linear trend lines

[bonzo]

can anyone help?
I have to put 95% confidence intervals around a linear trend line. How
do I work out the values for the trend line? I am using a scatter
graph with 2 sets of data.
Please help!

 

[Jerry W. Lewis]

The confidence bound for the estimated line at x is

=FORECAST(x,known_y's,known_x's) +/- TINV((1-conf)*2,COUNT(known_y's)-2)
*STEYX(known_y's,known_x's)
*SQRT(1/COUNT(known_y's)+(x-AVERAGE(known_x's))^2/DEVSQ(known_x's))

the confidence bound for an individual observation at x is

=FORECAST(x,known_y's,known_x's) +/- TINV((1-conf)*2,COUNT(known_y's)-2)
*STEYX(known_y's,known_x's)
*SQRT(1+1/COUNT(known_y's)+(x-AVERAGE(known_x's))^2/DEVSQ(known_x's))

Jerry

 

[Gary Klass]

bonzo:
If Jerry's solution may be too complicated for you, try this:
Run the regression program in Data Analsysi add-in.

use the "Standard Error" (of the estimate, underneath the R-square).

Calculate two new variables plus or minus 2 standard errors. and plot them.

The confidence bound for the estimated line at x is

=FORECAST(x,known_y's,known_x's) +/- TINV((1-conf)*2,COUNT(known_y's)-2)
*STEYX(known_y's,known_x's)
*SQRT(1/COUNT(known_y's)+(x-AVERAGE(known_x's))^2/DEVSQ(known_x's))

the confidence bound for an individual observation at x is

=FORECAST(x,known_y's,known_x's) +/- TINV((1-conf)*2,COUNT(known_y's)-2)
*STEYX(known_y's,known_x's)
*SQRT(1+1/COUNT(known_y's)+(x-AVERAGE(known_x's))^2/DEVSQ(known_x's))

Jerry

--
Gary Klass
(e-mail address removed)
Editor, PSRT-L
4600 Politics and Government
Illinois State University
Normal, Illinois 61790
http://LILT.ILSTU.EDU/gmklass
(309) 438-7852
(fax) 438-7638
The Chart of the Week: http://LILT.ILSTU.EDU/gmklass/cow