A new sample has to be drawn for each round of Afrobarometer surveys.
Whereas the standard sample size for Round 3 surveys will be 1200 cases, a larger sample size will be required in societies that are extremely heterogeneous (such as South Africa and Nigeria), where the sample size will be increased to 2400. Other adaptations may be necessary within some countries to account for the varying quality of the census data or the availability of census maps.
The sample is designed as a representative cross-section of all citizens of voting age in a given country. The goal is to give every adult citizen an equal and known chance of selection for interview. We strive to reach this objective by (a) strictly applying random selection methods at every stage of sampling and by (b) applying sampling with probability proportionate to population size wherever possible. A randomly selected sample of 1200 cases allows inferences to national adult populations with a margin of sampling error of no more than plus or minus 2.5 percent with a confidence level of 95 percent. If the sample size is increased to 2400, the confidence interval shrinks to plus or minus 2 percent.
The sample universe for Afrobarometer surveys includes all citizens of voting age within the country. In other words, we exclude anyone who is not a citizen and anyone who has not attained this age (usually 18 years) on the day of the survey. Also excluded are areas determined to be either inaccessible or not relevant to the study, such as those experiencing armed conflict or natural disasters, as well as national parks and game reserves. As a matter of practice, we have also excluded people living in institutionalized settings, such as students in dormitories and persons in prisons or nursing homes.
What to do about areas experiencing political unrest? On the one hand we want to include them because they are politically important. On the other hand, we want to avoid stretching out the fieldwork over many months while we wait for the situation to settle down. It was agreed at the 2002 Cape Town Planning Workshop that it is difficult to come up with a general rule that will fit all imaginable circumstances. We will therefore make judgments on a case-by-case basis on whether or not to proceed with fieldwork or to exclude or substitute areas of conflict. National Partners are requested to consult Core Partners on any major delays, exclusions or substitutions of this sort.
The sample design is a clustered, stratified, multi-stage, area probability sample.
To repeat the main sampling principle, the objective of the design is to give every sample element (i.e. adult citizen) an equal and known chance of being chosen for inclusion in the sample. We strive to reach this objective by (a) strictly applying random selection methods at every stage of sampling and by (b) applying sampling with probability proportionate to population size wherever possible.
In a series of stages, geographically defined sampling units of decreasing size are selected. To ensure that the sample is representative, the probability of selection at various stages is adjusted as follows:
The sample is stratified by key social characteristics in the population such as sub-national area (e.g. region/province) and residential locality (urban or rural). The area stratification reduces the likelihood that distinctive ethnic or language groups are left out of the sample. And the urban/rural stratification is a means to make sure that these localities are represented in their correct proportions.
Wherever possible, and always in the first stage of sampling, random sampling is conducted with probability proportionate to population size (PPPS). The purpose is to guarantee that larger (i.e., more populated) geographical units have a proportionally greater probability of being chosen into the sample.
The sampling design has four stages
A first-stage to stratify and randomly select primary sampling units;
A second-stage to randomly select sampling start-points;
A third stage to randomly choose households;
A final-stage involving the random selection of individual respondents
We shall deal with each of these stages in turn.
STAGE ONE: Selection of Primary Sampling Units (PSUs)
The primary sampling units (PSU's) are the smallest, well-defined geographic units for which reliable population data are available. In most countries, these will be Census Enumeration Areas (or EAs). Most national census data and maps are broken down to the EA level. In the text that follows we will use the acronyms PSU and EA interchangeably because, when census data are employed, they refer to the same unit.
We strongly recommend that NIs use official national census data as the sampling frame for Afrobarometer surveys. Where recent or reliable census data are not available, NIs are asked to inform the relevant Core Partner before they substitute any other demographic data. Where the census is out of date, NIs should consult a demographer to obtain the best possible estimates of population growth rates. These should be applied to the outdated census data in order to make projections of population figures for the year of the survey. It is important to bear in mind that population growth rates vary by area (region) and (especially) between rural and urban localities. Therefore, any projected census data should include adjustments to take such variations into account.
Indeed, we urge NIs to establish collegial working relationships within professionals in the national census bureau, not only to obtain the most recent census data, projections, and maps, but to gain access to sampling expertise. NIs may even commission a census statistician to draw the sample to Afrobarometer specifications, provided that provision for this service has been made in the survey budget.
