Data Quality and Cognitive Processing of Questions
Marek Fuchs
University of Kassel, Germany marek.fuchs@uni-kassel.de
© Marek Fuchs, Universität Kassel
Humanizing Elements in Survey Data Collection
© Marek Fuchs, Universität Kassel
Introduction
© Marek Fuchs, Universität Kassel
Layout kept constant
Same location for text and video
Response scale identical
Social Interface Theory
• Respondents ascribe social properties to computers
• React to questions administered by computers similarly to interviews with human interviewers.
• Higher levels of social presence
• More pronounced social desirability
• Increased underreporting of sensitive information
© Marek Fuchs, Universität Kassel
Survey Methods Research
• AV-CASI: More positive responses to sensitive questions (Katz et al., 2007)
• AV-CASI: Less contaminated by social desirability (Gerich,
2007)
• Web surveys: marginal effects of humanizing elements (Couper,
Tourangeau, & Steiger, 2003)
• Anthropomorphizing elements increase respondents‘ motivation and involvement
• Audio-visual channel will promote comprehensive question understanding
• Increased reporting
© Marek Fuchs, Universität Kassel
Methods
Universe
• Student online access panel at the University of Kassel, Germany
Sample
• N = 1.148 (880 for this experiment)
• Response rate (within the panel): 49%
Field work
• Summer 2007, e-mail invitation and 2 reminders
• Incentive: lottery drawing of book vouchers
Experiment
• Text vs. video/male vs. female
• Between subjects design
Fuchs/Funke 2007, 2008
© Marek Fuchs, Universität Kassel
6
7
8
Results
7,6
5,8
6,8 6,8
5,4 5,4
5
4
3
2
1
0
Social precence Social desirability Message understanding
© Marek Fuchs, Universität Kassel
5,7
6,1
Flow
Video version
Text version
5
Results
4,4
4
3
2
1
0
Sensitive behaviors
© Marek Fuchs, Universität Kassel
4,4
Video version
Text version
Structural Equation Models for Text vs. Video
Text-based version
• Negative effect
- Social presence
- Social desirability
• Positive effect
- Question understanding
- Flow
Video-enhanced version
• No effect
- Social desirability
- Question understanding
• Positive Effect
- Social presence
- Flow
© Marek Fuchs, Universität Kassel
Summary
• Social presence lower in video version
• Social desirability does not differ
• Message understanding does not differ
• Slightly higher flow in the video version
© Marek Fuchs, Universität Kassel
Summary
• No overall differences for sensitive behaviors
However, the cognitive processes differ
• Text version
- Social presence and social desirability promote underreporting
- Question understanding and flow increases reporting
• Video version
- Social presence and flow increase reporting
- Social desirability and message understanding have no effect
© Marek Fuchs, Universität Kassel
Discussion
Implications
• It is not only the visual presence of the interviewer
• It is also the communicative channel used to convey the question meaning
Caveats
• Student population restricts generalization
• Backchannel is keyboard and mouse only
Next steps
• More active interviewer behavior
• Heterogeneity of interviewer
• „Choose your interviewer“
© Marek Fuchs, Universität Kassel
End
Marek Fuchs
University of Kassel, Germany marek.fuchs@uni-kassel.de
© Marek Fuchs, Universität Kassel