W3_Stephon_Statistical Assessment of Bid Selection Process

1.    Problem Definition

The offshore construction section of an IOC in Nigeria manages a portfolio of “Minor Plant Modification” projects which are executed by “Local EPCI” contractors as a policy. The number of change orders as well as claims on these projects is unacceptable by management and the root cause(s) is/are to be investigated.

In W2 posting, we were able to attribute this challenge partly to inadequate scope definitions in “requests for proposals” developed in the division. For this week’s posting (W3) we will try to confirm if the contractor selection process contributes to this challenge. The assumption is that the selection process considers predominantly, or only the commercial bids, using the “lowest bidder” criterion which results in the selection of contractors with a potential of running into financial problems as the project progresses. Hence the claims and change orders.

 

2.    Feasible Alternatives

For this assessment, still based on the initial sets of bids for sixteen different projects, we will compare the company estimates with the selected contractor bids using the following statistical analyses:

1.     Find the coefficient of correlation between the company estimates and the winning bids

 

Given by:

 

 

2.     Simple Linear Regression Analysis of company estimates and selected bids.

 

Given by:

 

 

            Where

 

               And

 

 

 

3.     Find the Coefficient of Determination (R²) of the regression model developed to determine how well it describes the relationship between the estimates and the winning bids

 

Given by

 

          4.     Plot the a Scatter Diagram of the difference between the company estimate and the selected contractor bids

 
 3.    Develop the Outcomes for each Alternative

         The table below shows the company estimates and winning bids (adjusted by a factor) as well as the statistics calculated from them.

 

 

Fig 1:  Table of variables and computations

From the table above we calculate the following:

1.     r = 143,914685,467.77/ (SQRT134,397,903,535.55 * SQRT 169,077,491,094.27) = 0.95
 

2.      m =  143,914685,467.77/134,397,903,535.55 = 1.07

       c =  181,122.98 – 1.07*189,879.52 = -22202.01

 

  therefore

       ŷ = 1.07*x – 22,202.01 (this has been calculated for all values of x in the table)

 

3.     R² = 154,105,355,429.21/169,077,491,094.27 = 0.91

 

 

 

 

Fig 2:  Scatter Diagram of Company Estimates vs Winning/Selected Bids with Regression Line

   

 

Fig 3: Scatter Diagram of Company Estimates minus Winning/Selected Bids

 

 

 

4.    Acceptable Criteria

 We will accept the assumption that the winning bids are being selected based on just the lowest bidder criterion with little or no technical competence consideration if there is not enough evidence of correlation between the company estimates and the winning bids. Acceptance and rejection will be based on the following limits.

 

1.)   r  >= |0.85| :reject assumption, otherwise accept

2.)   R² >= |0.85|: reject assumption, otherwise accept.

3.)   If charts do not indicate adequate relationship between the two variables, accept assumption otherwise reject.

 

 5.    Analysis and Comparison of the Alternatives

Our calculated parameters are as follows:

 r = 0.95: therefore we reject the assumption

R² = 0.91: therefore we reject the assumption

 

The chart of differences between the variables shows that we have an equal number of points plotted above and below 0. Based on this, we also reject the assumption

 

 
6.    Select the Preferred Alternative

The Coefficient of correlation (r) obtained indicates a strong relationship between the two variables. The Coefficient of determination (R2) obtained indicates that the linear regression model developed describes that data adequately. These figures as well as the equal distribution of the differences above and below 0  implies that the lowest bidder criterion was not the only bases for selecting bids as the selected bids exhibit the same characteristics as the company estimates. So far, we have been able to identify only insufficient information in the request for proposals as a root cause of the unacceptable number of change orders and claims in the division’s minor projects.

 

 

7.    Performance Monitoring & Post Evaluation of Result

Further studies will be carried out on the contractor bids

 

References:

1.    Kvanli, A. H., Pavur, R. J. & Guynes C. S. (2000).  Introduction to Business Statistics. Chapter 14

2.    Amos, S.J. (2007). Skills & Knowledge of Cost Engineering Fifth Edition. Chapter 9

3.   R-Squared. Retrieved from http://www.hedgefund-index.com/d_rsquared.asp