key: cord-0815321-fygexbht
authors: Hussein, Mohamed; Zayed, Tarek
title: Critical Factors for Successful Implementation of Just-in-time Concept in Modular Integrated Construction: A Systematic Review and Meta-analysis
date: 2020-10-19
journal: J Clean Prod
DOI: 10.1016/j.jclepro.2020.124716
sha: a6517d8dac12cc3c98033b0defb0fefdae4ee3d6
doc_id: 815321
cord_uid: fygexbht

Modular integrated construction (MiC) is a revolutionary construction method. However, the logistics management of MiC has always been a major barrier to the wider adoption of MiC. Nonetheless, this challenge can be tackled by the application of lean techniques, namely, just-in-time (JIT). Numerous studies have identified and evaluated the critical factors (CFs) required to implement JIT; however, there is no consensus among the previous studies on these CFs and their level of importance. Therefore, this research, for the first time, provides a systematic review and meta-analysis of these CFs. The systematic review identifies 42 CFs. To further provide a synthesis analysis of previous studies, a meta-analysis approach is used. This analysis is conducted on the identified CFs to evaluate their importance level and hence rank them. The results indicate that all the 42 CFs are important for applying JIT, of which seven are highly significant for successfully implementing JIT in MiC. Although the ranking obtained by meta-analysis is much more reliable than that provided in the individual studies, however, there is still a high heterogeneity in the results, which depicts the uncertain nature of the construction field. Therefore, sub-group analysis is conducted to investigate this heterogeneity and uncover the hidden patterns in the literature. This is achieved by studying the influence of predictive factors (moderators) on the importance level of CFs. This analysis shows that the economy of a country and the type of project executed are influential factors. The results further indicate that developing economies, in contrast to advanced economies, should pay more attention to three CFs. Also, the results show that seven CFs are much more important in MiC projects than the other project types. This research work is highly beneficial for theory development and for practitioners by identification of significant CFs that warrant management dedication to best apply JIT. Researchers, in particular, can consider the recommendations given here for implementing future meta-analysis studies.

Modular integrated construction (MiC) brings radical changes to the traditional cast-in-situ 38 construction method (Wuni and Shen, 2020a) . In MiC, the whole building is divided into a 39 number of full-volumetric modules produced off-site in a factory environment. Next, these 40 modules are transported to the construction site for direct installation with minimal 41 construction activities. MiC has many advantages over the traditional method, including 42 reduction in construction duration and wastes (Wang et al., 2015) , and 43 improvement in quality and safety (Xu et Table 1. This table provides an example of three CFs where some studies consider  90 them as significant factors to the successful implementation of JIT, while other studies find 91 them as non-significant. Here, a factor is deemed to be significant if its mean value is higher 92 than "4" on the five-point Likert scale (Lam et al., 2007; Chan et al., 2004) . Therefore, there 93 is little consensus among the previous studies on the ranking and evaluation of CFs based 94 on their importance level. Although the lack of consensus among researchers may be attributed to the uncertain nature 106 of the construction management research field, an international review of the CFs required 107 for implementing JIT and a generalized understanding of the importance of these CFs are 108 both warranted and imperative. Therefore, this study aims to achieve these two objectives. 109

Firstly, the development of a checklist of CFs of JIT from an international perspective will 110 help researchers in conducting further empirical studies. Secondly, a generalized 111 that may affect the importance level of these CFs? If yes, what are these factors? And how 117 can they affect the evaluation of CFs? Thus, this study aims to address these questions, 118

which are yet to be answered. Answering these questions will help MiC stakeholders to 119 optimize their strategic resource allocation by investing in the significant CFs required for 120 implementing JIT successfully. 121

Therefore, the novelty of this study can be summarized in two points. Firstly, it provides a 122 systematic review of the CFs required for the successful implementation of JIT in MiC. 123 Secondly, this study aims to meta-analytically synthesize the previous studies that have 124 quantitatively evaluated the CFs of JIT. The meta-analysis approach adopted in this study 125 provides a more reliable evaluation of the CFs than previously reported. Moreover, this study 126 conducts a sub-group analysis to understand how different factors affect the evaluation of 127 the identified CFs. Although few studies have used meta-analysis in the construction 128 management research field (Alruqi and Hallowell, 2019) , the approach is well-established in 129 many research fields and considered as a key component for theory building (Hunter and 130 Schmidt, 2004) . 131

The rest of this paper is organized as follows; section 2 illustrates the importance of JIT in 132

MiC and summarizes the main findings of the previous studies. Then, the methodology 133 adopted to conduct the systematic review and meta-analysis is illustrated in section 3. The JIT is a lean construction approach, which aims to deliver required materials on-time, at the 141 right location, and with the required quality and quantity (Ballard and Howell, 2010) . JIT is 142 built on six fundamental principles. They are pull system, waste elimination, smooth workflow, 143 J o u r n a l P r e -p r o o f total quality management, supplier relations, and top management commitment (Pheng and 144 Chuan, 2001). The pull system requires that materials are sent only after receiving 145 authorized orders from the demand side, which eliminates the need for high inventory levels. 146

