In the absence of random and systematic errors, the average amount of student loans will equal to μ as shown below.  However, the real world data may not be μ.  Rather μ ± σ.  Therefore, the average amount is distorted by the standard deviation (σ).  The source of these distortions could have been caused by the 4 endogenous actors in the system as the Association has discussed in its BLOG.  Even the US lawmakers have admitted the existence of σ.  These are the proofs that σ does exist.  Hard evidences on the borrowers side also confirmed the problems.  Therefore, these issues cannot be denied any longer, and by anyone.

Systematic errors (SE) have distorted the student loans away from its population average amount by the factor of +σ minus errors due to randomness.  Distortion increases could have been caused by any factors, but systematic errors are one of the most damaging elements.  Letting these SE to occur is equivalent to pushing the current $1.5 trillions student loans to grow out-of control.

No analytics can cure for these type of errors for they may have been intentionally fed into the system by any of the players, or because moral is not in the equation when the decision or policy is made.

Before answering the above question, one needs to understand what is systematic error in the education industry, who are the producers and sources of these errors as the first step to understand the issues.  There are four major players (1). Regulators; (2). Administrators; (3). College decision makers and (4). American public i.e., students and their family.  The first-three group, as it has discussed in this BLOG are endogenous to the system, while students are exogenous, i.e., they do not have the power to influence the outcome.

When the regulator permits the for-profit institutions and the student loans servicing companies to go-public or selling their stock in Wall Street then we have another entity added as an exogenous force.  Therefore, all together there are 4 endogenous and one endogenous variable.

Let us define what could be the systematic errors in the education industry?  This type of errors are well known in the Physic research area where researchers collect information, data or events, based on tainted equipment that consistently produces biased outcomes.  One needs to know what is the desirable or optimal outcome before she or he is able to make the inference that other than the optimal outcome is the undesired or tainted results.  In the education industry the equivalent of “tainted outcome” are numerous such as lower graduation rate, pilled up student loans, unaffordable tuition, lower delivered education qualitys, higher operating cost, inefficiency and many others.

Now, let us take the following important questions:  Who are among these subsystems (endogenous variables) are the instruments that could have direct impact to generate the tainted outcomes?  The first-three members of the endogenous group have the power to make the public policy, while the fourth member of the group may have the power to affect the public interests through corporates’ policies.

• Regulator/law makers’, government and college administrator’s objectives should be consistent with the tax payers’ interests.  If not then they potentially are the sources and producers of the systematic errors.
• The Wall Street’s interests are leaning toward the corporates’ objective is to make profit.  Therefore, it may generate conflict with the tax payers’ interests.  This group will have higher probability to make systematic errors.

Things are out-of control if all these endogenous forces directly or indirectly working against the public interests.  The $1.5 trillion piled-up of current student loans which show no sign of slow down is nothing but showing the results of different policies that have been implemented by these endogenous groups caused the systematic errors.  Therefore, there is no cure for the current US student loans issue, until after these systematic errors are managed, minimized or eliminated.

The answer to this question is, it depends.  For public companies who sell their stock in the Wall Street—the CEOs often use the EPS (earning per share).  These companies are trying to beat the Wall Street analysts on the estimated EPS each quarter.  Otherwise, their stock price will be punished. Therefore, the motivation for these companies is profit margin.  This makes perfect sense for the for-profit entities.  The questions one may ask, will this EPS always has a positive correlation with time?  The answer is yes, so long there is a new innovation that will drive to the new product developments.  Otherwise, it will stagnant.

Let us turn our discussion to the education industry.  As the public may know–there are for-profit higher learning institutions and non-profit Colleges and Universities.  The for-profit certainly follows the Wall Street’s example.  Especially those institutions who are selling their stock in the Market.  Therefore, there is no surprise when the data show that attendees of these institutions are experiencing difficulties related to their student loans.  This may imply whoever permits these type of institutions to operate has made a suboptimal decision.

What about the non-profit higher educations?  What are the guiding principles that they look and apply in managing their institutions?  Do they have any guidelines?  Well, the School Board has the final say about a school policy.  Therefore, members of this Board also have to apply some sort of principles.  Did they?

