Journal of Operations Management 24 (2006) 779790 www.elsevier.

com/locate/jom

Six Sigma: The role of goals in improvement teams

Kevin Linderman a,*, Roger G. Schroeder a,1, Adrian S. Choo b,2
Operations and Management Science Department, Curtis L. Carlson School of Management, University of Minnesota, 3-150 CarlSMgmt Building, 321-19th Avenue South, Minneapolis, MN 55455, USA b Rensselaer Polytechnic Institute, Troy, NY, USA Received 1 May 2005; received in revised form 1 August 2005; accepted 3 August 2005 Available online 27 December 2005
a

Abstract The tenets of goal theory have been well established as a motivation mechanism in the management literature. However, some quality-management advocates, such as W. Edwards Deming, often criticize the use of goals. This research investigates the tension between goals and quality management in the Six Sigma context. We nd empirical support that goals can be effective in Six Sigma improvement teams when teams adhere to the Six Sigma tools and method. However, challenging goals are counterproductive when Six Sigma teams do not use the tools and methods rigorously. This research reconciles the differences between quality management and goal theory by showing that the Six Sigma tools and method interact with goals. # 2005 Elsevier B.V. All rights reserved.
Keywords: Quality management; Six Sigma; Goal theory; Teams; Process improvement

1. Introduction Much has been written about quality management over the last few decades (Ahire et al., 1995; Sousa and Voss, 2002). Most of the research focuses on studying quality-management practices and associated success factors (Kaynak, 2003; Sousa and Voss, 2002). However, research on how motivational factors inuence qualitymanagement outcomes is scant. Motivation is the process that accounts for an individuals intensity, direction, and persistence toward a goal (Robbins, 2003, p. 155). While goal theory research suggests that specic, challenging goals lead to higher performance (Locke and Latham, 1990), at least one quality-management authority has expressed a conicting view. For example, Deming
* Corresponding author. Tel.: +1 612 626 8632. E-mail addresses: klinderman@csom.umn.edu (K. Linderman), rschroeder@csom.umn.edu (R.G. Schroeder), achoo@csom.umn.edu (A.S. Choo). 1 Tel.: +1 612 624 9544. 2 Tel.: +1 612 626 9723. 0272-6963/$ see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jom.2005.08.005

(1986) viewed arbitrary numerical goals as counterproductive. Research in quality management from a goaltheoretic perspective can help resolve these conicting viewpoints. Reviews of the goal-setting literature have consistently demonstrated the effectiveness of individual goal setting on performance (Locke and Latham, 1990; Locke et al., 1981). In fact, Miner (1980) rated goal theory high in both criterion validity and usefulness in application. (Criterion validation indicates how well scores on a measure correlate with the criterion on interest (Singleton and Straits, 1993, p. 122). This gives an indication of practicality or usefulness of a measure.) Also Pinder (1984) said, goal theory has demonstrated more scientic validity to date than any other approach on motivation . . .. Moreover, the evidence indicates that it probably holds more promise as a motivational tool for managers than any other approach. Research on the effects of goals on group performance, however, is still emerging. OLeary-Kelly et al. (1994) performed a quantitative meta-analysis of

780

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790

goals in the group setting and found a signicant relationship between group goals and group performance. None of the studies in their meta-analysis considered the effect of goals on quality-improvement teams. Quality-improvement teams are central to quality improvement (Scholtes et al., 1996) because effective teams can perform signicantly better than a collection of individuals (Mohrman et al., 1995; Katzenbach and Smith, 1993). Setting challenging group goals can promote team effectiveness (Locke and Latham, 1990) because goals tell teams what needs to be done and how much effort to expend. Little research in quality management uses goal theory. Some quality-management movements, such as the Zero Defects, have argued for setting very high goals. However, some quality thinkers have been critical of the goal-theoretic perspective. For example, Deming (1994, p. 41) said . . . a goal that lies beyond the means of its accomplishment will lead to discouragement, frustration, and demoralization . . .. A numerical goal accomplishes nothing. Deming systematically rejected the use of goals as a source of motivation (Carson and Carson, 1993). Demings negative views on quantitative goal setting are at odds with both the historical management prescriptions and contemporary research on goal setting and motivation (Duncan and Van Matre, 1990, p. 5). Carson and Carson (1993) compare and contrast Demings views with goal theory and reconcile differences when possible. Hackman and Wageman (1995) considered research issues related to quality management and noted, It is not surprising that there is disagreement among the TQM authorities about goal setting, because the nature of the work done by quality teams raises some complex issues about how goals and objectives are properly framed. The conicting views of goals in quality management (e.g. Deming versus Zero Defects) suggest the need for more scholarly investigation. Empirical testing of goal theory in the quality context can help reduce the confusion between goal theory and quality management. Recently, Linderman et al., (2003) developed a set of propositions about Six Sigma from a goal-theoretic perspective. Schroeder et al., (2003) provide a denition of Six Sigma and compare it with other qualitymanagement approaches. These researchers argue that nothing is radically new in Six Sigma but Six Sigma does place a strong emphasis on challenging specic goals (see also Pande et al., 2000). In fact, the name Six Sigma suggests a goal of 3.4 parts per million defective. The Six Sigma approach to process improvement also employs numerous goals; for example, setting project

