Reading quant right now. Had a quick question: What is the relevance of seeing whether any coefficient is significantly different from zero? The formula is (B- 0)/Standard error. i keep seeing this mentioned as the test to see the significance of a variable but not sure how it keeps relating to “zero”. thanks!
If a coefficient is significantly different from zero, then you can conclude that that variable is significant in the regression. If it weren’t significantly different from zero, you could conclude that that variable is not important in explaining the dependent variable.
the formula is t= (expected value of b1 - b1)/Sd Err we are testing for the coeff to be different from zero ie H0 : b1=0 & Ha: B1 not eqaul to 0
simi it is the other way around (b1 - expected value of b1) / std err. expected value of b1 = 0 if you expect b1 to be insignificant.
CP, Just checked it over… b0 & b1 are the hypothesized values of the parameter & t= estimated value- hypothesized value/ Std Err… Thanks though, will now remember it for the exam 
yes but conceptually, what makes zero such a good number to compare for significance?
iregula Wrote: ------------------------------------------------------- > yes but conceptually, what makes zero such a good > number to compare for significance? If the coefficient on an independent variable was zero, changes in the value of the independent variable would have no effect on the value of the dependent value. In other words, it would be unrelated to the dependent variable. Of course, in other cases, you may be testing a different hypothesis, in which case your null will be different.
busprof, thank you very much for this. i should have figured this on my own, although i think the sentences you mentioned above would have made a world of a difference if they included it in the reading.