How Data Ethical Is Your Data Analytics?
by Angelica Lim
Scales of justice...people who utilise data analytics for their decision making should be ethically bound by how they gather data to be analysed, how they determine the categorisation of the collected information and ultimately how the insights from the analysis are to be used.
Data analytics is featuring more and more in Shared Service and BPO centres. This is an expected evolution in progressing up the value chain - moving from delivering operational efficiencies and cost savings to helping generate insights which deliver business value for organisations.
These centres, with their vast reservoirs of data captured through immense volumes of transactions processed, are well-positioned to mine these information to extract valuable insights. The benefits of data analytics in Shared Service and BPO centres are many - e.g. improving the efficiency of the centres’ itself by identifying bottlenecks in transaction processing, determining opportunities to reduce Travel & Expenses, improving contract negotiations by revealing customers who are also vendors and finally, but not least, providing “on-demand” data analytics services which provide actionable insights to the business.
However, there is an aspect to data analytics which is often ignored in the pursuit of delivering insights for business benefits. It is the question of ETHICS. There are many who categorise data analytics as an information technology tool, as thus as a tool, it is ethically neutral. But this is a fallacy as the people which utilise data analytics for their decision making should be ethically bound in how they gather data to be analysed, how they determine the categorisation of the collected information and ultimately how the insights from the analysis are to be used.
Whilst the obvious wrongdoings such as unauthorised accumulation of data or the illegal sale of individuals’ information, are condemned and the infractors are pursued to the limit of the law, it is the grey areas of data analytics which are the most concerning. Take the well-known case where Target used data analytics to correctly predict that a teenager was pregnant and sent her coupons in the mail for maternity and baby products - before the high-school girl informed her parents! Whilst Target may have thought that this was a great way to identify a customer who was facing a major life-changing event and a potential for cross-selling; was there any thought to the impact this would have to someone who had not shared her pregnancy widely; not with her parents; and certainly not with Target. As Potter Stewart, quoted about said, ethics is knowing… “what is right to do”.
“Ethics is knowing the difference between what you have a right to do and what is right to do.”
…Potter Stewart, former Associate Justice of the United States Supreme Court.
Which brings us to the question of how do individuals view “what is right”? What may be right for one person may be an absolute abhorrence to another person. This is one of the dilemmas in understanding ethics and may be one of the causes for the challenges in the ethical implications of data analysis. One school of thought is that there are 3 ethical frames a person can use when assessing making an ethical decision - what is the Duty which you are obliged to perform; what is the Consequence of the decision made and what exemplary Virtues are you trying to emulate?
In a Duty framework, there are guardrails - policies, rules, regulations, laws - which set out the actions which an individual is mandated to follow. The outcome of the actions which is not a consideration in the decision making; it is about following the mandates - i.e. being ethical is about performing one’s duty as specified. Whilst this framework is the most straightforward to follow; it may produce actions which cause harm (e.g. when rules are blindly followed with no thought to the impact); or may be used as a convenient excuse when there is an ethical breach. The latter is the defence which Facebook used in the early days of the Cambridge Analytica scandal. In a situation where personal data of 50 million users were used to gain political insights, Facebook executives have claimed that from the company’s perspective, there was no ethical data breach. Information was neither stolen nor hacked - the individuals willingly provided their information when they signed up to a research app and provided permission for their data to be used for research purposes. Facebook followed the rules that were in place.
If however a Consequentialist perspective were taken, there may be a different conclusion to Facebook’s ethical responsibilities in this Cambridge Analytic issue. Taking a Consequence view means assessing the potential impact of the decision on people who will be affected both directly or indirectly by the decision made. Someone who uses the Consequence framework would typically want to make a decision which produces the maximum good. This may be seen as a compromise approach and there is seldom consensus on what maximum good means.
At a crossroads...Someone who uses the Consequence framework would typically want to make a decision which produces the maximum amount of good.
Take for example a hypothetical situation where a government requests a social media organisation to provide access to their users’ data in order to analyse and mitigate potential sources of terrorists threats. The debate here would be the sacrifice of private information of the many in order to protect the safety of the few. There is no right or wrong or easy answer - the questions are which consequence is more acceptable; and more acceptable to which party?
And if the intention is to adopt an ethical approach which is acceptable to the vast majority, then the Virtue framework comes into play. This framework is one that is people, instead of action based. Someone who is concerned about being viewed as an ethically virtuous person would use this framework. An ethical action is defined as whatever decision a role model virtuous person would make if in the same situation.
Applying this framework to the Target example above may have resulted in a different outcome. The background to the prediction of the teenage pregnancy was a request by the Target marketing department to a statistician on the possibility on analysing female customer data to predict if a customer was pregnant - even if the customer did not want Target to know this information. Whilst having this information would potentially help to increase Target sales, and thus increase value to the company, an ethically virtuous person may have chosen to not do the analysis as it may have been construed on snooping on a person’s privacy. However, a Virtue framework is the most difficult to apply because different individuals and different societies have significant variations in perspectives of what are ideal virtuous behaviours. There is no knowing for sure what Target would have chosen to do if they had applied the Virtue framework.
In the final analysis, Data Analytics is indeed the future of Shared Services and BPO centres. It is however, a responsibility of those who manage these centres to ensure that the extraction of data insights is done with a lens of doing the right things; instead of just doing what is your right to do. Ethics must be an inescapable part of seeking for and using of data.
Angelica Lim is a Big Data Researcher and the ex-CFO of BP Petrochemicals.
1. Duhigg, C. “How Companies Learn Your Secrets,” The New York Times, February 16, 2012, available at http://www.nytimes. com/2012/02/19/magazine/shopping-habits.html
3. Kuchler, H. and Garrahan, M. “Facebook in storm over Cambridge Analytica scandal” The Financial Times, March 19, 2018, available at https://www.ft.com/content/828e50ac-2ace-11e8-a34a-7e7563b0b0f4
4. Duhigg, C. “How Companies Learn Your Secrets,” The New York Times, February 16, 2012, available at http://www.nytimes. com/2012/02/19/magazine/shopping-habits.html
5. BBC. “Ethics Guide” available at http://www.bbc.co.uk/ethics/introduction/virtue.shtml
