| Date: | Mon, 3 Oct 2005 18:04:13 +0200 |
|---|
| Reply-To: | Miroslav Pavelka <m.pavelka@SH.CVUT.CZ> |
|---|
| Sender: | "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU> |
|---|
| From: | Miroslav Pavelka <m.pavelka@SH.CVUT.CZ> |
|---|
| Organization: | Czech Technical University |
|---|
| Subject: | Re: How the "standard error" (denominator in t-value) is
calculated? |
|---|
|
| In-Reply-To: | <dhponf$u20$1@ns.felk.cvut.cz> |
|---|
| Content-Type: | text/plain; charset=us-ascii; format=flowed |
|---|
Problem solwed (matlab code):
(e^2/((n-k-1)*(sum(x-mean(x)))^2))^0.5
Miroslav Pavelka wrote:
> What is tho formula the SAS calculated the standard error in regression
> node? (Enterprise Miner 4.3) I found the t-value is calculated as
> Estimate/Standard Error. But I cann't find the formula for the Standard
> Error of Regression Coeficient
> Pleas advice
>
> Bolow is the example of results, I tried simulate calculation of
> Standard Error in Excel with no success.
>
> Miroslav Pavelka
>
>
> Analysis of Maximum Likelihood Estimates
>
> Standard
> Parameter DF Estimate Error t Value
> Pr > |t|
>
> Intercept 1 0.9746 0.0333 29.23
> <.0001
> a1 1 0.00487 0.00315 1.55
> 0.1225
> a2 1 1.0198 0.0316 32.27
> <.0001
> a3 1 6.9640 0.0310 224.73
> <.0001
> a4 1 1.9955 0.0313 63.71
> <.0001
>
>
>
>
>
>
| 
