Validation Evidence from using Generalizability Theory in a Basic-Science Course
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
DESCRIPTION OF THE PROBLEM: Reliability is critical validation evidence on which to base high-stakes decision-making. Many times, one exam in a didactic course may not be acceptably reliable on its own. But how much might multiple exams add when combined together? THE INNOVATION: To improve validation evidence towards high-stakes decision-making, Generalizability Theory (G-Theory) can combine reliabilities from multiple exams into one composite-reliability (G_String IV software). Further, G-Theory decision-studies can illustrate changes in course-grade reliability, depending on the number of exams and exam-items. CRITICAL ANALYSIS: 101 first-year PharmD students took two midterm-exams and one final-exam in a pharmaceutics course. Individually, Exam1 had 50MCQ (KR-20=0.69), Exam2 had 43MCQ (KR-20=0.65), and Exam3 had 67MCQ (KR-20=0.67). After combining exam occasions using G-Theory, the composite-reliability was 0.71 for overall course-grades-better than any exam alone. Remarkably, increased numbers of exam occasions showed fewer items per exam were needed, and fewer items over all exams, to obtain an acceptable composite-reliability. Acceptable reliability could be achieved with different combinations of number of MCQs on each exam and number of exam occasions. IMPLICATIONS: G-Theory provided reliability critical validation evidence towards high-stakes decision-making. Final course-grades appeared quite reliable after combining multiple course exams-though this reliability could and should be improved. Notably, more exam occasions allowed fewer items per exam and fewer items over all the exams. Thus, one added benefit of more exam occasions for educators is developing fewer items per exam and fewer items over all exams.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.021 | 0.151 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.023 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it