Download

Tracking Change Over Time Ensuring Comparability

Upload
0
Embed
Share
Upload and view presentations on any device and embed the player to your website! --- > >
Upload PPT
Upload PPT

Download this Presentation

latifa

2021-10-12 published at Celebrities category

Presentation Transcript

1.Tracking Change Over Time: Ensuring Comparability Kinnon Scott DECPI More Frequent, More Timely and More Comparable Data for Better Results, PREM Week, April 20, 2011
2.The consequences of not focusing on comparability Misallocation of resources
3.Asset Indices: Poorest Quintile Overlap Source: Filmer and Scott, “Assessing Asset Indices” Demographics, forthcoming.
4.The consequences of not focusing on comparability Misallocation of resources Loss of Credibility Effort to move to empirically based policy making Undermines all data (whole survey or NSO)
5.Comparability Across countries Income vs. consumption PPPs Education levels Within country at one point in time Spatial cost of living adjustments Seasonality (Afghan for example) Within country over time
6.It’s not rocket science Ask the same question Verbatim Categories Change with the changing world Reference periods Matters in consumption: collapsing, memory, etc. (Deaton and Grosh, 2000 for overview) Same time in the year- “Monga” Bangladesh Fixed reference period in moving survey: Uzbek education Respondent Random adult lack of knowledge (Cull and Scott, 2010; Bardasi et al, 2010) Head of hhlds attitude Aggregation More aggregation, lower level, memory issue
7. Use verbatim questions
8.CWIQ 2003 DHS 2003 CENSUS 2000 GLSS 1998 Measuring access to improved water sources in Ghana Use questions from previous surveys Examples taken from ADP, Neil Fantom, DECDG
9.Affects the trends in indicators: Nigeria
10.Uzbekistan-Reference Periods
11.Field Work Interviewers, supervisors, data entry staff (and tools) all affect data collection Quality control Training 2-3 days versus 2-3 weeks Supervision Administrative vs substantive Data entry Ex Post Concurrent data entry (Glewwe and Dang, 2008) CAPI (SERPro, 2003; Caeyers, Chalmers and De Weerdt, 2010)
12.Peru 2001-2004 Poverty numbers questioned Did an in-depth review of method of Questionnaire design Unit non-response Item non-response Imputations Non-sampling errors One of the key issue was decline in supervision and data quality controls in the field Changes in survey quality changed poverty levels!
13.How to improve questions? Add new questions without dropping old Overlap with new versions-Income Colombia Experiment with new questions in parallel Within existing survey (Brazil POF) At pilot phase (Niger) Ex post (Malawi)
14.Summary Comparability needs to be major focus Can be affected by all stages of the survey (questionnaire design through field work through imputations for non-response) Test all changes and overlap series Check for comparability when using existing data sets

Latest Release