Content area
Full text
The Statistics Online Computational Resource (www.SOCR.ucla.edu) was funded by the NSF to develop, validate and widely disseminate integrated tools for probability and statistics education. The purpose of the SOCR project is to provide a free, web-based and browser-independent suite of tools; to have a well-designed, extensible and open-sources library; to introduce a graphical user interface (GUI) to statistical resources; and to present an integrated framework for course-material, simulation and computation web-resources. An experimental study conducted with UCLA undergraduates as subjects, has show significantly higher performance of the SOCR treatment group versus the control group using traditional teaching methods. Another benefit is that student performance in the treatment group is more homogeneous. The treatment group reported more satisfaction and found the course more interesting to take than the control group (Dinov et al. 2008).
The first SOCR components included Experiments, Distributions and Games (Dinov 2006). Two new components, SOCR Analyses and Modelers, are newly developed. In this article, we will discuss the pedagogical utilization of SOCR Analyses. The authors were motivated to develop SOCR Analyses to provide a free web-based graphical resource for data mining (using either parametric or non-parametric models), residual diagnostics, and computation of power and sample size. One of the goals is to make available a collection of all the commonly used statistics analyses accessible in one package and provide seamless data transfer between SOCR Analyses and all other tools. Currently, for such types of analyses, students and scholars have to mostly rely on software that either requires a license or a steep learning curve, even for users with some statistical background. There are also some web-based tools available that do not provide code base or examples, which limit the options for code modification and broad use for variety of projects. Many other web-based tools scattered in the cyber-space are applicable for merely a single analysis, not integrated as collections and demand more of the user's time and effort on tool searching and interoperability.
According to some popular websites that compile statistics tools and provide evaluations (Pezzullo 2007; Shackman 2008; CAUSEWEB), many free tools tend to have one or more of the disadvantages described below: (1) some tools have only limited functionalities as free versions; (2) some have to be downloaded to...