Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Multinomial logit

From   Klaus Pforr <>
Subject   Re: st: Multinomial logit
Date   Mon, 25 Feb 2013 22:04:41 +0100

dear Andrei,

I enjoy your strategy to first blame the method rather the theory behind your statistical model very much, but I also want to try to give you a helpful answer. At first I want to ask about your data structure to clear things up. You seem to have panel data:

(read the following in fixed width format):

id | t | employment status| temp(t) | temp(t-1) |
1 | 1 | 1=temp | 1 | . |
1| 2 | 2=self | 0 | 1 |
1 | 3 | 3=unempl | 0 | 0 |
1 | 4 | 4=perm empl | 0 | 0 |

You would have variables like
ID, T, EMPL_T0(employment status), TEMP_T0, TEMP_Tm1, SELF_T0, SELF_Tm1, UNEMP_T0, UNEMP_Tm1, PERM_T0, PERM_Tm1.

You seem to want to analyze something like this
mlogit EMPL_T0 SELF_Tm1 UNEMP_Tm1 PERM_Tm1.

Considering the substantive contant of your variables, you might want to think about using survival analysis models. You basically are interested in the transition rates from temp. employment into other states. The different outcome are competing risks. You could include individual heterogeneity by using frailty models.

best wishes


Am 25.02.2013 18:26, schrieb Aндрей Протасов:
Dear STATA users,

I am working on multinomial logit model.

I would like to estimate the probability of leaving temporary employment in one year.

By now, I have generated 4 variables for (t+1) year.

temporary emp (1- temporary, 0 - others(self-emp, unemployment, permanent emp))
self-employment (1-self, 0 -others)
unemployment (1-unem, 0 - others)
permanent employment (1 - perm employment, 0 - others)

Do i have to create a dependent variable, which will include all  these categories to use in my regression?

Next_year (self-emp=1, perm_emp=2, unemp=3, temp emp=4)?
So, base is temp emp.

But, I am not sure if this approach is correct cause the coefficients still do not satisfy my hypothesis....

Thank You very much in advance for help.

Kind regards,
Andrei Protasov
*   For searches and help try:

Klaus Pforr
GESIS -- Leibniz Institut für Sozialwissenschaft
Postfach 122155
D - 68072 Mannheim
Tel: +49 621 1246 298
Fax: +49 621 1246 100

*   For searches and help try:

© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index