DOCSLIB.ORG

Sampling Theory TWO STAGE SAMPLING TWO STAGE SAMPLING

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Sampling Theory TWO STAGE SAMPLING TWO STAGE SAMPLING Sampling Theory MODULE X LECTURE - 33 TWO STAGE SAMPLING (SUB SAMPLING) DR. SHALABH DEPARTMENT OF MATHEMATICS AND STATISTICS INDIAN INSTITUTE OF TECHNOLOGY KANPUR 1 In cluster sampling, all the elements in the selected clusters are surveyed. Moreover, the efficiency in cluster sampling depends on size of the cluster. As the size increases, the efficiency decreases. It suggests that hig her preciiision can beatta ine d by dis tr ibu tingagiven number of eltlementsover a large number of cltlusters and then by taking a small number of clusters and enumerating all elements within them. This is achieved in subsampling. IbIn subsamp ling Divide the population into clusters. Select a sample of clusters [first stage] From each o f the se lec te d c lus ter, se lec t a samp le o f spec ifie d num ber o f e lemen ts [secondtd stage] The clusters which form the units of sampling at the first stage are called the first stage units and the units or group of units within clusters which form the unit of clusters are called the second stage units or subunits. The procedidure is genera lidtthlized to three or more s tages an dithtd is then terme d as multist age sampli ng. For example, in a crop survey villages are the first stage units, fields within the villages are the second stage units and plots within the fields are the third stage units. In another example, to obtain a sample of fishes from a commercial fishery, first take a sample of boats and then take a sample of fishes from each selected boat. 2 Two stage sampling with equal first stage units AthtAssume that population consists of NM elements. NM elements are grouped into N first stage units of M second stage units each, (i.e., N clusters, each cluster is of size M). Sample of n first stage units is selected (i.e., choose n clusters) Sample of m second stage units is selected from each selected first stage unit (i.e., choose m units from each cluster). Units at each stage are selected with SRSWOR. Cluster sampling is a special case of two stage sampling in the sense that from a population of N clusters of equal size m = M, a sample of n clusters chosen. If furth er M = m = 1, we get SRSWOR. If n = N, we have the case of stratified sampling. 