-Coefplot- with Multiple Models Output - Reshaping Graph

[Kirk Harvey]

Dear all my question is related to

coefplot

.
I ran a simple logit regression:

logit employ i.birth if id == "lu", or baselevels nolog
eststo lu
logit employ i.birth if id == "cy", or baselevels nolog
eststo cy
* etc...

where employ is a binary variable for being employed, and i.birth is a categorical variable covering regions of birth (AMERICA, ASIA, EUROPE, etc.) with the home country set as the base. The variable id represents specific countries (at, be, de, uk, etc.).
When I run coefplot:

coefplot lu cy be, label drop(_cons) eform xline(1)

I get this graph:

I would like to reshape this graph so that id (lu, cy, be, etc.) is on the y-axis and the coefficients for i.birth (AFRICA, ASIA…) are shown in this row (spreaded across x-axis).
I read forums related to coefplot but did not find a similar case.
This example is also not covered in the help documentation for coefplot.
Could someone please suggest a solution?

[Andreas Backhaus]

Try the

vertical

option as suggested by the coefplot help file.

[Kirk Harvey]

Hi Andreas,

thank you for your help, but I need something different.
When I use -vertical- I get this:

But i need something different that looks like this example:

Where the points America, Africa... represent coeffcients from the logit for particular countries (at, be, cy...).

[Kirk Harvey]

Here is the example:

employ    id    birth
1    at    NAT
0    at    NAT
1    at    NAT
0    at    NAT
1    at    NAT
1    at    NAT
1    at    NAT
1    at    NAT
1    at    NAT
1    at    NAT
0    at    NAT
0    at    NAT
0    at    NAT
0    at    NAT
0    at    NAT
0    at    NAT
0    at    NAT
0    at    NAT
1    at    ASIA
1    at    ASIA
1    at    ASIA
1    at    ASIA
1    at    ASIA
1    at    ASIA
1    at    ASIA
1    at    ASIA
0    at    ASIA
0    at    ASIA
1    at    ASIA
1    at    ASIA
1    at    ASIA
1    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    ASIA
0    at    EUROPE
0    at    EUROPE
0    at    EUROPE
1    at    EUROPE
0    be    EUROPE
0    be    EUROPE
0    be    EUROPE
0    be    EUROPE
0    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    EUROPE
1    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
0    be    ASIA
1    be    ASIA
1    be    ASIA
1    be    ASIA
1    be    ASIA
1    be    ASIA
1    be    ASIA
1    be    NAT
1    be    NAT
1    be    NAT
0    be    NAT
0    be    NAT
0    be    NAT
0    be    NAT
0    be    NAT
0    be    NAT
0    be    NAT
0    be    NAT
0    be    NAT
1    be    NAT
1    be    NAT
1    be    NAT
1    be    NAT
1    be    NAT
1    be    NAT
1    be    NAT

logit employ ib3.birth if id =="at", or baselevels nolog
eststo at

logit employ ib3.birth if id =="be", or baselevels nolog
eststo be

I wonder how to get results in a form like this:

[Andreas Backhaus]

Thanks for the clarification but I don't see what kind of information one would read from a plot like this, as there is no scale for the coefficients?

[Kirk Harvey]

Hi there should be a scale.
The best I could get so far is this graph.
But I would like to merge this graphs to one graph (have coefficients for ASIA, AFRICA) on one line for "ro", "se"... as they represent models for different states
. Is there any command that would allow that?

[Andreas Backhaus]

With the help provided here: https://repec.sowi.unibe.ch/stata/co...ted.html#h-3-6 and here: https://www.statalist.org/forums/for...ultiple-models, you should get close to something like you want. But note that plotting all regional coefficients on the line for e.g. ro would result in a huge overlap, as the coefficients are similar in magnitude. coefplot would add vertical space between each marker, but depending on the number of regions and countries, your graph might become very large in vertical direction.