W. Edwards Deming united scientific and humanistic approaches to management
When I returned from speaking at the annual conference of the Deming Institute in Los Angeles last month, the education sites were abuzz about a new Time magazine cover trumpeting “Bad Apples”, the latest example of what has become a new national sport–knee-jerk teacher bashing.
It was a sad reminder of how much our quick-fix, here-today-gone-tomorrow society has forgotten about what our leading institutions learned, less than four decades ago, about the best approach to improving quality—whether at companies, schools or other institutions. These were hard fought lessons learned during a period of deep economic malaise—during the late 1970s and early 1980s—from the man who may have been the most important, and most misunderstood, management thinker of the 20th century.
As I pondered the Time magazine cover and the national narrative of education failure, which scapegoats classroom teachers as the principle culprit for all that ails American education, I couldn’t help but think of W. Edwards Deming and how much has been forgotten since he “rescued” American manufacturing 35 years ago. (Deming’s influence on Japan and the U.S. was the subject of my first book The Man Who Discovered Quality.)
Mainstream education reformers want schools and educators to learn from business. The problem is that they are pushing the wrong business lessons. As the leaders of America’s major corporations learned from Deming, meaningful long-term improvement cannot come from top-down punitive solutions.
Deming’s breakthrough was in combining an understanding of how science–in particular statistical theory–can be used to achieve meaningful systems improvement with what had heretofore been seen as touchy-feely, unscientific approaches to empowerment. He used statistical theory to demonstrate how employees, properly trained, can serve as a key resource for identifying organizational problems, solutions and opportunities for improvement. Importantly, he believed that employees can only meet this challenge if they work in an atmosphere that is collaborative and free of fear.
He also believed that because only management can control the larger systemic factors that impact quality—everything from purchasing to hiring to organizational culture—the onus for creating the underlying conditions for quality improvement rests with management. Thus, Deming’s mantra, thundered again and again in his basso profundo: The responsibility for quality rests with senior management.
Deming’s approach to organizational improvement transformed entire industries in post-war Japan and, later, in the U.S. In the years leading up to his death, in 1993, he began turning his interest to education. He believed that the same principles he advocated for companies—systems thinking, collaborative improvement, understanding statistical variation, creating organizational cultures free of fear and conducive to creative problem-solving—could also transform schools.
Simply put, Deming would be appalled by much of what passes for education reform today. In this blog post, I will describe how the lessons Deming taught American industry might apply to education. His ideas are already being applied by a handful of schools and districts that are explicitly adopting his management ideas (more on this in a future post.) Others, see here and here, have used systems thinking to achieve remarkable improvements, and have created conditions where Deming’s ideas would flourish and, likely, produce even greater gains.
In Japanese schools, Deming’s work is associated with the wildly popular “lesson study,” a collaborative, continuous-improvement approach to lesson planning by small groups of teachers, which is just being discovered by American educators.
Deming’s work has important implications for education: First, it is based on management (everyone from principals to education bureaucrats) recognizing its responsibility for creating a climate conducive to meaningful improvement, including building trust and collaboration, and providing the necessary training; this involves hard work, Deming admonished, not quick-fix gimmicks, incentives or threats.
Second, for many teacher advocates, it means dropping the defensive—education-is-good-enough—posture and embracing a mindset of continuous improvement; it also may mean adopting union contracts that mirror the professional practices of many teachers and are based on more flexible work rules. (Though not the unsustainable sweat-shop hours that are common at many charters.)
Third, by ending the finger-pointing and building a more collaborative approach to improvement, schools and districts could create cultures that are far more rewarding and productive for both children and educators.
But first, a quick primer on Deming’s influence on American management. The time was 1979. U.S. industry was being beaten by foreign competition. Chrysler would be the subject of its first (but not last) government bailout; the Ford Motor Co. was about to lose $1 billion for that fiscal year, and at least as much in 1980; and GM’s profits were expected to plunge by a breathtaking $2.5 billion. Meanwhile, Japanese automakers were gaining market share; Toyota would soon surpass GM as the world’s largest car company.
Then, as now, the convenient scapegoat were rank-and-file employees—in Detroit’s case, the hourly workers whose high wages and ostensibly poor work ethic were initially blamed for the automakers’ problems. Only as Japanese wage rates reached parity with the U.S. and Japanese automakers began hiring American workers for their U.S. plants did some Detroit auto executives begin rethinking that narrative of blue collar failure.
