Over the past quarter century, sophisticated statistical techniques have been playing an increasing role in the analysis of particle-physics experiments. For one thing, the experiments have become much more elaborate and difficult, typically producing enormous volumes of data in search of very small effects.
In the decades after World War II, discoveries of new particles—for example, many of the early strange-quark-bearing mesons and hyperons—were often based on one or a few bubble chamber photographs. In the oft-told case of the J/ψ meson, the first known particle harboring charmed quarks, the cross section for its formation in an electron–positron collider rose so dramatically at the resonant energy that the discovery was clear within hours of the first hint of a signal.
Such discoveries were deemed obvious; there was no need to calculate the probability that statistical fluctuations had produced a spurious effect. Contrast that with today’s search for the Higgs...