If the current student loans data are the reflection of the results of these collective principles that have been applied in the industry for ages, then the answer is no.  Majority of the School Board and, therefore US Higher Learning institutions may not have a clear guideline when managing the school, except for using traditional metrics such as enrollment growth.  This student enrollment growth is equivalent with the EPS in the Wall Street.  Now that public may have a better picture what has happened.  This situation even more dire in the absence of strong regulations.

However, Adam Smith who is considered to be the Father of Capitalism has argued that including moral consideration when making strategic decisions produced results that are far superior than those of produced merely based on ratio or rational expectation.  Your thoughts?

Now that you may have or have not read the following article which we have shared and written in our BLOG many years ago.  Apparently, there may have some correlations why you have chosen to attend a Masters degree program in Business Analytics or to become a future data science professional with the information that we have shared in the past several years.  This article has confirmed AAEA’s early hypothesis, in that the US is experiencing serious shortages for data science professionals.  This is the reason why many higher ed institutions are offering the program, starting in the past a couple of years.

To the current graduate students, the Association has a little suggestion while you are still in your program.  It may help you to take useful classes or are thinking on doing a summer internship before finishing your graduate program.  First and foremost, your classes only fill a-half of the knowledge and expertise in preparing you as the future data scientist.  The other half is coming from the real world.  How can you fill and get the experience from the real world in data science, unless you have been or are employing in the industry or your current company.

If you start fresh the second career as a data scientist, a few things that will increase the odd (with disclaimer) that your dream will come true are:

1. Capitalize the real-world knowledge that you may have had in the past or currently.  For example, if you are, for whatever reason have received your undergrad in “soft” science which have forced you to work at a minimum wage company such as at one of the fast-food chain stores, you can and need to capitalize your past experience in the fast food industry.  Therefore, concentrate to understand what kind of data which you can analyze differently which will help the corporate office of that chain store may see your experience as important core competency.
2. If you never been working, fresh after your undergrad graduation to attend the grad school, you need to think what do you want to specialize in?  For example, after the second and third year, a medical student has a pretty clear idea what he wants to specialize in and where he wants to work and live.  Becoming a brain surgeon, while making the top dollars, may not be fun for those who cannot cope with the work related stress, plus it took extra years to finish the residency program.  If you are unsure, a good start is looking at a grocery store shelf.  The next time when you go shopping at one of your favorite stores, look what are available on the shelf.  This may give your some future ideas what industry or even a company you wanna to specialize in?  Great grocery stores have a wide range of products from the pharmaceutical to fresh produce.  From breakfast cereal to jewelry.
3. Do not avoid hard classes such as advanced econometrics or advanced mathematical statistics.  You got to have those, plus various programming classes where you can learn how to code using Python, R, SQL, C++, JavaScript,  etc.
4. You got to have advanced managerial/microeconomics, managerial finance, and if possible managerial accounting.  Depending on what the super-super specialist that you wanna to be?  Think what a fellow in medical school training meant.
5. You need mathematical programming classes where you will be taught how to model and run a stochastic simulation modeling.
6. If your have a two-year program, try to do a summer internship after finishing two semesters of your first year.
7. Be certified.  Example, SAS offers various options.
8. Write and present manuscripts at various professional annual meetings.  Net working with others in your area of interests.  A data scientist has to know how to write efficient reports – and most likely the readers may not have rigorous background in stat or math.
9. Try to teach undergrad classes–this is good training to be a great presenter and a public speaker.  Data scientists need to know how to present their works to others, which usually non-technical users of your findings.
10. Yes, you need to take 2 semesters qualitative theory, psychometric, survey design and advanced marketing research & consumer behavior courses.
11. Last, but certainly not the least important, work as hard as you can and avoid or minimize taking student loans 🙂

Now that everyone is talking about analytics.  When the Association introduced the approach for the first-time, different interest groups are rushing—try to take advantage of it.  The software companies are racing to get their first product in the market–try to get the competitive advantage to be the analytics product leader.  In 2013, there was only one institution which offer formal training in data analytics.  Now, there are more institutions offering Business Analytic programs than, may be the applicants.  They are mushrooming every where. This current development is paralleled with the booming of the MBA program back in the 80’s.  The Association glad be able to share and contribute to the development of the curricula not just in the US, but all over the world.

