Using statistical evidence in employment cases (Part II)

1) if there was no discrimination against the protected group members, what would be the probability (or sheer random chance) of observing the employment related disparity by random chance alone?

This question is typically used when assessing the statistical importance of an overall difference in selection rates or salary by protected group status. For instance, consider an example where African-American employees earned $10,000 less than similarly situated White employees. In this example, ,the statistical expert will use generally accepted methodologies to calculate the probability that sheer random chance generated the observed racial salary differential between the similarly situated employees.

The probability that is calculated from the expert’s statistical test is commonly referred to as a chance probability.

Consider the situation where the statistical test in the above example determined that there was a chance probability of 1% of observing the racial salary differential between the similarly situated employees. A chance probability in this setting would suggest that if there was NO discrimination against African-American employees, then we would only expect to see the observed racial salary differentials in 1 out of every 100 instances. This finding could be suggestive of discrimination against African-American employees.

A chance probability that is small and equal to or less than a pre-specified threshold is referred to as statistically significant.

Conversely, if the calculated chance probability of observing the racial salary differential was 50%, then we would expect to see the observed employment outcome in 1 out of every 2 instances. Since there is a relatively large chance of observing the employment outcome by random chance, this finding is not suggestive of discrimination against African-American employees.

A chance probability that is large and greater than a pre-specified threshold is referred to as statistically insignificant.

In many court room settings, an employment disparity finding that has less than a 1% chance of occurring by sheer random is viewed as statistically significant. The actual methodology used to calculate the chance probability depends on the type of employment question being addressed by the statistical expert.