# F value in hierarchal regression query

#### DavidH

##### New Member
hi,
I hope all is well. I am reading a paper that reports the following F values for a variable that enters a hierarchal regression with 2 variables already entered in block 1, and 2 variables already entered in block 2. For the third block, the authors report F (3,171) = 0.36 p < .001 and in another model for different DV, report F (3,171) = 0.43 p < .001.
I thought that F <1 in regression meant that the error term was bigger than the systematic factor, and therefore could not be statistically significant, so I am curious as to whether a F that is close to zero can be significant with n of 176.
Any thoughts or suggested reading would be very welcome
thanks,
David

#### obh

##### Well-Known Member
F (3,171) = 0.36
P( X > 0.36 ) = 0.781966.
@hlsmith do I miss something?

#### hlsmith

##### Less is more. Stay pure. Stay poor.
I know this is stats101 stuff, but I don't do a lot of f-dist tests. I also only run a multilevel model probably about every other year. If I read the description right, they may have 3-levels, which is pretty rare in application, since you need the right amount of data and structure. But I would imagine an f-test is an f-test. Like how you can conduct a t-test or a t-test in linear regression to test coefficients.

F-dist based test is comparing the variability between groups versus within groups (relative), looking for more explained variability in the numerator, so the 1 rule would kind of make sense. I got the same value as @obh

So the authors may have mis-reported the values or you may have mis-read them - but yeah it seems off.