When you perform linear regression in Excel, you have the option of adding a trendline to the chart. You can also add the linear equation and r square.
Reclaculation of the r square amount via the RSQ function yields the same number as displayed on the graph.
The same options are available for curvilinear regression (i.e. exponential, logrythmic, power, etc.).
However, the displayed r square value cannot be calculated via the RSQ function for the curvilinear regression. In fact, r square analysis is only applicable to linear regression.
My question is this: How is Excel calculating the r square value (on the trendline graph) for the curvilinear data and what does this r square value represent?
There are various methodologies for computing the equivalent to an r square value, i.e. correlation ratio or ETA coefficient. However, these are not an r square computation.
Reclaculation of the r square amount via the RSQ function yields the same number as displayed on the graph.
The same options are available for curvilinear regression (i.e. exponential, logrythmic, power, etc.).
However, the displayed r square value cannot be calculated via the RSQ function for the curvilinear regression. In fact, r square analysis is only applicable to linear regression.
My question is this: How is Excel calculating the r square value (on the trendline graph) for the curvilinear data and what does this r square value represent?
There are various methodologies for computing the equivalent to an r square value, i.e. correlation ratio or ETA coefficient. However, these are not an r square computation.