The new Macs released last week use Intel chips (all prior Macs used PowerPC chips). Will Prism 4 Mac run on such Macs? The quick answer is: Yes.

To allow all Mac programs to run, these Macs ship with "Rosetta" which automatically works in the background to allow programs written for PowerPC chips to work on an Intel chip. When I first heard of this, I was dubious that it would actually work. But we tried Prism 4.0c Mac on an early prototype of an Intel-based Mac, and it ran fine. You don't have to do anything special -- just launch Prism as usual and it works. So far, we've only spotted one minor glitch (if you give two-way ANOVA nonsense data with N=0, the new Macs give nonsense results while older Macs give blank results). We'll test more once we have a real Intel Mac (so far, we've only used a prototype made available to developers).

In many experiments, you collect replicate Y values at each X. I've written a new article that explains how to compare the variation among replicates to the scatter of points around a best fit curve. If the scatter around the curve is much higher than predicted from the variation among replicates, you have evidence that the model doesn’t fully describe your data. And you don't need to fit an alternative model in order to gather evidence that the model you are using is inadequate.

This method is nothing new -- at least not in advanced texts of linear regression -- but I am not aware of it being used in nonlinear regression.

I'd appreciate any comments you have on the article and the technique. Thanks.