When I fit an exponential model, Prism reports the rate constant (k) with both standard error and confidence interval, but reports the half life with only a confidence interval.

When Prism fits a model, it reports both the standard error (SE) and confidence interval (CI) for the parameters it fits (unless the Diagnostics tab is used to not report everything). With some models, it also reports transforms of those best fit values. For example, for exponential fits, it reports the half life as well, which is computed as ln(2)/k. Prism also reports the confidence interval of the half life, but not its standard error.

The confidence interval Prism reports is asymetrical around the rate constant, and corresponds exactly to the confidence interval of the rate constant. Each end of the confidence interval of k is transformed to determine the confidence interval of the half life. Because the confidence interval is asymetrical, there is no way to compute a SE value.

Prism can compute the confidence interval a different way. To do this:

1. Go to the nonlinear regression fit tab, with an exponential equation selected.
   
2. Click the New button and choose to clone the equation.
3. Edit the name of the cloned equation
4. Go to the last tab.
5. Change the method for computing the confidence interval to: symmetrical.
   

Now Prism will report both SE and CI for half life. It computes the SE of the half life just as if it had been the parameter being fit, and compute a symmetrical confidence interval.

Note that the SE really is not a very useful value to think about or report. Its only real use is to compute a confidence interval.