10. Kinds of Criterion-related Validity
Concurrent
C-R V.
Test =
criterion
EXAMPLES?Test
Predictive
Criterion
11. Examples of Criterion-related Validity
One of the objectives of a Spanish course is to measure speaking
(especially pronunciation of vowels) and writing (especially
comments on readings). Testing these abilities in detail could take up
to 2 hours per student (too impractical), but only 20 minutes can be
devoted to each student. Can a twenty-minute session give an
accurate estimate of the student’s ability in these functions?
After the students finish the 20-min test, the teacher decides to
choose a sample of the students taking the evaluation to subject
them to the full 2-hour one. Then he compares the results.
12. Examples of Criterion-related Validity
A student takes a placement test for English. He receives the results
and is placed in 3rd level. How well did the test predict the most
appropriate class for the student? Will he succeed and pass?
21. Reliability Exemplified
STUDENT
SCORE (A)
SCORE (B)
T
80
52
U
61
78
V
98
59
W
43
68
X
52
66
Y
40
98
Z
28
50
• 100-item test
• SCORE A: Monday
• SCORE B: Tuesday
STUDENT
SCORE (A)
SCORE (B)
A
91
97
B
88
83
C
46
50
D
54
52
E
99
99
F
66
70
G
27
25
28. Tips for more reliable tests
1. Use enough items
2. Do not give too much freedom
3. Write unambiguous items
4. Provide clear instructions
5. Ensure that tests are legible
6. Make candidates familiar with formats/techniques
7. Provide non-distracting and uniform conditions
8. Use items that allow objective scoring
9. Provide a detailed scoring key
10. Train scorers
11. Agree upon acceptable responses and scores before scoring
12. Identify candidates by number
13. Employ multiple, independent scoring
29. Reliability and Validity
A reliable test is not always valid and
viceversa.
WHY?
http://www.youtube.com/watch?v=5lG3R6-4T3k
32. Kinds of Washback
POSITIVE WASHBACK
NEGATIVE WASHBACK
Encourages students
Clarifies concepts
…
Makes students nervous, anxious
Makes teachers/ students ignore
«unimportant» topics
…
http://tapchi.vnu.edu.vn/nn_4_09/b.8.pdf
33. Tips for achieving beneficial washback
1. Test the activities whose development you
want to encourage
2. Sample widely and unpredictably
3. Use direct testing
4. Make testing criterion-referenced
5. Base achievement tests on objectives
6. Ensure the understanding of the test
7. Provide assistance if necessary