In support of many judicial and regulatory proceedings, as well as in the execution of government and corporate assignments, Moshman Associates has conducted a large variety of statistical analyses. The range of the underlying applications is broad. Some examples include:

• Health services
• Consumer product performance
• Employee time studies
• Accuracy of cause of death coding on death certificates
• Insurance claims
• Alaskan Pipeline product differentials
• Share of market estimation
• Voter behavior
• Projections of demographic characteristics
• Curriculum content and variation.

When contemplating a statistical study, we first ascertain the underlying problem and determine what implications may flow from the spectrum of possible results. At that point, we consider what data are required to answer the pertinent questions. Sources of existing data are then examined for their utility or, if no data are available, ways of collecting and compiling the necessary data are considered. The possible uses of surrogate data are investigated if cost implications should so dictate.

The choice of analytical techniques we use depends on the data characteristics and, if appropriate, the criteria to be used in the analysis. Many classical statistical methods are valid if the data follow a normal distribution, i.e., are characterized by the familiar bell-shaped curve, or have some other standard statistical distribution. If the assumptions underlying a specific method are not valid or nearly so, we may exploit one of the non-parametric or distribution-free techniques which may not be as sensitive but which have greater validity in their applicability. Alternately, we will devise or adapt a technique unique to the problem and data.

We usually prepare a report of findings and interpretation following the analysis and a discussion with our client. Our report may be written, oral or both, and is prepared without statistical jargon so that it is intelligible to the intended audience while fully disclosing the theoretical basis for the conclusions therein.  Laypersons, including judges and juries, should not be expected to be familiar with statistical concepts, let alone understand the subjectivity implicit in the choice of the criteria for statistical significance.