improvement goals to increase performance by a factor of 10 (10), hiring Black Belts on the basis of costsaving goals, and selecting improvement projects based on nancial and strategic goals. As a result, Six Sigma provides an ideal context to study the relationship between goal theory and quality management. Our research begins to test the tenets of goal theory in Six Sigma projects. Because Six Sigma is consistent with quality-management principles, these results should also be generalizable to other quality practices. 2. Theoretical development Goal-setting theory suggests that challenging goals lead to enhanced performance because they mobilize effort, direct attention, and encourage persistence and strategy development (Locke and Latham, 1990). According to goal-setting theory, goals are effective because they indicate the level of performance that is acceptable (Locke and Latham, 1990). Specic goals are critical to individuals because they establish a minimum acceptable performance level, but ambiguous goals either do not make clear the appropriate performance level or indicate to individuals that a range of performance levels is acceptable (Locke and Latham, 1990). Difcult goals, if accepted, lead to greater individual effort and persistence (Locke and Latham, 1990; Locke et al., 1981; Tubbs, 1986). However, assuming that the results of goal theory for individuals also apply to groups may be subject to the ecological fallacy (Singleton and Straits, 1993, p. 69). The ecological fallacy can occur when erroneous information from one level of aggregation (e.g. individuals) is used to draw inferences about another level of aggregation (e.g. groups). In this setting, the effect of goals motivating individuals may differ from that of groups. As a result, further research needs to be conducted at the group level in goal theory (Weingart and Weldon, 1991). In response to this phenomenon, scholars have been studying goal theory in group settings (Weingart and Weldon, 1991; Weldon et al., 1991; Weingart, 1992; OLeary-Kelly et al., 1994; Durham et al., 2000; Knight et al., 2001). In general, scholars have found a signicant relationship between group goals and group performance (OLeary-Kelly et al., 1994; Locke et al., 1997). Goal theory also suggests that specic goals result in higher levels of performance than vague nonquantitative goals such as Do best goals (Locke and Latham, 1990). Do best goals are goals that are implied by the task or occur when the subject is told to do the best he or she can. OLeary-Kelly et al. (1994) found that goal specicity also applies to group goals.

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790

781

Six Sigma is known for employing specic challenging process improvement goals (Pande et al., 2000). In fact, Pande et al. (2000, p. 44) indicates, A clear goal is the center piece of Six Sigma. It is extremely challenging, but still believable. Quantitative improvement goals used in Six Sigma included setting target levels for defects per million opportunities (DPMO) and/or Process Sigma. The 10 rule is one common approach used in Six Sigma to establish improvement goals. For example, if the current DPMO is 30,000 for a process, the improvement goal is to reduce the DPMO by a factor of 103000. Fig. 1 illustrates the proposed relationship between Six Sigma and goals under investigation in this research. The gure indicates a direct effect of challenging and explicit goals and a direct effect of the Six Sigma Tools/ Method on team performance along with a moderating effect or interaction between goals and Six Sigma. These proposed effects are explained below. Although explicit challenging goals can lead to higher performance, a drop in performance can occur if goals become too challenging (Erez and Zidon, 1984). If goals are viewed as unattainable, individuals may exert less effort, which would decrease performance (Erez and Zidon, 1984; Locke and Latham, 1984). Erez and Zidon (1984) found a bipolar relationship between goals and performance. When reviewing the literature goal difculty and performance relationship Locke and Latham noted, in all cases goals are linear except when subjects reach the limits of their ability at high goal difculty levels (1990, p. 27). OM scholars have also argued for considering bipolar (or non-linear) relationships when applying behavioral theory to operations management phenomena (Bendoly and Hur, 2005). In Six Sigma, excessively challenging goals could risk being viewed as unattainable and thus exhibit a non-linear relationship between goal difculty and performance. Fig. 2 illustrates that challenging Six Sigma goals lead to higher performance; however,

Fig. 2. Goal level and performance.

when goals become excessive, performance actually declines. Hypothesis 1. The degree of challenging goals used in a Six Sigma project improvement team is concavely related to project performance. Research in goal theory suggests that the ability to achieve goals moderates the relationship between goals and performance (Locke, 1982). Goal theorists have argued that difcult goals on complex tasks may actually prove to be more detrimental to performance than assigning no goals at all because they may create a level of anxiety that interferes with goal accomplishment (Earley et al., 1989). This perspective is consistent with the views of some quality-management authorities. According to Deming, A numerical goal accomplishes nothing. What counts is the methodby what method? (Deming, 1994, p. 41). Quality management has advocated the use of quality tools and methods in problem solving (Breyfogle, 1999; Gitlow et al., 1995; Hoerl, 1998; Ishikawa, 1985; Kume, 1985, 1995; Mizuno, 1988). Using quality improvement tools and methods should reduce the tasks complexity by guiding the search for solutions to complicated problems, which, in turn, facilitates goal achievement. Six Sigma advocates rigorous application of the quality tools in each step of the problem-solving methodology (Schroeder et al., 2003; Sanders and Hild, 2000). In Six Sigma, prescriptive tools are suggested for each step in the improvement methodology (Rath and Strong, 2000). Using Six Sigma tools and method provides a mechanism for improvement teams to achieve their goals, especially for challenging projects. As a result, the degree to which Six Sigma teams adhere to the tools and method should alter their ability to achieve improvement goals, more so when goals are challenging then when they are easy to achieve.

Fig. 1. Six Sigma and goals.

