An Experimental Study on the Dynamic Reconfiguration

of Software Projects

Maurício Covolan Rosito

, Marcelo Blois Ribeiro

and Ricardo Melo Bastos

Pontifícia Universidade do Rio Grande do Sul (PUCRS), Porto Alegre, RS, Brazil

GE Global Research, Porto Alegre, RS, Brazil

Keywords: Dynamic Reconfiguration, Empirical Evaluation, Organizational Workflows, Software Project

Management.

Abstract: Developing a software product is a complex activity that involves many uncertainties. Software projects

usually experience many modifications during their execution phases. These adjustments can be understood

as reconfigurations in the schedule, in the resources allocation and other design elements. The large amount

of information that the project manager must deal, combined with the frequent changes in the scope and

planning, makes this activity more challenging. In addition, the manager may need to consult other

departments in the organization during the execution of a software project. The distinction between the

specific activities in a project with activities that take part in the organization’s common activity flow can be

observed. In order to contribute to the solution of the noted difficulties, it was proposed a computational

model called SPIM. In this sense, this article presents the results of an experimental study related to

dynamic reconfiguration of software projects, with emphasis on the integration of project management with

organizational flows. A software tool was built to demonstrate and evaluate the results.

1 INTRODUCTION

Software development requires planning and

execution of activities, in which it is necessary to

deal with both technical and managerial issues. The

growing complexity and volume of projects that a

project manager must deal simultaneously

contributes to the increasing challenges related to the

development of projects (Kerzner, 2000); (Pressman,

2009). Particularly in software projects, aspects such

as uncertainties in the specification of requirements

and their instability throughout the project

development, in the use of applied technology and

human nature itself potentiate these difficulties.

During the planning and execution of software

projects, different types of tasks are assigned to

resources with different characteristics in order to

reach the goals related to time and costs of these

projects. In response to new information or

estimations, managers may need to make changes to

the project plan, such as reallocating resources or

canceling tasks (Joslin and Poole, 2005). These

adjustments, required for the project according to

over time changes, give rise to the term ‘project

reconfiguration’. More recently, research in this area

has addressed this problem from a dynamic

perspective, through which the projects adapt

themselves during their implementation. Such

changes often determine impact on costs and

previously established deadlines of the project.

In the same scenario, the manager may need to

interact with other departments of the organization

during the execution of software projects in order to

obtain relevant information to a specific project (for

example, contact the finance department). Thus, the

distinction between the specific activities in a project

with activities that take part in the organization’s

common activity flow (here called organizational

flow) can be observed. Therefore, the project

manager needs some kind of support to help in the

process of decision making taking into account the

integration of these different streams of activities

during the simultaneous execution of projects.

In order to contribute to the solution of the noted

difficulties, it was earlier proposed a computational

model called Software Planning Integrated Model

(SPIM). The SPIM allow supporting dynamic

reconfiguration of software projects considering the

planning and replanning of their activities. To

evaluate the model and embodiment of the proposal,

232

Covolan Rosito M., Blois Ribeiro M. and Melo Bastos R..

An Experimental Study on the Dynamic Reconﬁguration of Software Projects.

DOI: 10.5220/0004442902320239

In Proceedings of the 15th International Conference on Enterprise Information Systems (ICEIS-2013), pages 232-239

ISBN: 978-989-8565-60-0

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

a software tool was developed and used in an

experimental study.

2 PROJECT MANAGEMENT

The development of a software product requires an

effort that involves dealing with, among other

things, activities and resources to produce the

desired results (Schwalbe, 2010). Generally, a

project is intended to achieve a specific result and

involves the coordinated implementation of

interrelated activities. More than that, projects are

planned, executed and controlled by individuals, and

are restricted by limited resources.

Software companies often make use of project

management knowledge in order to build their

solutions with quality and within scope, time and

resource constraints. Project management practices

are responsible for monitoring the achievement of

project goals through the application of a group of

techniques and tools. Thus, project managers need

some kind of decision making support, usually based

on a project management methodology, to deal with

different responsibilities, tasks and project variables.

