Response-Improving Research on Computers in Science Learning: A Reaction.
Michael E. Hale · Journal of Research in Science Teaching · 1986
The substance of the commentary by Kracjik, Simmons, and Lunetta is a criticism of the over-generalizations by authors of computer-assisted instruction papers at the 1985 annual meeting of the National Association for Research in Science Teaching. KSL identified as the reasons for these over-generalizations the use of less than state of the art software and of poor criterion measures. This reaction will attempt to examine the validity of the criticisms made in their commentary. Many of their comments are accurate and worth heeding. But, there is some concern regarding a few of the comments made. To begin, one must agree with Kracjik, Simmons, and Lunetta about the reporting of relevant information of criterion measures used in a study. In all reports, well organized detailed information can help communicate a researcher’s findings. However, the criticism of Dillashaw and Bell in using the Test of Logical Thinking may be too severe for a pilot study. The analysis of scores on subtests of a general skill test may help identify the thinking skills enhanced by programming instruction alluded to by Kracjik, Simmons, and Lunetta. There is a strong intuitive belief that programming increases logical thinking skills. To study the effect of programming on general logical thinking skills can not be brushed aside. The nonsignificant result in the Dillashaw paper may be attributed to the power of the instruction in Creative Programming (1983) instead of the criterion measure selection or the length of the study. Concerning two other criticisms of criterion measures, Choi and Gennaro’s use of a paper and pencil test and Dillashaw and Bell’s reading ability bias, research suggests (Staver, 1984) that the method of testing (written versus interviews) does not effect the results of reasoning tasks. Kracjik, Simmons, and Lunetta are correct in stating that personal interviews provide rich data sets on individuals. However, if the questions chosen by the researcher are addressed then why must the study collect data for other questions. It can be argued that only a miniscule effort would be necessary to answer another two or more questions, but all the studies have limits.