782

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790

Hypothesis 2. The degree that Six Sigma teams adhere to the use of tools and method positively moderates the effect of challenging goals on project performance. The application of quality tools and methods is grounded in rational decision-making (March, 1999; Harrison, 1999). Rational decision-making emphasizes a systematic step-by-step approach to problem solving. Handeld et al. (1999) found that appropriate use of quality tools can lead to improved performance. To the degree that improvement teams follow the Six Sigma tools and method they can make better decisions, which improves project performance. Furthermore, this is consistent with prior research that suggests the degree of implementation of quality practices is positively related to organizational performance (Douglas and Judge, 2001). Hypothesis 3. The degree of adherence to Six Sigma tools and method is positively related to project performance.

project buy-in and ensures the availability of resources to the team. Most projects have a target savings of at least US$ 175,000, but this savings target does not necessarily apply to all projects. Some projects may have zero nancial savings but generate other important strategic benets that are difcult to quantify in nancial terms. Six Sigma project improvement teams at MFG employ a structured improvement method and use numerous quality tools. In MFG, the structured improvement approach is called DMAIC (Dene, Measure, Analyze, Improve, Control). This approach is patterned after PDCA (Plan, Do, Check, Act). In each improvement step, standard quality tools like failure modes effects analysis (FMEA), causeeffect charts, and Statistical Process Control are used. These tools include many of the seven classic tools of quality control and the seven new tools for problem formulation and diagnosis (Gitlow et al., 1995). 3.2. Data collection

3. Methods 3.1. Research environment The primary source of data for this study comes from a high-tech manufacturing rm using Six Sigma that we refer to as MFG. MFG is a Fortune 500 company with more than 60,000 employees worldwide. This leading manufacturer of electronic components has annual revenue of more than US$ 6 billion. It has been using Six Sigma for 3 years and is very advanced in its application. At the time of this research, MFG had about 200 full-time Black Belt specialists and had completed more than 1500 Six Sigma projects, resulting in savings of more than US$ 400 million from its Six Sigma efforts. Black Belts are highly trained full-time specialists in process improvement who receive more than 4 weeks of training and hands-on project improvement experience. MFG is also training most of its employees in Six Sigma basics. These individuals receive 2 weeks of training and are called Green Belts. In MFG, each Six Sigma project is assigned a team of employees who have substantial knowledge of the process or product to be improved. They serve on a parttime basis and often have Green Belt training. A fulltime Black Belt specialist leads each of these teams. The Black Belt usually reports to the team sponsor, called a Champion, who is from senior management and is trained in Six Sigma basics. The Champion provides a holistic view of the organization, which helps establish

A questionnaire was developed to collect data from a sample of MFGs 1500 completed Six Sigma projects. Most recently completed projects by the Black Belts were selected to minimize the measurement error due to recency effect. That is, the sampling time frame considered only projects completed with in the last 6 months to avoid problems associated with memory loss. Because Black Belts came from different business units in MFG, the projects selected were fairly random and representative of the population of all completed projects at MFG at the time of survey. Black Belt and team members from each selected project were invited to participate in this survey. Using multiple informants reduced the common method bias and increased validity compared with a single-informant survey design. The questionnaire was divided into parts: Black Belts answered one part of the survey that contained questions about performance. Team members answered the other part of the survey about goals, tools, and method. To a certain extent, the separation of theoretical constructs into different parts of the survey further reduced the common respondent bias while at the same time reduced the number of questions per survey, which tended to increase the response rate. The survey was web-based: personalized email invitations to participate in the survey were sent out by the Vice President and Executive Director for Six Sigma to 1233 team members and Black Belts from 324 projects. Each personalized email had a customized link

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790

783

that brought the respondent directly to the survey for a specic project. Extra efforts were made to ensure condentiality and avoid the negative effects of Social Desirability Bias (Nunnally and Bernstein, 1994). Responses collected via the Internet were stored in a protected Microsoft Access database. Follow-up reminders and thank-you emails were sent every week for 3 weeks before we concluded the data collection with a satisfactory number of responses. Some 951 persons completed the survey for about a 77% response rate. At the project-level, we collected data for 206 projects, which represents a 63.5% response rate. The data for each project are based on responses from the Black Belt and at least two team members. 3.3. Variables and measures Although we had multiple informants for all projects, not all theoretical constructs had multiple responses. The Performance construct had multiple responses from Black Belts and at least one team member for all projects, while other constructs had multiple responses from only a third of the projects. We assessed the inter rater reliability (IRR) (James et al., 1984) and deleted 18 projects with inconsistent responses (i.e. IRR 0, >1) and low agreement (i.e. IRR < .10). The remaining 188 projects had an average IRR greater than 0.87 for all constructs with multiple responses. We also compared, using a one-way ANOVA, within-project variance and between-project variance for the various constructs. F-statistics indicated signicant differences for between-project variances. Although we did not have multiple responses for all constructs, the results of the two tests showed sufcient evidence that the data were consistent at the project-level and that it was appropriate to aggregate the item scores for theoretical constructs with multiple responses. We aggregated multiple responses by taking averages for the respective item scores in each project to arrive at a sample of 188 projects. Then we removed all projects designated as a Design For Six Sigma (DFSS) because these projects require a methodology different from the DMAIC that traditional Six Sigma projects use. Finally, we removed all projects that involved soft dollar-savings and considered only hard dollar-saving projects. Soft dollar-savings projects are often tied to the rms long-term strategy, which often makes assessing performance difcult. This process resulted in a total of 128 projects. While quality-management measurement scales existed in prior research (e.g. Saraph et al., 1989), none was useful for this research, which required scales

