Very good, Donald. I appreciate the help dusting off the cobwebs...
although... I'm not sure I ever did know what n before the sigma was.

Also enjoyed your comment, "no need for capital C's, it's not a title of
nobility".


"Donald Burrill" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]
> On Sat, 13 Mar 2004, John Gregory wrote:
>
> > In Pearson's Correlation Coefficient formula, the Greek symbol for Sum
> > (backward E) has a small "n" preceeding it at half height.
>
> The mathematical symbol for "Sum" is the Greek letter sigma (a capital
> sigma), corresponding to the capital S in the word Sum (which is why it
> was chosen in the first place;  the corresponding symbol for "Product"
> in multiplication is a capital pi).  If there is a lower-case "n"
> preceding it, the value of "n" is to be multiplied by the sum, once the
> sum has been obtained.
>
> The standard methods of indicating the range of summation, which is what
> you described subsequently, are:
>  (1) when each individual value in the formula is indicated or implied
> by a subscript (say "i"), a line of small type under the capital Sigma
> reads "i = 1" and above the capital Sigma "n";  this would be read "Sum
> <the following quantities> for i = 1 to n".  This usage permits one to
> indicate partial summations (e.g., from 7 to 31 instead of from 1 to n).
>  (2) when the particular quantities to be summed are obvious from the
> context (e.g., for all the data in hand), the capital Sigma may be
> unadorned, the implication being "for all the data" or "from 1 to n".
>  (3) sometimes, as shorthand for (1) above, a subscript "i" attached to
> the Sigma.  More commonly used when the items to be summed are indexed
> in two dimensions (say "i" and "j") but the summing is to be carried out
> only with respect to one of them.
>
> > I think it's to be read something to the effect "Sum the following and
> > perform the other operations in sequence for each set of varible in
> > the list you're working from; or in other words... row by row from the
> > 1st to as many as there are (this "n")."
>
> Close.  Depends on whether "the other operations in sequence" are
> intended to be performed before summing or afterward.  This is made
> explicit by parentheses in the formula.  In particular, in one of the
> numerous formulas for the Pearson correlation coefficient (no need for
> capital C's, it's not a title of nobility) you may have an expression
> like
>   SUM((X_i - Xbar)(Y_i - Ybar))
>  (using "_i" for the subscript "i", and "Xbar" for the sample mean of X)
> in which case the algorithm implied is
>  subtract the mean of X from the current value of X;
>  subtract the mean of Y from the current value of Y;
>  multiply these two differences together;
>  and sum all such products.  Or, if you have a formula part of which
> reads
>  n SUM XY - (SUM X)(SUM Y)
>  the algorithm implied is
>  add up all the X values (= SUM X);
>  add up all the Y values (= SUM Y);
>  count the number of values (= n);
>  for each case, multiply X by Y (= XY) and sum all these products;
>  multiply this last by "n";
>  subtract from this the product of the first two sums.
>  (This quantity will be n times as large as the result of the previous
> formula, by the way.)
>
> > Is that how I'd express this verbally? Been a long, long time since
> > I've had to deal with the formula.
>
>  ------------------------------------------------------------
>  Donald F. Burrill                              [EMAIL PROTECTED]
>  56 Sebbins Pond Drive, Bedford, NH 03110      (603) 626-0816
> .
> .
> =================================================================
> Instructions for joining and leaving this list, remarks about the
> problem of INAPPROPRIATE MESSAGES, and archives are available at:
> .                  http://jse.stat.ncsu.edu/                    .
> =================================================================


.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to