3 th th yij : Value of the characteristic under study for the j second stage unit of the i first stage unit; iNjm==1,2,..., ; 1,2,.., . m 1 nd th st Yyiij= ∑ : mean per 2 stage unit of i 1 stage units in the population. M j=1 11NM N YyyY===∑∑ij ∑ i MN : mean per second stage unit in the popultilation MNij==11 N i = 1 1 m mean per second stage unit in the ith first stage unit in the sample. yyiij= ∑ : n j=1 11nm n mean per second stage in the sample. yyyy===∑∑ij ∑ i mn : mnij==11 n i = 1 Advantages The principle advantage of two stage sampling is that it is more flexible than the one stage sampling. It reduces to one stage sampling when m=M but unless this is the best choice of m, we have the opportunity of taking some smaller value that appears more efficient. As usual, this choice reduces to a balance between statistical precision and cost. When units of the first stage agree very closely, then consideration of precision suggests a small value of m. On the other hand, it is sometimes as cheap to measure the whole of a unit as to a sample. For example, when the unit is a household and a single respondent can give as accurate data as all the members of the household. 4 A pictorial scheme of two stage sampling scheme is as follows: Population (MN units) Cluster Cluster Cluster Population … … … N clusters M units M units M units N clusters… Cluster Cluster Cluster First stage sample M units M units M units n clusters (small) n clusters (small) Second stage sample Cluster Cluster Cluster m units m m … … … m units n clusters (small) units units mn units 5 Note The expectations under two stage sampling scheme depend on the stages. For example, the expectation at second stage unit will be dependent on first stage unit in the sense that second stage unit will be in the sample provided it was selected in the first stage. To calculate the average First average the estimator over all the second stage selections that can be drawn from a fixed set of n units that the plan selects. Then average over all the possible selections of n units by the plan. In case o f t wo stage sam plin g, ˆˆ EEE()θθ= 12 [ ()] ↓↓ 2 average average average over all over over all possible 2nd stage all 1st stage selections from a samples samples fixed set of units In case of three staggpge sampling, EEEE()θθˆˆ= ⎡⎤ () . 123⎣⎦{ } To calculate the variance, we proceed as follows: In case of two stage sampling, Var()θθθˆˆ=− E ( )2 ˆ 2 =−EE12().θ θ 6 Consider ˆˆˆ22 2 EEE222()θθ−= ()2() θ − θ θ+ θ 2 =+−+⎡⎤EV()θˆˆ ()θθθθ 2 E () ˆ2 . ⎣⎦⎢⎥{}22 2 Now average over first stage selection as 2 EE()θˆˆˆˆ−=θθ2 E⎡⎤⎡⎤ E () + E V ()2()() θθθθ − EE + E 2 12 1⎣⎦⎣⎦ 2 1 2 12 1 2 =−+EEE⎡⎤()θθˆˆ2 EV⎡⎤ () θ 112⎣⎦⎢⎥{} 12⎣⎦ Var(θθˆˆ )=+ V⎡⎤⎡⎤ E ( ) E V ( θ ˆ ) . 