Hi everyone,

I am not a stats man at all, just a programmer. so sorry if this is a
silly question

I was asked to compute something but I do not know really how to do it.
we have 52 weekly observations of variable A (indiviual patient) and of
variable B (average of the control group). A regression is run per
patient-year: A = a + b*B + e;

from this regression the keep the Error sum of squares (SSE) component
of the R2 is kept. I was then told that from SSE, we can get a close
approximation to the Standard deviation of the weekly values of
variable A.

I was asked to come up with a formula that demonstrates this
approximation.

Can you please give me some advice?

Thanks.

Hi everyone,

I am not a stats man at all, just a programmer. so sorry if this is a
silly question

I was asked to compute something but I do not know really how to do it.
we have 52 weekly observations of variable A (indiviual patient) and of
variable B (average of the control group). A regression is run per
patient-year: A = a + b*B + e;

from this regression the keep the Error sum of squares (SSE) component
of the R2 is kept. I was then told that from SSE, we can get a close
approximation to the Standard deviation of the weekly values of
variable A.

I was asked to come up with a formula that demonstrates this
approximation.

Can you please give me some advice?

Thanks.

On 4 Dec 2006 03:06:26 -0800, cp279@yahoo.com wrote:

Hi everyone,

I am not a stats man at all, just a programmer. so sorry if this is a
silly question

I was asked to compute something but I do not know really how to do it.
we have 52 weekly observations of variable A (indiviual patient) and of
variable B (average of the control group). A regression is run per
patient-year: A = a + b*B + e;

from this regression the keep the Error sum of squares (SSE) component
of the R2 is kept. I was then told that from SSE, we can get a close
approximation to the Standard deviation of the weekly values of
variable A.

I was asked to come up with a formula that demonstrates this
approximation.

Can you please give me some advice?

Thanks.

You probably want sqrt(SEE/(n-2))
-Dick Startz

Sorry,

yes R2 = R-Squared.

As I said I am not a stats man, and if I got the idea right this should
not be a sample dependent problem. What I was told is that you can
mathematically demonstrate that SEE (or SSE)

component of R-Squared can
be a very close approximation to the Standard Deviation of the
dependent variable in the regression... in this case A = a + b*B + e.

So basically I was asked to implement both ways and should that the two
options are basically giving the same information.

Thanks


richardstartz@comcast.net wrote:

On 4 Dec 2006 03:06:26 -0800, cp279@yahoo.com wrote:

Hi everyone,

I am not a stats man at all, just a programmer. so sorry if this is a
silly question

I was asked to compute something but I do not know really how to do it.
we have 52 weekly observations of variable A (indiviual patient) and of
variable B (average of the control group). A regression is run per
patient-year: A = a + b*B + e;

from this regression the keep the Error sum of squares (SSE) component
of the R2 is kept. I was then told that from SSE, we can get a close
approximation to the Standard deviation of the weekly values of
variable A.

I was asked to come up with a formula that demonstrates this
approximation.

Can you please give me some advice?

Thanks.

You probably want sqrt(SEE/(n-2))
-Dick Startz