Understanding third perception about Internet privacy risks

Published: 19 January 2019| Version 1 | DOI: 10.17632/rkfk5bz4st.1
Hongliang Chen


The current study used a sample of the U.S Mturk respondents in November 2016. Mturk self-selected respondents to participate the survey. A respondent who completed the questionnaire received $3. Such compensation is common in survey research as a means to enhance response rates (Largent, Grady, Miller, & Wertheimer, 2012). The sample size of the current study was 613. According to prior statistical research, if continuous variables play primary roles in data analysis and the alpha level is set a priori at .05, level of acceptable error at 3%, and standard deviation of scale is 1.167 (7/6), the minimum recommended sample size is 118 using Cochran’s sample size formula (Kotrlik & Higgins, 2001). Survey scientists often increase the minimum sample size by 40% to 50% as compensation for the incomplete submissions (Fink, 1995; Salkind, 1997). In the current study, the size sample of 613 should provide satisfactory statistical power.


Steps to reproduce

Using STATA 13.0 & SPSS 22.0, t-test was conducted to compare the means of the perceived Internet privacy risks on general others and self (H1). To test the effect of social distance on TPE perceptions, the TPE perceptions of seven categories of “others” were tested using repeated measures of ANOVA (H4). Consistent with prior studies, TPE perception was computed by perceived Internet privacy risks on general others subtracting the perceived Internet privacy risks on self (Blinded-In-Press-Citation; Gunther & Hwa, 1996). Then, a multi-stage structural equation model (SEM) was used to test the proposed model. The model consists of three stages: 1) four antecedents of TPE perceptions (excluding social distance), 2) TPE perceptions, and 3) intention of behavioral responses. Control variables were added into each cross-stage path. Chi-square, comparative fit index (CFI) and root mean square error of approximation (RMSEA) were used to measure the overall model fit.


Zhejiang University