The linear regression line is S=α+βP+E
E is random error
and E(s)=α+βP
SSS=∑S2=i=1∑8(Si−S)2=1205
SPP=∑P2=i=1∑8(Pi−P)2=55.9
SSP=∑(SP)=i=1∑8(Si−S)(Pi−P)=22.4
α=S−Pβ and β=(∑S2)1/2×(∑P2)1/2(∑SP) =(1205)1/2×(55.9)1/2(225.4)=0.8685
from the table S=∑Si/n=225/8=28.125
P=∑Pi/n=37/8=4.625
α=28.125−4.625×0.8685=24.1082
a) the estimated regression line is, S=24.1082+0.8685Pi
b) the standard error (SE) of α and β are
SE(α)=σ1/n+P2/Spp
and SE(α)=σ/Spp
σ2=1/(n−2)SSE=1/(n−2)[Sss−β2Spp]=1/(8−2)[1205−0.86852×55.9]
=1/6×1162.8351=193.8058
σ=193.8058=13.9214
SE(α)=σ1/n+P2/Spp=13.9214/55.9=13.9214/7.4766=1.86199
c) testing for hypothesis
H0:β=0 versus H1:β=0
at α=0.05
t=(β−0)/(SE(β))
t=0.8685/1.86199=0.4664
tcritical=2.4469
∣t∣<2.4469
we fail to reject H0 , that means at α=0.05,P doesn′t affect S
Comments