A) Form expectation about sign of coefficients
W = 0 + 1 S + 2 D + 3 L +
1 > 0
2 > 0
3 > 0
B) Estimate model and write the result in standard form
W = -478.607 – 12.8052 S + 1.89576 D + 35.9569 L
[131.5] [1.350] [0.4624] [2.881]
C) Interpret all coefficients
If number of student in polish (s) goes up by 1 unit (thousands) then number
of people with collage degree (w) will go down on average by 12.8059
If number of GNP dynamics (D) goes up by 1 unit (thousands) then number of
people with collage will go up on average by 1.8957
If number of people aged 20-24 (L) goes up by 1 unit (thousands) then
number of people with collage will go up on average by 35.9569
Variables (S,D,L)if the all independent variable are equal to 0, then W equal to
-478.607
D) Interpret all standard errors
When we calculate our coefficient 0 at value -478.607 we on average make
mistake about 131.5
When we calculate our coefficient 1 at value -12.8059 we on average make
mistake about 1.35
When we calculate our coefficient 2 at value 1.8957 we on average make
mistake about 0.4624
When we calculate our coefficient 3 at value 35.9569 we on average make
mistake about 2.881
E) Check precision of estimate of S and L
Si Standard Error
Si= | |* 100 = | |
I Coefficient
1.350
Ss= * 100 = 10.543 % 12.8052
2.881
SL= * 100 = 8.012 % 35.9569
F) Interpret and comment R^2
_ K (no of independent )
R2 = R2 - * (1- R2)
N –(K+1)
(no of observation)
_ 3
R2 = 0.964 - * (1- 0.964) = 0.959 => 95.9 %
25-3-1
R2 = 0.964
0.964 > 0.7 is good model
It means 96% of dependent variable is explained by independent variable and
is good model
R2 > 0.7 is good model
R2 < 0.5 is bad model
0.5 < R2 < 0.7 is in average
G) Test significance of L,D and of constant term
H0: 1=0 if parameters 1 is equal to zero variable j is not significant in the model.
H1: 1=/= 0 if parameters 1 is different to zero variable j is significant in the
model.
H) Test significance of all variables in the model
I) Test for autocorrelation
J) Check coincidence