Timeline for Comparison between Log-likelihood ratios and beta coefficients
Current License: CC BY-SA 3.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 30, 2014 at 9:46 | comment | added | Glen_b | I guess it depends on which test statistic you mean. For example, the z-test for an autocorrelation will likely be almost identical to the LR test (though I haven't worked out the LR test, I think from the form of the likelihood for an AR(p) (which apart from a term involving the first p observations, looks the same as a regression) I think it should work out. | |
| Jan 30, 2014 at 9:37 | comment | added | SamPassmore | Thanks Glen. An very thorough answer. +1. I have one follow up question. I am actually using this in regard to spatial auto regressive parameters, but feared no one would respond if i was more specific. Do you think that we could make the same conclusions? That the test statistic for an auto correlation parameter will be approximately the same as a LLR test statistic between a model with and without the parameter? Thanks again | |
| Jan 30, 2014 at 6:26 | vote | accept | SamPassmore | ||
| Jan 30, 2014 at 4:13 | history | edited | Glen_b | CC BY-SA 3.0 |
added 155 characters in body
|
| Jan 30, 2014 at 4:08 | history | answered | Glen_b | CC BY-SA 3.0 |