100 likes | 113 Views
Discover the stages of data capture in census processing, from entry to storage, focusing on data quality, integrity, and accuracy. Learn about maintaining questionnaire integrity and processing sequences for precise outcomes. A comprehensive approach addresses all aspects of data capture.
E N D
Session 9 – Data Capture: Process Stages UNSD-ESCWA Regional Workshop on Census Data Processing in the ESCWA region Doha, State of Qatar, 18-22 May 2008 Capture Data Quality Fred Highland Census Practice Architect Lockheed Martin Transportation & Security Solutions
Objective of a Population Census “ … count everyone, count them once only, and count them in the right place… ” Preston Jay Waite (US Census Bureau)
A Comprehensive View • It’s not just about paper – multi-channel data collection • Telephone • Voice capture accuracy (Service Quality v Data Quality) • Internet • Self capture – no capture accuracy or image quality issues • Inventory Control, Processing Integrity issues • Field Hand-held device • Similar to internet • Admin Records (not addressed here) • Similar to internet • Data validation • Many issues with linkage/coherence • A Comprehensive view must embrace all aspects and all channels
Delivery System Entry Storage Processing Disposition Inventory Control “Ensuring that questionnaires are accurately accounted for and managed” • Accurate capture of data is irrelevant if you lose the questionnaire
Questionnaire Pages Pages Pages Person 1 Household Person 2 Person n Item 1 Item 2 Item 1 Item n Item 2 Item n Questionnaire Integrity “Ensuring that the questionnaire and its component parts are kept together for the complete process” • Maintain the linkage of a questionnaire to its components • Pages/sheets (paper) • Household • Persons • Response items • Prevent mixing of data between questionnaires
Image Quality • What is image quality? • Qualitative not Quantitative • Capture of faithful images that are “fit for purpose” • This is subjective • No agreed-to measure exists ! • Operational definition • Images are of sufficient resolution for processing • Images represent the complete original document • Images are free of artefacts introduced by processing • Questionnaires must be represented as digital images for: • Automatic recognition (e.g. OCR, OMR) • Keying • Other image assisted functions (e.g. Coding) • Archive • Capture of high quality images is critical !
Processing Integrity “Ensure all responses are processed in the proper sequence, priority and completely through all appropriate steps” • Complicating factors • Multiple response modes • Paper, internet, telephone and hand-held device • Quantity of questionnaires • US - 140,000,000 • UK - 30,000,000 • Canada – 13,500,000 • Processing deadlines • Non-response follow-up • Census Coverage Survey (CCS) • Large number of processing steps….
Code Edit ASCII ASCII w/codes Data Capture Accuracy - Paper “Capture all of the data accurately, regardless of source and be able to assess and manage the accuracy” Questionnaire Scan 80% OCR 20% Image ASCII Keying 0.2% OMR 99.8%
Summary A comprehensive approach to quality must consider more than just data capture issues…. • Inventory Control • Questionnaire Integrity • Image Quality • Processing Integrity • Data Capture Accuracy … while ensuring accuracy and value for money A comprehensive approach identifies issues early and enables timely corrective actions. It requires an operational focus on data quality.