Why I am Getting

[Roman Johnson]

I am using Scott Long's user created command mtable which is from his Spost13 package.

I have generated a variable that I am trying to calculate the margins for a variable I have created (HispanicUnhealthyLifestyle) against my outcome variable (clinicaldepression) but it says there are no observations.

gen HispanicUnhealthyLifestyle = race == 2 & numcigs_r == 30 & exercise == 0

sum clinicaldepression if HispanicUnhealthyLifestyle

Variable | Obs Mean Std. Dev. Min Max
-------------+---------------------------------------------------------
clinicalde~n | 0



qui mtable if HispanicUnhealthyLifestyle, rowname(HispanicUnhealthyLifestyle)
> ///
> atmeans ci below
no observations
r(2000);


I am providing a data example below.



dataex HispanicUnhealthyLifestyle

----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. To install: ssc install dataex
clear
input float HispanicUnhealthyLifestyle
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
end

------------------ copy up to and including the previous line ------------------

Listed 100 out of 420 observations
Use the count() option to list more

My question is the following: How do I locate at which values there would be observations for my variable Hispanic?

[Carole J. Wilson]

I would start by using -summarize- to analyze subsets of the data:

sum numcigs_r if race==2 & exercise==0 
sum numcigs_r if race==2 & exercise==0 , detail

[Carlo Lazzaro]

Roman:
you may want to try something along the following lines:

sysuse auto.dta
bysort foreign: list make if foreign==1 & _n==1

[Roman Johnson]

Thank you Carole and Carlo for responding.

sum numcigs_r if race==2 & exercise==0

Variable | Obs Mean Std. Dev. Min Max
-------------+---------------------------------------------------------
numcigs_r | 5 2.2 1.788854 1 5

. sum numcigs_r if race==2 & exercise==0 , detail

numcigs_r
-------------------------------------------------------------
Percentiles Smallest
1% 1 1
5% 1 1
10% 1 1 Obs 5
25% 1 3 Sum of Wgt. 5

50% 1 Mean 2.2
Largest Std. Dev. 1.788854
75% 3 1
90% 5 1 Variance 3.2
95% 5 3 Skewness .84375
99% 5 5 Kurtosis 2.078125


It appears that there are 5 observations at the specifications I am interested. Why would it then report 0 observations against the outcome variable (clinicialdepression) in post 1?

[Carole J. Wilson]

I'm seeing that the maximum value for numcigs_r==5. You specified numcigs_r==30 in #1. There are no observations where numcigs_r==30.

[Roman Johnson]

Hi, Carole:


. sum numcigs_r if race==1 & exercise==1

Variable | Obs Mean Std. Dev. Min Max
-------------+---------------------------------------------------------
numcigs_r | 14 3.928571 5.210408 0 21

. sum numcigs_r if race==1 & exercise==1 , detail

numcigs_r
-------------------------------------------------------------
Percentiles Smallest
1% 0 0
5% 0 1
10% 1 1 Obs 14
25% 1 1 Sum of Wgt. 14

50% 2 Mean 3.928571
Largest Std. Dev. 5.210408
75% 5 5
90% 5 5 Variance 27.14835
95% 21 5 Skewness 2.717046
99% 21 21 Kurtosis 9.604744

I want to know which value I need for my race == 1 value (black) when exercise == 1. I put 4 for numcig_r but that returned no observations.



. sum numcigs_r if race==1 & exercise==1

Variable | Obs Mean Std. Dev. Min Max
-------------+---------------------------------------------------------
numcigs_r | 14 3.928571 5.210408 0 21

sum clinicaldepression if BlackModerateLifestyle

Variable | Obs Mean Std. Dev. Min Max
-------------+---------------------------------------------------------
clinicalde~n | 0

. mtable if BlackModerateLifestyle, rowname(BlackModerateLifestyle) ///
> atmeans ci below
no observations
r(2000);

[Carole J. Wilson]

Apparently there are no obs for those specific values. Since I don't know the goal for your study, I'll just suggest you use the median or 25th/75th percentile values since you know those exist. Alternatively, you can list all available values in the data for those groups:

tab numcigs_r if race==1 & exercise==1, nolab

[Roman Johnson]

Thank you!