How One Hospital Skewed The CDC's Gun Injury Estimate

[hack89]

For years, the estimates of nonfatal gunshot injuries published by the Centers for Disease Control and Prevention have grown increasingly unreliable — in 2017, they were more suspect than ever. But researchers have continued to cite the numbers as authoritative. Last year, a CDC spokesperson defended the data, saying the agency’s experts were “confident that the sampling and estimation methods are appropriate.”

Now the CDC is taking measures to curtail the spread of its most unreliable estimates. The 2016 and 2017 gun injury figures have been hidden on the agency’s public data portal, with a footnote stating “Injury estimate is not shown because it is unstable.” The CDC will hide unstable estimates for all injury types within the next six months, according to a spokesperson. Also, the option to include statistical information about how reliable or unreliable the estimates are is now enabled by default. Until recently, it was disabled by default.

The changes follow reporting by FiveThirtyEight and The Trace, a nonprofit news organization covering gun violence in America,1 that highlighted the unreliable estimates.

The CDC’s gun injury estimate was vulnerable to unreliability in part because of how few hospitals are surveyed in the data set that feeds it. When one hospital is replaced by another in the database the CDC uses, the changeover can cause the injury estimate to swing drastically. The CDC now says it is exploring the feasibility of collecting data from more hospitals, which would improve the estimate’s reliability.

https://fivethirtyeight.com/features/how-one-hospital-skewed-the-cdcs-gun-injury-estimate/

Given the emotions surrounding the subject, accurate information accepted by all sides is critical.

[jmg257]

1. With like 100,000 injuries a year, how much does it matter? How far are they off? nt

[hack89]

2. In 2017 they gave a range of 31,000 to 236,000

According to the CDC’s most recent figures, somewhere between 31,000 and 236,000 people were injured by guns in 2017. That range, which represents the confidence interval — the high and low ends of a range of estimates that probably contains the real number, whatever that number is — is almost four times wider than the one given in the agency’s 2001 estimate.

https://fivethirtyeight.com/features/this-years-cdc-gun-injury-data-is-even-less-reliable-than-last-years/

That kind of accuracy makes the stats useless - a delta of 200,000 tells you nothing meaningful.

[NickB79]

3. Holy standard deviations, Batman!

My biostatistics professor would have had a seizure if I presented him with numbers like me that.

[jmg257]

4. Wow - yep - very big margin for error. nt