Stratified Random Sampling Summary Notes in Progress

Similar documents
Objectives and Use of Stratification in Sample Design

ALLOCATING SAMPLE TO STRATA PROPORTIONAL TO AGGREGATE MEASURE OF SIZE WITH BOTH UPPER AND LOWER BOUNDS ON THE NUMBER OF UNITS IN EACH STRATUM

Estimating the Population Mean using Stratified Double Ranked Set Sample

Improved Estimation of Rare Sensitive Attribute in a Stratified Sampling Using Poisson Distribution

Chapter 22. Comparing Two Proportions. Copyright 2010 Pearson Education, Inc.

Chapter 22. Comparing Two Proportions. Copyright 2010, 2007, 2004 Pearson Education, Inc.

Statistical Inference (Chapter 10) Statistical inference = learn about a population based on the information provided by a sample.

Expectation and Variance of a random variable

MATH 320: Probability and Statistics 9. Estimation and Testing of Parameters. Readings: Pruim, Chapter 4

MBACATÓLICA. Quantitative Methods. Faculdade de Ciências Económicas e Empresariais UNIVERSIDADE CATÓLICA PORTUGUESA 9. SAMPLING DISTRIBUTIONS

Properties and Hypothesis Testing

BIOS 4110: Introduction to Biostatistics. Breheny. Lab #9

UCLA STAT 13 Introduction to Statistical Methods for the Life and Health Sciences

Element sampling: Part 2

Sampling Error. Chapter 6 Student Lecture Notes 6-1. Business Statistics: A Decision-Making Approach, 6e. Chapter Goals

Sampling, Sampling Distribution and Normality

Simple Random Sampling!

STA Learning Objectives. Population Proportions. Module 10 Comparing Two Proportions. Upon completing this module, you should be able to:

7-1. Chapter 4. Part I. Sampling Distributions and Confidence Intervals

Chapter 8: Estimating with Confidence

Statistics 511 Additional Materials

Topic 9: Sampling Distributions of Estimators

1 Inferential Methods for Correlation and Regression Analysis

Lecture 9: Regression: Regressogram and Kernel Regression

Confidence Intervals รศ.ดร. อน นต ผลเพ ม Assoc.Prof. Anan Phonphoem, Ph.D. Intelligent Wireless Network Group (IWING Lab)

UNIT 8: INTRODUCTION TO INTERVAL ESTIMATION

Random Variables, Sampling and Estimation

Topic 9: Sampling Distributions of Estimators

Topic 9: Sampling Distributions of Estimators

MATH/STAT 352: Lecture 15

Estimation of a population proportion March 23,

Comparing Two Populations. Topic 15 - Two Sample Inference I. Comparing Two Means. Comparing Two Pop Means. Background Reading

ACCESS TO SCIENCE, ENGINEERING AND AGRICULTURE: MATHEMATICS 1 MATH00030 SEMESTER / Statistics

µ and π p i.e. Point Estimation x And, more generally, the population proportion is approximately equal to a sample proportion

An Improved Warner s Randomized Response Model

Chapter 1 (Definitions)

Computing Confidence Intervals for Sample Data

Sampling Distributions, Z-Tests, Power

Chapter 8: STATISTICAL INTERVALS FOR A SINGLE SAMPLE. Part 3: Summary of CI for µ Confidence Interval for a Population Proportion p

6 Sample Size Calculations

If, for instance, we were required to test whether the population mean μ could be equal to a certain value μ

This is an introductory course in Analysis of Variance and Design of Experiments.

On stratified randomized response sampling

Use of Auxiliary Information for Estimating Population Mean in Systematic Sampling under Non- Response

There is no straightforward approach for choosing the warmup period l.

Statistical inference: example 1. Inferential Statistics

FACULTY OF MATHEMATICAL STUDIES MATHEMATICS FOR PART I ENGINEERING. Lectures

4 Conditional Distribution Estimation

Improved exponential estimator for population variance using two auxiliary variables

Interval Estimation (Confidence Interval = C.I.): An interval estimate of some population parameter is an interval of the form (, ),

Chapter 11 Output Analysis for a Single Model. Banks, Carson, Nelson & Nicol Discrete-Event System Simulation

Hypothesis Testing. Evaluation of Performance of Learned h. Issues. Trade-off Between Bias and Variance

Discrete Mathematics for CS Spring 2008 David Wagner Note 22

Big Picture. 5. Data, Estimates, and Models: quantifying the accuracy of estimates.

