Lab 4 Rubric

Total: 18 points
Each row is scored from 0–3 points based on the criteria.

Lab activity	3 points	2 points	1 point	0 points
Restate research question	Research question is clearly and correctly restated.	Research question is mostly clear, but minor issues with clarity.	Research question is vague or partially incomplete.	Research question is missing.
Step 1: SLR Screening	All 16 variables are tested against the outcome; results (F-stats/p-values) are clearly displayed using add1() or individual models.	Most variables are tested, but some are missing or results are difficult to find.	Only a few variables are tested or code contains significant errors.	No SLR screening performed.
Step 2: Preliminary Selection	Correct logic applied (p < 0.25); initial MLR model is fit with the correct subset and table is displayed.	Model fit with minor errors in variable selection logic or table formatting.	Model fit attempted but ignores the p < 0.25 selection rule.	No preliminary MLR model provided.
Step 3: Purposeful Selection Logic	Provides clear explanation for every dropped variable (common sense + p-value + \(\Delta\%\)) and includes code for at least one \(\Delta\%\) calculation.	Explains dropped variables but lacks the code example for \(\Delta\%\) or reasoning is inconsistent.	Variables are dropped without explanation or logical justification.	No evidence of the variable reduction process.
Step 5: Interaction Testing	Uses add1() correctly to test interactions; adds one subquestion interaction (if applicable) and runs the updated model.	Tests interactions but misinterprets significance or fails to run the final updated model.	Attempted interaction testing but code is broken or logic is fundamentally flawed.	Interaction testing is missing.
Final Visualization (Forest/Table)	Produces a clean, readable Forest Plot or tbl_regression(); labels are formatted (not raw code names) and axes are clear.	Visualization is produced but is messy (e.g., raw variable names like att7) or difficult to read.	Visualization is attempted but missing key components like Confidence Intervals.	No forest plot or regression table provided.