Regardless of who draws the sample, the NIs should thoroughly acquaint themselves with the strengths and weaknesses of the available census data and the availability and quality of EA maps. The country and methodology reports should cite the exact census data used, its known shortcomings, if any, and any projections made from the data. At minimum, the NI must know the size of the population and the urban/rural population divide in each region in order to specify how to distribute population and PSU's in the first stage of sampling. National investigators should obtain this written data before they attempt to stratify the sample.
Once this data is obtained, the sample population (either 1200 or 2400) should be stratified, first by area (region/province) and then by residential locality (urban or rural). In each case, the proportion of the sample in each locality in each region should be the same as its proportion in the national population as indicated by the updated census figures.
Having stratified the sample, it is then possible to determine how many PSU's should be selected for the country as a whole, for each region, and for each urban or rural locality.
The total number of PSU's to be selected for the whole country is determined by calculating the maximum degree of clustering of interviews one can accept in any PSU. Because PSUs (which are usually geographically small EAs) tend to be socially homogenous we do not want to select too many people in any one place. Thus, the Afrobarometer has established a standard of no more than 8 interviews per PSU. For a sample size of 1200, the sample must therefore contain 150 PSUs/EAs (1200 divided by 8). For a sample size of 2400, there must be 300 PSUs/EAs.
These PSUs should then be allocated proportionally to the urban and rural localities within each regional stratum of the sample. Let's take a couple of examples from a country with a sample size of 1200. If the urban locality of Region X in this country constitutes 10 percent of the current national population, then the sample for this stratum should be 15 PSUs (calculated as 10 percent of 150 PSUs). If the rural population of Region Y constitutes 4 percent of the current national population, then the sample for this stratum should be 6 PSU's.
The next step is to select particular PSUs/EAs using random methods. Using the above example of the rural localities in Region Y, let us say that you need to pick 6 sample EAs out of a census list that contains a total of 240 rural EAs in Region Y. But which 6? If the EAs created by the national census bureau are of equal or roughly equal population size, then selection is relatively straightforward. Just number all EAs consecutively, then make six selections using a table of random numbers. This procedure, known as simple random sampling (SRS), will ensure that each EA will have an equal probability of being sampled.
If the PSUs'/EAs have different population sizes, however, then random sampling must be conducted with probability proportionate to population size (PPPS). The idea here is that units with larger populations should have a proportionally greater chance (probability) of being chosen. The PPPS method is not difficult to use and is described in Appendix 6.
Once EA's have been randomly selected they should be plotted on a national map. Use this map to plan out the deployment routes for the various field teams. In some cases, a few EAs may be so inaccessible or so dangerous that substitution of PSUs becomes necessary. As long as PSU substitutions never constitute more than 5 percent of all PSU's it is acceptable to make them. The best method is to randomly draw another EA in the hope that it will fall in a more convenient location. Please record which EAs are substitutes and justify why they were substituted. If more than 5 percent of PSUs require substitution, then the NI should discard the entire Stage 1 sample and draw a new one.
In some countries, the NI may be concerned that a random sample might miss a politically important minority group. Or, even if this minority is represented in the sample in accordance with its share of the national population, there may be too few cases to make reliable generalizations about the attitudes of this group. Under these circumstances, over-sampling is permissible, as we did in Round 1 for the Toaureg, Ijaw, and Coloured minorities in Mali, Nigeria and South Africa respectively. Purposive over-sampling will also be required as a condition of one donor's funding in Round 3; USAID wishes to gather extra information on certain regions where their projects are located, probably in Mali, Mozambique, Senegal, South Africa, and Zambia. Note that the over-sample should be coterminous with a given sampling stratum, usually a region. The NI should consult the relevant Core Partner about any planned over-sampling and keep detailed records that allow correct weighting factors to be calculated to correct for over-sampling at the stage of data analysis.
Additional Cluster (Optional)
In countries where regions are too numerous or too scattered to provide a logistically feasible sampling frame, an additional stage of clustering can be considered, as follows:
* Choose a suitable geographic unit between region/province and EA: e.g. administrative district. In large countries, it may not be practical to visit all districts or even all regions. Number and stratify all districts and, using PPPS, randomly choose a subset of these districts. Preferably, the subset should not be less than half of the total number of districts in the country. And the subset should always cover all relevant social variations nationwide.