Homework 5 Solutions

Mathacle. PSet Stats, Concepts In Statistics Level Number Name: Date:

Recursive Computations for Discrete Random Variables

Stat 200 -Testing Summary Page 1

Estimation of Population Mean Using Co-Efficient of Variation and Median of an Auxiliary Variable

Chapters 5 and 13: REGRESSION AND CORRELATION. Univariate data: x, Bivariate data (x,y).

ME 501A Seminar in Engineering Analysis Page 1

Statistics Lecture 27. Final review. Administrative Notes. Outline. Experiments. Sampling and Surveys. Administrative Notes

Section 9.2. Tests About a Population Proportion 12/17/2014. Carrying Out a Significance Test H A N T. Parameters & Hypothesis

Topic 10: Introduction to Estimation

It should be unbiased, or approximately unbiased. Variance of the variance estimator should be small. That is, the variance estimator is stable.

Lecture 3. Properties of Summary Statistics: Sampling Distribution

Lecture 6 Chi Square Distribution (χ 2 ) and Least Squares Fitting

Econ 325/327 Notes on Sample Mean, Sample Proportion, Central Limit Theorem, Chi-square Distribution, Student s t distribution 1.

Overview. p 2. Chapter 9. Pooled Estimate of. q = 1 p. Notation for Two Proportions. Inferences about Two Proportions. Assumptions

STATISTICAL INFERENCE

(all terms are scalars).the minimization is clearer in sum notation:

Read through these prior to coming to the test and follow them when you take your test.

OPTIMIZED DESIGNS OF FRAMEWORKS AND ELEMENTS IN SPATIAL SAMPLING FOR CROP AREA ESTIMATION

Estimation of Population Ratio in Post-Stratified Sampling Using Variable Transformation

Lecture 6 Chi Square Distribution (χ 2 ) and Least Squares Fitting

Class 23. Daniel B. Rowe, Ph.D. Department of Mathematics, Statistics, and Computer Science. Marquette University MATH 1700

TABLES AND FORMULAS FOR MOORE Basic Practice of Statistics

(7 One- and Two-Sample Estimation Problem )

Inferential Statistics. Inference Process. Inferential Statistics and Probability a Holistic Approach. Inference Process.

t distribution [34] : used to test a mean against an hypothesized value (H 0 : µ = µ 0 ) or the difference

A New Mixed Randomized Response Model

KLMED8004 Medical statistics. Part I, autumn Estimation. We have previously learned: Population and sample. New questions

CHAPTER 11 Limits and an Introduction to Calculus

Estimation for Complete Data

1 Models for Matched Pairs

Continuous Data that can take on any real number (time/length) based on sample data. Categorical data can only be named or categorised

STAT 350 Handout 19 Sampling Distribution, Central Limit Theorem (6.6)

Resampling Methods. X (1/2), i.e., Pr (X i m) = 1/2. We order the data: X (1) X (2) X (n). Define the sample median: ( n.

Chapter 7 Student Lecture Notes 7-1

A quick activity - Central Limit Theorem and Proportions. Lecture 21: Testing Proportions. Results from the GSS. Statistics and the General Population

STATISTICAL PROPERTIES OF LEAST SQUARES ESTIMATORS. Comments:

Nonparametric regression: minimax upper and lower bounds

Output Analysis (2, Chapters 10 &11 Law)

Optimal Estimator for a Sample Set with Response Error. Ed Stanek

Response Variable denoted by y it is the variable that is to be predicted measure of the outcome of an experiment also called the dependent variable

Statistics 3. Revision Notes

Data Analysis and Statistical Methods Statistics 651

A General Family of Estimators for Estimating Population Variance Using Known Value of Some Population Parameter(s)

Chapter 6 Sampling Distributions

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 9

Transcription:

Stratified Radom Samplig Summar otes i Progress -3-09 ecture 3- Basic Estimatio Metods witi strata ad overall, Examples, Samplig Allocatio Rules. ecture 4- Samplig Allocatio Rules, Optimal Allocatio Proof Outlie, Example of Calculatios, Post Stratificatio

Probabilit Samplig We kow properties of samples Geeral Teor 6 Hase-Hurwitz wr Horvitz-Tompso a desig Simple Radom Samplig Witout Replacemet -5 simple samplig uits ested Samplig Uits Cluster Samplig Multi-Stage Samplig 3 Improvig Samplig Desig wit Auxiliar Iformatio Regressio Metods Ratio Estimator 7 Regressio Estimator 8 Stratified Radom Samplig Double Samplig 4 Practical Advatages

