Bioequivalence and Bioavailability Forum

Dear Raveendra,

In my personal opinion neither, these "cut-off" values are only a guideline to assess the mathematical goodness of fit, is 0.74999 really 'worse' than 0.75001?

Use them to help you select the points of the apparent elimination phase but remember to also consider the PK interpretation, plausibility, trend across the groups etc.

Simon

Dear Raveendra,

In my opinion you should use adjusted R square. Unlike R square, adjusted R square allows for the degress of freedom associated with the sums of the squares. Therefore, even though the residual sum of squares decreases or remains the same as new explanatory variables are added, the residual variance does not. For this reason, adjusted R square is generally considered to be a more accurate goodness-of-fit measure than R square.

Generally less than 0.75 means bad correlation, therefore data may not be reliable.

Regards,

Atish

Dear Raveendra!

❝ […] what is the minimum R² value […]. some are using 0.8 and some are using 0.75. which one correct?

Neither nor, because R² depends on the number of data points used in the estimation. In linear regression R² 0.9756 (n=3) ‘explains’ the same amount of information as R² 0.5314 (n=6)!
For a rather lengthy discussion see this thread.
If you truly want to set a cut-off, then R²_adj should be used. BTW a cut-off common to all studies is not justified. For drug A a R²_adj of 0.95 may be easy to get, whilst for drug B a R²_adj of even 0.75 very difficult.