The routine practice of hard data production is described in two human service institutions. Three modes of production are delineated: Direct counts of behavior, secondary production, and surveys of those in-the-know. Practical production is contrasted with the formal image of hard data held and used by staff members and regulating agents. The use of circumstantial rules and the use of glossing practices are shown to be two means by which hard data become available. The practical reality of hard data is discussed in relation to staff and outsider awareness of hard data as a matter of production.
