Multiple Indicator Cluster Surveys
Survey Design Workshop
Use of PDAs in MICS
MICS4 Survey Design Workshop
Mobile data collection
• In the recent years we have seen
development of many innovative data
collection tools using handheld PocketPC
personal digital assistants (PDAs),
smartphones or tablet personal
computers.
MICS4 Survey Design Workshop
Personal Digital Assistants
• Handheld computers
• Various operating systems
– Windows mobile
– Palm
– Others (e.g. iPhone, Nokia)
• Cell phone capability
– Plus: good communications
– Minus: security
MICS4 Survey Design Workshop
Tablet Personal Computer
• Tablet – sized computer with the key
features of a full – size personal computer
• Various operating systems
• Computer vs. cell
phone capability
MICS4 Survey Design Workshop
Tablet vs PDA
• Choosing a tablet over a PDA
- Tablets have much larger screens, better resolutions,
more space for on screen typing making things easier to
see and more useful for data entry
- Tablets and PDAs have similar battery life, though if
PDAs are used for making and receiving calls battery will
need more frequent charging
- Better security
MICS4 Survey Design Workshop
PDA Software
Variety of software available:
• CSPro
• Pocket PC creations
• Pendragon
• Techneos
• Visual CE
• Visual basic
• Available software varies by operating system
– Most available for Windows mobile
– Some only Palm
MICS4 Survey Design Workshop
Software requirements
• MICS survey requirements
– Keep track of the “data path”
– Multiple languages on the fly
– Roster manipulation
– Saving an interview in the middle
– Dynamically customized questions and response
categories
– Powerful programming language
• Able to handle complex consistency checks and skips
– Ability to run the same application in PDAs, tablets or
PCs
– Capability to administer instruments at different times
– Able to handle long/complex questionnaires
MICS4 Survey Design Workshop
CSPro
• Windows software tool for census and survey
processing
• Developed jointly by IPC-US Census Bureau,
Macro International, SerPro
• Public domain/funded by USAID
• Free download at
– www.cspro.org or
– www.census.gov/ipc/www/cspro/index.html
MICS4 Survey Design Workshop
CSPro
• Modules
– Data entry
– CAPI Data collection (laptops/tablets, PDAs)
– Data editing and imputation
– Frequencies and crosstabulations
– Data manipulation utilities
– Data dissemination
– Export to statistical packages
MICS4 Survey Design Workshop
MICS CAPI System
• All applications to collect and administer data
are written in CSPro
• Three systems:
– Interviewers - data collection – PDA/Tablet
– Supervisors – data monitoring and control –
PDA/Tablet
– Central Office - centralized data and monitor
fieldwork - PC
MICS4 Survey Design Workshop
Int 5
Int 2
Repair Utility
Interviewer’s
System
Update
Utility
Supervisor’s
System
Screen
Supervisor
Reports
Supervisor
Reports
System
Updates
System
Updates
MICS4 Survey Design Workshop
Central Office
Central Office
System
Questionnaires
• One question per
screen
– Question area
– Answer area
– (for text or numeric
responses)
– List of categories
(where appropriate)
MICS4 Survey Design Workshop
Customizable questions
• Multiple languages
– Language can be changed
during interviewing
• Questions customized to
the individual
• Response categories also
customizable
• Radio buttons for
categorical variables
• Check boxes for multiple
response variables
MICS4 Survey Design Workshop
A Typical Screen
Questions Area
Display/enter values
Possible Responses
Navigation/tools
MICS4 Survey Design Workshop
Virtual Keyboard
Values in ranges or textual text
needs to be typed
Soft Keyboard
MICS4 Survey Design Workshop
Dynamically Generate Answers
Colors have meanings
Value Sets populated
MICS4 Survey Design Workshop
Benefits
– Improved data quality?
