A computerized model for measuring and benchmarking the partnering performance of construction projects

Automation in Construction 18 (2009) 1099–1113

Contents lists available at ScienceDirect

Automation in Construction j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / a u t c o n

A computerized model for measuring and benchmarking the partnering performance of construction projects John F.Y. Yeung ⁎, Albert P.C. Chan, Daniel W.M. Chan Department of Building and Real Estate, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China

a r t i c l e

i n f o

Article history: Accepted 30 July 2009 Keywords: Partnering Performance Index (PPI) Key Performance Indicators (KPIs) Benchmarking Project monitoring Partnering Internet and information technology Hong Kong

a b s t r a c t Research into partnering performance measures for building and construction projects becomes crucial because a growing trend of client organizations has been observed to adopt partnering approach to procure their projects worldwide over the past decade. Although there are some related research studies and papers documented on this research area, few, if any, comprehensive and systematic research studies focus on developing a comprehensive, objective, reliable and practical performance evaluation model for partnering projects in construction. A Partnering Performance Index (PPI), which is composed of seven weighted Key Performance Indicators (KPIs), has been developed to measure, monitor, improve, and benchmark the partnering performance of construction projects in Hong Kong. A set of Quantitative Indicators (QIs) and well-defined ranges of Quantitative Requirements (QRs) for each QI have been further established using the Delphi survey technique and Fuzzy Set Theory. Evaluation of partnering success can now be based on quantitative evidences, thus tackling the subjectivity of performance evaluation. By making use of the Internet and database technology, PPI can be monitored on-line, thus saving much time, cost and efforts on data collection and retrieval than if they are done manually. As such, an Internet-based computerized partnering monitoring and benchmarking tool, namely the Computerized Partnering Performance Index System (CPPIS) has been developed. The Internet-based CPPIS enables project participants to input data at any time and location and the project administrators could perform data analysis via the Internet. The CPPIS also enables project managers to measure, monitor, improve and benchmark their partnering performances against those already stored in the database. Graphical presentations of data and various performance measures are also built in to assist various end-users to identify problematic areas and critical success factors for achieving partnering excellence. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Research into partnering performance measures for building and construction projects becomes vital because more client organizations have introduced partnering approach to their projects worldwide during the last decade [1]. Although some related research studies and papers have been documented on this research area [2–4], few, if any, comprehensive and systematic research studies focus on generating a comprehensive, objective, reliable and practical performance evaluation model for partnering projects in construction. Moreover, these research studies were conducted with some limitations. Crane et al. [2] launched extensive interviews with 21 successful partnering relationships and then classified partnering performance measures into three types: result, process and relationship measures. However, a performance index was not compiled and appropriate Quantitative

⁎ Corresponding author. Tel.: +852 2766 5873; fax: +852 2764 5131. E-mail address: [email protected] (J.F.Y. Yeung). 0926-5805/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.autcon.2009.07.003

Indicators (QIs) were not identified for assessing partnering performance, thus making project monitoring and benchmarking difficult. Cheung et al. [3] adopted eight partnering measures to develop a Partnering Temperature Index to measure the performance of partnering projects. However, the weightings by default were treated as equal for each measure and the selection of partnering measures was quite subjective. The evaluation was also subjective because it was solely based on the personal perceptions of respondents to measure whether the project has achieved its objectives in terms of different measures of Partnering Temperature Index (PTI) by using a five-point Likert scale ranging from 1 = strongly disagree, 2 = slightly disagree, 3 = no feeling either way, 4 = slightly agree, and 5 = strongly agree. Lo et al. [4] used a Balanced Scorecard approach to measure the partnering project performance through an extensive literature review and data analysis (principal components factor analysis) via a questionnaire survey. Although a comprehensive approach was applied to assess partnering performance of construction projects in Hong Kong, different industrial practitioners might interpret the same strategic objectives differently. In addition, corresponding weightings

1100

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

were not derived for different strategic objectives, thus rendering project monitoring and benchmarking difficult. It should be highlighted that these researchers encountered the common problem of subjectivity in selecting the most vital KPIs for measuring the partnering performance and the lack of scientific and objective resolution methods. To solve this problem, a Partnering Performance Index (PPI), comprising seven weighted KPIs, has been generated to measure, monitor, improve and benchmark the partnering performance of construction projects in Hong Kong [5]. A set of Quantitative Indicators (QIs) [6] and well-defined ranges of Quantitative Requirements (QRs) [7] for each QI have been further established to ensure that various assessors could evaluate partnering performance based on quantitative evidences, thus eliminating the subjectivity of performance evaluation. The purpose of this paper is to develop a computerized system for compiling the PPI of partnering projects for monitoring and benchmarking purposes. By making use of the Internet and database technology, PPI can be well monitored via the Internet, thus saving much time, cost and efforts on data collection and retrieval than if they are undertaken manually. Project team members can just input their individual project information and data and the computerized system can provide a PPI score to compare their project performances with other counterparts directly. The Computerized Partnering Performance Index System (CPPIS) not only provides better understanding to clients, contractors and consultants in running a successful partnering project, but it also helps set a benchmark for measuring the partnering performance of their projects. Both the clients and contractors can apply the system for monitoring purposes when the partnering-based project is implemented at the very early beginning of the construction phase. And the results can be used to compare the partnering performance of a certain project with its counterparts for benchmarking purposes. It should be noted that since project partnering is still dominant within the Hong Kong construction industry when compared with strategic partnering, the PPI model and the CPPIS are mainly applied to project partnering. 2. Research methods The research methods employed in this research study encompassed: (1) literature review; (2) content analysis; (3) face-to-face interviews with field experts; (4) Delphi questionnaire survey; (5) empirical questionnaire survey; and (6) Fuzzy Set Theory. Literature review and content analysis were firstly used to establish a conceptual framework of the performance measures for partnering projects. Literature review was used because it could consolidate all previous studies related to the research study undertaken by other researchers and the understanding of the prevailing partnering practice. After conducting an extensive literature review, both qualitative and quantitative content analyses were used to develop the conceptual framework of the performance measures for partnering projects. Fellow and Liu [8] asserted that content analysis is often used to determine the main facets of a set of data, by simply counting the number of times an activity occurs, or a topic is depicted. The initial step in content analysis is to identify the materials to be analyzed. The next step is to determine the form of content analysis to be employed, including qualitative, quantitative or structural; the choice is dependent on the nature of the research project. The choice of categories will also depend upon the issues to be addressed in the research if they are known. In qualitative content analysis, emphasis is placed on determining the meaning of the data (grouping data into categories) while quantitative content analysis extends the approach of the qualitative form to yield numerical values of the categorized data (frequencies, ratings, ranking, etc.) which may be subject to statistical analyses. Comparisons can be made and hierarchies of categories can be examined. After that, a PPI, which is composed of seven weighted KPIs, has been developed using four rounds of the first Delphi questionnaire survey [5]. This method was used because it could solve the problem of subjectivity