During a project’s lifetime, actual data, such as

the time or resources that were spent to perform a

particular task, are collected and entered by the

project manager. The manager usually creates a

project plan to specify and limit the scope of the

project describing the work breakdown structure

(WBS) and the project schedule. When creating a

project schedule, the manager begins with a set of

tasks in the WBS (Pressman, 2009). Then, he

specifies all project-related information, such as the

individual tasks, the execution’s sequence of these

tasks and the resources to perform these tasks.

However, the manager may not have all relevant

information up front, forcing him to interact with

other departments in the organization (such as the

human resources department). Hence, the flow of

activities in an individual project is usually related to

other common activity flows of the organization.

Both types of flows are executed in parallel, have

their own resources and may influence the timing of

activities and costs of software design (see Fig. 1).

It can found potential dependency relations

between the activities in both workflows. As an

example, the activity of developing a web site

(which fits in the project’s workflow) may depend

on the hiring of staff by the responsible department

(this activity fits in the shared workflows of the

company). Consequently, it was observed the need

for a decision support solution to anticipate these

requirements related to the other departments of the

company during the execution of software projects.

3 SPIM: AN INTEGRATED

MODEL

The SPIM was first developed considering the

integration of project management concepts

provided in Project Management Book of

Knowledge Guide (PMBOK) (Project Management

Institute, 2008) with the concepts of software

development provided in Rational Unified Process

(RUP) (Kruchten, 2000) and in Object-oriented

Process, Environment and Notation (OPEN)

(Graham et al., 1997). Details about these

integrations models can be seen in Callegari et al

(2008) and Rosito et al (2012). The detailed study of

the PMBOK, RUP and OPEN metamodels helped to

identify how their classes are organized and which

are the valid relations between the elements of each

model. It allowed the development of a methodology

for integrating models of project management with

models for software development processes.

A first experimental study using SPIM was

conducted by six undergraduate and four

postgraduate students of computer science (see

Rosito and Bastos, 2012). This experiment reveals

that the use of the SPIM model approach help

managers to create and conduct a more precise

project plan than the traditional method. However,

some limitations in this first experiment were

observed (such as the use of graduate students) and

advances in research are described below.

Process Engineering Meta-Model Specification

(SPEM) (Object Management Group, 2011), is

considered the reference metamodel for defining

software processes developed by the Object

Management Group. Thus, this research has

advanced to creating the metamodel PMBOK+SPEM.

In the PMBOK+SPEM metamodel (see Fig. 2),

the Organization class represents a company that is

organized by programs. The organizations, usually,

divide projects in several phases aiming a better

managerial control. Also, a necessary resource for

the project, such as people or equipment, is

represented by the Resource class. These resources

are divided into active resources (Stakeholder class)

and non-active (PhysicalResource class). Then, the

ActivityPhysicalResourceWork class associates

physical resources to activities. It establishes the

physical resources work load in that activity.

The PMBOK+SPEM metamodel was designed

AnExperimentalStudyontheDynamicReconfigurationofSoftwareProjects

233

Figure 1: Interdependency between an organizational workflow and a software project workflow.

considering the need of project managers to access

information from other departments of the

organization during the software project planning.

Then, this metamodel defines three different types of

activities: (a) productive activities: activities directly

related to the construction of the software product;

(b) managerial activities: activities that are only

required to coordinate the construction of the

software product; and (c) management supporting

activities: any other activities that do not belong to

an individual project’s activity workflow.

Stakeholders can play several roles during the

execution of project activities. Thus, for each

association between a role and activity

(ActivityStakeholderWork class) there must be an

association of this activity with a stakeholder able to

play that role. Then, managerial activities are

performed by managerial roles and productive

activities are performed by productive roles.

According to the SPEM metamodel, an activity

supports the nesting and logical grouping of

elements contained in the WBS (BreakdownElement

class). The self-defined relationship to the Activity

element allows reuse of content set to an activity in

another activity. Thus, it becomes possible to inherit

a structure defined for an activity in terms of its

elements nested in a second activity. The

relationship between the Activity class and the

ProcessParameter class establishes input and/or

output parameters to the activities in terms of work

products. The ProcessPerformer class establishes the

relationship between the activities and the roles in

the project. The ProcessResponsabilityAssignment

class establishes the responsibility relationship

between the roles and work products.

The PMBOK+SPEM metamodel helped to

provide the conceptual framework necessary to

develop a unique model, called SPIM, to assist in

project planning considering the concepts arising

from the software development processes. Also, it

was developed a tool called Software Planning

Integrated Tool (SPIT). SPIT aims to offer some

kind of support to help managers in the decision

making process of software projects through the

concepts proposed by the SPIM. In this experiment,

participants had access to the following modules:

Validator, BackOffice and Workflow Integrator.

The SPIM Validator acts as an add-in for

Microsoft Project and performs the SPIM validation

rules on software projects. The BackOffice is

responsible for managing the information required

by the SPIM, such as roles definition, types of

activities and associated work products. This

information is exported to Microsoft Project through

custom field to be used by the SPIM Validator. The

Workflow Integrator module is responsible for

synchronizing the information contained in

organizational workflows with those present in a

specific project. Currently, the organizational

workflows were developed through the Visual

Studio Workflow Designer tool.

4 EXPERIMENTAL STUDY

To perform the evaluation of models and products

where the human factor is considered, the literature

provides some approaches based on an experimental

strategy. Pfleeger and Atlee (2009) suggest the

following approaches to evaluate processes,

products and resources: feature analysis; case

studies, surveys, and experiments. Experiments

represent a more controlled type of study, usually

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

234

Figure 2: Part of the PMBOK+SPEM metamodel (more details not shown due to space reasons).

conducted in laboratories. In this research, the use of

a formal experiment method was chosen. However,

as the experiment has a quantitative approach

(Wohlin et al. 2000), an integrated survey was also

used in order to evaluate qualitative data. The

proposals of Juristo and Moreno (2003), Field

(2005) and Wohlin et al. (2000) were used as guides

to conduct this experiment.

4.1 Objective Definition

The Goal-Question-Metric technique (GQM)

(Solingen and Berghout, 1999) was used to define

the study, establishing the overall goal, the

objectives and the measurement. It was decided that

the purpose of this research is to compare, in the

Unified Process, the accuracy and the effort of

integrated planning model SPIM compared to the

traditional model of software project planning.

4.2 Design

In this stage the researchers should formalize the

hypotheses, determine the independent and

dependent variables, selection of participants,

preparation of the experiment and the conceptual

consideration of the validity of the experiment.

Then, these researchers selected an ‘in-vitro’ and

‘offline’ approach in which participants performed

the experiment in a controlled environment. To

conduct this experiment, the context involving

students of two distinct universities was chosen.

This approach can reduce risks and costs not

covered by the scope of the research at this time.

Thus, the experiment was conducted by thirty six

students of post graduation courses in Project

Management. After that, based on the previous

informal definition of the two issues in this research,

it was possible to formalize the two hypotheses and

a definition of its measures for evaluation.

The first hypothesis is related to the effort of

managers in planning activities and resources for

software projects. Then, the first null hypothesis

) is as follows: the effort involved in planning the

activities of the software project using the SPIM

integrated model is equal to the effort to do the

planning of activities according to the traditional

model. The effort should be measured by time spent

AnExperimentalStudyontheDynamicReconfigurationofSoftwareProjects

235

in minutes with the planning of activities for

software development projects in each approach.

The second hypothesis of this research is related

to the accuracy of managers to plan the activities and

resources in software projects. So, the second null

hypothesis (H

) is as follows: the accuracy in the

schedule planning of projects regarding to the

assignment of deadlines and resources considering

the integration with the organizational flows through

the SPIM integrated model is equal to the accuracy

accomplishing the planning in the traditional model.