* A population limit shall be set for districts that should be self-representing (i.e. large districts which must be represented in the sample). Self-representing districts will thus have a probability equal to one of inclusion in the sample.
* Once PPPS is applied, other districts will have a probability proportional to population size of inclusion in the sample.
Additional Stratum (Optional)
In urban areas that have extremely diverse housing patterns, the NI may choose to add an additional layer of stratification to increase the likelihood that the sample does not leave out high-density (especially informal) settlements. Using a street map, a city or town can be divided into high- medium- and low-density areas. It can then be required that PSUs are represented equally (or better yet, in proportion to population sizes, if these are known) within the sample for that city or town.
STAGE TWO: Selecting Sampling Start Points (SSP's)
Within each PSU/EA, Field Teams travel to a randomly selected sampling start point (SSP). Thus the number of start points is the same as the number of PSU's (150 or 300). A sampling start point (SSP) is required so that interviewers know where to start random walk patterns within each PSU (see next section). This procedure has the effect of further clustering the sample into manageable areas that are reachable on foot or by a short vehicle ride.
Either in the office or in the field, the Field Supervisor (FS) selects the SSP using one of the following three methods.
The ideal method
If the FS is able to obtain a list of all households in a selected EA, then this should be done. Possible sources include the national census bureau or the office of district administrator or local government authority. Once a list is available, the field supervisor should randomly (using a random numbers table) choose eight households, and send one Interviewer to each. A detailed map showing all households in the EA and matching them with the listed names is necessary for this method.
(Note: If this method is used, it is not necessary to apply Stage Three: Selection of Households. Go straight to Stage Four: Selection of Respondents).
An alternative method (where maps are available for the PSU)
If the census bureau has provided EA maps, the FS can randomly select a start point using a grid. The FS places a ruler with numbers along the top of the map and a ruler with numbers along the side of the map. He/she then uses a table of random numbers (or a set of numbered cards) to select a number for the top axis and a number for the side axis, resulting in a random combination (e.g. "9 and 6.") A line is then drawn on the map horizontal to the number chosen on the side, and another line is drawn vertical to the number chosen on the top. The point on the map where these two lines intersect is the sampling start point. The SSP is marked on the map, and given to the field team for that area. The fieldwork team then locates the nearest housing settlement to this point, and travels there (or as near as they can to the point). In rural areas, finding the SSP may require the field team to consult with local residents.
Because we never know in advance the actual condition on the ground in all the PSU's, the FS may need to choose a second sampling start point as a reserve or substitute if the SSP is inappropriate or inaccessible.
Another alternative (where maps are not available)
When maps are not available for the selected PSU, the following procedure should be used. The FS contacts a local government councilor or another official knowledgeable about the area. This person is consulted to determine how many housing settlements (e.g. villages) are in the PSU. These settlements must have identifiable boundaries that do not overlap with one another. These settlements are numbered and, using numbered cards, the FS asks the informant to randomly select one card. The settlement identified by the selected number is the settlement where the interviews will be conducted.
IMPORTANT: At the start point, then the FS must be certain to preserve randomness, by rotating the place where Interviewers begin their random walk pattern. If the Team starts on a main road at one SSP, they should start off the road at the next SSP. If the Team starts in a central place (like a school) in one EA, they should start in a peripheral place in the next EA. And so on. The logic of random sampling is to avoid ANY kind of pattern in the units selected at any stage.
STAGE THREE: Selecting Households
Having arrived at the sampling start point, the Team is ready to select households.
For the purposes of the Afrobarometer, a household is defined as a group of people who presently eat together from the same pot. By this definition, a household does not include persons who are currently living elsewhere for purposes of studies or work. Nor does a household include domestic workers or temporary visitors (even if they eat from the same pot or slept there on the previous night). And, in practice, we want to select our respondent from among persons in the household who will be available for interview on that same day.
In multi-household dwelling structures (like blocks of flats, compounds with multiple spouses, or backyard dwellings for renters, relatives, or household workers), each household is treated as a separate sampling unit.
IMPORTANT: The third (household) and fourth (respondent) stages of sampling are conducted by Interviewers. Interviewers must be carefully trained and supervised to ensure that they follow Afrobarometer sampling instructions to the exact letter. These sampling instructions are summarized below and spelled out on the first two pages of every questionnaire. Field Supervisors are responsible for ensuring that their teams of Interviewers understand their parts of the sampling methodology and execute them correctly.
