The efficiency of a selection step is defined as the fraction of desired events or objects that pass it. For instance, if 60% of genuine electrons in a sample pass the electron identification, the identification efficiency is 60%. In practice, the efficiency that it is feasible to achieve must be balanced against the purity of the sample after the selection: one generally wants high efficiency and purity, but it is often not possible to increase one without decreasing the other. Which balance works best depends on the analysis demands and strategy.
The opposite of efficiency is inefficiency, defined as 1 – efficiency.