RSQ function gives incorrect result

[imurch]

The RSQ function in my version of Excel (the 'About Excel' tab says it's
Version 11.2 (050714)) appears not to work properly.

For a set of numbers in cells I3:J9, I entered the following into cell J18:
=RSQ(J3:J9,I3:I9)
and got a result of -0.8804265 which is clearly wrong since R^2 cannot be
negative.
Interestingly =CORREL(J3:J9,I3:I9) gives exactly the same result, implying
that the implementation of this function lacks the step of squaring the
result!! (Minor oversight!!)
This is in agreement with the fact that graphing these points and displaying
the R^2 value on the chart gives a value of 0.7752, which is equal to
(-0.8804265)^2.

This is especially amusing since according to Microsoft Support articles,
the RSQ function has been 'fixed' in Versions 2003 and 2004 so that it does
not give negative numbers!

[Jerry W. Lewis]

Please post your data.

Jerry

imurch wrote:

> The RSQ function in my version of Excel (the 'About Excel' tab says it's
> Version 11.2 (050714)) appears not to work properly.
>
> For a set of numbers in cells I3:J9, I entered the following into cell J18:
> =RSQ(J3:J9,I3:I9)
> and got a result of -0.8804265 which is clearly wrong since R^2 cannot be
> negative.
> Interestingly =CORREL(J3:J9,I3:I9) gives exactly the same result, implying
> that the implementation of this function lacks the step of squaring the
> result!! (Minor oversight!!)
> This is in agreement with the fact that graphing these points and displaying
> the R^2 value on the chart gives a value of 0.7752, which is equal to
> (-0.8804265)^2.
>
> This is especially amusing since according to Microsoft Support articles,
> the RSQ function has been 'fixed' in Versions 2003 and 2004 so that it does
> not give negative numbers!