0

I have some questions about pre-tests and post-tests:

  1. Can I have different items (questions) for pre-test and post-test? Or is it necessary to have the same item?
  2. Can post-test be used as a summative assessment?
  3. Can post-test be used to evaluate if the students have achieved the desired learning outcomes? Can you provide me with references. Thank you for your time.
sammyyy
  • 51

2 Answers2

1

  1. You do not need to have the same questions on the pre- and post-test; whether you should have the same items depends on context.

  2. I don't know what a 'summative assessment" is

  3. There are problems with pre-test post-test designs, especially when the tests are not perfect measures of the quality they are testing. Specifically, you may be correlating change with error (statistical error, not error on the tests). Using a multilevel model with more than two time points deals with a lot of these.

For example, picture a test in a school. Suppose that, on the day of pre-test, John is having a terrible day. He fought with his parents the night before, didn't get sleep, skipped breakfast and is just not having a good day. Jill, on the other hand, is having an excellent day. Now, on the post-test, both have typical days; further suppose that they learned the same amount. John's score will go up by more than Jill's.

Peter Flom
  • 119,535
  • 36
  • 175
  • 383
  • 1
    "picture a test in a school. Suppose that, on the day of pre-test, John is having a terrible day. He fought with his parents the night before, didn't get sleep, skipped breakfast and is just not having a good day. Jill, on the other hand, is having an excellent day." Are you sure he wants to draw conclusions on the individual level? Otherwise, if you take say a 100 students instead of 2, they will not all randomly have a bad day at the same time. There might be still other reasons that invalidate your setup. For example if there was a national tragedy just before one of the test days. – David Ernst Oct 05 '17 at 11:14
  • You will nevertheless be correlating error. The kids who had bad days will be seen to be improving, the kids who had good days will be seen to be getting worse (or improving less). This is covered in detail in several books on longitudinal data. – Peter Flom Oct 06 '17 at 19:29
-1

  1. You need to use the same variables.

  2. Not quite sure what do you mean by summative assessment. You just need to compare the results between the two tests.

  3. Check this article if it's helpful: "The relationship between students' emotional intelligence and writing achivement. Using ANOVA is actually better the T-test."

user5226582
  • 205
user179484
  • 1