for project improvement goals, project performance, and adherence to Six Sigma tools and method. The researchers developed the scales based on a literature review and revised the scales through a pilot study at MFG. All multi-item measures in this study use 7-point Likert scales that ranges from 1-strongly disagree to 7strongly agree. The scales were pre-tested through a pilot study at MFG by six Black Belts and eight team members. Five university researchers also knowledgeable in Six Sigma pre-tested the instrument. The pilot study assessed three main characteristics of the surveyclarity (i.e. is the question clear and easy to answer?), content (i.e. does the question make sense and is it appropriate?), and the average time taken to answer the questionnaire. Appendix A provides a summary of the scales. 3.3.1. Team project goals Team project goals consisted of three items. This scale had an alpha of 0.59 (Cronbach, 1951). New scales often result in lower alphas making this scale minimally acceptable for further use (Carmines and Zeller, 1978). Using principle component analysis (PCA), the resultant eigenvalue associated with this scale is 1.6, above the minimum acceptable level of 1.0, and explains 54.9% of the variation. All items loaded into a single factor and the factor loadings for each item ranged from 0.633 to 0.791, all above the acceptable lower bound of 0.4 (Carmines and Zeller, 1978). 3.3.2. Six Sigma Tools/Method Six Sigma Tools/Method consisted of four items. This scale had an alpha of 0.68 (Cronbach, 1951), which is considered acceptable. Using principle component analysis, the resultant eigenvalue associated with this scale is 2.056 and explains 51.4% of the variation. All items loaded into a single factor and the factor loadings for each item ranged from 0.631 to 0.814. 3.3.3. Team performance Team performance consisted of ve items. This scale had an alpha of 0.90 (Cronbach, 1951). Using principle component analysis, the resultant eigenvalue associated with this scale was 3.659 and explained 73.2% of the variation. All items loaded into a single factor and the factor loadings for the items ranged from 0.784 to 0.919. All of the scales were evaluated using principle components analysis. For each scale factor scores were created using the regression method. Conceptually the factor score represents the degree to which each individual scores high on the group of items that have

784

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790 Table 1 Correlations Variable 1 2 3 4
*

high loadings on a factor. Thus, higher values on the variables with high loadings on a factor will result in a higher factor score (Hair et al., 1998, p. 119). The factor scores provide the basis for subsequent analysis. 3.3.4. Control variable Discussion with MFG executives indicated that there might be a difference in performance of Six Sigma projects between domestic and international sites. Domestic sites were more involved in R&D efforts and production was often initiated domestically, then transferred to international sites. MFG also observed a lot of enthusiasm for Six Sigma from the international sites; people at the domestic sites tended to be more skeptical. Scholars have noted the importance of considering the international context when studying operations management (Prasad and Babber, 2000), which suggests the need to control for domestic versus international location. 4. Results Mean, standard deviation, and correlations of the variables under consideration are displayed in Table 1. The collinearity diagnostics (including Variation Ination Factors) indicate that multicollinearity is not a problem. A signicant correlation exists between Team Performance and Six Sigma and between Team Performance and Domestic Projects. Table 2 provides the results of the regression analysis. Model 1 gives the regression of the control variable on the Project Performance variable, which is signicant (F = 15.541, P < 0.001). Model 2 adds in the main effects of Six Sigma Tools/Method and Goals,

Mean 4.09 0.16 2.01 3.07

S.D. 0.56 0.37 0.41 0.41

1 1.00 0.33* 0.11 0.38*

Project Performance Domestic Goals Six Sigma

1.00 0.08 0.13

1.00 0.13

1.00

Correlations signicant at level P < 0.01.

this also gives a signicant model (F = 10.298, P < 0.001) and suggests the main effects inuence Project Performance. Also, the R2 improves by 0.11 in Model 2, but the Goals2 term is not signicant. Model 3 adds in the interaction terms between Goals and Six Sigma Tools/Method. The model is signicant (F = 7.798, P < 0.001) and the R2 improves by 0.028. Model 4 in Table 2 drops the non-signicant predictor variables (Goals2, Six Sigma Tools/Method Goals2) from the regression equation. This model is also signicant (F = 11.669, P < 0.001) along with all the predictor variables. We removed the nonsignicant terms because they can alter the interpretation of the model (Neter et al., 1996, p. 308). Regression diagnostics indicate no problems with outliers, normality, or homoscedasticity. We nd that the R2 of 0.275 is acceptable given the limited number of variables used to predict performance. Subsequent analyses are conducted on Model 4. Model 4 suggest that performance of domestic projects is signicant and results in lower performance than international projects (t = 3.736, P < 0.001), All other models (Models 13) give similar conclusions. These results are consistent with our expectations that

Table 2 Regression analysis for Six Sigma goals and performance Variable Model 1 b Domestic Goals Goals 2 Six Sigma Tools/Method Six Sigma Tools/Method Goals Six Sigma Tools/Method Goals 2 R2 F DR2
* ** ***

Model 2 t 3.942
***

Model 3 t 3.672 2.278** 0.149 4.470***

***

Model 4 t 3.711 2.007** 0.602 3.683*** 2.166** 0.081

***

b 0.796 0.180 0.007 0.355

b 0.797 0.164 0.034 0.347 0.155 0.004 0.279 7.798*** 0.028*

b 0.830 0.160 0.355 0.138 0.275 11.669***

t 3.736*** 2.042** 4.546*** 2.032**

0.891

0.11 15.54***

0.251 10.298*** 0.141***

P < 0.1. P < 0.05. P < 0.001.

