differential vs. difference (nouns, outwith math)

The two have significantly different expectations with regard to the structure of what is being compared, and differential is much more restrictive.

Taking the insurance actuarial table as an example — suppose the insurance rate is determined by looking a number up in a table. There are three indices — age, sex, and geographical location. So you can talk about the rate differential between genders because the gender subtables both have a parallel data structure. You can make the comparison element by element.

Using differential implies that you know the data structures are the same, and that you are restricting your comparisons to an element by element comparison across the common indices.

It does not make sense to use differential when making a "black box" comparison. For instance, unless you know the rate table has an index for marital status, you can't talk about the differentials between married and unmarried rates. These can only be described as differences, and would normally be some statistical values such as averages over large samples of customers.

Looking at this sentence —

However, the Directive also stated differential pricing could be maintained where statistical evidence supported such an approach.

This works only if the tables for men an women share the exact same indices. If the table for men had age and location as indices, and the table for women had marital status and educational level as indices, the statement wouldn't make sense. You would talk of rate differences, not rate differentials.

So I understand the above sentence to mean that using different indices to actuarialize the rates of men and women would be considered discriminatory, using the same indices, but assigning different rates to the elements would be conditionally allowed.