# RADA

## Informal comments on the article: "Oakland's Speed Hump Program: Is It Really Working?" Journal of American Public Health

Table 2 is the crux of the study and it is very confusing. First, it is clear that the sample size is tiny. Though there are supposedly 100 child-auto accidents and 200 controls, only 14 child-auto accidents happened within the child's neighborhood and only 6 were on the child's street of residence when the child is within a block of a speed hump.

Here is the confusing part. There were 100 case patients and 200 controls. With 14 neighborhood injuries, that means 14 percent. Six index street injuries should mean 6 percent, but the table says 12 percent. With 200 controls, 46 neighborhood cases equal 23 percent, but 24 index street cases should be 12 percent, not 24 percent. I am not sure what the significance of this is other than they don't seem to be able to calculate simple percentages.

The next two columns are more revealing. The "adjusted OR" column says that, if a child lives on a street without speed humps, it is likely that 47 percent of his/her emergency room visits will be due to auto-child accidents. If the child lives on a street with speed humps, only 40 percent of his/her emergency room visits will be due to auto-ped accidents. This 7 percent difference is the basis of their conclusions that speed humps make streets safer.

Within a 95 percent confidence interval, however, the odds of an injury on a street without speed humps ranges from 24 to 95 percent, while the odds on a street with speed humps is 15 to 106 percent. That is a huge overlap. In fact, it is 100 percent overlap.

That amount of overlap means there is NO STATISTICAL SIGNIFICANCE to the results. In other words, the results are virtually IDENTICAL on streets with speed humps and streets without. I am shocked that they don't reveal this in the abstract, much less the text.

In short, the tiny sample size was inadequate to produce any findings of statistical significance. The writers' inability to calculate simple percentages seems to explain their failure to recognize the statistical insignificance of their results.

* * *

The original article was published in the April 2004 issue of the Journal of American Public Health (AJPH). The above comments listed were submitted to RADA on April 6, 2004.

Return to the Reviews page to link to comments from other reviewers.

© Rada Project, 2004
May 24, 2004 (Version 1)