Triangular distribution for random parameters
Posted: 31 Mar 2023, 09:45
Hi,
I'm estimating a mixed multinomial logit assuming a triangular distribution for the random parameters. Also, I keep two parameters fixed at 0: one on the upper bound, the other on the lower bound.
However, I have a problem with the random coefficient specification and interpretation.
In particular, if I don't divide the following formula by two, do I get half the range instead of the full range? Does the constrained bound affect the specification?
randcoeff[["x"]] = b_x_a + b_x_b * (draws_x_a + draws_x_b)
Thank you,
Marta
I'm estimating a mixed multinomial logit assuming a triangular distribution for the random parameters. Also, I keep two parameters fixed at 0: one on the upper bound, the other on the lower bound.
However, I have a problem with the random coefficient specification and interpretation.
In particular, if I don't divide the following formula by two, do I get half the range instead of the full range? Does the constrained bound affect the specification?
randcoeff[["x"]] = b_x_a + b_x_b * (draws_x_a + draws_x_b)
Thank you,
Marta