Vous êtes sur la page 1sur 5

Danielle Ann A.

Ranido February 2,
2016

11417110
EXPEDES

Let = mean number of percent increase in sales by the display design i


where i= 1, 2, & 3
H0: 1=2= 3

Ha: at least one i is different


=0.0 5
H0 <0.05
Decision rule: Reject if
=0.05 , we reject H0
At . Thus, at least one the mean numbers of
percent increase in sales by the display design is different.
2
Let = variance of percent increase in sales by the display design i
where i= 1, 2, & 3
H0: 21= 22= 23

Ha:
at least one 2i is different
=0.0 5
H0 <0.05
Decision rule: Reject if
=0.05 , we fail to reject H0
At . Thus, there is no sufficient
evidence that suggests that the assumption of homogeneity is
violated.

For both the LSD and HSD test, it is observed that the display design 3 is
significantly different from display design 1 and 2. On the other hand, display
design 2 and display design are not significantly different from each other. As
for the Dunnetts t test, display design 2 was used as the control variable. It
is shown below that the display design 3 is significantly different from the
control variable which is display design 2.
Let = mean number of percent increase in sales by the display design i
where i= 1, 2, & 3
H0: i=i ' where i i'
Ha: i i '

From the results in the tests, it is shown that the display design 3 has
the highest mean among the three display designs. It is observed that
display design 2 and 1 are in the same group. It signifies that they are not
significantly different from each other. Therefore, the distributor can
interchangeably use display design 1 and 2. On the other hand, using display
design 3 will yield the best results.
H0: The errors follow a normal distribution

Ha: The errors do not follow a normal distribution

=0.0 5
H0 <0.05
Decision rule: Reject if

At =0.0 5 , we fail to reject the null hypothesis. There is no sufficient

evidence that suggests that the assumption of normality is violated.

Codes:
data assignment;
input design sales;
datalines;
1 5.43
1 5.71
1 6.22
1 6.01
1 5.29
2 6.24
2 6.71
2 5.98
2 5.66
2 6.60
3 8.79
3 9.20
3 7.90
3 8.15
3 7.55
;
run;
proc glm data = assignment;
class design;
model sales = design;
means design/duncan dunnett('2') tukey lsd hovtest=bartlett;
output out = results r = resid p = yhat
h = leverage dffits = dfit
covratio =covR cookd = cooks
rstudent = Rstudent student = student;
run;
proc print data = results;
run;
proc univariate data = results freq normal plot alpha = 0.05;
var resid;
run;

Vous aimerez peut-être aussi