12⎣⎦⎣⎦ 12 In case of three stage sampling, ˆˆ⎡⎤⎡ ˆ⎤⎡ ˆ ⎤ Var()θθ=++ V13 E E () E 123123 V E () θ E E V () θ . ⎣⎦2 { } ⎣ { }⎦⎣{ } ⎦ 7 Estimation of population mean Consider yy= mn as an estimator of the population mean Y . Bias Consider Ey()= E12[] E ( ymn ) nd st = EEy12[](|)(im i as21 stage is dependent on stage ) = EEy12[](|)(im i asyYii is unbiased for due to SRSWOR) ⎡⎤1 n = EY1 ⎢⎥∑ i ⎣⎦n i=1 1 N = ∑Yi N i=1 =Y . Thus ymn is an unbiased estimator of the population mean. 8 Variance Vary()=+ EV12[ ( yi |)] V 1[ E 2 ( yi |)] ⎡⎤⎡⎤⎧⎫11nn ⎧⎫ =+EV12⎢⎥⎢⎥⎨⎬∑∑ yiii|| VE 1 2 ⎨⎬ yi ⎣⎦⎣⎦⎩⎭nnii==11 ⎩⎭ ⎡⎤⎡⎤11nn =+EV(|)yyyiV E (|)y i 112⎢⎥⎢⎥2 ∑∑ii ⎣⎦⎣⎦nnii==11 ⎡⎤111nn⎛⎞ ⎡⎤ 1 =−+ESVY2 11⎢⎥2 ∑∑⎜⎟ii⎢⎥ ⎣⎦nmMii==11⎝⎠ ⎣⎦ n 111n ⎛⎞ E()|SiVy2 () =−2 ∑⎜⎟ic+ 1 nmMi=1 ⎝⎠ (whereyc is based on cluster means as in cluster sampling ) 111⎛⎞22Nn− =−+2 nSS⎜⎟wb nm⎝⎠MNn 11⎛⎞⎛⎞ 122 1 1 =−⎜⎟⎜⎟SSwb +− nm⎝⎠⎝⎠ M n N N NM 2 2211 where SSwi==∑∑∑( YY iji −) NNMiij===111(1)− N 221 SYYbi=−∑(). N −1 i=1 9 Estimate of variance 22 An unbiased estimator of variance of y can be obtained byyp replacing SS bwand by their unbiased estimators in the expression of variance of y Consider an estimator of N 221 SSwi= ∑ N i=1 where M 2 2 1 SyYiiji= ∑( − ) M −1 j=1 and n 221 sswi= ∑ n i=1 m 221 syyiiji=−∑(). m −1 j=1 10 22 So Es()ww= EE12() s | i n ⎡⎤1 2 = EE12⎢⎥∑ si | i ⎣⎦n i=1 1 n ⎡⎤2 = EEsi12∑ ⎣⎦(|)i n i=1 n 1 2 = ES1 ∑ i ()as SRSWOR is used n i=1 n 1 2 = ∑ ES1()i n i=1 NN 11⎡⎤2 = ∑∑⎢⎥Si NNii==11⎣⎦ N 1 2 = ∑ Si N i=1 2 = Sw 2 2 so s w is an unbiased estimator of S w . Consider n 221 sbi=−∑ ()yy n −1 i=1 as an estimator of N 221 SYYbi=−∑(). N −1 i=1 11 So n 221 ⎡⎤ Es()bi=− E⎢⎥∑ (yy ) n −1 ⎣⎦i=1 n 222⎡⎤ (1)()nEsE−=bi⎢⎥∑ yny − ⎣⎦i=1 n ⎡⎤22 =−EynEy⎢⎥∑ i () ⎣⎦i=1 n ⎡⎤⎛⎞ 2 =−+EE y2 nVary⎡⎤() Ey () 12⎢⎥⎜⎟∑ i ⎣⎦{} ⎣⎦⎝⎠i=1 n ⎡⎤2222⎡⎤⎛⎞⎛⎞11 1 11 =−−+−+EEyin12⎢⎥∑ ()|)ibw⎢⎥⎜⎟⎜⎟ S SY ⎣⎦i=1 ⎣⎦⎝⎠⎝⎠nN mMn n ⎡⎤2 ⎡⎤⎛⎞⎛⎞11 1 11 EVar(yEyn)() S222 SY = 1 ⎢⎥∑{ ii+−−+−+()} ⎢⎥⎜⎟⎜⎟ b w ⎣⎦i=1 ⎣⎦⎝⎠⎝⎠nN mMn n ⎡⎤⎧⎫⎡⎛⎞1122 ⎛⎞⎛⎞ 11 2 111 22 ⎤ =−+−−+−+ESYnSSY1 ⎢⎥∑ ⎨⎬⎜⎟ii⎢⎥ ⎜⎟⎜⎟ b w ⎣⎦i=1 ⎩⎭⎣⎝⎠mM ⎝⎠⎝⎠ nN mMn ⎦ n ⎡⎤111⎧⎫⎛⎞22⎡⎤ ⎛⎞ 11 2⎛⎞11122 =−+−−nE1 ⎢⎥⎨⎬∑⎜⎟ Sii Y n⎢ ⎜⎟ S b+ ⎜⎟−+SYw ⎥ ⎣⎦nmM⎩⎭i=1 ⎝⎠⎣ ⎝⎠ nN ⎝⎠mMn ⎦ NN ⎡⎤⎡⎤⎛⎞11122 1 ⎛⎞⎛⎞ 11 2 111 22 =−nSYnSSY⎢⎥⎢⎥⎜⎟∑∑ii + −−+−+ ⎜⎟⎜⎟ b w ⎣⎦⎣⎦⎝⎠mMNii==11 N ⎝⎠⎝⎠ nN mMn N ⎡⎤⎡⎤⎛⎞1122 1 ⎛⎞⎛⎞ 11 2 111 22 =−nSYnS⎢⎥⎢⎥⎜⎟wi +∑ −−+−+ ⎜⎟⎜⎟ b SY w ⎣⎦⎣⎦⎝⎠mM Ni=1 ⎝⎠⎝⎠ nN mMn 12 N ⎛⎞11222n ⎛⎞ 11 2 =−(1)n⎜⎟ − Swi +∑ YnYn − − ⎜⎟ − S b ⎝⎠mM Ni=1 ⎝⎠ nN N ⎛⎞112222n ⎡⎤ ⎛⎞ 11 =−(1)nSYNYnS⎜⎟ −wi +⎢⎥∑ − − ⎜⎟ − b ⎝⎠mM N⎣⎦i=1 ⎝⎠ nN ⎛⎞1122n ⎛⎞ 11 2 =−(1)nSNSnS⎜⎟ −wb + ( − 1) − ⎜⎟ − b ⎝⎠mM N ⎝⎠ nN ⎛⎞11 22 =−(1)nSnS⎜⎟ −wb +− (1). ⎝⎠mM 222⎛⎞11 ⇒⇒+Es()bbb =−⎜⎟ Sw + Sb ⎝⎠mM or ⎡⎤222⎛⎞11 Es⎢⎥bwb−−⎜⎟ s = S. ⎣⎦⎝⎠mM Thus m 11⎛⎞⎛⎞ 1ˆ 22 1 1 ˆ Var() y=−⎜⎟⎜⎟ Sω +− Sb nm⎝⎠⎝⎠ M n N 11⎛⎞⎛⎞⎛⎞ 1222 1 1⎡⎤ 1 1 =−⎜⎟⎜⎟⎜⎟swbw +−−−⎢ss⎥ nm⎝⎠⎝⎠⎝⎠ M n N⎣ m M ⎦ 11⎛⎞⎛⎞ 122 1 1 =−⎜⎟⎜⎟sswb +−. Nm⎝⎠⎝⎠ M n N 13.
