2004 Census of Population Test Evaluation of the Internet option

1
2004 Census of Population TestEvaluation of the
Internet option

•  
• Danielle Laroche
• Statistics Canada
• UNECE Work Session
• Ottawa, May 1618, 2005

2
Census

• Conducted every 5 years (May 16, 2006)
• Mandatory
• Self-enumeration
• Two types of questionnaires
• Short Form (2A) gt 80 of households
• Long Form (2B) gt 20 of households

3
Characteristics of Internet questionnaires

• Access code/Web site address printed on paper
  questionnaires
• Comparable to the Paper
• Question wording, instructions and response
  choices
• Editing (Validation messages)
• Non-response and partial response
• Invalid response and amount verification
• Radio buttons and check boxes
• Automated skips

4
Qualitative Testing

• Census uses high level of security which has
  specific computer requirements
• Negatively impact Internet take-up rate
• Validation messages were effective
• Respondents assume they must answer each question
  to continue

5
Comparison between Paper and Internet
questionnaires

• Item non-response rates
• Internet has much lower rates for all questions
• in both 2A and 2B
• Questionnaire failed edit follow-up rates
• Internet rates are much lower for both 2A and 2B
• Internet questionnaires are more complete and
  less expensive in terms of follow-up

6
Conclusion

• Internet offers new possibilities and new
  challenges (results are encouraging)
• General public have high expectations with regard
  to Internet questionnaires
• We need to explore the reasons for the
  differences observed and how these affect the
  quality of responses?
• Other surveys / analysis.