in selecting the most vital KPIs and develop their appropriate weightings. The PPI can assist in developing a benchmark for measuring the performance of partnering projects. However, different assessors may have their own semantic interpretations on each KPI. In order to avoid any possible discrepancies in interpreting the meaning of each KPI, a set of QIs was then established by conducting five structured face-to-face interviews with field experts and 2 rounds of the second Delphi questionnaire survey [6]. Since the QIs selected are fuzzy in nature which requires assessors' subjective value judgment, another empirical questionnaire survey was used to define Fuzzy Quantitative Requirements (FQRs) for each QI [7]. A Fuzzy Set Theory approach, namely the Modified Horizontal Approach with the Bisector Error Method, was used for data analysis. Fuzzy Set Theory was adopted in this research study because it is used to tackle ill-defined and complex problems due to incomplete and imprecise information that characterize the real-world systems [9]. 3. Research Findings: development of KPIs and a PPI, identification of QIs, and definition of Fuzzy QRs Based on the comprehensive literature review with content analysis, a conceptual framework encompassing 25 various measures has been developed for assessing the performance of partnering projects in construction. The 25 performance measures were classified into four major categories, as follows [5]: (1) Result-oriented objective measures: • Time performance • Cost performance • Profit and financial objectives • Scope of rework • Safety performance • Environmental performance • Productivity • Pollution occurrence (2) Result-oriented subjective measures: • Quality performance • Professional image establishment • Client's satisfaction • Customer's satisfaction • Job satisfaction • Innovation and improvement (3) Relationship-oriented objective measures: • Litigation occurrence and magnitude • Dispute occurrence and magnitude • Claim occurrence and magnitude • Introduction of facilitated workshop (4) Relationship-oriented subjective measures: • • • • • • •

Trust and respect Effective communications Harmonious working relationships Long-term business relationship Top management commitment Employee's attitude Reduction of paperwork

The selection criterion of KPIs is similar to the research work of Chan et al. [10] that only KPIs which have been selected by 50% of experts or above will be selected for further consideration. Only seven KPIs met this criterion after conducting the first two rounds of Delphi survey. These include: (1) time performance; (2) cost performance; (3) quality performance; (4) trust and respect; (5) top management commitment; (6) effective communications; and (7) innovation and improvement. Safety performance, although commonly accepted as an important performance measure for project success, did not meet the “50% of experts” threshold in the Delphi survey, and hence not being included as a KPI for assessing partnering performance.

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

After that, the research team has conducted four rounds of the first Delphi questionnaire survey [5]. In the first round, 39 experts were asked to select a minimum of five to a maximum of 10 out of 25 KPIs that they believed to be the most important KPIs to evaluate the success of partnering projects (the respondents were encouraged to propose additional KPIs for partnering projects in Hong Kong if deemed appropriate.) Similar to Chan et al. [10], only KPIs which have been selected by at least 50% of experts will be selected for further consideration. Seven KPIs met this criterion in the first round of the Delphi study. The top seven KPIs were: (1) time performance; (2) cost performance; (3) quality performance; (4) trust and respect; and effective communications (equal frequencies for both); (6) harmonious working relationships; and top management commitment (equal frequencies for both). In addition, five new KPIs, which had not been identified from the literature, were suggested by the panel of experts. They included: (1) method of procurement and time for closing of final account; (2) job efficiency and reliability; (3) minimizing impact on operations; (4) commitment of staff at work level; (5) good public relations. However, they were not selected for further study since they did not meet the 50% cut-off criterion either.

1101

Similar to Round 1 of the Delphi survey, the second round of the Delphi questionnaire was forwarded to the group of panel members by both mail and email. In this round, the results of Round 1 were consolidated and presented and the experts were requested to reconsider whether they would like to change any of their original choices in the light of the consolidated results from Round 1. It should be pointed out that ‘harmonious working relationships’, originally rated as one of the top seven KPIs in Round 1, was dropped out and replaced by ‘innovation and improvement’. The descending order of the top seven KPIs was slightly changed as follows: (1) time performance; (2) cost performance; (3) quality performance; (4) trust and respect; (5) top management commitment; (6) effective communication; and (7) innovation and improvement. In the third round of the Delphi questionnaire, experts were asked to provide ratings on the top seven KPIs based on a five-point Likert scale to evaluate the success of partnering projects. In addition, the five-point Likert scale, ranging from 1 = least important, 2 = slightly important, 3 = important, 4 = very important, to 5 = most important, is used because the dimension for measuring KPIs should be unipolar, referring to different degrees of the same attribute, but not bipolar, referring to

Table 1 Correlation matrix amongst the seven weighted KPIs (for Round 3 of the Delphi questionnaire in Hong Kong). Source: Yeung et al. [5]; permission has been obtained for both print and online use from Taylor & Francis. Correlation matrix

Time performance

Cost performance

Quality performance

Trust and respect

Top management commitment

Effective communications

Innovation and improvement

Time performance Cost performance Quality performance Trust and respect Top management commitment Effective communications Innovation and improvement

1

0.505a 1

0.551a 0.520a 1

− 0.347 − 0.411b − 0.360b 1

− 0.248 − 0.261 − 0.418b 0.682a 1

− 0.241 − 0.388b − 0.129 0.674a 0.550a

− 0.138 − 0.242 0.172 0.249 0.248

a b

0.547a 1

1

Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed).

