HowtoLie2“How to Lie with Statistics” by Darrell Huff is one of my favorite books, because statistical analysis is often critical in ADA class action and other kinds of civil rights litigation, and statistics are so often abused in these cases. The difference between the manageable defense of one or two properties and an unbelievably expensive defense involving hundreds or thousands can turn on whether a particular class meets the “numerosity” requirement in Rule 23(a)(1) of the Federal Rules of Civil Procedure. That requires the use, and often mis-use, of statistics.

A good example of the misuse of statistics can be found in the majority opinion in Colorado Cross Disability Coal. v. Abercrombie & Fitch Co., 765 F.3d 1205, 1215 (10th Cir. 2014). The plaintiffs claimed that Hollister stores, which included a raised front porch at the entrance, violated the ADA because the porch was not accessible to those with disabilities. The 10th Circuit affirmed class action certification although it found the porch did not violate the ADA.

The majority decision affirming class certification found that there was sufficient evidence of numerosity because (1) there were almost 250 Hollister stores nationwide and (2) there are 43 million Americans with physical or other disabilities. The error is a common one. The majority assumed that because a particular set was huge (43 million) any subset (disabled persons denied access to a Hollister store) must be equally huge. This is an obvious fallacy. After all, there are 43 million numbers between 1 and 43 million, but the subset of numbers between one and five contains only five members. The fact that one set is big tells you absolutely nothing about the size of any given subset.

The dissenting opinion describes the problem in general terms, but it is worthwhile to look at the problem with specifics. The plaintiffs themselves admitted that the number of individuals confined to a wheelchair in the U.S. is only around 3.6 million individuals.  It is still a big number, but less than 1/10th of the number used by the majority. It is also clear that 3.6 million is over-inclusive. Hollister sell clothes aimed at a market of 14 to 18 year olds, and those individuals plus their parents are the only ones likely to be affected by an accessibility problem. According to a Cornell University report available at the number of individuals with “ambulatory” disabilities between 16 and 20 is only around 171,000. The specific 14 to 18 year group targeted by Hollister is not discussed, but it seems likely that this number is fairly close to the population of teens in wheelchairs to whom Hollister markets.

The population of teens that actually go to Hollister stores is without a doubt much smaller. The almost 250 stores are typically in urban malls, which excludes disabled teens in rural areas. Each store has its own geographic market area that excludes some urban areas. Los Angeles, for example, has a dozen Hollister stores, but none are located in South Central. Hollister also has a specific appeal to certain income groups, and teens outside those groups, like those living in South Central, are probably not potential customers.   The number that matters, which is not found anywhere in the discussion of numerosity, is the number of teens who actually shop at each Hollister store times the .8 percent of teens with ambulatory disabilities.

It is possible that after a careful analysis the likely number of customers with ambulatory disabilities would satisfy Rule 23’s numerosity requirement for all Hollister stores. It is certain, however, that a simple appeal to national or even regional statistics on disabilities cannot provide a reliable guide to the number of individuals affected by a particular accessibility problem at a particular store. Nonetheless, that is the usual approach to class action certification in ADA cases.  Courts rarely, if ever, do the kind of careful market analysis that a business would do before opening a store. When dealing with a handful of stores that sell highly specialized merchandise (high end jewelry, for example), or a single apartment complex in an FHA case, careful market analysis might well reveal that the class of affected consumers is not sufficiently numerous to meet Rule 23’s requirements. In my own experience with class action settlements that involved small cash payments to class members there might be no more than one or two claims made over a period of five or six years despite notices posted on the property and given to every tenant.

Although the cases are not encouraging, there is a real defense opportunity in cases where the plaintiff seeks to certify a class action. Statistics is a science, and much of what courts accept from plaintiffs in ADA and FHA class actions amounts to junk science. Defendants willing to invest in real statistical analysis and to attack plaintiff statistics based on their scientific defects have an opportunity to avoid certification, and turn an unmanageable mess into a defensible lawsuit.