Improvig Samplig Desigs wit Auxiliar Iformatio Regressio Metods Use relatiosip betwee ad x i liear regressio models. Stratified Radom Samplig Use auxiliar variable s to set up rougl omogeeous groups or strata Double or Two-Pase Samplig Practical Advatages as sometimes ot eoug iformatio o Frame to use eiter directl witout Two-pase samplig

samplig uits i populatio, i te sample, sample witout replacemet. Te ke result is te fiite populatio correctio factor o te variace ad stadard error of te sample mea or proportio due to te samplig beig witout replacemet. Te ke parameters are mea, proportio ad populatio total. Applicatios to surves of uma pops, area surves to estimate aimal or plat populatio parameters Modif to Sstematic Radom Samplig sometimes. Good samplig desig if te populatio is omogeeous. ow we move oto cosiderig ow to sample eterogeeous populatios usig stratified radom samplig to improve te precisio over simple radom samplig.

Stratified Radom Samplig Ver Widel used optio. Useful we te populatio is eterogeeous ad it is possible to establis strata wic are reasoabl omogeeous witi eac oe.

Break up a regio ito 3 omogeeous abitat strata uits uits 3 uits uits uits 3 uits

Te populatio is divided up ito omogeeous strata. Te stratum sizes are, Kow o Frame + + + Witi eac stratum a simple radom sample of size, is take. + It is importat to realize tat te samplig is idepedet i te differet strata. ote- Similar to subpopulatios but ow eac subpopulatio is sampled separatel ote: Aalogous to a radomised complete block desig i experimetal desig

Experimetal desig- Completel radom desig for omogeeous exptl uits Radomised complete block desig for eterogeeous exptl uits Samplig desig- Simple radom samplig for omogeeous samplig uits Stratified radom samplig for eterogeeous samplig uits

Stratified Radom Samplig-W Stratif? Te stratum parameters are of iterest i teir ow rigt Icrease efficiec of estimators of overall populatio parameters b coice of strata tat are omogeeous over te samplig uits witi eac. To make te surve easier to admiister operatioall.

Stratified Radom Samplig-W Stratif? Te stratum parameters are of iterest i teir ow rigt Habitat meas of iterest i a wildlife surve Regioal meas of iterest i a statewide political surve

Stratified Radom Samplig-W Stratif? Icrease efficiec of estimators of overall populatio parameters b coice of strata tat are omogeeous over te samplig uits witi eac. Te overall populatio total, for example, ca be estimated more efficietl.

Stratified Radom Samplig-W Stratif? To make te surve easier to ru admiistrativel. For example, a ealt surve migt ave special subpopulatios strata like studets livig i dorms or prisoers i jails or ospital patiets tat could be sampled differetl to te rest of te public.

Stratified Radom Samplig How to Stratif ad How Ma Strata? Pick Homogeeous Strata to icrease efficiec Usuall Use 5-0 Strata. If too ma te te sample size witi strata is too low

Stratified Radom Samplig Estimatio Metods Idividual stratum meas, totals ad teir stadard errors. Over all populatio meas ad totals ad teir stadard errors.

Stratum Estimatio Metods:Poit Estimates / / Sample Populatio j j j j j j s,..,.,,.., j respose of uit j i stratum µ σ µ τ µ τ µ µ

Stratum Estimatio Metods:Variaces ad SE s variaces te tat SEs are te square roots of Recall j j Were s s Var Var s Var Var µ τ µ

Overall Populatio Estimatio Metods Populatio Mea µ Populatio Total τ

Overall Populatio Mea Estimatio Metods / Sample Estimates / Populatio st st j j W W µ µ µ µ µ µ

Estimatio of Overall Populatio Mea µ. µ st Tis is a weigted average of te idividual stratum meas. Here te weigts are te relative stratum sizes. W st W /,,..,

Estimatio of Overall Populatio Mea µ. st st s Var W Var W E W E Idepede ce Uses Variace. Ubiased. Properties µ µ

Estimatio of Overall Populatio Total τ. variaces of roots square are SEs Recall Recall Variace Ubiased Properties / st st st st st s Var Var Var Var E E µ τ τ τ τ τ τ τ τ µ τ

Estimatio of Overall Populatio Proportio p. Estimate Variace / Estimate Poit st st p p p Var W p Var p W p p

Cofidece Iterval Estimatio Mea ad Total µ st ± z α / SE µ st p st ± z α / SE p st τ st ± z α / SE τ st Tere are better approximatios usig te t dist See P i te Text.Tis is referred to as Sattertwaite s approximatio.

