GEICO has been taking heat in New Jersey because of its use of education and occupation in determining insurance rates. Of course, GEICO is not alone in using these factors, but GEICO was the focus of the Star Ledger story that shed light on this matter, and so the company has received the brunt of the criticism.

As reported in last weeks Insurance Advocate e-newsletter, GEICO uses education and job status as rating factors and generally charges lower rates for persons with college degree and low risk jobs like an architect engineer, lawyer or teacher.

Conversely persons with just a high school diploma or working in unskilled and semi skilled jobs or long haul drivers and route men are charged higher rates.

As we go to press, Assemblyman Neil Cohen (D-Union County) has introduced legislation to forbid the use of occupation and education in determining rates.

These arguments appear similar to the debate that was in full swing at about this time last year: the use of insurance scores " based on credit scores " to determine rates. However, I believe that this instance is even more offensive.

With insurance scores, insurers were able to argue that credit scores are not discriminatory because they are neutral to factors such as race, age, and economic well-being. The same cannot be said in this case. Education and occupation are, in fact, related to economic well-being. That determining rates based on these factors does not discriminate against the poor while favoring the wealthy would be a difficult argument to make.

In January of last year, J. Robert Hunter, of CFA, was quoted in this column as saying that insurers use of credit scoring to determine rates was wrong because insurers had no thesis with regards to credit scoring, just a correlation. He declared that insurers should form a thesis, and then try to find correlations, rather than just using a computer to find correlations.

His argument held water on that issue, as it does in this instance. How can increases based on these factors be justified to consumers when the response they get is simply, our computer shows a correlation. The saying that applies here, I believe, is correlation does not imply causation.