Selfselection biases in the Internet survey:

1
Self-selection biases in the Internet survey A
case study of Northern Indiana
2002 TTRA Conference Arlington, Virginia June 24,
2002
Yeong-H Hwang Daniel R. Fesenmaier
National Laboratory for Tourism and
eCommerce University of Illinois at
Urbana-Champaign
2
Introduction
New Research Horizon

• New research opportunities by the Internet
• Travelers information technology use
• Design of efficient information systems

• Survey tools
• Tierney (2000)
• effectiveness of tourism website
• Litvin and Kar (2001)
• comparison between Internet and mall intercept
  survey

2002 TTRA Conference Arlington, Virginia
3
Introduction
Internet Survey

• Use Internet as a media for
• distributing survey instrument
• selecting sample

• Example
• e-mail list from diverse sources ? survey
• pop-up survey

2002 TTRA Conference Arlington, Virginia
4
Introduction
Internet Survey Pros and Cons
Demand for research on biases of Internet survey
2002 TTRA Conference Arlington, Virginia
5
Related Works
Survey Errors
(Dilman, 2000)
2002 TTRA Conference Arlington, Virginia
6
Related Works
Internet Survey

• Diverse presentation format of instrument
• enable to use multimedia instrument
• demand more decisions in instrument design
• ? more vulnerable to measurement error
• (ONeil and Penord, 2001)

2002 TTRA Conference Arlington, Virginia
7
Related Works
Selection of E-sample
(Bradley, 1999)

• Closed survey
• in advance selection of samples/sampling frames
• ? send survey/link to survey to those selected
• Open survey
• open to public
• banner advertisement
• Hidden survey
• pop-up survey
• enable systematic presentation (every nth
  visitor)

2002 TTRA Conference Arlington, Virginia
8
Related Works
Selection of E-sample
(Frequent use in Tourism)

• Open-and-closed survey
• sampling frame ?open to public
• sample selection from frame (closed)
• e.g.
• registration
• contact info ? sample selection
• sample from Internet conventional paper survey

2002 TTRA Conference Arlington, Virginia
9
Method
Method

• Conversion Study for Harbor/Amish Country,
  Indiana
• 4,688 mail out ? 1,845 return (39.3)

• Survey questions
• willingness to provide contact info. (name and
  e-mail)
• demographics
• travel planning behaviors
• travel related psychological characteristics
• actual trip behaviors

2002 TTRA Conference Arlington, Virginia
10
Method
Method

• Grouping by willingness to provide contact info.
• Not likely (47.3)
• Somewhat likely (25.4)
• Likely (27.3) of 1,460
  respondents

• Analyses
• Comparisons between groups in
• demographics
• travel planning behaviors
• travel related psychological characteristics
• actual trip behaviors

2002 TTRA Conference Arlington, Virginia
11
Results
Sample Characteristics

• Female 70
• Middle age respondent (36-55) 50
• Married 79
• Children at home 33
• Income (over 50K) 55

2002 TTRA Conference Arlington, Virginia
12
Results
Group Comparison Demographic Characteristics
Gender
(?2 .83)
2002 TTRA Conference Arlington, Virginia
13
Results
Group Comparison Demographic Characteristics
Age
(?2 47.27, plt.1)
2002 TTRA Conference Arlington, Virginia
14
Results
Group Comparison Demographic Characteristics
Marital Status
(?2 2.62)
2002 TTRA Conference Arlington, Virginia
15
Results
Group Comparison Demographic Characteristics
Living with Children
(?2 16.88, plt.01)
2002 TTRA Conference Arlington, Virginia
16
Results
Group Comparison Demographic Characteristics
Income
(?2 36.70, plt.01)
2002 TTRA Conference Arlington, Virginia
17
Results
Group Comparison Trip Planning Behavior
Information Request Mode
Based on Multiple responses Significant at
.01 level

• Likely group
• ? tend to use e-mail and website to
  request info.

2002 TTRA Conference Arlington, Virginia
18
Results
Group Comparison Trip Planning Behavior
Information Source for Planning
Based on Multiple responses Significant at
.1 level Significant at .01 level

• Likely group
• ? tend to use more Internet and less
  memory

2002 TTRA Conference Arlington, Virginia
19
Results
Group Comparison Trip Planning Behavior
Decision Timing
No significant difference
2002 TTRA Conference Arlington, Virginia
20
Results
Group Comparison Self-evaluated Travel Pattern
Measurement scale 1 (strongly
disagree) 5 (strongly agree)
Significant at .05 level Significant at
.01 level
2002 TTRA Conference Arlington, Virginia
21
Results
Group Comparisons Trip Behavior
Based on Multiple responses
No significant difference
2002 TTRA Conference Arlington, Virginia
22
Results
Summary of Findings
Self-selected e-samples
Age
Young
Old
Income
High
Low
Yes
No
Living with children
Internet use for trip planning
Yes
No
Travel pattern
Allocentric
Psychocentric
Actual trip behavior
Population ratio
2002 TTRA Conference Arlington, Virginia
23
Implication
Implications

• Coverage errors in self-selected e-sample
• Salience of error changes by different
  characteristics

• Be aware of Internet survey errors
• Extreme caution when generalizing findings
• Consideration for the purpose of study

2002 TTRA Conference Arlington, Virginia
24
Thank you!!!! Questions/Comments??