Six Sigma SPC OEM Products
Six Sigma and SPC (Statistical Process Control) Software
Applications
Six Sigma SPC - 2070 W. Washington St. #5
Springfield IL 62702 Ph: 217.698.0063
6 Sigma Statistical Process Control (SPC) Software for Windows
95/98/NT/2000/XP/ME

Graph used to show relationship or correlation
between two factors or variables. i.e. X-Y or XY or X Y chart. (can
be used in conjunction with correlation
coefficient below)

Each factor or variable should be drawn
on a graph, (a piece of grid paper would work), with X plotted
on the vertical axis and Y plotted on the horizontal axis, or
visa-versa. Each factor or variable should be graphed in such a way
that they intersect both axis making one point for each of the X-Y
paired numbers.

Be careful when using the correlation
coefficient. One should think about the following rules.

The result of the calculation, (r),
is a measure of the linear association between X and Y as it
exists in any given set of number pairs. Even when there is high
value of (r), it does not necessarily imply that X and Y are
related in any way as a cause and effect. It is not unheard of to
obtain a high degree of statistical association with no causal
connection whatsoever. It is also not impossible that two
variables which are not related may both be related to a third
variable. This could cause the first two variables to show a
statistical association when there are not is necessarily any real
cause and effect.

You should make sure that there are sound engineering reasons for
the relations before trying to draw any deductions from a
calculated (r).

Whereas unassociated pairs of data
could have zero correlation the average, specific pairs may have correlation
as a consequence of samplingvariation,
and thus will have values of (r) above or below zero. Due to the
fact that (r) is sometimes above or below zero does not
automatically point to two sets of numbers being related.

Do not draw any conclusions based
on the correlation coefficient until you have tested it for
significance as follows…

Multiply r by the square root of
n, where n is the number of pairs used to determine (r). We
will call this (t).

If (t) is larger than 3, deem
the correlation is significant. If (t) is smaller than 3, the
correlation may not be significant. The smaller the value of
(t), the lesser the probability that the correlation will be
significant. If the correlation is not absolutely significant,
as prescribed above, this could be due to…

(a) A genuine lack of
correlation

(b) Inadequate data. If you
should have reason to think that a correlation should
exist you may wish to acquire more data pairs, calculate a
new value of (r), and retest the new values for
significance.

Not all can see the relation between
the scatter plot and the correlation coefficient, but it can aid in
evaluations of a process.