Why Technology Hasn't Lowered Costs or Improved Quality in Education

Below is an article that was published in the Spring of '06 that I wrote for Threshold magazine, a publication of Cable in The Classroom. I am on their Advisory Board. In short, in every other industry, technological innovation is an effort to increase productivity. Productivity is greater output with the same input of labor or the same output with less input of labor. No matter the case, productivity is dependent upon integrating technology with labor -- something that education at every level has willfully ignored.

---------------------

OUTSIDE HELP: IMPROVING PRODUCTIVITY IN SCHOOLS

After capital expenditures, instructional labor is education’s biggest expense. Yet, despite the enormous potential of technology for improving the quality and quantity of instruction, and the enormous pressures on schools to reduce spending, little has been accomplished to enable greater teaching productivity. Put simply, to improve productivity, either more or better instruction must be provided to the same students or more students must be taught with the same resources.

Schools that expect technology investments to drive quality improvements and cost reductions would do better by focusing on products and services that answer questions such as:

* How do we free up more time for our most skilled teachers to perform the most complex instructional activities?

* What parts of the educational process are best done by computers instead of people?

* Are there parts of the educational process that could be performed more cheaply by others?

* What services does my school provide that cannot be done better or more cheaply by someone else?

The answers to these questions are neither simple nor easy, and they threaten traditional notions of appropriate student:teacher ratios, quality control, classroom design, funding formulas, salary structures, and job responsibilities. However, every successful technology innovation in the history of humankind has enabled people to do more with less. Education should be no exception.

PRODUCTIVE APPROACHES
Though these questions may be thorny and difficult, the traditional 20:1 student:lecturer model is giving way to more productive ways of offering instruction. While most examples are in higher education, they do offer models for K–12 education as well. Schools are combining their own instructors with other instructional services to create programs that are cheaper, more effective, and more scalable. For instance:

To free up professor time for course management, course design, and student intervention, the Kentucky Community and Technical College System (KCTCS) contracted with a commercial company to grade student essays within 24 hours and adopted a textbook that included an essay-review service. The benefits? For the students, they receive suggested grades within 24 hours, have a third-party review, and the graders have more consistent training and provide more consistent assessment than could be provided in a traditional structure. For KCTCS, one instructor can teach more students, the per-student cost is lower, and some of the cost of providing the service is borne by the textbook provider.

After deciding to offer a program focused on math and writing for underprepared students, Kaplan University partnered with an online tutoring company to provide the assessment, content delivery, and tutoring components. The students receive tutoring access up to 24 hours a day, seven days a week. The school is able to shorten the development time of a program, reduce its risk and development costs, and rely on the expertise of a partner company for specific instructional activities such as online tutoring.

The National Center for Academic Transformation (NCAT) has worked with more than 50 colleges to redesign high-enrollment courses. NCAT results demonstrate that, by rethinking instructional labor strategies, schools can increase student success and decrease per-student costs. When schools make greater use of digital content and courseware and rethink student: teacher ratios, staffing patterns, and faculty roles, students need less intervention from live instructors, and more of what they do need can be provided by tutors, teaching assistants, or course coordinators.

The Baltimore City School System is one of many public school systems nationwide to hire teachers overseas to affordably meet the “highly qualified teacher” requirement
of the No Child Left Behind Act (NCLB). The system hired as many as 50 Filipino teachers to teach in the city’s schools. Additionally, many online Supplemental Education Service (SES) providers under NCLB use tutors located overseas. Whether in person or online, tapping the global market for educational labor allows more flexible staffing, and either lower-cost or more-qualified staffing.

WHY NOT?
Opponents might argue that the intimate teacher/student relationship is lost if grading is outsourced. However, the reality is that the intimate teacher/student relationship is an ideal that is far less common than we would hope and is far more expensive than we would admit. Initial results from most of the NCAT’s projects and from the Kentucky grading project show that student performance actually improved, while per-student costs decreased. Another argument against such models is that outsourced instructors will have different teaching techniques and standards than those taught at a particular school. While there certainly are some teaching functions that are best not outsourced—particularly those that require
a high degree of socialization, such as most teaching of elementary students—there are many functions that can be easily outsourced. For instance, math, science, and writing fundamentals are essentially the same across schools, states, and countries. Most schools are already comfortable with outsourcing at least some elements of education—many schools that offer distance-learning courses do so through third-party providers, and textbooks and courseware are the result of outsourcing content development and delivery.

