Implication of Monotone Likelihood Ratio Property

triggg

New member
Joined
Dec 29, 2013
Messages
3
Dear all,

I am trying to prove the following, without success:

Consider a family of distribution functions that satisfy the MLRP property, i.e.

. . . . .\(\displaystyle \dfrac{\partial}{\partial x}\, \dfrac{f_t (x)}{f_{t'} (x)}\, \geq\, 0\)

or, equivalently,

. . . . .\(\displaystyle F_{t}(x)\, >\, F_{t'}(x)\)

. . . . .\(\displaystyle \forall\, x\, \mbox{ and }\, \forall \, t\, >\, t'\)

(where \(\displaystyle f\) is the pdf, and is assumed to exist, and \(\displaystyle F\) is the cdf; \(\displaystyle x\) belongs to the non-negative reals).

I want to show that

. . . . .\(\displaystyle \dfrac{\partial}{\partial\, t}\, \dfrac{\int_a^b \, \left(F_t (x)\, -\, F_{t'} (x)\right)\, dx}{F_t (b)\, -\, F_{t'} (b)}\, \geq\, 0\)

. . . . .\(\displaystyle \forall \, a\, \geq \, 0\, \mbox{ and }\, b\,>\,a\)

thanks for any advice you might give me!

G
 
Last edited by a moderator:
Top