Abstract:

The Rasch model was applied to calibrate the items of Food Chemistry exam in order to assess the performance of the indices of difficulty and discrimination. The test was applied to 216 students who mostly enrolled in the fourth academic year. The first group was consisting of 109 students and the second by 107. Both groups were evaluated with multiple-choice items randomly distributed. The first group had 90 minutes to solve only 25 items of the 49 and the second group had 60 minutes to solve all the items (49). Of all the items in both cases only 4 should be reviewed, because had lower levels of discrimination. By comparing the scores of the two groups, it was observed that there is a high significant difference (α = 0.01). In addition, in the first group a higher proportion of students (49%) resolved the test at a rate of 0.86 items /min (25 items / 29 minutes) compared to the proportion of students in the second group (39%) who met the test at a rate of 1.3 items / min (49 items/43 minutes). This can be attributed to that in the first group the highest proportion of items (51%) were grouped into the easy difficulty with an average of 0.65 ± 0.15 and greater proportion of items (86%) with high discrimination (41% ) to very high (45%) with a mean of 0.73 ± 0.18. Furthermore, the distribution of difficulty levels has a Gaussian curve behavior or normal curve. In the second group the largest proportion of the items (92%) are located on the easy difficulty (47%) to moderately easy (45%) with a mean of 0.77 ± 0.19, with the highest proportion of items (80%) with high discrimination (53%) to moderate (37%) and an average of 0.57 ± 0.25, in this case the distribution of difficulty levels do not follow a normal curve behavior. The analysis allowed to verify that the used items had varying degrees of difficulty and high reliability in presenting high significant difference in levels of discrimination (α = 0.01) which were based on the ability of students tested.

Keywords:

Item Response Theory (IRT), Rasch model, Characteristic Graphical of item (CGI), chemistry.