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790

785

international Six Sigma projects had a higher level of performance than domestic projects at MFG. Interpreting the other coefcients of the regression model is more difcult. Without interactions, the coefcients of the regression equation indicate the change in mean response with a unit increase in the predictor variable. However, the meaning of the coefcients is not the same in the presence of interaction terms (Neter et al., 1996, p. 308). Model 4 indicates a signicant interaction between Goals and Six Sigma Tools/Method (t = 2.032, P < 0.001). In this setting, the mean response, E [Project Performance], with a unit increase in Goals when Six Sigma Tools/ Method is held constant and given by the following: E Pro ject Per formance 0:16 0:138 Six Sigma Tools=Method Similarly, with a unit increase in Six Sigma Tools/ Method and Goals held constant is given by the following: EPro ject Per formance 0:355 0:138Goals The conditional effects plot (Neter et al., 1996, p. 310) provides a way to interpret the regression equation in the presence of an interaction effect. Fig. 3 gives the conditional effects plot for E [Project Performance] as a function of Goals conditioned on Six Sigma Tools/ Method. The plot considers values between the 1st and 99th percentile ranking for goals, conditioned on the 1st, 50th, and 99th percentile ranking of Six Sigma Tools/Method. In general, the conditional effects plot indicates that when Six Sigma Tools/Method is low, challenging goals result in lower performance. However, when Six Sigma Tools/Method is high, challenging goals result in higher performance. As a result, Six

Sigma Tools/Method has a reinforcement or synergistic (Neter et al., 1996, p. 311) effect on goals. Similarly, Fig. 4 provides the conditional effects plot for E [Project Performance] as a function of Six Sigma Tools/Method conditioned on Goals. The plot considers values between the 1st and 99th percentile ranking for Six Sigma Tools/Method, conditioned on the 1st, 50th, and 99th percentile ranking of goals. An equivalent story emerges where increasing Six Sigma Tools/ Method results in lower performance when goal difculty is low. However, when goal difculty is high, increasing Six Sigma Tools/Method results in higher performance. The conditional effects plots (Figs. 3 and 4) and the statistical signicance of the interaction term in Table 2 support Hypothesis 2. That is, empirical evidence indicates that Six Sigma Tools/Method positively moderates the effect of challenging goals on project performance. We are unable to fully support Hypotheses 1 and 3 as stated, since there is a signicant interaction term in model 4 that makes it difcult to interpret the main effects. This is because the main effects are conditioned by their interaction with another variable. That is, the effect of goals is conditioned on the level of Six Sigma Tools/Method and visa versa. The conditional effects plot in Fig. 3 indicates partial support for Hypothesis 1. Increasing goals linearly improves performance when teams make extensive use of Six Sigma Tools/Method. However, we found no support for the quadratic relationship between goals and performance as suggested in Fig. 2. Possibly all Six Sigma projects sampled in this survey did not have goals that were viewed unrealistic and unattainable, which in turn could have reduced motivation and lowered performance outcomes. The conditional effects plot in Fig. 4 also gives support to Hypothesis 3. The use of Six Sigma Tools/

Fig. 3. Goal difculty conditioned on Six Sigma.

786

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790

Fig. 4. Six Sigma conditioned on goal difculty.

Method results in higher performance except when goals are low. One reviewer noted the anomaly that extensive use of Six Sigma Tools/Method actually resulted in lower performance when goals are low. There could be an explanation for this outcome. In discussions with Six Sigma consultants, they note that sometimes organization inappropriately apply the Six Sigma Tools/Method. As the adage goes if you have a hammer, everything looks like a nail. Teams can waste valuable time and effort rigorously applying Six Sigma Tools/Method when the solution is obvious. In this situation, the excessive use of Six Sigma Tools/Method can actually reduce performance. As suggested by one consultant, the solution method should t the problem. This consultant advocates the use of a Do-It project when the solution is obvious and the focus is on implementation. (A Do-It project means take action immediately, there is no need for a root cause analysis.) This consultant argues that a Six Sigma project should only be used when the solution is not clear (that is, root cause analysis is required). Possibly projects that have low goals should have been Do-It projects rather than Six Sigma projects, which would be consistent with the results in Fig. 4. 5. Discussions and conclusions 5.1. Theoretical implications This study addresses the role of goals in Six Sigma project improvement teams. To the extent that Six Sigma reects quality-management practices in general, this research should be generalizable to other quality-management approaches. We found empirical support that goals do in fact affect project performance. However, analysis shows a positive interaction effect

between Goals and Six Sigma Tools/Method with performance. This supports the perspective advocated by some quality-management authorities. As Deming (1994, p. 41) said, . . . a goal that lies beyond the means of its accomplishment will lead to discouragement, frustration, demoralization. A numerical goal accomplishes nothing. What counts is the methodby what method? The results of this research are consistent with Demings perspective. Goals can have a positive effect on performance when used with Six Sigma Tools/ Method. As Carson and Carson (1993) note, Deming and traditional theorists are often arrayed on opposing sides. The truth, though, is somewhat different. This research provides empirical support that Deming and goal theory can be reconciled. Little empirical research has been done using goal theory in operations management and quality management in particular (Linderman et al., 2003), a surprising fact given that goal theory is a well-established management theory (Locke and Latham, 1990; Pinder, 1984; Miner, 1980). Incorporating theories from organizational behavior can help inform the practical consequences of implementing operations management practices. In this setting, we nd that behavioral theories interact with technical tools and method in interesting ways; that is, the use of technical tools and motivational factors must be managed jointly rather than in isolation. More broadly, operations management should not be understood as a purely technical problem but must be considered simultaneously with behavioral underpinnings. Recently operations management scholars have recognized that incorporating human behavior into OM models will yield more realistic insights (Boudreau et al., 2003). Some scholars have begun to examine the behavioral implications of technical tools in operations management (Doerr et al., 1996;

