Portland State University – Seattle Education

We’re rapidly entering a world of evidence-based decision making in public education. These decisions will be powered by vast amounts of data run through proprietary black boxes that parents will have no way of understanding. The approach is called Moneyball and the goal is to justify ration resources to students –while investors make a tidy profit.

One of the most difficult challenges I’ve had as a parent is convincing other parents that the endless collection of our kids’ data isn’t benign and technology isn’t inherently benevolent.

Big Data, Like Big Brother, Isn’t Your Friend

As adults, we’ve chosen to ignore this cold hard fact: that by using electronic devices, we are allowing ourselves to become a product. Von Shoshana Zuboff calls this evolution in big data mediated economics surveillance capitalism:

It’s now clear that this shift in the use of behavioral data was an historic turning point. Behavioral data that were once discarded or ignored were rediscovered as what I call behavioral surplus. Google’s dramatic success in “matching” ads to pages revealed the transformational value of this behavioral surplus as a means of generating revenue and ultimately turning investment into capital. Behavioral surplus was the game-changing zero-cost asset that could be diverted from service improvement toward a genuine market exchange. Key to this formula, however, is the fact that this new market exchange was not an exchange with users but rather with other companies who understood how to make money from bets on users’ future behavior. In this new context, users were no longer an end-in-themselves. Instead they became a means to profits in a new kind of marketplace in which users are neither buyers nor sellers nor products. Users are the source of free raw material that feeds a new kind of manufacturing process.

As adults we’re vaguely aware that certain choices we make will impact our credit report. The inputs seem arbitrary and frankly ridiculous. Unless, there’s a problem, THEN, the unfairness of the system quickly comes into focus.

How your credit report is determined is an example of a black box. Inputs go in, something happens inside the box, and then your credit report comes out. What happens inside the box? Who knows? It’s a proprietary predictive model.

What sorts of random digital bits could impacts your credit report? Things like what operating system you use, if you do your browsing using a desktop or cellphone, even what you decided to use as your email address.

This excerpt is from New Study Shows You Can Predict Credit Rating from Your Online Tech Fingerprint.

And these things evidently matter a lot in predicting default.

When combining information from both variables (“Operating system” and “Email host”), default rates are even more dispersed.15 We observe the lowest default rate for Mac-users with a T-online email address. The default rate for this combination is 0.36%, which is lower than the average default rate in the 1st decile of FICO scores. On the other extreme, Android users with a Yahoo email address have an average default rate of 4.30%, significantly higher than the 2.69% default rate in the highest decile of FICO scores.

In general, there were a few things that jumped out as predictors.

iOS vs. Android (iOS users were around half as likely to default)

Emails with name in them were better

Desktops defaulted far less than mobile

People with numbers their emails defaulted more

People with old domain emails (Hotmail, Yahoo) defaulted more

People who ordered at night instead of in the afternoon defaulted more

The most interesting thing about this is how easily this stuff can be (and is) gathered from users just during a regular browsing session. Especially if you’re on the website itself that is going to making the decision.

The same goes for healthcare. Did you know insurance companies are collecting bits of personal information you leave online and then determining if you’re worth the risk of insuring.

This excerpt is from Health Insurers Are Vacuuming Up Details About You – And it Could Raise Your Rates.

With little public scrutiny, the health insurance industry has joined forces with data brokers to vacuum up personal details about hundreds of millions of Americans, including, odds are, many readers of this story. The companies are tracking your race, education level, TV habits, marital status, net worth. They’re collecting what you post on social media, whether you’re behind on your bills, what you order online. Then they feed this information into complicated computer algorithms that spit out predictions about how much your health care could cost them.

Are you a woman who recently changed your name? You could be newly married and have a pricey pregnancy pending. Or maybe you’re stressed and anxious from a recent divorce. That, too, the computer models predict, may run up your medical bills.

Are you a woman who’s purchased plus-size clothing? You’re considered at risk of depression. Mental health care can be expensive.

Low-income and a minority? That means, the data brokers say, you are more likely to live in a dilapidated and dangerous neighborhood, increasing your health risks.

Remember the argument about the internet being a democratizing platform which will lift up all the boats? Well, it seems to me the algorithms being used by these proprietary predictive models are just as abelist, classist, sexist, and racist as the business are in the real world.

And about that television viewing. The streaming sites you pay for consider you the ultimate and most profitable product. This couldn’t be clearer than reading the headline: You Are the Product if You Use a Roku Streamer, Says Company CEO.

Oh, and if you used RealNetworks RealTimes slideshow app, your face may have helped develop the companies’ new facial recognition software. Don’t expect royalties or even an apology from the company.

Since its founding in 1994, the company has built a steady stream of consumer and commercial digital media applications. One of the more popular is RealTimes, a mobile app that creates slideshows out of the photos and videos on users’ smartphones.

The app’s popularity also had a spinoff benefit, Glaser said: It provided the company with a database of 20 million faces, half of which RealNetworks has been able to use as a “training set” to teach machine-learning algorithms how to recognize individual faces.