The waste elimination principle identifies and eliminates waste activities that do not add 147 value to the final product. Hence the waste elimination principle emphasizes the pull system 148 principle by considering the inventory as a waste activity. Since JIT targets zero inventory, 149 the principle of smooth workflow tries to maintain an uninterrupted workflow through the 150 different activities along the supply chain. The principle of smooth workflow cannot be 151 implemented without ensuring good quality of the received materials through the principle of 152 total quality management. To receive the required materials on-time, with the required 153 quality and quantity, mutually beneficial and long-term customer-supplier relationships are of 154 utmost importance. Finally, the benefits of JIT cannot be achieved without the commitment 155 of top management and involvement of employees in the process. These benefits are 156 inventory reduction, reduction of procurement costs, improvement of market competitiveness, 157 building long-term supplier relationships, and enhancement of forecasting (Akintoye, 1995) . Overall, the above-mentioned studies show that academic attention towards studying CFs of 189 JIT is increasing. Also, as MiC continues to draw more attention, many review studies have 190 been published, as mentioned in section 1. However, a systematic review of CFs required 191 for the successful implementation of JIT in MiC is missing. Moreover, the previous studies on 192

CFs for implementing JIT have quantified the importance of each CF based on the opinions 193 of its respondents. Therefore, a synthesis of the results presented in these studies is 194 required to provide a more reliable evaluation of CFs. Consequently, meta-analysis is 195 selected to achieve this objective in a transparent and replicable manner. Meta-analysis provides a statistical synthesis of results collected from different studies to 198

give a more reliable indication of the population mean than that provided by individual 199 studies (Horman and Kenley, 2005) . Meta-analysis is usually conducted with a systematic 200 review to combine consistent results (Borenstein et al., 2009) . The systematic review is 201 required to find relevant studies. In meta-analysis, each individual study is assigned a weight 202 based on the number of respondents (sample size) in it by using a standard statistical 203 approach (Borenstein et al., 2009) . In a narrative review, the researcher gives credence to 204 each included study, and hence his/her conclusions are subjective. Accordingly, the 205 obtained conclusions cannot be used without caution. Therefore, meta-analysis is deemed to 206 be more reliable and transparent than a narrative review. However, the adoption of meta-207 analysis is not free of some challenges (Glass et al., 1981) . Inconsistent studies that have 208 used different measuring techniques or addressed different topics may be erroneously 209 combined with each other. The risk of publication bias is another issue, where peer-review 210 journals tend to publish only significant results. However, these issues can be tackled by a 211

proper design of the systematic review and meta-analysis. meta-analysis has been used frequently in the medicine, social, and business research fields 228 (Hunter and Schmidt, 2004) , its adoption in the construction management research field is 229

To the best of the authors' knowledge, this study is one of the first that conducts a 231 systematic review of CFs for JIT implementation. Also, it performs a meta-analysis of results 232 from multiple quantitative studies. Hence, it provides a more reliable assessment of the 233 importance of the CFs required for implementing JIT in MiC. 

Identification of a clear and feasible research question is the first step in any meta-analysis 245 study (Tawfik et al., 2019) 

Conducting preliminary research is paramount to 1) ensure the novelty of the research idea; 254 2) assure that there are enough studies to perform systematic review and meta-analysis; 3) 255 gain a deep understanding of the field of study; 4) collect a wide variety of keywords related 256 to the study topic to help in retrieving relevant studies; and 5) help in defining the inclusion indicates that the statistical meta-analysis approach has not been used before to prioritize 267

CFs to apply JIT in either MiC or the traditional construction. 268

Specification of inclusion and exclusion criteria is a critical step in the meta-analysis study. strategy is a manual search method called snowballing search strategy (Wohlin, 2014) . The 296 first step of this strategy is to identify a start set of papers. These papers are identified after 297 filtering the studies obtained from the keywords-based search method. Then, each study in 298 the start set will be used to conduct backward and forward snowballing. In the backward 299 snowballing, the researcher searches for relevant studies in the reference list of each study 300 in the start set. In the forward snowballing, the researcher uses the list of papers that cited Screening the retrieved studies undergo three consecutive processes (Moher et al., 2009) . 310

Firstly, the title and abstract of each obtained study are evaluated according to the 311 predefined eligibility criteria. The eligibility criteria are inclusive rather than exclusive 312 because sometimes the title and abstract provide insufficient information. At this stage, the 313 authors have included any study which addresses logistics issues in the construction field. 314

Secondly, duplications are removed manually. In this step, duplications emerge only from 315 using multiple databases. Then, eligible articles are downloaded for full-text evaluation. 316

Thirdly, the evaluation of the full text is implemented according to the inclusion and exclusion 317 criteria discussed in section 3.3. Then, the included studies are classified into qualitative and 318 quantitative studies. The difference between them is that the former only discuss CFs to 319 apply JIT without providing any statistical data to evaluate their importance. Forty-six out of 320 150 studies report statistical data to assess the CFs, as shown in Fig. 2 . Only quantitative 321 studies are applicable to meta-analysis. 322

In this stage, the authors have collected data from the full-text articles in a structured Excel 324

sheet. In this study, the extracted information is as follows: authors name, publication type 325 (journal, conference, report, thesis, etc.), publication year, country of the conducted case 326 study, type of project under study (buildings, infrastructure or MiC), CFs to apply JIT, the 327 number of respondents (sample size), research method and the reported statistical data 328 (mean and standard deviation (SD)). Only these data will be used to conduct meta-analysis. 329

Before the commencement of meta-analysis procedures, an examination of the extracted 330 statistical data is essential to determine the dominant measuring technique among the 331 retrieved studies. Determination of this dominant measuring technique allows us to include 332 only studies that deploy this measuring technique. This step is necessary to prevent one of 333 the limitations of meta-analysis, where a researcher mistakenly combines studies that use 334 different measuring techniques or address different topics. In the meta-analysis jargon, this 335 limitation is called mixing "apples and oranges" (Borenstein et al., 2009) . 336

It is found that all of the 46 studies have used the Likert scale to evaluate the CFs except 337 three studies. Hence, these three studies have been excluded. Then, the authors find that 338 the remaining 43 studies have used different Likert scales. However, the five-point Likert 339 scale has been adopted by 30 out of the 43 studies. In the five-point Likert scale, the scale is 340 ranged from 1 to 5 where "1" indicates the least important and "5" represents the most 341 important. It is worth mentioning that it is possible to convert the results obtained by other 342