Motivatig Stratificatio: Example. A artificial example cotrasts: Simple radom samplig Stratified radom samplig were a perfect variable as bee foud to stratif o. ote- Te stratified estimate of te mea as 0 variace ad SE. Urealistic but makes te poit of te value of stratificatio!

Stratificatio Examples Text P0-. Ver simple example of calculatios 4 Strata example. Ver simple example of calculatios Mail Surve Example of Estimatio of Proportios Mule Deer Aerial Surve Example of Estimatio of Populatio Totals wit Cofidece Itervals.

Estimatio of Overall Populatio Mea µ: Calculatio Example 4 Strata. W Tis is a weigtedaverageof te stratum meas. Here te weigtsare te relativestratumsizes. W st µ st 90,, 0.0,W 0.50, st / 00,, 3, 0.,W 5.00, W 3,,..,4 400, 3 3 0, 83 0.44,W 30.00, 30, 90 0.35.00 0.00.5 + 0.5 + 0.4430 + 0.35 4.5 4 4 4 4 4 W i

Estimatio of Overall Populatio Mea µ : Example. ou like if 0.Ceck ourself 4.0. 9.0, 9.0, 6.0, 0.35 0.44, 0., 0.0, 83 0,,,, 90 30, 400, 00, 90, 4 3 4 3 4 3 4 3 st st Var s s s s W W W W s Var Var W Var

ecture 4. Brief Review of Stratificatio Estimatio Cocepts usig te Examples Surve Plaig- How to coose te overall sample size ad ow to allocate to te umber of samplig uits i eac idividual stratum.

Recurrig Teme: Improve Precisio of Estimates for Fixed Costs Oe could argue tat moe is cetral te fuctioig of our societ. We use stratified radom samplig because we are trig to improve te precisio of estimates wile keepig costs costat. If we fid a good stratificatio based o auxiliar iformatio we will do muc better ta if we use simple radom samplig. ote tis is te same idea we used we we developed ratio ad regressio estimators. M opiio is tat stratificatio is a bit more practical i ma cases ta usig te regressio estimators but bot approaces are ver useful.

Improvig Samplig Desigs wit Auxiliar Iformatio Regressio Metods Use relatiosip betwee ad x i liear regressio models. Stratified Radom Samplig Use auxiliar variable s to set up rougl omogeeous groups or strata Double or Two-Pase Samplig Practical Advatages as sometimes ot eoug iformatio o Frame to use eiter directl witout Two-pase samplig

Recurrig Teme: Improve Precisio of Estimates for Fixed Costs Bot use of regressio models ad stratificatio are ver useful for gaiig efficiec i estimatio i fiite populatio samplig. M opiio is tat stratificatio is a bit more practical i ma cases ta usig te regressio estimators but Q of te Week? W do ou tik tis migt be true or do ou disagree?

Q of te Week: Stratificatio vs Regressio Models

Estimatio of Overall Populatio Proportio p: Example

Estimatio of Overall Populatio Proportio p: Example Tis example illustrates ma ke poits:. Te stratum estimates are of value i teir ow rigt. Te overall populatio parameter estimate -i tis case te proportio is also importat. 3. Tere are importat practical issues i ruig a surve ad i tis case te orespose could be a issue causig bias.

Stratificatio Example: Helicopter Surve to Cout Mule Deer Kufeld et al 980 Joural of Wildlife Maagemet, 44, 63-639. 8 strata of differet sizes based o differet regios i differet abitats. Samplig Uit is a plot were a complete cout of mule deer is made. Ver good example of use of te stadard samplig metodolog. Idividual stratum estimates ad te overall populatio estimates sum of te stratum estimates. Stud desig is crucial i suc a expesive surve Reasoable precisio of estimates

Stratificatio Example: Helicopter Surve to Cout Mule Deer: Results Table.

Stratificatio Example: Helicopter Surve o Mule Deer: Summar. Tis example illustrates ma ke poits:. Te stratum estimates ma be of value i teir ow rigt. Te overall populatio parameter estimate -i tis case te proportio is also importat. 3. Tere are importat practical issues i ruig suc a complex ad costl surve. 4. Tikig carefull about te desig i advace is a o braier to use a colloquialism. 5. I tis case te odetectio of some aimals is a issue possibl causig bias. Tere are metods to adjust for odetectio wic are covered i m oter class ST 506.