Every year, the cost of education outpaces inflation with no increase in overall student performance. In "The World Is Flat," the best-selling book that attempts to define trends and technologies in the global marketplace, Thomas Friedman describes how companies in almost every industry are “insourcing” their logistics—letting partner companies manage the tracking and delivery of their products and services—and outsourcing elements of their production process. These trends are increasingly part of education as well.

No matter how it’s defined, education, like other hidebound industries before it, is about to become part of a global market. In other industries, this has resulted in products and services that are cheaper and of higher quality. Viewed one way, this threatens the cost and service structure of American education. Viewed another, this is an opportunity to rethink the components and functions of a school and all of the political, economic, and accountability structures that surround it. If doing so can improve productivity in schools, more students will ultimately receive more opportunities to learn, achieve, and succeed.

Service Level Chicanery

For online tutoring, students want on-demand help. Frankly, why else would they choose to be tutored online? As the online tutoring industry evolves into a larger market, it is interesting to note the lengths to which companies will go to to promise on-demand assistance. For instance, Tutor.com claims on-demand tutoring on its web site and in various corporate blogs. However, their service is only available from 11 AM - 10 PM (PT). Perhaps a more accurate claim would be "on-demand tutoring except for 13 hours per day." Worse, TutorVista (and many other small tutoring companies) claim 24/7 tutoring. In practice, this means that a student can schedule a tutor at any time for any time with sufficient notice. It does not mean that a student can get help exactly when they want it. At SMARTHINKING, we offer true on-demand math tutoring for the fall and spring semesters. During the summer, drop-in tutoring is available for 16 hours per day because there is not a sufficient volume of students to support full 24/7 access.

Why would companies spin these little white marketing lies? Because true on-demand tutoring -- 24/7, drop-in, live service -- is more expensive, requires greater scale, requires greater expertise, and requires greater data than pre-scheduled tutoring. These companies are trying to attract customers without investing in the tutoring force and management expertise necessary to offer on-demand service.

To understand why on-demand tutoring is more expensive and more complex, it helps to think of on-demand tutoring like a customer service center. When you call a company or utility, you expect to talk to someone on the other end. Running such a center is very similar to running an on-demand tutoring service. A critical operational variable in a call-center is "utilization capacity." This is the percentage of time that tutors spend tutoring students. It utilization capacity is too high then students have to wait a long time for tutors. If utilization capacity is too low, then the company is losing money. For instance, if a tutor is being paid $12 per hour and utilization capacity is 50%, the cost per hour tutored is $24. But, because one or more tutors are almost always available, there are no wait times. So, utilization capacity and wait times are inter-related variables. To do it right, a company needs to set a service level target and then determine the target utilization capacity to meet that target.

So, utilization capacity and service levels are mutually dependent variables. However, achieving a target service level also depends on the average length of a tutoring session. For instance, if a typical session in a call center is 2 minutes long, then a customer service center doesn't need to staff as many people to achieve a low wait time because customer service representatives are frequently available. However, for online tutoring, the average session length is around 30 minutes. This means that, with a small number of students per hour, an online tutoring company must have a very low utilization capacity to meet minimum desired service levels. As the number of students per hour rises, utilization capacity can rise while keeping service levels constant. Because utilization capacity can rise, it costs less to offer on-demand tutoring with a large number of students than with a small number of students.

Lastly, to manage all of this efficiently, an online tutoring company needs management sufficiently skilled to schedule tutors appropriately and sufficient data to know when the peaks and valleys of demand are likely to be.

So, to sum up, true on-demand tutoring generates higher labor costs because a portion of a tutor's time will not be used, requires scale to keep the unused portion of time as small as possible, requires management expertise, and good data systems. This is a significant investment that all small online tutoring companies are unwilling to make. The expense of true on-demand tutoring combined with its attraction to customers is the impetus behind the false advertising cropping up in the online tutoring industry.

