peng tai-quan and zhu jonathan---don\'t blame the internet anymore!
TRANSCRIPT
Tai-Quan Peng, Jonathan J. H. ZhuDepartment of Media and Communication
City University of Hong Kong
Internet VS. Use of Traditional Media◦ Displacing View (e.g.,
Kaye, 1998; Nie & Erbring, 2000)
◦ Complementary View (e.g., Stempel et al., 2000; Nguyen & Western, 2006)
Internet VS. Sociability◦ Pessimistic View (e.g.,
Kraut et al., 1998; Nie & Erbring, 2000)
◦ Optimistic View (e.g., Kraut et al., 2002; Lee & Zhu, 2002; Cole & Robinson, 2002) Criticisms(Nie & Hillygus, 2002):
Inadequate measurement of Internet use Limitation to bivariate analysis Causal Relationship (Nie, 2001)
Simple, but NOT trivial!◦ Uni-item to Multi-item
Sophistication of Internet Usage (SIU)◦ To quantify but NOT to
evaluate system usage (Burton-Jones and Straub, 2006)
◦ Exploratory to Confirmatory Approach
◦ Dimensionality & Construct Validity (Peng & Zhu, 2008)
Dimensions Examples
Time/Duration/Frequency
Amount of online time, in Zhu & He (2002a)
Frequency and Duration, in Compeau et al. (1999)
Year of experience, in Gravill et al. (2006)
Activities
Activity scope, in Jung et al. (2001)
Diversity of online time, in Zhu and He (2002a)
Actions, in PEW (2007)
Experience, in Gravill et al. (2006)
Skills
Skill of use, in Thompson et al. (1994)
Cognitive experience, in Rogers (2003)
ContextSite Scope, in Jung et al. (2001)
Assets, in PEW (2007)
Attitude/AffectiveEvaluation of Internet Influence, in Jung et al. (2001)
Attitudes, in PEW (2007)Source: Peng & Zhu (2008)
RQ1: Is there any difference between Internet non-user’s sociability and Internet user’s sociability?
RQ2: Is there any difference between Internet non-user’s use of traditional media and Internet user’s use of traditional media?
RQ3: With the new measurement of Internet use, SIU, what is the influence of Internet use on users’ sociability when his/her use of traditional media is controlled?
RQ4: With SIU, what is the influence of Internet use on users’ use of traditional media when his/her sociability is controlled?
Data Source: Hong Kong Internet Project (HKIP) (2003-2005)
Measurement◦ Internet Adoption: User VS.
Nonuser◦ Sophistication of Internet
Usage Online Activities – Composite
Score Online Time – Self-report Online Skill – Composite Score Diversity of Online Method –
Composite Score Diversity of Online Place – H-
Statistics ◦ Demographics: Age, Gender,
Education Degree, Family Income
Internet Adoption, Use of Traditional Media, and Sociability◦ Generalized Linear Modeling (GLM)◦ Measurement Equivalence/Invariance (ME/I) Test within
Multi-Group Confirmatory Factor Analysis (MGCFA) ME/I: comparison of the various components of
measurement models, and can be extended to structural models and mean structures (Cheung and Rensvold, 1999)
Application: Peng & Zhu (2005) Internet Use, Use of Traditional Media, and
Sociability◦ Structural Equation Modeling (SEM)
LISREL 8.0 (Joreskog & Sorbom, 1996)◦ Maximum Likelihood Estimation
Overall Test (F ratio for Wilks's λ) TV Radio
Newspaper
Chatting with Family Members
Playing/Exercising with FM
Shopping with FM
Internet Adoption(0 = Nonuser)
4.08 19.75** 2.99 6.67* 10.78* -0.20* 0.04
Age 19.32 -0.19 1.05*** 0.65*** -0.13 0.00 0.00* Gender(0 = Male) 5.53 14.29** -8.40 -7.77*** 3.96 0.04 0.07 Family Income 3.21 -2.72** -2.23 -0.06 0.57 0.03 0.03** Education Level 14.07 -14.50*** -7.83** 3.54*** -1.90 -0.10** -0.07** Survey Year 1.78 3.36 -2.58 -0.04 -3.87 0.07 0.07* R2 0.07 0.033 0.065 0.003 0.005 0.009*p < .05; **p < .01; ***p < .001
Internet Nonuser
Internet User
Δχ2 (Δdf = 1)
Latent Means
Equivalence Test
αSOC 0 -0.07 0.02αUTM 0 -49.77*** 119.5***
Factor-covariance
Equivalence Test
rUTM,SOC -0.02 0.15* 2.89†
Four Measurement Equivalence (ME) Tests: 1. Configural Equivalence: we are not comparing apple with orange2. Metric Equivalence: measurement scale of UTM and SOC are
invariant between user and nonuser3. Latent Means Equivalence: difference of UTM and SOC between
users and nonusers4. Factor-covariance Equivalence: Correlation between UTM and
SOC
On average Internet users significantly spend less time on traditional media than Internet nonusers;
Internet users spend the same amount of time on sociability as Internet nonusers;
Internet adoption moderates the relationship between use of traditional media and sociability.
Explanatory Variables
Sophistication of Internet Usage
Use of Traditional Media
Sociability
Age -0.41*** 0.37*** 0.07
Gender -0.25*** 0.005 0.10**
Education 0.47*** -0.22*** -0.21***
Family Income 0.07*** -0.20*** 0.14***
SIU N.A. -0.13 -0.05
Use of Traditional Media Sociability
Sophistication of Internet Usage
UTM1.00
SOC0.22* 1.00
SIU-0.22** -0.17*** 1.00
*p < .05; **p < .01; ***p < .001
Parameters Short adoption history group
Long adoption history group
Δχ2
(Δdf = 1)
β UTM,SIU 0.09 = -2.41 1.21β SOC,SIU -0.47 = 0.27 0.05α UMT 0 = 1.48 0.38α SOC 0 > -6.50 3.17†
α SIU 0 < 1.01***62.75***
†p < .10; *p < .05; **p < .01; ***p < .001
Strength-Level
Comparison
Mean-LevelComparison
Internet use will NOT influence users’ sociability and use of traditional media.
‘Strong effects based on bivariate analysis are almost always destined to evaporate after rigorous controls are exercised’ (Zhu & He, 2002a)
Internet users’ overall time budgets for media use may be larger than those of nonusers (Nguyen & Western, 2006), so that, even if they have to reallocate their traditional media budgets to compensate for Internet use, there must be a bottom line in their reduction.
Internet users’ use of traditional media and sociability will not be influenced by the sophistication of Internet usage, regardless of their adoption histories.
Nil-influence of the Internet? No! Zoom-in examination of delicate
mechanisms over long periods of time (Zhu & Wang, 2005)◦Time-based Measurement of UTM and Sociability◦Panel Study VS. Cross-sectional Study◦Zoom-in examination of specific Internet
activities (e.g., Web 2.0 technologies)