—=—
Ordinarily when we do a
206
regression
of this kind, we will call one variable the
Andependent variable and the other the depend
ent variable,
In other words, we are supposed to
have one variable which is exect.
It 1s not
subject
to some variability and the other
one, we state this in terms of this (indicating).
In this case, both are subject to
sampling error and so we use what we call the
third regression line;
You have a mean value for the acti-
vity collected on the gummed paper and the
mean value of the sample collected in the
roof pots and each one of the three regression
lines will pass through that point, one in
which we consider the roof pot as exact would
be the one in which we minimize the vertical
deviation from that line, another in which we
consider the gummed paper as being exact and
that we will be minimizing the horizontal
deviations and a third in which we consider
both as being variabie ig one which is true
in this case and that falla in between,
I have drawn an idealized case here,
Ore in which the variability is the same in the
—_
WED
poe ARCH