Issue with -churdle- command.

[Susan Paudel]

I'm using the following code:

churdle linear income_nonfarm_share_perT Age Gender married_unmarried above_primary literacy nadeq total_livestock_unit has_migrant area_in_ha i.agro_ecol, select(Age Gender married_unmarried above_primary literacy nadeq total_livestock_unit has_migrant area_in_ha i.agro_ecol) ll(0)

However, it is giving me an error saying "initial values not feasible". The histogram for the dependent variable is mostly distributed around 0 and 1.

The code runs if I change the dependent variable that is not heavily distributed around 1.

Any suggestion on why it's happening would be much appreciated!

Thanks!

[Leonardo Guizzetti]

On the little information provided, I do not think the -churdle- model is appropriate for your dataset. Certainly, having mostly values of 0/1 (and some in between) would be suggestive of some kind of -fracreg- or generalized linear model. These seem appropriate also if I interpret your dependent variable as a representing some kind of proportional share of income.

Maybe you can explain in words the general idea of what you are trying to do and provide us a data example following the instruction in the FAQ.

[Susan Paudel]

I am actually trying to find associates of the non-farm income share of households. However, there are households that do not take part in non-farm activities and do not have any non-farm income. So, I was trying to use double hurdle regression thinking variables influencing the initial decision to participate will be different than the variables affecting the income share after the participation.

My dataset has a total of 8388 households, of which 2300 households take part in non-farm activities. Households take part in various activities like the farm, wage, remittance, etc.

Will -fracreg- or generalized linear model work in this context too?

Thank you very much.

[Leonardo Guizzetti]

Yes, they still seem appropriate in this case. Consider the following fractional logistic regression model as a possible starting point.

fracreg logit income_nonfarm_share_perT Age Gender married_unmarried above_primary literacy nadeq total_livestock_unit has_migrant area_in_ha i.agro_ecol

[Susan Paudel]

Thank you very much. It worked.