Continuing the Comparison Between Graphical- and Text-based Programming Instruction

Students enrolled in the second quarter of the three-quarter sequence had the option of enrolling in a pilot section of programming fundamentals taught using primarily LabVIEW instead of the traditional C/C++ approach. Fourteen students enrolled in this section, and all gave their consent to be included in the study. From the over 200 students enrolled in parallel sections taught using C/C++, a matched comparison group was selected, based upon previous programming experience, intended major, and, where possible, gender. The courses were structured such that they addressed the same basic programming constructs, including basic arithmetic, repetition selection, and data structures. Both cohorts learned MATLAB for two weeks following the initial eight weeks of instruction in LabVIEW or C/C++. A brief illustration of how a simple problem can be approached in either LabVIEW or C/C++ is included as an appendix to this paper. To summarize the results of the previous paper, three comparisons of the groups were done. The first focused on the students’ ability to apply the language used in their programming instruction to problem solving. To do this, the final exam for each class contained an identically worded problem, solvable in either LabVIEW or C/C++. The C cohort outperformed the LabVIEW cohort by about 10%. A second area of interest was how well each cohort learned a subsequent language. Since both courses ended with the same MATLAB instruction, the final exams contained two identical MATLAB problems. Both cohorts performed equally well on the MATLAB component of the exam. Finally, a comparison of student epistemological views was carried out. To develop a suitable measurement instrument, a subset of questions from the Maryland Physics Expectations Survey (MPEX) 2 were modified to pertain to programming and administered as a pre- and post-survey. The student responses were compared to those of an expert group or programmers, some of whom were familiar with LabVIEW and some of whom were not. In the pre-test administration, the two cohorts were indistinguishable from each other and clearly different from the experts.