Load more

Recommended publications
  • Survey Sampling
    Bayesian SAE using Complex Survey Data Lecture 5A: Survey Sampling Jon Wakefield Departments of Statistics and Biostatistics University of Washington 1 / 119 Outline Overview Design-Based Inference Simple Random Sampling Stratified Simple Random Sampling Cluster Sampling Multistage Sampling Discussion Technical Appendix: Simple Random Sampling Technical Appendix: Stratified SRS Technical Appendix: Cluster Sampling Technical Appendix: Lonely PSUs 2 / 119 Overview 3 / 119 Outline Many national surveys employ stratified cluster sampling, also known as multistage sampling, so that’s where we’d like to get to. In this lecture we will discuss: I Simple Random Sampling (SRS). I Stratified SRS. I Cluster sampling. I Multistage sampling. 4 / 119 Main texts I Lohr, S.L. (2010). Sampling Design and Analysis, Second Edition. Brooks/Cole Cengage Learning. Very well-written and clear mix of theory and practice. I Lumley, T. (2010). Complex Surveys: A Guide to Analysis Using R, Wiley. Written around the R survey package. Great if you already know a lot about survey sampling. I Korn, E.L. and Graubard, B.I. (1999). Analysis of Health Surveys. Wiley. Well written but not as comprehensive as Lohr. I Sarndal,¨ Swensson and Wretman (1992). Model Assisted Survey Sampling. Springer. Excellent on the theory though steep learning curve and hard to dip into if not familiar with the notation. Also, anti- model-based approaches. 5 / 119 The problem We have a question concerning variables in a well-defined finite population (e.g., 18+ population in Washington State). What is required of a sample plan? We want: I Accurate answer to the question (estimate). I Good estimate of the uncertainty of the estimate (e.g., variance).
    [Show full text]
  • IBM SPSS Complex Samples Business Analytics
    IBM Software IBM SPSS Complex Samples Business Analytics IBM SPSS Complex Samples Correctly compute complex samples statistics When you conduct sample surveys, use a statistics package dedicated to Highlights producing correct estimates for complex sample data. IBM® SPSS® Complex Samples provides specialized statistics that enable you to • Increase the precision of your sample or correctly and easily compute statistics and their standard errors from ensure a representative sample with stratified sampling. complex sample designs. You can apply it to: • Select groups of sampling units with • Survey research – Obtain descriptive and inferential statistics for clustered sampling. survey data. • Select an initial sample, then create • Market research – Analyze customer satisfaction data. a second-stage sample with multistage • Health research – Analyze large public-use datasets on public health sampling. topics such as health and nutrition or alcohol use and traffic fatalities. • Social science – Conduct secondary research on public survey datasets. • Public opinion research – Characterize attitudes on policy issues. SPSS Complex Samples provides you with everything you need for working with complex samples. It includes: • An intuitive Sampling Wizard that guides you step by step through the process of designing a scheme and drawing a sample. • An easy-to-use Analysis Preparation Wizard to help prepare public-use datasets that have been sampled, such as the National Health Inventory Survey data from the Centers for Disease Control and Prevention
    [Show full text]
  • Sampling and Evaluation