T he method for selecting households is as follows:
In well-populated urban and rural areas, with single-dwelling units:
Starting as near as possible to the SSP, the FS should choose any random point (like a street corner, a school, or a water source) being careful to randomly rotate the choice of such landmarks. The four Interviewers should be instructed to walk away from this point in the following random directions:
The Walk Pattern : Interviewer 1 walks towards the sun, Interviewer 2 away from the sun, Interviewer 3 at right angle to Interviewer 1, Interviewer 4 in the opposite direction from Interviewer 3, etc. If the Team contains more than four Interviewers, then the FS should take them to another randomly selected place near the SSP to begin their walk patterns.
When interviews are to be conducted during the night by the whole team (excluding call backs),the team should use the moon or some other random landmark to begin the walk pattern (Field Supervisors should just make sure that interviewees disperse in directions opposite to each other).
Each Interviewer should use the day code to establish an interval (n) for household selection. The day code introduces randomness into the interval. It is calculated by adding together the numbers in the day of the month as follows. On the 5 th, 14 th and 23 rd of the month the interval would be 5, but on the 6 th, 15 th and 24 th it would be 6. And so on. On some days (the 1 st and 10 th of the month) the Interviewer moves to the adjacent dwelling structure (because the sampling interval is 1). On the 29th of the month the Interviewer must leave the widest gap, selecting only every eleventh house.
In every case, the Interviewer selects the nth house on the right.
In well-populated urban and rural areas, with multiple-dwelling units:
If the start point is a block of flats, or if the walk pattern includes a block of flats, then the Interviewer should start on the top floor and work his/her way downwards, stopping at every nth flat on the right. In an exception to the normal walk pattern, which only refers to blocks of flats, the Interviewer should only visit alternate floors of the block.
18.104.22.168 In sparsely populated rural areas, with small villages or single-dwelling farms:
In such areas, there may be only a few households around a given start point. We do not wish to over-cluster the sample by conducting too many (e.g. all 8) interviews in one small village. In these cases, the following guidelines shall apply: If there are 15 or fewer households within walking distance of the start point, the field team shall drop only one Interviewer there. If there are 16-30 households within walking distance of the start point, two Interviewers can be dropped there. (If there are more than 50 households, the whole team can be dropped off as usual). If only one or two Interviewers can be dropped at the start point, the rest of the team should drive to the nearest housing settlement within the same EA and closest to the SSP, where the next one, two or three Interviewers shall be dropped according to the same rule. And so on.
In sparsely populated rural areas, with commercial farms:
In countries where commercial farms are large and contain populous settlements of farm workers, effort should be made to avoid collecting all eight interviews for that EA on one farm. To do this, the field supervisor should drop two Interviewers at the first farm (either the first randomly chosen from a comprehensive list of farms within the EA, or the first nearest the randomly selected start point), and then drop the remaining two Interviewers at the next farm. Once the first two are finished, they are moved to another farm for two more interviews, and the same with the second pair, so that eight interviews are obtained from four separate farms in each EA. It is important that all selected farms are within the selected EA. Households should be chosen from lists of households on the farm, or by using a standard random walk pattern. Remember to include both the farm owner's and farm workers' dwellings on the lists or on the walk pattern. Once the teams' eight interviews are completed, the field supervisor should move the team on to the next selected EA and repeat the procedure.
Interviewer's second interview
In a Team of four, each Interviewer is to obtain two interviews per EA (4 Interviewers x 2 interviews = 8 interviews, the quota for the EA). After completing the first interview, he or she should follow the same procedure as before. He/she continues walking in the same direction and chooses the nth dwelling on the right (where n = the day code). And so on. If the settlement comes to an end and there are no more houses, the Interviewer should turn at right angles to the right and keep walking, again looking for the nth dwelling on the right. This procedure is repeated until the Interviewer finds an eligible dwelling containing an eligible household.
TAGE FOUR: Selecting Individual Respondents
Once the household is chosen, the Interviewer is responsible for randomly selecting the individual respondent within the household who will be interviewed.
