The Sampling.Kit is a web-tool designed for survey designers and samplers to perform several functions. The Sampling.Kit assumes a stratified multi-stage sampling design similar to the standard sampling designs in DHS surveys, where sample clusters are selected in the first stage, and households are selected in the second stage. In the selected households, all individuals of target population (such as women age 15-49 years) are eligible for the survey. More details about sampling design of the DHS surveys can be found here .


The current version of the Sampling.Kit has four tools:


1. Sample size calculates sample size to estimate a proportion with a pre-determined precision level. The calculations account for factors like response rates and number of target population individuals per household.


2. Sample allocation allocates the sample clusters or units using different allocation methods, such as proportional, equal and power allocation.


3. Expected precision calculates the expected precision of survey estimates given the sample size and allocation.


4. Significance testing performs significance testing for one statistic or two statistics from two independent samples.


Author and Maintainer

Mahmoud Elkasabi , (Email:Mahmoud.Elkasabi@icf.com)

Sample size of target population
$$n = Deft^2 \frac{1-p}{p.{\alpha}^2}$$
Expected proportion (%)
Relative Standard Error (%)
Design effect
Number of households to be selected
$$n_{HH} = \frac{n}{RR_{HH}.I_{HH}.RR_{indv}}$$
Household response rate (%)
Individuals per household
Individual response rates (%)

Households to be selected: 


              

                Households expected to respond:
              

              

              

                Eligible individuals expected to be found:
              

              

              

                Individuals expected to respond:
              

              

              

                This sample size is expected to measure:
              

              

              

            

          

        

      

      

        

          

          

            

              
1. Add total sample size to be allocated

              
2. Add total number of sample strara

              
3. Add the measure of size N_h in the table

              

                to add N_h values, double-click on each cell
              

              

                N_h can be frequency or percentage distribution of households or population
              

            

          

          

            

              
Equal allocation

              \(n_h = \dfrac{n}{H}\) 
              
              
     

              
     

            

          

          

            

              
Proportional allocation

              \(n_h = n. \dfrac{N_{h}}{\sum^H_{h = 1} N_{h}}\) 
              
            

          

          

            

              
Power allocation

              \(n_h = n. \dfrac{N_{h}^{\alpha}}{\sum^H_{h = 1} N_{h}^{\alpha}}\) 
              
            

          

        

        

          

          

            

              Total sample size
              

                
                
              

              Number of Strata
              

                
                
              

              Power value
              for power allocation
              

                
                
              

            

          

          

            

              

            

          

          

            

              

            

          

        

      

      

        

        

          

            Number of Domains
            

              
              
            

            Expected proportion (%)
            

              
              
            

            Design effect
            

              
              
            

            Household response rate (%)
            

              
              
            

            Individuals per Household
            

              
              
            

            Individual response rate (%)
            

              
              
            

          

        

        

          

            

              

                
1. Add total number of doamins D

                
2. Change p, Deft and response rates

                
3. Add domain-level p and Deft if available

                
4. add number of households selected in each domain

              

            

            

              

                
SE: expected standard error

                
RSE: relative standard error

                
LCI and UCI: lower and upper bounds of a 95% confidence interval

              

            

          

          

            

              

                

              

            

            

              

                

              

            

          

        

      

      

        

        

          

            

            

              
              

                
                
              

            

            

            

              Statistic
              

                
                
              

              Standard error
              

                
                
              

              

              Null hypothesis
              

                
                
              

            

            

              Statistic 1
              

                
                
              

              Standard error 1
              

                
                
              

              

              Statistic 2
              

                
                
              

              Standard error 2