Likert scales into the five-point Likert scale (IBM Support, 2018). This conversion is possible 343 only in the case that raw data are available and not only the statistical results such as the 344 mean and SD. This case is applicable only for one study, which is the study by Kawish 345 20 studies have been identified to undergo meta-analysis procedures. However, after 352 rechecking the statistical results of the 20 studies, it is found that two of them have precisely 353 the same results, which is not possible. After close examination, it is found that one of them 354 is a journal paper, and the other is a conference paper, but both of them belong to the same 355 researchers and the same case study. Therefore, the authors decide to exclude the 356 conference paper. In summary, we have 19 studies applicable to perform meta-analysis. 357

These studies will be mentioned in section 4.1. 358

Before starting meta-analysis, cleaning of the data extracted from the 19 studies is vital to 359 enter these data to the meta-analysis software effortlessly (Tawfik et al., 2019) . The data 360 cleaning can be done by organizing an excel sheet for each factor containing the information 361 mentioned above in a structured way. Again, the objective of doing meta-analysis in this study is the prioritization of CFs to apply 367 JIT in MiC. This objective is accomplished by the calculation of an aggregated mean and SD 368 for each factor. Consequently, the meta-analysis is applied to each of the identified CFs. 369

Each factor has to be reported in at least two studies to perform meta-analysis. The following 370 sections explain how the aggregated mean and SD are calculated. 371 3.7.1 Effect size, summary effect, and the adopted statistical model 372

The first step in performing meta-analysis is to decide on the effect size based on the 373 available type of data and the purpose of the analysis (Higgins and Green, 2011) . In this 374 study, the effect size represents the mean of the respondents' opinions reported for each 375 factor in the included studies. In these studies, the respondents have evaluated the CFs on 376 the five-point Likert scale. To make it clearer, for each factor, we have a number of studies 377 that have evaluated this factor. Each study reports the mean, SD, and the number of 378 respondents (sample size). The main idea of meta-analysis is to give each study a weight 379 based on the precision of this study, as shown in Eqs. (1) and (4) (Borenstein et al., 2009) . 380

The precision is a function of the sample size. For simplicity, we can assume that the larger 381 the sample size of a study, the higher the weight given to this study. In the meta-analysis, 382 the summary effect is the weighted mean of the individual effects of each study, as shown in 383

Eq. (5). The same idea can be used to calculate the weighted variance, as shown in Eq. (6). 384

The weighted mean (summary effect) and the weighted variance of each factor can be used 385 to rank the CFs. Hence, these statistics will answer the research questions mentioned in 386 section 1. Specifically, research questions related to the significance level and prioritization 387 of CFs. Again, these calculations are repeated for each CF. The mechanism used to 388 distribute the weights among the studies depends on whether the fixed-effects model or the 389 random-effects model is chosen to conduct the analysis. In this study, the random-effects 

Heterogeneity refers to variability in the effects of the different studies combined to perform 404 the meta-analysis (Borenstein et al., 2009 ). This variability is due to true variation among 405 studies rather than the random error (chance). Hence, it is crucial to check the presence of 406 heterogeneity by using statistical tests such as chi-squared ( 2 or ℎ ) test and "-value, 407 especially when confidence intervals of the individual studies do not overlap. A high "-value 408 or a low chi-squared provides evidence of homogeneity or lack of heterogeneity. However, 409

quantifying the heterogeneity is more important. The heterogeneity can be quantified by 410 calculating the ratio of true heterogeneity to total observed variance (# ), as shown in Eq. (7). 411

As a rule of thumb, # < 25 % implies low heterogeneity; # between 25 % and 412 75 % indicates moderate heterogeneity; and # > 75 % implies high heterogeneity (Higgins 413 and Green, 2011). In case of finding high heterogeneity, investigation to explore the reasons 414 behind such heterogeneity needs to be conducted. In this study, sub-group analysis and 415 sensitivity analysis are deployed for investigating heterogeneity. 416

Where # = the percentage of variation between studies that is due to true heterogeneity 417 rather than chance. 418

The sub-group analysis aims at dividing respective studies of a CF into sub-groups based on 420 a specific criterion. This criterion can be any factor that the researcher thinks that it might 421 have an impact on the results (i.e. importance level of CFs in this study). For example, these 422 criteria might be the project's type (e.g. MiC and traditional projects), project's budget (e.g. 423 low, medium and high budget), project's country, etc. Hence, sub-group analysis can 424 address our research questions related to the sensitivity of these CFs to specific factors. The 425

sub-group analysis can be used to investigate the heterogeneity and to make a comparison 426 between the sub-groups to answer specific questions (e.g. secondary research questions in 427 section 3.1). If the sub-group analysis results indicate that there is a significant difference 428 between the results of multiple sub-groups, it can be concluded that the dividing criterion 429 might be the reason behind the heterogeneity. Hence, this dividing criterion (e.g. project's 430 type) has an impact on the results, and consequently, insightful conclusions can be drawn 431 for each sub-groub (e.g. MiC and traditional projects). However, sub-group analysis is 432 restricted by the data available in the studies under investigation (Higgins and Green, 2011) . The purpose of sensitivity analysis in meta-analysis is to study the effect of different sources 447 of uncertainty in the participant studies to conclude robust findings (Higgins and Green, 448 2011). The sensitivity analysis is conducted by identifying studies whose effect sizes vary 449 substantially from the other studies. Hopefully, by removing such studies, the overall 450 heterogeneity would be reduced. Hence the findings become more reliable (tighter 451 confidence interval). However, removing a study should be backed by a justification. This 452 justification demonstrates why this study might be different from the others; otherwise, it 453 should be included. The difference between studies is represented by the four criteria 454 illustrated in section 3.7.3. 455

