# minimal sufficient statistics

#### Matan1692

##### New Member
Hello all.
Is the theorem about minimal sufficient statistics by the factorization theorem is a statement of if and only if?
Let theta be an unknown parameter, if for sufficient statistic s(x) i previously found, and for the likelihood's ratio of L(x,theta)/L(y,theta).
If i show that s(x)=s(y) force the likelihood ratio be independent of theta, will i prove minimal of the sufficient statistic? (the opposite direction is of course: the likelihood ratio will force s(x)=s(y))
proofs will be welcomed
Thanks allot

#### Buckeye

##### Member
it is if and only if. There is another theorem as well: The exponential families have complete, sufficient statistics. And any estimator based off these statistics is UMVUE.

#### Matan1692

##### New Member
Thank you, where can i find a proof for this theorem?