Deming was already an octogenarian when he got his first, storied invitation to help revive the sputtering fortunes of the Detroit auto makers. The catalyst was a documentary, If Japan Can, Why Can’t We, which introduced Deming’s methods and his influence on Japan, to an American audience, providing a much needed wake-up call for Detroit. Deming had helped Japanese companies rebuild following World War II, becoming a Japanese icon; the much-heralded Toyota production system grew out of a years-long dialogue between Deming and the automaker.
Responding to an urgent appeal from Donald E. Petersen, then the new president of Ford, Deming flew to Detroit and, during the next few years, proceeded to rip the lid off of the prevailing assumptions about the quality problems of U.S. companies. It is a testament to how desperate the auto executives were that they grudgingly embraced Deming’s message, despite the fact that he lay the lion’s share of the blame for quality problems on senior management instead of labor.
Petersen alone among the big-three CEOs embraced Deming’s message without reservation. At GM, executives in charge of Cadillac and Pontiac, fearful of how senior management would react, brought Deming in via the back door. In the coming years, his ideas about a bottoms-up systems-oriented approach to management helped transform both automakers and became associated with, among others, the revival of Cadillac and the creation of the highly successful Ford Taurus and Sable automobiles.
Yet, Deming was a quintessential outsider—both in his pedigree and his outlook. Raised in Wyoming, Deming was trained as an engineer and physicist, but became a pioneer in statistical sampling methods. At the peak of his popularity, he worked out of the basement of his modest home in Washington, D.C. Driven by a messianic belief in his ideas, Deming never sought to build a business or a fortune. And so he would never be as wholeheartedly embraced as his contemporaries, especially Peter Drucker, who never challenged the fundamental status quo and, thus, became the darling of 20th century CEOs.
At a time when American industry was becoming ever more siloed and finance focused, Deming advocated a collaborative, systems-focused, process-obsessed approach to management. While he was often derided as a mere statistician, Deming made a crucial breakthrough by linking the scientific explanation for how systems work (in particular, how to understand and manage the statistical variation that erodes the quality of all processes) and the humanistic (an intuitive feel for the organization as a social system and a collaborative, democratic vision of management.)
Both strains—the scientific and humanistic—could be traced to a single deceptively simple, and profoundly elegant, understanding of how all processes work. Every process is subject to some level of variation that is likely to diminish quality. Variation is the enemy of quality; yet it is as ubiquitous as gravity.
What makes variation a particular nuisance is that it comes in two distinct guises: “Special causes” of variation, which are the result of special circumstances or a temporary glitch in the system are opportunistic and, by definition, unpredictable; thus they can wreak havoc with a process and give management no basis on which to predict the quality level of a organization’s products or services. They can, however, be identified and eliminated by workers who have been properly trained to analyze the process.
A simple education example: On a recent New York State English test, students were required to answer questions based on a color map. Most schools, however, lacked color copiers and gave their students the test in black and white, which made the distinctions on the map difficult to read. As a result, many students couldn’t answer the map-based questions correctly. Teachers would have been best equipped to both identify and solve the problem—perhaps by finding a nearby Kinkos or arranging a temporary copier rental—assuming they were given the authority to do so.
“Common Causes” of variation, by contrast, are more difficult to isolate because they are inherent in the system. As such, they are, by definition predictable. While common causes can never be fully eliminated, they allow the process to function with a predictable level of variation. Thus an organization that has only common causes to contend with in its process will produce products of a level of quality that is predetermined by the capabilities of the system.
For example, at a school that doesn’t invest in new text books and chooses to recycle outdated ones, the decision is likely to diminish student performance. This decline will be predictable. While teachers might be able to identify the problem, only senior management (the principal) has the power to change the purchasing policy. Because they are systemic, common-cause variation holds the greatest opportunity for long-term improvement.
The key to reducing variation and improving quality, Deming believed, was to train employees who work with the system every day and who know it best to distinguish between special- and common-cause variation and to empower them to develop creative improvements. Ordinary employees—not senior management, or hired experts—are in the best position to see the cause-and-effect relationships in each process. The challenge for management is to tap into that knowledge on a consistent basis and to make that knowledge actionable. To do so, management must also shake up the hierarchy (if not eliminate it entirely), drive fear out of the workplace, and foster the intrinsic motivation of its employees.
