Hi Everyone:
In a recent posting to this list I complained about my graduating seniors
inability to analyze an issue well enough to determine whether they actually
faced an ethical issue. I used (perhaps unadvisedly) the VIOXX example.
Others pointed to the Race Car cases simulation of the Challenger disaster
which was right on the mark. At any rate I have added these responses below
for anyone that is interested. As an aside I will review expected value, chi-
squared, simple and logistic regression, normal and logistic distributions, to
see if it makes any difference in their ability to make a sound morally-
relevant decision.
>From Ann Bucholz:
I think it's a great idea, George. The challenge for me is that I always want
to do much more than the time allows. Still, there is no question that students
need more work in that area. In fact, apparently so do trained professionals.
The Shuttle Challenger might not have flown if trained engineers had not failed
to calculate simple probability correctly. The Carter Racing case is a great
tool for what you want to do. Its data mirror the Shuttle Challenger data and
so the case can have a real impact when students to realize they just launched
the Shuttle. I used it years ago but have not used it for a while.
>From T. Klein:
While some knowledge of statistical analysis is certainly useful in dealing
with this kind of issue, other considerations are probably more important.
The significance of the difference between 4 and 7 adverse events, as you
suggest, depends on sample size.
But the relevance of this difference depends on other matters:
How does this compare with alternative treatments - or no treatment at all?
What is the outcome of the adverse event, i.e., in how many cases, test vs.
control, was heart failure fatal?
What other effects, positive and negative, are attributable to the treatment
in question? Are we trading off 3 more deaths against hundreds or thousands of
improved lives?
What, if any underlying variables were associated with these differences?
(Clinical studies should and, I believe, typically do control for age, sex,
other medical conditions, etc.)
Also, as you suggest, economic implications would multiply these
differences. If each failure were given a value of, say, $500,000, that's a
$1.5 MM cost.
And, of course, there are other non-economic social and personal benefits and
costs to be weighed in a comprehensive evaluation of a treatment.
Finally, the treatment details may be at issue. Might screening the subjects
for certain conditions or modifying dosages, etc. change the results. This
case was presented in the press as "bad medication." Would some change in the
treatment protocol present a different picture.
Ethics instruction is a grand opportunity to stress critical thinking. We know
PR can put a positive spin on events that are not so positive - and the press
has a habit of making "aha" out of something on the basis of limited
information. Before we either indict or praise something, we ought to dig into
the situation a bit further.
In any event, business school seniors should have had enough statistics to
understand the point you want to make and taking class time to review
statistical techniques may be less useful than considering the kinds of issues
covered in my bullet list above.
>From Jean Pasquero
This is an interesting challenge.
I would formulate it a bit differently, though.
It's not stats techniques that you need to assign your students, it's
statistically based notions like acceptable risk.
For example: no medication will ever be 100% harmless on a large population.
There always will be at least a small subgroup whose particular conditions make
them over-react to the drug.
This group may be so small as to escape even the most carefully designed
battery of statistical tests.
For example, Vioxx was of great service to many suffering people.
It created side effects in (statistically) a few of them.
It caused impairments or even death for very few of them.
The questions then can be :
- what are the FDA requirements with respect to these potential cases ?
- how does the FDA decide in grey areas ?
- how far should we mandate drug companies to discover the possibility of
existence of outlier groups (like 3 standard deviations away, or more) ?
- what should the companies do when they discover outlier groups, before
launching the product, with or without FDA rules ?
- what's cover-up, where does it start and end ?
- when it comes to medication, should you use a utilitarian or a deontological
approach ?
- as a sufferer yourself, until how many % in a hundred would you accept the
risks associated with Vioxx ?
- should government prevent you from assuming this personal risk, and why ?
Maybe, one method could be to have students examine industry arguments in cases
like Vioxx, and ask them to weigh the validity of these arguments from both a
statistical and an ethical points of view.
jp
------------------------------------------------------------
>From Linda Trevino:
I'm frustrated by MY ability to quantitatively assess these outcomes. But, that
has little to do with statistics. In many, if not most situations, one simply
doesn't have enough information to project all outcomes. I call it the crystal
ball problem. We are pretty bad at foreseeing the future. So, we clearly need
other tools beyond quantitative ones (such as values). Also, Ann is right.
