Sample size calculation should be dependant on the sampling methodology

[Edouard-Legoupil]

Sampling size calculation is managed with https://github.com/unhcr-americas/rmsSampling/blob/master/R/mod_estimate.R

Currently it accounts for

• "ci" "Confidence Level",
• "moe" "Tolerable Margin of Error in % ",
• "p" "Anticipated response distribution",
• "power" "Desired Sample Power", (added in the interface but not yet in the server calculation...)
• "over" "Desired oversampling proportion",

This module should take the sampling method as a parameter and adjust the parameters to define the sampling size based on parameters specific to the method...

What additional parameters should be defined for each method that are selected by the method screener?

• nonprob 'Non-probabilistic methods';
• srs 'Simple Random Sampling \n(SRS) without stratification';
• ppsframe 'Multiple-Stage \nCluster Sampling\n with Frame';
• srsstrata 'Simple Random Sampling\n (SRS) within Strata';
• acs 'Adaptive Cluster Sampling';
• lts 'Time Location Sampling';
• rds 'Respondent Driven Sampling\n (RDS)';

[Edouard-Legoupil]

Some ideas similar to this one are available within REACH tool - https://github.com/oliviercecchi/Probability-sampling-tool