The accuracy should be evaluated by the ratio of the

participants’ score and the total score possible.

Considering that experimental units are the

objects upon which the experiment is run, five

different scenarios of software development project

were created, aiming to approach different software

project risks. The first scenario is related to the

assignment’s compatibility of the involved

stakeholder role with the type of activity (managerial

or productive). The second scenario is related with

the interaction among the organizational flows to the

acquisition of new hardware during the project. The

third scenario is related to the risk of identifying that

the most qualified staff is unavailable at critical

moments. The fourth scenario is related to the

situation where new employees are hired and must

perform some sort of training before joining the

project. In the fifth scenario, the manager identifies

that software components purchased from a third

party contain defects which limit their functionality,

so that he should contact the product’s supplier.

The outcomes of this experiment (response

variables) are concerned to the effort and accuracy

in planning activities of software projects. Also, any

project characteristics (called factors) intentionally

varied were identified during experimentation. Each

factor has several possible alternatives. In this

experiment, there is one factor to be analyzed

(project planning methods) and two alternatives: the

traditional method of project planning and the

method using the integrated planning model SPIM.

Considering the characteristics of this research, the

one-factor designs were chosen. This sort of design

involves comparing the variable response to each

alternative in a given number of experimental units.

Some characteristics, however, would be desirable

to be invariant, but they vary in an experiment

(blocking variables). Project management (much

more than other software development task) is an

activity which performances highly depends on the

person who does it, hence the risk that results are

highly influenced by ability and experience is high.

In this experiment, the level of experience in project

planning is a blocking variable.

If both alternatives are used in the same project,

two similar teams are required. The definition of

which participants would perform each approach of

software project planning (in the traditional way or

with the help of SPIM model) occurred randomly. In

this case, the experimenter took thirty six cards (half

red and half black) from the pack; the red cards

would correspond to the use of the traditional project

planning method and the black ones to the use of the

SPIM method. The experimenter shuffled the cards

and allowed each subject to take a card for each

experimental unit (software development project).

The balancing principle was also used so that each

propose of software project planning was performed

by the same number of participants (eighteen

participants to each proposal).

4.3 Execution of the Experiment

The realization of the experiment occurred in

December 2011, when the set of participants

performed the experiment in a controlled

environment (university's computer lab). Initially, all

participants received an email inviting them to join

this experimental study. In this invitation it was

explained that this event included a presentation of

SPIM and the realization of a practical activity

where participants would have the opportunity to

perform exercises based on typical situations of

project management. The experiment involved only

students that had some interest in the area of project

management. To take part in this event the invitees

had to access the link to the event and create an

access account. Thus, a web site was developed in

order to store the questionnaires of this experiment

aiming to maintain the integrity of the data obtained

during its execution. An access control system

ensured that each participant had access only to

questions that have been designated for them.

The problem studied corresponds to five

scenarios that simulated situations in software

development projects. At first, all participants

received a brief training in the SPIM model and had

the opportunity to test the main features of the SPIT

on a sample project. Later, they had the opportunity

to make the first questions about the proposed work.

Then, they were presented to the same description of

each scenario and were asked to perform the

corresponding project planning - some using the

traditional method and others with the SPIT tool. In

order to avoid possible distortions in the obtained

results both in the trial of SPIT and the

questionnaire’s resolution phase it didn’t occur

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

236

having any interaction with the interviewer.

4.4 Analysis

There are different analysis techniques depending on

the characteristics of the collected data and on the

applied design. The methods of analysis can roughly

be divided into two major blocks: parametric and

non-parametric methods. According to Miles and

Huberman (1994), parametric tests are statistically

more powerful than non-parametric methods.

However, if these parametric tests are not

conclusive, then the analysis will have to resort to

the application of non-parametric tests. Considering

these two types of analysis techniques, the drawing

of conclusions was attempted by rejecting the null

hypotheses with the parametric test or/and accepting

them with the non-parametric test.