Same for the 23andMe genetic testing kit. It appears your most personal of personal data was used to help ink a 4 year, $300 million deal between Galaxo and 23andMe.

Home DNA test results from the 5 million customers of 23andMe will now be used by drug giant GlaxoSmithKline to design new drugs, the two companies announced Wednesday.

It’s the biggest partnership yet aimed at leveraging the increasingly popular home genetic testing market, in which customers pay for mail-in saliva tests that are analyzed by various companies. 23andMe dominates the market.

A Predictive Model to Rule Them All

I recently went to a talk about child protective agencies using predictive analytics to help in the delivery of services.

One of the off-hand remarks, which I found stunning, came from the tech guy on the panel. He said it was difficult to use predictive analytics in schools because the environment wasn’t data rich and parents weren’t keen on their kids being treated as guinea pigs.

No one on the panel batted an eye.

So basically, while I and many education activist across the country have been working tirelessly to warn parents and school communities about the danger of data collection and the potential for misuse, this shocking bit of news was considered common knowledge in the academic world.

Think about that for a moment.

But then again, maybe the collection of personal data without consent is standard operating procedure in universities now days.

Consider the case of Portland State University and it’s violation of FERPA.

An internal review at Portland State University has found that a research project that gathered data from public K-12 school computers violated federal law.

As first reported by WW this winter, professors at the Graduate School of Educationimproperly asked student teachers in public school classrooms across Portland to collect and analyze demographic information and test scores (“Invasion of Privacy,” WW, March 7, 2018).

The university now concedes the project did not have proper authorization to use the data, and probably violated the Family Education Rights and Privacy Act, a federal law that protects student data.

What did Portland State do to fix the violation? It sent out a letter informing the school districts – not parents or guardians – their student data was used improperly.

Big deal.

What’s more, with such a light slap on the wrist you know there really is no reason for university researchers to change their practices for fear violating FERPA.

Predictive Modeling Can Be Wrong

Remember the financial melt down caused by the sub-prime mortgage scam?

The big financial institutions didn’t see it coming, because their predictive modeling said it was impossible.

How did that turn out?

Just because some model is fed data doesn’t mean its predictions are accurate. In the case of IBM’s Watson, those predictions can even be deadly.

Internal company documents from IBM show that medical experts working with the company’s Watson supercomputer found “multiple examples of unsafe and incorrect treatment recommendations” when using the software, according to a report from Stat News.

Stat reviewed documents that were included in two presentations given in June and July 2017 by IBM Watson’s former deputy health chief Andrew Norden. The documents were reportedly shared with IBM Watson Health management.

According to Stat, those documents provided strong criticism of the Watson for Oncology system, and stated that the “often inaccurate” suggestions made by the product bring up “serious questions about the process for building content and the underlying technology.”

Moneyball for Education

We’re rapidly entering a world of evidence-based decision making in public education. These decisions will be powered by vast amounts of data run through proprietary black boxes that parents will have no way of understanding.

The approach is called Moneyball and the goal is to justify ration resources to students. The old days of providing services to each individual student who needs them will be over. Instead, targeted interventions will be the name of the game.

The goal of moneyball is to create a compelling narrative that justifies and even celebrates austerity. Moneyball’s fundamental assumption is discretionary spending must continue to be cut and streamlined in the name of “funding what works”. This trick immediately removes from debate any discussion about cuts to non-discretionary spending –like the 50% of the federal budget that goes to defense.

The authors admit that denying services to everyone who needs them is unfortunate, but there’s always a silver lining: rationing services is a cheap way to create a randomized trial!

Why are districts pushing for kids on devices and most, if not all, of the curriculum delivered electronically under the catchy phrase of personalized learning?

Lots and lots of real time data.

Even better, the insatiable need by the state, impact investors, and non-profits for more and more student data makes the opt-out movement the perfect pawn in the data wars.

Let the opt-out movement push for the elimination of the end of year testing. Why?

Because tests like the Smarter Balanced Assessment just don’t deliver enough data to satisfy these outside groups.

How can the state ration resources if they don’t have enough data to run through their predictive models?

Further, how can the state decide which student needs a particular intervention or what student should be put on which career pathway without massive amounts of data?

How can social impact investors know if their Pay for Success or impact investment is paying off without massive amounts of data?

How can non-profits and other NGO’s prove they made their numbers to validate payment on their outcome based contracts without massive amounts of data?

It’s bad enough that most adults see nothing wrong with being turned into products that help Google, Facebook, and all of the other data harvesters meet their earning targets.

It’s another thing all together, when institutions and organizations who parents are inclined to trust, want to get in on the same game.

Treating kids as a data stream to be harvested and manipulated to meet the needs of managers and investors is wrong.

Some things should be more important than money. In a healthy society, our kids would be on the top of that list.

-Carolyn Leith

Category: Portland State University

Black Boxes, Student Data & Playing Moneyball for Education