I am given the following table:
Regression 3 ?
Residual 61 315.2025 Total 63 700.4500
Their answers are saying the MSE of the F stat = 315.2025/61
I did 315.2025/60
I think the problem is they have the total number of observations off. They stated 64, but they are calculating the MSE as if there was 65 observations. Am I crazy?
How many coefficients are in the model, not counting the intercept?
Did they state a sample size?
It’s possibly an old example where they forgot to change the numbers correctly.
MSE= SSE/(n-k-1)= 315.2025/(n-k-1) where n-3-1=61 --> n=61+3+1=65 So the observations should be 65.
MSE=315.2025/61
RSS=SST-SSE = 700.45-315.2025= 385.2475 —> MSR= RSS/k = 385.2475/3=128.416
F-stat= MSR/MSE = 128l.416/5.167= 24.85
So, it looks like they just put the wrong number of observations when they ran the regression (because i’m assuming they just copy/pasted the results of the table onto the book).
If they wouldn’t have given you the number of observations it would make sense, because they expect you to be able to know to just use SSE/df from the table.