What will be the right statistical approach for this design please?

#1
I have a data of:

30 medical students who have rated visibility of 40 brain anatomical structures (each structure has been rated on a Likert scale of 1-5, where 1 corresponds to "not visible" and 5- "clearly visible").

I am trying to figure out what will be the appropriate statistical test in this situation please to figure out the correlation among the ratings of the different students? I am not sure even if "correlation" will be the right test here, any advice for statistical analysis of this data please?
Thank you.
 

Karabiner

TS Contributor
#2
30 medical students who have rated visibility of 40 brain anatomical structures (each structure has been rated on a Likert scale of 1-5, where 1 corresponds to "not visible" and 5- "clearly visible").
Do you really mean Likert scale (consisting of several Likert-type items which are summed up, resulting in an interval scaled measure), or just a single Likert-type item (a rating item with a 1-5 response format; ordinal scaled)?

I am trying to figure out what will be the appropriate statistical test in this situation please to figure out the correlation among the ratings of the different students? I am not sure even if "correlation" will be the right test here, any advice for statistical analysis of this data please?
What is the research question?

With kind regards

Karabiner
 
#3
I mean the latter, a rating item with a 1-5 response format; ordinal scaled. Sorry for the ambiguity.

The research question is two-fold: 1) What is the overall correlation between the students in identifying these structures (on the 1-5 scale)? Is there any statistically significance difference in between their ratings?

2)Is year of medical school study associated with rating (I guess this will be a regression analysis)?
 

Karabiner

TS Contributor
#5
1) What is the overall correlation between the students in identifying these structures (on the 1-5 scale)?
As far as I know, in a situation with multiple raters and multiple objects and an ordinal scaled measure,
the general reliability of the rating (if this is what you mean by "overall correlation between students")
can be assessed using Krippendorf's alpha.
Is there any statistically significance difference in between their ratings?
I am not sure what exactely you mean by this. If you want to know wether students differ with
regard to their ratings of the same objects, then this is not a useful question. Of course they do
(mind that "statistically significant" would just mean that you reject the hypothesis "the difference
between students is exactely =0.00000000000 in the population where these ratings were sampled
from"). But maybe you want to know something else?

With kind regards

Karabiner