Academia.eduAcademia.edu

Figure 10 – uploaded by Gladys Castillo

See full PDFdownloadDownload figure
Figure 11: Bland-Altman plots for inter-agreement among expert average, BN test and BN exam

Figure 11 Bland-Altman plots for inter-agreement among expert average, BN test and BN exam

Related Figures (15)

Figure 1 illustrates the structure of the GBSM. Figure 1: Structure of the Bayesian Student model
Figure 1 illustrates the structure of the GBSM. Figure 1: Structure of the Bayesian Student model
Figure 2: Using G(x) to compute the probabilities 3. Bringing the model into practice in PMatE
Figure 2: Using G(x) to compute the probabilities 3. Bringing the model into practice in PMatE
Table 1: The CPT for the variable C11 in the BSM for knowledge diagnosis
Table 1: The CPT for the variable C11 in the BSM for knowledge diagnosis
Figure 3: An example of a generated multiple-choice question
Figure 3: An example of a generated multiple-choice question
Figure 4: The Bayesian network for knowledge diagnosis
Figure 4: The Bayesian network for knowledge diagnosis
Table 2: Computation of the conditional probabilities using function G
Table 2: Computation of the conditional probabilities using function G
Figure 6: State of the network after four questions have been answered 4. Evaluation with real students
Figure 6: State of the network after four questions have been answered 4. Evaluation with real students
Figure 7: An example of a question in the written exam
Figure 7: An example of a question in the written exam
Figure 8 illustrates the relationships of each concept with the different questions. Figure 8: Relationships among concepts and questions in the written exam
Figure 8 illustrates the relationships of each concept with the different questions. Figure 8: Relationships among concepts and questions in the written exam
Figure 9: Relationships among concepts and questions in the computer test
Figure 9: Relationships among concepts and questions in the computer test
Figure 10 shows the Bland-Altman plots for each of this categories. Each point in the Bland-Altman plots represents one of the concepts. The mean of the scores given by the two experts is represented in the abcissa (x-axis) value, while the difference among the two scores ins represented as the ordinate (y- axis) value. If the score tends to agree, the y-axis value will be near to 0. The lower and higher values of the abcissa represent concepts with lower and higher level of knowledge, respectively. Also, Table 2 shows the 0.05 confidence intervals for the mean difference, at each level of granularity. Figure 10: Bland-Altman plots for inter-agreement among experts
Figure 10 shows the Bland-Altman plots for each of this categories. Each point in the Bland-Altman plots represents one of the concepts. The mean of the scores given by the two experts is represented in the abcissa (x-axis) value, while the difference among the two scores ins represented as the ordinate (y- axis) value. If the score tends to agree, the y-axis value will be near to 0. The lower and higher values of the abcissa represent concepts with lower and higher level of knowledge, respectively. Also, Table 2 shows the 0.05 confidence intervals for the mean difference, at each level of granularity. Figure 10: Bland-Altman plots for inter-agreement among experts
Table 3: Confidence intervals for the mean difference among experts
Table 3: Confidence intervals for the mean difference among experts
Table 4: Confidence intervals for the mean difference among expert average, BN test and BN exam
Table 4: Confidence intervals for the mean difference among expert average, BN test and BN exam
Table 5: Confidence intervals for the mean difference among expert’s average, BN test anc BN exam, at the overall concept C12. We can see that, compared to the expert average, PCA tends to over- estimate. This is not surprising as there is a 0.5 guessing factor that has
Table 5: Confidence intervals for the mean difference among expert’s average, BN test anc BN exam, at the overall concept C12. We can see that, compared to the expert average, PCA tends to over- estimate. This is not surprising as there is a 0.5 guessing factor that has
ene RE EES Oe ee ee ene ne NN IL EE ES To this end, we assigned an overall score for each student, based on the percentage of correct answers (PCA) in the computer test, and used this measure as an estimation of the overall performance of the student in the test. This new estimation is then compared to the diagnosis obtained for the overall concept C12 for both the BSM in the computer test and the BSM in the written exam (BN test and BN exam, respectively), and to the experts average (Exp.Av.). Figure 12 shows the corresponding Bland-Altman plots. Figure 12: Bland-Altman plots for BN test, Bn exam and Expert Average (in the overall concept C12)
ene RE EES Oe ee ee ene ne NN IL EE ES To this end, we assigned an overall score for each student, based on the percentage of correct answers (PCA) in the computer test, and used this measure as an estimation of the overall performance of the student in the test. This new estimation is then compared to the diagnosis obtained for the overall concept C12 for both the BSM in the computer test and the BSM in the written exam (BN test and BN exam, respectively), and to the experts average (Exp.Av.). Figure 12 shows the corresponding Bland-Altman plots. Figure 12: Bland-Altman plots for BN test, Bn exam and Expert Average (in the overall concept C12)
Related topics:
Computer ScienceBayesian NetworksComputers in EducationComputerized Adaptive TestingBayesian NetworkStudent Modeling
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved