# Hypothesis test

An analyst constructed the following hypothesis test H0: b=0 H1: b > 0 The null hypothesis means a the dependant variable is sensitive to changes in the independant variable b the independant variable is sensitive to changes in the dependant variable c changes in the dependant variable do not explain changes in the independant variable d changes in the independant variable do not explain changes in the dependant variable

since regression equation is Y = a + bx and here we are saying b=0 Changes in independent variable do not explain changes in the dependent variable… Basically we are saying Y is a constant? Choice D

you are right.

agree with the reasoning, pretty straightforward. Aren’t the null and the alternative supposed to cover the entire universe of possible outcomes though?

lola Wrote: ------------------------------------------------------- > Aren’t the null and the alternative supposed to > cover the entire universe of possible outcomes > though? No - otherwise you wouldn’t have the choice of 1 or 2 tailed tests.

vickyhall4b Wrote: ------------------------------------------------------- > lola Wrote: > -------------------------------------------------- > ----- > > Aren’t the null and the alternative supposed to > > cover the entire universe of possible outcomes > > though? > > No - otherwise you wouldn’t have the choice of 1 > or 2 tailed tests. um, no. got nothing to do with it. two-tailed: Hnull doesnt equal something; Ha equals the same something one-tailed: Hnull is smaller(bigger) or equal to something; Ha is bigger(smaller) than the same something both tests cover the entire universe of outcomes

lola Wrote: ------------------------------------------------------- > vickyhall4b Wrote: > -------------------------------------------------- > ----- > > lola Wrote: > > > -------------------------------------------------- > > > ----- > > > Aren’t the null and the alternative supposed > to > > > cover the entire universe of possible > outcomes > > > though? > > > > No - otherwise you wouldn’t have the choice of > 1 > > or 2 tailed tests. > > um, no. got nothing to do with it. > two-tailed: Hnull doesnt equal something; Ha > equals the same something Got that reversed. > one-tailed: Hnull is smaller(bigger) or equal to > something; Ha is bigger(smaller) than the same > something > > both tests cover the entire universe of outcomes

JoeyDVivre Wrote: ------------------------------------------------------- > Got that reversed. > yeah, good pickup. thnx

just to clarify to reject the null hypothesis does the t-stat you calculate have to be outside the critical point or inside?

Outside…