Statistical Significance

Hi guys,

I am really struggling with Quant these days… and the source of confusion at times is the material itself…

Can anyone help me explain what statistical significance is all about? Can we do statistical significance of both intercept term and coefficient?


yes, you can. they are different tests though. I am at work can’t tell you which tests though.

Independent significance of intercept and slope is done using a t-test.

Joint significance of intercept and slope is done using the F-test.

Determining significance basically boils down to check whether the coefficient in question differs from a required (usually 0) value with a pre-set certainty.

You use an F-test to determine whether the slope is not zero (in a simple regression) or slopes are not all zero (in a multiple regression); _ the F-test does not include the intercept _. Indeed, there is no single test for whether the slope(s) and the intercept are not all zero.

what about the p-test? Can’t even use that for the intercept?

You’ll usually calculate a t-statistic for the intercept. You can either calculate the t-critical values using α and compare the t-statistic to the t-critical values, or they’ll give you a p-value and you can compare that to α.

Thanks guys!

I am gradually making sense of it.