Interaction variables in OLS and FE

[Gabriela Kalibatseva]

Dear all,

I am running OLS and FE estimation. I have difficulties creating the below interactions of variables. I need to have a variable that resembles if the reporting person in a survey is head or spouse in a household where children live (therefore more likely those are their children) and interaction between log of income and being child of the head of household.

Any ideas?



Kind regards,
Gabriela

[Gabriela Kalibatseva]

Here is the example:

[vignon gibran]

français
Hello, My name is Gibran. I am doing a panel data study on the causes of deforestation in UEMOA.
N = 17 and i = 8. With the fixed effect regression I see that there is presence of individual effects in my panel since the p-value associated with the fisher statistic is 0.00. the hausman test tells me that he chooses the fixed effect model, but the R squared of this model is too small.
within = 45.4
between = 18
overall = 17
what to do? it disturbs it? I'd like your advice since this is for my memory license.

[vignon gibran]

français
Hello, My name is Gibran. I am doing a panel data study on the causes of deforestation in UEMOA.
N = 17 and i = 8. With the fixed effect regression I see that there is presence of individual effects in my panel since the p-value associated with the fisher statistic is 0.00. the hausman test tells me that he chooses the fixed effect model, but the R squared of this model is too small.
within = 45.4
between = 18
overall = 17
what to do? it disturbs it? I'd like your advice since this is for my memory license.

[Carlo Lazzaro]

# 3 and 4 are a duplicate posts; see, with reply, at: https://www.statalist.org/forums/for...f-fixed-effect

[Alfonso Sánchez-Peñalver]

To add to Carlo Lazzaro's point, not only #3 and #4 are duplicate posts, they don't belong here.

Hi Gabriela,

I am a bit confused of what you're trying to do and, most of all the variables you are showing. In the variable list you are showing there is no indicator that there are children in the household. The only indicator you have is whether the survey was responded by one of the household's children. Clearly not all of the surveys in houses with children were responded by children, so the variable that you show "Child of the Head of the Household" doesn't seem to work for what you express you want in #1. For that you need to have an indicator variable that reflects whether there are children present in the household, and I really don't see that in the list of variables you provide.

If I have misunderstood something about your post, I am sorry. Let me know.

Alfonso.