Using programming protocols to investigate the effects of manipulative computer models on student learning

This study addressed two important instructional issues regarding programming education: what novices must learn in order to achieve programming success and how learning experiences should be organized to help novices acquire programming knowledge. The primary goals of this investigation were (1) to document programming behavior in an attempt to learn more about the novice\u27s preconceptions and intuitions about programming and (2) to evaluate the effects of a manipulative computer-based model on student learning;The study was conducted using a posttest quasi-experimental design. Students assigned to the treatment group completed a series of programming-like computer tasks using a manipulative model of computer memory operations while students in the control group completed a placebo computer activity. Two posttests were administered following instruction and programming activities on Pascal programming concepts. Students were allowed to enter three of their posttest solutions into the computer and run and modify them. On-line histories of these programming sessions were maintained;The primary evaluation technique used in the study was to document the characteristic features of each student\u27s initial and final solution efforts as well as any intermediate programming problems. Analyses that compared the number of students in the treatment group exhibiting a selected feature to the number of students in the control group exhibiting the same feature were then performed. Mean posttest scores of the two experimental groups were also compared for the second posttest;Due to the exploratory nature of the study and the small sample size employed, the findings were considered tentative. Conclusions of the study concerning the preconceptions and intuitions of novice programmers were that novices (1) possess inaccurate models of computer operations and (2) appear to expect information to be processed by computers as it is by humans. Conclusions about the effects of the manipulative model were that (1) the model did promote learning and (2) this learning was not measurable by using standard scoring procedures