View
237
Download
0
Tags:
Embed Size (px)
Citation preview
Consumer Expenditure Survey Redesign
Jennifer EdgarBureau of Labor Statistics
COPAFS Quarterly MeetingMarch 4, 2011
Motivation for Redesign
Reducing measurement error Maintaining/improving response rates Meeting user needs Changes in society and technology Need for flexibility
Measurement Error: Evidence
The ratio of CE aggregate expenditures to PCE aggregate expenditures for many spending categories has shown a declining trend
Subgroup comparisons show differences in expenditure reporting by mode of interview and use of recall aids
The 2008 CE Internal Program Review Report cites item nonresponse and measurement error as areas of concern
5
Measurement Error: Possible Sources
Complexity: detailed information Conditioning: 22 sections, 5 waves Proxy reporting: household respondent Recall effects: 3-month reference period Survey length: 60 minute average Telephone administration: 33%
6
Meeting User Needs
Varied users: CPI Bureau of Economic Analysis (BEA) IRS U.S. Defense Department U.S. Bureau of the Census Other federal and state agencies Businesses, researchers and the general
public
Meeting User Needs
Varied needs: Detailed monthly expenditure estimates Annual household expenditures
By demographic subgroupBy geographic region
Housing values and rental equivalence Sales tax information
Changes in Society and Technology
Shifts in consumer behaviors Automatic withdrawals Online bill paying Online shopping Big box stores (e.g. Target, Costco)
Technological advances Increasing respondent use of
technology (internet, smart phones) Emergence of new data collection
methods
Need for Flexibility
Allow for faster implementation of changes To respond to changes in marketplace To incorporate changes in data
collection technology
Allow for mixed-mode interviewing
Redesign Overview: Gemini
Mission Improve data quality through a
verifiable reduction in measurement error, with a particular focus on under-reporting
Objective Develop a detailed research plan for
implementing a redesign of the CE surveys, including piloting, evaluation, and transition
Gemini Tasks
2009-2010 Produce database of CE research Refine data requirements and
priorities Define data quality for CE
2010-2011 Sponsor internal and external events
for information gathering and expert review
Gemini Tasks
2010-2013 Conduct research2012 Assess user impact of design
alternatives2013 Recommend a survey redesign Propose a transition roadmap2013 + Piloting, Evaluation, Transition
15
Gemini Research
Current research addressing: Recall period Interview frequency Split questionnaire designs Interview length and reporting quality Records availability and utility Feasibility of a balance edit Individual diaries Online diaries
Gemini Project Website
www.bls.gov/cex/geminimaterials.htm
Gemini overview Gemini events Gemini presentations CE survey methods research CE methods and survey forms Other redesign related events and
studies 17
2010 Methods Workshop
Objectives:
Identify existing knowledge on key topics
Obtain recommendations for research to support and inform the CE redesign
Workshop Approach
Identified Five key survey topics Two experts for each topic
Write paper and give presentation Additional experts to participate in
discussion
BLS ‘issue papers’ Explaining why the topic is considered key Describing existing CE research Asking key questions for redesign process
Methods Workshop Topics
Proxy Reporting Recall Interview Structure Global Questions Split Questionnaire Designs CE “From Scratch” Next Steps
Methods Workshop Results
Papers and presentations provided useful information on each topic
Discussion allowed CE to ask follow-up questions and refocus discussion
No absolute answers or definite research projects identified, but promising leads in many directions
Key Recommendations 1
Identify objectives For the survey
Minimum CPI requirements−Can they be changed?
Key estimates produced−What is required in the future?
For the redesignIdentify metrics to evaluate redesign
alternatives and final implemented design
Key Recommendations 2
Reconsider recall Respondents may not be able to
accurately recall the expenditure information
Respondents may not have encoded the expenditure informationAutomatic paymentsProxy issues
Respondents may not align memories with CE categories or interview structure
24
Key Recommendations 3
“It depends...” The effectiveness/feasibility of
most issues vary by expenditure category, CU type, etc.Proxy reportingLong/short recall periodsFlexible or rigid interview structuresGlobal questionsUse of technology