For the testing of hypotheses, in a context of one

factor and two treatments, the literature suggests the

significance test called ‘T test’ for two independent

samples (if performed a parametric test) or ‘Mann-

Whitney test’ (if it is a non-parametric test). This

definition was taken after verifying if the

distribution was normal or not (by the ‘Shapiro-Wilk

test’) and checking the variance of the data obtained

by running the experiment (‘Levene Test’).

5 ANALYSIS OF RESULTS

According to the scope of this research, it was

necessary to evaluate these two hypotheses: effort

and accuracy. For the hypothesis analysis of this

research, we used the T test (suitable for comparing

the averages of a quantitative variable between two

independent groups) or Mann-Whitney test (if the

test is non-parametric). Then, the verification of

each null hypothesis for each developed scenario

was performed. The null hypothesis (H

) is related to

the randomness of the observed results, that is, if it

is true, statistically the results of the experiment

evidence to be occasional (no conclusion can be

drawn). The alternative hypothesis (H

) is one that

will be accepted if the null hypothesis is rejected.

Still, it must be noted that the level of the test

significance (p-value) was fixed in 5%. The analyses

presented in this experiment were made using the

Statistical Package for Social Sciences (SPSS).

5.1 First Hypothesis: Effort

Through an initial analysis of the distribution, the

behaviour of the samples could be evaluated.

Initially, we studied the behaviour of each sample

(traditional and SPIM) in order to find outliers. An

outlier is an observation that lies an abnormal

distance from other values in a random sample from

a population (Grubbs, 1969). According to the

boxplot graph it was observed that the effort variable

does not have outliers. After that, it could be verified

that the data has a normal distribution through

Shapiro-Wilk test. However, the T test also assumes

that the variability of each group is approximately

equal. With this goal, two hypotheses were defined:

: The variances are equal; H

: The variances are

not equal. The Levene's test (see Table 1) shows if

its assumption of the T test has been met.

Table 1: Levene's test for the effort variable.

Variable Assumption Significance

Effort Equal Variances assumed 0.271

Equal Variances not assumed 0.271

According to the results, the significance (p-

value) of Levene's test is 0.271. If this value is lower

than or equal to the significance level (α) for the test

(in this case 0.05), then the null hypothesis in which

the variability of the two groups is equal may be

rejected, implying that the variances are unequal. If

the p-value is greater than the α level, then, equal

variances are assumed. In this case, 0.271 is greater

than α, so the fact that the variances are equal was

assumed. Once it was identified that the distribution

was normal and variances were equal, the T test was

applied (see results in Table 2).

This is a two-sided test, in which the p-value =

0.140 is directly compared with α = 0.05

(significance level). Since p-value=0.140 > 0.05, H

is not rejected. Thus, there is no statistical evidence

to reject the hypothesis that the effort average to

accomplish the planning of the activities using the

traditional model is equal to the spent effort with the

SPIM model. It was observed that the effort average,

in minutes, to carry out the planning of the activities

using the SPIM model was around 48 minutes while

using the traditional model it was around 43

minutes. For the presented analyzes, the conclusion

is that, statistically, there is no significant difference

in relation to the effort to make the planning of

projects using the traditional method and the SPIM.

Table 2: T test for the effort variable.

Variable Criterion T Sig.

Effort

Equal Variances assumed -1.511 0.140

Equal Variances not assumed -1.511 0.140

AnExperimentalStudyontheDynamicReconfigurationofSoftwareProjects

237

5.2 Second Hypothesis: Accuracy

Similarly to the analysis of the first hypothesis, it

was observed that the accuracy variable does not

have outliers. However, using the Shapiro-Wilk test,

it was not possible to identify if that the data has a

normal distribution. Therefore, a parametric test

(like T test) could not be used. The Mann-Whitney,

however, test could be used. It is a non-parametric

analog to the independent samples T test and can be

used when we do not assume that the dependent

variable is a normally distributed interval variable.

