| Question |
Discussion |
| Did you pick a sensible model? |
Nonlinear regression adjusts the variables in the equation (model) for which you chose to minimize the sum-of-squares. It does not attempt to find a better equation. |
| Have you set appropriate parameters to constant values? |
Prism doesn't have to fit all the parameters in a model. In many cases, it is appropriate to set one or more parameters to constant values. See Fixing parameters to constant values . |
| Is the scatter Gaussian? |
Nonlinear regression assumes that the scatter is Gaussian. To test this assumption, see Have you made a common error when using nonlinear regression? |
| Is the scatter consistent? |
Nonlinear regression assumes that the scatter of points around the best-fit curve has the same standard deviation all along the curve. The assumption is violated if the points with higher (or lower) Y values also tend to be further from the best-fit curve. The assumption that the standard deviation is the same everywhere is termed homoscedasticity. |
| Is variability only in the Y direction? |
The nonlinear regression model assumes that X values are exactly correct, and that experimental error or biological variability only affects the Y values. This is rarely the case, but it is sufficient to assume that any imprecision in measuring X is very small compared to the variability in Y. |
| Does each data point contributes independent information |
The deviation of each value from the curve should be random, and should not be correlated with the deviation of the previous or next point. If there is any carryover from one sample to the next, this assumption will be violated. You will also violate this assumption if you tell Prism to treat each replicate as a separate point, when they are not independent. |
