Family Preferences Concerning Online Privacy, Data Mining, and Targeted Ads: Regulatory Implications

Abstract

Young Internet users engage in risky or inappropriate behavior online that could either be embarrassing or harmful to their future. As importantly, young Internet users engage in online activities that reveal a great deal about the cost to serve them and their willingness to pay for goods and services, which could be used against them by well-informed sellers. Educational applications that collect users’ information are becoming ubiquitous in the classroom, presenting the opportunity for students’ data to be mined. We are not aware of prior studies that examine parental or students’ attitudes and preferences toward data mining of educational application accounts, and how these attitudes differ across several countries. We used three survey instruments to measure parents’ and students’ attitudes toward data mining of educational applications. Parents in all countries studied prefer far less data mining of students’ online activities than seems to be the current practice. Most importantly, aversion to data mining does not seem to be correlated with awareness of current practices of data mining of teens’ activities. This study highlights regulatory alternatives and suggests future research and future data requirements for designing appropriate regulatory interventions. The nature of the intervention will be guided by the nature of the causes of inappropriate online behavior and inappropriate selection of educational software. Intervention could range from no regulation needed, through providing greater transparency, to new and detailed legal requirements that software providers must meet.

Key words and phrases::

• data mining
• educational software
• online behavior
• online teen behavior
• privacy
• privacy regulation

Notes

1. Several studies have found that nearly 20 percent or more of teenagers have sent sexually suggestive messages [83]. Other studies focus on whether teenagers are sending sexually explicit photographs; the results vary between 1 percent and 28 percent [46, 55]. Another study has documented that teens who sext are more likely to engage in sexual activities [84].

2. Regulators need to make complex trade-offs. Consumer surplus is generally seen as good because it represents consumers’ value above the prices they pay. In some markets, maximizing consumer surplus is the principal concern of regulators. Regulators generally view deadweight loss as bad because it represents a market inefficiency. Interestingly, complete and perfect price discrimination also reduce or eliminate deadweight loss. When sellers can charge only a limited number of prices, there will always be some consumers who are willing to pay more than the seller’s cost but less than the seller’s lowest price. As the seller becomes better informed, he may offer lower prices to consumers with the lowest true willingness to pay. Balancing the desire to maximize consumer surplus and the desire to reduce deadweight loss can be complicated. In some industries, such as air travel, with most aircraft largely filled to capacity, we might expect perfect price discrimination to increase airline revenues and reduce consumer surplus. Likewise, prices might increase for other services that consumers view as essential. It is probable that most faculty members would pay more for all of their telecommunications services than they now do, if service providers knew how much they valued the services and therefore could price these consumers more accurately. But we do not actually know what privacy policies would maximize total social or societal welfare. We are raising issues of fairness, which are explosively difficult for regulatory economists. Is it fair to force people with genetic predispositions to diseases to pay more for insurance? Is it fair not to do so, and to force the rest of us to subsidize them? The answer depends on which individuals you ask, and indeed, it depends on their ages when you ask them. Is it fair to charge some people more for the same product just because they announced that they really wanted it and Google overheard that? We surmise that most Americans would agree that this is not fair, even if it increases producer surplus by the exact amount that it decreases consumer surplus, and even if it leaves social welfare unchanged. This is a subject for further surveys.

3. That is, we trade off response rate in favor of having a survey that is representative. The response rate may appear low if we drop a large number of people because we do not need more people from their demographic group, but that low response rate is misleading.

4. We worked with an anthropologist who indicated that the best way to get accurate assessments of strength of feelings was to offer a five-point Likert scale and then combine categories. The term “strongly object” was created by combining the individual categories of “very strongly object,” “strongly object,” and “moderately object.” We excluded “weakly object” and “do not object.”

5. There will always be principal agent problems. School officials might violate parents’ wishes and ignore best practices, in exchange for payments from software vendors. Most jurisdictions already have laws against this, and define it as criminal behavior. In the worst cases, it is punishable by jail terms. No new laws should be necessary to address this.

6. The knapsack problem classically assumes that the value of each item is known, and is independent of the value of any other items that might be included in the knapsack. Carrying a tent without pegs might make sense, if pegs can be improvised from twigs encountered on the trail. Carrying pegs without a tent is harder to justify. Even when excluding interaction effects like these, solution of the knapsack problem is combinatorial and therefore intractable.

7. The Hippocratic oath starts by requiring that physicians must do no harm. Regulators should do no harm either, but assessing what regulatory actions cause harm, and to whom, is more complicated. We suggest that regulators should start by not reducing consumer surplus, by not increasing deadweight loss, and by not inhibiting markets’ ability to innovate. That is, regulators should not make markets less efficient, should not harm consumers, and should not harm market evolution and innovation.

Additional information

Notes on contributors

Eric K. Clemons

Eric K. Clemons is professor of operations and information management at the Wharton School of the University of Pennsylvania. His education includes an S.B. in physics from MIT, and an M.S. and a Ph.D. in operations research from Cornell University. His research for the past 30 years has involved the systematic study of the transformational effects of information on the strategy and practice of business. He was among the first scholars to study online global securities trading, business process outsourcing, channel conflict and e-commerce, and the effect of information on product proliferation and the transformation of consumer behavior in these new marketplaces. More recently, he has begun studying privacy and the challenges of applying current antitrust law to online business models. He is the founder and project director for the Wharton School’s Sponsored Research Project on Information: Strategy and Economics within the Program for Global Strategy and Knowledge Intensive Organizations. He has also held appointments at the Harvard Business School, the Johnson School of Management at Cornell, the Engineering College at Cornell, Hong Kong University of Science and Technology, Peking University Law School, and the Desautels Centre at the Rotman School of the University of Toronto. He has published over 100 scholarly papers and regularly publishes online in Huffington Post, Business Insider, and Tech Crunch.

Joshua S. Wilson

Joshua S. Wilson is research coordinator for Eric Clemons in the Department of Operations and Information Management at the Wharton School of the University of Pennsylvania. His research and teaching interests include strategic uses of information systems, information economics, online privacy, and the analysis of mandatory participation third-party payer markets such as search. He received his B.A. in philosophy, politics, and economics from the University of Pennsylvania.