Thus, the Mann-Whitney test for two independent

samples was used to verify that the observed

differences between the averages in two independent

groups are statistically significant. To achieve this

objective, the following hypotheses were defined:

: There is no difference between the mean of the

two samples; H

: There is a difference between the

mean of the two samples. The results of the Mann-

Whitney can be seen in Table 3.

Table 3: Mann-Whitney test for the accuracy variable.

Variable

Mann-

Whitney U

Wilcoxon W Z

Asymp

Sig

Accuracy 66.500 237.500 -3.046 0.002

Since the degree of significance (0.002) is

smaller than the level of significance given (0.05),

the hypothesis H

was rejected. Based on the results

presented for the accuracy variable it is understood

that there is a difference between the mean effort to

do the planning with the traditional and SPIM

methods. However, based on the results of the

Mann-Whitney test only the null hypothesis can be

rejected, but it was not possible to evaluate the

alternative hypotheses. Comparing the mean values

of the SPIM approach (84%) with the traditional

approach (52%) we conclude that the accuracy in

making the planning model using the SPIM is larger

than in the traditional model.

6 QUALITATIVE ANALYSIS

A qualitative exploratory evaluation has also been

conducted. At the end of the experiment’s execution

each participant answered a questionnaire, produced

in accordance to Rea and Parker (2005). The survey

had 17 questions where the first 8 were focused on

the managers individual knowledge mapping and the

remaining were used to estimate the SPIM model’s

contributions in the planning process from the

project managers’ point of view.

An analysis of the obtained results from the

questions related to the profile of respondent

individuals shows that 52.63% of these had a project

management experience between two and five years

and 21.05% had experience between five and ten

years. In addition to that experience of the

respondents, 34.21% of the sample reported their

experience in project management as little while the

remaining 65.79% declared it as moderate or

advanced. In addition, 72.42% of the subjects

classified their knowledge of software development

processes as moderate or advanced. This indicates a

sufficient range of experience regarding project

management by the subjects.

Table 4: Perceived benefits in performing the integrated

planning of managerial and productive activities.

Question %

Reduction in time during the project’s elaboration

process

52.63

Identification of the dependencies between the

management supporting activities and production

activities

100

Identification and measuring of the indirect costs of the

project, due to the management support activities

Being able to access enterprise workflow information 75

The capacity of avoiding distortions during planning

when support activities are involved

100

Helps to anticipate the needs stemming from support

areas of the organization during the project planning

87.50

Makes an explicit distinction between the activities of a

software project and the activities belonging to other

departments within the organization;

100

The analysis of the SPIM begins with the

respondents’ evaluation of the direct benefits in

performing the integrated planning of managerial

and productive activities in a project (see Table 4).

According to the second and the last rows in

Table 4, all participants found that the integrated

planning allows the identification of the hidden

dependencies between the management supporting

activities and the production activities, while

avoiding frequent distortions in the planning of the

projects due to the uninformed use of resources from

the management supporting activities. The visibility

of management supporting activities with the

activities in the software project (whether productive

of managerial) was also identified as a strong benefit

of SPIM by all of the interviewees.

When questioned whether they agreed or not on

the distinct nature of the three types of activities,

most of the respondents (75%) answered that the

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

238

SPIM model helps managers to access enterprise

workflow information. Also, 87.50% of the

interviewed subjects agreed that the SPIM model

contributes in identifying the dependencies of the

activities between the project workflow and the

organization workflow, which allows the prediction

of the needs that come upon the organizational

support areas during the planning of the project. As

a final consideration, the majority of the participants

found that the SPIM contributes in the identification

and measuring of the indirect costs of the project,

due to the management support activities.

7 CONCLUSIVE REMARKS

This paper presented the SPIM, a model to integrate

software project management with organizational

workflows. An experimental strategy was chosen to

evaluate the proposed model. This experiment aimed

to compare the accuracy and the effort of integrated

planning model SPIM compared to the traditional

model of software project planning, considering the

characteristics and particularities involved in the

Unified Process.

The experiment reveals that the use of the SPIM

helps managers to create and conduct a more precise

project plan than the traditional method. In certain

circumstances, the project manager only perceives

the need to have asked another department for some

information earlier just at the very moment the team

must execute a project’s activity that depends on that

other department. The obscurity in identifying this

kind of relationship during the planning and

execution of a software project can negatively affect

the project schedule. This evidence was clear during

this study while analysing accuracy variable.

An evidence related to the effort variable could

also be extracted: the time for planning the activities

using the SPIM is similar to the traditional model.

The idea behind the SPIM comes from the need to

reduce the complexity in visualizing the

interdependencies of both organizational workflows

and individual project’s workflow of activities. Most

of the effort of using the SPIM is related to filling

the extra information proposed by this model.

Nevertheless, the results of the effort variable did

not become favorable to the traditional method. The

results of this experiment reaffirm the benefits that

the SPIM provides in solving problems related to the

inadequate definition of tasks due to the obscurity in

visualizing the interdependency between the

organization’s and project specific workflows.

REFERENCES

Callegari, D., Rosito, M., Blois, M., Bastos R. (2008). An

Integrated Model for Managerial and Productive

Activities in Software Development, In: ICEIS - 10th

International Conference on Enterprise Information

Systems, Spain.

Field, A. (2005). Discovering Statistics Using SPSS. 2nd

edition. Sage Publications.

Graham I., Henderson-Sellers B. and Younessi, H. (1997).

The OPEN Process Specification”, Addison-Wesley.

Grubbs, F. E. (1969). Procedures for detecting outlying

observations in samples. Technometrics, 11, 1-21.

Joslin, D., Poole, W. (2005). Agent-Based Simulation For

Software Project Planning. In Proceedings of the 37th

Conference on Winter Simulation. Orlando, Florida,

pp 1059-1066.

Juristo, N., Moreno, A. (2003). Basics of Software

Engineering Experimentation. Kluwer academic

Publishers. 2nd edition.

Miles M.B., Huberman A.M. (1994). Qualitative Data

Analysis, 2nd edition. London: Sage Publications.

Kerzner, H. (2000). Applied project management: best

practices on implementation. NY: Wiley e Sons, 534p.

Kruchten, P. (2000). The Rational Unified Process: An

Introduction, Addison-Wesley, 2nd edition, 2000.

Object Management Group (2011). Software & Systems

Process Engineering Metamodel Specification 2.0,

viewed September 25, 2011, <http://

http://www.omg.org/spec/SPEM/2.0//>.

Pfleeger, S. L., Atlee, J. Software Engineering: Theory

and Practice. (2009). 4th edition, Prentice Hall.

Pressman, R. (2009). Software engineering: a

practitioner's approach. 7th edition, McGraw-Hill.

Project Management Institute (2008). PMBOK - A Guide

to the Project Management Body of Knowledge.

Newtown Square, PA: PMI, 4th edition.

Rea, L. M., Parker, R. (2005). Designing and Conducting

Survey Research: A Comprehensive Guide, Jossey-

Bass. 33rd edition

Rosito, M., Callegari, D., Blois, M., Bastos R. (2012).

Project Management and Software Development

Processes: Integrating PMBOK and OPEN. In:

WASET - World Academy of Science, Engineering and

Technology, Malaysia.

Rosito, M., Bastos, R (2012). A model to integrate

software project management with organizational

workflows. In: ISDA - 12th International Conference

on Intelligent Systems Design and Applications.

Schwalbe K. (2010). Information Technology Project

Management, Thomson Learning, 6th edition

Solingen, R., Berghout, E. (1999). The

Goal/Question/Metric Method. NY: McGraw-Hill.

Wohlin, C., Runeson, P., Höst, M., Ohlsson, M., Regnell,

B., Wesslén, A. (2000). Experimentation in software

engineering: an introduction. Boston: Kluwer

Academic Publishers, 204 p.

AnExperimentalStudyontheDynamicReconfigurationofSoftwareProjects

239