APPENDIX 1 (1)

SAMPLE MEASUREMENTS

Mean x=
x
n

( x) 
  2

Standard deviation
s=
1 

n −1 x − n  or
2

 

s=
1
n −1

( x − x)2 
s
Coefficient of Variation CV =  100%
x

Coefficient of Skewness =
Pearson’s Measure of Skewness
3(mean − median) mean− mode
OR
s tandard deviation s tandard deviation
 APPENDIX 1 (2)

CONFIDENCE INTERVAL

Parameter and description A (1 - ) 100% confidence interval

Mean , for large samples, s
x  z 2
σ2 unknown n
Mean , for small samples, x  t 2
s
; df = n – 1
σ2 unknown n
1 1
( x1 − x 2 )  t  2 sp + ; df = n1 + n2 – 2
n1 n2
Difference in means of two normal
distributions, 1 - 2
12 = 22 and unknown (n1 − 1)s12 + (n2 − 1)s22
sp =
n1 + n2 − 2

s12 s2
( x1 − x2 )  t  2 + 2 ;
n1 n2
2
Difference in means of two normal s12 s22 
 n +
distributions, 1 - 2 ,
 1 n2 
12  22 and unknown df = 2 2
 s12   s22 
   
 n 1  + n 2 
n1 − 1 n2 − 1

sd
Mean difference of two normal d  t 2 ; df = n – 1 where n is no. of
distributions for n
paired samples, d pairs
 APPENDIX 1 (3)

HYPOTHESIS TESTING

Null Hypothesis Test statistic

H0 :  = 0 x − 0
z=
σ2 unknown, large samples s n
H0 :  = 0 x − 0
t= ; df = n – 1
σ2 unknown, small samples s n
( x 1 − x 2 ) − ( 1 −  2 )
t= ; df = n1 + n2 – 2
1 1
sp +
H0 : 1 - 2 = 0 n1 n 2
12 = 22 and unknown (n1 − 1)s12 + (n2 − 1)s22
sp =
n1 + n2 − 2

( x1 − x 2 ) − (1 −  2 )
t=
s12 s 22
+
n1 n2
H0 : 1 - 2 = 0 s12
2
+ s22 
12  22 and unknown  n1 n2 
df = 
2 2
 s12   s22 
   
 n1  +  n2 
n1 − 1 n2 − 1
d − d
H0 : d = 0 t= ; df = n – 1, where n is no. of pairs
sd n

(oij − eij )2
Hypothesis for categorical data 2 =  eij
 APPENDIX 1 (4)

ANALYSIS OF VARIANCE FOR A COMPLETELY RANDOMIZED DESIGN

Let:
k = the number of different samples (or treatments)
ni = the size of sample i
Ti the sum of the values in sample i
=
n = the number of values in all samples
= n1 + n2 + n3 +...
x = the sum of the values in all samples
= T1 + T2 + T3 + ...
x 2
= the sum of the squares of values in all samples

Degrees of freedom for the numerator = k – 1

Degrees of freedom for the denominator = n – k

Total sum of squares: SST = x 2

−
(  x) 2

n
Between-samples sum of squares:

 T12 T22 T32


SSB=  + + + ...  −

 ( x)2
 n 1 n 2 n 3  n

Within- samples sum of squares = SST - SSB

SSB
Variance between samples: MSB=
(k −1)
SSW
Variance within samples: MSW=
(n − k )
MSB
Test statistic for a one-way ANOVA test: F =
MSW

M
u
k
a
s
u
r
a
t
 APPENDIX 1 (5)

SIMPLE LINEAR REGRESSION

Sum of squares of xy, xx, and yy:

SSxy =  xy−
 x)(y)
(
n
( x ) 2
y) 2
= x − = y −
(
2 2
SSxx and SSy y
n n

Least Square Regression Line:

Y = a + bx

Least Squares Estimates of a and b:

SSxy
b= and a = y − bx
SSxx

Total sum of squares: SST= y −

 y) 2
( 2

n
SSxy
Linear correlation coefficient: r =
SSxxSSy y