Stochastically Simulating Low Morale - A Risk Factor Effect on Risk Management

: One of the proposed purposes for software process simulation is the management of software development risks, usually discussed within the category of project planning and management. However, modeling and simulation primarily for the purpose of software development risk management has not been explored and is quite limited. This study describes an approach to simulate Low Morale - a risk factor, to analyze its effect on certain software development risk management activities


INTRODUCTION
One of the proposed purposes for software process simulation is the management of software development risks, usually discussed within the category of project planning/management [1] .
However, simulation primarily for the purpose of software development risk management has been quite limited. A notable exception is Madachy's model [2] , designed partially for the purpose of risk assessment. This study describes a different approach to simulation for managing software development risks.

Assessing uncertainty through risk factors:
Uncertainty gives rise to risk, the potential of loss. The categories of (Gemmer, 1997, Table 1) are helpful for thinking about various uncertainties in managing software projects.
One means of managing the risks arising from uncertainty is to characterise risky scenarios and identify the risk factors in those scenarios. Each scenario can then be associated with a probability of occurrence, a potential cost and its value (utility) to the administrators. However, analysis of scenarios begins with identification and quantitative description of the factors composing scenarios. These risk factors can then be arranged in various scenarios and using a vehicle for propagating their uncertainties, can be related to system outcomes.
A simulation model supports risk management to the extent that it supports the generic process.

Software development risk factors (SDRFs):
The software engineering literature on risk has focused on three topics: a. The appropriation of techniques from other disciplines In information Uncertainty about when certain events may occur or the ability to react to them Inadequate authority to make or influence decisions or inconsistency in processes Inadequate or inaccurate information on which decisions are based b. The development of risk management approaches for software development c. The identification of software development risk factors and their relationship to project outcomes. Although, the second topic i.e. risk management, has received the most attention, considerable attention has been and is being given to understanding the sources of risk.
A variety of approaches have been used to investigate SDRFs. From them ,a trend of collections of risk factors have emerged, including prioritized lists, taxonomies, questionnaires and matrices, for assessing software development risks. Some investigators have produced SDRFs lists numbering to the orders of 150 or more factors. Houston [3] found twenty-nine of these factors were cited most often in studies intended to identity the most important SDRFs. The selected SDRFs for simulation in this study are shown in Table 2.
In context of Table 2, this can be stated that the effect of low morale on productivity can vary during a project and it can be continuously recalculated and the distributions for its variables are sampled continuously throughout each run.

Effect of low morale on Efficiency:
Morale is modeled as a variable having an initial level set. This starting level, a value between 1 (lowest morale) and 11 (highest morale), typically reflects good morale, for example 7 (slightly "up") or 8 (fairly satisfied team). The morale level can be affected by high schedule pressure and by attrition. It may, in turn, produce effects on productivity, error generation and attrition.
With regard to the effect of staff morale on productivity, the survey results showed that a. When low morale affected Efficiency, productivity decreased b. The probability of an effect increased from 0.46 to 0.95 as morale decreased from "slightly down "to "lowest: open rebellion".
A typical software development project was considered for simulation experiment. In this case, morale is modeled.

INFORMAL DESCRIPTION OF THE ALGORITHM
An efficiency level of an employee in an organization may be defined as the measure of employee's satisfaction. If the schedule pressure is more, the efficiency will naturally be low. No management would like to have a policy with poor efficiency level of an individual even if it is most economical operating policy. The efficiency here is expressed in percentage. An initial level of "Morale" is set. The "morale level" can be affected by "high schedule pressure" and by "attrition". It may produce effect on efficiency and in turn, on productivity. The length of each simulation run was made NRUNS=500 weeks. We have assumed the initial morale level to be 11 units and no initial "incentive due" at the beginning of the simulation run.  The efficiency for different level of Morale threshold has been worked out and shown below.

Terms used
From the above Table we can analyze that by reduction in morale boosting period the performance of the person improves, i.e. efficiency level but this is at increased average moral level.
Increasing moral-level -addition can increase the Efficiency level. But this also results in increased average moral level.
Again, a program was written to simulate all combinations of Threshold morale from 5 to 25 in steps of 2 and morale addition level (Incentive) from 5 to 15 in steps of 2 .For each of these different combinations ,the average morale level and Efficiency level computed and printed The length of each simulation run was 500 weeks.
From the Table 4, a plot of efficiency level vs average morale level was made. The policies lying on the curve are the best policies. Those lying above are inefficient and ineffective policies, because input in the