I am beginning my thesis and I need some advice.
I am trying to estimate a probit model. The binary dependent variable is employment status and the independent variables include: network size, age, years since migration, network quality.
There are 18,000 migrants in the sample with ID''s, and there is a year variable (2001-11).
network size refers to the number of migrants living in the same area as that migrant
I have annual information (2001-11) for network size, age, years since migration, but only 2011 information for employment status and network quality.
So it a very unbalanced panel
Given this, can I estimate this model using an ordinary probit command?
"probit empstatus nwsize age years nwquality if migrant == 1"
I do not want to use random effects etc by using xtprobit
or, will I have to create 11 network size variables: nwsize01 nwsize02, ...., nwsize11
and regress
"probit empstatus nwsize01 nwsize02, ...., nwsize11 age years nwquality if migrant == 1"
Taking away the panel aspect.
Or, maybe both are possible?
Any help would be appreciated,
Dwayne.