0
$\begingroup$

Supposing I have a Null hypothesis $H_{0}$ that my data comes from model $M_{1}$, with parameter $\theta = \theta_{1} $ (so that the model is parametrized as $M(\theta_{1})$ ), and an alternative hypothesis $H_{1}$ for model $M_{1}(\theta)$ where $\theta = \theta_{2}$.

Can I use the likelihood ratio test in case like this, where I am comparing the likelihood between two models?

If not, can I use Bayes factor to do the hypothesis test as alternative ?

$\endgroup$
1
$\begingroup$

That is what a likelihood ratio test is used for: to compare the fit of two models. Suppose we have some parameter space $\Theta$. Let $\Theta_0 = \{\theta_1 \}$. Then the hypothesis test would be $$H_{0}: \theta \in \Theta_0 \\ H_{1}: \theta \notin \Theta_0$$

Then consider $L(\theta|x)$ and the likelihood ratio statistic $$\Lambda(x) = \frac{\sup \{L(\theta|x): \theta \in \Theta_0 \}}{\sup \{L(\theta|x): \theta \in \Theta \}}$$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.