RTI International is a trade name of Research Triangle Institute 3040 Cornwallis Road ■ P.O. Box 12194 ■ Research Triangle Park, North Carolina, USA 27709.

Slides:



Advertisements
Similar presentations
1 RTI International is a trade name of Research Triangle Institute 3040 Cornwallis Road P.O. Box Research Triangle Park, North Carolina, USA
Advertisements

EVAULATION OF THE NSCRG SCHOOL SAMPLE Donsig Jang and Xiaojing Lin Third International Conference on Establishment Surveys Montreal, Canada, June 21, 2007.
Census and Statistics Department Introduction to Sample Surveys.
Towards a Better Integration of Survey and Tax Data in the Unified Enterprise Survey Claude Turmelle Statistics Canada ICES-III Montréal, Québec, Canada.
STATISTICS FOR MANAGERS LECTURE 2: SURVEY DESIGN.
Statistics for Managers Using Microsoft® Excel 5th Edition
Chapter 7 Sampling Distributions
© 2002 Prentice-Hall, Inc.Chap 1-1 Statistics for Managers using Microsoft Excel 3 rd Edition Chapter 1 Introduction and Data Collection.
Who and How And How to Mess It up
Sampling.
Why sample? Diversity in populations Practicality and cost.
Sampling Prepared by Dr. Manal Moussa. Sampling Prepared by Dr. Manal Moussa.
Aaker, Kumar, Day Ninth Edition Instructor’s Presentation Slides
Consumer Expenditure Survey Redesign Jennifer Edgar Bureau of Labor Statistics COPAFS Quarterly Meeting March 4, 2011.
Statistical Methods Descriptive Statistics Inferential Statistics Collecting and describing data. Making decisions based on sample data.
The Excel NORMDIST Function Computes the cumulative probability to the value X Business Statistics: A First Course, 5e © 2009 Prentice-Hall, Inc
7-1 Copyright ©2011 Pearson Education, Inc. publishing as Prentice Hall Chapter 7 Sampling and Sampling Distributions Statistics for Managers using Microsoft.
Basic Business Statistics (8th Edition)
Sampling Methods.
Course Content Introduction to the Research Process
FINAL REPORT: OUTLINE & OVERVIEW OF SURVEY ERRORS
Marketing Research Aaker, Kumar, Day Seventh Edition Instructor’s Presentation Slides.
Sampling Designs and Sampling Procedures
Chapter Eleven: Basic Sampling Issues
Sample Design.
SHADES OF GREEN Growing Green Jobs for Florida 2010 Florida Green Jobs Survey February 24, 2012 Tallahassee, FL.
The new HBS Chisinau, 26 October Outline 1.How the HBS changed 2.Assessment of data quality 3.Data comparability 4.Conclusions.
Copyright 2010, The World Bank Group. All Rights Reserved. Agricultural Census Sampling Frames and Sampling Section A 1.
COLLECTING QUANTITATIVE DATA: Sampling and Data collection
Sampling. Concerns 1)Representativeness of the Sample: Does the sample accurately portray the population from which it is drawn 2)Time and Change: Was.
Definitions Observation unit Target population Sample Sampled population Sampling unit Sampling frame.
Sampling Basics Jeremy Kees, Ph.D.. Conceptually defined… Sampling is the process of selecting units from a population of interest so that by studying.
From Sample to Population Often we want to understand the attitudes, beliefs, opinions or behaviour of some population, but only have data on a sample.
Chapter 1 Introduction and Data Collection
12th Meeting of the Group of Experts on Business Registers
CHAPTER 12 – SAMPLING DESIGNS AND SAMPLING PROCEDURES Zikmund & Babin Essentials of Marketing Research – 5 th Edition © 2013 Cengage Learning. All Rights.
Multiple Indicator Cluster Surveys Survey Design Workshop Sampling: Overview MICS Survey Design Workshop.
Using the Dun & Bradstreet (D&B) Database as a Sampling Frame for Company Surveys Sarah Cotton, Anil Bamezai.
Lesli Scott Ashley Bowers Sue Ellen Hansen Robin Tepper Jacob Survey Research Center, University of Michigan Third International Conference on Establishment.
Sampling Methods. Definition  Sample: A sample is a group of people who have been selected from a larger population to provide data to researcher. 
Copyright ©2011 Pearson Education 7-1 Chapter 7 Sampling and Sampling Distributions Statistics for Managers using Microsoft Excel 6 th Global Edition.
Variables, sampling, and sample size. Overview  Variables  Types of variables  Sampling  Types of samples  Why specific sampling methods are used.
Sampling “Sampling is the process of choosing sample which is a group of people, items and objects. That are taken from population for measurement and.
© 2013 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.
Statistics for Managers Using Microsoft Excel, 4e © 2004 Prentice-Hall, Inc. Chap 1-1 Statistics for Managers Using Microsoft ® Excel 4 th Edition Chapter.
DTC Quantitative Methods Survey Research Design/Sampling (Mostly a hangover from Week 1…) Thursday 17 th January 2013.
Lecture 9 Prof. Development and Research Lecturer: R. Milyankova
Chapter Ten Basic Sampling Issues Chapter Ten. Chapter Ten Objectives To understand the concept of sampling. To learn the steps in developing a sampling.
1. Population and Sampling  Probability Sampling  Non-probability Sampling 2.
5-4-1 Unit 4: Sampling approaches After completing this unit you should be able to: Outline the purpose of sampling Understand key theoretical.
A Comparison of Variance Estimates for Schools and Students Using Taylor Series and Replicate Weighting Ellen Scheib, Peter H. Siegel, and James R. Chromy.
Basic Business Statistics, 10e © 2006 Prentice-Hall, Inc.. Chap 7-1 Chapter 7 Sampling Distributions Basic Business Statistics.
RTI International is a trade name of Research Triangle Institute 6110 Executive Blvd. ■ Suite 902 ■ Rockville, Maryland, USA Phone
Chapter Eleven The entire group of people about whom information is needed; also called the universe or population of interest. The process of obtaining.
Copyright © 2013, 2009, and 2007, Pearson Education, Inc. 1 In an observational study, the researcher observes values of the response variable and explanatory.
Chapter 6: 1 Sampling. Introduction Sampling - the process of selecting observations Often not possible to collect information from all persons or other.
Basic Business Statistics, 11e © 2009 Prentice-Hall, Inc. Chap 7-1 Chapter 7 Sampling and Sampling Distributions Basic Business Statistics 11 th Edition.
Do Monetary Incentives Increase Business Survey Response Rates? Results from a Large Scale Experiment Paul Biemer, RTI International and University of.
Basic Business Statistics, 8e © 2002 Prentice-Hall, Inc. Chap 1-1 Inferential Statistics for Forecasting Dr. Ghada Abo-zaid Inferential Statistics for.
Basic Business Statistics
CASE STUDY: NATIONAL SURVEY OF FAMILY GROWTH Karen E. Davis National Center for Health Statistics Coordinating Center for Health Information and Service.
Sampling technique  It is a procedure where we select a group of subjects (a sample) for study from a larger group (a population)
SAMPLING Why sample? Practical consideration – limited budget, convenience, simplicity. Generalizability –representativeness, desire to establish the broadest.
Population vs. Sample. Population: a set which includes all measurements of interest to the researcher (The collection of all responses, measurements,
1 STAT 500 – Statistics for Managers STAT 500 Statistics for Managers.
© Copyright McGraw-Hill CHAPTER 14 Sampling and Simulation.
Formulation of the Research Methods A. Selecting the Appropriate Design B. Selecting the Subjects C. Selecting Measurement Methods & Techniques D. Selecting.
AC 1.2 present the survey methodology and sampling frame used
Chapter Ten Basic Sampling Issues Chapter Ten.
Sampling: How to Select a Few to Represent the Many
Presentation transcript:

