To make sure she's making apple-to-apple comparisons among felons, Iyengar compares the actions of criminals whose rap sheets are identical except for the order in which they committed their crimes. Since you don't start counting strikes until the first record-activating offense, order is crucial. For example, someone who is convicted for armed robbery—a record-activating offense—followed by shoplifting will have strike three looming; if he had reversed the order of his offenses, he'd have only one strike. So if the two-striker commits fewer crimes (but more violent ones) relative to the one-striker, we know it's the effect of the three-strikes incentives and not something about the different offense records of the two criminals.