ARTICLE IN PRESS
Building and Environment 42 (2007) 2344–2359 www.elsevier.com/locate/buildenv
Toward a typology of construction mediator tactics Tak Wing Yiu, Sai On Cheung, Chau Ha Cheung Construction Dispute Resolution Research Unit (CDRRU), Department of Building and Construction, City University of Hong Kong, 83 Tat Chee Avenue, Hong Kong Received 17 February 2006; accepted 5 May 2006
Abstract The use of assisted negotiation plays a pivotal role in dispute settlement. Mediation is a notable example and has gained popularity because of the cost and time advantages. In a mediation, the tactics used by the mediator can be decisive. Construction has been described as a litigious industry and despite the vast numbers of mediation conducted, research on mediator tactics remains few. This paper reports a study directed at developing a typology of tactics used in construction mediation. Mediator tactics and mediation outcomes were firstly long listed from review of literatures. Data on the usefulness of tactics used and the outcome achieved were solicited from practicing mediator. Exploratory Factor Analysis (EFA) and Cluster Analysis (CA) were then used to identify the group structures. The results obtained from these two statistical tools were compared using the approach suggested by Newby et al. [Associations of empirically derived eating patterns with plasma lipid biomarkers: a comparison of factor and CA methods. American Journal of Clinical Nutrition 2004;80:759–67]. As such, the group structure obtained from the EFA was preferred than those from the CA. This finding supports the general observation that the structure of EFA is more adaptive. The grouping is further confirmed by a structural equation modeling (SEM) and provides a convenient platform for further research in construction mediation. The potential applications of the typology include effectiveness of construction mediator, behavior and strategic choice of tactics. Furthermore, with a typology of construction mediator tactics, the application can also be extended to study the responses of disputants respective to mediator tactics. It is believed that these applications are of both academic and practical value to the development of construction mediation. r 2006 Elsevier Ltd. All rights reserved. Keywords: Construction mediation; Mediator tactics; Typology
1. Introduction Mediation is a form of assisted negotiation where a mediator plays a pivotal role in facilitating settlement. As such, tactics used by mediators have been studied extensively by industrial specialists [1–2] and social scientists [3–11]. Typically, these studies examined the behaviors of mediators and their effect on the settlement process. Collectively, the roles of mediators were better understood as harvested from these studies. Nonetheless, there is a cogent need for a theoretical framework to further researches in mediation such as examining the interaction between mediators and disputants. Previous effort in this respect can be found in the exploration of mediator tactics typologies conducted by social scientists Corresponding author. Tel.: +852 34426779; fax: +852 27887612.
E-mail address:
[email protected] (T.W. Yiu). 0360-1323/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.buildenv.2006.05.005
[3,11–17]. Table 1 presents a summary of mediator tactics typologies derived from these studies. These typologies have put mediator tactics in perspective and thus facilitated a useful platform to further researches in mediation. For example, Bercovitch and Wells [18] suggest that categorizing mediator tactics is the first step to analyze mediation and shall provide a theoretical base for mediation researches. This paper reports a study that aimed to add to the portfolio of mediator tactics typologies but with a construction specificity. 2. Toward a typology of construction mediator tactics Construction projects are fluxed with disputes because of the technical and contractual complexities. In the last two decades, the use of mediation has received considerable acceptance by the construction industry in Hong Kong. This is because in mediation, the mediator aims at
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
2345
Table 1 Typologies of mediator tactics Citation
Typology
Description
Kressel [3]
(1) Directive tactics
Promoting specific solution to pressure or manipulate the parties directly into ending the dispute. The mediator attempts to orient himself to dispute and to establish the groundwork upon which his later activities will be built. Increasing the probability to achieve a mutually acceptable solution to the dispute with minimum manipulation or suggestion from the mediator.
(2) Reflective tactics (3) Nondirective tactics Bartunek et al. [12]
(1) Content tactics (2) Process tactics
Altering interactions of parties. Changing perception of parties.
Kochan and Jick [13]
(1) Non-contingent tactics (2) Contingent tactics
Non-contingent tactics are process-oriented. Contingent tactics are more active.
Kressel and Pruitt [4]
(1) Reflective tactics (2) Contextual tactics (3) Substantive tactics
Similar to the reflective tactics suggested by Kressel [3]. Altering the climate and conditions prevailing between the parties. Dealing directly with the issues in dispute.
Stein [14]
(1) Incremental tactics (2) Comprehensive tactics
Evolving a compromise in negotiation in which goals are narrowly defined, segmented and limited. Dealing with all issues in the dispute through the use of ‘single negotiation text’.
(1) Communication
Searching, supplying and clarifying information
(2) Formulation (3) Manipulation
Designing to help the mediator gain and retain control over the process of interaction. Changing the decision-making process of the parties.
(1) Integration (2) Pressing
(4) Inaction
Finding a solution that satisfies the parties’ goals and aspirations. Changing the goals or aspirations of the parties through the use of threats or minor sanctions. Promising rewards or positive benefits in exchange for concessions or agreement to compromise. Withdrawing from conflict and allowing the parties to handle the dispute by themselves.
(1) Facilitative
Taking the parties through a formatted process.
(2) Transformative (3) Evaluative (4) Solution-based
Realizing parties’ needs, wants and growth. Evaluating the merits of the parties. Examining the desired solutions of the parties.
(1) Disputants oriented (2) Disputant-disputant relationship oriented (3) Disputant-third party relationship oriented
Creating a platform under which resolution to be exercised. Choosing a gap between the disputants.
Communication-facilitation tactics
Adopting a fairly passive role, channeling information to the parties, facilitating cooperation but exhibits little control over the more formal process or substance of mediation. Exerting a more formal control over the mediation process with respect to the environment of the mediation. Affecting the content and substance of the bargaining process by providing incentives for the parties and changing their motivational calculus.
Touval and Zartman [11]
Carnevale [15]
(3) Compensation
Maurice and Robertson [68]
Wall et al. [16]
Bercovitch and Derouen [17]
Procedural tactics Directive tactics
Directing towards the disputants with the aim of enhancing a settlement.
facilitating voluntary agreement through a collective decision-making process instead of engaging in a win–lose fight [19]. This generally involves evaluating alternatives and making suggestions to the disputing parties directing toward a consensual agreement [20,21]. Within this framework, mediation has gained popularity in resolving construction dispute in Hong Kong [22]. The first formal introduction of mediation in Hong Kong dated back to the mid-1980s when mediation was included in the standard forms of contract for building works used for Hong Kong
Government projects as part of the contract dispute settlement procedure. Mediation was also used in the Quasi-government infrastructure projects like the Chek Lap Kok Airport. The conditions of contract for these projects employed mediation as the first of a 3-tier dispute resolution procedure. More recently, the Mediation Council of the Hong Kong International Arbitration Centre has launched a pro Bono Mediation Scheme aiming to promote the use of mediation to resolve construction disputes of smaller sum.
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
2346
2.1. An inspiration from the experience of social science With its increasing popularity, the study of construction mediation offers notable academic and practical value. With reference to the typologies of mediator tactics developed from previous studies as summarized in Table 1, it would be invaluable to construction mediation research to develop a construction specific mediator tactics typology in view of the extensive use of mediation to settle construction disputes. The proposition that a construction specific typology of mediator tactics shall provide a theoretical framework for mediation research is also generally supported by the experienced mediation researchers in the field of social science. The typologies of mediator tactics proposed in several studies were subsequently used as a fundamental basis for a number of notable mediation researches. Examples of these developments are illustrated in Fig. 1. As shown in Fig. 1, the mediation researches employing the mediator tactics typologies of Kochan and Jick [13], Bartunek et al. [12] and Kressel [3] are outlined. Kochan and Jick [13] categorized mediator tactics as either noncontingent or contingent. With this classification, the contingent use of tactics was explored by Shapiro et al. [23] who suggested that the effectiveness of mediator tactics is contingent to the nature of the dispute. Other studies, mostly in the field of labor disputes, had also achieved respectful results. For example, it is found that the use of forceful, substantive tactics was positively related to settlement when the hostility level is high [24–27]. Lim et al. [27] found that active mediator strategies like pressuring tactics are useful when conflict is tense. Moreover, Zubek et al. [28] reported that tactics such as challenging the parties to come up with ideas, requesting reactions to new ideas, suggesting new ideas was more
likely to reach agreement under condition of high hostility and little joint problem solving by the disputants. In contrast, if the disputants share a joint problem-solving attitude, agreement are more likely to be reached without the application of such tactics. The contingent use of mediator tactics was further considered in relation to the gender of the mediators. Eagly et al. [29] reported that males tend to be more task oriented than female and the use of pressing tactics from male mediators is more frequent than female mediators. On the other hand, women tend to be more socioemotional and concerned about approval from others. These studies generally concluded that tactics are the most essential tool of a mediator and its appropriate use shall have decisive effect on the mediation outcomes [1,3]. On the other hand, the typology developed by Bartunek et al. [12] covers a wider range of mediator behaviors [18]. The typology of Bartunek et al. [12] classifies mediator tactics into two types: content and process. Content tactics aim at altering interactions of parties (e.g. setting the agenda, suggesting concessions and deflating excessive demands), while Process tactics aim at changing the perception of parties (e.g. acting as communication link, educating the disputants in dispute resolution technique and providing save-saving mechanism). Among the typologies of mediator tactics cited, the work of Kressel [3] was the most commonly referenced. This typology categorizes mediator activities into three groups: directive, reflective and non-directive. Directive tactics promote specific solution and may employ pressure or manipulation with the aim of ending the dispute. Reflective tactics attempt to orient mediator to the dispute and establish groundwork upon which his later activities will be built. While non-directive tactics aim at increasing the probability to achieve mutually acceptable solution.
Mediator Tactics Typology fromKochan and Jick [13]
Shapiro etal. [23]
Eagly et al. [29]
(a)
Hiltrop [24-25], Zubeket al. [28] Donohue [26]
Lim et al. [9]
Mediator Tactics Typologyfrom Bartunek et al. [12]
(b)
Bercovitch and Wells [18]
Mediator Tactics Typology from Kressel [3]
(c)
Kressel and Pruitt [4]
Carnevale [15]
Carnevale et al. [31]
Lim et al. [27]
Esser and Marriott [35]
Fig. 1. Research developments from the typologies of (a) Kochan and Jick [13], (b) Bartunek et al. [12] and (c) Kressel [3].
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
However, the weakness of this typology is the difficulty of drawing precise lines between reflective and non-directive tactics, which appear to overlap in certain instances [18]. As such, Kressel’s [3] work was further expanded by Kressel and Pruitt [4], who identified another three categories of mediation tactics; reflexive, contextual and substantive. This typology was subsequently augmented by three major mediation studies: firstly, Carnevale [15] identified four fundamental tactics available to mediators. These are integration, pressing, compensation and inaction. Based on these, a model of mediator strategic choice was developed to predict the behavior of mediators’ choice of tactics. This model was supported and further extended by other researchers [9,30–34]. Secondly, the typology of Kressel and Pruitt [4] underpinned a research that examined the contingencies of mediator tactics [9,31]. This typology generalizes mediator behavior to form a taxonomy. The findings of this study indicated that some mediator tactics are effective in certain dispute types and not others. This provides a strong empirical support to the contingent use of mediator tactics perspective to the dispute sources. Finally, Esser and Marriott [35] applied the typology of Kressel and Pruitt [4] to compare the effectiveness of substantive and contextual mediator tactics in laboratory setting. The above collectively point to the proposition that the implementation of a typology of mediator tactics can facilitate further research on construction mediation. This paper reports a study for this purpose and the research design is shown in Table 2. As a brief introduction, the study involved four stages of work. Stage I was background studies backed by a review of literatures on mediation. Commonly used mediator tactics and mediation outcomes were long-listed. Based on these findings, a questionnaire was prepared to collect data. In Stage II, a questionnaire survey was conducted to solicit
2347
case-specific data from a panel of construction mediators. Stage III of the study explores the group structure of mediator tactics; two classification methods: Exploratory Factor Analysis (EFA) and Cluster Analysis (CA) were applied to the collected data. In Stage IV, the authenticity of the typology structure obtained in Stage III was tested by performing confirmatory factor analysis (CFA) using the structural equation modeling (SEM) methodology [49]. 3. Stage I: background study For the background study, a literature review on mediator tactics [6,31,36–42] and mediation outcomes was conducted [31,43–48]. These reviews identify a total of 32 mediator tactics, among which, eight are related to disputant’s perception, 13 are related to mediation procedure and the remaining 11 are related to settlement. In addition, a total of 16 mediation outcomes are also identified. The lists of mediator tactics and mediation outcomes arranged as described are shown in Tables 3 and 4, respectively. Based on these, a questionnaire for data collection was developed. 4. Stage II data collection For the purpose of this study, respondents were required to provide case specific data by reference to the most recently completed mediation cases. The questionnaire has three main sections. For the first section, the respondents were required to provide their background information; the particulars of the mediated cases such as the project nature, contract sum and the parties involved. The remaining two sections are designed to address mediator tactics and mediation outcomes. The respondents were asked to rate the degree of usefulness of mediator tactics on a Likert
Table 2 Toward a typology of construction mediator tactics: overall research plan Stages of study
Research tasks
Approach
Outcomes in relation to objectives
Stage I: Background study
Review of previous literature Long-listing of mediator tactics and outcomes
Desktop analysis
Questionnaire for the use of data collection
Stage II: Data collection
Use the expertise of construction mediators, collecting data for their used tactics and mediation outcomes
Establish a panel of construction mediators to respond the questionnaire.
A completed set of data for empirical analyses.
Stage III: Date analyses
Explore the group structures of mediator tactics used by the mediator.
Two classification methods; Exploratory Factor Analysis (EFA) and Cluster Analysis (CA) are used, and Comparing from EFA and CA.
A selected typology of mediator tactics explored by either EFA or CA.
Stage IV: Confirmatory analysis
Confirm the typology of mediator tactics.
Structural equation modeling (SEM).
A typology of construction mediator tactics, and Implications on the typology of mediator tactics in construction industry
ARTICLE IN PRESS 2348
T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
Table 3 List of mediator tactics Mediator tacticsa Disputants’ perception relatedb 1. Educate the parties about the bargaining or impasse process (T_EDUCATE) 2. Encourage the parties themselves to verbalize their willingness to respectfully listen to each other’s grievances (T_LISTEN) 3. Encourage the parties to meet each other’s needs (T_NEEDS) 4. Help the parties to ‘‘save face’’ (T_SAVESFACE) 5. Remind the parties that their position was unrealistic (T_UNREAL) 6. Suggest particular settlement for parties to consider (T_SUGG_SETTLE) 7. Try to change the expectation of parties (T_EXPECT) 8. Encourage the parties to apologize, and regret for harm suffered by another in the past (T_APOLOGIZE) Mediation procedure relatedb 1. Use humour to lighten the atmosphere (T_HUMOUR) 2. Keep in rapport with the parties (T_RAPPORT) 3. Argue one party’s case to the other (T_ARGUE) 4. Reduce the feeling of hostility towards each party (T_HOSTILITY) 5. Focus on the impasses issues during caucuses session (T_IMPASSES) 6. Attempt to speak their language (T_LANGUAGE) 7. Control the bargaining structure and timing (T_CONTROL) 8. Formulate clear goals before or during the process (T_GOALS) 9. Call for frequent caucuses during mediation (T_CAUCUSES) 10. Avoid taking sides on important issues in joint sessions (T_TAKES SIDES) 11. Assure each party that the other was being honesty (T_ASS_HONESTY) 12. Keep the negotiations focused on the issues only (T_NEGO) 13. Express pleasure or displeasure at negotiation progress (T_PLEASURE) Settlement relatedb 1. Settle simple issue first (T_SIMPLE) 2. Help the parties to devise a framework for negotiations (T_FRAMEWORK) 3. Make compromise suggestions to the parties (T_SUGGEST) 4. Suggest the parties to review their needs (T_REVIEW) 5. Mention the costs of disagreement (T_COSTS) 6. Simplify the agenda by eliminating or combining issues (T_COMB_ISSUES) 7. Discuss other settlements in comparable cases (T_COST_SETTLE) 8. Help the parties to establish priorities among the issues (T_PRIORITIES) 9. Suggest some tradeoffs among issues (T_TRADEOFFS) 10. Press the parties to make concessions (T_PRESS) 11. Make the parties to aware the destructiveness of the conflict (T_DESTRUCT) a
Mediator tactics were rated on a scale from (1) least useful to (5) most useful. Kerr [36]; Carnevale et al. [31]; Douglas [37]; Perez [38]; Stevens [39]; Pruitt [6]; Karim and Pegnetter [40]; Eiseman [41]; Young [42].
b
Table 4 List of mediation outcomes Mediation outcomesa,b 1. Agreement perceived to be devised from the parties (O_AGREE) 2. The parties gained satisfaction on the mediation as a tool of dispute resolution (O_SAT_MED) 3. Overall success (O_SUCCESS) 4. I felt the parties trust the mediator (O_MED_TRUST) 5. The underlying core conflict of the dispute was resolved (O_RESOLVE_CORE) 6. A mutually beneficial settlement was reached (O_REACH_SETTLE) 7. The needs and goals of mediator satisfied (O_SAT_GOAL) 8. The number of issues was reduced (O_ISSUES) 9. Nothing ambiguously stated (O_NOAMB) 10. The settlement was reached in reasonable time (O_TIME) 11. The inter-party relations improved (O_RELATION) 12. I acquired a reputation for the effectiveness in setting the dispute (O_REPUT) 13. Both parties felt no future problems expected (O_NOPROB) 14. Both parties learned to communicate (O_COMM) 15. I improved self-esteem after the settlement of the dispute (O_SELF_ESTEEM) 16. I improved my cultural sensitivity (O_CULTURE) a
Carnevale et al. [31]; Johnson et al. [43]; Bush [44]; Depner et al. [45]; Touval [46]; Day-Vines et al. [47]; Shulman [48]. Mediation outcomes were rated on a scale from (1) not achieved to (5) highly achieved.
b
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
scale of 1(least useful)–5 (most useful). Likewise, the mediation outcome achievements were rated on a Likert scale of 1 (not achieved)–5 (highly achieved). Before sending out the data collection questionnaire, a list of prospective respondents was first prepared. The accredited mediators registered at the Mediation Council of the Hong Kong International Arbitration Center (HKIAC) were the targeted respondents. The HKIAC is the leading organization for the provision of arbitration and mediation services in Hong Kong. Furthermore, to ensure relevancy of the responses, only those mediators with construction background were sent the questionnaire. As such, 85 construction mediators formed the target group. They were contacted first for their agreement to participate in the study. A total of 32 construction mediators responded positively and provided data for the study. The respondents are well-respected senior members of the industry and all have at least 5 years experience in construction mediation. The project nature of the mediated disputes are civil (50%), building (35%) and building services and maintenance related (15%).
5. Stage III: data analysis In Stage III, two classification methods, EFA and CA, were used to explore the group structures of tactics and outcomes. EFA explores the relationship structure of the data seeking to define a set of common underlying constructs, known as factors. Separate dimensions of the structure can firstly be identified. Interpretation of variables can be accomplished by summarizing the data according to the constructs [61]. The data obtained for mediator tactics (3 groups) are each subjected to EFA. All the EFA results satisfied the statistical fitness criteria of Kaiser–Meyer–Olkin (KMO) and Bartlett’s Test (BT). As shown in Table 5, the KMO values for the EFA fall within the range of 0.560–0.682 and are all above the threshold requirement of 0.5 [50–52]. In addition, following the rule of thumb suggested by Comrey et al. [53] on the selection of variables, a factor loading value of 0.71 is considered a good demarcation for variable selection within factors. Accordingly, variables with loading less than 0.71 were discarded. The discarded variables are given in Appendix A
2349
of this paper. Based on these analyses, a typology of mediator tactics is shown in Table 6. CA, another commonly used classification method, was applied to the same data set. In CA, following the cluster procedure as suggested by Hair et al. [61], the structure of the collected data can be obtained. Similar to EFA, CA is also widely used to investigate the grouping of data [78] in pharmacy [54], geochemistry [55,56], human cognition [57] and construction [58,59]. Moreover, this technique is identified more as an art of finding groups in data [60]. Typically, there are two types of cluster procedures: hierarchical and K-mean (non-hierarchical). Hierarchical cluster procedure is a stepwise clustering procedure that starts with the individual objects, which firstly group the most similar objects and eventually all objects are fused into a single cluster. K-mean cluster procedure produces only a single cluster solution for a set of cluster seeds. These seeds are then used to group objects within a prespecified distance of the seeds [61]. Due to the constraint of K-mean clustering procedure of providing only single cluster solution [61], hierarchical cluster procedure is more frequently used to investigate inherent clusters [54,56,57,59]. To perform the hierarchical cluster procedure, an agglomerative algorithm called Ward’s method is employed. With this method, the distance between two clusters is the total sum of squares between the two clusters over all variables. The within-cluster sum of squares is minimized over all partitions obtainable by combining two previous clusters in the clustering procedure. Ward’s method therefore tends to combine clusters with a small number of observations [59,61]. As it can minimize the within-cluster differences and avoid problems with ‘‘chaining’’ of the observations found in the single linkage method [61]. Therefore, this method has been commonly used in cluster procedures [54,55,59]. The output of a CA is presented as a tree diagram called dendrogram, a graphical ‘‘structure’’ showing objects that are similar to each other. Cluster solution can be identified by observing the large change in the distance coefficient [61]. The dendrograms generated for the mediator tactics listed in Table 3 are shown in Fig. 2. In these dendrograms, mediator tactics are listed along the left vertical axis. The horizontal axis shows the distance between clusters. Distances between nodes of the dendrogram are proportional to the difference
Table 5 Test of data suitability in EFA Mediator tactics
Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy Bartlett Test of Approx. w2 Sphericity df sig
Disputants’ perceptions related
Mediation procedures related
Settlement related
0.682
0.642
0.560
83.304
175.523
103.523
28 .000
78 .000
55 .000
ARTICLE IN PRESS 2350
T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
Table 6 Factor structure of mediator tactics from EFA Construction mediator tactics Mediation procedures related
Settlement related
Disputant’s perception related
Factor 1: Process control Control the bargaining structure and timing
Factor 1: Ice breaking Settle simple issue first
Express pleasure or displeasure at negotiation progress Use humour to lighten the atmosphere
Help the parties to establish priorities
Factor 2: Caucuses Call for frequent caucuses during mediation
Make compromise suggestions to the parties Suggest some tradeoffs among issues
Factor 3: Trust building Avoid taking sides on important issues in joint sessions Keep in rapport with the parties Factor 4: Analyzing
Press the parties to make concessions Factor 3: Pressing Settlement
Factor 1: Encourage for self-improve Encourage the parties to meet each other’s needs Encourage the parties to apologize, and regret for harm suffered by another in the past Encourage the parties themselves to verbalize their willingness to respectfully listen to each other’s grievances Help the parties to ‘‘save face’’ Educate the parties about the bargaining or impasse process Factor 2: Reality test Try to change the expectation of parties
Factor 2: Seeking progress
Mention the costs of disagreement Make the parties to aware the destructiveness of the conflict
Argue one party’s case to the other
Fig. 2. Dendrogram for hierarchical cluster analyses of (a) mediation procedures-related, (b) settlement-related and (c) disputants’ perception-related tactics.
(dissimilarity) between mediator tactics. The agglomeration schedules of hierarchical cluster are also produced and presented in Table 7. To select the number of clusters, the commonly used criterion is to set the demarcation where there is a significant increase of the agglomeration coefficients [54,61]. Hair et al. [61] suggested identifying large relative increases in the cluster homogeneity by calculating the
percentage change in the agglomeration coefficients. Table 8 reports the percentage changes of coefficient for having 3–1 cluster, respectively. It is observed that the largest percentage change occurs in going from two to one cluster for the mediation procedures-related tactics, and the other noticeable changes in the percentage increase occur in going from three to two clusters for the settlement and disputant’s perception-related tactics. Hence, a two-cluster
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
2351
Table 7 Agglomeration schedule of hierarchical cluster analyses (adopted by Hair et al. [61]) Stage
Cluster combined Cluster 1
Coefficients Cluster 2
Stage cluster first appears
Next stage
Cluster 1
Cluster 2
Mediation procedure related 1 1 2 6 3 5 4 3 5 3 6 1
2 7 6 5 4 3
14.00 33.00 68.00 123.75 188.80 395.43
0 0 0 0 4 1
0 0 2 3 0 5
6 3 4 5 6 0
Settlement related 1 1 2 6 3 1 4 2 5 2 6 1
3 7 5 4 6 2
25.00 62.50 101.50 147.00 221.00 321.14
0 0 1 0 4 3
0 0 0 0 2 5
3 5 6 5 6 0
Disputant’s perception related 1 2 2 5 3 4 4 2 5 1
3 6 5 4 2
14.00 30.00 68.00 115.20 189.33
0 0 0 1 0
0 0 2 3 4
4 3 4 5 0
Table 8 Analysis of agglomeration coefficient for the CA of mediator tactics (Adopted from Hair et al. [61]) Number of cluster
Agglomeration coefficients
Mediation procedure related 3 123.75 2 188.80 1 395.43
Percentage change in coefficient to next level
52.57 109.44a —
Settlement related 3 2 1
147.00 221.00 321.14
50.34a 45.31 —
Disputant’s perception related 3 2 1
68.00 115.20 189.33
69.42a 64.35 —
a
The largest percentage increase in coefficient to next level.
solution is considered appropriate for mediation procedure-related tactics and three-cluster solution is chosen for the settlement-related and disputant’s perception-related tactics. The details of cluster solutions can be found in Table 8. 5.1. Structure of mediation outcomes Similar to the forgoing procedures for identifying mediator tactics, EFA and CA were applied on the mediation outcomes as listed in Table 4. Following the
cluster procedures as suggested by Hair et al. [61], the resulting dendrogram is shown in Fig. 3. Furthermore, a two-cluster solution is considered appropriate for mediation outcome by analyzing the agglomeration coefficients (Table 9). However, the result obtained from EFA provides a more detailed classification with mediation outcomes are categorized into four groups. These results satisfied the statistical fitness criteria of KMO and BT. The KMO value is 0.685, which is above the threshold requirement of 0.5 [50–52]. In addition, the demarcation for variable selection of Comrey et al. [53] is also applied. Variables with loading less than 0.71 were discarded. The discarded variables are listed in Appendix A of this paper. The factor structure of mediation outcomes from EFA is shown in Fig. 5. The result obtained from EFA can be augmented by a CFA. SEM has been used for this purpose in a wide range of research areas [62–64]. SEM estimates a series of interrelated dependent relationships simultaneously, which explicitly account for the measurement errors in the variables [61], as such is a reliable tool to confirm factor structures. In this study, the software called AMOS 5 was used. In terms of the statisticality of a SEM model, AMOS offers five goodness-of-fit (GOF) indices [65]. These are: w2/df ratio, goodness-of-fit index (GFI), comparative fit index (CFI), root-mean-square error of approximation (RMSEA), and the Tucker–Lewis index (TLI). The summary of their threshold values is shown in Table 10. The factor structures are considered reliable if the following acceptance criteria are satisfied. Generally, the GOF indices of the initial models (Fig. 5a) are lower than the recommended acceptance criteria as
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
2352
shown in Table 10. According to the previous SEM studies [62,63], it is not uncommon that refinements to the initial model are needed. AMOS allows the use of modification indices to generate the expected reduction in the overall model fit w2 for each possible path that can be added to the model. The refined models are given in Fig. 4b. A summary of overall GOF measures is shown in Tables 11 and 12. In the final model, the GOF measures are of acceptable fit. w2s are statistically significant. The GFI reaches the
recommended level (i.e.40.8), indicating that the relative amount of variance and covariance in the data accounted for by the proposed model. The RMSEA is generally below the suggested threshold level and the TLI and CFI are higher than the recommended level of 0.90. Hence, the indices indicate that all three refined models are acceptable. In sum, the CFA statistically supports the factor solutions of mediation outcomes, which are shown in Table 12. 5.2. Grouping by factor and cluster analyses compared By observing both structures, the results of the EFA appear to be able to provide a more informative classification than the CA. To confirm this observation, the grouping results were further investigated by a two-step statistical approach suggested by Newby et al. [66], who studied food patterns devised from cluster and factor analyses with respect to five outcomes (plasma total cholesterol, LDL cholesterol, HDL cholesterol, the ratio of total HDL cholesterol and triacylglycerols) were compared by the following two steps:
Fig. 3. Dendrogram for hierarchical cluster analysis on mediation outcome.
Table 9 Analysis of agglomeration coefficient for the CA of mediation outcomes (adopted by Hair et al. [61]) Number of cluster
Agglomeration coefficients
Percentage change in coefficient to next level
3 2 1
338.63 411.75 523.38
21.59 27.11a —
a
The largest percentage increase in coefficient to next level.
Separate regression models were built for each individual factor or cluster for each outcome variable to examine the associations between the cluster (or the factor) and the outcomes (i.e. each mediation outcome as dependent variable and each factor (or cluster) as independent variable); and All factors were included in one model and all clusters were included in another model to see which model better predicted each of the outcomes (i.e. each mediation outcome as dependent variable and all the factors (or clusters) as independent variables).
In this connection, both structures are compared in terms of their association and prediction power to mediation outcomes. To achieve this, a table is used to summarize the contents of both group structures (Table 13 refers). Despite the methodological differences, both group structures are similar. In fact, those groupings for settlement-related tactics are identical. Hence, Newby et al. [66]’s approach is only applied to mediation proceduresrelated and disputant’s perception-related tactics. Similar to the approach employed by Newby et al. [66], mediation outcomes are used and acted as dependent variables. Firstly, linear regression is employed to examine individual associations between each cluster and each
Table 10 Fit Indices for structural equation model Goodness-of-fit measure
Acceptance criteria
References
w2/ degree of freedom Goodness of fit index (GFI) Root mean square error of approximation (RMSEA) Tucker–Lewis index (TLI) Comparative fit index (CFI)
Recommended level from 1 to 2 0 (no fit) to 1 (perfect fit) recommended Level40.8 o0.05 indicates a very good fit, the threshold level is 0.10 0 (no fit) to 1 (perfect fit) recommended Level40.9 0 (no fit) to 1 (perfect fit) recommended Level40.9
[63,69–70] [63,71–72] [64,71,73–75] [64,76] [75–77]
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
err-OWW
err-OWW
err-ors
err-oa
err-ors
O_REACH_SETTLE
O_REACH_SETTLE 0.82
0.77
0.66
0.62
err-oa
O_Win-win Settlement
O_AGREE
0.89 err-osm
O_SAT_MED
err-OP
O_Win-win Settlement
O_AGREE
-1.98
0.93 err-osm
2353
O_SAT_MED
err-OP
0.92
0.93 err-omt
err-orc
err-omt O_MED_TRUST
O_MED_TRUST
0.70 0.96
0.85 O_Progress
O_RESOLVE_CORE
err-orc
O_RESOLVE_CORE
0.58
0.82
O_Progress
0.58 0.86
0.79 err-ois
err-ois
O_ISSUES
Mediation Outcome
err-OI
err-ocu
Mediation Outcome
err-OI
0.51
O_CULTURE
O_ISSUES
err-ocu
O_CULTURE
err-oco
O_COMM
0.59 0.87
0.91 0.80 err-oco
O_COMM
0.82
O_Improvement
0.71
err-ose
err-ot
O_SELF_ESTEEM
O_TIME
O_Improvement 0.51
0.73
err-ose
0.69 err-ot
O_Time Advantage
err-OTA
O_SELF_ESTEEM
O_TIME
0.69 O_Time Advantage
err-OTA
(a)
(b) Fig. 4. (a) The EFA model and (b) the refined model of mediation outcomes.
Table 11 Factor structure of mediation outcomes from EFA
Table 12 Overall goodness-of-fit measures of structural equation model
Mediation outcomes
Goodness-of-fit measure
Factor 1: Win–win settlement A mutually beneficial settlement was reached Agreement perceived to be devised from the parties The parties felt satisfaction on the mediation as a tool of dispute resolution Factor 2: Progress I felt the parties trust the mediator The underlying core conflict of the dispute was resolved The number of issues was reduced Factor 3: Improvement I improved my cultural sensitivity Both parties learned to communicate I improved self-esteem after the settlement of the dispute Factor 4: Time advantage The settlement was reached in reasonable time
factor with each of mediation outcome variables (win–win settlement, progress, improvement and time advantage). Cluster 1 of disputant’s perceptions-related tactics can be
w2/ degree of freedom Goodness-of-fit index (GFI) Root mean square error of approximation (RMSEA) Tucker–Lewis index (TLI) Comparative fit index (CFI)
Structural equation models of mediation outcomes Initial model
Refined model
1.455 0.769 0.121
1.132 0.829 0.065
0.849 0.890
0.956 0.969
tested in 4 separate regression models to examine the associations between this cluster and the four mediation outcomes. Second step involves a final set of regression analysis, which includes all clusters in one model and all factors in another model, to examine which model better predicts each of the mediation outcomes. The results of the first step are summarized in Table 14. The coefficients of 44 regression models represent the individual associations between each cluster (and factor) with each of the mediation outcomes. Generally, the regression coefficients
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
2354
Table 13 The contents of typologies from EFA and CA Mediation procedures related tactics Cluster analysis
Factor analysis
Mediator tactics
Cluster 1 (Trust building)
Factor 3: Trust building
1. Keep in rapport with the parties 2. Avoid taking sides on important issues in joint sessions
Cluster 2
Factor 1: Process control
1. Control the bargaining structure and timing 2. Express pleasure or displeasure at negotiator progress 3. Use humor to lighten the atmosphere
Factor 2: Caucuses Factor 4: Analyzing
1. Call for frequent caucuses during mediation 1. Argue one party’s case to the other
Settlement related tactics Cluster 1 (Seeking progress)
Factor 2: Seeking progress
1. Make compromise suggestions to the parties 2. Suggest some tradeoffs among issues 3. Press the parties to make
Cluster 2 (Pressing settlement)
Factor 3: Pressing settlement
1. Mention the costs of disagreements 2. Make the parties to aware the destructiveness of the conflict
Cluster 3 (Ice breaking)
Factor 1: Ice breaking
1. Settle simple issues first 2. Help the parties to establish priorities among the issues
Disputants’ perceptions related tactics Cluster 1 Factor 1: Encourage to self-improvement
1. Encourage the parties themselves to verbalize their willingness to respectfully listen to each other’s grievances 2. Encourage the parties to meet each other’s needs.
Cluster 2
Factor 1: Encourage to self-improvement
1. Encourage the parties to apologize, and regret for harm suffered by another in the past 2. Educate the parties about the bargaining or impasse process 3. Help the parties to ‘‘save face
Cluster 3 (Reality Test)
Factor 2: Reality Test
1. Try to change the expectation of parties
Table 14 Regression coefficients (b) associating cluster and factor solution with mediation outcomes (adopted from Newby et al. [66]) Mediator tactics
Win–win settlement
Improvement
Progress
Time advantage
Mediation procedure related
C1 C2
.319 .513
F3 F1 F2 F4
.319 .517 .279 .225
C1 C2
.382 .170
F3 F1 F2 F4
.382 .238 .179 .046
C1 C2
.218 .251
F3 F1 F2 F4
.218 .559 .072 -.039
C1 C2
.185 .520
F3 F1 F2 F4
.185 .459 .258 .293
Disputant’s perception related
C1 C2 C3
.393 .520 .161
F1
.515
.034
.129
.219 .282 .075
.282
F2
C1 C2 C3
F1
.220
.476 .639 .129
.629
F2
C1 C2 C3
F1
.161
.052 .104 .220
F1
F2
C1 C2 C3
F2
.075
Where C1 ¼ Cluster 1, C2 ¼ Cluster 2, C3 ¼ Cluster 3 and F1 ¼ Factor 1, F2 ¼ Factor 2, F3 ¼ Factor 3, F4 ¼ Factor 4.
between each cluster and factor are similar. It is due to the fact that Cluster 2 of the mediation procedure-related tactics is the combination of Factor 1 (process control), Factor 2 (caucuses) and Factor 4 (analyzing), while Factor 1 (encourage to self-improve) of disputant’s perceptionrelated tactics is the combination of Clusters 1 and 2. These findings imply that the individual factor or cluster is having similar effect on the mediation outcome.
In the second step, the difference of the EFA and CA results can also be compared by observing the R2 of the separate regression model containing all of the factors and clusters as independent variables [66]. Owing to the similarity of these two results, the R2 of ‘all factors’ model is slightly different from that of ‘all clusters’ model (Table 15 refers). In general, the prediction power of ‘all factors’ model is therefore higher than the ‘all clusters’
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
2355
Table 15 R2 of two separate regression models containing all clusters or all factors, in predicting mediation outcome (adopted from Newby et al. [66])
‘All clusters’ Model ‘All factors’ Models
Win–win Settlement R2
Improvement R2
Progress R2
Time advantage R2
.546 .532
.474 .536
.491 .503
.405 .462
Table 16 Overall goodness-of-fit measures of structural equation model Goodness-of-Fit measure
Structural equation models of mediator tactics Mediation procedures related
Settlement related
Initial model
Refined model
Initial model Refined model Initial Model
Refined Model
1.207 0.927 0.082 0.902 0.953
1.751 0.848 0.156 0.632 0.755
1.312 0.937 0.100 0.929 0.927
2.069 w2/ degree of freedom Goodness-of-fit index (GFI) 0.821 Root mean square error of approximation (RMSEA) 0.186 Tucker–Lewis index (TLI) 0.496 Comparative fit index (CFI) 0.616
model. This is due to the fact that EFA partitions the variance from a single variable among the various factors, while CA does not. This seems to be a distinct advantage for EFA, since the results obtained from EFA is more accurate, efficient and yields more detail [67]. As a result, the group structure from EFA was selected. This choice is also in line with the proposition that the detail structure of EFA can be more adaptive. 6. Stage IV confirmatory analysis In Stage IV, SEM was employed to confirm the final structure of mediator tactics. Based on the results obtained from EFA, three SEM models are constructed as in Fig. 4 where the latent variables are ellipse shape and the observed variables are in rectangles. The resulting GOF indices of these initial models are shown in Table 16. For the initial models, their GOF indices are less than the recommended value when compared with the acceptance criteria in Table 10. Hence, model modifications are required to obtain better-fit models. The refined models are given in Fig. 5. In the final models, the GOF measures are acceptable statistically. In sum these CFA statistically confirms the grouping structure of typology of tactics in construction mediation developed from the EFA (Fig. 6). 7. Conclusions This paper aims at adding to the portfolio of mediator tactics typologies with one that is construction specific. With the use of two classification methods, (1) Exploratory
1.090 0.898 0.054 0.956 0.973
Disputant’s perception related
3.242 0.750 0.269 0.489 0.659
Factor Analysis (EFA) and (2) Cluster Analysis (CA), two group structures of mediator tactics were developed. The groupings derived from the EFA appear to be more informative. To further investigate this observation, the grouping structures obtained from the EFA were compared with that obtained from CA following the statistical approach suggested by Newby et al. [66]. This involved comparing the two groupings by their associations and prediction power to mediation outcomes. The associations between the cluster (or the factor) and the outcomes are firstly examined by building multiple regression models for each individual factor or cluster. Secondly, all factors are included in one regression model and all clusters are in another regression model to examine their respective prediction power. The result supports that the structure obtained from EFA offers higher predictive power when compared with that from the CA. To augment this finding and enhance the reliability of the proposed construction mediator tactics typology, the EFA structure, was further analyzed by the tool of structural equation modeling (SEM). The final models from SEM all passed the statistical test. Based on these final models, a typology of construction mediator tactics is proposed. This typology composes of three types of mediator tactics, mediation procedure-related, disputants’ perception-related and settlement-related tactics. Furthermore, the groupings of these types of tactics are depicted. The mediation procedurerelated tactics are grouped by four major generic techniques: analyzing, caucuses, trust building and process control. The disputant perception-related tactics are grouped by two major tactics: encourage self-improvement and reality
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
2356
e_IC
e_A err-arg
0.83
T_ARGUE
err-sim
Analyzing
T_SIMPLE 0.76
e_C Ice Breaking 0.81 err-cau
T_CAUCUSES
0.97
T_PRIORITIES
err-pri
0.71
Caucuses
0.71
e_SP e_TB
0.71 T_Settlement T_Mediation Procedure
err-rap
T_RAPPORT
0.98
T_SUGGEST
err-sug
0.72
0.62
0.99 Trust Building
0.81 err-tak
err-tra
T_TRADEOFFS
T_TAKE_SIDES
0.68
Seeking Progress
0.70 0.66
0.80 T_PRESS
err-pre
e_PS err-ple
T_PLEASURE 0.67
err-hum
T_HUMOUR
0.56
err-cos
T_COSTS
Process Control
0.88
1.03
Pressing Settlement 0.49
err-con
T_CONTROL
err-des
T_DESTRUCT
e_PC
(a)
(b)
e_I
err-lis
T_LISTEN
0.89 err-nee
T_NEEDS
err-apo
T_APOLOGIZE
0.92 0.68
Encourage Self-improvement
0.67
err-edu
0.68
T_EDUCATE 0.72
err-sav
T_Disputants' Perception
T_SAVEFACE 0.71
0.96 err-exp
T_EXPECT
Reality Test
e_RT
(c) Fig. 5. The EFA models (the initial models) of (a) mediation procedures-related, (b) settlement-related and (c) disputant’s perception-related mediator tactics.
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
2357
e_IC
-0.75 e_A
err-sim err-arg
T_ ARGUE
T_SIMPLE 0.82
1.05 Analyzing
Ice Breaking e_C
0.59 -1.04 err-pri
-1.36 0.31 err-cau
T_PRIORITIES 0.74
0.97 T_CAUCUSES
Caucuses
0.71 e_SP
-1.59
e_TB T_Settlement
0.71 T_Mediation Procedure err-rap
T_ RAPPORT
err-tak
Trust Building err-tra
T_TRADEOFFS
0.61
Seeking Progress
T_TAKE_SIDES 0.63
0.72 -0.55
err-ple
0.60
1.06
1.00 0.77
-5.84
T_SUGGEST
err-sug
0.73
0.59
err-pre
T_PRESS
e_PS
T_PLEASURE 0.64 0.68
err-hum
Process Control
T_HUMOUR
T_COSTS
err-cos
1.09
0.92 Pressing Settlement err-con
0.39
T_CONTROL err-des
T_DESTRUCT
e_I
(a)
e_PC
(b)
err-lis
T_LISTEN
err-nee
T_NEEDS
0.91 0.95
0.87 0.66
Encourage Self-improvement
err-apo
T_APOLOGIZE 0.61
err-edu -6.03
0.70
T_EDUCATE 0.62
0.56
-7.92
T_Disputants' Perception err-sav
T_SAVEFACE 0.71
1.00 err-exp
(c)
T_EXPECT
Reality Test
e_RT
Fig. 6. The refined models of (a) mediation procedures-related, (b) settlement-related and (c) disputant’s perception-related mediator tactics.
ARTICLE IN PRESS 2358
T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359
test. And the settlement-related tactics are grouped by three major tactics: ice breaking, seeking progress and pressing settlement. Acknowledgements The authors would like to express their appreciation to the accredited mediators of the HKIAC for their information and participation in the research. The work described in this study was supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region (Project No: CityU 111905). Appendix A As suggested by Comrey et al. [53], the following variables were discarded:Mediator tactics: Mediation procedure related 1. Reduce the feeling of hostility towards each party. 2. Focus on the impasses issues during caucuses session. 3. Attempt to speak their language. 4. Formulate clear goals before or during the process. 5. Assure each party that the other was being honesty. Settlement related 1. Help the parties to devise a framework for negotiations. 2. Suggest the parties to review their needs. 3. Simplify the agenda by eliminating or combining issues. 4. Discuss other settlements in comparable cases. 5. Keep the negotiations focused on the issues only. Disputants’ perception related 1. Remind the parties that their position was unrealistic. 2. Suggest particular settlement for parties to consider. Mediation outcomes: 1. Overall success. 2. Nothing ambiguously stated. 3. The inter-party relations improved. 4. I acquired a reputation for the effectiveness in setting the dispute. 5. Both parties felt no future problems expected. 6. The needs and goals of mediator satisfied. References [1] Kochan TA. Step-by-step in the Middle East from the perspective of the labor mediation process. In: Rubin JZ, editor. Dynamics of third party intervention. New York: Praeger; 1980. p. 122–35. [2] Peters E. Strategy and tactics in labor negotiations. New London, CT: National Foremen’s Institute; 1955. [3] Kressel K. Labor mediation: an exploratory survey. New York: Association of Labor Mediation Agencies; 1972.
[4] Kressel K, Pruitt DG. Themes in the mediation of social conflict. Journal of Social Issues 1985;41:179–98. [5] Kressel K, Pruitt DG. Mediation research. San Francisco: JosseyBass; 1989. [6] Pruitt DG. Negotiation behavior. New York: Academic Press; 1981. [7] Rubin JZ. Experimental research on third-party intervention in conflict: toward some generalizations. Psychological Bullet 1980;87: 379–91. [8] McLaughlin M, Carnevale P, Lim R. Professional mediators’ judgments of mediation tactics: Multidimensional scaling and cluster analysis. Journal of Applied Psychology 1991;76:465–72. [9] Lim RG. Framing mediator decisions through the development of expectations: a range-frequency explanation. PhD thesis. University of Illinois, Urbana, 1990, 114pp. [10] Wall JA. Mediation: an analysis, review, and proposed research. Journal of Conflict Resolution 1981;25:157–80. [11] Touval S, Zartman IW. International mediation in theory and practices. Boulder, CO: Westview Press; 1985. [12] Bartunek JM, Benton AA, Keys CB. Third party intervention and the bargaining behavior of group representatives. Journal of Conflict Resolution 1975;19:532–57. [13] Kochan TA, Jick T. The public sector mediation process. Journal of Conflict Resolution 1978;22:209–40. [14] Stein J. Structures, strategies, and Tactics of mediation: Kissinger and Carter in the Middle East. Negotiation Journal 1985;1:331–47. [15] Carnevale PJD. Strategic choice in mediation. Negotiation Journal 1986;Jan:41–56. [16] Wall JA, Stark JB, Standifer RL. Mediation—a current review and theory development. Journal of Conflict Resolution 2001;45(3): 370–91. [17] Bercivutch J, Derouen K. Mediation in internationalized ethnic conflicts: assessing the determinants of a successful process. Armed Forces & Society 2004;30(2):147–70. [18] Bercovitch J, Wells R. Evaluating mediation strategies: a theoretical and empirical analysis. Peace and Change 1993;18(1):3–25. [19] Carnevale PJD. Time pressure and strategic choice in mediation. Organizational Behavior and Human Decision Processes 1988;42: 111–33. [20] Folberg J, Taylor A. Mediation—a comprehensive guide to resolving conflicts without litigation. San Francisco, CA: Jossey-Bass Publishers; 1984. [21] Goldberg SB, Green ED, Sander FEA. Mediation. Journal of Dispute Resolutions 1985;22:38–44. [22] Cheung SO. The alternative dispute resolution movement in the construction industry in Hong Kong. The Australasian Dispute Resolution Journal 1999;10(2):98–112. [23] Shapiro D, Driegh R, Brett JM. Mediator behavior and the outcome of mediation. Journal of Social Issues 1985;41:101–14. [24] Hiltrop JM. Mediator behavior and the settlement of collective bargaining disputes in Britain. Journal of Social Issues 1985;41(2):83–99. [25] Hiltrop JM. Factors associated with successful labor mediation. Journal of Society Issues 1989;41:241–62. [26] Donohue WA. Communicative competence in mediators. Journal of Social Issues 1989;41:322–43. [27] Lim RG, Carnevale P. Contingencies in the mediation of disputes. Journal of Personality and Social Psychology 1990;58(2): 259–72. [28] Zubek JM, Pruitt DG, Peirce RS, Iocolano A. Mediator and disputant characteristics and behavior as they affect the outcome of community mediation. Presented at 2nd annual meeting of international association of conflict management, Athens, GA. 1989. [29] Eagly AH, Crowley M. Gender and helping behavior: a meta-analytic review of the social psychological literature. Psychological Bulletin 1986;100:283–308. [30] Carnevale PJD, Conlon DE. Time pressure and strategic choice in mediation. Organizational Behavior and Human Decision Process 1988;42:111–33.
ARTICLE IN PRESS T.W. Yiu et al. / Building and Environment 42 (2007) 2344–2359 [31] Carnevale PJD, Lim RG, Mclaughlin ME. Contingent mediator behavior and its effectiveness. In: Kressel K, Pruitt DG, editors. Mediation research—the process and effectiveness of third-party intervention. San Francisco, CA: Jossey-Bass Publication; 1989. [32] Chaudhry SS, Ross WR. Relevance trees and mediation. Negotiation Journal 1989;5:63–73. [33] Conlon D. Mediator behavior and interest: effects on mediator and disputant perceptions. PhD thesis. University of Illinois, Urbana, 1998, 85p. [34] Harris KL, Carnevale P. Chilling and hastening: the influence of third-party power and interests on negotiation. Organizational behavior and human decision processes 1990;47:138–60. [35] Esser JK, Marriott RG. A comparison of the effectiveness of substantive and contextual mediation tactics. Journal of Applied Social Psychology 1995;25(15):1340–59. [36] Kerr C. Industrial conflict and its mediation. American Journal of Sociology 1954;60:230–45. [37] Douglas A. Industrial peacemaking. New York: Columbia University Press; 1962. [38] Perez FA. Evaluation of mediation techniques. Labor Law Journal 1959;10:716–20. [39] Steven CM. Strategy and collective bargaining negotiation. New York: MaGraw Hill; 1963. [40] Karim A, Pegnetter R. Mediator strategies and qualities and mediation effectiveness. Industrial Relations 1983;22(1):105–14. [41] Eiseman JH. A third party consultation model for resolving recurring conflicts collaboratively. Journal of Applied Behavioral Science 1977;13:303–14. [42] Young OR. Intermediaries: additional thoughts on third parties. Journal of Conflict Resolution 1972;16:51–65. [43] Johnson DW, Johnson RT, Dudley B, Ward M, Magnuson D. The impact of peer mediation training on the management of school and home conflicts. American Educational Research Journal 1995;32: 829–44. [44] Bush RAB. What do we need a mediator for?: Mediation’s ‘‘valueadded’’ for negotiators. Ohio State Journal on Dispute Resolution 1996;12:1–36. [45] Depner CE, Cannata K, Ricci I. Mediated agreements on child custody and visitation. Family and Conciliation Courts Review 1995;32:306–25. [46] Touval S. Coercive mediation on the road to Dayton. International Negotiation 1996;1:547–70. [47] Day-Vines NL, Day-Hairston BO, Carruthers WL, Walls JA, Lupton-Smith HA. Conflict resolution: the value of diversity in the recruitment, selection, and training of peer mediators. The School Counselor 1996;43:392–410. [48] Shulman HA. Using developmental principles in violence prevention. Elementary School Guidance and Counseling 1996;30:170–9. [49] Sharma S. Applied Multivariate Technique. New York: Wiley; 1996. [50] Holt GD. Construction research questionnaire and attitude measurement: relative index or mean. Journal of Construction Procurement 1997;3(2):88–94. [51] Cheung SO, Yeung YW. The effectiveness of the dispute resolution advisor system: a critical appraisal. The International Journal of Project Management 1998;16(6):367–74. [52] Cheung SO, Tam CM, Ndekugri I, Harris FC. Factors affecting clients’ project dispute resolution satisfaction in Hong Kong. Construction Management and Economics 2000;18:281–94. [53] Comrey AL, Lee HB. A first course in factor analysis. 2nd ed. Hillsdale, NL: Erlbaum; 1992. [54] Dias VH, Pinto JF. Identification of the most relevant factors that affect and reflect the quality of granules by application of canonical and cluster analysis. Journal of Pharmaceutical Sciences 2001;91(1):273–81. [55] Norra S, Stuben D. Trace element patterns and seasonal variability of dust precipitation in a low polluted city—the example of Karlsruhe/
[56]
[57] [58] [59]
[60] [61]
[62]
[63]
[64]
[65]
[66]
[67] [68]
[69]
[70]
[71]
[72]
[73]
[74] [75]
[76]
[77] [78]
2359
Germany. Environmental Monitoring and Assessment 2003;93:203–28. Grande JA, Borrego J. Application of cluster analysis to the geochemistry zonation of the estuary waters in the Tinto and Odiel rivers (Huelva, Spain). Environmental Geochemistry and Health 2003;25:233–46. Bimler D, Kirkland J. Smoke and mirrors: mapping the dimensions of Cigarette Space. Quality and Quantity 2003;37:377–91. Holt GD. Applying cluster analysis to construction contractor classification. Building and Environment 1996;31(6):557–68. Componation PJ, Byrd J. Utilizing cluster analysis to structure concurrent engineering teams. IEEE Transactions on Engineering Management 2000;47(2):269–80. Kaufman L, Rousseeuw PJ. Finding data in groups: an introduction to cluster analysis. New York: Wiley; 1990. Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate data analysis, 5th ed. Englwood Cliffs, NJ, USA: Prentice-Halll; 1998 (p. 89–140). Johnson B, Stevens JJ. Exploratory and confirmatory factor analysis of the school level environment questionnaire (SLEQ). Learning Environments Research 2001;4:325–44. Newby M, Fisher D. A model of the relationship between university computer laboratory environment and student outcomes. Learning Environments Research 2000;3:51–66. Dorman JP. Cross-national validation of the what is happening in this class? (WIHIC) questionnaire using confirmatory factor analysis. Learning Environments Research 2003;6:231–45. Kelloway EK. Using LISREL for structural equation modeling: a researcher’s guide. USA: SAGE Publications, International Education and Professional Publishers; 1998 (p. 8–30). Newby PK, Muller D, Tucker KL. Associations of empirically derived eating patterns with plasma lipid biomarkers: a comparison of factor and cluster analysis methods. American Journal of Clinical Nutrition 2004;80:759–67. Bailey KD. Typologies and taxonomies: an introduction to classification techniques. London: Sage Publications; 1994. Maurice WML, Robertson DL. Essential Dispute Resolution and Mediation Principles in a Nutshell for Lawyers, The Official Journal of the Law Society of Hong Kong, August. 2000. Carmines E, McIver J. Analyzing models with unobserved variables: analysis of covariance structures. In: Bohrnstedt G, Borgatta E, editors. Socail management: current issues. Beverly Hills, CA: Sage; 1981. Bryne BM. Structural equation modeling with LISREL, PRSLIS, and SIMPLIS: basic concepts. Application, and programming, NJ: Lawrence Erlabum Mathwah; 1998. Wong PSP, Cheung SO. Trust in construction partnering: views from parties of the partnering dance. International Journal of Project Management 2004;22(6):437–46. Jedidi K, Jagpal HS, DeSarbo WS. Finite-mixture structural equation models for response-based segmentation and unobserved heterogeneity. Marketing Science 1997;16(1):39–59. Browne MW, Cudeck R. Alternative ways of assessing model fir. In: Bollen KA, Longs JS, editors. Testing structural equation models. Newbury Park, CA: Sage; 1993. Arbukle JL, Wothke W. Amos 4.0 user’s guide. USA: SmallWaters Corporation; 1995. Bruce J, Joseph JS. Exploratory and confirmatory factor analysis of the school level environments questionnaire (SLEQ). Learning Environments Research 2001;4:325–44. Meuleners LB, Lee AH, Binns CW, Lower A. Quality of life for adolescents: assessing measurement properties using structural equation modeling. Quality of Life Research 2003;12:283–90. Schumacker RE, Lomax RG. A beginner’s guide to structural equation modeling. Mahwah, NJ: Lawrence Erlbaum; 1996. Everitt B. Cluster analysis. 2nd ed. Oxford: Heinemann Educational; 1980.