Presents a quantitative and statistically validated study of employee withdrawal (turnover, absenteeism, etc.) and termination processes. Reliability theory, simultaneous equation (econometric) and other multivariate modeling techniques are joined to study these problems and how they are affected by other variables such as corporate policies, supervisory interaction, organizational climate, and individual or demographic attributes. Data from a sample of ten Pittsburgh firms in the NAB-JOBS program are used for empirical tests. Although this program was formed to find, hire, and train unemployed members of the urban poor and near poor--the secondary labor force--the study was not confined to this part of the labor market. Comparable behavior by members of the firms' regular or primary labor force is also studied in both the transient and steady-state modes. 33 pp. Ref.