ActivStats is a complete introductory statistics textbook on a CD. It is the most exciting development in statistics education I have ever experienced.

ActivStats has all the features of a traditional statistics textbook: coverage of statistical concepts with text, graphics, and exercises (plus a table of contents, index, and glossary). What's different is that ActivStats provides powerful features that are only available on a computer: interactive lessons, videos, and (amazingly) a built-in data analysis package (the award-winning Data Desk).

ActivStats emphasizes data visualization and data analysis. Less emphasis is placed on the mathematical mechanics of statistics, which can now be mostly handled by the computer, as in real data analysis. The sequence of topics is as follows: data and variables, univariate distributions, relationships between variables (correlation, regression, surveys, experiments), randomness, probability, random variables, sampling distributions, and statistical inference.

(The sequence of topics in ActivStats exemplifies the currently most popular sequence for the introductory course. I recommend a substantially different emphasis and sequence of topics for the introductory course [1996], but I shall not discuss that here.)

ActivStats consists of twenty-four "lessons", each of which is between two and six "pages" long. Each page in a lesson appears on the computer screen as a scrollable block of text and graphics. Each page gives a high-level summary of a sub-topic of the lesson topic.

Most pages contain between one and five icons, each representing an "activity". When a student clicks on an icon, it expands on the screen into a window containing a presentation. Activities come in four flavors:

1. Narrated Videos (average length around two minutes). These describe real-world problems that motivate the statistical topic under discussion on the page.
2. Narrated Mini-Lectures (average length around five minutes). These are on a single statistical topic and are carefully illustrated with innovative moving graphics. Many are punctuated with "hands-on" sessions in which students interact with the computer to gain experience with the topic.
3. Data Analysis Exercises (average length around eight minutes). These guide students through the use of Data Desk to display or analyze data to reinforce the statistical topic under discussion.
4. Concept-Review Exercises (average length under two minutes). These contain a set of sentences describing the main ideas presented earlier on the page. Students fill blanks in the sentences with supplied words, and the computer indicates the correctness of the answers. These help students to assess their understanding of the ideas.

Each lesson is accompanied by a set of homework questions (roughly nine per lesson), many of which are intended to be answered using Data Desk (with supplied data). Each lesson also has a set of projects (roughly two per lesson), which are similar to the homework questions except that students must actually collect data as part of each project.

Each ActivStats lesson has cross-references to eleven currently popular introductory statistics textbooks. Thus a teacher can teach a course with ActivStats in conjunction with a standard introductory text. One of these texts is by David Moore, a leading statistics educator and 1998 president of the American Statistical Association. His book, The Active Practice of Statistics (1997), is specifically tailored to be a companion textbook to ActivStats and employs the same philosophy and contains the same twenty-four lesson topics as ActivStats.

Of special note are the innovative "visualization tools" that accompany many of the mini-lectures. Each tool is a graphical display of data (e.g., a bar chart, scatterplot, or density plot) together with one or more ways the student can easily change properties of the data or display. (Many tools have "sliders" that allow the student to change the value of a property through a continuous range of values.) When a student changes the value of a property, the display immediately and (if necessary) continuously updates to reflect the new value(s) of the property. These fascinating tools are revealing windows for students into both data and statistical concepts. A controllable moving picture is worth ten thousand words.

Also of note is the tight integration between ActivStats and Data Desk. If a student clicks on an activity or exercise that uses Data Desk, the Data Desk program is automatically launched and loaded with (a) the step-by-step instructions for the student to follow to do the work and (b) the dataset to be analyzed.

ActivStats contains several other nice touches, including careful discussion of many stumbling blocks of statistics and superb graphics. These features all reflect the great care and tremendous effort Paul Velleman and his development team have put into this remarkable computer-based statistics "textbook".

ActivStats is the most notable of several new products that reflect a major turning point in statistics education: Students now have the opportunity to actively learn statistics, using their media of most frequent choice -- computers and television. Students who actively participate in their learning learn substantially more than students who passively attend lectures or engage in passive reading of a textbook (Garfield 1995).

