The use of NRET as a scoring method in this study represented a departure from normal testing routine for the students, although the method might have seemed familiar to them.
Based on the NRET scores, students' knowledge could be classified into full knowledge (4), partial knowledge (3, 2 or 1), absence of knowledge (0), partial misconception (-1 or -2), and full misconception (-3).
The students were trained to use NRET before sitting for a final MC test.
By the end of the training sessions, the subjects were able to follow the NRET test instructions smoothly.
For the first research question, the extent of guessing under the NRET method was assessed based on two procedures.
For the second research question, the NR and NRET scoring taxonomies were analyzed to gauge the ability of the NRET method in detecting partial knowledge and misconceptions.
For the third research question, reliability of the NR scores and NRET scores were compared.
Therefore, guessing was minimal under NRET since the data fit with the 2-parameter IRT models that assume minimal guessing.
For the second research, the results of the NR and NRET scoring taxonomies for the whole test in Table 5 show the ability of the NRET method to detect partial knowledge and misconceptions.
But the NRET scores could be divided into five different categories: FM for full misconception (-3), PM for partial misconception (-2 and -1), NK for no knowledge (0), PK for partial knowledge (1, 2, and 3) and FK for full knowledge (4), which is in line with the suggestion put forward by Ben-Simon et al.
However, the average number of items for full knowledge under NRET was only 23.32.
Similarly, the results in Table 7 show that their knowledge states for wrong answers could be further categorized into full misconception, partial misconception, no knowledge, and partial knowledge under NRET. Thus, NRET could detect partial knowledge and misconceptions of the students.