Defining Online Tutoring Quality

What is quality education? What is quality teaching? How can one measure it? These are some of the thorniest questions in education today. No single system or metric can determine it. For instance, standardized tests suffer from the fact that students may not test well, may have been trained on the wrong material, or may be having a bad day. Student survey data may reflect student opinions of the teacher rather than opinions of the learning. For instance, numerous studies in post-secondary education show a positive correlation between lenient grading and student satisfaction. Longitudinal data from schools, such as job placement rates or lifetime earnings of students, cannot be easily compared to each other because students at different schools enter and exit with different skill levels. Lastly, portfolio analysis -- the compilation and examination of a given student's work over a period of time -- suffers from the subjectivity of the teacher. Due to the flaws of any single metric, those that need to measure educational quality -- such as schools, accrediting agencies and parents -- are forced to rely on a meta-analysis of all of the metrics listed above, input analysis (credentials of instructors, training processes, and others), and reputation.

So what does this mean for online tutoring? A quick search of the Internet will demonstrate that every online tutoring company claims to have "high quality" tutors. Most will claim that their tutors are extensively trained. All list fabulous quotes from users. All show terrific survey results. All claim grade increases. So, how does a parent or a school determine who really is better? Using the above framework, they need to look at inputs, metrics, and reputation. Of these three, inputs are the least manipulable because inputs impact the cost structure of a business. They are also the least advertised, because they are a proxy for educational quality. However, they probably provide the best indication of the educational value of a service.

INPUTS
By looking at what goes into the tutoring process, one can get a sense of what should come out. The inputs that are relevant to online tutoring are tutor credentials, tutor training, tutor oversight, tutoring philosophy, service availability, service levels, breadth of service, and ease-of-use.

1. Tutor Credentials: Generally, tutors with advanced degrees in their discipline have a better understanding of the material than those that don't. This is particularly relevant in math and science. Companies that don't indicate the degree levels of their tutors are typically relying on current college students or graduates as opposed to masters level and PhD tutors.
2. Tutor Training: Just about everyone has been taught by someone who is brilliant in their discipline, but doesn't know how to teach. Tutor training on how to tutor is just as, if not more, important than subject level expertise. When students need help, they are typically frustrated and lack confidence. An effective tutor not only helps them with the subject matter, but encourages them as well. At SMARTHINKING, we call this the "affective" element of tutoring. Again, every company will claim to do this. However, if a company is serious about tutor training, its tutors will be part-time employees as opposed to independent contractors. Though hiring tutors as independent contractors is simpler and cheaper, IRS regulations require that any position that requires significant training as a job requirement must be filled by an employee.
3. Tutor Oversight: Tutoring is a one-to-one social experience that combines the delivery of subject knowledge with, hopefully, the social cues necessary to encourage the learner. However, as with any human interaction, there can be miscommunication between student and tutor, mistakes made in the provision of information, and differences in communication styles. Identifying and addressing these sorts of issues requires the subjective perspective of experienced educators. At SMARTHINKING, we have former college professors that oversee each of our disciplines. Further, they contribute to the scholarship surrounding online tutoring.
4. Tutoring Philosophy: An online tutoring service can be a learning service or an answer service. While not mutually exclusive, they are certainly not the same. Frequently, students want an answer, not a lesson. Tutoring companies that evaluate their tutors solely on student satisfaction ratings give tutors the incentive to do the work for the students. While this might make good business sense for a company trying to sell its services to consumers, it's not good education.
5. Service Availability, Breadth and Levels: All of these inputs are for naught if an online tutoring service doesn't provide service at the moment and in a subject that the student needs it and without an onerous wait time.
6. Ease-of-Use: Again, all of these inputs are for naught if the online tutoring technology is not user-friendly and capable of supporting educational interactions. For instance, Voice over the Internet (VOIP) is a feature that is frequently requested, but seldom used. This is because the set-up and performance for VOIP -- microphones, speakers, soundcards, volume settings, bandwidth, archiving, non-duplex service, voice lagtimes, and others -- make it more cumbersome than helpful. Also, math notation is notoriously difficult to do on the Internet. Technology that enables easy superscripts, subscripts, fractions, graphing, and other mathematical symbols enables effective online tutoring.
   

