Dear Risqia;

The choice of the significance level is a decision of the researcher.

Usually is 5%.

I suggest you use information criterion (AIC, BIC...), and out of sample tests to choose the best model.

Hope this helps.

Fernanda Maria Müller

Hi Fernanda, I have checked the AIC and log-likelihood if I dropped out the insignificant parameter, and the result is the model without those insignificant parameters 'worse' than the model with those insignificant parameters because have bigger AIC and smaller loglikelihood..

I agree with you, so I can still use the model because the first model have better information criterion ...

Thanks for your help Fernanda, I really appreciated it..

Risqia Fadhilah Syahrir.

Hi Dear,

A 10% significance level is also a valid significance level (usually refereed to as a 1 star significance level), you can use the model with mu which solves your problem.

Thanks Mohamed for your suggestion, yeah I used 10% significance level instead of 5%