Required reading

Heinzerling and Steinzor excerpts

Heinzerling, L. and R.I. Steinzor. 2004. A perfect storm: Mercury and the Bush administration. 34 ELR 10297.Read only the first section, "Unprecedented events."

Impediments to scientific progress: A generic description

Science for Business, Law and Journalism Chapter 19. Flaws in our ability to see the answer. Chapter 20. Some problems are intrinsically difficult.

A scientific paper on methylmercury

Grandjean, P. et al. 1998. Cognitive performance of children prenatally exposed to "safe" levels of methylmercury. Environmental Research, Section A 77: 165-177. (1) Read with a view of finding examples of the various impediments to scientific progress. (2) Also, come to class with your questions about the statistical models used in this paper.

Supplementary reading

There is no required reading on statistics, even though we will spend a fair amount of time discussing the statistical models in the Grandjean et al. paper. You might want to look at the statistics chapter in the Reference Manual on Scientific Evidence, 2nd edition, published by the Federal Judicial Center

Kaye, D.H. and D.A. Freedman. 2000. Reference Guide on Statistics. This contains much more material than what I intend to discuss. Just look at the basics.

Questions for class discussion

Scientific method (on asking the right questions)

1. Come to class with one good, substantive question about the science in Grandjean et al. (1998). (As a lawyer, you must be able to ask the right questions. The scientist you are speaking to should be able to provide the answers, but it is no easy task to ask the questions that will elicit the information that you really need. Here is your chance to practice asking a scientist questions.)

2. Some of scientific papers are quite good, while others have serious problems. The authors of the poor papers do not labeled them as such! How can you tell the good papers from the bad? Suppose one of the scientific papers in the course pack showed up on your desk, as part of a case you were litigating. How would you determine whether it was reliable? What questions would you want to ask the scientist who wrote the paper?

Statistical models

3. Craig will give a demonstration/presentation on statistical models (groan). Come to class with your questions about the statistics in this paper. Why do we need statistics? What questions do the authors answer with statistical models? What alternative do the authors have to the use of statistical models? What burning question about statistics did you always want to know the answer to?

Examples of impediments drawn from Grandjean et al. (2001)

4. The reading from Science for Business, Law and Journalism outlines four generic reasons that scientific problems are difficult, and sometimes even entirely intractable. Be prepared to cite examples of the problem of rare events and the ethical difficulties of experimenting on humans, drawn from Grandjean et al. (2001). Hint: There are obvious and immense ethical problems with a scientific study that directly studied the response of children to experimentally-imposed methylmercury.

5. What time lags between cause (methylmercury intake) and effect (cognitive performance) are a concern? Potentially, how long are these time lags? What model did Grandjean et al. use a model to "speed up" these time lags.

6. Observe that "complexity" is defined rather narrowly in Science for Business, Law and Journalism. Is there evidence that Grandjean et al. (2001) were (or could have been) overwhelmed with the sheer number of factors or objects that needed to be studied? If so, what model did they use to circumvent this problem? Hint: (1) How did they address PCBs? (2) Note that the human brain has a huge complexity. What models did Grandjean et al. employ to simplify human thought processes?

7. For each of the examples of impediments you cited above, discuss how Grandjean et al. (1998) used a model to circumvent it. What assumptions or limitations would be inherent in the use of each of these models? (This question overlaps somewhat with the previous questions.)

8. For each of the models you described in the previous question, discuss their accuracy, repeatability and convenience. Give a particular illustration (from Grandjean et al. if possible) of the tradeoff between choosing a model that is accurate, and one that is convenient.

"Good" and "Bad" models

9. In science, we are hardly ever able to say that a model is "good" or "bad". It all depends on the use to which the model is put. A model may be perfectly acceptable for one application, and horrible for another. Give and example, again drawn from Grandjean et al.