Table 2 Comparisons of rounds three and four of the Delphi questionnaire in Hong Kong. Source: Yeung et al. [5]; permission has been obtained for both print and online use from Taylor & Francis. KPIs for partnering projects in Hong Kong

Round 3 Mean rating

Time performance Cost performance Top management commitment Quality performance Trust and respect Effective communications Innovation and improvement Number (n) Kendall's Coefficient of Concordance (W) Level of significance

4.48 4.32 3.97 3.94 3.81 3.52 2.81 31 0.249 0.000

Round 4 Rank

Corresponding weighting

1 2 3 4 5 6 7

0.167 0.161 0.148 0.147 0.142 0.131 0.104

Mean rating 4.55 4.35 4.10 3.90 3.90 3.58 2.90 31 0.290 0.000

Rank

Corresponding weighting

1 2 3 4 4 6 7

0.167 0.160 0.150 0.143 0.143 0.131 0.106

Table 3 Correlation matrix amongst the seven weighted KPIs (for Round 4 of the Delphi questionnaire in Hong Kong). Source: Yeung et al. [5]; permission has been obtained for both print and online use from Taylor & Francis. Correlation matrix

Time performance

Cost performance

Quality performance

Trust and respect

Top management commitment

Effective communications

Innovation and improvement

Time performance Cost performance Quality performance Trust and respect Top management commitment Effective communications Innovation and improvement

1

0.464a 1

− 0.193 − 0.231 1

0.414b 0.528a − 0.271 1

− 0.181 − 0.416b 0.804a − 0.256 1

− 0.213 − 0.416b 0.426b − 0.121 0.571a 1

− 0.166 − 0.278 0.205 0.185 0.273 0.495a 1

a b

Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed).

1102

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

the presence of opposite attributes [11]. A statistical analysis was performed on the 31 questionnaires received in which the mean ratings for the top seven KPIs were computed. A preliminary series of weighted KPIs was developed based on the mean ratings advocated by the 31 experts. The weighting for each of the top seven KPIs was computed by using the following equation.

WKPIa =

MKPIa for a = 1 ∑ MKPIg

Partnering Performance Index (PPI) is developed which can be represented by the following formula: Partnering Performance Index ðPPIÞ for Round 3

PPI = 0:167 × Time Performance + 0:161 × Cost Performance + 0:148 × Top Management Commitment + 0:147 × Quality Performance

ð1Þ

g

+ 0:142 × Trust and Respect + 0:131 × Effective Communications + 0:104 × Innovation and Improvement

where MKPIa ∑ MKPIg g

WKPIa represents the weighting of a particular top-7 KPI represents the mean ratings of a particular top-7 KPI represents the summation of mean ratings of all the top-7 KPIs

The results for the weighting for each KPI, including: (1) time performance, with the weighting of 0.167; (2) cost performance, with the weighting of 0.161; (3) top management commitment, with the weighting of 0.148; (4) quality performance, with the weighting of 0.147; (5) trust and respect, with the weighting of 0.142; (6) effective communications, with the weighting of 0.131; and (7) innovation and improvement, with the weighting of 0.104. In order to compile a composite indicator to evaluate the success of partnering projects, a

ð2Þ

The PPI is composed of the top-7 weighted KPIs identified in the Round 3 of the Delphi questionnaire and the coefficients are their individual weightings, which are calculated by their individual mean ratings divided by the total mean ratings. The Index is derived based on the assumption that this is a linear and additive model. It is logical and valid to derive this linear and additive model because the correlation matrix as shown in Table 1 reveals that the top-7 weighted KPIs are not highly correlated with each other at 5% significance level (more than half of them are even insignificantly correlated with each other). In addition, the units of measurement for the top-7 weighted KPIs are different so it is not likely to have any multiplier effect between them. Though it seems more sophisticated

Table 4 Result of round two of the Delphi questionnaire in Hong Kong. Source: Yeung et al. [6]; permission has been obtained for both print and online use from Taylor & Francis. Quantitative indicators for measuring the performance of partnering projects in Hong Kong

Average ratings of experts in Round 2 Importance Measurability Obtainability Mean ratings

Quantitative indicators for measuring time performance Variation of actual completion time expressed as a percentage of finally agreed completion time 4.56 Time improvement: measuring how much time improvement of a project is delivered to previous similar projects 3.80 Subjective assessment by using Likert scale (say ahead of schedule, on time, or behind schedule) 3.12

4.64 2.92 3.30

4.38 3.04 3.46

4.53 3.25 3.29

Quantitative indicators for measuring cost performance Variation of actual project cost expressed as a percentage of finally agreed project cost Cost improvement: measuring how much cost improvement of a project is delivered to previous similar projects Subjective assessment by using Likert scale (say within budget, on budget, or overrun budget)

4.56 4.00 3.12

4.48 3.08 3.62

4.32 3.32 3.68

4.45 3.47 3.47

Quantitative indicators for measuring top management commitment performance Partnering development cost of project expressed as a percentage of total project cost 3.00 Percentage of top management attendance in partnering meetings 4.32 Measuring level of top management commitment by using Likert scale (say high level, moderate level, or low level) 3.62

4.08 4.60 3.80

4.12 4.60 3.88

3.73 4.51 3.77

Quantitative indicators for measuring quality performance Cost of rectifying major defects or non-conformances of a project expressed as a percentage of total project cost Average number of non-conformance reports generated per month Perceived end-users' satisfaction scores by using Likert scale

4.26 3.94 3.74

3.76 4.28 3.46

3.22 4.08 3.74

3.75 4.10 3.65

Quantitative indicators for measuring trust and respect performance Average duration for settling variation orders Frequency of meeting another party's expectation Perceived key stakeholders' satisfaction scores by using Likert scale

3.64 3.50 3.82

3.56 2.46 3.66

3.48 2.64 3.84

3.56 2.87 3.77

2.92

2.64

2.95

2.76

2.70

2.77

3.34

3.58

3.51

4.26 3.80

3.72 3.80

3.56 3.80

3.85 3.80

3.44

3.52

3.72

3.56

Quantitative indicators for measuring effective communications performance Reduction of written communication: measuring how much written communication is reduced as compared to 3.28 previous similar projects Variation of the number of formal letters and emails sent between parties per month against the number with 2.84 previous similar projects Perceived key stakeholders' satisfaction scores by using Likert scale 3.62 Quantitative indicators for measuring innovation and improvement performance Cost saving resulted from innovation expressed as a percentage of total project cost Number of innovative initiatives introduced (e.g. construction techniques, procurement approaches, management strategies) Perceived key stakeholders' satisfaction scores by using Likert scale Kendall's Coefficient of Concordance (W) Level of significance Remark 1: Rating 1 = very unimportant/very difficult and 5 = very important/very easy.

