Longbranch Research Associates

Statistical Analysis In Litigation

What you can do at this site:

Longbranch Research Associates was born in 1979 as Econometric Research, Inc., and that is actually still its official name. It changed what it is called because its scope is much broader than econometrics. For example, how many econometricians know what "multiple pools analysis" is, or proportional hazards models? More importantly, if econometrics is a hammer, econometricians think every problem is a nail. We find such people uncreative, and usually wrong. One needs to understand the problem posed in litigation, the laws under which the plaintiff has brought charges, and the workings of any institution relevant to this complaint. Only then can one determine the appropriate data and analysis.

Go to our about page to learn about some of the methods we have pioneered, our values and how we go about our work.

Go to our services page, to learn more about what we do and have done for clients. There is also a contact form which will send us an email. Tell us who you are, something about your case, and how to contact you.

Go to our publications page, to read about and purchase the 2006 book: "The Expert: The Statistical Analyst In Litigation." There may be other material there for sale and even some for free.

This is a web site dedicated to what LRA does, which is analyze data to inform a decision-maker (usually a judge) about an issue (usually constrained by law) in a context (often an institution). Our product is usually a report and expert testimony. The more technical issues that could be presented in this LRA web site have been moved to StatisticsInLaw.com, to allow more extensive presentation and discussion. For example, we believe that almost all "analyses" of discrimination are based on an outcome (say, the race composition of a police force), where discrimination is a process (hiring from among applicants). Most such analyses---whether they "won" or "lost"---are wrong. We name names and cases where that has happened. Failure to understand and analyze the process has led many an "expert" into misleading his/her client, who then loses a case that perhaps could have been settled.

Cases are organized in this site in two ways. For one thing, we take on current issues, current cases. They organize themselves by date. Second, we also try to organize by subject, to generate a string of posts. That is, one can just ask "what are current cases in this area," or one can ask "What do these people think of this particular question or subject." Over time, we expect to build an understanding of how to generate and use relevant data, how to frame the legal question in terms of data analysis, etc. You ask, "Don't experts know how to do these things?" You will find out that few do.

We will discuss admission or rejection of expert testimony as a broader topic than just in statistics. In this topical organization, we go as far back as the literature takes us.

Dr. Michelson was appalled, when he started working in law (in the 1960s), at the poor quality of work coming from economists, mathematicians, statisticians, and social scientists whatever their label. Was this field particularly inept? Looking around, he found that the problem of an attorney finding a statistical analyst, that is, finding a good one, was similar to problems many people face every day. He learned this the first time he hired an architect, to whom he specified that the bedroom would be in the back of the house. The architect showed up with a model in which the bedroom was in the front (street side).

In due course, Michelson learned that attorneys are not looking for a good expert, only a confirming one. Again, he found many similar situations elsewhere. Drug testing anyone? Legislation? So-Called Experts is a 15 chapter book about these two issues: Does the person making the selection actually want an expert? If so, how would he know who is one? It is free, because Michelson found the convolutions of publication too difficult, and the services that advertise that they will help too, let's say, inexpert.