The original claim would be hyperbolic, except that someone killed in a car crash doesn't increase the risk to other people on the road, but an Ebola infection certainly does. In the mathematics of exponential increase, a small initial amplification factor can produce very impressive results in a short time. Consider the hypothesis that just 1% of Ebola sufferers infect other people. Using this well-established mathematical method:

y' = y (1+p/100)^t

y = infected population % at time zero.

y' = infected population % at time t.

t = time in consistent units

p = percentage increase for one unit of time

So, if it takes a week for the infected population to increase by 1%, then:

http://i.imgur.com/TDPd7ug.png

IOW if after a week the infected population has increased by 1%, after a year (52 weeks) the infected population has increased by 67%. After 2 years (104 weeks) the infected population has increased to 281% of the original. And so forth.

This isn't true for car crashes, and the above is what is meant by an "exponential increase".

If the ghosts of people who died in the road began to spook drivers and cause more accidents, you would see a rapid spike of mortality in the short term, followed maybe by change of behavior where ~90% of traffic is composed of professional bus/truck drivers that are hardened and not easily affected by ghosts.