Comparing the force and motion conceptual evaluation and the force concept inventory

Abstract

In this paper we compare and contrast student’s pretest/post-test performance on the Halloun-Hestenes force concept inventory (FCI) to the Thornton-Sokoloff force and motion conceptual evaluation (FMCE). Both tests are multiple-choice assessment instruments whose results are used to characterize how well a first term, introductory physics course promotes conceptual understanding. However, the two exams have slightly different content domains, as well as different representational formats; hence, one exam or the other might better fit the interests of a given instructor or researcher. To begin the comparison, we outline how to determine a single-number score for the FMCE and present ranges of normalized gains on this exam. We then compare scores on the FCI and the FMCE for approximately 2000 students enrolled in the Studio Physics course at Rensselaer Polytechnic Institute over a period of eight years (1998–2006) that encompassed significant evolution of the course and many different instructors. We found that the mean score on the FCI is significantly higher than the mean score on the FMCE, however there is a very strong relationship between scores on the two exams. The slope of a best fit line drawn through FCI versus FMCE data is approximately 0.54, and the correlation coefficient is approximately r=0.78 , for preinstructional and postinstructional testings combined. In spite of this strong relationship, the assessments measure different normalized gains under identical circumstances. Additionally, students who scored well on one exam did not necessarily score well on the other. We use this discrepancy to uncover some subtle, but important, differences between the exams. We also present ranges of normalized gains for the FMCE in a variety of instructional settings

Validating two questions in the Force Concept Inventory with subquestions

“…However, we expect that our results can be extrapolated because they are consistent with the results of the interviews with students at Tokyo Gakugei University, Nagoya University, and Meijo University. Further, Thornton et al and Scott et al found similar results for Q.16 [13,14].…”

“…Similar shortcomings have been highlighted by other studies. For example, Thornton and colleagues used the interview method to highlight the inadequacy of Q.37 in force and motion concept evaluation [13], and Scott and colleagues performed a factor analysis on an FCI data set and hypothesized why students who answer Q.16 correctly treat it as a question pertaining to Newton's first law rather than to his third law [14].…”

The puzzling reliability of the Force Concept Inventory

“…For instance, in think-aloud studentinterviews, Thornton et al showed that some students chose a correct FCI response using incorrect reasoning. 10 In a previous study using latent Markov chain modeling, 18 we found statistical evidence of the same false-positive (FCI question 16) reported in the qualitative analysis in Ref. 10.…”

“…A reasonably high correlation of r ¼ 0.78 was found, indicating that the score on one instrument is a good predictor of the score on the other. 10 Internal consistency reliability is measured by correlating the total scores on two distinct halves of a test to establish whether the sub-parts are consistent with each other. Internal consistency reliability assumes that a test measures a unique construct across all test items.…”

Gender fairness within the Force Concept Inventory

“…Henderson [44] also examined test-retest reliability between the FCI as a graded posttest and as an ungraded quiz given the following semester; excellent test-retest reliability was measured in a sample of 500 university students. The FCI has also been compared with an alternate test of conceptual knowledge of mechanics, the FMCE [14]; a high correlation of overall test scores, r ¼ 0.78, was demonstrated [51].…”