Cognitive biases that plague all humans as limited information processers
interfere significantly with this type of decision. In my textbook I deal with
those in a chapter that follows the prescriptive frameworks chapter and I use
the Pinto Fires case to illustrate that. If anyone is interested, Denny Gioia
(my colleague and Ford recall coordinator in the early 1970s) teaches the case
as a live case on a DVD we created that is available to purchase at a
reasonable price from Penn State media services. It's a phenomenal teaching
tool (if I do say so myself).
A seemingly quantitative decision like Vioxx doesn't get easier with statistics
either. To this day, there continue to be rheumatologists who would like to
have Vioxx available for their patients who are in chronic pain and who are
willing to take the longer term risk of cardiovascular problems in exchange for
the ability to get out of bed every day. Some people claim that Vioxx worked
better for them than anything else. Interestingly, Pfizer continues to sell a
similar Cox-2 inhibitor with a strengthened black box warning. Merck decided
not to reintroduce Vioxx (which they recalled voluntarily). I would have loved
to be a fly on the wall in those board rooms.
>From Beverly Gay
Another method that is taking this further and deeper is using 3D Immersive
Learning Environments (ILEs) or Virtual Worlds.
You set up a case study simulation that is unscripted for the participants (not
a "game"); it enables them to experience first hand the situation themselves.
Because it creates a sense of presence (sharing an environment with others in
real time), a mixed reality state occurs allowing the mind to extend into the
virtual environment. The participant feels it is real resulting in
psychological and physical reactions.
The learning is deeper and longer lasting because of these kinetic qualities.
Much like we learn in real life. It is not a passive experience, but a dynamic
interactive one instead.
--
From: John W. Dienhart
Identifying ethical issues is one hardest issues to breach for those who under-
identify them. I have a couple of ideas I would like to throw into the pot.
1. Ed Soule in are recent article in BEQ talks about imposing unknown
risks on others as one component of an ethical issue. This connects a well worn
business concept, risk, which business students respect, with ethics.
Discussing risk opens a door, as it were, to the world of ethics. I am third
author on this paper, but it is really Eds baby, and I was privileged to work
with him on it.
2. I would like to join with Linda endorsing videos. This connects to Pat
Werhanes famous and effective names and faces approach if we put names and
faces to business cases, the ethical issues become salient sooner.
3. One theme in Lynn Sharp Paines work is the structural similarity of ethical
issues at different levels analysis. For example, case where incentives that
induce copier salespeople to lie and cheat is structurally similar to cases
where incentives that make CEOs lie and cheat. We can use this in the
classroom by having students identify ethical problems they have encountered in
the workplace and use their experiences to see how the same issues are
replicated in other cases.
4. Metrics are important, but we need to know when to apply them and what
metrics to use. I agree with others that better stats does not make for more
ethical awareness. In fact, it could have opposite effect. Linda mentioned the
Pinto case, where the focus on metrics was a major factor in Dennis Gioias not
noticing the ethical issues.
Linda K. Treviño
I like the idea of connecting with "risk."
In this transparent world of ours, risk to the person's or company's reputation
becomes a huge factor, and one that students can identify with.
John W. Dienhart
Jean, great point. The only thing I would add is that that who bear the risk
should be informed and can refuse the risk if they so desire. I am not sure
those producing the product should impose their view of what acceptable risk
is. This is the responsibility of users of the product. A wrinkle here is that
some of the riskiest products we use are procured or chosen for us by others.
-------------------------------------------------
SIUE Web Mail
_______________________________________________________________________
To send a message to the list, send your email to
SIM@aomlists.pace.edu
_______________________________________________________________________
Visit the SIM Division website at:
http://sim.aomonline.org
_______________________________________________________________________
If you wish to unsubscribe from this list or change your delivery
options, you can do so online at:
http://aomlists.pace.edu/scripts/wa.exe?SUBED1=sim&A=1