Background and Research Biography:

Centex Homes - Large Corporation working on large projects. Centex Homes, La Sera Home Development site, San Ramon CA. Lead Field Manager - Steve Dawe

Centex Homes was used as the large corporation. Steve Dawe, Lead Field Manager, kept constant track of a randomly selected subcontractor over the period of 6 weeks. This analysis consisted of time quotations for an entire production team rather than an individual tasks. The employee's on the team were consistently the same day to day and the tasks were unique and varied from day to day.

TechXpress - Small Company working on small tasks. TechXpress, Web Department, San Luis Obispo, CA. COO - Alex Minicuci.

The Website Development department was used to reference individual tasks. Programmer's time was used and individual tasks that never repeated were analyzed. In each case the programmer gave an estimation of time and then once the task is complete the project manager recorded the actual time completed.

These two datasets were gathered in a way such that the teams/individuals were not aware they were being studied. This made sure that the data collected is legitimate and can be considered pure.

Project Management:

A project manager is responsible for successful completion of any given project and that projects inherent quality and delivery time. For the purposes of this research paper the definition of a project will be any scale of work that an individual who is an expert in their field is required to do and such work is required to be complete before another person can continue with their work. Therefore, projects include work, that multiple humans do, that the tasks involved in the work build on one another. These tasks are known as predecessors. The main goal of a project manager is to deliver a high quality product in the shortest time possible. Therefore, it is important for a project manager to develop efficiencies into the production schedule.

A project manager will be able to analyze the all the tasks that are required, all the resources required, and all the pertinent predecessors to establish a production schedule. This schedule is often based on the experts time estimations on how long a set task will take. Some times these estimations are based on historical data if the tasks are repetitive and have little variation which can lead to tighter and smaller ranged estimations. Other times the tasks are never repetitive and have much variation and thus require broader range estimations. With this in mind the project manager must build a production schedule based on one of those factors.

This research paper focuses on the latter of these prediction methods. Each of the tasks recorded were tasks that the expert gave a time estimation for having never done anything like that task before. In the case of Elite Tile each house has unique plans, special order tile, and options based on customers preferences. In the case of TechXpress each of the tasks recorded was a bug found in programming. This is a unique task that never repeats itself and requires an intuitive sense of problem solving and immense knowledge of the industry standards of programming and methodologies to complete the tasks. In each case a project manager requested an initial time estimation from the expert, assigned time in the production schedule, and then recorded the actual time to completion. This is the general method of workflow for standard, human based work within a project setting.

In general, a project manager is encouraged to complete the project in the shortest time possible without sacrificing quality while still keeping on or below budget. However, if a project manager builds a schedule that is too tight and some of the tasks go over their time estimations the overall project timeline gets pushed back. This is an all too frequent issue with major projects as there are too many external variables that can affect the expert's abilities to complete their tasks. Sometimes these variables can be controlled; sometimes these variables are in the external environment and are out of the project managers or the expert's control. It is also not to say that the expert cannot finish the task before the estimated time. This can also create issues as now resources and other experts are not being used most efficiently. To combat this effect most project managers use what is known as buffer time. Buffer time is essentially an estimated amount of time built into each task so as to build in enough time so when any given task goes over its estimated time the overall project timeline is not affected. This method works well but is often done arbitrarily and at the project manager's discretion. This is not a poor method of dealing with these issues but it is not the most efficient. The amount of time between the actual completion time and the estimated completion time is known as Failure Time. The effects of buffer time and failure time are illustrated in Figure 1.1 below.

With this concept in mind, Failure Time can be an added value method of estimating the exact buffer time needed for an individual expert. To estimate Failure Time the project manager must have historical data to reference for each individual expert. This also works for a team of experts that continually operate together (e.g. Elite Tile). The project manager takes the difference between the estimated time and the actual time to delivery of each task and records this data. Then the difference is divided by the estimated time to develop a ratio as a percent of difference to estimated time. This ratio as a percent is then run through a regression analysis which then gives a 99% cortile range of a more accurate failure time estimation.

Figure 1.1 shows how estimating buffer time in an arbitrary way gives the inevitable project completion time +30 or -30 minutes in project 2 and 3 respectively. This is assuming that no one in the project fails to hit the estimated time. Estimating this Buffer Time on Project 2 gives a total of 12.5% overage. This example is only a ½ hour over the initial project estimated time which is small when considered to a big picture. However, if this trend continued for a project that was estimated at 1920 hours (1 work year) the inefficiencies would be 240 hours or 6 weeks. There are not many organizations that can afford to have 6 weeks of wasted time on the books per year. Great value can be added to an organizations production schedule if historical data can be analyzed and more accurate Buffer Time can be estimated.

Using simple statically analysis a project manager can make a more accurate estimation of failure time and thus build a more accurate production schedule. The following figures show actual data taken from real projects as recorded by project managers. The statistical analysis of this data follows the datasheets.

Statically Analysis of TechXpress:

What we can take from this analysis is that the project manager can make a 99% accurate estimate that any given task estimated will have an actual completion time between 42.623% below and 83.136% above that of the estimated time. Considering the data, if the project manager wants to make absolute certain that the tasks will be completed the project manager will need to schedule 133% above the estimated time to accommodate the highest overage of Task 19. Considering this, the project manager can estimate Buffer Time at a more accurate level and eliminate the wasted time inefficiencies.

Statistical Analysis of Elite Tile:

What we can take from this analysis is that the project manager can make a 99% accurate estimate that any given task estimated will have an actual completion time between 11.3886% below and 23.0380% above that of the estimated time. Considering the data, if the project manager wants to make absolute certain that the tasks will be completed the project manager will need to schedule 33% above the estimated time to accommodate the highest overage of Task 21 and Task 11. Considering this, the project manager can estimate Buffer Time at a more accurate level and eliminate the wasted time inefficiencies.

Conclusion:

Considering the statistical analysis done with real world data Failure Time is an element that can easily be implemented into the daily scheduling of a project managers job. With this information project managers can eliminate some time inefficiencies by estimating the needed buffer time better. While there will never be an exact and 100% accurate way of estimating Buffer Time there is an effective way. With the evidence provided the thesis statement is proven correct.