To ensure that women are not underrepresented, the Afrobarometer sets a gender stratum of an equal number of men and women in the overall sample. To accomplish this stratum, the gender of respondents is alternated for each interview. First, the Interviewer determines from the previous interview whether a man or a woman is to be interviewed. The Interviewer then lists (in any order) the first names of all the household members of that gender who are 18 years and older, even those not presently at home but who will return to the house that evening. From the list (which is numbered, see p. 2 of the questionnaire), the interviewer randomly selects the actual person to be interviewed by asking a household member to choose a numbered card from a blind deck of cards. The interviewer should interview only the person whose number is selected and no one else in that household.
If the person selected refuses to be interviewed, the Interviewer replaces the household by continuing the walking pattern and again selecting the nth dwelling on the right (where n = the day code).
Note: In the Afrobarometer, we substitute households, not respondents. Under no circumstances must the interviewer substitute another individual in the same household for a respondent selected randomly by means of the numbered card method. It is not acceptable, for example, to substitute a spouse, parent, child, sibling (or domestic worker or visitor) in the same household for a selected respondent who happens not to be at home at the time.
If there is no one at home in the selected household on the first try, the respondent should make one call-back later in the day. Or, if the designated respondent is not at home, the Interviewer should make an appointment to meet them later in the day. Again, a call-back will be necessary in order to find the selected respondent and to conduct the interview. It is also acceptable for the Interviewer to enquire about the whereabouts of the selected respondent (they may perhaps be at work) and, if nearby, to walk to that place to conduct the interview.
If the call-back is unsuccessful, say because the respondent has still not returned home for the appointment, then, and only then, the Interviewer may substitute the household. If the house is still empty or the selected respondent is not at home at the time of the call-back, the Interviewer must substitute that household with the very next household found in the direction of the walk pattern. This slight change in the walk pattern is necessary under these circumstances since the Interviewer may already have had a successful call earlier in the day in the household that is located at the sampling interval.
Reducing Household Substitutions
Round 3 draws on experiences from Round 2. All substitution figures above 5 percent are considered high in the Afrobarometer surveys. We would urge NIs to reduce the substitutions, whether for Primary Sampling Units (PSUs) or households through better planning.
Many household substitutions seem to occur because of the timing of the interviews. Our data show that most interviews take place between 8:00 am and 6:00pm. We can minimize substitutions through the following means:
a. Plan around the working timetables of rural or urban communities. This means scheduling interviews to take place perhaps towards the end of the day in some areas.
b. In urban areas, gender strata are often difficult to meet because a lot of men are at work, especially when interviews are conducted during the week. We therefore advise that interviews in urban areas be spread to include weekends. When planning deployments in urban areas, ensure that at least one day of interviews falls on a weekend.
c. If a minority language group is in the sample, NIs need to plan ahead to ensure that field teams have the right translations of the questionnaire. This means drawing the sample well before the other fieldwork
Table 1. Afrobarometer Surveys, 1999-2006
Round 1 Round 2 Round 3
Fieldwork Dates Sample Size Fieldwork Dates Sample Size Fieldwork Dates Sample Size
Botswana Nov-December 1999 1200 July-August 2003 1200 May-June 2005 1200
Ghana July-August 1999 2004 Aug-September 2002 1200 March 2005 1197
Lesotho April-June 2000 1177 February-April 2003 1200 July-August 2005 1161
Malawi Nov-December 1999 1208 April-May 2003 1200 June-July 2005 1200
Mali January-February 2001 2089 Octr-November 2002 1283 June-July 2005 1244
Namibia Sept-October 1999 1183 Aug-September 2003 1200 February-March 2006 1200
Nigeria January-February 2000 3603 Sept-October 2003 2400 Aug-December 2005 2363
South Africa July-August 2000 2200 Sept-October 2002 2400 February 2006 2400
Uganda May-June 2000 2271 Aug-September 2002 2400 April-May 2005 2400
Tanzania March-May 2000 2198 July-August 2003 1200 July-August 2005 1304
Zambia Oct-November 1999 1198 June-July 2003 1200 July-August 2005 1200
Zimbabwe Sept-October 1999 1200 April-May 2004 1200 October 2005 1048
Cabo Verde May-June 2002 1268 March-April 2005 1256
Mozambique August-October 2002 1400 June 2005 1198
Kenya Aug-September 2003 2400 September 2005 1278
Senegal Nov-December 2002 1200 Sept-October 2005 1200
Madagascar May-June 2005 1350
Benin April-May 2005 1198