Stratificatio Example: Agler Surve o Small ake: Aget preset all da ad iterviews aglers as te leave. Use two strata WD, WE. M T W T F S S 3 4 5 6 7 8 9 0 3 4 5 6 7 8 9 0 3 4 5 6 7 8 9 If Exted to mo. surve te 4 strata x WD-Weekdas -sample 8 WE-Weekeds 8 -sample 4 Higer Rate

Stratified Radom Samplig:Allocatio Surve Plaig- I simple radom samplig we just ad to figure out ow large te sample eeded to be to a acieve a precisio goal. ow i stratified radom samplig it is a little more complex: - How to coose te overall sample size. - How to allocate to te umber of samplig uits i eac idividual stratum.

Stratified Radom Samplig: Sample Allocatio Rules to Obtai Better Precisio -Equal Allocatio allocate te sample size equall i all te strata. -Proportioal Allocatio allocate proportioal to te size of te strata-ver widel used -Optimal allocate proportioal to size ad stratum variaces ad iversel proportioal to costs i te differet strata.

Stratified Radom Samplig: Sample Allocatio Rules to Obtai Better Precisio Equal Allocatio allocate te sample size equall i all te strata. ot usuall sesible uless all strata are equal size i terms of overall estimates precisio. However, mabe good if ou wat to compare stratum meas as our primar focus

Stratified Radom Samplig: Sample Allocatio Rules: Equal Allocatio /

Stratified Radom Samplig: Sample Allocatio Rules: Equal Allocatio / Earlier 4 Stratum Example 4 / 0.5 Use 83x0.5, 0.75,,, 83x0.5, 0.75, 3, 3 4 3 83x0.5, 0.75, 0 4 4 0.75 83x0.5

Stratified Radom Samplig: Sample Allocatio Rules: Proportioal Allocatio Allocate proportioal to te size of te strata-ver widel used / W

Stratified Radom Samplig: Sample Allocatio Rules: Proportioal Allocatio / W 83x0.0, 83x0., 36.5, ote weigts give earlier. Use rouded values 8.3, 8, 3 9., 9, 3 37, 4 4 3 83x0.44, 9.05 9 for total 4 83 83x0.35

Stratified Radom Samplig: Sample Allocatio Rules to Obtai Better Precisio Optimal Allocatio Takes accout of stratum sizes, differet variaces, ad differet costs of samplig i differet strata Optimal Allocatio is ot used as muc as proportioal allocatio but it ca result i a gai i precisio if te costs ad variaces are kow or well estimated from a prior stud or a pilot surve.

Geeral Optimal Allocatio Metodolog σ Var st We wat to miimise te variace for fixed Total Cost were te Cost Fuctio is : C c c 0 is is c 0 + te overead cost te per samplig uit cost. We cosider te costraied fuctio i terms of ad use te agrage multiplier approac. Tis will be outlied o te witeboard ad is discussed i simpler form i text o P6. c te

Geeral Optimal Allocatio Results l c c c C c c 0 / OverallSample Size / / Relative Sample Sizes σ σ σ σ

Fiseries Agler Surve Book Example of te Allocatio Calculatios 000, Total Cost C $00 000, 3 500, 3500. W 0.57, W 0.86, W3 0.43. Prior St Deviatio Values σ 0, σ 0, σ 3 0, Cost Values Overead Cost co $00 Per Uit Cost c $.00, c $.00, c 3 $0.64,

Self Weigted Stratified Samples allocatio i.e We are usig proportioal / / ad ol if if / / W st j j st

Post Stratificatio.6 P4 Used sometimes o a simple radom sample we te stratum sizes are ot kow. Post Stratified Variace Sowig Extra Pealt Term Var st Var prop st + [ / ] σ

Stratificatio Brief Summar Slides

Stratified Radom Samplig-W Stratif? Te stratum parameters are of iterest i teir ow rigt Icrease efficiec of estimators of overall populatio parameters b coice of strata tat are omogeeous over te samplig uits witi eac. To make te surve easier to admiister operatioall.

Stratified Radom Samplig-Advatages Advatages ad Disadvatages -Icreased efficiec -Coveiece -Focus o Subpopulatios of special iterest Over sample tem Disadvatages - Ma ot alwas ave te iformatio ou eed o our frame so ou kow stratum sizes. - Ma ot kow stratum st deviatios or costs well to use optimal allocatio

Stratified Radom Samplig- We is it Better? Te Stratum meas differ widel from eac oter. Te stratum stadard deviatios are small. Tat is te variatio witi eac of te stratum are small. Recall motivatig example

Compariso of Sample Allocatio Rules c c / / eed Pilot Ifo Costs, Allocatio Uequal Optimal Geeral eed Pilot Ifo Costs, ema Allocatio Equal AllocatioMost Commo Proportioal AllocatioRarel Used Equal σ σ σ σ