Standard Error of Estimate (SEE)

[Iamking]

Is the standard error of estimate (SEE) the relationship between the independent and dependent variables or is it the measure of variablity of actual Y-values relative to estimated Y-values from a regression equation?
Also, SEE is the standard deviation of the error terms in the regression. Does this mean if the data is considered linear one standard deviation to the left or right can be found the the T table and used in the test statistic? IF yes, do I use the same test statistic when creating a hypothesis test for the correlation coefficient (t=(r(n-2)^(1/2))/(1-r^2)^(1/2)).
Thanks in advance.

 

[Harrogath]

Standard error of estimate (SEE) is the sd of the errors resulting the regression calculation, thats all.
Error (i) = Actual Y(i) - Estimated Y(i)
Error (i) = Actual Y(i) - b0 - b1 * X1(i) ……………. where (i) = 1, 2, 3, … n
SEE = SD( error(i)^2 )
Remember the data is not linear, the relationship between the independent and dependent variable is considered linear. If not, your linear regression would have no sense.
I don’t understand what you mean after. Can you explain it more clearly?
Regards

 

[Iamking]

Okay, I understand that part but my next question is how does it tie in with simple linear regression?
As for the second part, my question is would I have to do a test statistic and find a critical value for SEE?

 

[S2000magician]

The assumption we use is that the error terms are normally distributed.
Because the SEE is a standard deviation, you’d use a chi-square statistic to test it for a specific value. Recall your Level I test for variance: ns²/σ².
There’s nothing in the Level II curriculum about doing any sort of statistical test on SEE.

 

[Iamking]

Thank you, now I need help conceptualizing the whole topic before moving on the multiple regression.
In laymen’s terms can you help me sum the chapter up in a few sentences?

 

[Harrogath]

Iamking wrote:
Thank you, now I need help conceptualizing the whole topic before moving on the multiple regression.
In laymen’s terms can you help me sum the chapter up in a few sentences?

A regression is an econometric method to desaggregate a variable into others. In simple regression you decompose a dependent variable into an independent variable and an error. In multiple linear regression you decompose it in many indepedent variables and an error. This is the fundamental of regression calculations.
Why would you want to decompose a variable? You want to modelate a variable behavior using other variables, which are much more easy to observe. This aids to make forecasts using historical data.
In real world, linear regression is not used bacause variables do not have linear relationships. Advanced econometrics use other types of calculations and also the topics we learn in CFA 2, so keep them in mind anyway.

 

[CMLSML]

Iamking wrote:
In laymen’s terms can you help me sum the chapter up in a few sentences?

Realize that a layman aint passin this exam. Just sayin