This section provides a numerical example on the statistical meta-analysis procedures. The 457 results provided here are related to CF32 "Contract and incentives". Firstly, Table 2 shows 458 statistical data (i.e. sample size, mean, SD and standard error (SE)) of CF32 obtained from 459 studies which have addressed this CF. These data are identified to the meta-analysis 460 software (RevMan 5.3). Table 3 Excluding this study reduces the heterogeneity from I² = 97% to I² = 79%. Table 4  468 summarizes the results of sub-group analysis based on the four criteria mentioned in section 469 3.7.3. The results indicate that the project type is the only influential criterion because there 470 is a significant difference (P-value < 0.05) in the importance level of CF32 between general 471 building projects and MiC. These procedures have been replicated for the remaining of 42 472

CFs. 473 Table 5 shows the 19 studies which meet the inclusion criteria to conduct the meta-analysis. 479

Also, Table 5 lists the information required to implement sub-group analysis such as 480 publication type, publication year, country, and project type. Moreover, the table shows that 481 the included studies have evaluated the importance of CFs by using questionnaires, 482 interviews and case studies. As shown in Table 5 , the publication year of the included 483 studies ranges between 2009 and 2019. Therefore, the authors divided the included studies 484 equally into two groups to conduct the sub-group analysis based on the publication year. 485

The first group contains studies published from 2009 to 2015, while the second group 486 includes studies published from 2016 to 2019. 487 Table 5 488 Table 6 Table 7 and Fig. 3 summarize the results of the meta-analysis for each CF. Meta-analysis 501 has been performed by using Review Manager (RevMan) 5.3, which is a standard meta-502 analysis software (RevMan Web, 2020). The second column of Table 7 shows the results of 503 heterogeneity tests before and after conducting the sensitivity analysis. The third column 504

shows whether there is a statistically significant difference between sub-groups in each of 505 the four criteria mentioned in section 3.7.3. Fig. 3 shows the weighted mean (summary effect) 506 and the weighted SD for each CF in an error bar. 507 Table 7 508 is the oldest study among the studies reported CF2, while the second one is unique by its 520 project type which is MiC. The potential reasons behind these high levels of heterogeneity 521 and how such heterogeneity has less impact on the findings of this study are discussed in 522

The results of the sub-group analysis for each CF are summarized in the third column of 524 Table 7 . Again, the sub-group analysis is applicable only if each group has at least two 525 studies. For the publication type criterion, the results indicate that there is no statistically 526 significant difference between the results published in journal papers and the results 527 sub-group analysis based on the project type is applicable only for 21 out of the 42 CFs. This 535 means that the majority of studies focused mainly on building projects and ignored other 536 project types. Interestingly, the sub-group analysis reveals that the importance levels of eight 537

CFs vary significantly depending on the project type. These eight CFs are CF1 "Top 538 management commitment", CF3 "Extensive planning and scheduling", CF15 "QA/QC", CF16 539 "Agreed implementation methodology to implement JIT", CF25 "Employee morale and 540 motivation", CF27 "Knowledge and awareness of JIT", CF32 "Contract and incentives" and 541 CF39 "Supportive governmental regulations". For the country's economy criterion, the sub-542 group analysis indicates that whether having the country an advanced or stable economy 543 (within the G20) does not affect significantly the importance of the majority of CFs except 544 three CFs. These CFs are CF6 "Decentralization of decision making", CF29 "Education and 545

research" and CF39 "Supportive governmental regulations". 546

Finally, Fig. 3 shows the weighted mean and the weighted SD for each CF after conducting 547 the sensitivity analysis. Apparently, all CFs have mean values higher than "3" on the five-548 point Likert scale. As a result, all of the 42 CFs deserve to be considered for successful JIT 549 implementation. However, only seven CFs have weighted means higher than "4" which 550

indicates that these factors have significant importance. This threshold has been adopted in 551 previous studies by Lam et al. (2007) and Chan et al. (2004) . These seven CFs are CF21 552 "Production planning", CF42 "Route planning", CF14 "Willingness to invest in JIT practices", 553

CF16 "Agreed implementation methodology to implement JIT", CF40 "Political and economic 554 stability", CF7 "Market strategy to adopt JIT" and CF27 "Knowledge and awareness of JIT". 555

This section summarizes the CFs and their statistical results reported in studies on MiC. 557

Only three out of the 19 included studies focused on MiC projects. Table 8 summarizes the 558

CFs mentioned in these studies and the weighted mean and SD of each CF. As shown in 559 Table 8 , 30 out of the 42 CFs are mentioned in the three studies. As for the CFs mentioned 560 in only one study, we have just put the mean and SD reported in the study. On the contrary, 561 the weighted mean and weighted SD are calculated for CFs mentioned in multiple studies. 562

All the 30 CFs have mean values higher than "4", as shown in Table 8 . Hence these CFs are 563 significant for JIT implementation in MiC, except six CFs. These CFs are CF5 "Integrating 564 risk management and supply chain management", CF18 "Site layout planning", CF20 565 "Standardization of activities", CF34 "Collaboration among supply chain members", CF35 566 "Communication and information sharing" and CF41 "Inventory management". 567

The high heterogeneity is observed despite the small number of studies and the fact that all 569 the three studies were conducted in Malaysia. Interestingly, it is observed that the study by In the previous section, the meta-analysis results indicate that seven CFs have mean values 578 higher than "4" on the Likert scale and the other CFs have mean values between "3" and "4". 579 Where ./ = the signal to noise ratio; 0̅ and 2 = the sample mean and sample standard 588 deviation, respectively. 589 Table 9 shows the ranking of CFs with considering all project types (global rank) and MiC subtracted by four, which is the adopted threshold, to judge the significance of CFs. 606 Table 9 607

Despite that Table 9 provides a prioritization of the 42 CFs, it does not show whether there is 609 a significant difference between the importance of these CFs. Also, if there are some CFs 610 that are more important than the others significantly, it would be valuable to identify these 611 factors. Therefore, a one-way analysis of variance (ANOVA)-F test has been performed to 612 test if there is a significant difference between the importance of the 42 CFs. If there is a 613 significant difference, then Tukey's method is used to identify which CFs are more important 614 than the others. Tukey's method is a multiple-comparison method in ANOVA. 615

The first step to implement ANOVA is to collect the mean values of each CF from the studies 616 which addressed this factor. Then, these data have been identified to Minitab ® 18 statistical 617 software to conduct the one-way ANOVA and the Tukey's method. Fig. 5 shows the results 618 of the ANOVA-F test and Tukey's method. Before the discussion of the ANOVA results, it is 619 essential to check the conditions required for a valid ANOVA F-test. Fig. 5(a) shows that the 620 However, considering only the mean values and ignoring the variability associated with 643 these values may lead to flawed prioritization. Therefore, the SNR is used to tackle this issue. 644

After that, a one-way ANOVA F-test is conducted to test if there is a statistically significant 645 difference between the importance of the 42 CFs. With a P-value < 0.05, it is found that 646 there is a significant difference between their levels of importance. However, Tukey's 647 The results summarized in Table 10 show that ranking CFs based on their frequency of 652 occurrence is significantly different from the ranking obtained from weighted means and 653 SNR values. This finding numerically demonstrates that the frequency of occurrence of any 654 factor in the literature does not necessarily reflect its importance level. Consequently, the 655 ranking of factors based solely on their frequency of occurrence may be misleading. The 656 different ranking of CFs based on weighted means and SNR values may be attributable to 657 the high heterogeneity associated with the majority of the 42 CFs. 658

The high heterogeneity may be ascribed to multiple reasons. Firstly, the adoption of the 660 random-effects model and the few numbers of studies in the case of some CFs increase the 661 heterogeneity. Secondly, the sub-group analysis fails to investigate the heterogeneity of 662 some CFs. This might be due to that studies which are combined in a sub-group based on a 663 specific criterion may require further division based on another criterion to be more 664 homogeneous. For instance, dividing the 15 studies, which evaluate CF22 based on a 665 country's economy (seven studies related to G20 countries and eight studies from countries 666 out of the G20), reduces the heterogeneity from # = 98% to # = 95.6% among studies in the 667 G20 group. Dividing these seven studies in the G20 group based on the project type and 668 country (additional criteria), further reduces the heterogeneity between studies, as shown in 669 Fig. 6 . This figure shows that the heterogeneity between studies on infrastructure projects in 670 the USA, studies on general building projects in the UK, and studies on general building 671 projects in other G20 countries have been reduced to 69%, 0%, and 0%, respectively. 672 The sub-group analysis, however, can uncover similar behaviours and validate such 691 arguments by considering multiple sources of information. In short, the high heterogeneity of 692 the data extracted from the included studies complicates the ranking of the 42 CFs. 693 However, the forest plots of the 42 CFs indicate that the high heterogeneity has less impact 694 on the findings of this study. In the majority of these forest plots, there is a relative 695 consistency among the studies of each CF. This consistency is related to the significance of 696 a CF. For example, Fig. 7 shows the forest plots of some CFs whose respective studies 697 agree that the importance of these factors is not significant. This consistency is despite the 698 high heterogeneity reported in each CF. 699

The sub-group analysis is conducted to achieve the secondary objectives of this study. The show that the importance of eight CFs differs significantly between project types. Table 11  720 shows these eight CFs accompanied by their weighted mean values for each project type. 721

Interestingly, all these CFs, except CF39 "supportive governmental regulations", are more 722 important in MiC projects than in the other projects. As for MiC projects, only 12 CFs are 723 evaluated by multiple studies, as discussed in section 4.3. These 12 CFs are CF1, CF3, CF5, 724 CF15, CF16, CF18, CF21, CF25, CF27, CF34, CF35 and CF41. The meta-analysis results 725

indicate that seven of them are significant for implementing JIT. These seven CFs are CF1 726 "top management commitment", CF3 "extensive planning and scheduling", CF15 "QA/QC", 727 CF16 "agreed implementation methodology to implement JIT", CF21 "production planning", 728 CF25 "employee morale and motivation" and CF27 "knowledge and awareness of JIT". 729

The fourth secondary objective can be achieved by identifying if the importance of CFs 731 differs significantly based on the country's economy. The sub-group analysis indicates that 732 the importance of the majority of the 42 CFs does not differ significantly between countries 733 with and without advanced economy except three CFs. These CFs are CF6 "decentralization 734 of decision making", CF29 "education and research" and CF39 "supportive governmental 735 regulations". The results of this sub-group analysis show that these CFs are more important 736 to developing economies than advanced economies. Hence, developing economies may 737 concentrate effort on these CFs to successfully implement JIT. Thus, the sub-group analysis 738 can uncover and emphasize such pattern more reliably and transparently. 739

This study provides a generalized assessment of the importance level of CFs required for 

Given the challenges of MiC logistics and the potential of JIT to address them, the 778 managerial implications of this study are invaluable. The lack of manuals to guide 779 practitioners to implement lean techniques, including JIT, is one of the main barriers to adopt 780 lean techniques in the construction industry (Demirkesen and Bayhan, 2019). Therefore, this 781 study provides two important deliverables to convert JIT from mere theory to practice. Firstly, 782 Table 6 provides checklists of CFs that when thoroughly considered, will aid in the 783 successful application of JIT. Secondly, Table 9 provides a meta-analytical prioritization of 784 these CFs. This ranking helps MiC stakeholders to optimize resource allocation strategies 785 for a better implementation of JIT. Through a sub-group analysis, this study shows that the 786 importance of these CFs is changing with different countries and project types. Moreover, 787 the high heterogeneity between the included studies highlights that other factors might have 788 an impact on the assessment of CFs. Therefore, stakeholders implementing JIT should note 789 that ignoring the project characteristics (e.g. type, duration, budget, etc.) and the 790 environment in which JIT is implemented may not end with obtaining the desired outcomes. 791

Despite that this study examines CFs required for implementing JIT, many of these CFs are 793 also associated with lean construction. JIT is a fundamental part of lean, and the concepts 794 and techniques developed under lean are similar to that of JIT (Schonberger, 2007) . 795 Therefore, this study provides valuable theoretical and managerial implications for lean 796 construction. In particular, this study shows that the importance of some CFs is sensitive to 797 the type of project and the territory it is implemented. There is also the possibility that these 798

CFs are also sensitive to additional criteria or moderators. Therefore, it would be valuable 799 that researchers examine the sensitivity of CFs of other lean construction techniques (e.g. 800 last planner system, six sigma, 5S, visual stream mapping, etc.) to potential moderators. 801

Also, the study findings imply that lean construction practitioners should be aware that the 802 importance of CFs to apply lean construction is dynamic with the project's characteristics 803 and environment. This would avoid additional costs associated with repairing of defective modules on-site or 823 returning them to the factory (Enshassi et al., 2019) . Other factors such as CF28 "training", 824 CF20 "standardization of activities" and CF25 "employee morale and motivation" also 825 contribute to defect and cost reduction (Kumar et al., 2020) . Last but not least, through CF7 826 "market strategy to adopt JIT", successful JIT application could provide the organization with 827 a sustainable competitive advantage and improve its market image which in turn increase its 828 market share (Green et al., 2019) . On the other hand, the successful implementation of JIT 829 requires initial costs to consider factors such as CF28 "training", CF31 "availability of 830 resources", CF33 "incentives and reward systems", CF37 "existence of certified and qualified 831 lean personnel" and CF38 "experienced and skilled workers". 832 JIT implementation contributes significantly to environmental sustainability (Klassen, 2000) . The successful application of lean techniques, including JIT, contributes to the social 852 sustainability of the built environment (Mellado and Lou, 2020). For instance, factors such as 853 CF18 "site layout planning" and CF20 "standardization of activities" help in providing a safe 854 working environment (Sanad et al., 2008) . Also, consideration of CF42 "route planning" 855 would reduce the social impacts (e.g., emissions and traffic accidents) associated with the 856 transportation of such bulky materials in urban areas (Kong et al., 2018) . Besides, reduction 857 of inventory levels (through CF41 "inventory management"), especially for constriction sites 858 in dense urban areas, contributes to the social welfare of their neighbours by reducing social 859 impacts such as the noise and traffic congestion. CF35 "communication and information 860

sharing" would help in avoiding disputes and claims among stakeholders and mitigating 861 corruption risks (Faisal, 2010) . In addition, paying attention to factors such as CF22 "culture 862 and human attitude", CF24 "knowledge sharing", CF25 "employee morale and motivation", 863 CF26 "trust among supply chain members", CF28 "training", CF8 "human resources 864 management", CF6 "decentralization of decision making", CF10 "well-established long-term 865 supplier-client relationships" and CF4 "stakeholder management" would increase the successful application of lean techniques would add more workloads over employees (Hines 870 et al., 2004) . Such social concerns are applicable to factors such as CF3 "extensive planning 871 and scheduling", CF13 "adopt new management concepts", CF20 "standardization of 872 activities" and CF27 "knowledge and awareness of JIT". However, considering factors such 873 as CF1 "top management commitment", CF6 "decentralization of decision making" and CF25 874 "employee morale and motivation" help in reducing the work stress (Conti et al., 2006) . Despite the obvious importance of this study, some limitations do exist. Firstly, combining 884 and assessing CFs from several international studies overlook the sensitivity of these CFs to 885 project characteristics such as the project's type, budget, territory, etc. However, these 886 sensitivities may be overlooked for theoretical development or situations where bespoke 887 research is not feasible or economical, but considering project characteristics is imperative in 

This study contributes to the knowledge domain of application of lean techniques in the 902 construction field, and its novelty is summarized as follows 1) CFs for implementing JIT in 903 Similarly, the results of the sub-group analysis illustrate that the country's economy has not 928 influenced significantly the importance level of the majority of CFs except three CFs. These 929

CFs are CF6 "decentralization of decision making", CF29 "education and research" and 930 CF39 "supportive governmental regulations". The results highlight that these CFs are more 931 important for developing economies than advanced economies. On the contrary, the results 932 of the sub-group analysis based on project type reveal that the importance levels of eight 933

CFs differ significantly between multiple project types. Seven of them are more important in 934

