In effect the CPSC report only looks at the emergency room visits from 2 of the over 200 amusement parks in the United States, since only 2 of the emergency rooms in their national sample are near major amusement parks. I fact, the CPSC's increase was entirely due to an increase in the visits to one emergency room. The CPSC made no effort to determine why this increase occurred. They just claim that it indicates a national trend.
This is gross abuse of statistics. A sample size of 2 is not sufficient to be a statistical sample of over 200 parks. Their 100 sample emergency rooms are adequate for sampling accidents due to toys or barbeque grills where their are millions of points of use. It doesn't work for examining safety when the exposure occurs at only a few hundred locations.
Their are many reasons why the number of emergency visits may increase from 1 park.
1)The park has a real safety problem.
2)The park has had a major increase in attendance.
3)A 2nd park has opened in the area.
4)The park has changed its policy and now sends minor injuries previously treated on site to the emergency room.
5)The emergency room has started coding all or too many injuries from the park as ride related. This might include falls while walking, heat exhaustion, numerous types of bobos, and heart attacks as ride injuries.
The report only shows an increase in the number of emergency room visits, not an increase in injuries. As mentioned above, the increase may be due to many other reasons. Correlations alone are not sufficient to show causality, there must also be some information on what is actually happening. It is a classic example in statistics that there is a strong correlation in Europe between the number of stork's nests in an area and the birth rate. Ingnore the actual facts, and one could conclude that storks must bring babies based on the correlation alone.
When you do a statistical study and one data point is so far out from the rest that it reverses the result of your study, the only responsible thing to do is to determine why this point is so far out and if it represents something important. Also, no decent statistician would attempt to use a sample size of 2 for just about anything. The CPSC needs to have their statistics reviewed by impartial statisticians before releasing reports.
Ride safe,
Jim Fisher