Error: only one level in treatment variable

[Yuanyuan Zhuang]

Dear statelister,

I run into a problem while running the following code:
teffects nnmatch (_mbdty_diarrhea_new c_age m_age_atbirth m_married m_ed_yr_cat c_sex m_place_delivery_cat) (bf_dummy6), nneighbor(3) biasadj(c_age m_age_atbirth) ematch(m_married m_ed_yr_cat c_sex m_place_delivery_cat)

the error message is:there is only one level in treatment variable bf_dummy6; this is not allowed
r(459);

however the variable bf_dummy6 is a dummy variable like following:
bf_dummy6 | Freq. Percent Cum.
------------+-----------------------------------
0 | 217,465 18.13 18.13
1 | 981,928 81.87 100.00
------------+-----------------------------------
Total | 1,199,393 100.00

can anyone help and tell me what is wrong in the code? Thanks in advance

Best
Yoyo

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(bf_dummy6 mbdty_diarrhea_new) long c_age byte c_sex float(m_age_atbirth m_ed_yr_cat)
1 0 3 1 17 0
1 0 4 1 21 0
1 0 1 1 25 0
1 0 1 2 25 0
0 0 0 2 35 0
1 1 1 2 33 0
1 0 0 2 22 0
0 0 0 1 18 .
1 1 1 2 26 0
1 0 3 2 19 0
1 1 1 2 37 0
1 0 4 1 41 0
1 1 1 2 20 0
1 0 2 1 32 .
1 0 0 2 29 0
1 0 3 2 35 0
1 0 1 1 35 .
1 1 1 1 42 0
1 0 1 1 25 0
1 0 4 2 19 0
1 0 3 2 17 0
1 0 3 1 27 0
0 1 0 2 29 0
1 0 3 2 25 0
0 1 0 1 32 0
1 0 0 1 29 0
1 1 2 2 33 .
0 1 0 1 22 0
1 1 3 1 22 0
1 0 3 2 33 0
1 0 3 2 30 0
0 0 0 1 23 0
1 1 3 1 21 0
1 0 1 2 32 0
1 0 2 2 24 0
1 0 4 1 23 0
1 0 0 2 25 0
1 0 2 1 18 0
1 1 3 2 22 0
1 1 2 2 27 0
1 1 0 2 25 .
1 0 1 1 30 0
1 0 3 2 22 0
1 0 3 1 25 0
1 1 1 1 22 0
1 1 2 1 16 0
1 0 1 2 31 0
1 0 3 2 39 0
1 0 2 2 21 0
1 1 2 2 25 0
1 0 2 1 35 0
1 0 1 1 28 0
1 0 2 2 38 0
1 0 1 1 34 0
1 0 3 1 33 0
1 0 2 1 20 0
1 0 2 1 29 0
0 0 0 1 27 0
1 1 1 1 29 0
0 0 0 2 22 0
1 1 1 2 19 .
1 1 1 2 32 0
0 0 0 2 20 .
1 0 3 2 26 0
1 0 1 2 21 .
1 0 0 1 25 0
1 0 1 2 36 0
0 0 0 1 24 0
1 0 1 2 30 0
1 0 2 2 28 0
1 0 1 2 22 0
1 0 4 2 26 0
0 0 0 1 26 0
1 0 0 2 35 0
1 0 3 1 29 0
1 0 1 1 30 0
1 1 0 1 34 0
1 0 2 1 19 0
1 1 1 1 18 0
1 0 1 2 42 0
0 0 0 2 30 0
1 0 4 2 23 0
1 0 0 1 36 0
1 0 1 2 22 .
1 1 0 2 18 0
1 0 2 2 20 0
1 0 3 2 33 0
1 0 1 2 25 0
1 0 1 1 24 0
1 1 1 1 18 0
1 0 4 1 18 0
0 0 0 1 25 0
1 0 3 2 31 0
0 0 0 1 27 0
0 1 0 1 30 0
1 0 3 1 40 0
1 0 3 1 36 0
1 0 4 2 26 0
1 1 0 1 22 0
1 0 1 2 27 0
end
label values mbdty_diarrhea_new mbdty_diarrhea_new
label def mbdty_diarrhea_new 0 "No", modify
label def mbdty_diarrhea_new 1 "Yes", modify
label values c_sex c_sex
label def c_sex 1 "Male", modify
label def c_sex 2 "Female", modify
label values m_ed_yr_cat m_ed_yr_cat
label def m_ed_yr_cat 0 "No to Primary education", modify

[Clyde Schechter]

I don't see anything wrong with the code. The problem is probably with your data.

-teffects-, like all estimation commands, excludes from the estimation sample any observation where any of the variables in the command is missing. In your example data the variable m_ed_yr_cat has some missing values. So all of those observations will be excluded from the -teffects- estimation. There are other variables mentioned in your -teffects- command that do not appear in your example, but my guess is that there are some missing values for some of those as well. I think what is happening is that after you omit all the observations that have any missing values on the mentioned variables, the remaining observations all have the same value of bf_dummy6. You can test that easily enough:

egen mcount = rowmiss((_mbdty_diarrhea_new c_age m_age_atbirth m_married m_ed_yr_cat c_sex m_place_delivery_cat bf_dummy6)
tab bf_dummy6 is mcount == 0

[Yuanyuan Zhuang]

Hi Clyde(I was hoping that you would help), thanks for your answer and you are totally right about this. I got only bf_dummy6=0 left in the observations. Best Yoyo