RTI International is a trade name of Research Triangle Institute 3040 Cornwallis Road ■ P.O. Box ■ Research Triangle Park, North Carolina, USA Phone The O*NET Data Collection Program: Improving Efficiency in a Multistage Complex Establishment Survey Marcus Berzofsky, Brandon Welch, Susan McRitchie, and Rick Williams Third International Conference on Establishment Surveys Montreal, Canada June 21, 2007

2 Outline O*NET Background and Goals Design Challenges Initial Design Introduction of Wave Design Introduction of Model Assisted Sampling Conclusions

3 O*NET Study Goals Produce nationally representative estimates on 810 occupations plus new and emerging occupations Deductive approach which utilizes a common set of prespecified items to determine the critical characteristics for each occupation Respondents are incumbents in the occupation Estimates produced for four domains Skills Work Context Work activities Knowledge

4 O*NET Background Conducted by the National Center for O*NET Development and RTI International Sponsored by the U.S. Department of Labor Data collection began in June 2001 To date Over 143,000 establishments selected Over 100,000 employee respondents

5 Design Challenges: Basic Question How do we identify incumbents in all 810 occupations?

6 Design Challenges: Simple Solution Generate list of incumbents for each occupation Will work for some occupations, but not all Lawyers – list from ABA Secretaries no such list exists

7 Design Challenges: More Complete Solution Take advantage of the likelihood of occupations to be employed within the same industries and, hence, establishments Conduct general population survey of establishments and then sample incumbents within selected establishments

8 Design Challenges: Challenge of Linking Occupations to Industries How to determine which occupations are found in which industries?