0.401 0.000

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

to use a nonlinear model to fit the data obtained, overfitting is a common problem with nonlinear models especially when the sample size is not sufficiently large [12,13]. That is why a linear, but not nonlinear model is recommended if the relationship amongst variables is not proved to be nonlinear. In fact, a linear model is assumed to be a linearized model of an unknown nonlinear model if it really exists [14,15]. Practically speaking, it is simpler and easier to use this model to measure the partnering performance of construction projects in the Hong Kong construction industry. In order to obtain a measure of consistency, a statistical test was applied involving the calculation of the Kendall's Coefficient of Concordance (W) for the KPIs provided by the 31 experts [10] with the aid of the Statistical Packages for Social Sciences (SPSS) computer software. If the Concordance Coefficient is equal to 1, it means that all the experts rank the KPIs identically. In contrast, if the Concordance Coefficient is equal to 0, it means that all the experts rank the KPIs totally different. Table 2 also shows that Kendall's Coefficient of Concordance (W) for the rankings of the top-7 weighted KPIs was 0.249, which was statistically significant at 1% significance level. The null hypothesis that the respondent's ratings within the group of panel experts are unrelated to each other would have to be rejected. Therefore, it can be concluded that significant amount of agreement among the respondents within the group of panel experts is found. For Round 4 of the Delphi survey, the experts were provided with the consolidated results obtained in Round 3. The average ratings of the 31 experts for each KPI and the respondent's own ratings in Round 3 were

1103

provided. The respondents were asked to reassess their ratings in the light of the mean scored by the 31 experts. Most experts had reconsidered their ratings provided in the previous round and had made some adjustments to their ratings. The consistency of the experts' weightings was again computed using the SPSS package to calculate the Kendall's Coefficient of Concordance (W). Table 2 shows that there is no change for the order of their mean ratings except that trust and respect is upgraded from the fifth rank to the fourth rank. In addition, their corresponding weightings are similar with those of Round 3. It also reveals that the consistency of the experts' rankings for the top-7 weighted KPIs was improved by 16.5% to 0.29, which was again statistically significant at 1% level. Partnering Performance Index ðPPIÞ for Round 4 PPI = 0:167 × Time Performance + 0:160 × Cost Performance + 0:150 × Top Management Commitment + 0:143 × Quality Performance + 0:143 × Trust and Respect + 0:131 × Effective Communications + 0:106 × Innovation and Improvement ð3Þ Similar to the Index derived in Eq. (2), this PPI is composed of the top7 weighted KPIs identified in the Round 4 of the Delphi questionnaire and the coefficients are their individual weightings, which are calculated

Fig. 1. The Modified Horizontal Approach adopted in this research study in defining the Fuzzy Quantitative Requirements (adapted from Ng et al. [20] and Chow [21]).

1104

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

by their individual mean ratings divided by the total mean ratings. Same as the previous index, this Index is derived based on the assumption that this is a linear and additive model. The correlation matrix as indicated in Table 3 manifests that the top-7 weighted KPIs are not highly correlated with each other at 5% significance level (more than half of them are even insignificantly correlated with each other). Therefore it is valid to assume this linear and additive model. After that, the research team has conducted five structured face-toface interviews with construction experts in procuring partnering projects and then another two rounds of Delphi questionnaire survey. By doing so, a series of Quantitative Indicators (QIs) have been established for each of the seven weighted KPIs [6]. The results are shown in Table 4. Nevertheless, the establishment of a set of QIs cannot fully eradicate the subjectivity of performance evaluation. For instance, in terms of time performance, ahead of schedule by 2% may be perceived as ‘good performance’ in one case while ahead of schedule by 5% may be regarded as ‘very good performance’ in other case. Should a partnering project be classified as ‘good’ or ‘very good’ in case of ahead of schedule by 3.5%? For the sake of rectifying this deficiency, well-defined quantitative ranges are needed to objectively, reliably and practically measure the partnering performance of construction projects. After that, a Fuzzy Set Theory approach, namely the Modified Horizontal Approach with the Bisector Error Method (Fig. 1) was used to define the Fuzzy Quantitative Ranges (FQRs) [7]. Table 5 shows the results of FQRs. It should be noted that although four of the partnering measures (time performance, cost performance, quality performance, and innovation and improvement) will not be fully known until project completion, an interim and regular assessment of these measures could facilitate the project managers to take timely remedial actions to bring the project back to the right track. Moreover, the other three partnering measures (top management commitment, trust and respect, and effective communications) can be assessed during the project life. Therefore, the PPI can be used to measure, monitor, improve, and benchmark the partnering performance of construction projects. In addition, the PPI for one project could be used to better plan for future projects.

4. Case studies—application of the PPI model In order to demonstrate the application of the PPI model to measure the partnering performance of construction projects in Hong Kong, three case studies were examined and all the data were provided by some survey respondents from another questionnaire survey undertaken by the same research team. Scope of analysis under each case study covers the project performance in terms of time, cost, quality, top management commitment, effective communications and innovation and improvement. Table 6 shows the summary of the background information and the results of different KPIs and PPI of these three Hong Kong case studies. It should be noted that the KPIs and the PPI are applied together with the Quantitative Indicators (QIs) and the Fuzzy Quantitative Requirements (FQRs) in order to mitigate the subjectivity problem in interpreting each of the identified KPIs. The details of each case study are discussed in the following subsections.

4.1. Case 1—a new private building project It is a building work, constructed with total contract duration of 26 months and total contract sum of HK$1150 million (the details were provided by a respondent of a questionnaire). The project was procured with sequential traditional procurement system and the form of contract is guaranteed maximum price. It is a very successful partnering project, in which it was constructed ahead of schedule by 0.63% and it was estimated to be under-run by more than 20% (the information was provided by the respondent). There was 95% of top management attendance in partnering meetings as perceived by the respondent and the average number of non-conformance reports generated per month was 3. The trust and respect score was 9 and the effective communications score was 8. In addition, it received 9% of cost saving resulted from innovation expressed as a percentage of total project cost. As a whole, the PPI was scored at 4.243 out of a total of 5.