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790

787

Rungtusanatham, 2001; Schultz et al., 1998). Our research helps advance this effort. This research also contributes to goal theory. Although goal theory is well established for individuals, it is still developing for groups (Locke and Latham, 1990; Weingart and Weldon, 1991; Weldon et al., 1991; Weingart, 1992; OLeary-Kelly et al., 1994). Little goal theory research has considered different types of groups (OLeary-Kelly et al., 1994). Most studies consider work-teams where group membership is stable and employees work together on a daily basis. To our knowledge, this is the rst study to consider problemsolving groups (Six Sigma teams) and goals. In this setting, problem-solving teams are formed to solve a problem and then disbanded upon completion of the project. We also consider the inuence of using problem-solving tools and method on goals, which other goal theory studies have not considered. 5.2. Managerial implications Quality-management practices can often become the latest management fad (Abrahamson, 1996), which can also create distortions between the rhetoric and reality of what is practiced (Zbaracki, 1998). Managers may set high goals when deploying quality but may not follow through on fully implementing the prescribed techniques. Based on this research, setting challenging goals without fully utilizing the tools and method can lead to sub-optimal results. Organizational leaders need to make sure that quality-improvement teams are, in fact, using the tools and method espoused by quality theorists and practitioners. For example, MFG developed a Project Evaluation System that performs a pre- and post-project review of the improvement projects to assess the appropriate use of the tools and method. The Project Evaluation System helps make sure that improvement teams use the tools and methods in the prescribed fashion. Using these deployment techniques not only ensures proper use of tools and method but also facilitates achievement of project improvement team goals. Recently, Steve Kerr, Chief Knowledge Ofcer at General Electric, noted Most organizations do not have a clue about how to manage stretch goals. It is popular today for companies to ask their people to double sales or increase speed to market threefold. But then they do not provide their people with the knowledge, tools, and means to meet such ambitious goals (Sherman, 1995, p. 231). This research supports the belief that challenging goals must be supplemented with tools and method to solve difcult problems. Kerr

further suggested that giving employees challenging goals without a means to achieve them is immoral (Sherman, 1995). In the quality context, leaders need to be cognizant of the importance of training and supporting the use of problem-solving tools and methods. Otherwise, challenging improvement goals could demoralize employees involved in Six Sigma projects. Organizational leaders pursuing quality practices often take a rational approach to improvement of the organizational system (Kast and Rosenzweig, 1972). Improvement of rational systems (Scott, 1987) is governed by both knowledge and motivation. Without knowledge, improvement only occurs by chance events that are rarely understood. In Six Sigma, the creation of knowledge occurs through intentional or explicit learning that employs formal improvement methods. Intentional learning requires regulation of actions taken by organizational members. Goals serve as regulators of human action by motivating project improvement teams. Thus, improvement goals motivate teams to engage in intentional learning activities that create knowledge (Linderman et al., 2003). As a result, organizational leaders that make effective use of goals can regulate how much organizational knowledge is created through Six Sigma. Also, organizations often use heuristics when setting goals for project improvement teams. One common heuristic used in Six Sigma is the 10 rule, which sets an improvement goal of reducing defects by a factor of 10 (Harry and Schroeder, 2000). However, from a goal theory perspective, such heuristics may not be effective. If managers do not ensure teams are trained and use the appropriate problem-solving tools, challenging goals might actually lead to lower performance. 5.3. Limitations and conclusions Despite the interesting results of this study, several limitations need to be emphasized. First, the study considered data from a single rm. Although this does help control for confounding factors like organizational culture, it lacks external validity. Further replications of this study are required to fully test the theory. Second, factors such as group cohesion (Levine and Moreland, 1990) and social loang (Price, 1987) were not considered. These contextual variables could inuence a teams ability to achieve goals; however to date, they have not been tested (OLeary-Kelly et al., 1994). As research on goal theory in a group setting matures, we should better understand contextual variables in teams that inuence goal achievement.

788

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790

Third, other types of improvement projects beyond Six Sigma were not considered. Possibly the type of improvement project employed could effect the relationship between goals and performance. For example, the Do-It project discussed earlier could have a different effect on the relationship between goals and performance. Other improvement approaches such as Lean were also not considered. Possibly using alternative improvement tools/methods has a different effect on relationship between goals and performance. Fourth, the role of national culture is not considered. Possibly, different national cultures affect how goals and Six Sigma Tools/Method affect performance. One reviewer noted that this relationship could be investigated by examining the interaction effects of the Domestic variable with all other variables in the model. However, this created signicant multicollinearity problems that made interpretation of the coefcients impossible. Future research can study the effect of national culture on these relationships in more depth. Finally, project size and duration, prior group project experience, or project group size could also affect performance. Variation of these variables at MFG were somewhat limited since they typically scoped projects to last between 4 months and half a year, and each project had typically four to six team members. In addition, they had a 2-year rotation system for Black Belts, which meant that Black Belts had a 2-year assignment in Six Sigma and then were re-integrated back into the regular organization. This would control somewhat for prior experience with respect to Black Belts. As a result, these variables were somewhat xed at MFG. Future research could investigate how changes in these variables inuence performance. Despite the limitations, this study makes several important contributions. Notably, it helps to clarify the controversial relationship that goal theory has had with quality-management authorities. It shows that goals can be effective when used with quality tools and method. This is particularly important for Six Sigma, which has a strong goal orientation toward improvement (Linderman et al., 2003). In addition, very little research in operations management has considered goal theory (Boudreau et al., 2003). This is surprising given that goal theory is well established in the management literature (Locke and Latham, 1990). This research helps illustrate that operations management is not just a technical problem but also requires behavioral consideration. Future research in operations management should consider theories from organizational behavior. Organizational behavior theories might serve to enhance future research in operations management.

