Combining information from two columns of a matrix into a variable

[Daniel Rosler]

Dear Statalist - users,

I'm working with the package ginidesc that i installed from ssc to compute gini-coefficients for subpopulations within countries.

Here are the commands I used and parts of the resulting output:

ginidesc income_equivalised if incrank_2 == 1, by(country) gk(i1)

Gini Coefficient by subgroups
of country

-----------------------------
               K |     Gini_k
-----------------+-----------
Austria          |      0.298
Belgium          |      0.228
Bulgaria         |      0.270
Cyprus           |      0.294
Czech Republic  |  0.290
Germany        |      0.330
Denmark        |      0.284
Estonia           |      0.222

Stored in matrix i2

I tried to store the Gini-coefficients in a variable with the following commands:

matrix list i1
svmat i1, names(inc)

This created 2 variables, one for each column, but the gini-coefficients are saved as the category and not the value of the variable. Also the information is now disconnected from the country - variable in the data set.

tab inc2

       inc2 |      Freq.     Percent        Cum.
------------+-----------------------------------
   .1230075 |          1        3.03        3.03
   .1402118 |          1        3.03        6.06
   .1471672 |          1        3.03        9.09
   .1478555 |          1        3.03       12.12
   .1688081 |          1        3.03       15.15
   .1694496 |          1        3.03       18.18
   .1719411 |          1        3.03       21.21
   .1736457 |          1        3.03       24.24
   .1751534 |          1        3.03       27.27

I would greatly appreciate if someone could tell me how to create one variable that has the same gini-coefficient for each observation within a country. Thanks a lot in advance!

[Andrew Musau]

It is difficult to follow what you want as the output indicates that the Gini coefficients are stored in matrix i2 and you svmat matrix i1. Most people including myself do not use this command so I suggest that you do the following:

ginidesc income_equivalised if incrank_2 == 1, by(country) gk(i1)
mat l i1 
mat list i2
clear
svmat i1
svmat i2
dataex

Then copy and paste the outputs of the matrix list commands and the dataex.

[Daniel Rosler]

Thank you for your reply Andrew. Indeed, I have mixed up the annotation. I have meant to only include the matrix i1.

Here is the output to the commands you suggested:

. mat l i1

i1[33,2]
           k     Gini_k
k         29  .38146576
k         33  .36469457
k         32  .30149865
k         19  .28763977
k         24  .24093089
k         17  .28083768
k          3  .26932833
k         35   .2533761
k          9  .24810015
k         26  .19667381
k         23   .2387253
k          8  .23858768
k         10  .20662333
k         14  .20646381
k         13   .2387725
k         27   .1986126
k         22  .21360622
k         36  .18896493
k          4  .19449703
k         16   .2619614
k         20  .20656714
k          5   .1694496
k         15   .1751534
k         28  .18700944
k         12  .16880809
k         21  .22658722
k          2   .1736457
k          1  .18417412
k          6  .14785549
k          7  .17194112
k         11  .14716721
k         25  .14021179
k         18  .12300753

. clear

. svmat i1
number of observations will be reset to 33
Press any key to continue, or Break to abort
Number of observations (_N) was 0, now 33.

* Example generated by -dataex-. For more info, type help dataex
clear
input float(i11 i12)
29  .3814658
33  .3646946
32 .30149865
19 .28763977
24  .2409309
17 .28083768
 3 .26932833
35  .2533761
 9 .24810015
26  .1966738
23  .2387253
 8 .23858768
10 .20662333
14  .2064638
13  .2387725
27  .1986126
22 .21360622
36 .18896493
 4 .19449703
16  .2619614
20 .20656714
 5  .1694496
15  .1751534
28 .18700944
12  .1688081
21  .2265872
 2  .1736457
 1  .1841741
 6 .14785549
 7  .1719411
11  .1471672
25 .14021179
18 .12300753
end

[Andrew Musau]

but the gini-coefficients are saved as the category and not the value of the variable.

I do not understand this statement. The variable i12 appears to be continuous to me and not a categorical variable. This will merge the country names with the matrix output.

preserve
decode country, g(name)
contract country name
tempfile names
save `names'
restore
ginidesc income_equivalised if incrank_2 == 1, by(country) gk(i1)
mat l i1
clear
svmat i1
rename (i11 i12) (country gini)
merge 1:1 country using `names', nogen
l, sep(0)

[Daniel Rosler]

Your approach has worked! Thank you so much, Andrew.
What I meant with categorical is that the gini-coefficients after svmat were stored as the categories of a variable and not as the values or frequencies of it. Like here:

 tab i12

        i12 |      Freq.     Percent        Cum.
------------+-----------------------------------
   .1230075 |          1        3.03        3.03
   .1402118 |          1        3.03        6.06
   .1471672 |          1        3.03        9.09
   .1478555 |          1        3.03       12.12
   .1688081 |          1        3.03       15.15
   .1694496 |          1        3.03       18.18
   .1719411 |          1        3.03       21.21
   .1736457 |          1        3.03       24.24
   .1751534 |          1        3.03       27.27
   .1841741 |          1        3.03       30.30
   .1870094 |          1        3.03       33.33
   .1889649 |          1        3.03       36.36
    .194497 |          1        3.03       39.39
   .1966738 |          1        3.03       42.42
   .1986126 |          1        3.03       45.45
   .2064638 |          1        3.03       48.48
   .2065671 |          1        3.03       51.52
   .2066233 |          1        3.03       54.55
   .2136062 |          1        3.03       57.58
   .2265872 |          1        3.03       60.61
   .2385877 |          1        3.03       63.64
   .2387253 |          1        3.03       66.67
   .2387725 |          1        3.03       69.70
   .2409309 |          1        3.03       72.73
   .2481001 |          1        3.03       75.76
   .2533761 |          1        3.03       78.79
   .2619614 |          1        3.03       81.82
   .2693283 |          1        3.03       84.85
   .2808377 |          1        3.03       87.88
   .2876398 |          1        3.03       90.91
   .3014987 |          1        3.03       93.94
   .3646946 |          1        3.03       96.97
   .3814658 |          1        3.03      100.00
------------+-----------------------------------
      Total |         33      100.00