Statistical theory also led directly to Deming’s most controversial ideas on pay incentives. Deming built on Abraham Maslow’s ideas about intrinsic motivation and the hierarchy of needs: He believed that if you create the conditions that allow people to do their best, most people will rise to the occasion. (The obverse is also true: A climate of fear and insecurity is the surest way to kill intrinsic motivation.)
Again, Deming invoked the power of statistical theory: If management is doing its job correctly in terms of hiring, developing employees and keeping the system stable, most people will do their best. Of course, there will always be fluctuations—human beings, after all, aren’t automatons. Deming understood that an employee with a sick child, a toothache or some other “special cause” problem may not function at peak performance all the time. However, in a well-designed system, most employees will perform around a mean.
There will also be outliers who perform above or below the mean—though well-run organizations will have the fewest outliers because they’re hiring and training practices will guarantee a consistent level of performance. The work of high performers, Deming believed, should be studied; their work can serve as a model for improving the system.
Low performers, Deming believed, represent a failure of management to perform one of its key functions. Deming believed that hiring represents a moral and contractual obligation. Once hired, it is management’s responsibility to help every employee succeed whether via training or relocation. While it might occasionally be necessary to fire a poor performer, Deming believed this option should be a last resort.
Of course, Deming’s focus on the role of leadership and organizational culture presents special challenges for schools working under strict union contracts. In New York City, the best public schools are led by strong principals who consciously built collaborative cultures, in many cases because they were able to select their teachers and establish common expectations early on.
Transforming a troubled school can be harder—but not impossible; at Brockton High in Massachusetts, strong leadership and a core group of teachers who were willing to work together made remarkable change possible. But for many schools, building a collaborative culture will be challenging; it will require strong leadership and some flexibility for school principals to move the people they want into key positions.
In their new book A Smarter Charter, Richard Kahlenberg and Halley Potter highlight charter schools that are “breaking the mold and providing explicit means for teachers to collaborate and participate in school decisions.” Many of these charters, like Amber Charter in New York City, are unionized and have exceptionally high teacher-satisfaction ratings and an 89 percent retention rate. Some, including Amber, have specially tailored union contracts that are designed to balance teacher protections with flexibility.
One strategy that almost certainly does not promote collaboration is individualized incentive pay, which according to Deming, “nourishes short-term performance, annihilates long-term planning, builds fear, demolishes teamwork, nourishes rivalry and politics.”
Deming’s opposition to merit pay made him so unpopular among business leaders that many shunned him, a key reason he is not better known today. Even at Ford, Petersen’s immediate successors dismantled Deming’s legacy; but, when the company once again fell on hard times, it was revived by a new CEO, Allan Mulally .
Indeed, contrary to popular belief, and the magical thinking of business people who love individualized incentives, there is no evidence that pay is a driver of long-term performance improvement in industry. Decades ago, Frederick Herzberg, whose 1968 treatise against incentive pay, “One More Time: How Do You Motivate Employees?”, was the most-requested Harvard Business School article for decades, explained why money doesn’t motivate in the long term. Money, he argued, is a “hygiene factor”: Not enough of it causes distress, but money alone has little to do with job satisfaction or performance. Not surprisingly, companies are perpetually dissatisfied with their incentive systems, which leads to constant tinkering and more business for compensation consultants.
In education, the research on this is clear. A 2010 Vanderbilt University study shows that the performance of teachers who were offered a bonus of up to $15,000 was no better than that of teachers who were offered no incentive. And a survey of 40,000 teachers funded by Scholastic and the Bill and Melinda Gates Foundation found that only one-quarter of teachers felt that performance pay was likely to have a strong impact on student achievement; instead, what the teachers valued the most, according to the study, was “supportive leadership, family involvement in education, access to high quality curriculum and student resources, and time for collaboration with colleagues.”
