2Session Objectives To introduce Multistage sampling To help participants appreciate when to apply multi-stage sampling
3Multi-Stage Sampling Design Most commonly used sampling design in practiceInvolves more than one stage of sampling and/or a combination of two or more sampling designs
4Why apply multi-stage sampling Structure of the populationSize of population (large)Fund available for the survey
5Examples of multi-stage sampling A Systematic sample of households within Enumeration Areas (EAs) within DistrictsDistricts – Strata – 1st StageEAs- Clusters- 2nd StageHouseholds -3rd StageWithin each District take a sample of EAsWithin each EAs, take a sample of h’holdsWithin each household, sample 2 individuals
6Cont’dDetermine from students at a university their opinion of the university.Stratify students according to their Faculty (Science, etc.) (strata 1st )Take simple random sample of courses in each of these Faculties (clusters 2nd )From each course take a random sample of students.
7Sampling and Analysis issues using the Paddy Game Module 3Session 9(b)
8Session ObjectivesTo introduce basic sampling concepts using the Paddy gameTo demonstrate how to select a sample from a populationTo demonstrate how population values can be estimated from samples
9What is Paddy?Paddy is a game that has been developed to demonstrate how to design and analyse survey dataThe game uses an example of a survey of rice productionOpen the paddy game on your computer by clicking on the PADDY icon.
10Background Information In a small district there are:10 villagesWith a total of 160 farmersEach with one field in which to grow riceA census has been undertaken and the acreage cultivated by each farmer is known.
12Aim of the SurveyThe main aim of the survey is to estimate the mean yield of rice per hectare and hence the total production of rice in the districtIn order to do this you will carry out a crop cutting survey. This is a survey where you visit fields, harvest the crop from one or more plots and measure the yieldA further objective is to investigate the use of fertiliser and varieties by the farmers
13ResourcesThe resources available allow you to visit and collect information from 30 plots.A plot covers an area of 1/80th of an acre.Yield from each plot harvested is then measured in bushels per acre.
14ResourcesTransport costs allow your survey team to visit a maximum of 5 villages, although you might decide to visit less than 5.These costs dictate that the survey should use a multistage sampling design
15Resources The game gives you the possibility of three levels: Villages Farmers within VillagePlots within farmers.
16Pilot RunWe begin with a pilot run to give you familiarity with the Paddy gameThe Paddy game has been installed on your computersThe sampling scheme suggested is the following:Select 2 villages using simple random samplingFrom each village select 3 fields, again with simple random samplingFrom each field, select 1 plot at random for harvesting
21Probability Proportional to Size (PPS) Sampling using the Paddy Game Module 3Session 9(c)
22Session ObjectivesTo demonstrate how to select a sample from a population using PPSTo demonstrate how population values can be estimated from samples
23Pilot Run Suppose the sampling scheme suggested is the following: Select 2 villages using PPS sampling where the sample depends on the number of farmers in the villageFrom each village select 3 fields, with simple random samplingFrom each field, select 1 plot at random for harvesting