You are on page 1of 5

Danielle Ann A.

Ranido March 2,
2017
11417110
EXPEDES

a.

Let = mean decrease in pain intensity due to drug i where i= A, B, C, &


D
H0: A =B =C = D (The four treatment means are all equal)
Ha: at least one i is different (The treatment means are not all
equal)
=0.0 5

Source DF Sum of Squares Mean Square F Value Pr > F


Model 12 169.0540000 14.0878333 52.36 <.0001
Error 27 7.2650000 0.2690741
Source DF Sum of Squares Mean Square F Value Pr > F
Corrected Total 39 176.3190000

In the table above, it was shown that the p-value is less than .
Therefore, we have sufficient evidence to reject the null hypothesis. The
means decrease in pain intensity are not all equal. There is a drug effect on
the reduction of pain intensity.

b.

Comparisons significant at the 0.05 level are indicated by ***.


drug Difference Simultaneous 95% Confidence
Comparison Between Limits
Means
D-B 0.5400 -0.0372 1.1172
C-B -0.5200 -1.0972 0.0572
A-B -1.0800 -1.6572 -0.5028 ***
Using drug B as a control, it was shown in the table that it has a
significant difference in terms of mean decrease in pain intensity with drug A.

c.

Sphericity Tests
Variables DF Mauchly's Criterion Chi-Square Pr > ChiSq
Transformed Variates 5 0.0032945 44.136315 <.0001
Orthogonal Components 5 0.0132156 33.4091 <.0001

MANOVA Test Criteria and Exact F Statistics for the Hypothesis of no DRUG
Effect
H = Type III SSCP Matrix for DRUG
E = Error SSCP Matrix
S=1 M=0.5 N=2.5
Statistic Value F Value Num DF Den DF Pr > F
Wilks' Lambda 0.20840667 8.86 3 7 0.0088
Pillai's Trace 0.79159333 8.86 3 7 0.0088
Hotelling-Lawley 3.79831100 8.86 3 7 0.0088
MANOVA Test Criteria and Exact F Statistics for the Hypothesis of no DRUG
Effect
H = Type III SSCP Matrix for DRUG
E = Error SSCP Matrix
S=1 M=0.5 N=2.5
Statistic Value F Value Num DF Den DF Pr > F
Wilks' Lambda 0.20840667 8.86 3 7 0.0088
Pillai's Trace 0.79159333 8.86 3 7 0.0088
Trace
Roy's Greatest 3.79831100 8.86 3 7 0.0088
Root

Source DF Type III SS Mean F Pr > F Adj Pr > F


Square Value G-G H-F
DRUG 3 14.47500000 4.82500000 17.93 <.0001 0.0014 0.0012
Error(DRUG) 27 7.26500000 0.26907407

Greenhouse-Geisser Epsilon 0.3712


Huynh-Feldt Epsilon 0.3861
The MANOVA test statistics clearly suggest the significance of the drug
effect among the subjects. This result is validated in the univariate ANOVA
result above indicating that the mean numbers of the decrease in pain
intensity of the subjects are not all equal. The adjusted p-value was used
since the assumption of sphericity was violated as indicated in the Mauchlys
spericity test.

SAS CODES
data eight;
input subject drug $ pain;
datalines;
1 A 1.6
1 B 3.4
1 C 2.7
1 D 4.3
2 A 2.3
2 B 5.1
2 C 4.2
2 D 6.5
3 A 4.2
3 B 5.3
3 C 4.6
3 D 6.0
4 A 7.1
4 B 8.9
4 C 7.8
4 D 9.4
5 A 3.5
5 B 3.7
5 C 3.4
5 D 3.9
6 A 5.8
6 B 6.5
6 C 6.2
6 D 7.1
7 A 4.9
7 B 5.6
7 C 5.4
7 D 6.2
8 A 6.0
8 B 7.2
8 C 6.3
8 D 7.3
9 A 1.2
9 B 1.4
9 C 1.3
9 D 1.7
10 A 2.7
10 B 3.0
10 C 3.0
10 D 3.1
;
run;

proc print data=eight;


run;

proc glm data=eight;


class subject drug;
model pain = subject drug / solution;
means drug / tukey dunnett('B')
hovtest = bartlette hovtest = levene alpha = 0.05;
run;

data eight2;
input subject druga drugb drugc drugd;
datalines;
1 1.6 3.4 2.7 4.3
2 2.3 5.1 4.2 6.5
3 4.2 5.3 4.6 6.0
4 7.1 8.9 7.8 9.4
5 3.5 3.7 3.4 3.9
6 5.8 6.5 6.2 7.1
7 4.9 5.6 5.4 6.2
8 6.0 7.2 6.3 7.3
9 1.2 1.4 1.3 1.7
10 2.7 3.0 3.0 3.1
;
run;
proc print data=eight2;
run;

proc glm data=eight2;


class subject;
model druga drugb drugc drugd = / nouni;
repeated DRUG 4 / mean printh printe printm summary;
run;

You might also like