Human Assets

How Accurate is the 4/5ths Rule for Determining Adverse Impact?

Adverse impact is present when selection decisions are made that have an unjustified and disproportionately negative effect on a particular group.  The most widely used approach to assessing adverse impact is the 4/5ths rule, institutionalised in the late 1970s by the US Equal Employment Opportunity Commission. At its simplest, this method compares the group that is selected less (lower selection ratio) to the group that is selected  more (higher selection ratio). For example, four out of ten (0.4) women selected compared to six out of ten (0.6) men.  Adverse impact is indicated if the selection ratio for one group is less than 80% (4/5ths) of the other group’s selection ratio.  So for the above example, since 0.4 (the selection ratio for women) is less that 80% of 0.6 (the selection ratio for men), it would be concluded that the selection process adversely impacts women.

This approach is used in a variety of contexts, almost without question, as the standard measure for determining whether or not there is adverse impact.  For example, most academic literature uses the rule to test differences in hiring rates, organisations use the rule to self-monitor their hiring programmes, and courts may also use the rule in litigation processes.

However, a recent article by three US researchers, Philip Roth, Philip Bobko and Fred Switzer, questions the use of the 4/5ths rule and considers the role of statistical tests of significance as a more accurate indicator of adverse impact.
 
False-Positive Errors
 
In their study, Roth and his colleagues looked at the probability of the 4/5ths rule erroneously indicating adverse impact in selection processes.  They applied the 4/5ths rule to a selection system and tested its accuracy by applying a statistical test to the same data.  They found that there was a greater chance of the 4/5ths rule indicating adverse impact when none existed (a false-positive error) when the number of applicants was low, e.g. 200 applicants rather than 2000, and also when the selection ratio was low, e.g. 10% of the applicants are selected rather than 30%.

On reflection, it seems logical that these errors would exist when applying such a simple rule of thumb.  For example, if a selection process recruits two men and one woman out of a group of ten men and ten women, the 4/5ths rule would indicate adverse impact whereas a statistical test would take into account that only three people were recruited in total.

It is clear, therefore, that the use of a statistical test would be a much more rigorous method of measuring adverse impact compared to the 4/5ths rule.  However, one reason for developing such a simple technique as the 4/5ths rule was to avoid the complexities of the use and interpretation of statistical tests, which many people may struggle to understand (both those who are enforcing the employment law and those who are affected by such decisions).

Implications

These findings challenge the use of the 4/5ths rule in situations when the applicant size is small and the selection ratio is low.  Organisations who use the 4/5ths rule under these conditions may conclude that adverse impact is present when it is in fact not.  This could cause organisations to consider redesigning a selection process unnecessarily that is currently working.  Furthermore, the use of the 4/5ths rule in tribunal cases could lead to unnecessary financial loss as well as create unwarranted damage to the corporate image.

This presents several problems for the many organisations whose selection process only includes a small number of applicants and for those who may not have the knowledge or ability to conduct statistical tests as an alternative to the 4/5ths rule.

How can Human Assets help?

At Human Assets, we have the expertise to design and deliver a tailored selection process to meet your needs fully and ensure fairness and equality throughout.  Furthermore, we can conduct an evaluation of the process in a statistically rigorous way to verify its validity and accurately assess for adverse impact.  This avoids any errors from calculations based on the simple 4/5ths rule.

If you would like to find out more please contact our consultants on +44 (0)20 7434 2122 or at enquiries@humanassets.co.uk
 
Reference
 
Roth, P.L., Bobko, P. & Switzer III, F.S. (2006).  Modeling the Behavior of the 4/5ths Rule for Determining Adverse Impact: Reasons for Caution.  Journal of Applied Psychology, 91, 507 - 522.

Newsletter :February 2007