After hand coding the data logs of 6 students from our intensive field test in MV, I developed specifications for batch processing the mass of data logs collected in the various implementations. Ed Burke and I negotiated what was possible given the state of the data logs, and Ed wrote the necessary program to process them. The table below presents the results of statistical analysis of the processed logs. I've organized them by implementation because both the activities and the extent of data logging varied.

The implementations (MR, MV, L, and C) are in chronological order. As you scan the table you get an idea of when activities were added and when data logging for particular activities was implemented. You can also tell which activities do not yet include essay questions: Meiosis, Monohybrid, Dihybrid, Sex-Linkage, Scales, Plates, and Invisible Dragons.

For each implementation you'll find the number of logs generated and the number of valid logs. Valid logs are those logs that have a date and a total time greater than 0. Invalid logs arise from bugs in the program or aborted launches.

For each activity, you'll find the number of data logs analyzed to produce the numbers that follow.

• Total time (median) is not the average time that students used an activity, but the amount of time in the middle of the range; that is, half the students used the activity for that amount of time or more and half used it for that amount of time or less.
• Index of Interaction is calculated by dividing the length of time a student used the activity by the number of interactions (mouse clicks and other selections). Again the median value is used.
• # Questions is the median number of questions that students encountered when using an activity.
• Words/answer is the number of words entered into the answer divided by the number of questions encountered. Again the median value is used.

This table is an intermediate research result. It is intended as a display of data that will provoke questions. We invite you peruse the table and generate your own questions.

What the production of this table demonstrates is that we can automatically process at least some of the data from the data logs which in turn can be incorporated into a report for teachers and students. Because students logged in under a variety of names, we cannot at this point correlate specific logs with specific students. When we have implemented a structured log in procedure, we will be able to generate a class roster for the teacher that includes which activities students have completed, how long they took and a rough indicator of engagement; current candidates are Index of Interaction and Words per Answer. Student reports will include the questions encountered and answers entered.