Academia.eduAcademia.edu

Figure 7 – uploaded by Shauna Sweet

See full PDFdownloadDownload figure
Table 5. Summary of Simulation Conditions Statistical investigations of AIG are unique in that a population of items needs to be

Table 5 Summary of Simulation Conditions Statistical investigations of AIG are unique in that a population of items needs to be

Related Figures (39)

of) individual instances which are the actual tasks presented to students on an assessment. Increased processor speed and greater programming flexibility make it possible to imagine
of) individual instances which are the actual tasks presented to students on an assessment. Increased processor speed and greater programming flexibility make it possible to imagine
below in Figure 2. essential surface features. A graphical representation of this item generation process is shown and the development of the RSM (Equation 2.4) by placing structure on the mean difficulty of
below in Figure 2. essential surface features. A graphical representation of this item generation process is shown and the development of the RSM (Equation 2.4) by placing structure on the mean difficulty of
describe groupings of items at Level 2, categories (Cho et al, 2014) versus families (Geerlings et The AMIS model decomposes the mean family-wise discrimination and difficulty
describe groupings of items at Level 2, categories (Cho et al, 2014) versus families (Geerlings et The AMIS model decomposes the mean family-wise discrimination and difficulty
Educational Objectives and Cognitive Task Models. Whether automatically generated
Educational Objectives and Cognitive Task Models. Whether automatically generated
Figure 9. Notional graphic illustrating the item generation process underlying the Generative Process Model The proposed conceptual framework can be readily translated into a pre-calibration
Figure 9. Notional graphic illustrating the item generation process underlying the Generative Process Model The proposed conceptual framework can be readily translated into a pre-calibration
Figure 10. Illustration of generative process and products for example QSC, Item Form, and Item Model 4.2.1 Components of the Item Generation Process
Figure 10. Illustration of generative process and products for example QSC, Item Form, and Item Model 4.2.1 Components of the Item Generation Process
Table 1. Distribution of Items and Observations by QSC Item Forms. Aligned to the eleven QSCs, a total of fifteen item forms were identified, where each item form is defined — as illustrated in Figure 10 - by the unique combination of an
Table 1. Distribution of Items and Observations by QSC Item Forms. Aligned to the eleven QSCs, a total of fifteen item forms were identified, where each item form is defined — as illustrated in Figure 10 - by the unique combination of an
Table 2. Distribution of Items and Observations by Item Form
Table 2. Distribution of Items and Observations by Item Form
Form-level Characteristics and Item Models. From each item form, the MIG produces
Form-level Characteristics and Item Models. From each item form, the MIG produces
Table 3. Distribution of Items across QSC, Operation, and Coded Content Features
Table 3. Distribution of Items across QSC, Operation, and Coded Content Features
Content Integration. As noted at the beginning of the chapter, the item generation process used by MetaMetrics was developed to systematically manipulate only a few key
Content Integration. As noted at the beginning of the chapter, the item generation process used by MetaMetrics was developed to systematically manipulate only a few key
Item Families. The total number of items generated from each parent item (within each item family) ranges from | to 186. These families of items generated from each parent item are
Item Families. The total number of items generated from each parent item (within each item family) ranges from | to 186. These families of items generated from each parent item are
Figure 12. Distribution of Items and Mean Empirical Log Odds by Item Family and Item Form
Figure 12. Distribution of Items and Mean Empirical Log Odds by Item Family and Item Form
Table 4. Generation Process Quality This variations in the quality of the item generation process were achieved by applying different understood, and through a process that generates items of variable quality.
Table 4. Generation Process Quality This variations in the quality of the item generation process were achieved by applying different understood, and through a process that generates items of variable quality.
Table 6. Generating Values and Parameter Estimates from “Large Sample” Condition Generating parameters were satisfactorily recovered, as shown in Table 6 below.
Table 6. Generating Values and Parameter Estimates from “Large Sample” Condition Generating parameters were satisfactorily recovered, as shown in Table 6 below.
Table 7. Relative Bias for Template Means, Fixed and Random Effects Across Conditions, Study 1
Table 7. Relative Bias for Template Means, Fixed and Random Effects Across Conditions, Study 1
Table 8. Variability of Estimates for Template Means, Fixed and Random Effects Across Conditions, Study 1 Note: Not all combinations of factors were fully explored. The shaded areas of the table indicate where no data is available
Table 8. Variability of Estimates for Template Means, Fixed and Random Effects Across Conditions, Study 1 Note: Not all combinations of factors were fully explored. The shaded areas of the table indicate where no data is available
