electronic questionnaire collection at statistics canada milana karaganis marc st-denis
TRANSCRIPT
Electronic Questionnaire Collection at
Statistics Canada
Milana KaraganisMarc St-Denis
Outline
Electronic Questionnaire project Electronic Questionnaire for business surveys
23-04-21Statistics Canada • Statistique Canada2
Electronic Questionnaire Project Build generic EQ service to be used by all surveys
• EQ platform
• Standard approach to EQ Create a critical mass of EQ service users
• 160+ surveys
• Business surveys and some social surveys Reduce collection and operations costs Timeline: April 1, 2010 -> March 31, 2016
23-04-21Statistics Canada • Statistique Canada3
Electronic Questionnaire Project
E-questionnaire solution
• Support all EQ applications for all surveys
• Use Statistics Canada infrastructure
• Comply with security and confidentiality standards
• Automate EQ survey application development -> build an engine to render EQ applications based on provided inputs, E-questionnaire Generation System (EQGS)
23-04-21Statistics Canada • Statistique Canada4
Electronic Questionnaire Project
23-04-21Statistics Canada • Statistique Canada5
Respondent
EQ Collection Portal (with EQ application for a
survey)
EQ backend:- Respondent repository
- Database- E-mail service
- Data management service- Survey management service
- Questionnaire generation system
etc.
PUBLIC ACCESS ZONE
Statistics Canada internal network
Electronic Questionnaire Project
Standard approach to EQ – corporate solution• Develop EQ applications according to EQ standards
• Apply standard look and feel
• Build functionalities required for all surveys
• Use standard collection strategies
EQ as initial collection mode in sequential mode collection
• Use standard development process
Shorten development cycle
23-04-21Statistics Canada • Statistique Canada6
23-04-21Statistics Canada • Statistique Canada7
23-04-21Statistics Canada • Statistique Canada8
23-04-21Statistics Canada • Statistique Canada9
Progress To-Date
EQ solution implemented• Change Management process
• First set of EQ standards released
• Survey development process created
• Initial collection strategies established
Deployed 39 EQ applications and realized efficiencies
Overall success:• Lessons learned
• Big cultural shift
EQ Among Business Surveys
Outline• Evolution of EQ collection strategies
• Results observed to date
• Survey collection strategy guidelines
• Results of tests
23-04-21Statistics Canada • Statistique Canada10
Collection Strategies Among Business Surveys
Moving from conventional collection modes (PAPI & CATI) to EQ (over 100 surveys in 5 yrs)
Evolution of multi-modal collection with EQ as primary mode
Measuring results and improving • Respecting data quality, timeliness, budgets
Reviewing collection strategies
23-04-21Statistics Canada • Statistique Canada11
Reviewing Collection StrategyQuarterly Services Indicators (QSI) Survey
23-04-21Statistics Canada • Statistique Canada12
Services Indicator Survey (quarterly ) (QSI) 2011
Q4 2011 Q1 2012
Mail out / E-invitation Oct 3, 2011 Mail out / E-invitation Jan 4, 2012e-reminder / fax Oct 17 e-reminder Jan 18Phone reminder (all) Oct 24 e-reminder Jan 25Phone reminder (all) Nov 7 e-reminder Feb 8e-reminder / fax Nov 14 Phone reminder (all) Feb 15e-reminder / fax Nov 28 e-reminder Feb 29Fax to EQ Dec 5 Phone and fax (all) Mar 7End of collection Dec 9, 2011 End of collection Mar 9, 2012
Q4 2011 EQ Mail Total
Final Sample 1,525 98.3% 27 1.7% 1,552
Response Rate 58.5% 50.0% 58.0%
# of cases in NRFU 905 59.3% 22 81.5% 927
23-04-21Statistics Canada • Statistique Canada13
Q1 2012 EQ Mail Total
Final Sample 1,351 96.4% 50 3.6% 1,401
Response Rate 65.6% 46.2% 65.6%
# of cases in NRFU 683 50.6% 41 82.0% 724
Reviewing Collection StrategyQuarterly Services Indicators (QSI) Survey
EQ Collection Strategy Guidelines
Developing tools to ensure optimal collection strategies (monthly; quarterly; annual)• Follow-up schedule geared to multi-mode collection
Standardise approach towards efficiencies Considering respondent burden, data quality,
response rates, costs
23-04-21Statistics Canada • Statistique Canada14
Date Days after Mail out Type of FUPre-contact n/aMail-out / E-invitation send date: 0Collection Start date:
Bounce-backs follow up 21 week max to complete then switch to mail for those unable to get correct email
1st Follow up attempt date: Phone 30 only for those without e-mail address (mail units)EQ 30 1st reminder emailFax X no fax unless requested
Mail X no mail unless requested
2nd Follow up attempt date: Phone 60 only for those without e-mail address (mail units)EQ Every 2nd week 2nd reminder emailFax X no fax unless requested
Mail X no mail unless requested
3rd Follow up attempt date: Phone 90 only for those without e-mail address (mail units)EQ Every 2nd week 2nd reminder emailFax X no fax unless requested
Mail X no mail unless requested
4th Follow up attempt date: Phone 120 only for those without e-mail address (mail units)EQ Every 2nd week 3rd reminder email
Fax 120fax reminder + copy of questionnaire to those without email address
Mail 120 mail only to those without email address or fax number
5th Follow up attempt date: Phone 140 all outstandingEQ Every 2nd week 5th reminder email
Fax 140fax reminder + copy of questionnaire to those without email address
Mail 140 mail only to those without email address or fax number
6th Follow up attempt date: Phone 160 all outstanding unitsEQ 105 6th and final reminderFax X no fax
Mail X no mail
EQ Experimental Design for the UES
Mailout Mar 18E-invitation Mar 28
T1 (Standard)
T2(E-mails only)
T3(phone attempt)
T4(phone attempt + e-mails)
Reminders E-mail only E-mail only 1st attempt – phone;3 next – E-mail
1st attempt – phone;3 next – E-mail
Reminder schedule
1st April 262nd May 263rd June 234th July 25
1st April 262nd May 93rd May 264th June 7
1st April 26-292nd May 263rd June 234th July 25
1st April 26-292nd May 93rd May 264th June 7
Phone follow-up Yes No (firm appt only)
Yes No (firm appt only)
Experiment end date
August 1 June 20*July 11
August 1 June 20*July 11
Sample size allocated to the treatment
1125 1102 1148 1114
23-04-21Statistics Canada • Statistique Canada16
Results of the experimental design
T1 T2 T3 T4 %u %u %u %uApril 26 13% 12.5% 15% 12%May 9 24% 22% 24% 22%May 26 36% 41% 37% 45%June 7 44% 43% 46% 49%June 20 51% 46% 53% 52%August 1 66% 54% 68% 58%Sept 12 77% 76% 78% 75%Oct 14 82% 81% 82% 81%
Progressive unweighted return rate (%) for each treatment (2011)
T1 T2 T3 T4
3.8 2.8 4.3 3.9
Number of call attempts by survey treatment
17 Statistics Canada • Statistique Canada 23-04-21
Highlights
Treatments consisting of 4 e-reminders only resulted in 46+% as of June 20th
At end of collection (Oct 14) the 4 treatments had very similar response rates
T2 showed the lowest attempts and lowest TPU Limited worth conducting telephone NRFU for
first few months; still reach similar final response rates with fewer phone attempts
23-04-21Statistics Canada • Statistique Canada18
Access Code Letter
Piloted for UES 2012 collection Conducted pre-contact for birth cases in sample Unable to contact cases = mailed an access
code letter providing hyperlink to an EQ More difficult cases since unable to reach during
pre-contact After 30 days, telephone NRFU
23-04-21Statistics Canada • Statistique Canada19
Access Code Letters: Results
StatusAccess code letter
# units (total)
Resp. (EQ) Resp. (Mail)
Non-Resp. Out of Scope
In Progress
Total 765 150 (20%) 39 (5%) 52 (7%) 74 (9%) 450 (60%)
Collection status on July 16, 2012 for units that received an access code letter
20 Statistics Canada • Statistique Canada 23-04-21
The Future Successful in achieving its objectives Mission critical monthly surveys UES (60 surveys) + LFS Add new functionalities to EQ platform Respondent EQ = Interviewer EQ Assess respondent’s level of tolerance to edits Streamline processes and operations Monitor quality, response rates, costs Research
23-04-21Statistics Canada • Statistique Canada21
23-04-21Statistics Canada • Statistique Canada22