Skip to content

Kaiser Fung on the ethics of data analysis

Kaiser gave a presentation and he’s sharing the slides with us here. It’s important stuff.


  1. zbicyclist says:

    These are great. I got talked into teaching a graduate marketing research course this fall, with ethics as one of the required units.

    I’m imagining a matrix with the examples on columns and the ethical philosophies as rows, and using this to drive the discussion.

    There are other examples, e.g. the relatively ordinary data mining described in the “Target knows when you’re pregnant before your parents do” NYT article a few years ago.

    I agree — it’s important stuff. This is particularly true when we get out of the realm of target marketing and into the realm of hiring and firing.

  2. Chris G says:

    Please indulge my cynicism for a moment… Chart #8: “Engineers aspire to make the world better.” Some do, sure. As with many professions there’s a fraction who are in it to do good but my experience is that most engineers do their job for the paycheck. Many find satisfaction in producing tangible solutions to tangible technical problems and in doing their job well but the suggestion that engineers in general aspire to make the world better is a laugher. (My experience is that scientists are much more inclined to want to make the world a better place.)

    • zbicyclist says:

      If so, then why are there far more evil scientist movies than evil engineer movies?

      • Corey says:

        There aren’t — movie makers just don’t understand the distinction.

      • Chris G says:

        I hadn’t considered Corey’s explanation. That’s food for thought.

        My thought was that ‘evil engineer’ movies would be too boring. A movie which captures the whole product development life cycle? Box office kryptonite. System requirements specification, allocation of system requirements to subsystems, requirements verification, requirements reviews, writing of specifications, system test plan development, subsystem test plan development, Preliminary Design Review and the 17 other gate reviews… And that’s before you even get to detailed design let alone doing subsystem fabrication and testing, system integration and testing, or any performance characterization. And let’s not forget things like software version control, configuration management, and environmental safety and health certifications. Urgh. It’s enough to make a normal person want to commit seppuku. (Well, at least it makes me consider it.) @#$% engineering process. People just want to see the damn death ray in action. I think that’s where evil scientists have the edge. They articulate the vision, build and test the first prototype, i.e. do the fun stuff. In contrast, engineering is a slog. By the time the death ray works (16 years and $2.4 Billion over budget) the audience doesn’t care anymore.

  3. Chris G says:

    Needs some Rawls.

    Looking at the “Major Ethical Philosophies” charts made my think “There’s a huge body of work in moral philosophy to draw on.” Rawls was the first person who came to mind. For lack of references on the subject, the the charts conveyed the sense that this is new ground. Obviously it’s not but my reaction those 3+ charts was that pointers to related reading would be constructive. Who’s the audience for this material? (What kind of people are enrolled in the Applied Analytics program?) Are they likely to go home and pull A Theory of Justice off their shelf or is this new turf for them? Would they benefit from some guidance?

  4. Rahul says:

    Perhaps it is productive to consider the ethics of data collection separately from data analysis?

    Also, there’s a peculiar issue of scale. Some data collection we didn’t object to in the past now seems iffy only due to scale.

  5. Bill Harris says:

    When reading this, Ulrich’s “critical system heuristics” came to mind as a way to think about who gets to make these sorts of decisions and who should get to make them. See, for example, and, for more depth,

Leave a Reply