Implementing technology in Census Data Management - UNSD

Implementing technology in Census Data Management - UNSD

Implementing technology in Census Data Management NSA goes CAPI It is better to prevent errors than to cure them later (Redman 2001) Why NSA goes CAPI

Accuracy of data collection Improved project completion time Timeliness and availability of data Flexibility of technology Cost effective ( especially for large and reoccurring surveys)

NSA and CAPI Successful stories Agriculture Census 2013/14 Used laptop Acer 15inch Namibia Household Income Expenditure Survey 2015/16 10600 HHs Namibia Inter-censual Survey 2016 - 12000HHs Labour Force Survey (NIDS) 2016 - 12000HHs NIDS and LFS 2016 took 3 months for data processing.

Census 2021 Planning Data collection tools design Considerations Considerations of technology Software main features Case management In field automation sampling process Advanced programming languages

Case tree (menu) Selection of handheld device GPS and mapping Data synchronization and backups Storage capacity Device performance in terms of processing power Compatibility with CSPro (Window & Android platform) user friendliness in terms of weight and size

NSA & Census and Survey Processing System (CSPro) CSPro and NSA CSPro Census and Survey Processing System Developed by U.S. Census Bureau CSEntry Android application, First release in 2014 Data entry, edits and imputations and tabulation

Main CSPro features Support multiple languages Sophisticated programming languages Tightly controlled path Data synchronization Why CSPro? Case Management Segregation of roles , HH assignment, completeness

and accuracy controls Interview assignments Field supervisor can assign interviews Field Supervisor overseen the completeness and quality of interviewer output Transferring data to the central server

Why CSPro? Improve data quality through.. Automated routing (skips patterns)

Direct consistency checks and data validation during fieldwork Precoded drop down menus or coded drop down menus or radiocoded drop down menus or buttons More accurate measurement..e.g... GPS coordinates Why CSPro? Data Transfer Data is

transmitted via FTP including transfer protocols that encrypt and protect sensitive survey data Why CSPro? Cont Case tree makes navigation easier

Visibility of entered records on the same screen Lessons learned from first implementation CAPI DP- Lessons learned Planning Phase

New technology without proper methodology assessment done < resources , time & costs) Advanced programming required (CSPro etc..) ,support from U.S. Census New technology for a complex survey on very tight schedule/timeline Lack of Process /Methodology change management < awareness & understanding> Design Phase Questionnaire & System specifications plan ( Business rules, edits rules, recodes, derived variables, Tabulation) needed earlier

Incomplete project process data flow leads to risky continual application updates during fieldwork Not sufficient built in buffer time in the project plan Lack of project change control measures limited accountability DP - Lessons learned Build / Execution Android CSEntry application was being upgraded and more features added during production

Insufficient testing plan and procedures < no test end-toend > Pilot study done partially Supervisors & enumerators need basic IT skills IT risks (loss, crash, fallback) support needed CAPI and Process Change Management ICT Infrastructures Backup and Storage of data. Need to have access to internet or USB / External drives. Data is kept on the tablets until completion of

project data analysis Power Availability. Computer batteries need to be recharged. a bit difficult with remote areas with no electricity. Availability of technological support ( timely availability of IT equipments, Networking and intenet access for transfering and updating of files ) CAPI and Process Change Management

Things must be done earlier than usual at least six month for a survey Tabulation plan Finalise questionnaire content Define business process flow and field structure(e.g. sampling frame) Data editing rules & specifications ( rules, upper /lower data range, derived variables and so on) Subject Matter and Data Management must co-operate more closely Invest more time on planning for all project KEY stakeholders Ensure comprehensive system testing by both stakeholders

CAPI is (only) a tool - Project management & people stay essential Getting ready for Census 2021 Implemented measures Adopted a standard data processing tool - CSPro Invested in capacity training to Key DP staff Sponsors UNFPA ,USAID, Ukaid Instructor U.S. Census Bureau 90% of the programmers are trained in Android CSPro

programming Version.6.3 Improved process documentations supported by templates( specification requirements, change control, system testing procedures + sign off) ICT implemented mitigations Secure file transmission server (sftp) Centralized encryption for data and devices and remote device management

Assets inventory management Backup plans, recovery plans, redundancy and high systems availability. Internet provision for the field users, only until team supervisor level. ICT support staff during fieldwork ICT office for each region Envisioned System Improvement Planning Start earlier with the planning and setting up project architectural for Census 2021 Questionnaire design ( indicators and edit rules) finalized earlier prior to

system development atleast a year before Increase permanent data processing staff Sharing resources ( skills and devices) with other African countries Improve field quality assurance and control

Develop a web-based report tool enable real time reporting at field level Monitor the enumerator progress per EA as linked to the GIS-component Timely correction and solving field problems as they occurs Management daily monitoring and progress report CAPI Data flow ICT Infrastructures Thank you

Recently Viewed Presentations

  • Random generalized linear model: a highly accurate and

    Random generalized linear model: a highly accurate and

    Random generalized linear model: a highly accurate and interpretable ensemble predictor Song L, Langfelder P, Horvath S. BMC Bioinformatics 2013 ... An RF is a collection of tree predictors such that each tree depends on the values of an independently...
  • Price collection with scanner data for the Swiss (H)CPI

    Price collection with scanner data for the Swiss (H)CPI

    Price collectors go to the outlets to collect prices manually. Must be put into effect by the 14th day of a month if. no data is delivered at all (if the data from the first week is delivered, this one...
  • Part 3 Module 7 Real-world problems involving distance

    Part 3 Module 7 Real-world problems involving distance

    Part 3 Module 7 Real-world problems involving distance Two useful facts From elementary geometry we have two familiar facts that can be useful if we are trying to calculate a distance.
  • CMS Measure Development Education &amp; Outreach

    CMS Measure Development Education & Outreach

    An ongoing process to engage the public in quality measure development. Elicit feedback that will help CMS design resources that can help all of those interested in healthcare quality improvement better understand the goals of quality measurement.
  • Drug Interactions Every Health Care Provider Should Know

    Drug Interactions Every Health Care Provider Should Know

    The focus of the subsequent slides will be to outline the role of the cytochrome P450 isozymes in metabolism of commonly prescribed drugs, and to identify approaches and tools that can be used in clinical practice to avoid cytochrome P450-mediated...
  • Religion and Human Relationships - Weebly

    Religion and Human Relationships - Weebly

    Religion and Human Relationships. Roles of men and women within a Christian family. ... In Leviticus it says: 'Do not lie with a man as one lies with a woman; that is detestable' Quotes from Scripture.
  • CS211 - Cornell University

    CS211 - Cornell University

    >5 <5. Binary Search Tree (BST) A binary search tree is a binary tree with a class invariant:. All nodes in the left subtree have values that are less than the value in that node, and. All values in the...
  • CS 594: Empirical Methods in HCC Introduction to

    CS 594: Empirical Methods in HCC Introduction to

    Planned contrasts revealed that Flow menu (3466 ms) took significantly more time than the Finger Count menu (3095 ms), p <. 001 with a high effect size d = 0.8. Planned contrasts revealed that Marking menu (3646 ms) took significantly...