0.9% of CPS tax units have age_head < 18

[MaxGhenis]

Including some as young as zero (notebook):

(cps.groupby('age_head').s006.sum().iloc[:18] / 100).astype(int)

age_head
0      34752
1      39530
2      35297
3      29054
4      25753
5      34461
6      16484
7      28085
8      24733
9      22542
10     20762
11     23471
12     23411
13     34202
14     31689
15    209089
16    319426
17    527100
Name: s006, dtype: int64

cc @prrathi

[andersonfrailey]

Completely speculating here, some of these may be dependent filers. Some could be issues with the data where ages are input wrong. I'd want to look into some of the other variables associated with those units to make a more informed guess. Thanks for pointing this out!