Variable construction help

[Muhammad Ramzan]

Hi I am working on a survey data having nine variables.



I have to construct these two variables.


1: avg. of nonmissing. We average over the non-missing individual measures. We then convert this average into a standard normal variable.
2: avg., no missing. We average over the individual measures. We keep only individuals who gave valid answers to all nine survey questions, and so have no missing values. We then convert this average into a standard normal variable.

I am confused what the difference between nonmissing and no missing is.

* Example generated by -dataex-. For more info, type help dataex
clear
input long id double(var_30 var1_30 var_feel12 var_inter12 var_ener12 var_feel2w var_most var_app var_slow)
24100002 1 1 5 5 5 5 5 5 5
24100003 1 1 5 5 5 5 5 5 5
24100005 2 2 5 1 5 5 5 5 5
24100007 3 3 5 5 5 5 5 5 5
24100008 1 1 5 5 1 1 1 1 5
24100009 1 2 5 5 5 5 5 5 5
24100012 1 2 5 5 5 . . . .
24100014 . . . . . . . . .
24100015 3 3 5 5 5 5 5 5 5
24100018 . . . . . . . . .
24100020 1 1 5 5 5 . . . .
24100021 3 3 1 1 1 5 1 1 1
24100022 2 2 5 5 5 5 5 5 5
24100023 3 2 5 5 5 . . . .
24100024 3 3 5 5 5 5 5 5 5
24100026 1 1 5 5 5 5 5 5 1
24100027 4 3 1 1 5 5 5 5 5
24100028 2 3 5 5 5 5 5 5 5
24100029 1 2 5 5 5 5 5 5 5
24100030 1 1 5 5 5 5 5 5 5
24100031 3 4 5 5 5 5 5 5 5
24100032 1 2 5 5 5 . . . .
24100033 1 2 5 5 5 5 5 5 5
24100036 1 1 5 5 5 5 5 5 5
24100037 4 3 5 5 5 5 5 5 5
24100038 1 1 5 5 5 5 5 5 5
24100039 . . . . . . . . .
24100040 . . . . . . . . .
24100042 3 1 1 1 5 5 1 1 5
24100045 1 1 5 5 5 5 5 5 5
24100046 1 1 5 5 5 5 5 1 5
24100047 1 1 1 1 5 5 5 5 5
24100049 2 4 1 5 5 5 5 1 5
24100052 1 1 5 5 5 5 5 5 5
24100053 . . . . . . . . .
24100054 . . . . . . . . .
24100056 . . . . . . . . .
24100058 2 2 1 5 5 5 5 1 5
24100059 3 4 5 5 1 1 1 5 5
24100060 . . . . . . . . .
24100063 3 3 1 5 5 . . . .
24100065 4 4 1 1 5 5 1 5 5
24100067 4 4 1 1 1 1 1 1 5
24100070 3 3 1 1 5 5 5 1 5
24100072 1 1 5 5 5 5 5 5 5
24100081 . . . . . . . . .
24100083 3 2 1 1 1 1 1 1 1
24100084 1 1 5 5 5 5 5 5 5
24100085 1 3 5 5 5 . . . .
24100087 2 2 5 5 5 5 5 5 5
24100089 1 1 5 5 5 5 5 5 5
24100092 1 1 5 5 5 5 5 5 5
24100093 1 1 5 5 5 5 5 5 5
24100097 1 1 5 5 5 5 5 5 5
24100099 3 4 1 1 1 5 5 1 1
24100100 2 2 5 5 5 5 5 5 5
24100103 1 1 5 5 5 . . . .
24100105 1 3 1 5 1 5 1 1 1
24100109 1 1 5 5 5 . . . .
24100110 1 1 5 5 5 . . . .
24100113 1 1 5 5 5 . . . .
24100116 1 1 5 5 5 . . . .
24100117 1 1 5 5 5 . . . .
24100118 1 1 5 5 5 . . . .
24100119 . . . . . . . . .
24100120 3 2 5 5 5 . . . .
24100121 5 5 1 1 1 1 1 1 1
24100122 1 1 5 5 5 5 5 5 5
24100123 . . . . . . . . .
24100125 . . . . . . . . .
24100129 4 4 1 1 1 1 5 1 1
24100130 1 1 5 5 5 5 5 5 5
24100132 1 1 5 5 5 5 5 5 5
24100134 1 1 5 5 5 5 5 5 5
24100135 3 2 1 1 5 5 5 5 1
24100136 1 4 5 5 5 5 5 5 1
24100137 . . . . . . . . .
24100138 4 4 . . . . . . .
24100139 3 4 1 1 1 1 1 1 1
24100140 1 1 5 5 5 5 5 5 5
24100141 1 1 5 5 5 . . . .
24100142 4 3 5 1 1 5 1 5 5
24100143 1 1 5 5 5 . . . .
24100144 4 3 1 5 1 5 5 5 5
24100147 1 1 5 5 5 . . . .
24100148 . . . . . . . . .
24100151 5 5 1 1 1 1 1 1 1
24100153 3 3 5 5 5 . . . .
24100154 1 1 5 5 5 . . . .
24100157 1 1 5 5 5 . . . .
24100158 3 3 1 1 1 1 1 1 1
24100159 1 1 5 5 5 . . . .
24100162 1 1 5 5 5 . . . .
24100164 4 3 1 1 1 1 1 1 5
24100165 4 3 1 1 1 5 1 1 1
24100166 2 2 5 5 5 . . . .
24100167 5 3 1 1 5 5 5 1 1
24100168 1 1 5 5 5 . . . .
24100170 4 4 1 1 5 5 5 1 5
24100171 2 2 5 5 5 . . . .
end

[Ken Chui]

The first is weighted mean. For example, for the last case, the non-missing mean will be (2+2+5+5+5)/5. However, it'd have a missing for the "no missing" mean because it does not have all the 9 answers.

egen validnum = rownonmiss(var_30-var_slow)
egen mean_non_miss = rowmean(var_30-var_slow)
egen mean_no_miss  = rowmean(var_30-var_slow) if validnum == 9

With that you can carry on with the standardization.

[Muhammad Ramzan]

thanks what is the difference between rowmean and rmean please

[Ken Chui]

thanks what is the difference between rowmean and rmean please

https://www.stata.com/statalist/arch.../msg00087.html

Same, please an older exchange above. I believe rowmean() is the formally documented term.