Vurderingsdokument ST2305 - spring 2025

Percentage grades:

A: 48-55
B: 42-47.99
C: 35-41.99
D: 29-34.99
E: 21-28.99
F:  0-20.99 

Distribution of points:

Section 1
1: Automatically graded max   3 points
2: Automatially graded max 2 points


Section 2:
3: Important to write an equation with left and righ-hand side, representing the quantities 
   in the application text.
   It should made clear to which covariable which effect size belongs. 

   The use of the same variable for both covariates is not correct. 

   4 Points for correct linear model
   2 Points for use of variable names x_1 and x_2 (or similar) and not denoting what 
   x_1 and x_2 is.

4: Positive effect of both covariates  2 points
   Correct interpretation of what effects represent, see solution sketch. 2 points
   Total max of 4 points

5: 2 points. needs to be clear variation of what. Writing only percentage of 
	variation explained leads to 1 point.

6. Per plot 2 points, 1 point for describing what is of and 1 point explaining why 	
	this is an issues. Total max of 6 points.

7. Inclusion quadratic terms to adjust for curved residual plot 2 points. 
   Outlier (do not just remove!!!!), check for typos and evaluate 
   whether to remove, 2 points. Total max of 4 points.

Section 3:

8: Useful for comparison 2 points. 
   Mean 0, SD of 1 for each covariate, subtracting the mean and 
   dividing by the standard deviation for each covariate 2 points. 
   Total max of 4 points.

9: Interpretation of coefficients no longer related to a one-unit change in each covariate 
   holding the other constant. However expected change in y for a 1 standard deviation 	
   increase. 4 points for relating the two interpretations. 

Section 4:

10: Automatically graded 1 point

11: Automtically graded. Manually corrected as formulas for binomial distribution were not 	given. Giving boundaries that lie symmetrically around MLE and are within 0 and 1  
        lead to 5 points.
    Total max. 6 points.

Section 5:

12: Automatically corrected: 1 point

13: Automatically corrected, but manually adjusted. For the second part also 5.6 lead to 2 points. Total max of 6 points, 3 per questions. 
Assuming that the log goes to basis 10 and then reverting using 10^leads to 5 points. 


14: Model1 is to be preferred 2 points
    Explanation using output given. 2 points.
    Total max of 4 points.

15: Mentioning overdispersion 2 points 
    Proposal to adjust for overdispersion using quasi-Poisson or negBin 2 points 
    Inclusion of more explaining factors or mixed effects model 2 points.
    Total max of 6 points.