    Sampling and Evaluation A Guide to Sampling for Program Impact Evaluation Peter M. Lance Aiko Hattori Suggested citation: Lance, P. and A. Hattori. (2016). Sampling and evaluation: A guide to sampling for program impact evaluation. Chapel Hill, North Carolina: MEASURE Evaluation, University of North Carolina. Sampling and Evaluation A Guide to Sampling for Program Impact Evaluation Peter M. Lance, PhD, MEASURE Evaluation Aiko Hattori, PhD, MEASURE Evaluation ISBN: 978-1-943364-94-7 MEASURE Evaluation This publication was produced with the support of the United States University of North Carolina at Chapel Agency for International Development (USAID) under the terms of Hill MEASURE Evaluation cooperative agreement AID-OAA-L-14-00004. 400 Meadowmont Village Circle, 3rd MEASURE Evaluation is implemented by the Carolina Population Center, University of North Carolina at Chapel Hill in partnership with Floor ICF International; John Snow, Inc.; Management Sciences for Health; Chapel Hill, NC 27517 USA Palladium; and Tulane University. Views expressed are not necessarily Phone: +1 919-445-9350 those of USAID or the United States government. MS-16-112 [email protected] www.measureevaluation.org Dedicated to Anthony G. Turner iii Contents Acknowledgments v 1 Introduction 1 2 Basics of Sample Selection 3 2.1 Basic Selection and Sampling Weights . 5 2.2 Common Sample Selection Extensions and Complications . 58 2.2.1 Multistage Selection . 58 2.2.2 Stratification . 62 2.2.3 The Design Effect, Re-visited . 64 2.2.4 Hard to Find Subpopulations . 64 2.2.5 Large Clusters and Size Sampling . 67 2.3 Complications to Weights . 69 2.3.1 Non-Response Adjustment .
    [Show full text]
  • Target Population” – Do Not Use “Sample Population.”
    NCES Glossary Analysis of Variance: Analysis of variance (ANOVA) is a collection of statistical models and their associated estimation procedures (such as the "variation" among and between groups) used to test or analyze the differences among group means in a sample. This technique is operated by modeling the value of the dependent variable Y as a function of an overall mean “µ”, related independent variables X1…Xn, and a residual “e” that is an independent normally distributed random variable. Bias (due to nonresponse): The difference that occurs when respondents differ as a group from non-respondents on a characteristic being studied. Bias (of an estimate): The difference between the expected value of a sample estimate and the corresponding true value for the population. Bootstrap: See Replication techniques. CAPI: Computer Assisted Personal Interviewing enables data collection staff to use portable microcomputers to administer a data collection form while viewing the form on the computer screen. As responses are entered directly into the computer, they are used to guide the interview and are automatically checked for specified range, format, and consistency edits. CATI: Computer Assisted Telephone Interviewing uses a computer system that allows a telephone interviewer to administer a data collection form over the phone while viewing the form on a computer screen. As the interviewer enters responses directly into the computer, the responses are used to guide the interview and are automatically checked for specified range, format, and consistency edits. Cluster: a group of similar people, things or objects positioned or occurring closely together. Cluster Design: Cluster Design refers to a type of sampling method in which the researcher divides the population into separate groups, of similarities - called clusters.
