مغالطه مقیاس گذاری تیم:تخمین غیرواقعی بهره وری رو به کاهش تیم های بزرگتر
|کد مقاله||سال انتشار||تعداد صفحات مقاله انگلیسی||ترجمه فارسی|
|4578||2012||11 صفحه PDF||سفارش دهید|
نسخه انگلیسی مقاله همین الان قابل دانلود است.
هزینه ترجمه مقاله بر اساس تعداد کلمات مقاله انگلیسی محاسبه می شود.
این مقاله تقریباً شامل 10934 کلمه می باشد.
هزینه ترجمه مقاله توسط مترجمان با تجربه، طبق جدول زیر محاسبه می شود:
- تولید محتوا با مقالات ISI برای سایت یا وبلاگ شما
- تولید محتوا با مقالات ISI برای کتاب شما
- تولید محتوا با مقالات ISI برای نشریه یا رسانه شما
پیشنهاد می کنیم کیفیت محتوای سایت خود را با استفاده از منابع علمی، افزایش دهید.
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Organizational Behavior and Human Decision Processes, Volume 118, Issue 2, July 2012, Pages 132–142
The competitive survival of many organizations depends on delivering projects on time and on budget. These firms face decisions concerning how to scale the size of work teams. Larger teams can usually complete tasks more quickly, but the advantages associated with adding workers are often accompanied by various disadvantages (such as the increased burden of coordinating efforts). We note several reasons why managers may focus on process gains when they envision the consequences of making a team larger, and why they may underestimate or underweight process losses. We document a phenomenon that we term the team scaling fallacy—as team size increases, people increasingly underestimate the number of labor hours required to complete projects. Using data from two laboratory experiments, and archival data from projects executed at a software company, we find persistent evidence of the team scaling fallacy and explore a reason for its occurrence.
Across a wide range of industries and functions, from construction to consulting and from healthcare to new product development, work is delivered to customers in the form of projects completed by teams (Edmondson and Nembhard, 2009 and Ilgen et al., 2005). Organizations turn to teams for many reasons, one of which is the increased speed with which projects can be completed when work is divided among many people. Organizations also rely increasingly on teams because knowledge is evolving so rapidly that in many settings, no single person has the depth of knowledge required to adequately serve customer needs. Teams also allow for specialization of member roles through the division of labor and can increase the knowledge resources available both within a team and through members’ external connections (Haleblian and Finkelstein, 1993, Moreland et al., 1996 and Reagans and Zuckerman, 2001). In many project-based organizations that rely on teams, an important key to competitive success is accurately estimating and adhering to project budgets and deadlines. For a business that delivers projects to customers, missing promised budget and deadline estimates can tarnish a previously good reputation with patrons, resulting in lost business. Such errors in forecasting may also turn projects that should have generated profits into money-losing ventures (Heskett et al., 1997 and Wheelwright and Clark, 1992). Despite the importance of meeting deadlines and correctly estimating costs, industry statistics suggest that many project-based organizations struggle with these activities. For example, studies in the construction, healthcare, aerospace, and information technology industries have found that anywhere from 33% to 88% of projects are delivered late and over budget (Knight, 2011, Standish, 2009 and Watson, 2008). One possible explanation for these budget and deadline overruns is that process challenges arise when people work together, yet estimators do not properly account for them. Research on teams has shown that although increasing a team’s size provides the potential for many benefits (e.g., through increased specialization and expanded knowledge networks), the team’s actual productivity may suffer due to process losses ( Levine and Moreland, 1998 and Steiner, 1972). Increasing a team’s size can hamper its coordination, diminish its members’ motivation, and increase conflict among team members ( Hare, 1952, Ingham et al., 1974 and Moreland et al., 1996). An interesting question is whether estimators are sufficiently sensitive to these problems. In this paper, we investigate whether estimators exhibit a bias that we term the team scaling fallacy—a tendency to increasingly underestimate task completion time as team size grows. We confirm the hypothesis that the team scaling fallacy plagues estimators in both the laboratory and the field. We also identify and test an important driver of this phenomenon: the tendency to focus too much on the process gains associated with increasing team size, relative to the process losses.
