1

I have obtained some parameter estimates and corresponding t-values. How can I use the t-values to say more about the significance of these estimates? For example, if t = \hat{\beta} / s.e.(\beta), then are larger values for t better or worse? What is the estimation error? I am confused in how to interpret their values.

Can someone recommend me a website or book where this is explained?

Thanks!

PS. If in the hypothetical equation Y = aX + e the error e is normally distributed with mean zero and variance (\sigma_e)^2, does that mean that a is normally distributed as well? Because from my understanding one needs a distribution for the parameter in order to apply a significance test, right?

Sean
  • 61
  • 6
  • This is pretty broad... I guess you should start with some (any) statistics handbook, since your question asks in general how does hypothesis testing work and this is covered in any introductory book. – Tim Jun 30 '18 at 12:11
  • Are you working on empirical data ? t- statistic is computed for different purposes e.g. to test difference between sample-statistics and population statistic. Also, you are talking about fixed-effects model. Try to indicate a concrete problem. –  Jun 30 '18 at 12:18
  • @subhash I am dealing with parameter estimates from a Kalman Filter procedure. I started from a theoretical model, obtained data, and estimated the parameters including unobserved states. The question now is, how 'valuable' are these parameter estimates? – Sean Jun 30 '18 at 15:28
  • What is extra or more in your mind ? Do you have anything to say on closing as duplicate. –  Jun 30 '18 at 16:27

0 Answers0