Table 5 The Fuzzy QRs of each QI against the five different performance levels. The most important Quantitative Indicator (QI) for each of the seven Performance level The seven selected KPIs (with corresponding weightings) selected KPIs Poor Average (total weighting is equal to 1) Time performance (0.167)

Cost performance (0.160)

Top management commitment (0.150) Quality performance (0.143)

Quality performance (0.143)

Trust and respect (0.143)

Effective communications (0.131) Innovation and improvement (0.106)

Variation of actual completion time expressed as a percentage of finally b− 3.1% agreed completion time b− 2.0% b− 2.6% Variation of actual project cost expressed as a percentage of finally b– 2.4% agreed project cost b− 5.3% b− 2.9% Percentage of top management attendance in partnering meetings b56.7% b57.0% b56.8% Average number of non-conformance reports generated per month N8 (for civil works) N8 N8 Average number of non-conformance reports generated per month N15 (for building works) N14 N15 Perceived key stakeholders' satisfaction scores by using Likert scale b 4.8 b 4.7 b 4.8 Perceived key stakeholders' satisfaction scores by using Likert scale b5 b5 b5 Cost saving resulted from innovation expressed as a percentage of total b 0.6% project cost b 0.6% b 0.6%

Good

Very good

− 3.1%–1.5⁎% 1.5%–7.4⁎% 7.4%–10.3⁎% − 2.0%–1.8⁎% 1.8%–7.9⁎% 7.9%–10.3⁎% − 2.6%–1.6⁎% 1.6%–7.6⁎% 7.6%–10.3⁎% − 2.4%–0⁎% 0%–4.3⁎% 4.3%–10.4⁎% − 5.3%–0⁎% 0%–4.6⁎% 4.6%–10.4⁎% − 2.9%–0⁎% 0%–4.4⁎% 4.4%–10.4⁎% 56.7%–72.1⁎% 72.1%–81.4⁎% 81.4%–96.1⁎% 57.0%–72.0⁎% 72.0%–82.1⁎% 82.1%–96.4⁎% 56.8%–72.1⁎% 72.1%–81.7⁎% 81.7%–96.3⁎% 4+–8 3+–4 1+–3 4+–8 3+–4 1+–3 4+–8 3+–4 1+–3 11+–15 6+–11 2+–6 11+–14 8+–11 3+–8 11+–15 7+–11 3+–7 4.8–6.5⁎ 6.5–8⁎ 8–8.9⁎ 4.7–6.0⁎ 6.0–8⁎ 8–8.9⁎ 4.8–6.2⁎ 6.2–8⁎ 8–8.9⁎ 5–6.8⁎ 6.8–8.3⁎ 8.3–9⁎ 5–6.3⁎ 6.3–8.3⁎ 8.3–9⁎ 5–6.5⁎ 6.5–8.3⁎ 8.3–9⁎ 0.6%–2.8⁎% 2.8%–6.6⁎% 6.6%–9.7⁎% 0.6%–3.1⁎% 3.1%–6.7⁎% 6.7%–9.4⁎% 0.6%–2.9⁎% 2.9%–6.7⁎% 6.7%–9.6⁎%

Excellent ≥10.3% ≥10.3% ≥10.3% ≥10.4% ≥10.4% ≥10.4% ≥96.1% ≥96.4% ≥96.3% ≤1 ≤1 ≤1 ≤2 ≤3 ≤2 ≥8.9 ≥8.9 ≥8.9 ≥9 ≥9 ≥9 ≥9.7% ≥9.4% ≥9.6%

Note: M⁎% represents “less than M%” while M+ represents “greater than M”. The first figure of each cell is calculated by the Vertical Error Method, the second figure by the Horizontal Error Method, and the third figure by the Bisector Error Method.

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

1105

Table 6 Case studies—application of the PPI model.

Background Nature of project Type of project Procurement method Tendering method Form of contract Total contract duration Total contract sum KPIs survey result Time performance Cost performance Quality performance Top management commitment Trust and respect Effective communications Innovation and improvement

Case 1

Case 2

Case 3

Private building work Partnering project Sequential traditional Negotiated tendering Guaranteed maximum price 26 months HK1150 million

Public housing work Partnering project Management contracting Open tendering Fixed price lump sum 24 months HK600 million

Infrastructure work Partnering project Sequential traditional Selective tendering Target cost contracting 26 months HK280 million

Ahead of schedule by 0.63% Under-run budget by more than 20% 3 average number of non-conformance reports generated per month 95% of top management attendance in partnering meetings 9 out of 10 trust and respect scores 8 out of 10 effective communications scores 9% cost saving resulted from innovation expressed as a percentage of total project cost

On time On budget 5 average number of non-conformance reports generated per month 90% of top management attendance in partnering meetings 8 out of 10 trust and respect scores 8 out of 10 effective communications scores 7% cost saving resulted from innovation expressed as a percentage of total project cost

Ahead of schedule by 12.5% Under-run budget by 5% 1 average number of non-conformance reports generated per month 85% of top management attendance in partnering meetings 9 out of 10 trust and respect scores 9 out of 10 effective communications scores 3% cost saving resulted from innovation expressed as a percentage of total project cost

3.375 out of 5 scores

4.335 out of 5 scores

Partnering Performance 4.243 out of 5 scores Index (PPI)

4.2. Case 2—a new public housing project It is a public housing work, constructed with total contract duration of 24 months and total contract sum of HK$600 million (the details were provided by a respondent of a questionnaire). The project was procured with management contracting and the form of contract is fixed price lump sum. It is a partnering project with average performance, in which it was constructed on time and it was estimated to be on budget (the information was provided by the respondent). There was 90% of top management attendance in partnering meetings as perceived by the respondent and the average number of non-conformance reports generated per month was 5. The trust and respect score was 8 and the effective communications score was 8. In addition, it received 7% of cost saving resulted from innovation expressed as a percentage of total project cost. As a whole, the PPI was scored at 3.375 out of a total of 5. 4.3. Case 3—a new infrastructure project It is an infrastructure work, constructed with total contract duration of 26 months and total contract sum of HK$280 million (the details were provided by a respondent of a questionnaire). The project was procured with sequential traditional procurement system and the form of contract is target cost contracting. It is an extremely successful partnering project, in which it was constructed ahead of schedule by 12.5% and it was estimated to be under-run budget by 5% (the information was provided by the respondent). There was 85% of top management attendance in partnering meetings as perceived by the respondent and the average number of non-conformance reports generated per month was 1. The trust and respect score was 9 and the effective communications score was 9. In addition, it received 3% of cost saving resulted from innovation expressed as a percentage of total project cost. As a whole, the PPI was scored at 4.335 out of a total of 5. 5. Validation of the research findings In order to ensure that the proposed performance evaluation model for partnering projects in Hong Kong is comprehensive, objective, reliable and practical enough for evaluating the partnering performance of construction projects, the model was presented to another seven independent experts who had not previously been

