June 15, 2001 Update: Log Files at High School "C"
The computer lab is quiet. Fifteen high school students are sitting at computers staring at the dragons on the screen. Some speak softly to their neighbors, commenting on what the dragons look like or which genes are responsible for which trait. We observers are pleased - the class is going well, the kids are engaged, BioLogica is teaching Genetics!!
Log files are collected from each computer at the end of the implementation. Each action of the student, each answer to a question, is time-stamped and stored for later transfer to our server. The files are long, repetitious, but full of information waiting to be mined for clues about how the students are reacting to our genetics hypermodel..
The implementation at High School "C" took place last week, from June 4th through 12th. It involved two lower-level biology classes, both taught by the same first -year teacher. Here are our preliminary lab notes:
So far I have only made a fast read-through of the log files, just to see what the kids were doing and how they did it. A more detailed analysis will come later this summer. For now, let me tell you, things are never what they seem.
The general impression I have is that most of the kids did not finish the scripted programs. They would start, work their way through a few of the questions or puzzles, then quit before the end of the period. My impression is that the kids just got frustrated. In the "Horns Dilemma" activity, for instance, they would click on an answer, get a message that it's wrong, try another wrong answer, and then give up. Sometimes they clicked on an allele again and again - sometimes 10 or 12 times! When the dragon didn't change (presumably) they closed down the activity. We are evidently not giving them the right error message - we should be telling them that there is more than one answer.
In several of the activities the students are asked to write very short essays. They are writing one or two lines, no more. Most wrote "I don't know" or left the space blank.
Punnett squares are used again and again in our activities, and for the most part the students did well with them. But if they get something wrong in a Punnett square, the only clue we give is that something is wrong with the answer. They tried various combinations of answers, still were told that they were wrong, and closed the activity. Again, our error messages are not explicit enough.
As various people pointed out at our project meeting on Tuesday, this group of students hadn't received much scaffolding in Genetics, and their previous experience with this kind of computer-based investigation was limited. Clearly, we must take such factors into account before we jump to conclusions. The good news is that we now have a tool for really finding out how the students are working with the program. And this kind of data we are beginning to perceive the importance of the teacher's role in a hypermodel class.