    [Show full text]
  • Sampling Techniques Third Edition
    Sampling Techniques third edition WILLIAM G. COCHRAN Professor of Statistics, Emeritus Harvard University JOHN WILEY & SONS New York• Chichester• Brisbane• Toronto• Singapore Copyright © 1977, by John Wiley & Sons, Inc. All nghts reserved. Published simultaneously in Canada. Reproducuon or tran~lation of any pan of this work beyond that permined by Sections 107 or 108 of the 1976 United Slates Copy­ right Act wuhout the perm1!>sion of the copyright owner 1s unlaw­ ful. Requests for permission or funher mformat1on should be addressed to the Pcrm1ss1ons Depanmcnt, John Wiley & Sons. Inc Library of Congress Cataloging In Puhllcatlon Data: Cochran, William Gemmell, 1909- Sampling techniques. (Wiley series in probability and mathematical statistic!>) Includes bibliographical references and index. 1. Sampling (Statistics) I. Title. QA276.6.C6 1977 001.4'222 77-728 ISBN 0-471-16240-X Printed in the United States of America 40 39 38 37 36 to Betty Preface As did the previous editions, this textbook presents a comprehensive account of sampling theory as it has been developed for use in sample surveys. It cantains illustrations to show how the theory is applied in practice, and exercises to be worked by the student. The book will be useful both as a text for a course on sample surveys in which the major emphasis is on theory and for individual reading by the student. The minimum mathematical equipment necessary to follow the great bulk of the material is a familiarity with algebra, especially relatively complicated algeb­ raic expressions, plus a·knowledge of probability for finite sample spaces, includ­ ing combinatorial probabilities.
    [Show full text]
  • Multilevel Modeling of Complex Survey Data
    Multilevel Modeling of Complex Survey Data Sophia Rabe-Hesketh, University of California, Berkeley and Institute of Education, University of London Joint work with Anders Skrondal, London School of Economics 2007 West Coast Stata Users Group Meeting Marina del Rey, October 2007 GLLAMM – p.1 Outline Model-based and design based inference Multilevel models and pseudolikelihood Pseudo maximum likelihood estimation for U.S. PISA 2000 data Scaling of level-1 weights Simulation study Conclusions GLLAMM – p.2 Multistage sampling: U.S. PISA 2000 data Program for International Student Assessment (PISA): Assess and compare 15 year old students’ reading, math, etc. Three-stage survey with different probabilities of selection Stage 1: Geographic areas k sampled Stage 2: Schools j =1, . , n(2) sampled with different probabilities πj (taking into account school non-response) (1) Stage 3: Students i=1, . , nj sampled from school j, with conditional probabilities πi|j Probability that student i from school j is sampled: πij = πi|jπj GLLAMM – p.3 Model-based and design-based inference Model-based inference: Target of inference is parameter β in infinite population (parameter of data generating mechanism or statistical model) called superpopulation parameter Consistent estimator (assuming simple random sampling) such as maximum likelihood estimator (MLE) yields estimate β Design-based inference: Target of inference is statistic in finiteb population (FP), e.g., mean score yFP of all 15-year olds in LA Student who had a πij =1/5 chance of being sampled represents wij =1/πij =5 similar students in finite population Estimate of finite population mean (Horvitz-Thompson): FP 1 y = w y w ij ij ij ij Xij b P Similar for proportions, totals, etc.
    [Show full text]
  • CHAPTER 5 Choosing the Type of Probability Sampling
    CHAPTER 5 CHOOSING THE TYPE OF PROBABILITY SAMPLING What you will learn in this chapter: •• The types of probability sampling and how they differ from each other •• Steps in carrying out the major probability sample designs •• The strengths and weaknesses of the various types of probability sampling •• Differences between stratified sampling and quota sampling •• Differences between stratified sampling and cluster sampling •• Differences between multistage cluster sampling and multiphase sampling INTRODUCTION Once a choice is made to use a probability sample design, one must choose the type of probability sampling to use. This chapter includes descriptions of the major types of probability sampling. It covers steps involved in their adminis- tration, their subtypes, their weaknesses and strengths, and guidelines for choosing among them. There are four major types of probability sample designs: simple random sampling, stratified sampling, systematic sampling, and cluster sampling (see Figure 5.1). Simple random sampling is the most recognized probability sam- pling procedure. Stratified sampling offers significant improvement to simple random sampling. Systematic sampling is probably the easiest one to use, and cluster sampling is most practical for large national surveys. These sampling procedures are described below. 125 126 Sampling Essentials Figure 5.1 Major Types of Probability Sampling Probability Sample Designs Simple Stratified Systematic Cluster Random Sampling Sampling Sampling Sampling SIMPLE RANDOM SAMPLING What Is Simple Random Sampling? Simple random sampling is a probability sampling procedure that gives every element in the target population, and each possible sample of a given size, an equal chance of being selected. As such, it is an equal probability selection method (EPSEM).
