The Carnegie Mellon Genetics Cognitive Tutor
       
Home research people contact
   

Development
The development of a Genetics Cognitive Tutor module involves these activities:

• Cognitive Task Analysis. Development begins with a task analysis of the learning goals
in problem solving, the problems posed to students, the knowledge required to solve the problems, and the way(s) students think about solving problems.

• Design and Programming of the Problem Solving Interface. The Cognitive Task Analysis guides the design of the problem solving interface. The goal is to design an authentic environment that provides tools to support the various solution paths students pursue and to scaffold student learning of the challenging task components that are identified in the cognitive task analysis.

• Development of the Cognitive Model. Problem-solving knowledge and strategies are represented in the form of a cognitive model. The components of this model are cognitive problem-solving rules that associate problem states with appropriate actions. The model enables the tutor to follow each student’s solution path, interpreting each problem-solving action.

• Design and Coding of Tutor Assistance. Multi-level advice messages are associated with each of the rules in the cognitive model. These help messages context-specific advice that engage the student in thinking about how the current problem state is related to the goal structure of the problem and help the student encode the relevant features of the problem state, draw on relevant genetics knowledge, and identify an appropriate problem solving action.

• Classroom Piloting and Evaluation
We have conducted focused evaluations of the individual modules in a diverse set of 12 post-secondary institutions, including public and private institutions, liberal arts institutions and research universities, and minority-serving universities. The tutor has been piloted in a variety courses across these institutions, ranging from introductory biology to upper level genetics courses.

Across all these contexts, we have conducted and analyzed 27 pretest-posttest evaluations of the available modules. Each of these single-unit evaluations consisted of (1) a proctored paper-and-pencil pretest (about 20 minutes); (2) completion of the Genetics Tutor problem solving module (roughly one hour); and (3) a proctored paper-and-pencil posttest (about 20 minutes). The paper-and-pencil tests consisted of problems similar to the Cognitive Tutor problems and two paper test forms were constructed for each module. The tests were designed to be very challenging for students, in order to avoid ceiling effects in measuring learning gains. Each form serves as the pretest for half the students, who then switch to the other form for the posttest, so that the pretests and posttests are matched across the full set of students, but for each student the pretest and posttest are different. A total of 801 students participated in these evaluations.

Overall, the Cognitive Tutor modules proved very effective.

Across the thirty-six evaluations, students averaged 39% correct on the pretest and 61% correct on the posttest and this average gain of 22 percentage points is the equivalent of 2 letter grades

Topic % of Evaluations Gain % Points
• Gene Interaction & Epistasis 5 evaluations 31.0
• Pedigrees - Basic Analysis 7 evaluations 9.1
• Pedigrees - Carrier Probabilities 4 evaluations 20.7
• Three Factor Cross 7 evaluations 21.6
• Population Genetics - Equilibrium 1 evaluations 28.4
• Gene Regulation 3 evaluations 4.3
• OVERALL 27 evaluations 22.1
 

For more information, see: Corbett, A.T., Kauffman, L., MacLaren, B., Wagner, A., and Jones, E. (2010). A Cognitive Tutor for genetics problem solving: Learning gains and student modeling. Journal of Educational Computing Research, 42, 219-239. Read the paper here.

 
 
 
hhmi nsf
fipse
ies
 
 
©2013 Genetics Cognitive Tutor Project, Carnegie Mellon University