Various measurement statistics are used to summarize the behavioral observations. Some behavioral safety providers record the actual number of occurrences; some calculate a percentage of safe behaviors among all recorded observations. The occurrence of behaviors also can be recorded as an all-or-nothing measure or as a proportional measure (e.g., percentage of unsecured ladders or fully boarded scaffold platforms). One study found that proportional measures have advantages over all-or-nothing approaches because,among other reasons, they are more sensitive to minor changes in safety performance (Cooper, Phillips, Robertson, & Duff, 1993). Another report showed that an aggregate measure of safety behavior was effective in monitoring performance across several specific housekeeping behaviors (Saari, 1987). Despite some research in this area, the safety literature lacks sufficient information to guide the selection of an appropriate measurement statistic.