SOFTWARE TOOLS FOR INTRODUCTORY STATISTICS: A STUDENT-BASED COMPARISON

We investigate experiences of first-year, non-science students with three data analysis tools during a pre-Calculus, introductory statistics course at the American University in Cairo, Egypt. Students could choose between DataDesk, Excel or StatCrunch, and were required to use one of these packages to analyze data collected for a semester project. It was especially important to evaluate these software packages from the point of view of a first-time user with no previous experience in either Statistics or advanced computer usage; several students were at first somewhat apprehensive of using a computer to analyze data. Among other outcomes, this investigation led to the development of student-based comparison of three software packages, from the perspective of a large group of potential users. BACKGROUND The benefits of including projects in introductory Statistics courses have been well documented. For example, Fillebrown (1994) states that a project facilitates statistical reasoning through open-ended questions, as well allowing students to work with real data and put into practice theory learned in the classroom. Chance (1997) notes that doing a semester project increases understanding of the statistical process, improves communication skills, and promotes student interest in Statistics. Smith (1998) modified the traditional course format to consist primarily of a series of projects undertaken throughout the semester under the theme ‘Learning Statistics by Doing Statistics’. Smith found that student test scores improved ‘dramatically’, and that students were overwhelmingly positive towards the use of projects. Thorme and Root (2002) found that community-based research projects enable students to absorb material by constructing their own meaning for what they learn, by practicing and sharing classroom theory, and by increasing motivation through social relevance. McGillivray (2005) used own-choice projects and concluded that these were highly beneficial, by giving students a sense of ownership of their collected data, and by increasing relevance and motivation. Similarly, Kurji (2002) found a great deal of success implementing student-designed experiments as an important part of the introductory statistics course, with emphasis on understanding and interpreting the results of computer output, rather than on mechanical computations that have traditionally been the focus of this course. Thus, it seems that projects of varying designs, intents and lengths play an integral part of the learning process particularly among students specializing in non-scientific fields. For the course in Statistical Reasoning at the American University in Cairo, a first-year, pre-calculus courses taken by non-science majors, it was decided to require a relatively short, multi-section project as part of students’ overall assessment. This project consisted of two parts: Section A dealt with two categorical variables in which the aim was to determine independence through graphical examination of the conditional distributions, while Section B consisted of a simple linear regression and interpretation of results. Students had to collect their own data through a small but random survey, enter the data into a statistical computing package, make the graphs and most importantly, interpret this output. Goals for this project included gaining practical knowledge of surveys and the need for randomized selection, solidifying the concepts of conditional distributions and independence, enhancing understanding of correlation, regression and the role of residuals, developing computer literacy specifically in the realm of statistical analysis including graphical output, and enhancing writing skills including the presentation of information in a logical way. Of particular interest was the use of a computer to analyze data; quite a few students seemed rather apprehensive at the beginning of the semester about learning to use a new program. To help overcome this temporary obstacle, the instructor gave students a choice between three software tools: StatCrunch, Excel and DataDesk. StatCrunch was included in the list of choices because it won the MERLOT Classics Awards in 2005 (MERLOT is an acronym for Multimedia Educational Resource Learning and Online Teaching, more information