involved in the Delphi questionnaire surveys for proper validation. Table 7 shows the details of expert interviewees. It should be noted that the surveyed group and validation group were composed of separate and independent groups of experts so that no biases existed for the validation results. Besides, the interviewees are highly qualified and representative enough to validate the research findings derived because they had derived abundant hands-on experiences in procuring partnering projects in Hong Kong. The validation process began with reviewing the appropriateness of KPIs, QIs and FQRs used in the model with the help of the CPPIS. Most of the interviewees believed that the KPIs and their individual weightings assigned are appropriate to measure the partnering performance of construction projects in Hong Kong. Moreover, most of them agreed that the QIs selected for measuring the seven weighted KPIs are objective, reliable and practical. However, some interviewees stated that the QI ‘Average Number of Non-conformance Reports Generated Per Month’ is not good enough to measure the quality performance and the QI ‘Variation of Actual Project Cost Expressed as a Percentage of Finally Agreed Project Cost’ should be clearly defined. On the other hand, the majority of them concurred that the QRs identified reflect the reality and are logical and reasonable, and the performance evaluation model built on the CPPIS is innovative and simple for industrial practitioners to adopt in practice. Some of them recommended that the CPPIS should be updated more frequently so that it is easier for project Table 7 Details of the expert interviewees for validation exercise. Interviewee Position

Organization

Role

Sector

1

Swire Properties Ltd

Client

Private

Realty Cheng & Partners Co Ltd Hong Kong Housing Authority Chevalier Construction Co Ltd Hsin Chong Construction Co Ltd Hong Kong Polytechnic University Mass Transit Railway Corporation Ltd

Client

Private

Client

Public

2 3 4 5 6 7

Project Manager Project Manager Senior Architect Project Manager Project Manager Senior Project Manager Project Manager

Main Private contractor Main Private contractor Client University Client

Infrastructure

1106

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

Table 8 Mean ratings of the validation results. Validation aspect

Validation expert 1 2 3 4 5 6

Degree of appropriateness Degree of objectivity Degree of replicability Degree of practicality Overall reliability

4 4 4 3 4

4 4 5 5 5

4 4 4 4 4

4 4 4 4 3

4 4 4 4 3

3 4 4 3 3

7

3 4 3 4 cannot judge Overall suitability to be adopted to measure 4 4 4 3 3 2.5 4 the partnering performance of construction projects in Hong Kong

Mean rating 3.71 4.00 4.00 3.86 3.67 3.50

Note: 1 = poor and 5 = excellent.

managers to control and monitor their projects, thus achieving continuous improvement. And it is desirable to set up a ‘Performance Reward System’ such as issuing a Certificate of Merit to the top-10 partnering projects which have been listed on the league table for six consecutive months. By doing so, more users are likely to be motivated and encouraged to apply this system in monitoring, measuring, improving, and benchmarking their project performance levels as a whole. In addition, these seven experts were invited to complete a validation scoring sheet according to a 5-point Likert scale (1 = poor and 5 = excellent). Table 8 indicates the mean ratings of each validation aspect. Most of the validation items were highly rated

(mean ≥ 3.5). It should be highlighted that the ‘degree of objectivity’ and ‘degree of replicability’ received very high ratings. The validation results have confirmed that the PPI model could improve the comprehensiveness, objectivity, reliability, and practicality when a construction partnering project is assessed.

6. Design of a computerized partnering performance monitoring system After developing a series of KPIs, a PPI, a set of relevant QIs, and their associated Fuzzy QRs, a computerized PPI model was developed. With the advancement of information technology and the wide application of the Internet, an automated CPPIS was designed. By doing so, the Internet-based CPPIS allows instant access, data collection and analysis, and simultaneous dissemination of the collated data. In addition, it can remove geographic barriers and reduce time in transferring data [16,17]. In fact, the usage of the Internet facilitates exchange of a huge amount of information at very high speed but with low cost. With the support of the Internet, project managers or their representatives can enter data of their current partnering performance via the Internet instead of faxes, postal mails and emails. The web-based CPPIS can also help reduce human and mathematical errors because data can be directly entered by project participants and data collection and analysis is then performed by the computerized system instead of the manual operation of project managers.

Fig. 2. Data input: Project Details (Part I).

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

6.1. Choice of programming languages The first task to design the CPPIS is to choose an appropriate programming language. Three major programming languages are available, encompassing Active Server Page (ASP), Java Server Page (JSP) and PHP (PHP Hypertext Pre-processor). The CPPIS has to provide database support, user-friendly interface, good security system, stable and speedy Internet connection. PHP Version 4.3 was chosen to develop the CPPIS because it requires comparatively less resources and it is free of charge [18]. PHP is a widely-used Open

1107

Source general-purpose scripting language that is particularly suited for web development and can be embedded into HTML [18]. In fact, it is an advanced programming language, which can facilitate interactive interface and supports powerful database. In order to support the PHP in establishing a database for data collection, storage and retrieval, an Open Source Database known as ‘MySQL’ was adopted [19]. MySQL is a relative database management system (RDBMS). The program runs as a server providing multi-user access to a number of databases [19]. It is a popular Open Source Database, which is designed for speed, power and precision in mission critical and heavy

Fig. 3. Data input: Project Details (Part II).

1108

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

load use. There are two major benefits of using it. The first one is that it can facilitate fast searching and the second benefit is to enable short processing time. Therefore, it can manage a large amount of data with high stability and reliability with low cost. These merits are crucial to the CPPIS. The following sections illustrate the usage of the CPPIS in practice. The illustration focuses on five major areas, including: (1) User Interface; (2) Project Details; (3) Web-based Questionnaire Survey; (4) Graphical Presentations for Project Control, Monitoring and Benchmarking; and (5) Updated Automatic Cumulative Database together with Categorization Functions. The CPPIS is operated through a user interface from which access is via the Internet domain address (http://yeungw.rdcw.com/demo/ppi/ login_admin.php). By entering a valid username and authorized password, the user can access to the different built-in functions. After logging into the CPPIS, a total of three major pull down menus will be displayed at the top of the first page, including: (1) Data Input; (2) Data Output; and (3) Help. It should be highlighted that a total of five major pull down menus are displayed at the top of the first page for the project administrators (the research team). Three of the five functions are same as users and the different two are: (1) Data Search/Edit; and (2) User Management.

