An investigation of fair information practices on American websites

This research used snowball sampling with a first line of six hundred online adult Americans viewing eight websites---having various fair information practices (FIP)---and completing an online survey. The data were used to evaluate a three-stage model relating published privacy policies on the websites (NOTICE), what personally identifiable information (PII) is collected (SENSITIVITY), how it is shared (SHARING), and Internet experience (EXPERIENCE) to perceived information risk (INFORISK); INFORISK to consumer trust (TRUST); and TRUST to consumer intention to purchase (INTENT---a proxy for Internet demand). The model was developed to integrate FIP into previous researchers' models of consumer trust and intention to purchase.;Of 505 responses, 314 subjects thirty or older or college graduates were retained for the analysis. The average age category was 40-49 years. Privacy pragmatists showed more INFORISK sensitivity to implementation of FIP than is generally reported, indicating better practices might in fact garner their business.;The research developed reliable (Cronbach alphas of 0.7 or better) multi-item instruments for fifteen constructs. These included FIP elements, risk perceptions, trust, website and Internet comfort, and intention to purchase.;Effects of NOTICE, EXPERIENCE, SHARING, and SENSITIVITY on INFORISK were confirmed at the 99% confidence level using multiple linear regression. The effect of INFORISK on TRUST and the effect of TRUST on IN TENT were confirmed at the 99% confidence level using simple linear regression.;The research indicates INFORISK increases as SENSITIVITY and SHARING increase. TRUST and INTENT correspondingly decrease, so consumers who think the collected data are more sensitive and the sharing policy is too liberal probably purchase less.;The multi-item instruments for NOTICE (locating, reading and understanding the privacy policy) and EXPERIENCE (general confidence and comfort with the Internet and with purchasing on websites) do not correlate with INFORISK. The Flesch Reading Ease score of each website's privacy policy does, however, as does each subject's personal Internet purchase estimate. Quite apart from their feelings, the subjects' INFORISK moves inversely with single, relatively quantitative measures of NOTICE and EXPERIENCE.;In summary, the research suggests buyers (including privacy pragmatists) respond ultimately with greater purchase intentions to good implementations of FIP.