Web Log Analysis: A Study of Instructor Evaluations Done Online

Article excerpt

Introduction

The rapid development of Internet-based technology and an increased desire to reduce the cost and time required to administer and record students' evaluations of faculty members has prompted some schools to move towards the use of online instructor evaluations. The ability to authenticate a user and to securely transmit and store data has helped make this possible. Additional benefits (beyond cost savings) include the ability to effectively administer evaluations in a wide range of learning contexts and locations (e.g., distance learning and other online learning environments) and the ease of compiling summary data by instructor, by course, by department, by question, or in any other way desired.

Initial experimentation has been done at the College of Business and Economics (COBAE) at California State University, Northridge (CSUN). As with any change that can have major implications for many people (including promotion and tenure decisions), there are a number of questions, concerns, and related issues with which to deal. This research focuses on some of the technical aspects of implementation, but other concerns are also discussed below.

In addition to the interest in instructor evaluations, this research also considers implications for analysis of web-logs in general. Use of the World Wide Web is becoming an important addition to traditional business and organizational service encounters (e.g., Haksever, Render, Russell, & Murdick, 2000; Turban, Lee, King, Warkentin, & Chung, 2000). Accordingly, a number of strategic advantages for improving service through technology are possible (Berry, 1996). While the sage advice "you can't manage what you can't measure" (Drucker, 1993) is still very much true, in an e-Business environment it is also true that "you can't measure what you can't monitor". Inspection, productivity, or quality control techniques in traditional service environments may be inadequate, inappropriate, or simply impossible in an electronic service environment. Learning about precise user behavior on web sites is of significant interest to both academics (Spiliopoulou, 2000; Joshi, Joshi, Yesha, & Krishnapuram, 1999) and practitioners (Davenport, 1999; Kimball, 2000).

Despite the abundance of available data that are related to the use of web sites, few organizations have yet developed the knowledge and skills necessary to study these data and transform them into information for the purposes of improving operations. This is partly because the Internet environment is relatively complex and data about transactions and events are generated so quickly; each time a user accesses ("hits") a web page, a web-log entry is generated. Since typical users jump from page to page quite quickly, changes happen quickly and a plethora of information is generated. Thus, "web-log analysis to improve web page content and design is not an easy task" (Drott, 1998, p. 50). Recording web hits on even a relatively small web server can result in log files with hundreds of thousands of lines of data--or more.

Because individual customers cannot be physically observed on a web site, studying user behavior on a web site is different than studying customers in a traditional service. On the Internet, customers move quickly, and many customers can be on a site at any given point in time, making it impossible to observe individual behavior in real time. However, it is possible to study the weblogs that were generated by the users. Web servers can be set up to maintain access logs that are recorded passively (i.e., without user disruption or intervention). It may be tempting to treat these logs as accesses originating from "machines" (i.e., computers), utilizing concepts from manufacturing operations. However, this approach is less than ideal since there are many personal characteristics involved in the way individuals use websites.

The COBAE instructor evaluation site captures and records individual student responses regarding course mission, curricular content and faculty evaluation. …