After computing the standard error of regression, my answer is 3.55 what does the value mean
Asked
Active
Viewed 658 times
0
-
It depends on which quantity you're calling "the standard error of regression". – Glen_b Oct 02 '14 at 03:13
-
@ Glen: here is how I defined the standard error of the model Se = √ Σ(Yi – Yhat)2 / n-2 – Anie Oct 02 '14 at 03:20
1 Answers
1
Assuming by standard error of regression you mean the residual standard deviation, $s$, where $s^2=\frac{\text{RSS}}{n-p}$ -- e.g. like this (edited regression output from R):
then that quantity, often called $s$, represents an estimate of the standard deviation of the error term in your regression model, a typical distance of points from the regression line.
It's measured in the units of the original observations.
Glen_b
- 257,508
- 32
- 553
- 939
-
Thanks a bunch Glen_b. It follows then from my computed value that the typical distance of points from the regression line for my data is 3.55. Sorry I really don't understand the last part - "measured in units of the original observation". My sample population is 12 with degree of freedom = 10.Thanks so far for your assistance just a bit confused cos different write ups have different explanations – Anie Oct 02 '14 at 03:35
-
What I meant is that if your response is measured in hours or meters or units of IQ, then so is $s$. If you write a new question outlining what contradictions you see in the writeups (preferably with quotes), you might be able to get those particular issues clarified. – Glen_b Oct 02 '14 at 04:26
-
@Glen_b : how we can interpret residual standard deviation? does Residual SD helps in drawing any inference from regression output? – user269867 Oct 06 '14 at 12:51
-
Besides the interpretation I mentioned in my answer, it's used in finding every parameter standard error, and in the F-statistic. – Glen_b Oct 06 '14 at 12:54