| Author |
 Topic  |
|
|
hauli
Austria
Posts |
 Posted - 02/26/2007 : 09:42:41 AM
|
Origin Version (Select Help-->About Origin): 7.5
Operating System: Windows XP
Dear all,
I am analyzing results from a toxicological test (AMES test). Empirically I know that the resulting dose-resonse curve starts linearly, then flattens and finally shows a downward trend.
I am only interested in the starting linear region, therefore I follow the following procedure to decide on which points ly on the linear part:
I fit a polynom of 2nd order to the data, and check with a t-test, if the regression coefficient of the quadratic term is significantly different from zero. If so, I omit the data at the highest dose and start the polynomial fit again. This procedure is done untill the coefficient of the quadratic term is not statistically different from zero.
Using Origins tool polynomial fit, I can choose "advanced results" and the regression coefficients are tested with a t-test for significance and the regression ANOVA is displayed.
My question now is, how can I perform the t-test for B2 in
y= A + B1*x + B2*x^2 programatically with Origin C?
Very best regards,
Hauli
|
|
|
zachary_origin
China
Posts |
Posted - 02/26/2007 : 9:40:17 PM
|
Hi Hauli,
Now in Origin75, there is no explicit OriginC function for getting the values of t-test. However, LabTalk script can be used for this (see stat.pr in LabTalk Language Reference). From OriginC, the result of the LabTalk script can be accessed with the notation of _LT_Obj which acts as a pragma. You can open \OriginInstallation\OriginC\System\internal.c and go to the function fitpoly to see how it works. Hope it helps.
Zachary
OriginLab Technical Services.
 |
 |
|
| |
Topic |
|
|
|
t-test for significance of regression coefficients