reg. ana. F sig undefined but p values are ok ?

#1
Hi people,
this is my first message.

I try to use regression in my studies and there are some difficulties that I met. I use excel for reg analysis.

I have 3 independent vars, a,w and t (and additional power of the same vars - totalling eight vars)

y= a + w + a^2 + a^4 + w^2 + w^3 + t^3 + t^4

and I investigate the effects on y dependent var.

first of all; I found four set of coefficents that can make the model very significant (as I thought). In those sets,
1- p values of all vars are less than 0.05, :tup:
2- SSR are found to be very low (y mean is around 1000, and SSR is calculates as min 3 and max 30) :tup:
3-R2s are all 0.999999 :tup:
4-F rat is max around 401000 and min around 50000 (so very high) :tup:
5-but all four sets have a undefined Fsig (#NUM!) :confused:

I also upload the anova table results in case you need to see :

1. http://a.imageshack.us/img8/9827/slide1ak.jpg
2. http://img839.imageshack.us/img839/7734/slide2m.jpg
3. http://a.imageshack.us/img28/8939/slide3d.jpg
4. http://a.imageshack.us/img130/2723/slide4sh.jpg

Can you help me please ?

note: I jus know the name of homoscedastic - dont know any test, or even what is that exactly

Thank you in advance.