used:
Patience is one of the most important qualities of an enumerator. As target groups are often not accustomed to being asked questions, interviewees can take a long time to answer, and the answers may be limited by the vocabulary of the interviewee and can be very ambiguous.
The communication between the "survey partners" will be enhanced if the enumerator looks at the interviewee with a friendly expression, uses encouraging words, and does not stare incessantly at the questionnaire.
The first concern of the survey leader is to instill in the enumerators the importance of their future activities.
When the person in charge of training has made sure that the enumerators have grasped the importance of the survey for the subsequent program or project, the next step is to make the participants aware of the inherent risks of mistakes caused by:
To assure standardization, the individual questions on the questionnaires should be asked using a consistent predetermined wording. This can be reinforced through practice with different interview partners during training. The other enumerators and the supervisor can then make critical observations of the dialogue and discuss its strengths and weaknesses. These exercises will introduce the enumerators to a variety of interview situations and help them to gain confidence in asking questions and in completing the questionnaires. The amount of time needed during an average survey can be estimated by timing the practice interview session. However, in a field study, it is necessary to allow additional time for the enumerator to establish contact with the target person.
Next the anthropometric measurements
and investigation methods relevant to nutritional physiology (e.g., measurement
of blood hemoglobin levels, examination for xerophthalmia and goiter) will
be practiced, and the data entered on the questionnaires. At this point,
there should be a knowledgeable person supervising the measuring in order
to eliminate potential errors. Measurement of the same person by several
enumerators can help to check for uniformity of measuring results.
A proposed schedule for the training of enumerators is shown on the next page. At the end of the training period, a pilot study must be carried out (see subchapter 3.1.6). Therefore, additional time has to be made available.
Adequate space and audiovisual materials are necessary for the training of survey teams.
Table 5. Proposed training schedule
|Day 1||Enumerators arrive, register and are briefed, etc.|
Introductions between the enumerators and training personnel, time to become acquainted, introduction of the training program and discussion.
Lecture on the general concept of the project, the survey objectives, and the consequences for the target groups and the entire project.
Introduction of the survey plan.
Introduction of the questionnaire.
Step-by-step presentation and memorization of the individual questions (in the predetermined wording).
Interview practice in pairs with survey colleagues (e.g., through role play) followed by discussion of positive and negative types of behavior during the interviews.
Demonstration of anthropometric measurement techniques
Practice with anthropometric measurement techniques.
Determination of intra- and inter-observer errors through repeated measurement of same individual at given intervals by both same and different enumerators
Analysis of measurements with the enumerators, checking for errors, and discussion of any unclear points
"General test" - completion of three questionnaires (including anthropometric measurements) with different survey partners .
Analysis of the surveys with the enumerators, checking for errors, and discussion of any unclear points.
|Day 5||Payment of daily allowances.
Procedure for the collection of information in a household
In the training session the following items should be emphasized:
3.1.6 Pilot testing
Shortcomings and unforeseen factors
are always discovered during a survey, despite careful planning. Such unforeseeable
deficiencies can be considerably reduced by conducting a pilot survey.
The pilot survey has the following objectives:
After the completion of the pilot survey,
a final consultation is held with the enumerators. The aim of this discussion
is to clarify any remaining obscure points and to reach an agreement on
the final procedure and survey to be used.
3.1.7 Scheduling of a survey
The scheduling of the survey is subject to factors outside the control of the planning but which nevertheless have to be considered. In table 6, some circumstances are listed under which a survey needs more or less time.
Table 6. Factors influencing the
time needed for a survey
|Less time spent on the survey||More time spent on the survey|
End of the project
Follow up survey
Experienced survey team
Part of a socioeconomic survey
Family travels to the enumerator
Good transportation infrastructure
Closed village settlements and urban areas
Functioning group structures
Time schedules not interrupted for religious or cultural reasons
Beginning of the project
Inexperienced survey team
Enumerator travels to the family
Poor transportation infrastructure
Open village settlements and nomadic settlements
Non-functioning group structures
Time schedules interrupted for religious or cultural reasons
The chronological progression of an isolated nutrition survey is presented in table 7. In this example, the schedule is based on the maximum time required for a purely nutrition project - estimated to be about 9 weeks (2 months). However, the time taken may be considerably less, depending on local circumstances. The search and study of literature, as well as further statistical analysis, which do not have to take place in the project area, are not included in this table.
