Documentation is incorrect for the NHIS CIGMENTHOL variable

Hello MPC friends,

I’m currently working with an IHIS export of data related to smoking and I’ve discovered an error in the documentation for the CIGMENTHOL variable. According to the current docs a code of 1 corresponds to those who smoke plain cigarettes and a code of 2 corresponds to those who smoke menthols. However, using this mapping I get eactly backward results from what I expect. I find that in 2010 69.9% of whites smoked menthols and 17.8% of blacks. Obviously this is incorrect. I reviewed the the “survey text” section of the docs and this coding is, in fact, backwards. In the original question a code of 1 indicated menthol smokers and a code of 2 those that smoke plain cigarettes.



Actually, it seems like there is a second problem here. According to the Universe tab, the 2010 and 2015 samples both cover “Sample adults 18+ who are current smokers,” however, according to the survey text in 2015 the question covered only *former* smokers. This makes 2010 and 2015 non-comparable. I’ve confirmed this does appear to be the case in the 2015 sample — when restricting to current smokers the CIGMENTHOL variable is 0 for all sampled adults.

Was the wrong underlying variable mapped for 2015 or did the CDC really change the universe of the question to make it non-comparable?

Yup this is definately an error. The codes in 2005, 2010, and 2015 are simply flipped around. In the 2015 sample, however, the issue is not a universe change. Rather the wrong source variable is feeding into the data. This is causing the issues you observe. We will have both of these issues fixed and deployed to the online data sometime next week.

We like to reward users who identify errors such as this with an IPUMS coffee mug. To claim yours, just email with your mailing address and I’ll have one sent out to you.

Excellent! Glad to hear it’s an easy problem to sort out. I’ll send along my address. :slight_smile: