Poverty Variables in NHIS

Hello team,

I have been looking at variable related to ratio of income to poverty threshold. I looked at POVERTY and POVERTY2 variables from 2017 NHIS. I see that there are several thousand counts for each level (they add up to more than the total sample). Do the categories overlap? (ex: Less than 1.00 and Under .50, what is this the difference here?)

Thank you.

You are correct that the categories overlap. This is a result of harmonizing poverty measures with the intent to maximize comparability over time but retain year-specific detail. Beginning in 2007, a series of response options were added that are larger intervals where more specific detail wasn’t available. These broad categories (e.g., Less than 1.0) allow for more precision than categorizing the cases as “undefined” or “unknown” in cases where a more exact ratio could not be calculated (e.g., “.50 to .74”). The specific ranges that fall within the broader category will NOT total to the less detailed category. You may choose to combine all values within a certain range to increase the comparability over time; you can do this easily by using only the first digit of the codes, where all codes that begin with a 1 fit into the broad “Less than 1.0” category, all codes that begin with a 2 fall into the “1.0-1.99” category, etc.

Thank you, Kari. I appreciate your time.

So if I wanted to create the following categories:

  • <1.00
  • 1.00-1.99- 2.00-3.99
  • = 4

Would I select the broad categories of:

  • “Less than 1”
  • Sum the categories of “1-1.99” + “2-3.99”
  • Sum the categories of “4.00 and over” + “5.0 and over”

Or should I sum the subcategories that find into my intervals? I am only looking at 2017 data should I won’t need to compare across time.

Thank you again,


Happy to share information, Jorge! You should combine all subcategories rather than using the broad categories. The broad categories would only provide you with a small fraction of the total eligible responses within the range they describe.