MiC projects. These seven CFs are CF1 "top management commitment", CF3 "extensive 935 planning and scheduling", CF15 "QA/QC", CF16 "agreed implementation methodology to 936 implement JIT", CF25 "employee morale and motivation" and CF27 "knowledge and 937 awareness of JIT" and CF32 "contract and incentives". Hence, the sub-group analysis can 938 be used to uncover such hidden patterns in the literature. Overall, this study contributes to 939 the theory development of JIT in the construction industry by identifying CFs to apply JIT 940 from an international perspective. This would furnish researchers with the basis for further 941 empirical studies. Also, the assessment of these CFs will help practitioners to optimize 4) It is preferable to provide not only the statistics of collected data but also data details. 962

These include data that the researcher thinks that they may affect the obtained 963 results, such as project type, project budget, project duration, characteristics of 964 respondents, country, etc. Providing data details facilitates conducting sub-group 965 analysis and meta-regression, which can uncover hidden patterns and provide 966

This is to acknowledge that the project leading to the publication of this paper is fully funded 969 J o u r n a l P r e -p r o o f J o u r n a l P r e -p r o o f Improper planning of resources leads to wastes and an imbalanced workflow. (Enshassi et al., 2018) , (Yunus et al., 2017) , (Olamilokun, 2015) . 3

Adopt different supplier selection criteria to select partners who are willing to adopt JIT rather than focusing only on the lowest price. Also, the selection of fewer and nearby suppliers support JIT success. (Kanafani, 2015) , (Aigbavboa et al., 2016) , (Olamilokun, 2015) . 3

Adopt new management concepts that focus more on productivity and quality, not only on time and cost. (Sarhan and Fox, 2013) , (Kanafani, 2015) , (Abdullah et al., 2009) . 3

Risk aversion by fear from invest in new strategies such as JIT is a critical barrier to adopt JIT.

(Demirkesen and Bayhan, 2019), (Olamilokun, 2015) . 2

Technical CF15 Quality assurance and quality control (QA/QC) Rejection of defective materials will disrupt the MiC supply chain. Therefore, the QA/QC strategy and robust performance measurement system need to be applied through the whole MiC supply chain. (Olamilokun, 2015) .

CF36 Collaboration between the design and construction stages JIT implementation is hampered by the dichotomy between design and construction stages, which leads to rework in both of them and deterioration of quality and productivity, and hence JIT is missed. 7.24 0.007 86% NA* NA* NA* NA** NA** NA** NA** * = sensitivity analysis is not applicable; ** = sub-group analysis is not applicable due to insufficient number of studies; 2 = chi-squared test statistic; and = percentage of variation. 

Towards producing best 973 practice in the Malaysian construction industry: the barriers in implementing the Lean 974 Construction Approach

A meta-977 analytic study of the impact of Lean Production on business performance

Resampling tests for meta-analysis of 980 ecological data

Drivers and barriers of lean construction practice 983 in South African construction industry

Just-in-Time application and implementation for building material 988 management

Prioritising lean construction barriers in Uganda's construction 991 industry

Accessed 31 th

Critical Success Factors for Construction Safety: Review 996 and Meta-Analysis of Safety Leading Indicators

Barriers to Lean Adoption for Construction Projects

Key factors of successful JIT 1002 integration with IBS -An overview

Success 1005 factors of JIT integration with IBS construction projects-a literature review

Barriers to successful 1010 implementation of lean construction in the Ghanaian building industry

The effect of motivators, supply, and lean 1014 management on sustainable supply chain management practices and performance: 1015 Systematic literature review and modeling

Toward construction

Introduction to Meta-1021 Analysis

Relationships between lean and sustainable construction: 1025 Positive impacts of lean practices over sustainability during construction phase

Exploring Critical Success Factors for Partnering in Construction Projects

The effects of lean production 1031 on worker job stress

A hybrid approach using 1034 entropy and TOPSIS to select key drivers for a successful and sustainable lean 1035 construction implementation

Critical Success Factors of Lean Implementation in the 1037 Construction Industry

Identifying 1040 supply chain vulnerabilities in industrialized construction: an overview

Barriers of lean tools implementation to eliminate 1043 construction waste

Integrated Risk Management 1047 Framework for Tolerance-Based Mitigation Strategy Decision Support in Modular 1048 Construction Projects

Sustainable Data Analytics for Environmental 1051 Performance Monitoring in Dynamic Supply Chain Infrastructure

Sustainable supply chains: A study of interaction among the enablers

Meta-assessment of bias in science

New advancements, 1058 challenges and opportunities of multi-storey modular buildings -A state-of-the-art 1059 review

Exploring the relationship between lean construction and 1061 environmental sustainability: A review of existing literature to decipher broader 1062 dimensions

Confidence intervals for estimating the population signal-1064 to-noise ratio: A simulation study

Meta-analysis in social research

Lean production theory-based simulation of modular construction 1069 processes

Impact of JIT, TQM and green 1071 supply chain practices on environmental sustainability

Building information 1074 modelling-based framework to contrast conventional and modular construction methods 1075 through selected sustainability factors

A novel 1078 mathematical optimisation model for the scheduling of activities in modular construction 1079 factories

Energy consumption and carbon emissions 1082 assessment of integrated production and erection of buildings' pre-fabricated steel 1083 frames using lean techniques

Cochrane handbook for systematic reviews of interventions

Learning to evolve: A review of contemporary lean 1088 thinking

Signal to Noise Ratio -What is the right size

Quantifying Levels of Wasted Time in Construction with 1095

Optimal logistics planning for modular 1098 construction using two-stage stochastic programming

Methods of meta-analysis: Correcting error and bias in 1101 research findings

Transforming different Likert scales to a common scale

Accessed 31 th

Lean Methodologies and Techniques for Modular 1106 Construction: Chronological and Critical Review

JIT Practices from the Perspective of Malaysian IBS 1109

Quantifying the waste reduction potential of using 1111 prefabrication in building construction in Hong Kong