Just a few years after Deming died, a hot new industry—software design—demonstrated another break in the connection between pay and performance. I have long wondered what Deming would have thought had he lived to see the open-source software movement in which thousands of volunteers produced higher-quality software than the private sector. Think Mozilla’s Firefox Web browser vs. Microsoft’s Explorer. The no-hierarchy, all-meritocracy culture of open source violates every assumption about the link between pay and performance: In open source, many software developers collaborate without any monetary compensation at all. Although some designers eventually find a way to monetize their contributions, few experts dispute that the principle incentive for open-source developers is reputational—the opportunity to do their best work.
If pay for performance is such a problem, why hasn’t this become more apparent in practice? At some highly competitive organizations, such as securities firms or companies like GE, pay incentives foster a culture of competitiveness that is considered important to the company’s organizational DNA. Then, too, incentive pay seems to work beautifully during good times when budgets are fat and there is enough money available to give most employees a “merit raise” or bonus. The problem with incentive-based pay is exposed during bad times, such as the Great Recession, when it becomes a zero-sum game that produces more losers than winners.
Indeed, the biggest problem with incentive pay is that it is inevitably seen as unfair. Evaluation systems linked to single metrics, like test scores, are easily gamed. More-nuanced approaches that include multiple measures, such as graduation and attendance rates, are often seen as too subjective. (While group incentives are more successful, they are not as popular.)
Consider what is happening in Louisiana, where education reform has focused on value-added measurement of teachers and privatization. With the help of Race-to-the-Top funding, which promotes incentive pay tied to student performance on test scores despite a continuing vigorous critique of such methods, the State of Louisiana instituted a bonus system in 2012/2013 school year.
The bonus plans were opaque, widely seen as unfair and based on constantly shifting criteria. As Mercedes Schneider, an English teacher, education blogger and recent bonus recipient explains, the incentive scheme is a crap-shoot, and one that unfairly disadvantages some teachers relative to others. (Schneider donated her $427.76 bonus to a friend raising an autistic child.)
To understand the Louisiana bonus system, it helps to know that by state law, 50 percent of teacher evaluations are based on student-learning measures. Louisiana began using highly controversial value-added measurements (VAM) to evaluate some teachers in 2012-13, but put the plan on hold in 2013-14.
The new bonus system is also based, in part, on student-learning measures. During 2013-14, the first year the bonuses were given, Schneider’s district administered a pretest at the start of the semester, which was intended to serve as a benchmark for the teacher-evaluation measure; but the bonuses were ultimately based solely on end-of-semester test results that were decoupled from the pretest.
This school year, 2014-15, the pretests were scrapped entirely and teacher performance was pegged to performance on end-of-semester tests with arbitrarily determined cut-off scores. This year, too, some teachers of non-tested subjects are being judged on core-subject tests they had nothing to do with.
Then consider the bizarre results of a bonus system at Pierre Capdau, a New Orleans charter school that is part of the New Beginnings network, where a handful of teachers got gargantuan bonuses. The highest award, for $43,000—not a typo—went to a fourth grade teacher who increased her student’s test scores by 88 percent. Meanwhile, the kindergarten teacher who had the highest test score improvement at the school—165 percent on the so-called DIBELS test, which is administered to grades K-2—got a fraction of that amount, $4,086, because kindergarten scores aren’t factored into state evaluations.
But things got even stranger. The stated purpose of the bonuses at Capdau, a failing school that got an F on its most recent state report card (see appendix here,) was to promote teacher retention. In New Orleans, a virtually all-charter district where large numbers of inexperienced teachers are not only overworked and underpaid, but have little training–a five-week Teach-for-America summer course for new college grads is typical—both morale and retention are real problems.
So, Capdau’s policy was to deliver the bonuses only if the recipients signed up to teach again the following fall. Yet, at least one teacher who was told he would get a bonus, was later informed that there had been a miscommunication and there would be no bonus after all. And the kindergarten teacher with the highest test score improvement was suddenly fired “without cause.”
These Louisiana’s examples reveal the pitfalls of a system that relies on individualized incentives to motivate employees to do their jobs. Deming’s work teaches us that schools, like all organizations, are interconnected systems that require close collaboration among stakeholders to improve. Yet, what possible incentive do teachers have to cooperate with each other if bonus systems are set up to benefit the very few? And imagine what $772,000, the total bonus pool allocated to New Beginnings’ three elementary schools, could have done in terms of hiring subject- matter experts, coaches etc.?
But don’t ask the Capdau principal who developed the bonus scheme—she’s already left for another job. (Indeed, the most rational minds in the charter community recognize that the key to retention isn’t bonuses, but better working conditions and training.)
