Testing the accuracy of my device has been quite an interesting ride, and it has lead me to one conclusion: Breathalyzers are far from an ideal way of testing Blood Alcohol Content (BAC).

This must be why in real serious cases, Police Departments use blood or urine tests as they are directly measuring the concentration of alcohol in your blood.

Breathalyzers however, are a one-off test. Quite simply, they are measuring the amount of alcohol gas in your mouth. Thats an important distinction. There is the obvious factor that could throw off the test, that you could have liquid alcohol in your mouth. Some of this will be in gas form, and thus what you will be measuring is the concentration of that liquid. Which has absolutely ZERO to do with your BAC.

It's obvious, but let me really drive this point home: you can take a mouth full of Vodka, spit it out, even rinse your mouth, and still blow a deadly level on a Breathalyzer, mean while your BAC is still zero.

While this was apparent to me, I wasn't sure at first how much I would have to control for it. For instance, if I take a drink of beer immediately before testing, I knew that would skew the results, but how long between drinking and testing was long enough? One minute, five minutes, thirty minutes? In my first set of tests, this proved to be an overwhelmingly important factor that all but invalidated my results. However they were useful in that they taught me how to test in the future.

My second round of testing, I controlled for this by having everyone wait at least a minute between drinking and testing, and having everyone thoroughly rinse out their mouth with water immediately before testing. This clamped our values to much more linear and sensible readings as we went on. But we still had outliers throughout the night. And I discovered that:

1. Burps produce readings similar to drinking right before testing.
2. When you're drunk, you burp a lot without even thinking about it.

This was much more difficult to control for as the night went on... (also of note, data taking became increasing difficult toward the end of the night ;) )

The other thing I found, which may or may not seem intutive, is that Breathalyzers essentially need a male/female setting. Example: when my girlfriend was blowing an 18 (don't worry what that means, I'm just using it as a point of refrence), we calculated her BAC to be ~0.123%. (Also imperically verified her as: pretty drunk). Mean while, I blew a 19, and calculated my BAC to be ~0.034%. Even after controling for all of the skewing factors, her readings simply needed to be interpreted differently than mine or the other male subjects.

After discovering all of this I did some searching and found Breathalyzers are actually contested based on these exact grounds. They are bias against females, and there can be strongly mitigating factors during the test.

Oh well, I can still get rough approximations, and it's still been loads of fun

[cross-posted from: darkrockstudios.com]