Setting intercept to zero with SLOPE/STEYX/etc

[garbagesquirrel]

I am using Excel 2003 to calculate standard curves from
spectrophotometry data. The intercept of the linear regression
equation has to be set to zero. Generating trendlines from an x-y plot
of the data, this is no problem; I can just set the intercept in the
trendline options. In order to use the slope, equation, or R^2 values
so generated in another notebook, I have to cut-and-paste the values
from the chart. I would like to be able to use SLOPE on the source
data instead, but without being able to manually adjust the intercept,
the values it generates are useless. Here's an example:

[p-NPP] (μM) - x values
0
10
20
30
40
50

A400nm - y values
0.000
0.206
0.397
0.589
0.781
0.970

Slope from trendline equation, intercept set to 0: 0.0195
Slope from SLOPE function: 0.019334286

The difference between the two is significant enough to affect further
calculations based on those values. Is there some way to adjust the
intercept outside of the chart trendline? Thanks in advance.

 

[Guest]

Use the array formula
=LINEST(ydata,xdata,FALSE,TRUE)
array entered with a 5x2 array selected. Help documents the various
outputs, including slope, R^2 and STEYX. Note that Excel 2003 is the first
Excel version where LINEST correctly calculates R^2, F, or SSreg when the
intercept is forced to zero.

Note that in Excel 2003 the chart trendline R^2 is still incorrect when the
intercept is forced to zero.

Jerry

 

[garbagesquirrel]

Note that in Excel 2003 the chart trendline R^2 is still incorrect when the
intercept is forced to zero.

Jerry

So the R^2 value generated for the trendline -- the one Excel displays
next to the trendline along with the equation -- is off? Is the one
that LINEST spits out any more accurate? Thanks again for the help.

 

[Guest]

Prior to Excel 2003, LINEST and the chart trendline made the same mistake in
concept regarding R^2 when the intercept is forced to zero. In Excel 2003,
the concept was corrected in LINEST, but not in the chart trendline. Example
data with certified results are available from NIST

http://www.itl.nist.gov/div898/strd/lls/data/LINKS/DATA/NoInt1.dat
http://www.itl.nist.gov/div898/strd/lls/data/LINKS/DATA/NoInt2.dat

Jerry