Goodness of fit test

[stats4today]

I am doing polynomial curve fitting in Excel (2003 and 2007). I have used the
r-squared function, but want the F-value, or F-statistic function. Can
someone help?

 

[Mike Middleton]

stats4today -

One approach is to create the X^2, X^3, etc. values on your worksheet that
has the Y and X data. Then use the Regression tool of the Analysis ToolPak,
which provides ANOVA results with and F statistic. (For the Regression tool,
the X, X^2, X^3, etc. values must be in adjacent columns.)

- Mike

http://www.MikeMiddleton.com

 

[Xt]

I am doing polynomial curve fitting in Excel (2003 and 2007). I have usedthe
r-squared function, but want the F-value, or F-statistic function. Can
someone help?

p = number of predictors = power in this case
n = number of points
R2 = r-squared value from Excel trendline

F = R2/(1-R2)*(n-p-1)/p

xt

 

[Jerry W. Lewis]

Note that this tests whether the entire model fits better than just a simple
mean. It does not test whether there are too many or too few polynomial
terms in the model. For that, see Mike's answer.

Jerry