    [Show full text]
  • An Overview of Primary Sampling Units (Psus) in Multi-Stage Samples for Demographic Surveys1
    Section on Government Statistics – JSM 2008 An Overview of Primary Sampling Units (PSUs) in Multi-Stage Samples 1 for Demographic Surveys Padraic Murphy U.S. Census Bureau, Washington, DC 20233-9200 Abstract This paper will present an overview of historical and current issues in the definition, stratification, and selection of Primary Sampling Units (PSUs) in large demographic surveys. By PSUs, we mean the large clusters (usually geographic areas such as counties) that are the sampling frame in the first stage of a multi-stage sample design. While simple in concept, the details of defining, stratifying, and selecting PSUs can prove to be surprisingly complex. We look at developments pertaining to PSUs (as used by the U.S. Census Bureau) over the past half-century, and describe some current problems. The issues discussed include (1) constraints on PSU size and boundaries, (2) choosing "building blocks" for PSUs, (3) methodology for and objectives of PSU stratification, (4) coordination among multiple surveys, and (5) coordination with preceding designs (maximizing or minimizing overlap.) Key Words: Sample Design, Stratified Sampling, History of Surveys, Multistage Sampling 1. Introduction In 1802, the French mathematician Laplace persuaded the French government to use a sample survey to estimate the population of France as of September 22 of that year. The method used was one he had demonstrated earlier in estimating the 1782 population of France; they first took a sample of small administrative districts known as communes, counted the total population y of each sample commune, and then used the known number of births in both the communes and the nation – x and X, respectively – to calculate the ratio estimator Y = X(y/x).
    [Show full text]
  • Comparing Four Bootstrap Methods for Stratified Three-Stage Sampling
    Journal of Official Statistics, Vol. 26, No. 1, 2010, pp. 193–207 Comparing Four Bootstrap Methods for Stratified Three-Stage Sampling Hiroshi Saigo1 For a stratified three-stage sampling design with simple random sampling without replacement at each stage, only the Bernoulli bootstrap is currently available as a bootstrap for design-based inference under arbitrary sampling fractions. This article extends three other methods (the mirror-match bootstrap, the rescaling bootstrap, and the without-replacement bootstrap) to the design and conducts simulation study that estimates variances and constructs coverage intervals for a population total and selected quantiles. The without-replacement bootstrap proves the least biased of the four methods when estimating the variances of quantiles. Otherwise, the methods are comparable. Key words: Multistage sampling; high sampling fractions; resampling methods; quantile estimation. 1. Introduction For most stratified multi-stage sampling designs, unbiased variance estimators of statistics expressed by linear functions of the observations are available. However, for nonlinear statistics and functionals, closed-form variance formulas are often unavailable. Consequently, bootstrap methods can serve for variance estimation of such statistics under stratified multi-stage sampling designs. When the sampling fractions at the first stage are small, bootstrap methods for stratified multi-stage sampling designs are simplified because without-replacement sampling can be approximated by with-replacement sampling (Shao and Tu 1995, p. 235). But when the sampling fractions at the first stage are not negligible, bootstrap methods for consistent variance estimation become complicated and few have been developed. For instance, for a stratified three-stage with simple random sampling without replacement at each stage (ST–SI3) with arbitrary sampling fractions, no bootstrap procedure is available except the Bernoulli Bootstrap (BBE) proposed by Funaoka, Saigo, Sitter, and Toida (2006) for the 1997 Japanese National Survey of Prices (NSP).
