I’m having trouble figuring this out. I’ve read the book and watched the lectures but it’s more in the application of determining what I’m supposed to use when answering a question/problem.
What are the queues or keys that you use when reading through a problem to determine what to use between a one or two tailed test?
I choose the wrong one basically every single time, just don’t get it.
“Cues”, not “queues”.
The question to ask yourself is whether the null hypothesis is that some parameter is equal to a given value (2-tailed test), as opposed to being greater than or equal to, or less than or equal to, a given value (both 1-tailed tests). Some of that can be chalked up to common sense, and some comes simply from practicing a number of problems.
As an example, suppose that Bob, a competitor of yours, says that his fund returns 1% per month. It would be silly to think that he means exactly 1%, no more, no less, so you presume that he means 1% or more. To prove Bob wrong, you conduct a 1-tailed test.
Thank you for the explanation. Can’t believe I used queue there. The burnout must be setting in for the day…
It happens to all of us.
Best of luck! Hang in there!
Just as a follow-up; one of the questions I’m looking at discusses the effect that a debt ratio has on the short interest ratio.
“…She also estimates a simple regression to investigate the effect of the debt ratio on a company’s short interest ratio.”
There isn’t any other information that would swing me one way or another on whether that would be a one-tail or two-tail test. But, based on the explanation you gave me, using common sense my thought is that effect can be viewed similar to how we see an equal sign, and thus make it a two-tailed test. Sure enough, I answered the question with this in mind and got it right this time…
Good job!
I would read that as asking whether there’s no effect vs. some effect (which could be positive or negative).