is there any rule or pattern that how to set or select the "null hypothesis value"? as you use 0 as the null hypothesis value. As mostly we already know the mean or average value, so can we hypothesize the null value around the mean value?
The null hypothesis is not based on the estimates. It is an a priori belief you are trying to test. You can test any null you want. The closer to the estimate, the lower the t stat. There would be no point in choosing a null equal to the estimate, as the t stat would always be zero.
Thank you for posting this video. My only question is what if I have to do a one-tailed t-test for a non-zero beta (i.e. for e.g. beta1=1). I ask this because stata only shows 2-tailed t statistics for default bypothesis beta = 0.
Great I came across this video before my finals.
How did it go?
Thank you for this!!!!
brilliant... thanks
is there any rule or pattern that how to set or select the "null hypothesis value"? as you use 0 as the null hypothesis value. As mostly we already know the mean or average value, so can we hypothesize the null value around the mean value?
The null hypothesis is not based on the estimates. It is an a priori belief you are trying to test. You can test any null you want. The closer to the estimate, the lower the t stat. There would be no point in choosing a null equal to the estimate, as the t stat would always be zero.
what does a high p value for the constant mean?
Thank you for posting this video. My only question is what if I have to do a one-tailed t-test for a non-zero beta (i.e. for e.g. beta1=1). I ask this because stata only shows 2-tailed t statistics for default bypothesis beta = 0.
Oh, I actually figured that out by connecting dots with the help of your video. Thanks very much.
Can I compare 2 coefficients from diferent regressions?
Yes, as long as the dependent variable is the same.