Why don the log likelihood tests (ll.pm) and Pearsons test (x2.pm) report the significance values associated with their raw scores?
We don’t assign significance values to these tests since setting the p-values seems to be a fairly arbitrary choice. It isn’t clear that p should be .01, .05, or something else. This strikes us as an important limitation of significance tests. Rather than setting an arbitrary cutoff on these values, we recommend looking at the raw scores from these tests in order to establish cutoffs. Having said all this, it is very likely that future versions of NSP will have this capability.
Related Questions
- I created a test suite from a model using Reactis, but when I run the tests within Reactis, the output values are different than those stored in the tests. Why is this?
- Why don the log likelihood tests (ll.pm) and Pearsons test (x2.pm) report the significance values associated with their raw scores?
- What is the difference between ESL/Generalist EC-6 TExES test and Generalist EC-6 with ESL Supplemental TExES tests?