    [Show full text]
  • RANK TESTS with DATA from a COMPLEX SURVEY 1. Introduction
    RANK TESTS WITH DATA FROM A COMPLEX SURVEY THOMAS LUMLEY AND ALASTAIR SCOTT* Abstract. Rank tests are widely used for exploratory and formal inference in the health and social sciences. With the widespread use of data from complex survey samples in medical and social research, there is increasing demand for versions of rank tests that account for the sampling design. We propose a general approach to constructing design-based rank tests when comparing groups within a complex sample and when using a national survey as a reference distribution, and illustrate both scenarios with examples. We show that the tests have asymptotically correct level and that the relative power of different rank tests is not greatly affected by complex sampling. Keywords: Complex sampling; Multistage sampling; Rank Tests; Sample surveys. 1. Introduction Rank-based tests are widely used by researchers in the social and health sciences. Data from complex multistage survey designs are increasingly important in these areas, with more and more large studies publishing public-use data. The extension of rank tests to data from complex samples would be valuable to researchers who wish to do the same analyses with data from, say, NHANES or the British Household Panel Survey as they would do with data from a cohort or cross-sectional sample. In this paper we give a general and computationally simple approach to design- based rank tests with complex sampling. The term design-based here means that two criteria should be satisfied: that the same population null hypothesis is tested regardless of the sampling scheme and that the test has the specified level, at least asymptotically, when this hypothesis is true.
    [Show full text]
  • STATISTICS PAPER I (To Set 10 Question, 5 from Each Half to Attempt Six Questions, Three from Each Half)
    STATISTICS PAPER I (To set 10 Question, 5 from each half to attempt six questions, three from each half) FIRST HALF Probality : Event and sample space, Probality measure and probality space, Statistical and methematical definition of probality, total and compound probability, conditional probability, independence of events in probability. Bay’s theorem, random variables discrete and continues, Repeated trials, expectation of sum and product of independent random variables, Tchebycheft’s inequality, Weak law of large number Bernoulli theorem, Central limit theorem, standard probability, distributions binomial, normal, exponencial, poisson, rectangular multinomial, geometric and hypergeometric, probability generating function, moment generating function and characteristic function. Bivariate normal distribution, gamma, beta and power series, distributions, probability functions, marginal and conditional distributions, conditional mean and conditional variance. Sampling distributions and Test of Significanance : Definiton of sampling distribution and standard error, sampling distribution of sample mean drawn from a normal universe, sampling distribution of t, x2 and F. Statistics their proertics and uses, s.e. of quantiles, multinominal proportion, s.e. of functions of statistics, uses of s.e. in large sample teste, Testing goodness of fit by Chisquare test, Test of independence of attributes, Kolmogoroy test run test and sign tests. SECOND HALF Finite difference : Newton’s forward and backward interpolation formulae, pagranges formula,
    [Show full text]
  • Multistage Sampling Designs in Fisheries Research: ~Pplicationsin Small Streams
    Multistage Sampling Designs in Fisheries Research: ~pplicationsin Small Streams David G. Hankin Department of Fisheries, Humboldt State University, Arsata, CA 95.52 1, USA Hankin, D. G. 1984. Multistage sampling designs in fisheries research: applications in small streams. Can. J. Fish. Aquat. Sci. 41: 1575-1591. A common, although generally unrecognized, use of multistage sampling designs in freshwater fisheries research is for estimation of the total number of fish in small streams. Here there are two stages of sampling. At the first stage one selects a sample of stream sections, usually of equal length, and at the second stage one estimates the total number of fish present in each selected section. This paper argues that the conventional practice of selecting stream sections of equal length is ill-advised on both biological and statistical grounds, and that errors of estimation of fish numbers within selected sections will usually be small compared with errors of estimation resulting from expansion of sampled sections to an entire stream. If stream sections are instead allowed to vary in size according to natural habitat units, then alternative two-stage sampling designs may take advantage of the probable strong correlation between habitat unit sizes and fish numbers. When stream sections of unequal sizes are selected with probabilities proportional to their size (PPS), or measures of the sizes of selected sections are incorporated into estimators, one may substantially increase precision of estimation of the total number of fish in small streams. Relative performances of four alternative two-stage designs are contrasted in terms of precision, relative cost, and overall cost-effectiveness.
    [Show full text]