Or, this research has demonstrated the usefulness of organizational behavior theories for future operations management inquiries. Acknowledgement This research was supported in part by a National Science Foundation Grant NSF/SES-0080318 Appendix A. Measurement scales All responses range from 1-strongly disagree to 7strongly agree
Team project goals 1 We found it very difcult to achieve the project goals 2 It was relatively easy to achieve the project goals (reverse scaled) 3 The project goals were challenging to us Six Sigma Tools/Method 1 The project strictly followed the sequence of DMAIC steps 2 Each step in DMAIC was faithfully completed 3 There was an emphasis on applying various analysis tools wherever applicable in this project 4 This team frequently used Six Sigma tools to analyze data and information Team performance 1 We met or exceeded customers expectations in this project 2 This team had superb results on this project 3 This team did not meet the project goals (reverse scaled) 4 The cost savings or strategic impact of the project were signicant 5 The project was effective in improving the process or product

References
Abrahamson, E., 1996. Management fashion. Academy of Management Review 21 (1), 254285. Ahire, S., Landeros, R., Golhar, D., 1995. Total quality management: a review and an agenda for future research. Production and Operations Management 4 (3), 227307. Bendoly, E., Hur, D., 2005. Bipolarity in the constraint-performance relationship: Amalgamating interdisciplinary lessons. Working Paper Emory University. Boudreau, J., Hopp, W., McClain, J.O., Thomas, L.J., 2003. On the interface between operations and human resources management. Manufacturing and Service Operations Management 5 (3), 179 202. Breyfogle III, F.W., 1999. Implementing Six Sigma: Smarter Solutions Using Statistical Methods. John Wiley & Sons, New York, NY. Carmines, E.C., Zeller, R.A., 1978. Reliability and Validity Assessment. Sage, Beverley Hills, CA. Carson, P.P., Carson, K.D., 1993. Deming versus traditional management theorist on goal setting: can both be right? Business Horizons (SeptemberOctober), 7984.

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790 Cronbach, L., 1951. Coefcient alpha and the internal structure of tests. Psychometrica 16, 297334. Deming, W.E., 1986. Out of the Crisis. MIT Press, Cambridge, MA. Deming, W.E., 1994. The New Economics for Industry, Education, Government, second ed. MIT Center for Advanced Engineering Study, Cambridge, MA. Doerr, K.H., Mitchell, T.R., Klastorin, T.D., Brown, K.A., 1996. Impact of material ow policies and goals on job outcomes. Journal of Applied Psychology 81 (2), 142152. Douglas, T.J., Judge, W.Q., 2001. Total quality management implementation and competitive advantage: the role of structural control and exploitation. Academy of Management Journal 44 (1), 158 169. Duncan, W.J., Van Matre, J.G., 1990. The gospel according to Deming: is it really new? Business Horizons (JulyAugust), 39. Durham, C.C., Locke, E.A., Poon, J.M., McLeod, P.L., 2000. Effects of group goals and time pressure on group efcacy, informationseeking strategy and performance. Human Performance 13, 115 138. Earley, P.C., Connelly, T., Ekegren, G., 1989. Goals, strategy development, and task performance: some limits on the efcacy of goal setting. Journal of Applied Psychology 74, 2433. Erez, M., Zidon, I., 1984. Effect of goal acceptance on the relationship of goal difculty to performance. Journal of Applied Psychology 69, 6978. Gitlow, H.S., Oppenheim, A., Oppenheim, R., 1995. Quality Management: Tools and Methods for Improvement, second ed. Irwin, Burr Ridge, IL. Hackman, J.R., Wageman, R., 1995. Total quality management: empirical, conceptual, and practical issues. Administrative Science Quarterly 40 (2), 309342. Handeld, R., Jayaram, J., Ghosh, S., 1999. An empirical examination of quality tool deployment patterns and their impact on performance. International Journal of Production Research 37 (6), 1403 1426. Hair Jr., J.F., Anderson, R.E., Tatham, R.L., Black, W.C., 1998. Multivariate Data Analysis. Prentice Hall, New Jersey. Harrison, E.F., 1999. The Managerial Decision-Making Process, fth ed. Houghhton Mifin, Boston, MA. Harry, M.J., Schroeder, R., 2000. Six Sigma: The Breakthrough Management Strategy Revolutionizing the Worlds Top Corporations. Doubleday, New York. Hoerl, R.W., 1998. Six Sigma and the future of the quality profession. Quality Progress 31 (6), 3542. Ishikawa, K., 1985. In: Lu, D.J. (Ed.), What is Total Quality Control? The Japanese Way. Trans., Prentice-Hall, Englewood Cliffs, NJ. James, L.R., Demaree, R.G., Wolf, G., 1984. Estimating within-group interrater reliability with and without response bias, Journal. Applied Psychology 69 (1), 8598. Kast, F.E., Rosenzweig, J.E., 1972. General systems theory: applications for organization and management. Academy of Management Journal (December), 447465. Katzenbach, J.R., Smith, D.K., 1993. Wisdom of Teams. Harvard University Press, Boston, MA. Kaynak, H., 2003. The relationship between total quality management practices and their effects on rm performance. Journal of Operations Management 21 (4), 405435. Knight, D., Durham, C.C., Locke, E.A., 2001. The relationship of teams goals, incentives, and strategic risk, tactical implementation, and performance. Academy of Management Journal 44 (2), 326338.