Meanwhile, Capdau’s fired bonus recipient, Ashleigh Pelafigue, who immediately found a job in another parish, had this to say: “It was the best thing they ever did for me…I am flourishing and becoming even better in a supportive, appreciative and engaging environment that is well on its way to becoming an A school and leading the way to our parish’s continued success.”
If bonus plans, as a way to retain valued employees, tend to backfire, then ed-reform fantasies of forcing out ostensibly lackluster employees and replacing them with a cadre of superstars are not much better. Joel Klein, in his new book Lessons of Hope, proudly invokes the mantra of Jack Welch, the former GE CEO who encouraged New York City’s former schools chancellor to “hire slow and fire fast.” At GE, Welch was known as Neutron Jack for his penchant for firing employees. (More on Klein’s book in a future post.)
But even without a union contract, schools can’t realistically fire (and hire) their way to better results, according to a report by the conservative (and pro-charter) American Enterprise Institute. The total number of college graduates from Barron’s “highly competitive” or “most competitive” institutions in the United States is approximately 141,956 annually, according to AEI. If fully 10 percent entered into teaching for a two year period before moving onto other careers, it would provide 27,655 such educators annually, only 6 percent of the (438,914) teachers at work in the nation’s largest school districts.
Simply put, schools have no choice but to work with the teachers they’ve got. Let’s concede that teachers—like other professionals—can and should continuously improve their craft; and that some teachers should probably never have gone into the profession in the first place. (It’s also true that teacher preparation programs need to be improved. See Arthur Levine’s devastating critique of teacher education.)
Yet, in most school districts the percentage of teachers who are poor performers is quite low. Grover J. Whitehurst, director of the Brown Center on Education Policy at the Brookings Institution, recently estimated that a 5 percent ineffectiveness rate would be typical. For the sake of argument, double or even quadruple that number for the most ineffective districts. Yet, the education-reform movement is largely focused on weeding out a relatively few “bad apples,” rather than on finding ways too help the vast majority—85 to 95 percent of the work force—who are at-least-competent improve their practice.
Today’s education-reform consensus is a reflection of the ideology and outlook of the business people and philanthropists who fund the movement and who bring to it the same top-down, blame-the-rank-and-file mindset of the auto executives of the late 1970s and 1980s. Today’s quick-fix authoritarian strategies—from testing regimes to the failed $1 billion iPad gamble in Los Angeles—mirror the war footing of Detroit when GM’s CEO Roger Smith wasted billions of dollars on robotics as a way to solve the “people problem” in Detroit.
Then as now, the more authorities seek to blame rank-and-file employees, the worse things get. Fear and loathing led, in Detroit, to look-alike cars, poor quality, lost market share. In schools it has resulted in endless testing and test-prep, a narrowing of curriculum, and, no doubt, a deterioration in meaningful education—the kind that is difficult to test.
Then as now, the cognoscenti have sought to quantify all outcomes by looking at a narrowly defined bottom line. For automakers this meant focusing on short-term profits—even if that meant eroding quality and, eventually, market-share. In schools, they look at an ever growing array of test scores—most recently in New York State kids were subjected to at least three different standardized tests per year (one State test and two city tests designed for the sole purpose of evaluating teachers.) One problem is that because of budget constraints, the more you test, the lower the quality of the tests and the less meaningful they become. And given that the tests are cloaked in secrecy, they are stripped of their only pedagogical purpose—the ability to help teachers analyze what their kids know so they can improve.
Deming had little to say about labor unions. But he had famously excellent relationships with rank-and-file employees. Having said that, he would object to work-rules that make it difficult to foster collaboration and problem-solving. He knew that when quality improvement became a team project in the auto industry, it was associated with a much more motivated workforce, as well as more flexible work rules. Hourly workers who had been written off by the automakers rose to the challenge when given an opportunity to make meaningful contributions to quality. See here.
The lessons for education are clear: Quality improvement must begin with senior management (principals and education bureaucrats) establishing the conditions for collaboration and iterative problem-solving. It requires flexibility and professionalism from both teachers and education leaders. Finally, a climate of fear and finger-pointing will do nothing to improve schools; indeed, it is likely to set back the effort for years to come.