MQM fixed residual variance

[Nathaniel Burner]

Hi Karl,

I am having trouble understanding how MQM fixes residual error to that of the full model as described by Jansen in the chapter in "Quantitative Trait Loci in Inbred Lines" (attached).

 Is this similar to comparing the change in SSE during the drop-one analysis, in that the MS of the change in SSE is divided by the MS for the full model to obtain the F-statistic?

Or, does MQM compare the MS of the reduced model to the MSE of the full model and test for significance, or lack thereof, that way?

Thanks