Probability sampling is considered as a gold standard approach for household service and it requires to properly sample your primary sampling units. Typically, the cluster as well as the household. This is necessary to ensure unbiased selection of household and produce a sample that represent the whole population. However, people often struggle to sample both cluster in a household appropriately. We develop a desktop-based application called RADAR Sample selector or you can download from the RADAR project website. This is the website or page from which you will be able to download this app. RADAR Sample Selector is a step-by-step or Windows-based or desktop application of selecting a cluster and selecting household when you choose the appropriate Window tab. You can decide on the cluster tab or the household tab. You're also able to select a desired language from the language tab. There are two language English and French or you will go to selector area of language your interest in. For this presentation, we are going to focus on the use of the tool to select a household since subject cluster sampling is described elsewhere. After clicking on the household tab, from here, you will have access to a RADAR household sampling application welcome window. From the welcome page, read carefully the information and instruction about the sampling material that will be used by the tool as well as the required inputs and guidance on the sampling method and how to save outputs. The main input for household sampling is the household sampling frame, which is a complete and updated list of all the household and the location in selected cluster of the study area. Typically, the household sampling frame is obtained from an enumeration mapping list of all the household in your study area. Mapping or household listing may not be required if there is an existing or recent and up-to-date or complete list of household according to the definition of household in your study. Let's see how a household sampling framework looks like. This is an example of household sampling frame where there are some data related to the region, the cluster, the dwelling, and also to know whether your dwelling is a residential or non-residential dwelling. Also, assign a household number to each of the household that are being enumerated within the cluster. This household number is very important since this is the one we are going to sample and they decided which household we are going to use for our study. We also may be able to have additional information on the household sampling frame like the name of the head of the household location description of the household. To access to the next page of your radar sample selector tool, just click to continue to householder page. On this page, which is very important, you'll be allowed to enter the inputs. As input enter the cluster number, here, the cluster name, and also the total number of households you found during mapping or enumeration for each cluster. An easy way to do that is to have a pre-prepared input template where you already enter these data and you just need to copy paste. This is an example where we just need to copy all this and then paste on our tool. We are able to paste our information from the pre-prepared input template where we have the cluster number and also the cluster name. For each cluster, we also have the total number of household that are been enumerated during the mapping or enumeration. After completing this data you need to indicate how many household you want to sample by cluster. You can sample up to 40 household by cluster. But typically we recommend to not sample more than 30 household per cluster to avoid excessive clustering or larger design effect. For this example let's use the default value which is 30 household for the demo. You can also select a sub-sample of household from your initial sample of household. This is relevant when you have a sub-sample of household for which specific questionnaires or service module apply. Sub-sampling is often carried out for surveys, including, for instance, anthropometry module, are many questions which require to interview eligible individual from a sub-sample of household. For a this demo let's pick 20 household as a sub-sample per cluster. After entering the required input you can save your output of file from here. But by clicking on Continue to Sampled Household, you will be able now to preview the sampled household. Let's go and preview the sampled household by clicking here. Once you click on continue to sampled household, you are able to preview the household which has been sampled by cluster. Let's pick the example of cluster number 1. For cluster number 1, for which there were 77 household enumerated during the enumeration. We ask the tool to sample 30 household. From the list of 77 household the tool was able to select the household for which the enumeration number was 2, 5, 7, 10, 12, etc. You remember this number I was talking about. The household number from the mapping. It will be now the corresponding number to this column. We have the 30 households that have been selected for this cluster. You also have additional information like the sampling interval, as well as the random number to select of affairs of the household. Since we use systematic random sampling metadata to sample household. To preview sub-sampling the household, click on Continue to sub-sample the household. You will also be able to see the 20 households we asked the tool to sub-sample. These are for cluster number 1, household of the enumeration number 2, 7, 10, 15, 17, 23, and so and so. The tool is going to do that for all of the 67 clusters that we have for this example. We have a number of the different household which had been selected within each cluster. You can now save your output of file for both sampled and sub-sampled household. The sample selector app will create an Excel file containing both for sampled and sub-sampled households. Let's save for our example. We are going to call for our case, let's say for instance output 1. The tool sent indication that it already processed and there is a file that was created color output 1. When we open output 1 file, this is when we will be able to get that. It's going to provide the information related to the cluster name, the total number of households within each cluster, the sampling interval, the random number for selecting of the first household, since we are using a systematic random sampling, and also provide the number of the selected 30 households for each cluster. Since we ask the tool to sample 30 households for each of the 67 clusters. Also on the right side, we are able to see the 20 sub-sample household. This sub-sample household have been selected from the 30 sampled household. Now we have the number of each of the selected household, as well as the number of each of the sub-sampled household. Using this information we should now able going back to our sampling frame to know which of the household had been selected. For instance, we have here household. If you pick the Excel file it says that the household number 2, 5, 7, and 10 had been selected. It means that from our sampling frame we are going now to get this household from this number, so household number 2. This one is already selected. Then we have number 2. After number 2, we also have number 5, 7. We have number 2, 5, 7, and 10, which had been selected. We also have all the information related to the household which are the name of the head of a household, the description of the location for the household as well as additional information that could be used to locate the household during data collection. This is how this tool can be used in order to sample households or to sub-sample households. Hoping that this will be helpful and you will be able to carry out your own household sampling. Thank you very much.