Table 7. Example of scheduling of
a baseline survey
|A C T I V I T I E S||
|Review of scientific
Review of reports from national and international organizations
Preparation of the training program for the enumerators
Gathering information at local institutions and from local persons
Collection of structural data
Initial calculation of sampling size
Design of questionnaires
Organization of the survey
Recruitment of enumerators
Talks with target groups and representatives of local organizations and institutions
First test of the questionnaire
Training of the enumerators
Preliminary study and determination of the final version of the questionnaires (pilot study)
Final determination of the sample size
Data collection in the communities
Data entry into the computer
Initial analysis of the survey
Informing the target groups and technical personnel of the survey results
More detailed analysis (analysis of causes and determinants)
3.1.8 Resources needed
The following list is a checklist for quick reference to see if the necessary resources are in place for a nutrition survey. No quantities are given, as these depend on the sample size. The resources in parentheses are not essential, but they may be needed depending on the specific situation.
Equipment and supplies
Despite all desires for standardization, it is necessary for each survey to develop its own questionnaire to be completed by the enumerators with data obtained from the field.
There are three distinct types of questionnaires:
Variables applying equally to individuals within a household, such as the size of the family or the amount of living space, should be recorded on a separate questionnaire for the family (household). Similarly, observable variables applying equally to all families in a village, such a climatic data, should not be recorded on a household questionnaire, but should be entered on a questionnaire prepared for a village or city suburb.
The questionnaires should be standardized to simplify their completion and subsequent use. For this purpose, the following rules should be explained on the following example of a page from a questionnaire (figure 4).
Figure 4. Example of a page from a questionnaire for a nutrition survey
In some cultures, men may have several wives. If a husband has several wives with children, an additional variable, "family unit," must be included in the survey. Thus all family units with the same male head of family are assigned the same family unit number, but individual mothers are assigned unique household numbers.
(Mother with cooking facilities)
In summary, each page of a questionnaire
must bear a household number to avoid accidental exchange during subsequent
EXAMPLE: The age of a child can be ascertained by asking the mother during the interview or by reading a birth certificate. The data obtained will likely be more reliable in the case of reading the birth date from the birth certificate.
For standardization, the type of recording of each variable must be defined and stated after the number of variable.
Variables that result from calculations carried out during the data analysis by computer (e.g., persons per living room, z-scores, or BMI) do not appear in a questionnaire.
The local language of the target community often differs from the national language. In these cases, the questions and answers in a questionnaire should be written in the language that the people of the target community use to communicate among themselves. If the questions were initially written in the national language, the quality of the translation into the local language should be checked by reverse translation back into the national language.
With closed questions, possible answers are predetermined during the pilot study, and therefore, there is the risk of loss of information. Nevertheless, using standardized questions with corresponding answer categories allows for a rapid and meaningful analysis of the entire data set, which outweighs the disadvantages of any possible loss of information. This is particularly true when several enumerators are used.
For these reasons, where the purpose
of a nutrition survey is to identify problems and suggest solutions and
not to carry out basic research, closed questions must be used.
7 or 77 when an answer is given which has not been provided for in the coding, it should be recorded as "other"
8 or 88 when the interviewed person answers "I don't know," or in the case of an observation when it was not possible to identify a clear symptom
9 or 99 when the interviewed person gives no answer or it was not possible to carry out an observation or measurement
- this adds considerably to the workload of data entry and the later analysis, and
- when several enumerators are engaged, it is impossible to maintain standardization.
However, it may be extremely valuable for the improvement of the survey procedure and the later data analysis if additional, qualitative observations are written down. For this purpose, all enumerators should have some additional sheets of paper where notes can be reported.
e.g., Variable code: GENDER
In contrast to the individual, household and structural questionnaires, the variable codes of the supervision questionnaires can be selected within each country since the reproducibility of the collected data should be calculated within each country separately.
Samples of questionnaires can be produced
automatically by the Nutrition Baseline software.
3.1.10 Ethical considerations
The implementation of surveys has ethical implications that need to be considered thoroughly as early as the planning stage. Although a baseline survey is intended to establish the basis for interventions to improve conditions and has no scientific research component, ethical implications must be considered. Ethics are strongly associated with cultural norms and values that may differ from society to society; however, there are some universal issues that have to be respected.