METRICS
Every tutoring company, SMARTHINKING included, touts its student survey results, quotes from satisfied customers, and sample tutoring interactions. Frankly, all of these are easily manipulated for marketing purposes. The only outcome metric that is more objective than the others are independent studies conducted by clients or others. While these will certainly have their share of methodological flaws, the bias of the company is removed.

REPUTATION
Because outcome data is so manipulable and input data is hard to discern, reputation and, by extension, brand, plays a role in determining educational quality. In large part, reputation is created by quality service over time. It is also reflected by a company's client base. For instance, to determine educational quality, one might look at the tutoring services chosen by other schools and educators. Presumably, other educators will vet a tutoring company on its educational features more fully than a parent, student or library might.

Despite the emphasis on testing in No Child Left Behind, the pressure being put on college accrediting agencies to measure school quality, and the growing demand for consumer educational services, the question of what is educational quality is not likely to be solved soon. In the meantime, those who need to look at quality are best served by evaluating an education provider holistically on inputs, metrics, and reputation, with an emphasis on inputs.

The Online Tutoring Landscape

In a previous post I noted that there are 2 kinds of online tutoring: Prescriptive tutoring and Drop-In tutoring. SMARTHINKING focuses on the Drop-In model. Within the drop-in market, there are 2 established companies -- SMARTHINKING and Tutor.com -- and at least one well-funded new company who may offer drop-in tutoring (TutorVista). Tutor.com has built its business by selling to public libraries who target middle school students. SMARTHINKING has built its business by selling to colleges, some high schools, and bundling with textbook providers. SMARTHINKING has tried selling to public libraries with limited success. Tutor.com has tried selling to colleges with limited success. Each company has tailored its service and pricing to fit its market.

Though SMARTHINKING has little or no competition in the higher education market, I have always thought that some company would emerge to compete with us. However, 2 dynamics have emerged over the past six months that will make it very difficult for new entrants to compete successfully in the college market.

The first dynamic is the growth of our partnerships with publishers. Starting in June, 4 of the 6 largest college textbook publishers representing over 70% of the college textbook market will bundle SMARTHINKING's services with selected products. These publisher partnerships will allow the 300+ schools who contract directly with SMARTHINKING to offer more tutoring services without incurring extra costs. For publishers, the addition of SMARTHINKING helps sell more books because schools want to extend their existing SMARTHINKING services. Schools have an incentive to contract with SMARTHINKING because publisers offer ways to defray costs. SMARTHINKING can now offer a school a complete solution on a single platform that can integrate SMARTHINKING tutoring, a school's own tutors, and tutoring provided by publishers. In addition to all of this, the branding that results from the marketing efforts of these publishers cannot be underestimated.

The second dynamic is the dramatic service level improvements that come with scale. Running a drop-in tutoring service is like running a call center without the call and without the center. The biggest difference is that session lengths -- the time spent on a call -- in most call centers is short. For online tutoring, the session lengths are long. This means that, to achieve consistently low wait times without signficant overstaffing, a drop-in tutoring center must have a lot of users. Our models indicate that a center needs about 35 sessions per hour to achieve average wait times under 4 minutes with an efficient staffing model. In the past year, SMARTHINKING has grown such that we meet these service levels across almost all of our available hours. Competitively, this means that any new entrant will need to signficantly over-staff prior to achieving student volume if the new entrant plans to compete on service levels. For SMARTHINKING, we will begin to integrate service levels as a marketing message to our current and future clients.

These 2 dynamics -- the interdependence of our schools and publishers and the predictability of service levels -- makes me increasingly confident in SMARTHINKING's ability to grow within and defend the college market from new competitors. These dynamics, plus the college focused service elements that we already offer such as 24/7 drop-in math tutoring and the world's largest online writing lab, make us the only choice in the college market.

Colleges (and High Schools) Will Compete On Academic Service Levels

Recently, I have given several presentations to higher education administrators about how call center theory can be applied to educational services. An online, drop-in tutoring service, like SMARTHINKING, is essentially a call center, but without the "call" and without the "center." To create this presentation, we developed a fairly complex model to analyze the interaction between the following variables:

• The number of tutors staffed;
• The average number of students expected;
• The length of a tutoring session;
• The desired average wait time for a student (the service level);
• And the variance embedded within all of these variables.

With this model, we can predict the number of tutors that need to be staffed (ie. 15) to meet a desired service level (ie. average wait time less than 4 minutes) with a desired tutor efficiency (ie. 75% of the time a tutor is working with a student). The upshot of all of this was that, to run a cost-effective, drop-in tutoring service with reasonable service levels, you need to have A LOT of expected students. This is becaues the session length for tutoring is far longer than that of your typical call center. I found this model fascinating, and I was sure that others would as well.

Others didn't. Perhaps I need to polish my presentation skills, but when I was explaining the model and the results I saw a lot of vacant stares and more than a few nodding heads. However, rather than ascribe the lackluster response to my own performance, I'll use it as a pretext to draw some generalizations about education. At a minimum, this will rationalize away my disappointment. Perhaps it will even be insightful!

As I thought about my presentation, I think the real reason that listeners were unmoved is that the presentation correlated with almost nothing in their daily work lives. In fact, at the beginning of the presentation, I asked how many schools had defined service level goals for faculty members to meet. For instance, were faculty required to return a paper or an e-mail in X amount of time? Only 1 school out of 150 indicated they had such service levels. It occurred to me that these basic principles of service are simply not considered when schools deliver education. This, I think, will change dramatically in the next 10 years.

As more classes are taken online, higher education increasingly becomes a commodity. Online, geographic barriers to student choice are gone. The remaining differentiators -- price and quality -- remain. In my opinion, the traditional pricing structure of higher education will soon crumble as well. If students can take an english 101 course at a community college for 1/3 of the price of the 4 year college, and the credit is comparable AND the student can take it online, pricing will eventually become more rational. With these changes, the only element left that an institution controls is its academic quality. Within academic quality, the content of general education subjects rarely change. So, for those schools not competing on student selectivity or brand, all that's left is the level of service that they provide to the student.

This is a lesson that has already been learned by higher education administrators in admissions and technology. Colleges understand the impact of service levels when recruiting students and the impact of service levels as it relates to tech support. It is ironic that the real product of education, student learning from courses and services, hasn't incorporated any of these lessons. For most schools -- particularly public institutions -- this is a result of the traditional higher education governance structure. Traditionally, academic decisions and business decisions are made by the faculty and the administators respectively. For schools that want to compete successfully online, academic and administrators will need to work together to focus on the services that provide the greatest benefit to their customers -- the students.

The Direct To Consumer Conundrum

I'm digging this blog idea. Prior to starting SMARTHINKING, I was a freelance writer for education and technology issues. I wrote for a variety of magazines and institutions -- Converge magazine (now defunct), Wired (a couple of very small pieces), CEO Forum on Education and Technology, National School Board Association, and others. This feels like those days again...

In the last six months, there has been resurgent interest in selling online tutoring services directly to consumers. One company plans to offer low-cost tutoring from tutors located in India and has raised over $10 million to try it. Another company is trying to raise a similar amount to aggressively expand into the direct-to-consumer market. This investor fervor hearkens back to the days of the late '90's when investors and companies showed similar irrational exuberance over the online tutoring market (from which my company benefited, to a degree).

The last decade is littered with companies that have tried to sell online tutoring services directly to consumers. In the late '90's one company raised close to $20 million on the theory that it could create a market for tutors offering their services and students needing tutors. Another well-known company invested close to $40 million to develop a direct-to-consumer online tutoring service. While both of these companies are still around, their business models have radically changed. After plowing through its $20 million, the first company bought a much smaller company that allowed them to sell services directly to public libraries -- effectively abandoning the direct-to-consumer model. It is now trying to resurrect the direct-to-consumer channel. The second company integrated its online tutoring offerings into it's place-based tutoring services -- effectively creating a hybrid tutoring option. Neither company has come close to justifying the original investment. From my own experience, SMARTHINKING ran several well-executed pilot programs targeting the direct-to-consumer market with very little success.

