CUSTOMIZATION VS. PERSONALIZATION: THE ROLE OF POWER USAGE & PRIVACY

Student researchers:

Sampada Marathe (graduate student)
Christen Reese (graduate student)

Faculty Supervisor:

Dr. S. Shyam Sundar

This paper was presented at the annual convention of the Association for Education in Journalism and Mass Communication, Washington DC, 2007.

Introduction

This paper attempts to explain why power users prefer content that they customize themselves while lowend users would like it when the system personalizes content for them. It experimentally demonstrates that this interaction between self vs. system tailoring and power-usage occurs only under conditions of low privacy. It also explores other theoretical mechanisms via mediators such as self-efficacy, sense of control, convenience, and utility, with implications for theory and design of tailoring sites and systems.

hypothesEs

H1: A three-way interaction will occur between tailoring (personalization, customization), privacy (low, high), and power usage on site attitudes such that the transverse pattern of personalization vs. customization effects for power usage (customization higher than personalization at high levels of power usage while personalization higher than customization at low levels of power usage) is true only under conditions of low privacy, but not true under conditions of high privacy.

H2: Perception of utility of content will mediate the relationship between tailoring (personalization, customization) and site attitudes, especially for those high on power usage.

H3: Perception of control will mediate the relationship between tailoring (personalization, customization) and site attitudes, especially for those high on power usage.

H4: Self-efficacy will mediate the relationship between tailoring (personalization, customization) and site attitudes, especially for those low on power usage.

H5: Perception of convenience will mediate the relationship between tailoring (personalization, customization) and site attitudes, especially for those low on power usage.

Key mediators for Power Users

Key Mediators for Non-power Users

Method

Our study design is a 2 (Customization, Personalization) X 2 (high privacy, low privacy) fully-crossed factorial experiment in which participants (whose level of power-usage as the measured independent variable) attended two sessions and were exposed to Google News each time. For the first session, all participants were exposed to the as-is version of the news Website. For the second session, they were either exposed to a personalized version of the Website (created by the researchers based on browsing history from the first session) or a customizable version (wherein participants could tailor content themselves). They were then asked to complete an online questionnaire measuring their attitudes toward the news content on the site.

Results

Power users favored customization under conditions of low privacy and personalization under conditions of high privacy. This lends support strong support to H1. The overall regression was not significant. However the mediating analyses showed some interesting results.

For low privacy, ease of use increased with increasing power usage in the customization condition and decreased with increasing power usage in the personalization condition. For high privacy, ease of use increased with power usage, but did not differ in the personalization and customization conditions.

For the variable computer self-efficacy, a significant main effect for power usage was observed, F(1, 62)=19.76, p<.001, such that as power usage increased, computer self-efficacy increased.

For low privacy, perceived control increased with increase in power usage in the customization condition, but decreased with increase in power usage in the personalization condition. For high privacy however, perceived control increased with increase in power usage, but did not differ for the two tailoring conditions.

For the utility variable, the main effect for power usage approached significance, F(1, 62)=2.81, p=0.09, such that as power usage increased perceived utility of the Website increased. A two way interaction between privacy and tailoring also approached significance, F(1, 62)=2.00, p=0.08, such that users felt the same level of perceived utility in both customization and personalization in the low privacy condition. However, in the high privacy condition, perceived utility was greater in the customization condition than the personalization condition.

Conclusions

Privacy turns out to be a key predictor of user attitudes toward personalization and customization, with clear implications for site and system design. Providing high privacy has tangible psychological benefits, by imbuing users with a greater sense of control and stronger attitudes toward personalization for power users and stronger attitudes toward customization for non-power users. Low privacy, which unfortunately is the default assumption, appears to be a key psychological concern for users, forcing power users to take personal control of the system by engaging in effortful customization and discouraging non-power users from effectively using the system.

For more details regarding the study contact,

Dr. S. Shyam Sundar by e-mail at sss12@psu.edu or by telephone at (814) 865 2173

Click here to get back to the abstracts listing page.

Click here to get back to the main research page.

Media Effects Research Lab at College of Communications, Penn State University