what could be a benefit for the forcing of the trendline through the
origin?
Simple answer: there is never a benefit in imposing constraints on
data. In the context of published work, this would be academic
misconduct.
Many of the explanations have already been stated by others, but I
summarize mine below:
* it assumes zero systematic errors
* it gives an erroneous slope
* a non-zero intercept, when theory says it should be zero, it
provides a great opportunity to discuss systematic errors and
uncertainty in science
* if it is an advanced class, teach them to calculate the uncertainty
in the intercept. In most cases, zero is within the uncertainty. If
not, then there is likely a systematic error.
A classic CHEMISTRY example is the Beer-Lambert law. The law states
that A = e b c. That is: absorbance is linear with concentration and
has a zero intercept.
This law is only valid at low concentrations and for a single
wavelength. At high concentrations, there is a limiting absorbance
associated with light leakage, detector noise, and bandpass. yet
students valiantly put a straight line through their data, missing
every single data point. <sigh>
There are numerous reasons for a non-zero intercept:
* instruments are also prone to drift, especially cheap ones
* the blank may have a different composition (and absorbance) than
the standards
* the cuvette orientation may change slightly and/or scratch with
ever insertion (we had an instrument that scratched the face if the
cuvette if the cuvette was pulled out at an angle)
As the laboratory coordinator in a previous life, I instructed TAs to
take marks off if it was apparent that the students forced the data
through zero.
Please forward this to your chemist friend.
Dr. Roy Jensen
(==========)-----------------------------------------¤
Lecturer, Chemistry
W5-19, University of Alberta
780.248.1808