I have variance of response. How can I find it's MSE?

  • Thread starter HF08
  • Start date
  • #1
39
0

Main Question or Discussion Point

To All,

I did a study and my response is defined as y = b1x1 + b2x2 + e where e ~N(0,1).
I have y~N(4,33). In my data results, I did an ordinary least squares regression model
for y = b1x1+b2x2+ e. The ANOVA is telling me the mean of y is 4, but MSE is 1.

So here is my question. If I know y~N(4,33). How can I determine the ANOVA MSE
is going to be 1?

Thanks,
HF08
 

Answers and Replies

  • #2
283
3
I think you are getting a bit confused here. [tex]\underline{y}=X\underline{\beta} + \underline{\epsilon}[/tex] is your model? There is no way that [tex] y [/tex] could be [tex] N(4,33) [/tex] when your [tex] \epsilon_i [/tex] are [tex] N(0,1) [/tex]. Also that ANOVA does not make sense if you have [tex] \epsilon_i [/tex] as there is no variance to analyse. In this case ANOVA is right. As
[tex] Var(y_i)=Var(\mu_i+\epsilon_i)=Var(\epsilon_i)=1 [/tex].


What sofware are you using for this? Try using R. Don't bother with saying that [tex] \epsilon_i [/tex] are [tex] N(0,1) [/tex] unless you have a very good reason to know this. Usualy [tex] \epsilon_i [/tex] are [tex] N(0,\sigma^2) [/tex] where you will have to estimate [tex] \sigma^2 [/tex]. In R the model you have can be set up by lm(y~b1+b2-1,data=dataname).

Hope this helps
 

Related Threads for: I have variance of response. How can I find it's MSE?

Replies
2
Views
2K
Replies
3
Views
2K
Replies
2
Views
2K
  • Last Post
Replies
1
Views
981
Replies
14
Views
9K
  • Last Post
Replies
4
Views
21K
Replies
3
Views
3K
Top