• Pro – checking of data in real time during interview
• Con – possibility for interviewer’s to ‘avoid’ work
• Overall – positive improvements, but only with good quality
controls in place
–
–
–
–
Eliminates missing information
Shortens interviews
Data ready almost immediately
Survey indicators monitored early in fieldwork
MICS4 Survey Design Workshop
Equipment
• Power - battery charging is an issue
–
–
–
–
PDAs generally give close to a full day of work
May require charging or swapping batteries during the day
Spare battery needed for each PDA
Separate charger is good but many only charge with battery in
PDA
• Need for extra tablets/PDAs
– Needed for all trainees (not just number of
interviewers/supervisors planned for fieldwork), plus IT staff
– In case of problems/loss in the field
• Cases/covers
– Good protection needed depending on expected conditions
MICS4 Survey Design Workshop
Equipment
• Security issues
– Theft of PDAs/Tablets is a risk
– Increased risk if PDAs are also cell phones
(smartphones)
• Virus risk
– PDA viruses from downloaded programs
– PC viruses infecting SD cards (e.g. when SD cards
are used in PCs in Internet cafés)
• Cell phones to communicate with field teams
– Built in if PDA is smartphone
MICS4 Survey Design Workshop
Data loss risk
• Data loss risk
–
–
–
–
Lost/stolen PDAs/Tablets
Hardware failure
Viruses
User error
• Risk management
– Backup to SD card immediately after interview
– Copy to supervisor at end of day
– Paper questionnaires can be used in emergency
MICS4 Survey Design Workshop
PDA Costs
• Basic PDAs/Tablet cost approx $450 with no
accessories
• Plus approx $100 in needed accessories:
•
•
•
•
•
•
SD card
Spare battery
Case
Screen protector
Spare stylus
Vehicle chargers (1 per team)
– Total approx $55k
MICS4 Survey Design Workshop
Costs
• Costs balanced out by
–
–
–
–
Printing of only 10% of questionnaires
No field editors
No data entry
No computers required for data entry
• However
– Technical assistance costs are higher!
• Cost savings
– if PDAs are re-used
– Savings on secondary processing, data editing
MICS4 Survey Design Workshop
Technical assistance
• Significant extra programming compared with
classic data entry
– Questionnaire programming is more complicated
•
•
•
•
Addition of question text
More customized response categories
More complicated flow to facilitate interviewing
Possible need for multiple languages
– Data management programming much more
complicated (even than questionnaires)
• Three systems (interviewer’s, supervisor’s, central)
compared with two (data entry, data coordinator)
• Logistics of data management much more complicated
MICS4 Survey Design Workshop
Preparation and Pretesting
• Early design of questionnaires needed
– Last minute changes to questionnaires are difficult
• Programs prepared well in advance and thoroughly
pretested
• Full pretest required to properly test all aspects of the
system
• Programs must be ready and thoroughly tested before
training begins
– Updating applications during training is problematic
• Making the change is easy, but
• Preparing update for all PDAs takes time
• Updating all PDAs is very time consuming
MICS4 Survey Design Workshop
Training
– Paper questionnaires still important
• Discussions and version control
• Documentation
• Training
– Facility for training (power, desks)
– Good documentation
– Train on paper first, then PDA
• Gives the ‘big picture’
• Paper can also be used in emergency in the field
– Fieldwork testing very important
– Recommended – 4 week training,
• Including 3 separate days of field practice
• Extensive practice during training is needed
MICS4 Survey Design Workshop
Data transfer
• Logistics of transferring data to central office
– Cell phone
• Only if PDA is smartphone
– Wireless email
• Limited Wifi locations
– Internet cafés
• Virus risk on PCs
– Field control teams
• Via bluetooth or SD cards
• More reliable, but slower
MICS4 Survey Design Workshop
Field Staff
– Interviewers should be computer literate
• Younger interviewers who are comfortable with cell phones
quickly learn to use PDAs
– Interviewers need an introduction to computers and to
PDAs
– Supervisors need to be very comfortable with
applications
– Balance between good interview experience and
good computer knowledge
MICS4 Survey Design Workshop
DP staff
• Several local data processing staff needed to
troubleshoot in the field
–
–
–
–
–
–
–
–
User errors in using the system
Program malfunctions – usually related to user error
Lost data problems – usually found on the backups
PDA viruses
PC viruses on SD cards
Frozen PDAs
Problems transferring data
Problems related to data inconsistencies that cannot be resolved
by the team
– Upgrading programs
• Central office staff for secondary processing
MICS4 Survey Design Workshop
Conclusions
• Quicker, better results are possible
• Many complications
– Costs likely higher
– Logistics much more complicated
– Technical assistance needs increased
• Don't underestimate how much work it takes to
properly implement a survey, even with PDAs.
– It is very easy to conduct a poor survey
MICS4 Survey Design Workshop
Thank you
MICS4 Survey Design Workshop
Descargar

Experiences with using PDAs in household surveys.