نتیجه گیری انگلیسی
Our findings suggest that estimators are not sufficiently sensitive to the effect of team size on the hours of labor required to complete a software development project. These results are consistent with the findings from Experiments 1 and 2 and provide additional evidence of the team scaling fallacy. Specifically, we found that % effort optimism correlated positively with team size, even after controlling for estimated person–hours to complete a project. The R-squared values in our regression models were small. But as a SoftCo manager told us, “Even a couple percentage points is very important. That money goes straight to the bottom line and can be the difference in a profitable or an unprofitable project. Also, if you get the estimate right, then you don’t have to stress the team out in hitting the schedule.” Our field data illustrate the importance of the team scaling fallacy in a natural context where opportunities abound for learning and feedback and the economic consequences for inaccurate forecasts are substantial. However, we were not able to observe planned team size—only actual team size. As a result, we cannot conclusively eliminate the possibility that some projects were behind schedule and added team members as a result — thus creating an operational problem that contributed to poor performance. Although our robustness checks examining alternative measures of team size helped to address this concern, and also supported our main findings, we acknowledge that there could be some underlying, unobserved heterogeneity in teams driving our results. However, such an explanation cannot account for the converging results from our two laboratory experiments, in which we were able to exogenously assign project-team size while holding all other characteristics of a group task constant. Future work might examine the team scaling fallacy in a field experiment where team size can be manipulated. General discussion and conclusion In three studies, we found evidence that the team scaling fallacy is a real and persistent bias. Optimistic errors in forecasting the total labor required to complete team projects increase as the size of the project team increases. This bias is exhibited by both outsiders estimating the effort required to complete a project and by insiders who will be completing the project themselves. The bias is evident whether estimates about teams of varying size are made between- or within-subjects. We also found that the bias was more pronounced when estimators placed more weight on process gains and less weight on process losses in teamwork. Our research thus provides the first direct empirical support for a specific implication of the coordination neglect hypothesis proposed by Heath and Staudenmayer (2000). The studies in this paper have complementary strengths and weaknesses. Our experiments, carried out in a laboratory setting, allowed us to examine the causal impact of exogenously varied team size on estimation error, while holding each team’s task constant. The experiments also allowed us to explicitly measure estimators’ intuitions concerning process gains and losses, which allowed us to document a possible mechanism for the team scaling fallacy. However, the experiments did not allow us to establish the external validity of our findings. Meanwhile, our field study established the external validity of our findings, but did not allow us to conclusively establish a causal relationship between team size and optimistic estimation error, or to examine the mechanism responsible for our findings. Together, the three studies provided compelling evidence that the team scaling fallacy is a meaningful phenomenon associated with a tendency to underestimate process losses and/or overestimate process gains. Questions for future research Our work suggests several questions for future research. First, it remains to be seen how team member interdependence affects the team scaling fallacy. If a team project requires little coordination, then coordination neglect is unlikely to bias estimators. Coordination costs vary across team projects, depending on task and team member interdependence. In this paper, we examined estimates of the total hours of labor required to complete projects that required significant coordination among team members. Future research could examine the sensitivity of the team scaling fallacy to interdependence of partitioned task components. For instance, one intriguing avenue of inquiry would be to examine the impact of increasing project modularity on the team scaling fallacy. By dividing work into modules that interconnect only through well-defined interfaces, modularity decreases the global consequences of local changes and thus limits both the number of interconnections and the need for them among team members (Baldwin & Clark, 2000). Thus, modularity may reduce potential process losses sufficiently to attenuate or eliminate the team scaling fallacy. In any case it would be interesting to examine the extent to which estimators are sensitive to task modularity when making their forecasts. A second question raised by the present work is how the team scaling fallacy might be affected by one’s familiarity with the team completing a project. In our research, estimators were generally unfamiliar with the specific individuals who would be working on team projects. It would be interesting to understand how estimates change if an estimator knows, for example, whether the people on a team have worked together previously, or the demographic diversity of team members. Such factors have been shown in previous studies to affect team performance (Espinosa et al., 2007, Harrison and Klein, 2007 and Huckman et al., 2009), but it remains to be seen whether estimators are aware of these effects, and to what extent estimators incorporate them into effort forecasts. Although we focused broadly in this paper on the impact of attending to process gains rather than losses