Project Management Scheduling and Estimating Techniques This paper compares two scheduling and estimating techniques – PERT and CPM – and argues that PERT is, generally, a better and more structured technique to use for email upgrading at the given organization. Program Evaluation Review Technique (PERT) is an extension of CPM (Critical Path Method), which incorporates variability in activity durations into project network analysis. PERT has been used extensively and successfully in practice. At the same time, precedence relationships in a CPM network fall into the three major categories listed below: 1. Technical precedence 2. Procedural precedence 3. Imposed precedence Technical precedence requirements are caused by the technical relationships among activities in a project. For example, in conventional construction, walls must be erected before the roof can be installed. Procedural precedence requirements are determined by policies and procedures. Such policies and procedures are often subjective, with no concrete justification. Imposed precedence requirements can be classified as resource-imposed, state-imposed, or environment-imposed.

For example, resource shortages may require that one task be before another. The current status of a project (e.g., percent completion) may determine that one activity be performed before another. The environment of a project, for example, the effects of concurrent email systems, may determine the precedence relationships of the activities in a project (Donaldson 1965). In real life, activities are often prone to uncertainties which determine the actual durations of the activities. In CPM, activity durations are assumed to be deterministic. In PERT, the potential uncertainties in activity durations are accounted for by using three time estimates for each activity. The three time estimates represent the spread of the estimated activity duration. The greater the uncertainty of an activity, the wider the range of the estimates (Golenko-Ginzburg 1989a). PERT uses the three time estimates and the simple equations to compute the expected duration and variance for each activity. The PERT formulas are based on a simplification of the expressions for the mean and variance of a beta distribution.

The approximation formula for the mean is a simple weighted average of the three time estimates, with the end points assumed to be equally likely and the mode four times as likely. The approximation formula for PERT is based on the recognition that most of the observations from a distribution will lie within plus or minus three standard deviations, or a spread of six standard deviations. The major steps in PERT analysis are summarized below: 1. Obtain three time estimates a, m, and b for each activity. 2. Compute the expected duration for each activity by using the formula for te. 3. Compute the variance of the duration of each activity from the formula for s2. It should be noted that CPM analysis cannot calculate variance of activity duration, since it uses a single time estimate for each activity. 4. Compute the expected project duration, T,. As in the case of CPM, the duration of a project in PERT analysis is the sum of the durations of the activities on the critical path. 5. Compute the variance of the project duration as the sum of the variances of the activities on the critical path.

The variance of the project duration is denoted by S2. It should be recalled that CPM cannot compute the variance of the project duration, since variances of activity durations are not computed. 6. If there are two or more critical paths in the network, choose the one with the largest variance to determine the project duration and the variance of the project duration. Thus, PERT is pessimistic with respect to the variance of project duration when there are multiple critical paths in the project network. Source: Johnson and Schou (1990) Several researchers including Golenko-Ginzburg (1988, 1989a, 1989b), Donaldson (1965), Johnson and Schou (1990), as well as Littlefield and Randolph (1987) have addressed the deficiencies in the PERT estimation procedure. The pitfall of using estimates furnished by an individual is that they may be inconsistent, since they are limited by the experience and personal bias of the person providing them. Individuals responsible for furnishing time estimates are usually not experts in estimation, and they generally have difficulty in providing accurate PERT time estimates.

Yet, PERT still proves to be a better technique since it incorporates variability in activity durations into project network analysis as opposed to CPM, which only gives limited analysis.