Regression Analysis And Causation Analysis

  • strict warning: Declaration of flag_handler_relationship_content::ui_name() should be compatible with views_handler::ui_name($short = false) in /home/evidentl/public_html/sites/all/modules/flag/includes/flag_handler_relationships.inc on line 171.
  • strict warning: Declaration of flag_handler_relationship_counts::ui_name() should be compatible with views_handler::ui_name($short = false) in /home/evidentl/public_html/sites/all/modules/flag/includes/flag_handler_relationships.inc on line 226.

Although correlation of two variables is not conclusive proof that one variable causes the other, correlation is one of the more common technical tools that can be utilized for measuring a cause and effect relationship. Evidence of correlation is often cited as support for admission of expert testimony under FRE 702. The Fifth Circuit's examination of the role of correlation analysis in United States v. Valencia, 600 F.3d 389 (5th Cir. March 10, 2010) (No. 08-20546) provides a useful overview of some of the fundamental considerations guiding the use of such evidence

In Valencia the circuit considered an appeal by the defendant who was convicted of wire fraud for "manipulat[ing] natural gas markets." One problem inherent in the case was to demonstrate some connection between her allegedly fraudulent acts and the public harm which supposedly resulted. Valencia worked as a natural gas trader. Her duties involved involved surveying gas price indices published by private newsletters. She used these estimates of future gas prices to facilitate her trading activities for her employer. Her duties also involved supplying pricing reports to the indices' publishers regarding the specific trades she made for her employer. This information was used by the publishers to estimate periodic changes in the indices' prices. Valencia, 600 F.3d at 399.

The prosecution claimed that the defendant reported false information to the publishers, so that she could "raise the index price if she had excess gas to sell, or lower[ ] the price if she needed to purchase more gas." To achieve these price movements, the defendant reported "trades which never occurred, mistat[ed] the price [of trades that did occur] or volume of real trades, and omitt[ed] real trades. By swaying gas indices one way or another at certain locations," the defendant "could allegedly boost her monthly performance and increase profits for her employer." Not coincidentally, as she influenced transactions that turned out better for the employer, the defendant received promotions and higher year-end bonuses as a reward. Valencia, 600 F.3d at 399-400.

After the defendant's conviction, she appealed citing the trial court's admission of testimony by a prosecution expert witness regarding the impact of the false data the defendant supplied. In particular, the expert's analysis. .

The circuit was not persuaded that a correlation study was necessary. It cited a number of cases that concluded that "[e]vidence of mere correlation, even a strong correlation, is often spurious and misleading when masqueraded as causal evidence, because it does not adequately account for other contributory variables."

  • Huss v. Gayden, 571 F.3d 442, 459 (5th Cir. 2009) (dictum that evidence of injury by negligent prescription must be shown “to a reasonable degree of medical certainty” because "[a]ny scientist or statistician must acknowledge ... that correlation is not causation.”)
  • Munoz v. Orr, 200 F.3d 291, 300 (5th Cir. 2000) (for Title VII disparate impact case, “the evidence ... focus[ed] on the degree of statistical disparity between protected and non-protected workers in regards to employment or promotion”)
  • Sheehan v. Daily Racing Form, Inc., 104 F.3d 940, 942 (7th Cir. 1997) (Posner, J.) (in age discrimination case, “equating a simple statistical correlation to a causal relation ... indicates a failure to exercise the degree of care that a statistician would use in his scientific work” as well as [c]ompletely ignore[ ] ... the more than remote possibility that age was correlated with a legitimate job-related qualification, such as familiarity with computers.”).

While proof of correlation was not the indicator of conclusive proof, it was clear that often "evidence of correlation itself is potentially relevant and unlikely to mislead the jury." In such a situation "an expert who reliably discerns this relationship can present such conclusions to the jury," despite other limitations of statistical correlation evidence.

  • Pirlott v. NLRB, 522 F.3d 423, 435-36 (D.C. Cir. 2008) (expert witness allowed to present data concerning conditions under which a union can charge objecting nonmembers union dues through “a positive correlation between wages and union density in the relevant market at issue”)
  • United States v. W.R. Grace, 504 F.3d 745, 765 (9th Cir. 2007) (“the fact that a study is associational—rather than an epidemiological study intended to show causation—does not bar it from being used to inform an expert's opinion about the dangers of asbestos releases”)
  • United States v. White, 356 F.3d 865, 870 (8th Cir. 2004) (“Our court recognizes the known correlation between drug dealing and weapons, and accepts that they are closely and integrally related to the issue of possession of a firearm.”)
  • United States v. Hopkins, 310 F.3d 145, 151 (4th Cir. 2002) (expert testimony that defendant's behavior was consistent with dealing crack was relevant and reliable in light of expert's law enforcement experience and analysis of germane facts in the case).

Whether a particular opinion is relevant and reliable thus does not simply turn on whether the expert asserts a causal or correlative relationship, but is closely tied to the law and facts at issue in a given case. Hodges v. Mack Trucks, Inc., 474 F.3d 188, 195 (5th Cir. 2006) (“Of course, whether a proposed expert should be permitted to testify is case, and fact, specific.”) (Nos. 04–41362, 04–41764, 05–40686).

______________________________

Subscribe Now To The Federal Evidence Review

** Less Than $25 Per Month ** Limited Time Offer **

subscribe today button

Federal Rules of Evidence
PDF