associated with increasing team size, it would also be interesting to examine estimator sensitivity to specific kinds of process gains and process losses in group work (for example, motivation problems and conflicts among members). Future work could explore which specific factors contribute to the team scaling fallacy, and compare the relative contributions of each factor. Future research might also examine whether characteristics of the estimator or the estimator’s setting moderate the team scaling fallacy. For example, although we provided an accuracy bonus to all of the estimators in Experiment 1, future research could examine whether stronger incentives increase accuracy. In field settings, authority or status sometimes depends on accurate estimation, so it could be interesting to explore the impact of nonpecuniary incentives, such as social accountability, on the team scaling fallacy. Further, we examined the team scaling fallacy in the context of individuals’ estimates of group performance. But groups of estimators also could be studied. Would groups exhibit the team scaling fallacy as well (cf. Buehler et al., 2005 and Sanna et al., 2005)? Finally, future research could consider the team scaling fallacy in the context of not only effort estimation, but also schedule estimation (i.e., missed deadlines). Future research might also investigate the team scaling fallacy in other field settings. Although the archival dataset we examined was both large and detailed, it came from a single company in a single industry. Prior studies of small groups have shown that shifts in team size have implications for team performance across a wide range of settings (Hackman and Katz, 2010 and Moreland et al., 1996). Thus, we would expect the team scaling fallacy to arise across a wide range of settings. We found that the team scaling fallacy is associated with greater attention to process gains relative to process losses. One additional avenue for future research is to explore the extent to which this observation is driven by overweighting of process gains relative to process losses, or overestimation of process gains relative to process losses. That is, to what extent do people incorrectly estimate the magnitude of process gains and losses associated with increasing team size, and to what extent do they underweight the impact of these factors? Our data did not permit us to disentangle these different possibilities; future research could follow up on this question. Implications for project management A natural question about the team scaling fallacy is how can it be remedied. It might be helpful to measure and model the relationship between increasing team size and the number of hours of labor required to complete team projects in particular settings. Such information would allow estimates to be scaled by an appropriate, pre-determined factor, based on the expected team size, providing a cognitive repair for the team scaling fallacy (Heath, Larrick, & Klayman, 1998). Indeed, forecasters might benefit by estimating completion times using base rates that incorporate the true effect of team size on task-completion times, based on similar past projects (Kahneman & Lovallo, 1993). Our findings have a number of direct implications for the field of project management. Although estimators and project managers are aware of the need for work integration within teams, our research shows the time required for such integration may be increasingly underestimated as a team grows. This problem may be compounded by common tools used in estimation, such as the work breakdown structure (WBS), which helps estimators to unpack the elements of projects that must be completed, so they can more accurately estimate total time required to complete the project. Benefits of a WBS focus around decomposition because “when the work is decomposed to greater levels of detail, the ability to plan, manage, and control the work is enhanced (PMI, 2008, p. 120).” Indeed, research on the planning fallacy has found that unpacking complex individual tasks into subtasks tends to increase the accuracy of predictions (Kruger & Evans, 2004). However, the WBS focuses solely on subdividing work, so there is risk that an estimator using the WBS may be particularly likely to neglect the costs of integration and team interaction. This could exacerbate the team scaling fallacy. Both anecdotal and survey data suggest that organizations often struggle to accurately estimate the effort that will be required to complete the projects that they undertake. More work than ever is now being delivered to customers in the form of projects completed by fluid teams. Accurately estimating the costs of executing this type of work is therefore of growing importance (Edmondson and Nembhard, 2009 and Huckman and Staats, 2011). Inaccuracy can have significant external consequences because missing promised budget and deadline estimates weaken potentially profitable projects and may result in lost repeat business (Heskett et al., 1997 and Wheelwright and Clark, 1992). Additionally, although project leaders may attempt to meet deadlines by adding team members late in a project, or by asking existing team members to work longer hours, both solutions can have substantial negative consequences. Increased staffing may inflate labor costs and can still result in missed deadlines due to the time required for integrating new team members (Brooks, 1975 and Chen, 2005). Likewise, overtime work can be an expensive solution due to both the higher rate of overtime pay and the risk that when team members work longer hours, stress levels can increase, resulting in worse performance and an increased threat of voluntary departure (DeMarco, 2002 and Humphrey et al., 2007). We hope that our research on the team scaling fallacy can lead to new procedures and tools that improve the accuracy of estimates of group effort and minimize negative organizational consequences of errors in these forecasts.