But, the business consideration and profit motive alone is not enough-companies need to educate their users.  Selling analytics is not just another hamburger store–try to sell as many burgers as possible.  It is more than that.  Most of the users think, analytics is just a point-and-click kind of thing which gives amazing results without knowing how the capabilities are structured and built. It is not just something one can produce and visualize, and show the output to their supervisor without knowing what is going on, and how the algorithm works, and what the assumptions are in finding those solutions. Weather the solution is unique or multiple?  In other words, it is not a canned software driven profession–rather the other way around.  The need to solve the real world problems dictates what kind of software needs to be built.  It is not software that derives the needs!  On other account, it is pretty funny that some users think data visualization and data analytics are equivalent.  However, they are not.  For example, an article written about a couple of years ago using the words “data czars“, yet some of them use data visualization and canned programs which is a bit of inflated to portray them as the “czar”.  However, the Association is glad as the innovator–the first entity to promote the application of analytics in higher ed, which then followed by the laggards.  The use of analytics may not exist, limited, unpopular or never been known or applied until after the article was presented and published.  By the way, predictive analytics is just only one component among many of the IRI-Education Analytics.

These are the flaws, facing most of the users who are lack of understanding about the statistical theory or mathematics behind the software.  Most of the analytics, not data visualization, is built, based on the unwritten assumptions in that the Central Limit Theorem is satisfied and that the data are random and the residual terms behave normally plus some other basic assumptions either for univariate or multivariate regression theory.

Sadly, our previous BLOG has mentioned pretty clear that majority of the events are not random, but systematic.  For example, a college administrator may constantly have made sub-optimal decision (in other words, wrong decision) simply because she or he does not apply any historical data to support her of his decision or using the :best guess” or because the past data are generated by the wrong or suboptimal policy.  In such a case we heard what has been mentioned as GIGO-or Garbage In, Garbage Out. If one applies any type of analytics based on tainted data that contained systematic errors, what kind of results will they produce?  Only garbage. A wise man suggests that knowing your data is the first step to use any analytics applications. Test if the events that generate those data satisfy and are consistent with the Central Limit Theorem.

Can a-non decision maker such as Business analytics or IR professionals make a correction or validating the tainted data pulled from Oracle or any database system?  If you are an analytics user, think about it, before getting too excited in applying point-and-click canned program in your cubical 🙂

Let us bring our discussions on student loans (SL) to the next level.  There are many sub-system entangled one-to-another that affect the student loans which all the American public has seen these days. If the System is broken, then one needs to know what are the reasons or what factors have caused the problems.  To avoid the unproductive discussions, let us approach this important scholarly work by using the real data with approaches that are academically sound, justifiable and verifiable by who ever that have the interest on one of the gigantic tasks to be solved in the modern history of the USA.  We will apply econometrics and multivariate statistics in these efforts.  To do that, let us bring what the possible relationship among the subsystems may look like Figure 1, as shown below.  Basically, there are four main elements–the US lawmakers, US regulator, the US higher ed institutions and the American public and students.  The first three are endogenous to the system, while the student (American public) sub-system is exogenous.  Therefore, the public sub-system is excluded in Figure 1.  So, none of the students has the ability to shape up the policy directly, but through the representative/lawmakers that they may have voted for, iff (if only if) these lawmakers truly represent them on such an important issue.

• The lawmakers who have the power to set the general policy has almost a full-control which direction for Uncle Sam to move to, for example NDEA. If things works as planned, this subsystem supposed to operate for the interests of the students or the public as their constituents.  Did they? may be not, read here.
• The legislator, based on the purpose why this country is established as stated in the Declaration of Independence, is also to work for the people of the country?  Did they?  May be not.  Read here for detail.
• Higher ed, especially at public universities and colleges are supposed to be managed for the interest of the tax payers.  Did they?  May be not.  Click here.

The overwhelmingly evidences show that none of these subsystems actually directed or managed for the benefit of the American public.  Therefore, the student loans are ballooning.  The real world evidences show that all the errors are non-random, but systematic or Bias E(β)≠β head. 

From the statistical theory, random errors, can always be fixed by adding the observations.  In statistical, mathematical concepts or measurement’s theory, there is no cure for systematic errors.  For example, in industrial engineering (IE) field, if a machine, say B that is used to produce the final product name A, consistently yield an output (product A) that cannot pass the quality control standard, then the factory manager got no choice, but to replace that particular machine (B).  But IE and the system of government is not a comparison apple-to-apple.  So, the readers can find their own answers….