A holistic review of off-site 1114 construction literature published between

Life cycle performance of modular buildings: A critical review

Barriers to the Implementation of Lean Thinking in the Construction 1120 Industry -The Case of UAE

Identifying and Prioritizing Barriers and Overcoming Strategies in 1125

Implementing Lean Construction Principles and Methods within Transportation Projects. 1126 Master of Science, Construction Management

Dynamic production scheduling model under due date 1130 uncertainty in precast concrete construction

Just-in-Time Manufacturing and Pollution Prevention Generate Mutual 1133 Benefits in the Furniture Industry. Interfaces (Providence)

An integrated framework for reducing precast fabrication inventory

Sustainable performance of just-in-time 1138 (JIT) management in time-dependent batch delivery scheduling of precast construction

Optimal single-machine 1141 batch scheduling for the manufacture, transportation and JIT assembly of precast 1142 construction with changeover costs within due dates

Modelling the interrelationship 1145 between factors for adoption of sustainable lean manufacturing : a business case from 1146 the Indian automobile industry 7038

Benchmarking the performance of design-1148 build projects: Development of project success index

What is the proper way to apply the multiple comparison test?

Meta-heuristic algorithm for solving vehicle routing problems with time 1154 windows and synchronized visit constraints in prefabricated systems

The impact of lean practices on performance: 1157 based on meta-analysis and Bayesian network

Cross-functional executive involvement and worker 1160 involvement in lean manufacturing and sustainability alignment

Investigating waste reduction potential in the upstream processes of 1163 offshore prefabrication construction

Ultra-rapid delivery of specialty field 1166 hospitals to combat COVID-19: Lessons learned from the Leishenshan Hospital project 1167 in Wuhan

Relationship between just-in-time manufacturing 1169 practices and performance: A meta-analytic investigation

4D BIM integrated construction 1172 supply chain logistics to optimise on-site production

Lean management, supply chain 1175 management and sustainability: A literature review

Statistics, Twelfith e

Modeling just-in-time purchasing in the ready mixed concrete 1182 industry

Preferred Reporting Items 1184 for Systematic Reviews and Meta-Analyses: The PRISMA Statement (Reprinted from 1185 Annals of Internal Medicine)

Design and Analysis of Experiments, Eighth

Application of discrete-event simulation to investigate 1190 effects of client order behaviour on off-site manufacturing performance in Australia

Critical Success Factor for 1194 Industrialized Building System Process Protocol Map by Lean Construction 1195 management

Investigating Facilitators and Barriers for Adopting Lean Construction 1197 Principles in the Nigerian Building Consulting Firms

Barriers to prioritizing lean construction in the Libyan 1200 construction industry

Just-in-time management in precast concrete construction: 1204 A survey of the readiness of main contractors in Singapore

Just-in-Time Management of Precast Concrete

Insights into The Adoption of Lean Management in Industrialised 1211 Building System (IBS) Implementation: The Drivers and Challenges

Lean, green, and the quest for superior 1214 environmental performance

Optimal Construction Site Layout 1217

Considering Safety and Environmental Aspects

Meta-analysis of studies examining long-term construction injury rates

Chic. Ill)

A Strategy for Overcoming Barriers to the Successful Implementation of 1223

Lean Construction in the UK. Doctoral dissertation

Barriers to Implementing Lean Construction in the UK 1226 Construction Industry. Built Hum

Japanese production management: An evolution-With mixed 1230 success

Barriers to lean implementation in the construction industry 1232 in China

AMSTAR 2: A critical appraisal tool for 1236 systematic reviews that include randomised or non-randomised studies of healthcare 1237 interventions, or both

Toward a holistic view on lean sustainable construction : A 1239 literature review

The role of modularity in 1242 sustainable design: A systematic review

Effective material logistics in urban construction sites: A 1245 structural equation model

Evaluation of the efficiency and sustainability of timber-1248 based construction

Analysis of selected building 1250 constructions used in industrial construction in terms of sustainability benefits

Best practice of 1253 prefabrication implementation in the Hong Kong public and private sectors

A step by step guide for conducting a systematic review and meta-analysis 1257 with simulation data

Lean thinking in the highways construction sector: 1260 motivation, implementation and barriers

The prevalence of 1263 musculoskeletal symptoms in the construction industry: a systematic review and meta-1264 analysis

Identifying best design strategies for construction waste 1267 minimization

Precast supply chain management in 1269 off-site construction: A critical literature review

Guidelines for snowballing in systematic literature studies and a replication 1272 in software engineering

Barriers to the adoption of modular integrated construction: 1275 Systematic review and meta-analysis

Fuzzy modelling of the critical failure factors for modular 1278 integrated construction projects

Critical success factors for modular integrated construction 1281 projects: a review

Critical risk factors in the application of 1284 modular integrated construction: a systematic review

Comparative analysis of modular construction practices in 1287 mainland China, Hong Kong and Singapore

Effect of stakeholder collaborative 1290 management on off-site construction cost performance

Critical Success Factors for Lean Thinking in the Application of 1294 Industrialised Building System (IBS)

A stochastic optimization approach of flow 1297 shop sequencing problem for on-time delivery of precast components

Proc. 8th Int. Conf

A simulation based optimization study for 1301 optimum sequencing of precast components considering supply chain risks

The hindrance to using prefabrication

Hong Kong's building industry

Questionnaire & Case study * Study ID in meta-analysis is written as follows: last name of the first author + year + optional key information (publication type) (country) (project type). ** Sample size refers to the number of respondents. *** This study discussed three case studies; two of them were in Ireland, and the third was in the USA. J o u r n a l P r e -p r o o f

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:J o u r n a l P r e -p r o o f