6.2. Project Details There are two Project Details (Parts I and II) icons under the Data Input pull down menu. The first icon (Project Details: Part I) links with the respondent's details. These include: (1) type of organization in which the respondent is working; (2) size of his/her organization; and (3) his/her experience in participating partnering projects (Fig. 2). The second Project Details icon (Project Details: Part II) links with the project details. These include: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

project name (optional); type of client organization; project nature; total contract duration; total contract sum (HK$); type of procurement system; type of tendering method; form of contract used; total number of partnering workshops held; involvement of facilitator for partnering workshops;

Fig. 4. Web-based questionnaire survey form regarding time performance.

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

(11) mechanism to monitoring partnering performance; (12) partnering arrangement specified/included in the contract (yes/no); (13) party introducing to use partnering scheme; and (14) stage of partnering implementation (Fig. 3). The ‘Project Details’ icons allow the project manager to insert specific project details. These project details are useful as the base for retrieval of project information.

1109

boxes according to the project participants' assessment on the achievement of the partnering measures, the completed questionnaire survey will be automatically transmitted to the MySQL database. Data stored within the database is partitioned in accordance with the project details. Fig. 4 shows the web-based questionnaire survey form regarding time performance under the default environment. There is a ‘Help’ icon labeled after the brief description of time performance. After clicking on the ‘Help’ icon, the calculation of time performance (time variation) will be shown. Similar icons have been designed for other Performance Indicators.

6.3. Web-based questionnaire survey 7. Dissemination of analyzed data through graphical presentations Having completed the project details successfully, the web-based questionnaire survey under the Data Input pull down menu is ready for use. After entering the authorized user login and password, project managers can access to the web-based questionnaire survey. The questionnaire can be completed by marking on the appropriate circles or/and providing exact numerical values. After choosing the click

The CPPIS allows the project managers to measure, monitor, improve and benchmark the partnering performance of their current construction projects. Charts and figures can be generated instantly from the MySQL based database. After clicking on the ‘Graph’ icon under the Data Output pull down menu, the PPI formula will appear at the top of the

Fig. 5. A graph showing the PPI score of an inputted partnering project together with its relative position.

1110

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

screen. On the left hand side of the screen, there will be a graph showing the overall performance level of the inputted partnering project (the PPI score), together with its relative position (e.g. 28th percentile) in comparison with the performance of other partnering projects stored in the database (Fig. 5). By doing so, a benchmark is developed and the performance of different partnering projects can be compared objectively. The seven KPIs icons are shown on the right hand side of the screen. After clicking on each of them (for example: time performance), a graph will be generated with the following results (Fig. 6): (a) actual performance level of individual KPI of the inputted partnering project (e.g. on time); (b) relative performance level of the inputted partnering project when compared with other partnering projects using Fuzzy Set Theory (e.g. good time performance); and (c) benchmarking (e.g. standing at the 82nd percentile in the database).

7.1. Useful icons under ‘Help’ pull down menu There are a plethora of useful icons under ‘Help’ pull down menu and they provide further information regarding the CPPIS. They include: (1) About Us; (2) KPIs; (3) Quantitative Indicators (QIs); (4) Quantitative Requirements (QRs); (5) Research Approach; and (6) Glossary. The ‘About Us’ icon provides the details of contact person and the research team members. The ‘KPIs’ icon provides the seven identified KPIs, together with their corresponding weightings. The ‘QIs’ icon shows the full descriptions of the 7 QIs. The ‘QRs’ icon indicates the results of Fuzzy QRs for each QI using Vertical Error Method, Horizontal Error Method, and Bisector Error Method. The ‘Research Approach’ summarizes the research methods adopted in this research study, including: (1) literature review; (2) Delphi questionnaire survey; (3) empirical questionnaire survey; (4) face-to-face interviews; (5) concordance analysis; and (6) Fuzzy Set Theory. The ‘Glossary’ icon provides a brief description of Delphi survey method, concordance analysis and Fuzzy Set Theory.

Fig. 6. A graph showing the time performance of the inputted partnering project.

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

Fig. 7. The ‘Overall Data’ icon enabling the project administrators to ‘Activate’, ‘Save’, ‘De-activated’ and ‘Delete’ the information of any partnering projects stored in the database.

1111

1112

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113

8. Data verification and validation

10. Conclusions

After demonstrating all the key functions of CPPIS for users, some vital functions of CPPIS for project administrators will be introduced in this section, especially for data verification and validation. In fact, in addition to the three functions designed for users, there are two other different functions designed for administrators. They include: (1) Data Search/Edit; and (2) User Management. After pulling down the Data Search/Edit menu, there are 4 functions/icons, which are used for managing the database. They encompass: (1) Project Details; (2) Consolidated Data; (3) Individual Data; and (4) Overall Data. The ‘Project Details’ icon shows the project details of each inputted partnering project. The ‘Consolidated Data’ icon provides grouping function for the partnering projects, such as Private, Government and Quasi-government. The ‘Individual Data’ icon enables the administrators to select and view the details of individual partnering project. The ‘Overall Data’ enables the administrators to view the details of all partnering projects. After viewing the project details of all partnering projects, the administrators can decide to activate, save, de-activate and delete the information of any partnering projects stored in the database. If some ‘abnormal’ data is found, the administrators can contact the respondent for further verification and then decide whether the data is activated, saved, de-activated or deleted in the database (Fig. 7). By doing so, all the data can be verified and validated. This is a very vital function because it ensures that all the data entered are valid and correct. In addition, the administrators can list out all users and add or delete any users through the User Management menu.

The design and usage of PPI model formalize the partnering monitoring and benchmarking procedures. By combining it with the Computerized Partnering Performance Index System (CPPIS), they provide an effective and efficient tool to measure, monitor, improve and benchmark the partnering performance of construction projects through an openly accessed, web-based platform where authorized project participants can access the CPPIS at any time and at any location. In fact, the CPPIS is a very powerful tool for monitoring and benchmarking partnering projects. It reduces a lot of time, cost and efforts for collecting raw project data and disseminating useful analyzed data through useful graphs. In addition, it assists project managers and administrators to evaluate the partnering performance of construction projects in an effective way within a very short period of time. The usage of CPPIS enables the project managers to compare and present data in user-friendly graphs and figures. By doing so, it helps them to identify critical success factors for partnering projects and to resolve problems identified after data analysis. Therefore, the managerial and leadership skills of project managers are likely to be enhanced.

9. Significance and limitations of the research study This research study has made a significant contribution to the research area of construction partnering in that it has developed a comprehensive, objective, reliable and practical web-based Computerized Partnering Performance Index System (CPPIS) to measure, monitor, improve, and benchmark the partnering performance of construction projects in Hong Kong. The development of the PPI Model enhances the understanding of clients, contractors and consultants in implementing a successful partnering project. The design of the CPPIS further enhances the practicality and applicability of the PPI Model. Project team members can just input their individual project information and data and the CPPIS can directly provide a PPI score to compare their project performances with other counterparts. Both clients and contractors can use the model for monitoring purposes when a partnering-based project is implemented at the very early beginning of the construction phase. And the results can be further applied to compare the partnering performance of a project with its counterparts for benchmarking purposes. However, there is a limitation of this research study in that it is likely that the variability of project nature could affect the applicability of PPI. For instance, a particular range of cost savings may be appropriate to assess one project type but less appropriate for another. Therefore, it is vital to note that project and environmental specifics at the time may have significant effect on the adopted ranges of QIs. For example, the effect of different project sizes, such as small, medium-sized and large-scale partnering projects; as well as different project natures, such as private, public and infrastructure sector partnering projects, may have an impact on the success of the partnering project. Although the CPPIS was primarily developed in Hong Kong, it can be taken as a prototype and the same research methodologies can be replicated to develop similar performance evaluation models in other geographic locations in the future for international comparisons. By doing so, different performance evaluation models, such as private, public and infrastructure sector partnering projects locally and internationally can be developed and compared to identify their similarities and differences.

Acknowledgements The work described in this paper was fully supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (RGC Project No. PolyU 5158/04E). Special thanks are devoted to Mr. Raymond Yeung, Technical Support Assistant of the Department of Industrial and Systems Engineering, The Hong Kong Polytechnic University who assisted the research team to design the Computerized Partnering Performance Index System. Gratitude is also extended to those industrial practitioners for their kind cooperation and generous contributions to the interviews and a set of Delphi and a separate empirical questionnaire surveys launched in Hong Kong. References [1] A.P.C. Chan, D.W.M. Chan, K.S.K. Ho, An analysis of project partnering in Hong Kong, Research Monograph, Department of Building and Real Estate, The Hong Kong Polytechnic University, ISBN: 962-367-363-9, October 2002, 96 pp. [2] T.G. Crane, J.P. Felder, P.J. Thompson, M.G. Thompson, S.R. Sanders, Partnering measures, Journal of Management in Engineering, ASCE 15 (2) (1999) 37–42. [3] S.O. Cheung, H.C.H. Suen, K.K.W. Cheung, An automated partnering monitoring system—Partnering Temperature Index, Automation in Construction 12 (3) (2003) 331–345. [4] T. Lo, P.S.P. Wong, S.O. Cheung, Using Balanced Scorecard (BSC) approach to measure performance of partnering projects, Surveying and the Built Environment 17 (1) (2006) 45–57. [5] J.F.Y. Yeung, A.P.C. Chan, D.W.M. Chan, L.K. Li, Development of a Partnering Performance Index (PPI) for construction projects in Hong Kong: a Delphi study, Construction Management and Economics 25 (12) (2007) 1219–1237. [6] J.F.Y. Yeung, A.P.C. Chan, D.W.M. Chan, Establishing quantitative indicators for measuring the partnering performance of construction projects in Hong Kong, Construction Management and Economics 26 (3) (2008) 277–301. [7] J.F.Y. Yeung, Developing a Partnering Performance Index (PPI) for construction projects: a Fuzzy Set Theory approach, Unpublished PhD Thesis, Department of Building and Real Estate, The Hong Kong Polytechnic University, HKSAR, 345 pp. [8] R. Fellow, A.M.M. Liu, Research Methods for Construction, Blackwell Science Ltd, Oxford, United Kingdom, 1997. [9] D. Baloi, A.D.F. Price, Modeling global risk factors affecting construction cost performance, International Journal of Project Management 21 (4) (2003) 261–269. [10] A.P.C. Chan, E.H.K. Yung, P.T.I. Lam, C.M. Tam, S.O. Cheung, Application of Delphi method in selection of procurement systems for construction projects, Construction Management and Economics 19 (7) (2001) 699–718. [11] N. Schwarz, Cognition and Communication: Judgmental Biases, Research Methods, and the Logic of Conversation, Lawrence-Erlbaum, Mahwah, New Jersey, 1996, pp. 43–46. [12] J. Neter, M. Kutner, C. Nachtsheim, W. Wasserman, Applied Linear Statistical Models 5th Ed, McGraw-Hill, 2005. [13] S. Weisberg, Applied Linear Regression 3rd Ed, John Wiley and Sons, 2005. [14] F. Morrison, The Art of Modeling Dynamics Systems: Forecasting for Chaos, Randomness and Determinism, John Wiley & Sons, New York, 1991. [15] H.B. Griffiths, Mathematics of Models: Continuous and Discrete Dynamics Systems, Ellis Horwood Ltd, New York, 1993. [16] R.H. Lloyd, G.S. Boughton, K.C. Siu, A project information management system, providing greater collaboration and improved communications on major

J.F.Y. Yeung et al. / Automation in Construction 18 (2009) 1099–1113 construction works, Proceedings of International Conference on Construction, Hong Kong, vol. 1, 2001, pp. 462–473. [17] Z.M. Deng, H. Li, C.M. Tam, Q.P. Shen, P.E.D. Love, An application of Internet-based project management system, Automation in Construction 10 (2) (2001) 239–246. [18] P. Olson, PHP Manual, the PHP Documentation Group, 2009. [19] MySQL, available at http://www.mysql.com, 2009.

1113

[20] S.T. Ng, D.T. Luu, S.E. Chen, K.C. Lam, Fuzzy membership functions of procurement selection criteria, Construction Management and Economics 20 (2002) 285–296. [21] L.K. Chow (2005). “Incorporating fuzzy membership functions and gap analysis concept into performance evaluation of engineering consultants—Hong Kong study,” Unpublished PhD thesis, Department of Civil Engineering, The University of Hong Kong, HKSAR, 225 pages.