\magnification=1200
\baselineskip=20pt
\nopagenumbers
\font\big=cmr12 scaled \magstep2

\centerline{\bf STANFORD UNIVERSITY}
\centerline{\bf DEPARTMENT OF STATISTICS}
\centerline{\big DEPARTMENTAL SEMINAR}
\bigskip

\baselineskip=12pt
\centerline{4:15 p.m., Tuesday, May 9, 2000}
\centerline{Sequoia Hall Rm. 200}
\centerline{(Cookies at 3:45 in 1st Floor Lounge)}

\bigskip
\baselineskip=15pt
\centerline{\sl Judea Pearl}
\centerline{\sl Professor of Computer Science}
\centerline{\sl UCLA}
\bigskip
\centerline{\bf ESTIMATING PROBABILITY OF CAUSATION}
\bigskip

According to common judicial standard, judgment in favor of the plaintiff
should be made if and only if 
it is ``more probable than not'' that the defendant's action was "the cause
for" the plaintiff's injury (or death). How can we estimate probabilities
of such hypothetical events as "was the cause for"?

I will propose a formal semantics, based on structural models, for the
probability PN(x,y) that event y would not have occurred if it were not for
event x, given that x and  y did in fact occur.  Armed with this semantics,
I will then explicate conditions under which PN(x,y) can be estimated from
statistical data, and show how data from both experimental and
nonexperimental studies can  be combined to yield information that neither
study alone can provide. Finally, I will examine when the standard
epidemiological criteria (e.g., Excess-Risk-Ratio) are adequate for
measuring the probability of causation, and how they can be corrected for
confounding bias.

Reference: J.Pearl, Causality, Cambridge University Press, 2000. Also, Tech
Report R-271, www.cs.ucla.edu/~judea/ 

\bye