Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.

# st: Using margins and marginsplot with the ologit command‏

 From Thomas Boquet To Subject st: Using margins and marginsplot with the ologit command‏ Date Fri, 17 Aug 2012 15:14:10 -0400

```Hello everybody,

I'm actually working with the command margins after ologit for a class exemple and I have a few questions on the interpretation of the output when dydx and at are used. Here is my code, I'm using the database attend.dta from the manual Econometric Analysis of Cross Section and Panel Data by Jeffrey Wooldridge avalaible on the Stata website :

set more off
set linesize 230

use "http://www.stata.com/data/jwooldridge/eacsap/attend.dta";, clear

gen termgpac=1 if termgpa>=0 & termgpa<=1
replace termgpac=2 if termgpa>1 & termgpa<=2
replace termgpac=3 if termgpa>2 & termgpa<=3
replace termgpac=4 if termgpa>3 & termgpa<=4
replace termgpac=. if termgpa==.
label variable termgpac "GPA , 1: entre 0 et 1, 2: entre 1 et 2, 3: entre 2 et 3, 4: entre 3 et 4"

gen hwrtec=1 if hwrte>=0 & hwrte<=25
replace hwrtec=2 if hwrte>25 & hwrte<=50
replace hwrtec=3 if hwrte>50 & hwrte<=75
replace hwrtec=4 if hwrte>75 & hwrte<=100
replace hwrtec=. if hwrte==.

ologit termgpac c.priGPA c.atndrte i.hwrtec i.frosh

margins, dydx(*) predict(outcome(4))

margins, dydx(*) atmeans  predict(outcome(4))

margins, dydx(*) at((median)c.priGPA c.atndrte) predict(outcome(4))

margins i.hwrtec, dydx(priGPA) at(priGPA=(1(0.2)4)) predict(outcome(1))
marginsplot, recast(line) recastci(rscatter) title(Effet marginal à la moyenne de atndrte) yline(0) xlabel(, angle(90)) ylabel(, angle(0)) ytitle("")

Q1 : I would like to know if the output of margins with the option at((median)..... is the average calculation when priGPA and atndrte are at their median.

Q2 : I'm not sure if it's the correct calculation of the average marginal effect of priGPA for different values of the variable (last calculation and graph).

I'm then showing a twoway table with the observed dependent variable against the prediction of the model.

predict p1 p2 p3 p4

gen termgpachat  = 1 if( p1>p2 & p1>p3 & p1>p4) & e(sample)
replace termgpachat = 2 if( p2>p1 & p2>p3 & p2>p4) & e(sample)
replace termgpachat = 3 if( p3>p1 & p3>p2 & p3>p4) & e(sample)
replace termgpachat = 4 if( p4>p1 & p4>p2 & p4>p3) & e(sample)
replace termgpachat=. if termgpac==. & e(sample)

tab termgpac termgpachat, row nofreq

Q3 : Are the values in percentage in the diagonal, the percentage of correctly classified observations?

Thank you very much

Thomas BOQUET

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```