Dear List,
I have five instructors teaching 45 students at two different levels distributed as follows:
Instructor 1: 8 students level A
Instructor 2: 8 students level A
Instructor 3: 9 students level B
Instructor 4: 9 students level B
Instructor 5: 9 students level B.
Students are asked to self assess their writing in English and in their native language using Likert-type scale (1= not at all, 5 = very well; better than most of people, and 6 = I don't know).
Instructors are asked to assess the students' writing in the
course using Likert-type scale as above.
There are four items (two in respect to English and two in regard with native language) in both students and instructors questionnaire. Instructors do not cross teach.
Here is an example from Instructor 1:
Student1 4 6 4 3
Student2 4 6 4 3
Student3 3 2 3 2
Student4 5 3 4 6
Student5 3 6 3 6
Student6 4 3 4 2
Student7 4 2 4 6
Student8 5 3 4 1
Instruct1 6 2 6 6
First thing I did is to put all "I don't know" answers with "1 = not at all" as there is no room for "no response or don't
know" in Likert scale. Likert scale should be at least ordinal by general agreement.
Next, I wish I would be able to do further analysis as correlation analysis, concordance between raters, etc..
Has any got any suggestion?
Ray.
|