789

Kume, H., 1985. Statistical methods for quality improvement. In: Loftus, J.H. (Ed.), The Association for Overseas Technical Scholarship. Trans., AOTS, Tokyo, Japan. Kume, H., 1995. In: Loftus, J.H. (Ed.), Management by Quality. Trans., 3A Corporation, Tokyo, Japan. Levine, J.M., Moreland, R.L., 1990. Progress in small group research. In: Rosenzweig, M.R., Porter, L.W. (Eds.), Annual Review of Psychology, vol. 41. Annual Reviews, Palo Alto, CA, pp. 585634. Linderman, K., Schroeder, R., Zaheer, S., Choo, A., 2003. Six Sigma: a goal theoretic perspective. Journal of Operations Management 21 (2), 193203. Locke, E.A., 1982. Relation of goal level to performance with a short work period and multiple goal levels. Journal of Applied Psychology 67, 512514. Locke, E.A., Latham, G.P., 1984. Goal Setting: A Motivational Technique That Works. Prentice Hall, Englewood Cliffs, NJ. Locke, E.A., Latham, G.P., 1990. A Theory of Goal Setting and Task Performance. Prentice Hall, Englewood Cliffs, NJ. Locke, E.A., Shaw, K.M., Saari, L.M., Latham, G.P., 1981. Goal setting and task performance 19691980. Psychological Bulletin 90, 125152. Locke, E.A., Durham, C.C., Poon, J.M.L., Weldon, E., 1997. Goal setting, planning, and performance on work tasks for individuals and groups. In: Friedman, S.L., Scholnick, E.K. (Eds.), The Developmental Psychology of Planning. Why, How, and When do We Plan. Lawrence Erlbaum Associates Inc., Mahwah, NJ, pp. 239262. March, J.G., 1999. A Premier on Decision Making. Free Press, New York, NY. Miner, J.B., 1980. Theories of Organizational Behavior. Dryden, Hinsdale, IL. Mizuno, S. (Ed.), 1988. Management for Quality Improvement: The 7 New QC Tools. Productivity Press, Cambridge, MA (originally published in Japanese, 1979). Mohrman, S.A., Cohen, S.G., Mohrman, A.M., 1995. Designing Team-Based Organizations. Jossey-Bass, San Francisco, CA. Neter, J., Kutner, M.H., Nachtsheim, C.J., Wasserman, W., 1996. Applied Linear Statistical Models, fourth ed. Irwin, Chicago, IL. Nunnally, J.C., Bernstein, I.H., 1994. Psychometric Theory, third ed. McGraw-Hill, New York, NY. OLeary-Kelly, A.M., Martocchio, W., Frink, D.D., 1994. A review of the inuence of group goals on group performance. Academy of Management Journal 37 (5), 12581301. Pande, P.S., Neuman, R.P., Cavanagh, R.R., 2000. The Six Sigma Way: How GE, Motorola, and Other Top Companies are Honing Their Performance. McGraw-Hill, New York. Pinder, C.C., 1984. Work Motivation. Scott Foresman, Glenview, IL. Prasad, S., Babber, S., 2000. International operations management research. Journal of Operations Management 18 (2), 209247. Price, K.H., 1987. Decision responsibility, task responsibility, identiably, and social loang. Organizational Behavior and Human Decision Processes 40, 330345. Rath, Strong, 2000. Rath and Strongs Six Sigma Pocket Guide. Rath and Strong Management Consultants, Lexington, MA (ISBN 0970507970). Robbins, S.P., 2003. Organizational Behavior, 10th ed. Prentice Hall, Upper Saddle River, NJ. Rungtusanatham, M., 2001. Beyond improved quality: the motivational effects of statistical process control. Journal of Operations Management 19 (6), 653673. Sanders, D., Hild, C.R., 2000. Six Sigma on business processes: common organizational issues. Quality Engineering 12 (4), 603610.

790

K. Linderman et al. / Journal of Operations Management 24 (2006) 779790 Sousa, R., Voss, C.A., 2002. Quality management re-visited: a reective review and agenda for future research. Journal of Operations Management 20 (1), 91109. Tubbs, M.E., 1986. Goal setting: a meta-analytic examination of the empirical evidence. Journal of Applied Psychology 71, 474483. Weingart, L.R., 1992. The impact of group goals, tasks component complexity, effort, and planning on group performance. Journal of Applied Psychology 77, 682693. Weingart, L.R., Weldon, E., 1991. Processes that mediate the relationship between group goal and group member performance. Human Performance 4, 3354. Weldon, E., Jehn, K.A., Pradhan, P., 1991. Processes that mediate the relationship between a goal and improved group performance. Journal of Personality and Social Psychology 61, 555569. Zbaracki, M.J., 1998. The rhetoric and reality of total quality management. Administrative Science Quarterly 43 (3), 602636.

Saraph, J., Benson, G., Schroeder, G., 1989. An instrument for measuring the critical factors in quality management. Decision Sciences 20 (4), 810829. Scholtes, P.R., Joiner, B.L., Streibel, B.J., 1996. The Team Handbook, second ed. Oriel, Madison, WI. Schroeder, R., Linderman, K., Liedtke, C., Choo, A., 2003. Six Sigma: denition and theory. Working Paper. University of Minnesota. Schultz, K., Juran, D., Boudreau, J., McClain, J.O., Thomas, L.J., 1998. Modeling and worker motivation in JIT production systems. Management Science 12 (44), 15951607 (part 1). Scott, R.W., 1987. Organizations: Rational, Natural, and Open Systems. Prentice-Hall, Englewood Cliffs, NJ. Sherman, S., 1995. Stretch goals: the dark side of asking for miracles. Fortune 132 (10), 231232. Singleton, R.A., Straits, B.C., 1993. Approaches to Social Research. Oxford University Press, New York, NY.