Transcript Intent to Participate in Internet Survey

IWIS 2012
September 2012
Daejeon, Korea
Social Participation and
Intent to Participate in Internet Surveys
U-Seok Seo
徐佑錫
Department of Urban Sociology, University of Seoul
[email protected]
Gihong Yi
李圻洪
Department of Sociology, Hallym University
[email protected]
2
Introduction
• Internet Surveys in Korea increased quite consistently
▫ Internet survey: 6.4% in 2005, 16% in 2011 (KORA)
▫ Marketing analysis using online panels / Web-based social surveys
▫ Government statistics collected through Internet surveys (e.g., Statistics
Korea, formerly the National Statistical Office)
Proportion of Internet Surveys in Surveys of Statistics Korea (%)
Year
Surveys
2005
2007
2008
2009
2010
2011
58.8
52.6
61.3
62.5
63.9
-
1
13
14.9
19.7
Cyber Shopping Survey
27.9
39.7
45.8
49
48.3
Household Income and Expenditure Survey
38.6
42.4
48.8
46
46.8
-
-
16.9
24
40.4
Mining and Manufacturing Survey
Census on Service Industry
Food Grain Consumption Survey
Populatioan and Housing Census
0.9
47.9
3
Reasons for Increase in Internet Surveys
• Increase in Internet use in Korea
▫ The rate of Internet use in Korea: 78.0% (July in 2011, KISA)
▫ 99% use Internet in the age groups 10s through 30s
▫ Age gaps disappearing
• Increasing difficulties for traditional surveys
▫ Increase in the refusal rate of face-to-face surveys, due to
lifestyle diversification and privacy concerns
▫ The usage of landline phone drops, which leads to difficulties in
contacting young people.
4
Internet Surveys
as an Alternative Mode of Data Collection
• Advantages of Internet Surveys
▫ Data collection with less cost and time
▫ Increasing accessibility with certain groups of population
▫ Enhanced monitoring of the ongoing process of data collection
• Sociologists and other academic researchers are reluctant to use
online survey data.
▫ Cf. marketing research
• Sample representativeness
▫ Mostly non-probability sampling
▫ ‘Volunteers’ or ‘convenience’ sampling
5
Issues regarding Representativeness
•
Two points
▫ Decreasing coverage error
▫ Self-selection bias remains and gains more importance
•
Harris Interactive’s prediction on the US 2000 presidential election
▫ Online panelists
▫ Reference survey based on probability sampling towards the same target population
▫ Propensity score adjustment
•
Propensity Score Adjustment
▫ Many studies since Harris Interactive’s prediction (Schonlau et al., 2009; Lee & Valliant, 2009;
Valliant & Dever, 2011)
▫ Studies in Korea (Kim & Lee, 2003; Lee & Jang, 2009; Huh and Cho, 2010)
•
Selection of variables
▫ Details about the selection of variables for adjustment purposes often remain undisclosed.
▫ Unfeasible in other research situations
▫ Insufficient theoretical justification and generalization
6
Social Participation and Intent to Participate
in Internet Social Survey
• “The Societal Trend Toward Self-Administration” (Dillman, 2000)
• Strong belief in relationship between public opinion and democracy
• Civic duty and survey participation (e. g., Couper, Singer & Kulka,
1998)
• Topic interest and survey participation (Groves, Presser & Dipko,
2004)
• The impact of online activities on social/political participation in
Korea
7
Social Participation and
Social Survey Participation
Groves, Singer & Corning (2000)
“Leverage-Salience Theory”
8
• Data
Social Survey 2009,
Statistics Korea
(formerly the National
Statistical Office, ROK)
• Research Q
Who shows
intent to participate in
Internet surveys?
9
INTENT TO
PARTICIPATE IN
INTERNET SURVEY
Men + (vs. Women)
Education +
Employed + (vs. else)
Unmarried + (vs. married)
Internet newspaper reading frequency +
Non-political donation +
Social group participation +
Volunteering experience +
10
INTENT TO
PARTICIPATE IN
INTERNET SURVEY
• Those who expressed intent
to participate in internet
survey are relatively:
▫ Young
▫ Well-educated
▫ Often donate non-politically.
▫ Participate in many social
groups.
▫ Volunteer often.
11
Logistic Regression of
Intent to Participate in
Internet Surveys
• First, just with the
demographic
variables.
• Next model includes
Internet newspaper
reading, donation
experience, group
participation, and
volunteering.
12
Logistic Regression of
Intent to Participate in
Internet Surveys
•
Confirms the results of previous
descriptive analyses.
•
LR increases (SS).
•
Those who express intent to
participate in Internet surveys
differ (from those who do NOT)
▫
Demographically
; and also
▫
•
By the degree of social
participation.
ISSUES
▫
REPRESENTATIVENESS
▫
DISCRIMINATION
13
Implications for Issues of Internet
Surveys
• Representativeness
▫ In addition to demographic variables and digital divide,
diverse social activities affect the intent.
• Potential Discrimination
▫ Socially-inactive groups may be undersampled.
▫ This may potentially lead to discrimination against the less
visible groups.