Table 9. Relative Bias for Template Means, Fixed and Random Effects Across Conditions, Study 2
Table 9. Relative Bias for Template Means, Fixed and Random Effects Across Conditions, Study 2
Table 10. Variability of Estimates for Template Means, Fixed and Random Effects Across Conditions, Study 2 Note: Not all combinations of factors were fully explored. The shaded areas of the table indicate where no data is available.
Table 10. Variability of Estimates for Template Means, Fixed and Random Effects Across Conditions, Study 2 Note: Not all combinations of factors were fully explored. The shaded areas of the table indicate where no data is available.
Note: Not all combinations of factors were fully explored. The shaded areas of the table indicate where no data is available. Table 11. Relative Bias for Template Means, Fixed and Random Effects Across Conditions, Study 2
Note: Not all combinations of factors were fully explored. The shaded areas of the table indicate where no data is available. Table 11. Relative Bias for Template Means, Fixed and Random Effects Across Conditions, Study 2
Note: Not all combinations of factors were fully explored. The shaded areas of the table indicate where no data is available. Table 12. Variability of Estimates for Template Means, Fixed and Random Effects Across Conditions, Study 3
Note: Not all combinations of factors were fully explored. The shaded areas of the table indicate where no data is available. Table 12. Variability of Estimates for Template Means, Fixed and Random Effects Across Conditions, Study 3
Table 13. Distribution of Items and Observations by QSC
Table 13. Distribution of Items and Observations by QSC
Table 14. Deviance Information Criteria for Six Analytic Models nature of the data generation process.
Table 14. Deviance Information Criteria for Six Analytic Models nature of the data generation process.
Table 15. Correlation Between Item Difficulty Estimates Using Different Calibration Models, by Templat
Table 15. Correlation Between Item Difficulty Estimates Using Different Calibration Models, by Templat
Table 16. Parameter Estimates for Item Generation Process Components Using the Unconstrained Generative Process Model for Item Calibration
Table 16. Parameter Estimates for Item Generation Process Components Using the Unconstrained Generative Process Model for Item Calibration
Echoing the previous study, the within-family variance was manipulated to be structure. The analytic models considered in this study are shown in Table 18 below
Echoing the previous study, the within-family variance was manipulated to be structure. The analytic models considered in this study are shown in Table 18 below
Response Data Generation. Response data were generated for each condition below.
Response Data Generation. Response data were generated for each condition below.
Note: The shaded areas of the table indicate where no data is available. Table 21. Parameter Non-Convergence as a Percentage of Replications Per Condition
Note: The shaded areas of the table indicate where no data is available. Table 21. Parameter Non-Convergence as a Percentage of Replications Per Condition
Note: The shaded areas of the table indicate where no data is available. Table 22. EAP Parameter Estimates by Condition, Median Values Across Replications
Note: The shaded areas of the table indicate where no data is available. Table 22. EAP Parameter Estimates by Condition, Median Values Across Replications
Note: The shaded areas of the table indicate where no data 1s available. Note: Relative bias was calculated for all parameters for each replication using the generating parameters and EAP estimates of those parameters. The median values within each condition are shown in this table. Table 23. Median Relative Bias Across Replications by Simulation Condition
Note: The shaded areas of the table indicate where no data 1s available. Note: Relative bias was calculated for all parameters for each replication using the generating parameters and EAP estimates of those parameters. The median values within each condition are shown in this table. Table 23. Median Relative Bias Across Replications by Simulation Condition
Note: The shaded areas of the table indicate where no data is available. Note: The empirical variability of the EAP estimates was calculated within each condition across r=50 replications Table 24. Empirical Variability Across Replications, by Simulation Condition
Note: The shaded areas of the table indicate where no data is available. Note: The empirical variability of the EAP estimates was calculated within each condition across r=50 replications Table 24. Empirical Variability Across Replications, by Simulation Condition
// varying item family diffculty across item families within models // decomposition of family means into a model mean and fixed effects family mu = model mu[mm] + fammu_offset[mm] .* fam_resid[mm]; family mu[multten] = family mu[multten] + char_ml0[mm[multten]]; family mu[threedig] = family mu[threedig] + char _d3[mm[threedig] ];
// varying item family diffculty across item families within models // decomposition of family means into a model mean and fixed effects family mu = model mu[mm] + fammu_offset[mm] .* fam_resid[mm]; family mu[multten] = family mu[multten] + char_ml0[mm[multten]]; family mu[threedig] = family mu[threedig] + char _d3[mm[threedig] ];
Related topics:
PsychologyComputer ScienceStatisticsAutomatic Item Generation
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved