Silicon Valley Musings

During a presentation on suicide prevention that I attended a part of my teaching credential program, some Santa Clara County suicide statistics were presented. Here are my comments on the statistics.

————————-
In the suicide statistics presented, suicide rates are calculated using the formula:

Number of Suicides / Size of Population * 100,000

This formula and its use of an absolute normalizing value (e.g. 100,000) has the unfortunate psychological effect of making suicides in small populations look worse, especially if the consumer of the information is unaware of the underlying population sizes.

For example, look at the table of fictitious suicide statistics using the rate formula provided above.

Fictitious Suicide Rates for Counties A, B, and C for 2000 and 2001

I would suspect most people would say County A had many more suicides than B or C and that given the change from 10 to 30, it also had a much higher rate of change.

What the table doesn’t show is that County A’s population is 10,000, County B’s is 100,000, and County C’s is 1,000,000.

So in fact all three counties had only one suicide each in 2000 and
only three suicides each in 2001. All three counties had the same 300%
increase in suicides from 2000 to 2001.

Now I would agree that three suicides in a population of 10,000 is obviously a much higher per capita rate than three suicides in 1,000,000. But the example statistics seem to suggest that a person in County A is one hundred times more likely to kill themselves than a
person in County C. Practically speaking, I wouldn’t think this is the
case.

For example, if in 1999 County C was split in two and half was merged
with County A and half with B (and all of the former County C’s suicides now occur the bigger County B). The statistics would look like:

Same number of suicides, same overall population, and same 300%
increase from 2000 to 2001. The only thing that changed are the county boundaries. I would suspect, however, most people would say these numbers look much better than the previous statistics.