two faces of causality: a small case study of the admission of scientific evidence to show causality...
TRANSCRIPT
Two Faces of Causality: A Small Two Faces of Causality: A Small Case Study of the Admission of Case Study of the Admission of
Scientific Evidence to Show Scientific Evidence to Show Causality in a Bias and a Toxic Causality in a Bias and a Toxic
Tort Case in the 4th CircuitTort Case in the 4th Circuit
Christina Kirk PikasChristina Kirk PikasLBSC 735: Legal Issues in Information LBSC 735: Legal Issues in Information
ManagementManagementDecember 11, 2002December 11, 2002
OverviewOverview
Review of the efforts made to form Review of the efforts made to form the admissibility of scientific evidence the admissibility of scientific evidence
Discussion of causality and the Discussion of causality and the scientific and the statistical methods scientific and the statistical methods used to proveused to prove
Case studies of two cases:Case studies of two cases: Product liabilityProduct liability Pay discriminationPay discrimination
Admission of Expert EvidenceAdmission of Expert Evidence
1919thth century century FryeFrye (1923) (1923) Federal Rules of EvidenceFederal Rules of Evidence (1975) (1975) DaubertDaubert Trilogy Trilogy
DaubertDaubert (1993) (1993) JoinerJoiner (1997) (1997) KumhoKumho (1999) (1999)
CausalityCausality Definition: “The principle of causal Definition: “The principle of causal
relationship; the relation between cause relationship; the relation between cause and effect” (and effect” (Black’s Law DictionaryBlack’s Law Dictionary))
Cause: “To bring about or effect” (Cause: “To bring about or effect” (Black’s Black’s Law DictionaryLaw Dictionary))
Correlation, association, or statistically Correlation, association, or statistically significant relationship is not enoughsignificant relationship is not enough
Primary issue in Primary issue in Toxic tortsToxic torts Product liabilityProduct liability Discrimination Discrimination
General vs. Specific CausalityGeneral vs. Specific Causality
General (examples: toxicology, General (examples: toxicology, epidemiology)epidemiology) anecdotal evidenceanecdotal evidence observational studiesobservational studies controlled experimentscontrolled experiments
SpecificSpecific Treating DoctorTreating Doctor Series of specific details such asSeries of specific details such as
•Biological plausibility•Consideration of alternate hypotheses•Cessation of exposure
•Temporal relationship•Strength and specificity of association•Dose-response relationship•Consistent with other knowledge
Case 1: Nettles v. Proctor & Case 1: Nettles v. Proctor & GambleGamble
Ms. Nettles used Vicks Sinex Nasal Ms. Nettles used Vicks Sinex Nasal Spray and later became blindSpray and later became blind
A neuro-opthalmologist was A neuro-opthalmologist was produced to give evidence on her produced to give evidence on her casecase No studies existed linking the main No studies existed linking the main
ingredient to her conditioningredient to her condition Only temporal connection was foundOnly temporal connection was found
As per Joiner – court did was neither As per Joiner – court did was neither arbitrary or capricious, decision was arbitrary or capricious, decision was affirmed affirmed
Case 2: Smith, et al v. Virginia Case 2: Smith, et al v. Virginia Commonwealth UniversityCommonwealth University
VCU employed a committee to VCU employed a committee to determine if there was a discrepancy determine if there was a discrepancy in pay between male and female in pay between male and female tenure and tenure-track professorstenure and tenure-track professors
The committee used a multiple The committee used a multiple regression analysis and determined regression analysis and determined that there was a $1,300 difference. that there was a $1,300 difference. Another committee was started to Another committee was started to review CVs and give deserving female review CVs and give deserving female employees appropriate raises.employees appropriate raises.
Case 2: continuedCase 2: continued
Plaintiffs AllegePlaintiffs Allege Not fair because raises based only on genderNot fair because raises based only on gender Inflated pool – more males had been Inflated pool – more males had been
administrators and therefore had higher payadministrators and therefore had higher pay Analysis not valid because did not take into Analysis not valid because did not take into
account major factors relating to pay, account major factors relating to pay, namely performancenamely performance
Trial CourtTrial Court Proxies were sufficient, regression study Proxies were sufficient, regression study
valid, pay handed out fairly, to correct valid, pay handed out fairly, to correct inequityinequity
Summary Judgment awarded to VCUSummary Judgment awarded to VCU
Case 2: ContinuedCase 2: Continued
Appeals CourtAppeals Court Regression did not take into account Regression did not take into account
performance factors, not invalid, but performance factors, not invalid, but probative value in questionprobative value in question
If material issues exist, should not have If material issues exist, should not have been a Summary Judgment, reversed.been a Summary Judgment, reversed.
AnalysisAnalysis If the lower court had employed If the lower court had employed DaubertDaubert
factors, the summary judgment was factors, the summary judgment was correctcorrect
The initial study was invalid – it poorly fit The initial study was invalid – it poorly fit the real situation under studythe real situation under study
ConclusionConclusion
Complexity of new cases, commingling Complexity of new cases, commingling of evidence, junk science make the of evidence, junk science make the gatekeeper role very importantgatekeeper role very important Judges see expert evidence 90 days Judges see expert evidence 90 days
before trialbefore trial Many courses, books, and studies exists Many courses, books, and studies exists
to help train judgesto help train judges Judges can appoint neutral experts to Judges can appoint neutral experts to
help interpret the evidencehelp interpret the evidence
More ConclusionsMore Conclusions
Scientific methods and statistics are Scientific methods and statistics are being used for purposes for which being used for purposes for which they were not designedthey were not designed Statistics don’t prove anything – give Statistics don’t prove anything – give
relative probabilityrelative probability Toxicology and epidemiology – give Toxicology and epidemiology – give
relative riskrelative risk Statistical significance and practical Statistical significance and practical
significance are not the samesignificance are not the same
FinallyFinally
DaubertDaubert provides a useful framework provides a useful framework if flexibly employedif flexibly employed
Resulting summary judgments save Resulting summary judgments save time and moneytime and money
It’s still easy to lie with statisticsIt’s still easy to lie with statistics