Results of Behavioral Segmentation Study

Results from the survey instrument of the first part of Project Meta are now available. This post will outline the ABCs of social science research and how Test Pilot fits in, the goals and importance of this preliminary quantitative study, the study design, and the results and implications. This video describes the behavioral segmentation in depth.

When conducting research in social psychology, researchers are often interested in attitudes, behaviors, and cognition.  Attitudes are one’s reaction toward something, usually exhibited in one’s beliefs or feelings. Behaviors are the actual actions that an individual performs, e.g. clicks on a certain part of an interface or browser chrome. Cognition refers to how one thinks and acquires knowledge.

In the current study, I used Test Pilot to measure users’ attitudes via a survey instrument and to log behaviors in the form of browser interaction data. Cognition will be measured in a subsequent qualitative study. The overarching goal of this study was to create a preliminary Firefox user typology based on the secondary research outlined in the previous post. I sought to answer the question of who uses Firefox, in terms of demographics and self-identified expertise levels, and how do they use it, in terms of actual interaction with the browser.

The survey measured four psychological factors (latter three from Johnson & Kupla, 2007): privacy (cookies, companies’ use of browsing info), sociability (meeting new people online, ‘friending’), utility (using the internet primarily for work or primarily for fun), and reciprocity (passive vs. active browsing, need for complexity). In addition, the survey measured satisfaction, engagement, and various advanced demographic data. The behavioral data measured interaction with the browser chrome, such as clicks on a + to open a new tab, etc. The survey data was linked via an anonymous ID to the behavioral data enabling attitudes and beliefs to be linked to behaviors and actions. The following results describe the major findings from 3701 survey responses. I used Python in combination with R to analyze and visualize the data.

Results.

1. Users who enjoy complexity love Firefox; others are significantly less satisfied. The largest difference in satisfaction is between high reciprocity and low reciprocity users.
2.  Privacy-oriented users and users who prefer to browse for “fun” are more satisfied with Firefox than their counterparts, but this difference is less significant than difference in reciprocity.
3. “Low utility” users, who believe the Internet should be used primarily for fun, are the most engaged with Firefox. They tend to use fewer other browsers in combination with Firefox. No other factor showed a significant difference in engagement.
4. There is no difference in satisfaction or engagement between high and low groups in sociability. Sociability is the only one of the four factors which showed no significant differences in either satisfaction or engagement.
5. Country comparison. Users in the USA are more privacy oriented and tend to like complexity more than those in other countries. Satisfaction is higher with Firefox in the US than in other countries, but engagement is almost the same. Sociability and utility is higher in other countries than the USA. See below plot.

6.  Satisfaction by self-ID expertise. Expert users are significantly more satisfied with Firefox than intermediate ones, and intermediate users are much more satisfied than beginner ones. See below plot.

7.  Psychological factors by self-ID expertise. Users become more privacy oriented, more comfortable with complexity, and use the Internet for more social activities as their expertise with Firefox increases. Users view the Internet as used primarily for fun than for work as they go from beginner to expert users. See below plot.

The results of this study will feed into ongoing and future qualitative and quantitative work aimed to understand Firefox users. Future research will look into the link between the survey data and behavioral data, as well as examine the behavioral data in depth. Further research could test if the findings hold on a representative sample of Firefox users, if users in different groups (e.g. high privacy vs. low privacy) exhibit different behaviors, and future qualitative research such as diary studies and fieldwork can provide deep insights into these factors as well as user cognition.