Volume 1 Number 3 - March 2004

Games of Chance

In case you haven't been near a television lately, the game of Texas Hold-em has become quite the rage these days. This version of stud poker has begun to rival video games for the attention of teenage boys. The popularity of this gambling activity leaves me with mixed feelings. While wary of its potentially addictive lure, I can appreciate the educational value of poker as an introduction to applied probability theory.

A basic understanding of the laws of chance is a fundamental prerequisite for functioning in modern society. Litigators in particular are often involved in high-stakes gambles on which huge monetary and other consequences may depend. For example, decisions about whether and how to utilize experts often hinge on judgments about the admissibility and persuasiveness of potential testimony. Knowing the odds of success (see article below) is valuable. So, perhaps there is more virtue than vice in this current fad. One can only hope.

More on Daubert and Statistics

In the previous issue of Statistical Retorts we discussed the admissibility of statistical evidence in light of recent court decisions stemming from the landmark Daubert v. Merrell Dow Pharmaceuricals case. At that point we had found only one very limited source of data on the rate at which statistical experts have been successful against Daubert challenges. Recently, however, we learned of a study by Deloitte & Touche that has examined the Daubert challenges in federal court for financial expert testimony, including that of statisticians from 1993 through 2002. This study reveals some interesting patterns.

First, the frequency of challenges has increased from a mere handful in the mid-1990's to approximately 50 per year over the last three study years. It will be interesting to see whether the level has stabilized or will continue to increase. Second about half (51%) of the 232 challenges found resulted in admission of the testimony, with another 16% partially excluded, and the remaining 33% totally excluded.

For statistical evidence, the results were very similar to the overall breakdown. Of 27 challenges, there were 14 admitted, 4 partially excluded and 9 totally excluded.

If we can assume that the total number of attempts to introduce statistical experts has not changed markedly, then these results show that Daubert challenges are much more frequent than several year ago. However, since nearly half of the challenges resulted in at least a partial exclusion, attorneys are far from indiscriminate. Taken together, these findings suggest that attorneys are becoming more sophisticated in their approach to proposed testimony by opposing experts. Accordingly, experts must anticipate the possibility of a challenge and be fully prepared to withstand it.

Statistical Evidence of Insurance Fraud

Insurance fraud and statistics? The connection may be far from apparent. Yet, statistical evidence can often play a key role in demonstrating the existence and/or the extent of fraudulent activity. Indeed, in prosecuting large-scale organized fraud, complex statistical analyses may be essential.

The focus of this article will be on auto injury claims, a fertile and well publicized area of organized fraud operations. The basic idea in auto injury scams is for an accident "victim" to exaggerate or even fabricate the injury. With the help of an unscrupulous attorney and medical providers, a minor or non-existent injury can be parlayed into an allegedly serious problem with extensive medical diagnosis and treatment. The result is sometimes known as winning the "insurance lottery" for all involved.

Until recently, this type of fraud was primarily the province of small-time professionals and individual claimants. In the last decade, however, organized fraud rings have become increasingly dominant. In some cases, the scale of the criminal enteprises has become enormous, and the sophistication of the perpetrators in operating these "businesses" and veiling their organizational structures has grown. Complex networks of medical clinics owned by holding companies and working with captive law firms generate thousands of claims, generate tens of millions of revenue dollars.

To discover suspected organized fraud operations is not terribly difficult. To prove fraud and demonstrate damages can be a lot harder. However, insurers are increasingly willing to accept the challenge of pursuing criminal or civil prosecution, as the effects of fraud on premiums spiral out of control in some jurisdictions.

In addition to traditional investigation techniques, the proof of fraud depends on establishing a pattern of practice for the medical clinics which serve as the engine driving large-scale fraud operations. Most "medical mills" have a very specific modus operandi by which large bills can be generated routinely.

Reviewing a valid statistical sample of the claim files and/or medical bills can reveal clearly the particular type of treatment pattern that is being "manufactured" by the clinic.

Demonstrating such uniformity of treatment across hundreds of cases, and tying this signature pattern to qualitiative evidence from medical experts and investigators, can paint a compelling picture. For example, in a widely publicized case in Massachusetts, we were able to show that certain diagnostic tests were applied in nearly all cases, although no useful information was being derived from them. In addition, patients were reported by the chiropractor to be disabled, despite evidence in the files to the contrary in a high percentage of these case.

In the Massachusetts case, the estimation of damages was based primarily on the payments to the chiropractor under first-party no-fault coverage. However, in many cases, insurers pay out under third-party liability coverage. These payments are in the form of settlements that include both reimbursement for medical and lost wage expenses and "general damages" that compensate the victim for pain and suffering. So, it may not be straightforward to determine how much the settlement has been inflated by virtue of completely or partly unnecessary treatment rendered. In this situation, statistical modeling can be useful to provide a scientifically credible estimate of excess payments by the insurer.

The general message is that insurance fraud has become big-time, and its prosecution therefore requires sophisticated methods to reveal and prove patterns of fraudulent activity. Statistical sampling and analysis can sometimes play a critical role in efforts to prosecute large-scale fraud rings.

Herbert I. Weisberg, Ph.D.
61 Pheasant landing Road
Needham, MA 02192-1000

Phone (781) 455-6850

Fax (781) 444-9563

email: hweisberg@correlation.com
Please let us know what you think about our newsletter. Thanks.