This is the way this graph should look.
In trying to understand the relationship between the CMSP, the helmet law and whether helmets have been effective in saving lives directly, neither this study nor the CHP point out that the most significant drop in fatalities occurred between 1987 when the CMSP program began and up to 1992 before the helmet law took effect. According to the above graph, there was a drop from 1.1 to .6 deaths per 1,000 between 1987 and 1992. Then, there was only a drop from .6 to .5 deaths per 1,000 from 1992 to 1995 after the Helmet Law. So, the most significant drop happened BEFORE the helmet law. Regarding this mild but continued decrease after 1992 that this study likes to attribute to the Helmet Law, one cannot assume that helmets were the only cause because you have to consider the helmet law in combination with the existing CMSP.
Here's another way of reading this graph: the fatality rate was declining steadily and fast from the commencement of the CMSP from 1987 to 1992. Interestingly, this decline rate leveled off dramatically at the introduction of the Helmet Law. So what is the CHP bragging about? The death rate was doing better before the Helmet Law!
Further, deaths per 1,000 declined while registrations increased during 1987-1992. After the Helmet Law was introduced however, registrations began to decline and the death rate per 1,000 slowed. This would indicate that the death rate is directly proportional to the declining registration rate. Therefore one could conclude that the Helmet Law has had an insignificant effect on reducing fatalities directly, but that the Helmet Law has had an indirect effect by reducing the number of registrations and the aggregate number of miles ridden. Less bikes ridden less often is the REAL reason why there are less accidents and therefore less deaths. If the Helmet Law were really effective in reducing the death rate directly, one would see a significant drop in the rate. We've seen just the opposite. The death rate has slowed.