I'm having some trouble conceptualizing the following in a paper I've come across:

A remark is made on using a likelihood ratio test for the following hypothesis

$$\begin{align} H_0: \beta_{1,1} = \beta_{1,2} \\ H_{\alpha}: \beta_{1,1} \neq \beta_{1,2} \end{align} $$

where we have the model setup of

$$Y_{i,j} = \beta_{0,j} + \beta_{1,j}x_i + \epsilon_{i,j},$$

with $\epsilon_{i,j} \sim N(0, \sigma^2)$ for all $i = 1, \dots, n$ and $j \in \{1,2\}$.

I was under the impression that we could only do this for nested-models (i.e. one model has more parameters), or am I misinterpreting that/the model above? How can I specify the "full" model here to develop a test-statistic?

  • $\begingroup$ Can you add a full reference to the paper? $\endgroup$ Jan 4, 2021 at 23:25


Your Answer

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

Browse other questions tagged or ask your own question.