Team performance and information system implementation

Abstract

Although the Earned Value Method (EVM), a multi-dimensional project control system, has been widely used since 1960, there has been little research focused specifically on its application in IS implementation. The management of IS projects—including the accurate forecasting of project duration—is complicated by the strong impact of the learning curve on the effectiveness of EVM. This study makes new contributions by advancing formulas for the accurate prediction of project duration, and by developing a decision support model in which the learning curve is fully integrated with EVM, and performance changes due to learning are isolated from other factors affecting project duration. The study makes three contributions to the understanding and use of EVM in IS implementation: (1) It provides an extended version of EVM, (2) it provides an illustration of the application of EVM to a real project situation based on empirical data from an ERP implementation project, and (3) it provides a theoretical basis for empirical studies of IS project control techniques. The decision support model can be used to determine both the learning curve coefficient and the project completion date during the early stages of a project, and it thus offers a significant practical contribution to the management of IS projects.

Appendices

Appendix A

In this Appendix, we derive the equation for the Work Completed Index (WCI), which is used in the case study presented in Section 6. WCI is defined as the ratio of work completed to work planned, where Work Planned = p ₀*T ₀ (Assumption 1). Note that in the case study, the resource base was increased after $t_2^ * = 2\;{\text{months}}$ to compensate for the low Schedule Performance Index (SPI ₂ = 0.8). If RI is the Resource Index and represents the level of change in the resource base, and the work accomplished during the project is equal to the area under the learning curve, then ${\text{Work Completed}} = p_1 * \left[ {\int\limits_0^{t * } {P_L \left( t \right)dt + RI} \int\limits_{t * }^{T_0 } {P_L \left( t \right)dt} } \right]$ and

$$WCI = \frac{{p_1 * \left[ {\int\limits_0^{t * } {P_L \left( t \right)dt + RI\int\limits_{t * }^{T_0 } {P_L \left( t \right)Pdt} } } \right]}}{{p_0 T_0 }}$$

(A.1)

In the case study, k was assessed from a test in which the time required to complete data conversion procedures was measured at two different points during the progression of the project (Appendix B). Consequently, PRI was determined from equation (4.3.3) and SPI ₁ was calculated from equation (4.3.1). Therefore p ₁ can be substituted in equation (A.1) from (4.1.4) to give

$$WCI = \frac{{p_0 * SPI_1 * \left[ {\int\limits_0^{t * } {P\left( t \right)dt + RI} \int\limits_{t * }^{T_0 } {P\left( t \right)Pdt} } \right]}}{{p_0 T_0 }}$$

(A.2)

After substituting the L-curve and rearranging terms, we get equation (A.3), which is used to calculate the ratio of work completed to work planned if the resource base on the project was increased by RI after time t*.

$$WCI = \frac{{SPI_1 }}{{kT_0 }}\left[ {\left( {1 - RI} \right)kt^ * - 1 + RI * kT_0 + \left( {1 - RI} \right)e^{ - kt * } RI * e^{ - kT_0 } } \right]$$

(A.3)

Note that in the case study, the Resource Index is at 1.25 (RI = 1.25), since 25 percent more resources were working on the project for the last 7 months.

Appendix B

In this Appendix, we present a procedure that can be used to assess the Learning Curve Coefficient k. Performance was measured as the time required to develop data conversion procedures at various times during the progression of the project. If CONV_PROC represents the amount of work required for development, $\Delta t_1^ *$ and $\Delta t_2^ *$ are times required to finish the procedures when the first and second measurements were taken (after times $t_1^ *$ and $t_2^ *$ elapsed on the project), and α is the ratio between the productivities of the project team at times $t_1^ *$ and $t_2^ *$, then equation B.1 can be used to estimate the ratio.

$$\alpha = \frac{{P\left( {t_1^ * } \right)}}{{P\left( {t_2^ * } \right)}} = \frac{{\frac{{CONV\_PROC}}{{\Delta t_1^ * }}}}{{\frac{{CONV\_PROC}}{{\Delta t_2^ * }}}} = \frac{{\Delta t_2^ * }}{{\Delta t_1^ * }}$$

(B.1)

Equation (B.1) can be rewritten as (B.2) after substituting the functional form of P _L (t) from equation (4.2.1).

$$\alpha = \frac{{p_1 \left( {1 - e^{ - kt_1^ * } } \right)}}{{p_1 \left( {1 - e^{ - kt_2^ * } } \right)}} = \frac{{1 - e^{ - kt_1^ * } }}{{1 - e^{ - kt_2^ * } }}$$

(B.2)

Equation (B.2) can be then transformed into

$$\alpha e^{ - kt_2^ * } - e^{ - kt_1^ * } + 1 - \alpha = 0$$

(B.3)

In the case study the productivity was measured after 1 month and then after 2 months, so $t_2^ * = 2t_1^ *$ and (B.3) can be transformed further into $\alpha e^{ - 2kt_1^ * } - e^{ - kt_1^ * } + 1 - \alpha = 0$, which can be solved to give

$$k = \frac{{{\text{In}}\left( {\frac{{1 \pm \sqrt {1 - 4\alpha \left( {1 - \alpha } \right)} }}{{2\alpha }}} \right)}}{{t_1^ * }}$$

(B.4)

In the case study the first data conversion procedure was completed in 10 days ($t_1^ * = 1$, $\Delta t_1^ * = 10$). A similar procedure for another set of accounts was completed in less than 8 days ($t_2^ * = 2$, $\Delta t_2^ * = \left[ {7.3,\;7.5} \right]$).⁷ Substituting the data into (B.4) gives the range for k = [0.6, 0.7].