The importance of equal opportunity and equal treatment has increasingly become a theme of discussions in the mortgage industry in recent years. In fact, heightened awareness of these issues was the impetus behind fair- lending initiatives such as the Department of Housing and Urban Development's best-practices agreements.
Many industry leaders have responded with aggressive programs to ensure that the principles and practices of fair-lending are administered to end disparate treatment and to promote equality among consumers.
In the mortgage industry, matched-pair testing is a good tool for obtaining a snapshot of potential disparate treatment of similarly situated consumers, when one of the matched pair is a member of a protected class. This testing methodology is often confused with "mystery shopping." However, there are very distinct differences. Mystery shopping is often used to determine quality of service or sales acumen, but matched-pair testing is primarily designed to identify discriminatory treatment.
Much controversy has surrounded the appropriate use and design of matched-pair testing. Social scientists use matched-pair testing to determine discrimination in settings such as insurance, apartment rentals, and mortgage lending.
Some say matched-pair testing does not require standardization of sampling, pre-test design, or statistical validation. But without standardization these tests identify only potential incidents of poor customer service. To truly test for disparate treatment, matched-pair testing must be standardized.
As enforcement agencies plan to increase funding for matched-pair testing, it is critical that they ensure the validity and standardization of the testing methodologies. It is also critical that organizations selected to use this type of measurement conduct fair and unbiased testing.
The standard for proper application of matched-pair testing for differential treatment discrimination was established by HUD-funded testing of real estate professionals in 1977. Since then, changes have been made in research design, but some of the basic tenets of statistical methodology have not been consistently employed by the various organizations engaged in conducting matched-pair testing.
Three of these standard and widely accepted elements of sampling theory and test design include:
Establishing the statistical significance of any observed differentials in treatment.
It is essential that matched-pair testing use statistical methodologies to distinguish between systematic and random differences in treatment.
A statistically valid study design must precede actual testing.
Matched-pair testing must identify, in advance, a specific type of differential-treatment discrimination. The questionnaires, profiles, and analysis of the results should be designed to obtain information about one specific type of differential-treatment discrimination.
Analysis of several or all aspects of the interaction can lead to invalid conclusions.
Matched pairs must be matched in all aspects except minority status.
This minimizes any effect of confounding variables and better isolates the target variable.