9 Design Challenges: Solution to Linking Occupations to Industries U.S. Bureau of Labor Statistics (BLS) conducts the Occupational Employment Statistics (OES) survey Obtains estimates on the number of employees in an occupation that are found in a particular industry Identifies industries that employ each occupation Dun and Bradstreet (D&B) provides a frame of establishments by industry Used D&B frame to select establishments based on information provided by OES

10 Initial Design Target Population: All non-military, non- institutionalized incumbents in the 50 United States plus DC Traditional probability based design Targeted a large set of diverse occupations for initial sample Sample size of 12,000 establishments targeting 210 occupations

11 Initial Design Two-stage cluster design First stage: Select establishments Second stage: Select employees Linked up to 10 occupations to a single establishment Used PPS random sampling to select up to 10 occupations employed in the establishment’s industry Reduces cost and number of establishments sampled by identifying more than one occupation at a time compared to sampling one occupation at a time under a list design Reduces total burden by only asking about up to 10 occupations likely employed by establishment

12 Initial Design: Selection of Establishments Establishments stratified by size (# of employees), industry groupings Establishments selected by Sequential PPS sampling

13 Initial Design: Selection of Employees Ask point of contact (POC) at establishment about occupations on list Selection of employees For occupations present, POC rosters employees SRS of employees selected from each occupation present

14 Initial Design Occupation s Select Establishments Select Employees OES/D&B Occupation-Industry linkage

15 Introduction of Wave Design: Motivation Large single sample: Inefficiently covered all occupations Limited the information available to inform future follow-up samples Was not able to adequately target all industries which meant that some occupations were not linked to many of the selected establishments

16 Introduction of Wave Design: Wave Design Split occupations into smaller sets of approximately 50 occupations Released sample in sub-waves Used information from prior sub-waves to inform sample allocation of future sub-waves Smaller sample size per sub-wave (approx. 3,000 establishments) More easily allowed sample to be allocated to more difficult to find occupations

17 Introduction of Wave Design: Clustering of Occupations Desired smaller set of occupations to contain occupations found in common industries Cluster analysis conducted to determine the most efficient manner to group occupations Clustered occupations by the distribution of industries associated with occupation from OES survey Reduced the dimensionality by excluding industries that contained less than 1% of the occupation for all occupations Uses SAS clustering procedures to group occupations Combined groups of occupations into sets of approximately 50 occupations

18 Wave Design: Targeting Industries and Coverage OES lists all industries in which an occupation is found Initial targeting of all occupations leads to a more efficient sample Coverage requirements ensure that all industry groupings that employ an occupation are sampled Based on the cluster of occupations in a wave, industries were targeted based on their likelihood of finding the occupations Substantive experts (I/O psychologists) assign concentration level to each industry to which occupation is linked Industries with a higher level concentration of the occupations were given a greater chance of selection

19 Wave Design: Targeting Industries and Coverage Coverage analysis indicated that coverage requirements could be relaxed from a high coverage level while not introducing additional coverage bias and simultaneously reducing costs by eliminating occupations with little likelihood of being found in the establishment’s industry

20 Wave Design: Burden Two types of burden on O*NET Establishment burden:  Time spent by POC at establishment determining if occupations are employed by establishment  Rostering employees as needed by POC  Passing out questionnaires to selected employees Employee burden:  Time spent by employee completing questionnaire (approx. 30 min)

21 Wave Design: Burden Methods used to minimize burden Establishment burden:  POC asked to roster only the first five occupations found at the establishment  Selected establishments (physical location) not eligible to be selected again for 12 months  Selection algorithm  Select no more than 20 employees from an establishment  Select no more than 8 employees per occupation Employee burden  Only asked respondents to complete one domain questionnaire (skills, work context, knowledge, or work activities)

22 Wave Design Occupations Wave Wave n Select Employees Select Estabs Targeting industries Cluster Analysis Select Employees Sample Size Completed ?

23 Model Assisted Sampling: Motivation Level of effort to obtain desired number of questionnaires varied greatly across occupations E.g., Secondary school teachers were easy to find and responded at very high levels E.g., Roustabouts are more difficult to identify and respond at a lower level Wanted method to control level of effort while ensuring that employee sample was representative of the occupation

24 Model Assisted Sampling: Definition Uses known information about the occupation to model how it is distributed by Census region (4 domains) Establishment size (4 domains) Industry division (12 domains) Sampling of an occupation is completed only after minimum targets are met for each of the occupation’s domains

25 Model Assisted Sampling: Application Sample yield monitored by domain Employee sample selection ceased for an occupation in domains where model parameters have been achieved Future sub-wave sample establishment selection eliminated occupations in completed domains Analysis indicated that estimates under MAS are not substantively different than estimates produced under traditional sampling paradigm

26 Model Assisted Sampling Occupations Wave Wave n Controlling Sample Allocation Select Employees Define MAS distribution Select Estabs Targeting industries Cluster Analysis Define MAS distribution Select Employees Controlling Sample Allocation MAS Domains Complet e?

27 Conclusions Multi-year studies often require modifications over time O*NET illustrates how relatively small changes can greatly improve the efficiency of a sample Often helpful to test or simulate changes to determine their impact on a study

28 Questions?