Velleman has produced a first-rate computer-based statistics "textbook" -- a significant contribution. However, he has also made a much more important contribution: he has given us a rich new paradigm for teaching statistics.

An important feature of the paradigm is that it is extendible -- teachers can add their own material to ActivStats. If Velleman provides good standards, openness, and easy-to-use procedures, it is likely that many statistics teachers will develop activities that run in the ActivStats environment for teaching many statistical topics, possibly at all levels. This will enable a sharing and natural selection of activities by statistics teachers, which will help us to zero in on optimal approaches to teaching statistical concepts.

Because products like ActivStats deliver substantially more functionality than standard (paper-based) textbooks, it is clear that these products will, in time, completely replace standard textbooks. However, the following problems must be solved before computer-based textbooks will be the norm:

1. The low "pixel count" of current computer screens detracts from computer-based statistics textbooks. A computer screen is a dense matrix of tiny dots called "pixels" (picture elements). The computer generates images on the screen by controlling the color and brightness of each pixel. A typical screen may be 800 pixels wide and 600 pixels high, giving roughly a half-million pixels on the screen. The printed page in a typical standard textbook (ignoring margins) is equivalent to 5400 x 9300 pixels, giving roughly fifty million pixels on the page -- a one-hundred-fold increase over a typical computer screen. The pixel count of the printed page of current textbooks evolved over the centuries as optimal for displaying "printed" information. Thus, as many readers will know from experience, the printed page is substantially more effective than the computer screen for displaying (static) information. (ActivStats makes the best of this unfortunate situation by generally using only small blocks of static text, and by using motion to help focus attention.)
2. Some students may be tempted to "surf" computer-based textbooks, jumping somewhat randomly from topic to topic, like surfing on the web. This approach leads to minimal learning since the later topics in a statistics course invariably require knowledge of earlier topics. Similarly, there is a sense among many students that when using a computer one must move quickly, without pausing to reflect. But statistics students must learn to pause and reflect on what they are learning, and sometimes even backtrack for review. Thus the challenge in designing a computer-based textbook is to keep students focused on the appropriate material until it is properly learned. ActivStats uses several techniques to help students stay focused, but more techniques may be needed.
3. Computer-based textbooks can only be used in courses if students have adequate access to computers. For ActivStats, I suspect the average college or university student will need five or more hours of readily available access in some weeks.

As a practicing statistician, I am attracted to careful empirical research. Therefore, I look forward to research by Velleman and others as they compare courses using ActivStats with other introductory statistics courses. I believe this research will demonstrate unequivocally that carefully designed "multimedia" teaching (supplemented with some lecturing) is superior to other approaches in two critical areas: (1) careful multimedia teaching gives students substantially better understanding of the use of statistics in empirical (scientific) research, and (2) careful multimedia teaching gives students substantially more respect for our field.

With this last thought, I heartily recommend that all introductory statistics teachers evaluate this extraordinary teaching tool with a view to using it in their courses.

Garfield, J. (1995), "How Students Learn Statistics," International Statistical Review, 63, 25-34.

Macnaughton, D. B. (1996) "The Introductory Statistics Course: A New Approach." Available at http://www.matstat.com/teach/

Moore, D. S. (1997), The Active Practice of Statistics, New York: Freeman.

ActivStats runs on Windows and Macintosh computers. Minimum requirements under Windows are Windows 95, 486/66, 12 Mb RAM, 640 x 480 x 256 color monitor, double-speed CD-ROM drive, and an 8- bit sound card that is QuickTime compatible. Minimum requirements on the Mac are System 7.5, 68040/25 or Power Macintosh, 8 Mb free RAM, 640 x 480 x 256 color monitor, and a double-speed CD-ROM drive.

This review originally appeared in the Winter (March) 1998 issue of The Statistics Teacher Network, a newsletter published three times a year by the American Statistical Association / National Council of Teachers of Mathematics Joint Committee on the Curriculum in Statistics and Probability. For a free subscription send a request with your name and snail-mail address to ASAINFO@amstat.org