Table 5.

A comparison of the goodness of fit (R²) values from logistic regression analyses (GraphPad Prism 5.0 software) obtained for the “best-fit” model versus the R² values obtained when the data were constrained to the dose addition (DA) and response addition (RA) model parameters. If either the DA or RA model perfectly predicted the effects seen with the mixture then the R² value would be equal to the R² value from the “best-fit” model when the observed data were fit to a logistic regression model using DA or RA model parameters. The difference among “best-fit” R² values and DA and RA R² values indicates the lack of fit of the observed effects to the DA and RA predictions. In all cases shown in the far right column of the table, DA R² values were closer than RA R² values to the actual effects of the mixture. In some cases, the DA model was a considerably better model than was the RA model. For other endpoints like LABC and epididymal weights (data not shown), the DA and RA R² values are similar, as they are for seminal vesicle weight.

Endpoint	R2 “goodness of fit” of the observed data to “best-fit”, DA and RA models
	Logistic model			R2
	Best-fit	DA	RA	DA > RA
Hypospadias	0.94	0.85	0.25	60%
Number of permanent nipples	0.96	0.89	0.65	24%
Seminal vesicle weight	0.89	0.83	0.65	18%
Ventral prostate weight	0.80	0.75	0.63	12%
AGD AT day 3	0.89	0.87	0.85	2%
% of 12 nipples on PND 13	0.89	0.87	0.85	2%