Tutoring seems like a market ripe for "disintermediation." According to Eduventures, the tutoring market is worth $4.5 billion and growing at 15% per year. It's also highly fragmented with price points ranging from $8 - $200 per hour, varying levels of quality, and imperfect mechanisms for quality assurance. In short, it seems like a perfect market for the aggregating ability of the Internet. In theory, some company should be able to offer a consistent level of service at a relatively low price to create a respected online tutoring brand that would aggregate both tutors and students. In practice, this has not been the case.

So, what happened? One answer might be that students don't really like online tutoring. Perhaps face-t0-face tutoring provides a level of personal interaction that is lacking on the Internet. While this is undoubtedly true, online tutoring offers other advantages like convenience of time and place, anonymity, archiving, and others. Further, usage of SMARTHINKING's online tutoring services is growing by more than 50% per year with most students using the service more than one time. Student surveys show that students really like the service. So, again from our experience, this isn't the case.

I believe that that the direct-to-consumer market hasn't worked because there is a mismatch between the market to whom online tutoring appeals and the consumer purchasing patterns for educational services. There are 2 types of tutoring. They are "prescriptive" and "drop-in." Prescriptive tutoring is where a student attends regularly scheduled tutoring sessions, frequently with the same tutor. Often, this tutoring will be tied to a pre-tutoring assessment to identify student weaknesses. The largest market for Prescriptive tutoring seems to be in the K-7 grades. Drop-in tutoring is where students get help from tutors when they need help. Essentially, this is a call center model for tutoring. Almost the entire consumer market for tutoring resides in the Prescriptive market. All of the well-know tutoring companies offer Prescriptive tutoring services. Prior to the advent of the Internet, Drop-in tutoring was restricted to places where students could be aggregated to create sufficient volume to offer a drop-in service. In practice, this was limited to learning assistance centers (ie. math labs) at colleges and universities. The Internet has allowed Drop-in tutoring to expand beyond the residential school. Because Drop-in tutoring requires students to initiate the interaction, this is most appropriate for high school and college students.

This is a rough and incomplete generalization, but I believe that education has 2 functions. These are socialization and knowledge transfer. When students are young, socialization is more important. As students age, knowledge transfer gains in importance. No matter how well constructed, online tutoring is simply a less powerful socialization experience than face-to-face tutoring. Therefore, online tutoring has not been as popular or successful with younger students. On the other hand, the convenience of the drop-in model of online tutoring is excellent at knowledge transfer. This model is used successfully by older students -- typically high school and college students. So, to sum up, online tutoring works well for older students, but not as well for younger students.

This is another rough and incomplete generalization, but consumer purchasing of educational products and services is restricted to three general categories. These are:

1. Educational objects and toys -- like reading software or Leapfrog's toys;
2. High stakes test prep -- SAT prep, etc...
3. "Get Ahead" services -- Tutoring for young students.

Numbers 1 and 3 appeal to parents of younger students. Number 2 appeals to parents and students in the high school and college levels. Parents of high school students are willing to pay for test prep because there is a very clear goal in mind. They have not been willing to purchase Drop-in tutoring because the value isn't as clear. College students will spend a ton of money on tuition and textbooks, but don't buy much of anything after that. Again, because the relationship between an external tutoring service and passing a class isn't extremely obvious. What this means is that, the element of online tutoring that works well -- Drop-in tutoring-- is not well matched to the buying power and habits of the students for whom it works well.

It is entirely possible that I have misjudged the market dynamics. Perhaps the consumer market for online tutoring in the late '90's and early '00's wasn't mature enough yet. Perhaps the new marketing power of search engines is enough to create this consumer market. Perhaps today's investment in the consumer online tutoring market will prove to be rational exuberance. However, I don't think student and parent buying patterns have changed very much. Here are the lessons that I draw:

1. Drop-in online tutoring works well.
2. Drop-in online tutoring is best sold as an "add-on" to a school or as part of a bundled solution with another educational product.
3. Students will come to expect Drop-in online tutoring as part of their educational experience, rather than purchase it independently.

The next year will tell if there really is a consumer market for online tutoring.